From devi at tenet.res.in Tue Mar 1 10:04:05 2005 From: devi at tenet.res.in (devi) Date: Tue, 1 Mar 2005 12:04:05 -0300 Subject: [uClinux-dev] keyboard Message-ID: <20050301150405.30518.qmail@mail.tenet.res.in> hi, The output on the console is done in keyboard.c in the procedure handle_scancode(unsigned char scancode, int down). The codeline tty = ttytab? ttytab[fg_console]: NULL; which should get the tty for output. Do I need to add a device tty0 or vc/0 .When I added tty0 it says no such device.Can anyone help me to solve this problem.So that I can get the typed character displayed. Regards Devi From yashi at atmark-techno.com Tue Mar 1 00:28:25 2005 From: yashi at atmark-techno.com (Yasushi SHOJI) Date: Tue, 01 Mar 2005 14:28:25 +0900 Subject: [uClinux-dev] attached romfs with linux with mmu In-Reply-To: <1108717764.18774.76.camel@justakiss> References: <20050217051453.27C80E8FE@mail2.atmark-techno.com> <421585CD.6040508@snapgear.com> <1108717764.18774.76.camel@justakiss> Message-ID: <20050301052825.51598636A@mail2.atmark-techno.com> At Fri, 18 Feb 2005 10:09:24 +0100, Erwin Authried wrote: > > > I don't see any reason it can't be made to work... > > The only real downside is that the start up code > > usually needs to move the romfs around so that it > > can set up its bss. > It should be possible to adjust the bss with objdump so that it starts > after the romfs image so that the romfs can be left as is. yeah, it's possible. but that'd make the image larger. it's not always a good idea, but it'll boost the boot time on a system with slow memory. -- yashi From andrea.scian at wawnet.biz Tue Mar 1 02:31:47 2005 From: andrea.scian at wawnet.biz (Andrea Scian) Date: Tue, 01 Mar 2005 08:31:47 +0100 Subject: [uClinux-dev] Re: Porting a JVM & OSGi Framework on uClinux Message-ID: <6.2.1.2.0.20050301082232.01d14ba8@192.168.2.1> Hi, >Hi all , > > have someone already ported a JVM ( CDC )& an OSGi Framework on uClinux > ( ARM ??P ) , > If someone have more Information please contact me. We have successully ported Kaffe JVM (www.kaffe.org) to our ARM7TDMI/uClinux platform (B2). Kaffe was initially developed for x86/Linux platform but it's enough flexible to fit an embedded system too! :-) I have not tryied to port CDC due the fact that it's not an opensource project. Unfortunately I have never worked with OSGi Framework so I don't really know if it's needs some special feature of the JVM (is it pure Java code or has some native code too?). Maybe you can ask to Kaffe mailing list too. Best Regard, Andrea Scian From wnlee at cad.csie.ncku.edu.tw Tue Mar 1 05:38:07 2005 From: wnlee at cad.csie.ncku.edu.tw (Wayne Lee) Date: Tue, 1 Mar 2005 18:38:07 +0800 Subject: [uClinux-dev] almost boot Linux, but fail at "kernel BUG at kernel/exit.c:934!" after doing "/bin/expand" Message-ID: <200503011040.j21AeSfx008702@cad.csie.ncku.edu.tw> Hi, folks, I try to port uClinux 2.6.5 on an arm926 board. (without chip's MMU capability) However, it really disappointed me that uClinux 2.6.5 failed to boot after issuing this command: "/bin/expand /etc/ramfs.img /dev/ram1" The error message is: kernel BUG at kernel/exit.c:934! It is too difficult to figure out what is the kernel bug. Could anyone kindly give me some hints? I have ever adopted the same packages to port uClinux 2.6.5 on arm940t board. In this time, the main different is that arm926 cpu is configured, and "armv5te" compile option is used. The packages I adopted: uClinux-dist-20040408.tar.gz linux-2.6.5.tar.gz linux-2.6.5-hsc2.patch.gz (hyok.choi's patch) The toolchain I adopted: arm-elf-tools-20040427.sh The boot messages: Linux version 2.6.5 (root at wayne_linux) (gcc version 2.95.3 20010315 (release)(ColdFire patches - 20010318 from http://fiddes.net/coldfire/)(uClinux XIP and shared lib patches from http://www.snapgear.com/)) #116 Tue Mar 1 17:37:10 CST 2005 CPU: ARM926EJ-Sid(wb) [41069263] revision 3 (ARMv5TEJ) CPU: D VIPT write-back cache CPU: I cache: 32768 bytes, associativity 4, 32 byte lines, 256 sets CPU: D cache: 32768 bytes, associativity 4, 32 byte lines, 256 sets Machine: QPT200 board On node 0 totalpages: 16384 DMA zone: 0 pages, LIFO batch:1 Normal zone: 16384 pages, LIFO batch:4 HighMem zone: 0 pages, LIFO batch:1 Built 1 zonelists Kernel command line: root=/dev/ram0 initrd=0x02000000,800K keepinitrd PID hash table entries: 512 (order 9: 4096 bytes) Memory: 64MB = 64MB total Memory: 63248KB available (608K code, 81K data, 44K init) Calibrating delay loop... 0.88 BogoMIPS Dentry cache hash table entries: 8192 (order: 3, 32768 bytes) Inode-cache hash table entries: 4096 (order: 2, 16384 bytes) Mount-cache hash table entries: 512 (order: 0, 4096 bytes) checking if image is initramfs...it isn't (ungzip failed); looks like an initrd POSIX conformance testing by UNIFIX Linux NoNET1.0 for Linux 2.6 ikconfig 0.7 with /proc/config* Initializing QPT200 Internal UART ttyS0 at I/O 0x80000100 (irq = 17) is a QPT200 Internal UART RAMDISK driver initialized: 16 RAM disks of 8192K size 1024 blocksize loop: loaded (max 8 devices) RAMDISK: romfs filesystem found at block 0 RAMDISK: Loading 772 blocks [1 disk] into ram disk... |/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/ -\|/-\|/-\|done. VFS: Mounted root (romfs filesystem) readonly. Freeing init memory: 44K Shell invoked to run file: /etc/rc Command: hostname QPT-200 Command: /bin/expand /etc/ramfs.img /dev/ram1 init/expand.c argv[0] is [/bin/expand] argv[1] is [/etc/ramfs.img] argv[2] is [/dev/ram1] read(fdi,&len,4); fdi=[0] len=[34422656] len = ntohl(len); lseek(fdo, 0, SEEK_SET); fdo=[3] close fdi[0] close fdo[3] free(buf) exit(0); kernel BUG at kernel/exit.c:934! From geral at MicroD.info Tue Mar 1 07:02:26 2005 From: geral at MicroD.info (Fabio Martins) Date: Tue, 01 Mar 2005 09:02:26 -0300 Subject: [uClinux-dev] Palm and uclinux In-Reply-To: <6.2.1.2.0.20050301082232.01d14ba8@192.168.2.1> Message-ID: <42242FA2.9969.F09DF@localhost> Hi all, In that current Palms I can install uclinux? Thanks! Fabio Martins http://www.MicroD.info Date sent: Tue, 01 Mar 2005 08:31:47 +0100 To: Ayoub ZAKI From: Andrea Scian Copies to: uClinux development list Subject: [uClinux-dev] Re: Porting a JVM & OSGi Framework on uClinux > Mime-Version: 1.0 > Content-Transfer-Encoding: quoted-printable > Content-Type: text/plain; charset=iso-8859-1; format=flowed > > Hi, > > >Hi all , > > > > have someone already ported a JVM ( CDC )& an OSGi Framework on uClinux= > =20 > > ( ARM =C2=B5P ) , > > If someone have more Information please contact me. > > We have successully ported Kaffe JVM (www.kaffe.org) to our=20 > ARM7TDMI/uClinux platform (B2). > Kaffe was initially developed for x86/Linux platform but it's enough=20 > flexible to fit an embedded system too! :-) > I have not tryied to port CDC due the fact that it's not an opensource= > project. > > Unfortunately I have never worked with OSGi Framework so I don't really=20 > know if it's needs some special > feature of the JVM (is it pure Java code or has some native code too?). > Maybe you can ask to Kaffe mailing list too. > > Best Regard, > > Andrea Scian > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org > > > > -- > Nenhum v_rus encontrado nesta mensagem. > Checked by AVG Anti-Virus. > Version: 7.0.300 / Virus Database: 266.5.0 - Release Date: 25/02/2005 > From scoppys at yahoo.com Tue Mar 1 07:11:59 2005 From: scoppys at yahoo.com (scoppy scoppy) Date: Tue, 1 Mar 2005 04:11:59 -0800 (PST) Subject: [uClinux-dev] new user of uclinux, some questions! Message-ID: <20050301121159.71043.qmail@web51110.mail.yahoo.com> Hi i'm new, i ahve some questions about uclinux: 1) Can kernel run in flash rom or it must be full loaded in ram? 2) how cai i do to make linux running in flash rom? 3) 3 mbyte of rom ad 2 mbyte of ram is enaugth for run uclinux (i'm try uclinux for mips32 2.4.16) tnx for help --------------------------------- Do you Yahoo!? Yahoo! Mail - Helps protect you from nasty viruses. -------------- next part -------------- An HTML attachment was scrubbed... URL: From vanitha at agilis.stengg.com Tue Mar 1 15:09:37 2005 From: vanitha at agilis.stengg.com (Vanitha Ramaswami) Date: Tue, 1 Mar 2005 20:09:37 -0000 Subject: [uClinux-dev] How to compile NET-SNMP Subagent for Snapgear kernel Message-ID: <01C51E9A.98992F60.vanitha@agilis.stengg.com> Hi All, I am using snapgear 3.0 distro on my IXP425 custom platform. I have downloaded the 5.2.1 NetSNMP package. Can anydone tell me the options to run the configure script..Previously i have ported the libxml & libpopt to the snapgear kernel & then i ran the configure script from within the Makefile itself checking for the config.status, as done in open ssl libraries.. But i have some problem in using the same way for the Makefile for the user space application..Have anyone done this earlier..? Can anyone tell me how to compile the SNMP sub Agent for snapgear kernel. How to use the net-snmp-config script..i have some issues with that. The Net-SNMP 5.2.1 & the subagent is working fine for me on the i686 enivironment. Thanks Vanitha [This e-mail is confidential and may be priviledged. If you are not the intended recipient, please kindly notify us immediately and delete the message from your system; please do not copy or use it for any purpose, nor disclose its contents to any other person. Thank you.] ---ST Electronics Group--- From carlos at embedded.cl Tue Mar 1 07:57:24 2005 From: carlos at embedded.cl (carlos at embedded.cl) Date: Tue, 01 Mar 2005 12:57:24 -0000 Subject: [uClinux-dev] m68k gdb problem Message-ID: <20050301125727.E335C14D9FD@mail07.powweb.com> An embedded and charset-unspecified text was scrubbed... Name: not available URL: From jsujjavanich at syntech-fuelmaster.com Tue Mar 1 08:18:36 2005 From: jsujjavanich at syntech-fuelmaster.com (Jate Sujjavanich) Date: Tue, 1 Mar 2005 08:18:36 -0500 Subject: [uClinux-dev] m68k gdb problem Message-ID: This would be the 5235. Yesterday, I also tried GDB 5.3 with no BDM patches, but there was the same error. - Jate S. -----Original Message----- From: carlos at embedded.cl [mailto:carlos at embedded.cl] Sent: Tuesday, March 01, 2005 7:57 AM To: Jate Sujjavanich; uClinux development list Subject: RE: [uClinux-dev] m68k gdb problem Hi, what coldfire platform are you using? there is a problem with 5407 bdm, so if you're using a 5407 you may encounter this problem. (it is documented in processor's errata) cheers ---------- Original Message ------------- Subject: RE: [uClinux-dev] m68k gdb problem Date: Mon, 28 Feb 2005 10:04:30 -0500 From: "Jate Sujjavanich" To: "uClinux development list" I have been having precisely the same problem. One person in the past had the same error. I tried to get around this error by compiling gdb 6.3 without the bdm patches, and I was able to set and stop at a breakpoint. However, another error cropped up when I tried stepping, and I think it has to do with instruction misalignment. The error message is: "Program received signal SIGILL" This post is more detailed than my last one. Any tips? - Jate S. -----Original Message----- From: uclinux-dev-bounces at uClinux.org [mailto:uclinux-dev-bounces at uClinux.org] On Behalf Of Josef Baumgartner Sent: Monday, February 28, 2005 5:13 AM To: uclinux-dev at uclinux.org Subject: [uClinux-dev] m68k gdb problem Hi, I tried to use the new gdb from Bernardo Innocentis toolchain (m68k-uclinux-tools-gdb-glibc22-20040604.sh) and detected a problem. If i try to set a breakpoint and run the application gdb says "Software breakpoints not implemented for this target.". The same works with gdb5.0. I attached the output of the two gdb sessions below. Has anybody used the new gdb? Any idea, what could be wrong? Thanks, Josef ------------------------------------------------------------------------ ---------------- $ /usr/local/bin/m68k-bdm-elf-gdb GNU gdb 5.0 Copyright 2000 Free Software Foundation, Inc. GDB is free software, covered by the GNU General Public License, and you are welcome to change it and/or distribute copies of it under certain conditions. Type "show copying" to see the conditions. There is absolutely no warranty for GDB. Type "show warranty" for details. This GDB was configured as "--host=i686-pc-linux-gnu --target=m68k-bdm-elf". (gdb) file nml_hnd.gdb Reading symbols from nml_hnd.gdb...done. (gdb) target remote 10.30.55.104:3000 Remote debugging using 10.30.55.104:3000 0xdf0048 in _start () (gdb) b main Breakpoint 1 at 0xdf0446: file main.c, line 338. (gdb) c Continuing. Breakpoint 1, main (argc=2, argv=0xddff44) at main.c:338 338 int temp, option, debug_mask = DBG_MASK_NONE; (gdb) ------------------------------------------------------------------------ ---------------- $ /usr/local/m68k-uclinux-tools/bin/m68k-bdm-elf-gdb GNU gdb 6.1 Copyright 2004 Free Software Foundation, Inc. GDB is free software, covered by the GNU General Public License, and you are welcome to change it and/or distribute copies of it under certain conditions. Type "show copying" to see the conditions. There is absolutely no warranty for GDB. Type "show warranty" for details. This GDB was configured as "--host=i686-pc-linux-gnu --target=m68k-bdm-elf". (gdb) file nml_hnd.gdb Reading symbols from nml_hnd.gdb...done. (gdb) target remote 10.30.55.104:3000 Remote debugging using 10.30.55.104:3000 0x00e06048 in _start () (gdb) b main Breakpoint 1 at 0xe06492: file main.c, line 338. (gdb) c Continuing. Software breakpoints not implemented for this target. (gdb) _______________________________________________ uClinux-dev mailing list uClinux-dev at uClinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev at uclinux.org _______________________________________________ uClinux-dev mailing list uClinux-dev at uClinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev at uclinux.org From gerg at snapgear.com Tue Mar 1 08:36:30 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Tue, 01 Mar 2005 23:36:30 +1000 Subject: [uClinux-dev] new user of uclinux, some questions! In-Reply-To: <20050301121159.71043.qmail@web51110.mail.yahoo.com> References: <20050301121159.71043.qmail@web51110.mail.yahoo.com> Message-ID: <42246FDE.4060001@snapgear.com> Hi Scoppy, scoppy scoppy wrote: > Hi i'm new, i ahve some questions about uclinux: > > 1) Can kernel run in flash rom or it must be full loaded in ram? Yes. > 2) how cai i do to make linux running in flash rom? Exactly what is required depends on the architectures and how it was implemented. For m68k/ColdFire you just choose the kernel "Run from ROM" option. (It needs appropriate code support behind this to work - and not all platforms currently have that). Fundamently you need a linker script and start up code that is setup to run from ROM/flash. > 3) 3 mbyte of rom ad 2 mbyte of ram is enaugth for run uclinux (i'm try > uclinux for mips32 2.4.16) That is enough. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From pwilshire at cox.net Tue Mar 1 09:55:40 2005 From: pwilshire at cox.net (Phil Wilshire) Date: Tue, 01 Mar 2005 09:55:40 -0500 Subject: [uClinux-dev] uClinux Training Munich and San Jose Message-ID: <4224826C.2030104@cox.net> Hi All, I am conducting a couple of uClinux workshops in Europe and the USA Munich ( March 22nd -24th ) and San Jose (April 12th - 14th ). These will be using the Analog Devices Blackfin DSP BF533 as a target system. Class details are available here: http://www.sysdcs.com/SDCSucbf.html Topics include: u-Boot JTAG 2.6.8 kernel effectively using the uClinux distribution Working with Cygwin ( and Colinux ) Source Management with CVS Debugging ( including kgdboe ) Intro to Device Drivers Networking Web Based User Applications There is an online registration form here: http://www.sysdcs.com/SDCSBFreg.html Please email me at pwilshire at cox dot net if you have any qusetions of need any more details regards Phil Wilshire From ashish_ibm at rediffmail.com Tue Mar 1 10:27:52 2005 From: ashish_ibm at rediffmail.com (ashish anand) Date: 1 Mar 2005 15:27:52 -0000 Subject: [uClinux-dev] does handle_bridge receives everyhing.. Message-ID: <20050301152752.8050.qmail@webmail10.rediffmail.com> in net/core/dev.c it appears that, net_rx_action passes every packet to handle_bridge() if CONFIG_BRIDGE is set and interface is a bridge port, why can't it check if this is bpdu ,then only pass the frame to handle_frame(). Best Regards, Ashish ? -------------- next part -------------- An HTML attachment was scrubbed... URL: From ben at thingmagic.com Tue Mar 1 12:28:10 2005 From: ben at thingmagic.com (Ben Giddings) Date: Tue, 01 Mar 2005 12:28:10 -0500 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like) system Message-ID: <4224A62A.3040909@thingmagic.com> Hi all, I am trying to upgrade from a Snapgear 3.0 distribution to a Snapgear 3.2 distribution. In the Snapgear 3.0 tree I'm using Rob Ranslam's patches from: http://ixp4xx-osdg.sourceforge.net/ I tried to port these over so they work with snapgear 3.2 and I'm having pretty good luck with that. Unfortunately, now when I try to build a kernel I get errors when building libnet and zlib. The first error is: /snapgear-3.2/snapgear/lib/glibc/include/stdio.h:40:28: stdio/stdio.h: No such file or directory Now, I found that there is a stdio/stdio.h in lib/glibc, and I tried adding lib/glibc itself to the include path, but then it couldn't find some other headers. I'm feeling that this might be the wrong tangent. Has anybody else got snapgear 3.2 working with an Xscale target, or has anybody else seen these types of errors? Ben From davidm at snapgear.com Tue Mar 1 16:52:09 2005 From: davidm at snapgear.com (David McCullough) Date: Wed, 2 Mar 2005 07:52:09 +1000 Subject: [uClinux-dev] almost boot Linux, but fail at "kernel BUG at kernel/exit.c:934!" after doing "/bin/expand" In-Reply-To: <200503011040.j21AeSfx008702@cad.csie.ncku.edu.tw> References: <200503011040.j21AeSfx008702@cad.csie.ncku.edu.tw> Message-ID: <20050301215209.GA24554@beast> Jivin Wayne Lee lays it down ... > > Hi, folks, > > I try to port uClinux 2.6.5 on an arm926 board. (without chip's MMU > capability) > However, it really disappointed me that uClinux 2.6.5 failed to boot after > issuing this command: > "/bin/expand /etc/ramfs.img /dev/ram1" > > The error message is: > kernel BUG at kernel/exit.c:934! > > It is too difficult to figure out what is the kernel bug. > Could anyone kindly give me some hints? > > I have ever adopted the same packages to port uClinux 2.6.5 on arm940t > board. > In this time, the main different is that arm926 cpu is configured, and > "armv5te" compile option is used. It's hard to say, things are looking pretty good. My guess is that you have expanded into the ramdisk that is bring used by your root filesystem ? Remove the expand line for now or move it to ram2/ram3 throughout the script and see if that helps. Otherwise I would double check that your RAM is all ok, Cheers, Davidm > The packages I adopted: > uClinux-dist-20040408.tar.gz > linux-2.6.5.tar.gz > linux-2.6.5-hsc2.patch.gz (hyok.choi's patch) > > The toolchain I adopted: > arm-elf-tools-20040427.sh > > The boot messages: > > Linux version 2.6.5 (root at wayne_linux) (gcc version 2.95.3 20010315 > (release)(ColdFire patches - 20010318 from > http://fiddes.net/coldfire/)(uClinux XIP and shared lib patches from > http://www.snapgear.com/)) #116 Tue Mar 1 17:37:10 CST 2005 > CPU: ARM926EJ-Sid(wb) [41069263] revision 3 (ARMv5TEJ) > CPU: D VIPT write-back cache > CPU: I cache: 32768 bytes, associativity 4, 32 byte lines, 256 sets > CPU: D cache: 32768 bytes, associativity 4, 32 byte lines, 256 sets > Machine: QPT200 board > On node 0 totalpages: 16384 > DMA zone: 0 pages, LIFO batch:1 > Normal zone: 16384 pages, LIFO batch:4 > HighMem zone: 0 pages, LIFO batch:1 > Built 1 zonelists > Kernel command line: root=/dev/ram0 initrd=0x02000000,800K keepinitrd > PID hash table entries: 512 (order 9: 4096 bytes) > Memory: 64MB = 64MB total > Memory: 63248KB available (608K code, 81K data, 44K init) > Calibrating delay loop... 0.88 BogoMIPS > Dentry cache hash table entries: 8192 (order: 3, 32768 bytes) > Inode-cache hash table entries: 4096 (order: 2, 16384 bytes) > Mount-cache hash table entries: 512 (order: 0, 4096 bytes) > checking if image is initramfs...it isn't (ungzip failed); looks like an > initrd > POSIX conformance testing by UNIFIX > Linux NoNET1.0 for Linux 2.6 > ikconfig 0.7 with /proc/config* > Initializing QPT200 Internal UART > ttyS0 at I/O 0x80000100 (irq = 17) is a QPT200 Internal UART > RAMDISK driver initialized: 16 RAM disks of 8192K size 1024 blocksize > loop: loaded (max 8 devices) > RAMDISK: romfs filesystem found at block 0 > RAMDISK: Loading 772 blocks [1 disk] into ram disk... > |/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/ > -\|/-\|/-\|done. > VFS: Mounted root (romfs filesystem) readonly. > Freeing init memory: 44K > Shell invoked to run file: /etc/rc > Command: hostname QPT-200 > Command: /bin/expand /etc/ramfs.img /dev/ram1 > init/expand.c > argv[0] is [/bin/expand] > argv[1] is [/etc/ramfs.img] > argv[2] is [/dev/ram1] > read(fdi,&len,4); fdi=[0] len=[34422656] > len = ntohl(len); > lseek(fdo, 0, SEEK_SET); fdo=[3] > close fdi[0] > close fdo[3] > free(buf) > exit(0); > kernel BUG at kernel/exit.c:934! > > > > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From doug at intellimetrix.us Tue Mar 1 16:56:06 2005 From: doug at intellimetrix.us (Doug Abbott) Date: Tue, 01 Mar 2005 14:56:06 -0700 Subject: [uClinux-dev] undefined references: __CTOR_LIST__ and __DTOR_LIST__ Message-ID: <4224E4F6.1030001@intellimetrix.us> Hi all, I've installed uClinux 2.4.x on a uCdimm module from the Arcturus distribution dated November 2003. Now when I try to compile and link a single file program, by just invoking m68k-elf-gcc, I get two undefined references: __CTOR_LIST__ in __do_global_ctors and __DTOR_LIST__ in __do_global_dtors I know this problem has popped up on the list from time to time, but I haven't seen a satisfactory solution for it. Most of the previous queries seem to involve the ARM port. It appears that these symbols have something to do with global constructors and destructors. I find it rather odd that these references get pulled in since my code is pure C. Any ideas? Thanks, Doug From davidm at snapgear.com Tue Mar 1 17:15:21 2005 From: davidm at snapgear.com (David McCullough) Date: Wed, 2 Mar 2005 08:15:21 +1000 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like) system In-Reply-To: <4224A62A.3040909@thingmagic.com> References: <4224A62A.3040909@thingmagic.com> Message-ID: <20050301221521.GE24554@beast> Jivin Ben Giddings lays it down ... > Hi all, > > I am trying to upgrade from a Snapgear 3.0 distribution to a Snapgear > 3.2 distribution. In the Snapgear 3.0 tree I'm using Rob Ranslam's > patches from: > > http://ixp4xx-osdg.sourceforge.net/ > > I tried to port these over so they work with snapgear 3.2 and I'm having > pretty good luck with that. Unfortunately, now when I try to build a > kernel I get errors when building libnet and zlib. > > The first error is: > > /snapgear-3.2/snapgear/lib/glibc/include/stdio.h:40:28: stdio/stdio.h: > No such file or directory > > Now, I found that there is a stdio/stdio.h in lib/glibc, and I tried > adding lib/glibc itself to the include path, but then it couldn't find > some other headers. I'm feeling that this might be the wrong tangent. > > Has anybody else got snapgear 3.2 working with an Xscale target, or has > anybody else seen these types of errors? You might be interested in the "SnapGear IXP400 Access Library patch" that is on the snapgear.org, should make the integration a bit easier. Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From bob at softmach.com Tue Mar 1 17:34:17 2005 From: bob at softmach.com (Bob Rubendunst) Date: Tue, 01 Mar 2005 16:34:17 -0600 Subject: [uClinux-dev] Input/output error by calling ioctl() In-Reply-To: <20050301170006.EFCF219F74@uclinux.org> References: <20050301170006.EFCF219F74@uclinux.org> Message-ID: <6.1.2.0.0.20050301162846.02d71930@127.0.0.1> Make sure there isn't a getty running on the port. When getty sees a drop of CD, it does a hangup on the port, and all opens are unceremoniously closed. At 11:00 AM 3/1/2005, Yegor Yefremov wrote: >Subject: [uClinux-dev] Input/output error by calling ioctl() >To: >Message-ID: > <7535214293BFB44BAA91FC85E643442401B943 at kallisto.VisionSystems.local> >Content-Type: text/plain; charset="us-ascii" > >Hi > >I'm writing a program that communicates with a terminal via serial port. >I need notice the change of DCD or DSR signals. >When I start my programm and read port state with ioctl(fd, TIOCMGET, >&controlreg) it succeeds whether DCD is high or low, >but when I then turn DCD off or unplug the cable the ioctl() call ends >with "Input/output error". Can anyone help me? Is there >any possibility to notice the change of DCD without an error? > > int ret, fd; > struct termios ser_termios; > > ret = open(SERIAL_DEV_COM1, (O_RDWR | O_NOCTTY | O_NDELAY)); > if (ret < 0) > { > perror("open"); > exit(1); > } > fd=ret; > > cfmakeraw(&ser_termios); > ser_termios.c_cflag &= ~(CSIZE | PARENB); > ser_termios.c_lflag &= ~(ECHO | ICANON | IEXTEN | ISIG); > ser_termios.c_iflag &= ~(BRKINT | ICRNL | INPCK | ISTRIP | IXON); > ser_termios.c_cflag |= CS8; > ser_termios.c_cc[VMIN] = 1; > ser_termios.c_cc[VTIME] = 0; > > ret = cfsetispeed(&ser_termios, B1200); > ret = cfsetospeed(&ser_termios, B1200); > ret = tcsetattr(fd, TCSANOW, &ser_termios); > > ret = ioctl(fd, TIOCMGET, &controlreg); > if (ret < 0) > { > perror("ioct: "); //here I get error if I turn DCD >off before calling ioctl() > } > >Best regards, > > >Yegor Yefremov -Bob Rubendunst Soft Machines http://www.softmach.com "Everything serial from start to stop" From davidm at snapgear.com Tue Mar 1 17:42:05 2005 From: davidm at snapgear.com (David McCullough) Date: Wed, 2 Mar 2005 08:42:05 +1000 Subject: [uClinux-dev] undefined references: __CTOR_LIST__ and __DTOR_LIST__ In-Reply-To: <4224E4F6.1030001@intellimetrix.us> References: <4224E4F6.1030001@intellimetrix.us> Message-ID: <20050301224205.GJ24554@beast> Jivin Doug Abbott lays it down ... > Hi all, > > I've installed uClinux 2.4.x on a uCdimm module from the Arcturus > distribution dated November 2003. Now when I try to compile and link a > single file program, by just invoking m68k-elf-gcc, I get two undefined > references: > __CTOR_LIST__ in __do_global_ctors and > __DTOR_LIST__ in __do_global_dtors My guess is that you are using a new tool chan with an old distribution or vice versa ? Which exact dist are you using and which exact toolchain ? There were changes made at some point in time that would cause this. > I know this problem has popped up on the list from time to time, but I > haven't seen a satisfactory solution for it. Most of the previous > queries seem to involve the ARM port. > > It appears that these symbols have something to do with global > constructors and destructors. I find it rather odd that these > references get pulled in since my code is pure C. Despite what you may think, all gcc compiled 'C' code links in the ctor/dtor support code, Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From tesis_uclinux at microe.udea.edu.co Tue Mar 1 19:11:32 2005 From: tesis_uclinux at microe.udea.edu.co (tesis_uclinux at microe.udea.edu.co) Date: Tue, 1 Mar 2005 19:11:32 -0500 Subject: [uClinux-dev] Big image.bin with uClinux-dist-20041215 for M68VZ328 ucdimm Message-ID: <200503020011.j220BWOw005326@microe.udea.edu.co> When we build the distribution for M68VZ328 ucdimm with all defaults of the vendor, it launchs a big image.bin with 270Mb of size for kernel 2.6.x and a normal image size for kernel 2.4.x. We have read the document "Absolute linking with gcc & friends". We understand which is the problem but we don't know exactly what modifications we have to do and in what file in the kernel 2.6.x because the related files are very diferent in 2.4 y 2.6. Thanks for your help From philipc at snapgear.com Tue Mar 1 20:02:16 2005 From: philipc at snapgear.com (Philip Craig) Date: Wed, 02 Mar 2005 11:02:16 +1000 Subject: [uClinux-dev] does handle_bridge receives everyhing.. In-Reply-To: <20050301152752.8050.qmail@webmail10.rediffmail.com> References: <20050301152752.8050.qmail@webmail10.rediffmail.com> Message-ID: <42251098.5050908@snapgear.com> ashish anand wrote: > in net/core/dev.c it appears that, > net_rx_action passes every packet to handle_bridge() > if CONFIG_BRIDGE is set and interface is a bridge port, > > why can't it check if this is bpdu ,then only pass the frame to handle_frame(). br_handle_frame() needs to be called for all packets on the bridge, not just bpdu. And handle_bridge() must always be called before calling br_handle_frame(), so it should be called for all packets as well. -- Philip Craig - SnapGear, A CyberGuard Company - http://www.SnapGear.com From gerg at snapgear.com Tue Mar 1 20:26:21 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Wed, 02 Mar 2005 11:26:21 +1000 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like) system In-Reply-To: <4224A62A.3040909@thingmagic.com> References: <4224A62A.3040909@thingmagic.com> Message-ID: <4225163D.4010601@snapgear.com> Hi Ben, Ben Giddings wrote: > I am trying to upgrade from a Snapgear 3.0 distribution to a Snapgear > 3.2 distribution. In the Snapgear 3.0 tree I'm using Rob Ranslam's > patches from: > > http://ixp4xx-osdg.sourceforge.net/ > > I tried to port these over so they work with snapgear 3.2 and I'm having > pretty good luck with that. Unfortunately, now when I try to build a > kernel I get errors when building libnet and zlib. > > The first error is: > > /snapgear-3.2/snapgear/lib/glibc/include/stdio.h:40:28: stdio/stdio.h: > No such file or directory If you compile the 3.2 code "out-of-the-box" does it work? Is it only after adding the ixp code and patches that this starts failing? > Now, I found that there is a stdio/stdio.h in lib/glibc, and I tried > adding lib/glibc itself to the include path, but then it couldn't find > some other headers. I'm feeling that this might be the wrong tangent. > > Has anybody else got snapgear 3.2 working with an Xscale target, or has > anybody else seen these types of errors? I use the 3.2 code base extensively on xscale targets... Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Tue Mar 1 20:32:28 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Wed, 02 Mar 2005 11:32:28 +1000 Subject: [uClinux-dev] Big image.bin with uClinux-dist-20041215 for M68VZ328 ucdimm In-Reply-To: <200503020011.j220BWOw005326@microe.udea.edu.co> References: <200503020011.j220BWOw005326@microe.udea.edu.co> Message-ID: <422517AC.2090704@snapgear.com> Hi, tesis_uclinux at microe.udea.edu.co wrote: > When we build the distribution for M68VZ328 ucdimm with all defaults > of the vendor, it launchs a big image.bin with 270Mb of size for kernel > 2.6.x and a normal image size for kernel 2.4.x. > > We have read the document "Absolute linking with gcc & friends". > We understand which is the problem but we don't know exactly what > modifications we have to do and in what file in the kernel 2.6.x > because the related files are very diferent in 2.4 y 2.6. In the 2.6 code we have gone to a single linker script for all m68knommu targets (instead of the dozens that where there before). I may have missed something specific for the ucdimm when converting it over. Check the common linker scripts in 2.6: ~/linux-2.6.x/arch/m68knommu/kernel/vmlinux.lds.S And compare that to the original ucdimm linker script in: ~/linux-2.4.x/arch/m68knommu/platform/68VZ328/ucdimm/ram.ld Start by checking what sections are linked where in your binaries with: m68k-elf-objdump --headers vmlinux Allowing for size differences the sections for each of the 2.4 and 2.6 kernels should be in roughly the same addresses in memory. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From robert.e.ranslam at intel.com Tue Mar 1 20:41:07 2005 From: robert.e.ranslam at intel.com (Ranslam, Robert E) Date: Tue, 1 Mar 2005 17:41:07 -0800 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like)system Message-ID: <8D5F43CD1194B04CAD2AC656844E8F2B0496203E@orsmsx409> Hello Ben, Have you tried the shar file from http://www.snapgear.org/downloads.html? Greg, et al, did a nice job on the v1.4 integration (an fixes!) for the IXDPG425 ( Monte Jade) This should work generally for any ixp4xx target in the sg3.2 release. My patches shouldn't cause a problem related to glib though. But I have Not updated the patches for sg3.2 RR >-----Original Message----- >From: uclinux-dev-bounces at uClinux.org >[mailto:uclinux-dev-bounces at uClinux.org] On Behalf Of Ben Giddings >Sent: Tuesday, March 01, 2005 9:28 AM >To: uclinux-dev at uClinux.org >Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale >(Coyote-like)system > >Hi all, > >I am trying to upgrade from a Snapgear 3.0 distribution to a Snapgear >3.2 distribution. In the Snapgear 3.0 tree I'm using Rob >Ranslam's patches from: > >http://ixp4xx-osdg.sourceforge.net/ > >I tried to port these over so they work with snapgear 3.2 and >I'm having pretty good luck with that. Unfortunately, now >when I try to build a kernel I get errors when building libnet >and zlib. > >The first error is: > >/snapgear-3.2/snapgear/lib/glibc/include/stdio.h:40:28: stdio/stdio.h: >No such file or directory > >Now, I found that there is a stdio/stdio.h in lib/glibc, and I >tried adding lib/glibc itself to the include path, but then it >couldn't find some other headers. I'm feeling that this might >be the wrong tangent. > >Has anybody else got snapgear 3.2 working with an Xscale >target, or has anybody else seen these types of errors? > >Ben >_______________________________________________ >uClinux-dev mailing list >uClinux-dev at uClinux.org >http://mailman.uclinux.org/mailman/listinfo/uclinux-dev >This message was resent by uclinux-dev at uclinux.org > From lucy.wang at philips.com Tue Mar 1 21:28:55 2005 From: lucy.wang at philips.com (lucy.wang at philips.com) Date: Wed, 2 Mar 2005 10:28:55 +0800 Subject: [uClinux-dev] [PATCH]uClinux patch for Philips LPC22xx Message-ID: Hi All, Attached is the uClinux patch for Philips LPC22xx, including the vendors/Philips/LPC22xx directory and a new config.in file. A readme file is also here. Together with the linux-2.6.5 patch, you can begin the first trial of uClinux+LPC22xx development. Note that the device drivers, e.g. Ethernet driver(although I've finished the rtl8019as driver), are not here. Best regards, Lucy -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: uClinux-Philips-LPC22xx.tar.gz Type: application/octet-stream Size: 28666 bytes Desc: not available URL: From doug at intellimetrix.us Tue Mar 1 21:46:34 2005 From: doug at intellimetrix.us (Doug Abbott) Date: Tue, 01 Mar 2005 19:46:34 -0700 Subject: [uClinux-dev] undefined references: __CTOR_LIST__ and __DTOR_LIST__ In-Reply-To: <20050301224205.GJ24554@beast> References: <4224E4F6.1030001@intellimetrix.us> <20050301224205.GJ24554@beast> Message-ID: <4225290A.9030503@intellimetrix.us> Hi David, David McCullough wrote: >Jivin Doug Abbott lays it down ... > > >>Hi all, >> >>I've installed uClinux 2.4.x on a uCdimm module from the Arcturus >>distribution dated November 2003. Now when I try to compile and link a >>single file program, by just invoking m68k-elf-gcc, I get two undefined >>references: >> __CTOR_LIST__ in __do_global_ctors and >> __DTOR_LIST__ in __do_global_dtors >> >> > >My guess is that you are using a new tool chan with an old distribution >or vice versa ? > >Which exact dist are you using and which exact toolchain ? >There were changes made at some point in time that would cause this. > > > The compiler identifies itself as 2.95.3. The uClinux version is 2.4.22. Is there any other info you need? Thanks, Doug From davidm at snapgear.com Tue Mar 1 22:03:32 2005 From: davidm at snapgear.com (David McCullough) Date: Wed, 2 Mar 2005 13:03:32 +1000 Subject: [uClinux-dev] undefined references: __CTOR_LIST__ and __DTOR_LIST__ In-Reply-To: <4225290A.9030503@intellimetrix.us> References: <4224E4F6.1030001@intellimetrix.us> <20050301224205.GJ24554@beast> <4225290A.9030503@intellimetrix.us> Message-ID: <20050302030332.GB29427@beast> Jivin Doug Abbott lays it down ... > Hi David, > > David McCullough wrote: > > >Jivin Doug Abbott lays it down ... > > > > > >>Hi all, > >> > >>I've installed uClinux 2.4.x on a uCdimm module from the Arcturus > >>distribution dated November 2003. Now when I try to compile and link a > >>single file program, by just invoking m68k-elf-gcc, I get two undefined > >>references: > >> __CTOR_LIST__ in __do_global_ctors and > >> __DTOR_LIST__ in __do_global_dtors > >> > >> > > > >My guess is that you are using a new tool chan with an old distribution > >or vice versa ? > > > >Which exact dist are you using and which exact toolchain ? > >There were changes made at some point in time that would cause this. > > > > > > > The compiler identifies itself as 2.95.3. The uClinux version is > 2.4.22. Is there any other info you need? I was after the original packages you downloaded, for example: m68k-elf-tools.... uClinux-dist.... If you are using entirely Arcturus provided files then they are probably best to answer, Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From it_bhuvan at rediffmail.com Tue Mar 1 23:26:14 2005 From: it_bhuvan at rediffmail.com (bhuvan eshwari) Date: 2 Mar 2005 04:26:14 -0000 Subject: [uClinux-dev] transcoding HDTV Message-ID: <20050302042614.18753.qmail@webmail28.rediffmail.com> ?Hi, I'm trying to transcode a HDTV stream to MPEG2TS format.. Is there any tool supporting this transcoding!! We were using VLC for transcoding bitrate around. 3072Kbps!! It does not 've provision to transcode very High bitrate like HDTV!! Any info to add!! reg, bhuvana. -------------- next part -------------- An HTML attachment was scrubbed... URL: From srinivas.vutukuri at gmail.com Tue Mar 1 23:58:38 2005 From: srinivas.vutukuri at gmail.com (srinivas naga vutukuri) Date: Wed, 2 Mar 2005 10:28:38 +0530 Subject: [uClinux-dev] transcoding HDTV Message-ID: Hi, check this link any useful... I just googled... http://www.videohelp.com/tools?tool=HDTVtoMPEG2 regards, srinivas. From gerg at snapgear.com Wed Mar 2 00:31:17 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Wed, 02 Mar 2005 15:31:17 +1000 Subject: [uClinux-dev] [PATCH] armnommu 2.4.27 vc547x fixes In-Reply-To: References: Message-ID: <42254FA5.6020604@snapgear.com> Hi Harald, Harald K?the wrote: > this patch is against uClinux dist 20041215. > It is for the C5471 port of the linux 2.4 kernel. > > The actual port does not even compile. > Here's a short list of changes: > - changed CONFIG_ARCH_C5471 to CONFIG_ARCH_VC547X so that it is in sync with arms mach-types.h file > - generates (i.e. does not remove) the linux.gz file which can be mkimage(d) an started via uboot, so no need to start from ROM > - changed system startup (i.e start from RAM), receives tag list via register from e.g. u-boot and uses the given commandline parameters. > - workaround in serial driver which fixes non clearing RX timeout interrupt. > - support for vc547x EVM flash (mtd map). > > What works: uboot boots kernel via nfs of mtd partion, sash, busybox (mostly) telnetd > The system is not really stable but I'm working on it. > An uboot diff and vendors/... diff can be supplied This looks good. I have applied to CVS. Please check that all changes are merged. Thanks Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Wed Mar 2 00:55:08 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Wed, 02 Mar 2005 15:55:08 +1000 Subject: [uClinux-dev] kernel 2.4.x for Mattel JuiceBox In-Reply-To: <000901c51b25$9c0b66e0$6f01a8c0@4010cdt> References: <000901c51b25$9c0b66e0$6f01a8c0@4010cdt> Message-ID: <4225553C.8000100@snapgear.com> Hi Martin, Martin Siu wrote: > If you are interested in putting uclinux on Mattel JuiceBox, here is the link to the its source code: > http://www.emsoftltd.com/customers.html > > Unpack the tarball, you will find the linux-2.4.x source code and a small bootloader. > The boot loader, used on JuiceBox, simply perform h/w init and copy the kernel image to sdram for execution. > A few modifications is made on the kernel source based on the original distribution (20040408): > - fixed up RAM/ROM kernel configs for s3c44b0x. > - detached romfs image from the kernel > - added 4096 color STN char driver, with backlight and contrast control. > - added cs43l43 DAC driver with volume wheel control > - added juicebox key driver Can you generate a patch against CVS, or uClinux-dist-20041215, so that we can apply it? Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From msiu at emsoftltd.com Wed Mar 2 01:17:11 2005 From: msiu at emsoftltd.com (Martin Siu) Date: Wed, 2 Mar 2005 14:17:11 +0800 Subject: [uClinux-dev] kernel 2.4.x for Mattel JuiceBox References: <000901c51b25$9c0b66e0$6f01a8c0@4010cdt> <4225553C.8000100@snapgear.com> Message-ID: <005a01c51eef$79406ed0$6f01a8c0@4010cdt> No probs, I'll prepare a patch file in the next few days. regards, Martin ----- Original Message ----- From: "Greg Ungerer" To: "Martin Siu" Cc: "uClinux" Sent: Wednesday, March 02, 2005 1:55 PM Subject: Re: [uClinux-dev] kernel 2.4.x for Mattel JuiceBox > Hi Martin, > > Martin Siu wrote: > > If you are interested in putting uclinux on Mattel JuiceBox, here is the link to the its source code: > > http://www.emsoftltd.com/customers.html > > > > Unpack the tarball, you will find the linux-2.4.x source code and a small bootloader. > > The boot loader, used on JuiceBox, simply perform h/w init and copy the kernel image to sdram for execution. > > A few modifications is made on the kernel source based on the original distribution (20040408): > > - fixed up RAM/ROM kernel configs for s3c44b0x. > > - detached romfs image from the kernel > > - added 4096 color STN char driver, with backlight and contrast control. > > - added cs43l43 DAC driver with volume wheel control > > - added juicebox key driver > > Can you generate a patch against CVS, or uClinux-dist-20041215, so > that we can apply it? > > Regards > Greg > > > > ------------------------------------------------------------------------ > Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com > SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 > 825 Stanley St, FAX: +61 7 3891 3630 > Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From sumac at myw.ltindia.com Wed Mar 2 02:51:51 2005 From: sumac at myw.ltindia.com (Suma Choudhary) Date: Wed, 02 Mar 2005 13:21:51 +0530 Subject: [uClinux-dev] 3 pids for a process using thread! Message-ID: hi all In a test code( thread.c) , I use pthread_create to create a thread. Tthe process table(using ps) has three instances of the executable with 3 different pids(successive numbers).why is that? The same code I tested on my pc and the process table has an entry for only one process. Is the way the uClinux handles the threads different? thanks in advance suma ps:the code contains a pthread_create and a pthread_join functions. From andrea.tarani at gilbarco.com Wed Mar 2 02:59:00 2005 From: andrea.tarani at gilbarco.com (andrea.tarani at gilbarco.com) Date: Wed, 2 Mar 2005 08:59:00 +0100 Subject: [uClinux-dev] RE: m68k gdb problem Message-ID: Hi have posted the solution (for GDB 6.0) some time ago. Hi lso have send the patch for GDB 6.3 and some patch for GCC 3.4 to Bernardo Innocenti but they doesn't have updated theweb server. Can you ping Bernardo?? From mohuaidong at gmail.com Wed Mar 2 04:38:25 2005 From: mohuaidong at gmail.com (Huaidong Mo) Date: Wed, 2 Mar 2005 17:38:25 +0800 Subject: [uClinux-dev] ppp on uclinux,help Message-ID: <6df4decc05030201389398c96@mail.gmail.com> Now I want to connect to internet with wireless modem on Samsung 44B0. I think I have to use pppd to dial the modem.But I have found pppd is very difficult to use. and chat is also bothered me. Has anyone do the same work? I wonder whether I have to use pppd+chap? Is there any useful tools which can make the work simplified? Is pppd+chap means that ISP must used chat authentication too? From Harald.Kuethe at controlware.de Wed Mar 2 05:08:38 2005 From: Harald.Kuethe at controlware.de (=?ISO-8859-1?Q?Harald=20K=FCthe?=) Date: Wed, 02 Mar 2005 11:08:38 +0100 Subject: [uClinux-dev] [PATCH] vc547x ethernet fixes Message-ID: Hello, this patch is against uClinux dist 20041215. It is for the ethernet driver of the C5471 Here's a short list of changes: - reworked most of the register/bit definitions according to manual - added option to enable/disable the use of the ESM (Ethernet State machine) - added option to give all TX descriptors of a packet to HW at once which fixes TX underrun problems, only implemented if ESM is not used so far. Regards Harald -------------- next part -------------- A non-text attachment was scrubbed... Name: eth-vc547x.diff.gz Type: application/octet-stream Size: 9089 bytes Desc: not available URL: From Harald.Kuethe at controlware.de Wed Mar 2 05:16:33 2005 From: Harald.Kuethe at controlware.de (=?ISO-8859-1?Q?Harald=20K=FCthe?=) Date: Wed, 02 Mar 2005 11:16:33 +0100 Subject: [uClinux-dev] [PATCH] vc547x vendors/ti/vc547x dir Message-ID: Hello, here is the vendor/Ti/vc547x directory with an example configuration for jffs2. Regards Harald -------------- next part -------------- A non-text attachment was scrubbed... Name: uClinux-vendors-vc547x.diff.gz Type: application/octet-stream Size: 19103 bytes Desc: not available URL: From atlarge80 at 163.com Wed Mar 2 07:12:53 2005 From: atlarge80 at 163.com (=?gb2312?B?YXRsYXJnZTgw?=) Date: Wed, 2 Mar 2005 20:12:53 +0800 (CST) Subject: [uClinux-dev] Trouble to complie busybox-1.00 with arm-elf-tools-20030314.sh Message-ID: <4225ADC5.00002D.05840@m229.163.com> Hi , When I complie busybox-1.00 with arm-elf-tools-20030314.sh(I get it from: http://www.uclinux.org/pub/uClinux/uclinux-elf-tools/arm-elf-tools-20030314.sh) I meet with such problem . I can not fix it up, Can anybody help me? Thanks in advance. Here is the problem: /usr/local/bin/arm-elf-gcc -s -Wl,-warn-common --static -o busybox -Wl,--start-group /usr/lpf/busybox/busybox-1.00/applets/applets.a /usr/lpf/busybox/busybox-1.00/archival/archival.a /usr/lpf/busybox/busybox-1.00/archival/libunarchive/libunarchive.a /usr/lpf/busybox/busybox-1.00/coreutils/coreutils.a /usr/lpf/busybox/busybox-1.00/console-tools/console-tools.a /usr/lpf/busybox/busybox-1.00/debianutils/debianutils.a /usr/lpf/busybox/busybox-1.00/editors/editors.a /usr/lpf/busybox/busybox-1.00/findutils/findutils.a /usr/lpf/busybox/busybox-1.00/init/init.a /usr/lpf/busybox/busybox-1.00/miscutils/miscutils.a /usr/lpf/busybox/busybox-1.00/modutils/modutils.a /usr/lpf/busybox/busybox-1.00/networking/networking.a /usr/lpf/busybox/busybox-1.00/networking/libiproute/libiproute.a /usr/lpf/busybox/busybox-1.00/networking/udhcp/udhcp.a /usr/lpf/busybox/busybox-1.00/procps/procps.a /usr/lpf/busybox/busybox-1.00/loginutils/loginutils.a /usr/lpf/busybox/busybox-1.00/shell/shell.a /usr/lpf/busybox/busybox-1.00/sysklogd/sysklogd.a /usr/lpf/busybox/busybox-1.00/util-linux/util-linux.a /usr/lpf/busybox/busybox-1.00/libpwdgrp/libpwdgrp.a /usr/lpf/busybox/busybox-1.00/coreutils/libcoreutils/libcoreutils.a /usr/lpf/busybox/busybox-1.00/libbb/libbb.a -Wl,--end-group /usr/lpf/busybox/busybox-1.00/sysklogd/sysklogd.a(klogd.o): In function `klogd_main': klogd.o(.text+0x29c): undefined reference to `daemon' /usr/lpf/busybox/busybox-1.00/sysklogd/sysklogd.a(syslogd.o): In function `syslogd_main': syslogd.o(.text+0x9a0): undefined reference to `daemon' /usr/local/lib/gcc-lib/arm-elf/2.95.3/libgcc.a(__main.o): In function `__do_global_ctors': __main.o(.text+0xc4): undefined reference to `__CTOR_LIST__' /usr/local/lib/gcc-lib/arm-elf/2.95.3/libgcc.a(__main.o)(.data+0x0): undefined reference to `__DTOR_LIST__' collect2: ld returned 1 exit status make: *** [busybox] Error 1 -------------- next part -------------- An HTML attachment was scrubbed... URL: From g.vinay at harita.com Wed Mar 2 07:49:31 2005 From: g.vinay at harita.com (G.Vinay(Embedded/Harita)) Date: Wed, 2 Mar 2005 18:19:31 +0530 Subject: [uClinux-dev] Need help on building MCF5249 kernel. Message-ID: <777E8D57B5332843A08EB103769A36D420D0DC@BLREX01.tvse.int> Hi all I am new member to the group aswell as to the uclinux. I am compiling the kernel for the MCF5249 C3 board, source downloaded from the uclinux.org. and followed the same procedures given in readme. then I am getting this error as.......... -fno-builtin -nostdinc -D__KERNEL__ -I/uClinux-dist/linux-2.4.x/include -Wall -Wstrict-prototypes -Wno-trigraphs -O1 -g -fno-strict-aliasing -fno-common -I/usr/local/lib/gcc-lib/m68k-elf/2.95.3/./include -pipe -DNO_MM -DNO_FPU -m5200 -Wa,-S -Wa,-m5200 -D__ELF__ -DMAGIC_ROM_PTR -DUTS_SYSNAME="uClinux" -D__linux__ -nostdinc -iwithprefix include -DKBUILD_BASENAME=acct -c -o acct.o acct.c acct.c: In function `do_acct_process': acct.c:320: structure has no member named `mmap' acct.c:323: structure has no member named `vm_next' acct.c:318: warning: `vma' might be used uninitialized in this function make[3]: *** [acct.o] Error 1 make[3]: Leaving directory `/uClinux-dist/linux-2.4.x/kernel' make[2]: *** [first_rule] Error 2 make[2]: Leaving directory `/uClinux-dist/linux-2.4.x/kernel' make[1]: *** [_dir_kernel] Error 2 make[1]: Leaving directory `/uClinux-dist/linux-2.4.x' make: *** [linux] Error 1 [root at localhost uClinux-dist]# please any one can help in resolving this problem. Thanks and Regards Vinay G Software Engineer Embedded software division Harita infoserve ltd. "Ozone" No.10, 3rd Main, Ashwini Layout, Koramangala, Ejipura BANGALORE 560 047 Phone: 56601924 -------------- next part -------------- An HTML attachment was scrubbed... URL: From davidm at snapgear.com Wed Mar 2 08:03:56 2005 From: davidm at snapgear.com (David McCullough) Date: Wed, 2 Mar 2005 23:03:56 +1000 Subject: [uClinux-dev] Need help on building MCF5249 kernel. In-Reply-To: <777E8D57B5332843A08EB103769A36D420D0DC@BLREX01.tvse.int> References: <777E8D57B5332843A08EB103769A36D420D0DC@BLREX01.tvse.int> Message-ID: <20050302130356.GA26804@beast> Jivin G.Vinay(Embedded/Harita) lays it down ... > > Hi all > > I am new member to the group aswell as to the uclinux. > > I am compiling the kernel for the MCF5249 C3 board, source downloaded from the uclinux.org. and followed the same procedures given in readme. then I am getting this error as.......... > > -fno-builtin -nostdinc -D__KERNEL__ -I/uClinux-dist/linux-2.4.x/include -Wall -Wstrict-prototypes -Wno-trigraphs -O1 -g -fno-strict-aliasing -fno-common -I/usr/local/lib/gcc-lib/m68k-elf/2.95.3/./include -pipe -DNO_MM -DNO_FPU -m5200 -Wa,-S -Wa,-m5200 -D__ELF__ -DMAGIC_ROM_PTR -DUTS_SYSNAME="uClinux" -D__linux__ -nostdinc -iwithprefix include -DKBUILD_BASENAME=acct -c -o acct.o acct.c > acct.c: In function `do_acct_process': > acct.c:320: structure has no member named `mmap' > acct.c:323: structure has no member named `vm_next' > acct.c:318: warning: `vma' might be used uninitialized in this function > make[3]: *** [acct.o] Error 1 > make[3]: Leaving directory `/uClinux-dist/linux-2.4.x/kernel' > make[2]: *** [first_rule] Error 2 > make[2]: Leaving directory `/uClinux-dist/linux-2.4.x/kernel' > make[1]: *** [_dir_kernel] Error 2 > make[1]: Leaving directory `/uClinux-dist/linux-2.4.x' > make: *** [linux] Error 1 > [root at localhost uClinux-dist]# > > please any one can help in resolving this problem. You have CONFIG_BSD_PROCESS_ACCT enabled, seems like you are the first to do this on uClinux ;-) If you didn't turn it on then I would guess that the kernel config for that target is a little broken. Either way you should revisit the kernel configuration for your target and make sure all is well, disabled CONFIG_BSD_PROCESS_ACCT as you go ;-) Cheers, Davidm > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From dhowells at redhat.com Wed Mar 2 09:20:35 2005 From: dhowells at redhat.com (David Howells) Date: Wed, 02 Mar 2005 14:20:35 +0000 Subject: [uClinux-dev] Need help on building MCF5249 kernel. In-Reply-To: <20050302130356.GA26804@beast> References: <20050302130356.GA26804@beast> <777E8D57B5332843A08EB103769A36D420D0DC@BLREX01.tvse.int> Message-ID: <5941.1109773235@redhat.com> David McCullough wrote: > You have CONFIG_BSD_PROCESS_ACCT enabled, seems like you are the first > to do this on uClinux ;-) > > If you didn't turn it on then I would guess that the kernel config for > that target is a little broken. > > Either way you should revisit the kernel configuration for your target > and make sure all is well, disabled CONFIG_BSD_PROCESS_ACCT as you go ;-) That might actually work on 2.4.29-uc0 plus my mmnommu changes. David From H.Haeberle at pilz.de Wed Mar 2 11:16:21 2005 From: H.Haeberle at pilz.de (Heinz Haeberle) Date: Wed, 2 Mar 2005 17:16:21 +0100 Subject: [uClinux-dev] m68k using gcj Message-ID: Does anybody have experience with compiling java for coldfire? I have built the gcc 3.4.3 environment added the java target to the gcc build process Thanks to the amazing build script David McCullogh this wasn't really hard work. The m68k-uclinux-gcj ist not able top compile now. It is missing the 'java.lang' package. Any ideas? Heinz Haeberle From ashish_ibm at rediffmail.com Wed Mar 2 11:43:48 2005 From: ashish_ibm at rediffmail.com (ashish anand) Date: 2 Mar 2005 16:43:48 -0000 Subject: [uClinux-dev] side-affect of lenght timer botom half Message-ID: <20050302164348.30788.qmail@webmail10.rediffmail.com> I guess bottom halfs are atomic and it can be scheduled on only one cpu even in a Multi-processor system. though I understand that any lengthy bottom half will bring down the general performance down as they are not preemptable from application view . what can be the worst /destructive side-affect of running a lengthy TIMER bottom half on a single cpu system..? Best Regards, Ashish -------------- next part -------------- An HTML attachment was scrubbed... URL: From roberto.farina at cefriel.it Wed Mar 2 12:48:52 2005 From: roberto.farina at cefriel.it (Roberto Farina) Date: Wed, 02 Mar 2005 17:48:52 +0000 Subject: [uClinux-dev] 3 pids for a process using thread! In-Reply-To: References: Message-ID: <4225FC84.1010308@cefriel.it> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The third thread u see il the manager thread: it comes from the internal implementation of Linux threads. It is created the first time u invoke pthread_create; it creates and coordinates between all threads owned by each process. Best regards Roberto Suma Choudhary wrote: | hi all | In a test code( thread.c) , I use pthread_create to create a thread. | Tthe process table(using ps) has three instances of the executable with 3 different pids(successive numbers).why is that? | The same code I tested on my pc and the process table has an entry for only one process. | Is the way the uClinux handles the threads different? | thanks in advance | suma | | ps:the code contains a pthread_create and a pthread_join functions. | | | _______________________________________________ | uClinux-dev mailing list | uClinux-dev at uClinux.org | http://mailman.uclinux.org/mailman/listinfo/uclinux-dev | This message was resent by uclinux-dev at uclinux.org - -- - -------------------------------------- Roberto Farina CEFRIEL - Embedded Systems Design Unit Via Fucini, 2 - 20133 MILANO (ITALY) Phone: +39 02 23954 345 Fax: +39 02 23954 545 ~ 178 606 5693 - -------------------------------------- -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.7 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFCJfyEzF5ONZxALKMRAiQHAKDKGwFoNrq1ToS4x7FHWwnnPwS1DgCfcl72 CzKr1qli5rFbhM1ZF5d4JjQ= =5Rqc -----END PGP SIGNATURE----- From mbeaulieu at eocycle.com Wed Mar 2 15:34:38 2005 From: mbeaulieu at eocycle.com (Martin Beaulieu) Date: Wed, 2 Mar 2005 15:34:38 -0500 Subject: [uClinux-dev] Booting romfs with blkmem in ROM Message-ID: <000e01c51f67$42675d10$8000a8c0@MartinLogiciel> Hi, I have a 5407 chips on my own board. I path uClinux-dist version 2.4.x for run on my board. UClinux kernel boot correctly. I boot in rom memory. But when i try to load the romfs, i have some problems. 1- When i try to load the romfs in ram, i have the following message : VFS: Mounted root (romfs filesystem) readonly. Warning: unable to open an initial console. Kernel panic: No init found. Try passing init= option to kernel. 2- When i try to load the romfs in rom, i can't load the ramfs.img in ram1 or ram0, but i can have the path. When i try to load RTAI with insmod, the kernel freeze :( I can boot romfs in ram ? What is the best solutions ? I need help thanks Martin -------------- next part -------------- An HTML attachment was scrubbed... URL: From ben at thingmagic.com Wed Mar 2 18:23:44 2005 From: ben at thingmagic.com (Ben Giddings) Date: Wed, 02 Mar 2005 18:23:44 -0500 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like) system In-Reply-To: <4225163D.4010601@snapgear.com> References: <4224A62A.3040909@thingmagic.com> <4225163D.4010601@snapgear.com> Message-ID: <42264B00.3060904@thingmagic.com> Hi Greg, I hoped you might answer. :) Greg Ungerer wrote: > If you compile the 3.2 code "out-of-the-box" does it work? > Is it only after adding the ixp code and patches that this > starts failing? "Out of the box", if I choose an ADI coyote board, a 2.4 kernel, and glibc, and don't change anything else, it works. As someone else suggested, I'm going to try the "IXP400 Access Library patch" from the snapgear site. Is it an alternative to the patches at the ixp4xx-osdg site? When you're building, are you using Intel's library, and if so what method are you using to integrate their patches? Btw, briefly today when I tried to get the files off the snapgear site the site looked like a default RedHat apache install. It seems to be fixed, but in case you didn't know about that, I thought I'd mention it. Ben From ben at thingmagic.com Wed Mar 2 18:30:41 2005 From: ben at thingmagic.com (Ben Giddings) Date: Wed, 02 Mar 2005 18:30:41 -0500 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like)system In-Reply-To: <8D5F43CD1194B04CAD2AC656844E8F2B0496203E@orsmsx409> References: <8D5F43CD1194B04CAD2AC656844E8F2B0496203E@orsmsx409> Message-ID: <42264CA1.4040402@thingmagic.com> Ranslam, Robert E wrote: > Have you tried the shar file from > http://www.snapgear.org/downloads.html? > > Greg, et al, did a nice job on the v1.4 integration (an fixes!) for the > IXDPG425 > ( Monte Jade) > > This should work generally for any ixp4xx target in the sg3.2 release. Hi Rob, I just tried it, and although I have ixp400AccessLibrary-1_4.zip, I don't have DSR2.5Release.zip. Is this another bit of Intel software? If so, where do I get it? Do I need it? > My patches shouldn't cause a problem related to glib though. But I have > Not updated the patches for sg3.2 It could be a problem on my end, I could have messed up my attempt to port the patches. Ben From gerg at snapgear.com Wed Mar 2 20:13:01 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Thu, 03 Mar 2005 11:13:01 +1000 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like) system In-Reply-To: <42264B00.3060904@thingmagic.com> References: <4224A62A.3040909@thingmagic.com> <4225163D.4010601@snapgear.com> <42264B00.3060904@thingmagic.com> Message-ID: <4226649D.9050200@snapgear.com> Hi Ben, Ben Giddings wrote: > I hoped you might answer. :) :-) > Greg Ungerer wrote: >> If you compile the 3.2 code "out-of-the-box" does it work? >> Is it only after adding the ixp code and patches that this >> starts failing? > > > "Out of the box", if I choose an ADI coyote board, a 2.4 kernel, and > glibc, and don't change anything else, it works. Ok... > As someone else suggested, I'm going to try the "IXP400 Access Library > patch" from the snapgear site. Is it an alternative to the patches at > the ixp4xx-osdg site? When you're building, are you using Intel's > library, and if so what method are you using to integrate their patches? The snapgear patch was primarily designed to add support for the IXDG425 (montejade) board to the AccessLibrary. Not so much to replace Rob's patch on ixp4xx-osdg. As you have found out though the snapgear AccessLibrary patch also requires the Intel DSR library (the ixdg425 has a bunch of voice ports on it). > Btw, briefly today when I tried to get the files off the snapgear site > the site looked like a default RedHat apache install. It seems to be > fixed, but in case you didn't know about that, I thought I'd mention it. I'll check into that, thanks :-) Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From vanitha at agilis.stengg.com Thu Mar 3 07:55:30 2005 From: vanitha at agilis.stengg.com (Vanitha Ramaswami) Date: Thu, 3 Mar 2005 12:55:30 -0000 Subject: [uClinux-dev] Net-Snmp 5.2.1 Message-ID: <01C51FF0.47CFCCE0.vanitha@agilis.stengg.com> Hi All, Have any one ported Net-Snmp 5.2.1 to the snapgear kernel.? Thanks Vanitha [This e-mail is confidential and may be priviledged. If you are not the intended recipient, please kindly notify us immediately and delete the message from your system; please do not copy or use it for any purpose, nor disclose its contents to any other person. Thank you.] ---ST Electronics Group--- From it_bhuvan at rediffmail.com Tue Mar 1 23:30:51 2005 From: it_bhuvan at rediffmail.com (bhuvan eshwari) Date: 2 Mar 2005 04:30:51 -0000 Subject: [uClinux-dev] Multicasting option in Wiscore kit!! Message-ID: <20050302043051.28256.qmail@webmail47.rediffmail.com> ?Hi, I'm trying to run a socket prgm. using multicast option using the Wiscore NetstartLite kit which is using uCLinux as Embedded OS and ARM7TDMI core. I'm unable to captute the packets correctly using Ethereal!! at the destination. There is ICMP msg. and RUDP msg(Malformed packet) msgs. coming alternately!! Any idea to progress further!! reg, bhuvana. -------------- next part -------------- An HTML attachment was scrubbed... URL: From tathagata at alumnux.com Thu Mar 3 03:50:14 2005 From: tathagata at alumnux.com (Tathagata Das) Date: Thu, 03 Mar 2005 14:20:14 +0530 Subject: [uClinux-dev] WAN link status Message-ID: <4226CFC6.5010103@alumnux.com> Hi all, I am new to this group and as well as to uClinux. I have one h/w board which has ARM processor and uClinux as the operating system. I want to show the status of wan port (interface eth0) depending on whethere one ethernet cable is plugged in to the wan port or not. To do this I need the etherenet driver code which is used in uClinux. For example ---- if ethernet cable is plugged in to the wan the status of eth0 interface should be "Connected" otherwise "Disconnected". Can anybody help in this regards. Thanks in advance Tathagata From devi at tenet.res.in Thu Mar 3 16:41:42 2005 From: devi at tenet.res.in (devi) Date: Thu, 3 Mar 2005 18:41:42 -0300 Subject: [uClinux-dev] console Message-ID: <20050303214142.24325.qmail@mail.tenet.res.in> hi, how do I configure my frame buffer device to be the virtual console.How do I attach a console on the /dev/fb0 device. Regards Devi From srinivas.vutukuri at gmail.com Thu Mar 3 06:52:15 2005 From: srinivas.vutukuri at gmail.com (srinivas naga vutukuri) Date: Thu, 3 Mar 2005 17:22:15 +0530 Subject: [uClinux-dev] console Message-ID: Hi, please look this link, any useful http://www.faqs.org/docs/Linux-HOWTO/Framebuffer-HOWTO.html#s3 regards, srinivas. From rdkehn at yahoo.com Thu Mar 3 09:32:54 2005 From: rdkehn at yahoo.com (Doug Kehn) Date: Thu, 3 Mar 2005 06:32:54 -0800 (PST) Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like)system In-Reply-To: <42264CA1.4040402@thingmagic.com> Message-ID: <20050303143254.14928.qmail@web51910.mail.yahoo.com> --- Ben Giddings wrote: > I just tried it, and although I have > ixp400AccessLibrary-1_4.zip, I > don't have DSR2.5Release.zip. Is this another bit > of Intel software? > If so, where do I get it? Do I need it? > This is available from the Intel site. This file has export restrictions. __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com From rdkehn at yahoo.com Thu Mar 3 09:55:08 2005 From: rdkehn at yahoo.com (Doug Kehn) Date: Thu, 3 Mar 2005 06:55:08 -0800 (PST) Subject: [uClinux-dev] Net-Snmp 5.2.1 In-Reply-To: <01C51FF0.47CFCCE0.vanitha@agilis.stengg.com> Message-ID: <20050303145508.19892.qmail@web51908.mail.yahoo.com> --- Vanitha Ramaswami wrote: > Have any one ported Net-Snmp 5.2.1 to the snapgear > kernel.? > I have not but let me make a couple of suggestions. The Snapgear/uClinux distro already has net-snmp included (users/net-snmp). Thus, you could just take the 5.2.1 source and diff/patch. Another option would be to configure net-snmp 5.2.1 as outlined in Chapter 10 of Building Embedded Linux Systems. Then port the configured net-snmp 5.2.1 tree into your Snapgear/uClinux tree. You could also just configure/build net-snmp outside the Snapgear/uClinux tree. Regards, ...doug __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com From rdkehn at yahoo.com Thu Mar 3 10:03:46 2005 From: rdkehn at yahoo.com (Doug Kehn) Date: Thu, 3 Mar 2005 07:03:46 -0800 (PST) Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like)system In-Reply-To: <20050303143254.14928.qmail@web51910.mail.yahoo.com> Message-ID: <20050303150346.30406.qmail@web51904.mail.yahoo.com> > --- Ben Giddings wrote: > > > I just tried it, and although I have > > ixp400AccessLibrary-1_4.zip, I > > don't have DSR2.5Release.zip. Is this another bit > > of Intel software? > > If so, where do I get it? Do I need it? > > > This is available from the Intel site. This file > has > export restrictions. > Oopps, I hit accidentally hit send. To continue ... I use uClinux20041215 with ixp400AccessLibrary-1_4.zip and no DSR2.5Release.zip just fine. But I don't need it. I believe, I don't know for sure, if you want to do voice stuff and/or use the encryption engine you will need it. Regards, ...doug __________________________________ Celebrate Yahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of the Web http://birthday.yahoo.com/netrospective/ From ben at thingmagic.com Thu Mar 3 12:46:56 2005 From: ben at thingmagic.com (Ben Giddings) Date: Thu, 03 Mar 2005 12:46:56 -0500 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like)system In-Reply-To: <20050303150346.30406.qmail@web51904.mail.yahoo.com> References: <20050303150346.30406.qmail@web51904.mail.yahoo.com> Message-ID: <42274D90.7020800@thingmagic.com> Doug Kehn wrote: > Oopps, I hit accidentally hit send. To continue ... > > I use uClinux20041215 with ixp400AccessLibrary-1_4.zip > and no DSR2.5Release.zip just fine. But I don't need > it. I believe, I don't know for sure, if you want to > do voice stuff and/or use the encryption engine you > will need it. I don't want or need voice stuff, and at the moment the only encryption I care about is SSL/SSH. I'm not sure what the encryption engine gets you, but I probably don't need it. Thanks! Ben From rdkehn at yahoo.com Thu Mar 3 15:07:55 2005 From: rdkehn at yahoo.com (Doug Kehn) Date: Thu, 3 Mar 2005 12:07:55 -0800 (PST) Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like)system In-Reply-To: <42274D90.7020800@thingmagic.com> Message-ID: <20050303200755.60975.qmail@web51904.mail.yahoo.com> --- Ben Giddings wrote: > > I don't want or need voice stuff, and at the moment > the only encryption > I care about is SSL/SSH. I'm not sure what the > encryption engine gets > you, but I probably don't need it. > There is a hardware encryption engine in the IXP. Very nice if you're building a VPN/security device. BTW, I'm using both SSL and SSH. Regards, ...doug __________________________________ Celebrate Yahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of the Web http://birthday.yahoo.com/netrospective/ From ben at thingmagic.com Thu Mar 3 15:13:10 2005 From: ben at thingmagic.com (Ben Giddings) Date: Thu, 03 Mar 2005 15:13:10 -0500 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like)system In-Reply-To: <20050303200755.60975.qmail@web51904.mail.yahoo.com> References: <20050303200755.60975.qmail@web51904.mail.yahoo.com> Message-ID: <42276FD6.9050200@thingmagic.com> Doug Kehn wrote: > There is a hardware encryption engine in the IXP. > Very nice if you're building a VPN/security device. > > BTW, I'm using both SSL and SSH. Ok, how do you interface to them? What SSH server/client combo are you using? How are you using SSL? Would it work with an embedded web server, like AppWeb? Ben From ben at thingmagic.com Thu Mar 3 15:46:15 2005 From: ben at thingmagic.com (Ben Giddings) Date: Thu, 03 Mar 2005 15:46:15 -0500 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like)system In-Reply-To: <20050303150346.30406.qmail@web51904.mail.yahoo.com> References: <20050303150346.30406.qmail@web51904.mail.yahoo.com> Message-ID: <42277797.6010806@thingmagic.com> Doug Kehn wrote: > I use uClinux20041215 with ixp400AccessLibrary-1_4.zip > and no DSR2.5Release.zip just fine. But I don't need > it. I believe, I don't know for sure, if you want to > do voice stuff and/or use the encryption engine you > will need it. Hmm, when I tried to apply the patch, it wouldn't apply: patching file modules/ixp425/ixp400/ixp400_xscale_sw/Makefile Reversed (or previously applied) patch detected! Assume -R? [n] I looked at modules/ixp425/ixp400/ixp400_xscale_sw/Makefile and it looks like the patches are already there. Patch: --- modules-csr1.4-crypto-dsr2.5/ixp425/ixp400/ixp400_xscale_sw/Makefile 2004-11-24 16:19:26.000000000 +1000 +++ modules/ixp425/ixp400/ixp400_xscale_sw/Makefile 2004-11-25 10:41:04.000000000 +1000 @@ -1,6 +1,8 @@ # # Top level xscale_sw Makefile # +# File Version: $Revision: 1.291 $ +# # -- Intel Copyright Notice -- # # Copyright 2002-2003 Intel Corporation All Rights Reserved. @@ -45,26 +47,25 @@ else ifeq ($(WIND_HOST_TYPE), sun4-solaris2) IX_HOST_OS := solaris + else + # Otherwise, try OSTYPE + IX_HOST_OS := $(OSTYPE) endif endif Makefile out of ixp400AccessLibrary-1_4.zip ... # In a Tornado environment, WIND_HOST_TYPE should be set ifeq ($(WIND_HOST_TYPE), x86-win32) IX_HOST_OS := windows else ifeq ($(WIND_HOST_TYPE), sun4-solaris2) IX_HOST_OS := solaris else # Otherwise, try OSTYPE IX_HOST_OS := $(OSTYPE) endif endif When you use the patch in that shar file, how are you applying it? Are you following the directions in the readme? Ben From jsujjavanich at syntech-fuelmaster.com Thu Mar 3 17:36:20 2005 From: jsujjavanich at syntech-fuelmaster.com (Jate Sujjavanich) Date: Thu, 3 Mar 2005 17:36:20 -0500 Subject: [uClinux-dev] Busybox link error, usually works on 2nd try Message-ID: I am compiling uClinux 2.4.x with GCC 3.4 with uClibc, and it errors out during linking on Busybox. Usually by doing another 'make,' it will finish. Other times I have to turn off busybox. The error is below. Another user posted about this same problem/workaround previously, but I desire a neater solution. Can any Busybox pros help? /home/jate/uClinux-dist-feb18/include/sys/select.h:54:1: warning: "__FDMASK" redefined In file included from /home/jate/uClinux-dist-feb18/user/busybox/libbb/loop.c:22: /home/jate/uClinux-dist-feb18/include/linux/posix_types.h:34:1: warning: this is the location of the previous definition m68k-elf-ar: creating /home/jate/uClinux-dist-feb18/user/busybox/libbb/libbb.a busybox.elf2flt(.data.rel.ro+0x4): undefined reference to `test_main' busybox.elf2flt(.data.rel.ro+0x54): undefined reference to `false_main' busybox.elf2flt(.data.rel.ro+0x112): undefined reference to `test_main' busybox.elf2flt(.data.rel.ro+0x11c): undefined reference to `true_main' collect2: ld returned 1 exit status - Jate S. -------------- next part -------------- An HTML attachment was scrubbed... URL: From randy.rude at gdcanada.com Thu Mar 3 17:48:25 2005 From: randy.rude at gdcanada.com (Randy Rude) Date: Thu, 3 Mar 2005 15:48:25 -0700 Subject: [uClinux-dev] Busybox link error, usually works on 2nd try - Email found in subject In-Reply-To: References: Message-ID: <200503031548.25319.randy.rude@gdcanada.com> On March 3, 2005 03:36 pm, Jate Sujjavanich wrote: > busybox.elf2flt(.data.rel.ro+0x4): undefined reference to `test_main' > busybox.elf2flt(.data.rel.ro+0x54): undefined reference to `false_main' > busybox.elf2flt(.data.rel.ro+0x112): undefined reference to `test_main' > busybox.elf2flt(.data.rel.ro+0x11c): undefined reference to `true_main' > collect2: ld returned 1 exit status I ran into this recently too - it's not your toolchain. Just apply the following patch to user/busybox/defconfig (I'm not sure why these aren't on by default?). Randy diff -r1.1.1.1 defconfig 106c106 < # CONFIG_FALSE is not set --- > CONFIG_FALSE=y 150c150 < # CONFIG_TEST is not set --- > CONFIG_TEST=y 158c158 < # CONFIG_TRUE is not set --- > CONFIG_TRUE=y The information contained in this e-mail message is PRIVATE. It may contain confidential information and may be legally privileged. It is intended for the exclusive use of the addressee(s). If you are not the intended recipient, you are hereby notified that any dissemination, distribution or reproduction of this communication is strictly prohibited. If the intended recipient(s) cannot be reached or if a transmission problem has occurred, please notify the sender immediately by return e-mail and destroy all copies of this message. Thank you. From rdkehn at yahoo.com Thu Mar 3 17:54:08 2005 From: rdkehn at yahoo.com (Doug Kehn) Date: Thu, 3 Mar 2005 14:54:08 -0800 (PST) Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like)system In-Reply-To: <42276FD6.9050200@thingmagic.com> Message-ID: <20050303225408.19226.qmail@web51903.mail.yahoo.com> --- Ben Giddings wrote: > > Ok, how do you interface to them? What SSH > server/client combo are you > using? How are you using SSL? Would it work with > an embedded web > server, like AppWeb? > SSH Target: OpenSSH_3.7.1p2 Host: OpenSSH_3.8.1p1 SSL Target: OpenSSL 0.9.7c Host: OpenSSL 0.9.7d I mainly use SSL through sslwrap. I haven't used AppWeb but I would think if you could compile AppWeb with SSL support it would work. Regards, ...doug __________________________________ Celebrate Yahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of the Web http://birthday.yahoo.com/netrospective/ From rdkehn at yahoo.com Thu Mar 3 18:11:08 2005 From: rdkehn at yahoo.com (Doug Kehn) Date: Thu, 3 Mar 2005 15:11:08 -0800 (PST) Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like)system In-Reply-To: <42277797.6010806@thingmagic.com> Message-ID: <20050303231108.32192.qmail@web51910.mail.yahoo.com> --- Ben Giddings wrote: > Doug Kehn wrote: > > I use uClinux20041215 with > ixp400AccessLibrary-1_4.zip > > and no DSR2.5Release.zip just fine. But I don't > need > > it. I believe, I don't know for sure, if you want > to > > do voice stuff and/or use the encryption engine > you > > will need it. > > Hmm, when I tried to apply the patch, it wouldn't > apply: > > patching file > modules/ixp425/ixp400/ixp400_xscale_sw/Makefile > Reversed (or previously applied) patch detected! > Assume -R? [n] > > I looked at > modules/ixp425/ixp400/ixp400_xscale_sw/Makefile and > it looks > like the patches are already there. > > Patch: > > --- > modules-csr1.4-crypto-dsr2.5/ixp425/ixp400/ixp400_xscale_sw/Makefile > > 2004-11-24 16:19:26.000000000 +1000 > +++ modules/ixp425/ixp400/ixp400_xscale_sw/Makefile > 2004-11-25 > 10:41:04.000000000 +1000 > @@ -1,6 +1,8 @@ > # > # Top level xscale_sw Makefile > # > +# File Version: $Revision: 1.291 $ > +# > # -- Intel Copyright Notice -- > # > # Copyright 2002-2003 Intel Corporation All Rights > Reserved. > @@ -45,26 +47,25 @@ > else > ifeq ($(WIND_HOST_TYPE), sun4-solaris2) > IX_HOST_OS := solaris > + else > + # Otherwise, try OSTYPE > + IX_HOST_OS := $(OSTYPE) > endif > endif > > > Makefile out of ixp400AccessLibrary-1_4.zip > > ... > # In a Tornado environment, WIND_HOST_TYPE should be > set > ifeq ($(WIND_HOST_TYPE), x86-win32) > IX_HOST_OS := windows > else > ifeq ($(WIND_HOST_TYPE), sun4-solaris2) > IX_HOST_OS := solaris > else > # Otherwise, try OSTYPE > IX_HOST_OS := $(OSTYPE) > endif > endif > > When you use the patch in that shar file, how are > you applying it? Are > you following the directions in the readme? > I followed the steps as given in the ADI-COYOTE-HOWTO for uClinux20041215 (Snapgear 3.2 has the same HOWTO and Snapgear 3.2 is I believe equivalent to uClinux20041215). I used the following packages when following the HOWTO: arm-linux-tools-20031127.tar.gz uClinux20041215.tar.bz2 snapgear-3.1.x-ixp400-v1.4.2-support.tar.gz ixp400AccessLibrary-1_4.zip Regards, ...doug __________________________________ Celebrate Yahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of the Web http://birthday.yahoo.com/netrospective/ From ben at thingmagic.com Thu Mar 3 18:36:22 2005 From: ben at thingmagic.com (Ben Giddings) Date: Thu, 03 Mar 2005 18:36:22 -0500 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like)system In-Reply-To: <20050303231108.32192.qmail@web51910.mail.yahoo.com> References: <20050303231108.32192.qmail@web51910.mail.yahoo.com> Message-ID: <42279F76.6090306@thingmagic.com> Doug Kehn wrote: > I followed the steps as given in the ADI-COYOTE-HOWTO > for uClinux20041215 (Snapgear 3.2 has the same HOWTO > and Snapgear 3.2 is I believe equivalent to > uClinux20041215). I used the following packages when > following the HOWTO: > > arm-linux-tools-20031127.tar.gz > uClinux20041215.tar.bz2 > snapgear-3.1.x-ixp400-v1.4.2-support.tar.gz > ixp400AccessLibrary-1_4.zip Aha! Excellent, thanks! I'll try that out. (In case others are wondering, it's in snapgear/Documentation) Ben From ben at thingmagic.com Thu Mar 3 18:38:53 2005 From: ben at thingmagic.com (Ben Giddings) Date: Thu, 03 Mar 2005 18:38:53 -0500 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like)system In-Reply-To: <20050303225408.19226.qmail@web51903.mail.yahoo.com> References: <20050303225408.19226.qmail@web51903.mail.yahoo.com> Message-ID: <4227A00D.8020804@thingmagic.com> Doug Kehn wrote: > SSH > Target: OpenSSH_3.7.1p2 > Host: OpenSSH_3.8.1p1 > > SSL > Target: OpenSSL 0.9.7c > Host: OpenSSL 0.9.7d > > I mainly use SSL through sslwrap. I haven't used > AppWeb but I would think if you could compile AppWeb > with SSL support it would work. Hmm, and do you have to do anything in particular to get them to use the hardware encryption features of the XScales? Any flags or anything? I'm not sure I can use OpenSSL/OpenSSH though. I am tight on space, and may have to use "dropbear ssh" and an as-yet-to-be-determined small SSL library for the web server. Ben From davidm at snapgear.com Thu Mar 3 18:42:18 2005 From: davidm at snapgear.com (David McCullough) Date: Fri, 4 Mar 2005 09:42:18 +1000 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like)system In-Reply-To: <42274D90.7020800@thingmagic.com> References: <20050303150346.30406.qmail@web51904.mail.yahoo.com> <42274D90.7020800@thingmagic.com> Message-ID: <20050303234218.GA21675@beast> Jivin Ben Giddings lays it down ... > Doug Kehn wrote: > >Oopps, I hit accidentally hit send. To continue ... > > > >I use uClinux20041215 with ixp400AccessLibrary-1_4.zip > >and no DSR2.5Release.zip just fine. But I don't need > >it. I believe, I don't know for sure, if you want to > >do voice stuff and/or use the encryption engine you > >will need it. > > I don't want or need voice stuff, and at the moment the only encryption > I care about is SSL/SSH. I'm not sure what the encryption engine gets > you, but I probably don't need it. If you get a copy of OCF for linux: http://lists.logix.cz/pipermail/cryptoapi/2004/000261.html then you can HW accelerate your SSH/SSL connections on the IXP4xx platforms (and many other platforms :-) Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From thinh.nguyenduc at vitek.com.vn Thu Mar 3 22:26:42 2005 From: thinh.nguyenduc at vitek.com.vn (Nguyen Duc Thinh) Date: Fri, 4 Mar 2005 10:26:42 +0700 Subject: [uClinux-dev] Font Unicode Message-ID: <20050304032810.BEC2D19FFB@uclinux.org> Hi All, I want to use Unicode in my application, but I don't know where i can download this module. Please tell me more Thanks and best regards -------------- next part -------------- An HTML attachment was scrubbed... URL: From yashi at atmark-techno.com Fri Mar 4 00:33:06 2005 From: yashi at atmark-techno.com (Yasushi SHOJI) Date: Fri, 04 Mar 2005 14:33:06 +0900 Subject: [uClinux-dev] Installing uClinux on ConnectME In-Reply-To: References: Message-ID: <20050304053307.2EF1DE966@mail2.atmark-techno.com> At Tue, 22 Feb 2005 17:46:01 -0600, Chapman Mays wrote: > > So, finally my questions: Has anyone described and/or documented the process > for doing this? If I take the process that I used on my development board > will that work? haven't we documented on user's manual? check chapter 12 in user's guide. -- yashi From atlarge80 at 163.com Fri Mar 4 01:33:00 2005 From: atlarge80 at 163.com (=?gb2312?B?YXRsYXJnZTgw?=) Date: Fri, 4 Mar 2005 14:33:00 +0800 (CST) Subject: [uClinux-dev] HELP:Problem with building cross-complier for uclinux Message-ID: <4228011C.000113.06099@m148.163.com> Hi all: I met a problem when I complie the toolchains' source code downloaded from www.uclinux.org. When I run /.build-uclinux-tools.sh, it works well until the stage4,I look into the uClibc-0.9.26/Makefile and find the following lines: oldconfig: extra/config/conf $(RM) -r include/bits $(INSTALL) -d include/bits @./extra/config/conf -o extra/Configs/Config.in but I still don't know what cause the error. Here is the problem: make -C extra/config conf make[1]: Entering directory `/usr/local/xchains/uClibc-0.9.26/extra/config' ../../.config:6: *** missing separator. Stop. make[1]: Leaving directory `/usr/local/xchains/uClibc-0.9.26/extra/config' make: *** [extra/config/conf] Error 2 [root at localhost uClibc-0.9.26]# Can anybody help me? Thanks in advance!!! -------------- next part -------------- An HTML attachment was scrubbed... URL: From gerg at snapgear.com Fri Mar 4 01:52:36 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Fri, 04 Mar 2005 16:52:36 +1000 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like)system In-Reply-To: <42277797.6010806@thingmagic.com> References: <20050303150346.30406.qmail@web51904.mail.yahoo.com> <42277797.6010806@thingmagic.com> Message-ID: <422805B4.4060900@snapgear.com> Hi Ben, Ben Giddings wrote: > Doug Kehn wrote: >> I use uClinux20041215 with ixp400AccessLibrary-1_4.zip >> and no DSR2.5Release.zip just fine. But I don't need >> it. I believe, I don't know for sure, if you want to >> do voice stuff and/or use the encryption engine you >> will need it. > > > Hmm, when I tried to apply the patch, it wouldn't apply: > > patching file modules/ixp425/ixp400/ixp400_xscale_sw/Makefile > Reversed (or previously applied) patch detected! Assume -R? [n] > > I looked at modules/ixp425/ixp400/ixp400_xscale_sw/Makefile and it looks > like the patches are already there. Had you already applied Rob's patch first? You would only need to apply one patch or the other... Regards Greg > Patch: > > --- modules-csr1.4-crypto-dsr2.5/ixp425/ixp400/ixp400_xscale_sw/Makefile > 2004-11-24 16:19:26.000000000 +1000 > +++ modules/ixp425/ixp400/ixp400_xscale_sw/Makefile 2004-11-25 > 10:41:04.000000000 +1000 > @@ -1,6 +1,8 @@ > # > # Top level xscale_sw Makefile > # > +# File Version: $Revision: 1.291 $ > +# > # -- Intel Copyright Notice -- > # > # Copyright 2002-2003 Intel Corporation All Rights Reserved. > @@ -45,26 +47,25 @@ > else > ifeq ($(WIND_HOST_TYPE), sun4-solaris2) > IX_HOST_OS := solaris > + else > + # Otherwise, try OSTYPE > + IX_HOST_OS := $(OSTYPE) > endif > endif > > > Makefile out of ixp400AccessLibrary-1_4.zip > > ... > # In a Tornado environment, WIND_HOST_TYPE should be set > ifeq ($(WIND_HOST_TYPE), x86-win32) > IX_HOST_OS := windows > else > ifeq ($(WIND_HOST_TYPE), sun4-solaris2) > IX_HOST_OS := solaris > else > # Otherwise, try OSTYPE > IX_HOST_OS := $(OSTYPE) > endif > endif > > When you use the patch in that shar file, how are you applying it? Are > you following the directions in the readme? > > Ben > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org > -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From lars.friedrich at wago.com Fri Mar 4 01:57:06 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Fri, 4 Mar 2005 07:57:06 +0100 Subject: [uClinux-dev] side-affect of lenght timer botom half Message-ID: > I guess bottom halfs are atomic and it can be scheduled on only one cpu even in a > Multi-processor system. This is why bottom half handlers became obsolete (or you might say became a superset term for softirqs, tasklets and queues) and were replaced by tasklets in recent[TM] kernel versions. > what can be the worst /destructive side-affect of running a lengthy TIMER bottom half > on a single cpu system..? Weird question. The worst side-effect of bad kernel design is always a non-responsive system. Best regards, Lars Friedrich From lars.friedrich at wago.com Fri Mar 4 02:01:01 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Fri, 4 Mar 2005 08:01:01 +0100 Subject: [uClinux-dev] WAN link status Message-ID: > To do this I need the etherenet driver code which is > used in uClinux. [...] > Can anybody help in this regards. Any ethernet driver code that exists was surely shipped with your uClinux distribution, so I fail to understand your request. Best regards, Lars Friedrich From lars.friedrich at wago.com Fri Mar 4 02:11:19 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Fri, 4 Mar 2005 08:11:19 +0100 Subject: [uClinux-dev] Font Unicode Message-ID: > I want to use Unicode in my application, but I don't know where i can download > this module. Please tell me more http://www.cl.cam.ac.uk/~mgk25/unicode.html Best regards, Lars Friedrich From tathagata at alumnux.com Fri Mar 4 06:27:39 2005 From: tathagata at alumnux.com (Tathagata Das) Date: Fri, 04 Mar 2005 16:57:39 +0530 Subject: [uClinux-dev] WAN link status In-Reply-To: References: Message-ID: <4228462B.2070403@alumnux.com> Hi Lars, thanks for your response. Actually i can not find out the ethernet driver code in uClinux distribution. Can someone tell me the exact path of ethernet driver. Besides this, can anybody give some suggesion of how to get information of ethernet port connectivity. regards Tathagata Friedrich, Lars wrote: >>To do this I need the etherenet driver code which is >>used in uClinux. >> >> >[...] > > >>Can anybody help in this regards. >> >> > >Any ethernet driver code that exists was surely shipped with your >uClinux distribution, so I fail to understand your request. > >Best regards, >Lars Friedrich >_______________________________________________ >uClinux-dev mailing list >uClinux-dev at uClinux.org >http://mailman.uclinux.org/mailman/listinfo/uclinux-dev >This message was resent by uclinux-dev at uclinux.org > > > From Markus.Klama at elcat.de Fri Mar 4 08:00:04 2005 From: Markus.Klama at elcat.de (Markus Klama) Date: Fri, 04 Mar 2005 14:00:04 +0100 Subject: [uClinux-dev] Problems with USB in a non PCI system Message-ID: Hi, I'm currently developing a new device based on the Atmel AT91M55800A controller. Using kernel 2.4.29-uc0, I have the system up and running. I wrote my own boot loader and added processor support to the kernel source. I also did some enhancements like adding support for the internal high speed ram. The system is running so far, using a compact flash card as an IDE harddisk (with an ext3 file system). Now I want to enable USB support. My hardware has the SL811 controller. The problem ist that when I have USB enabled, I get compile errors. Obviously hcd.c includes pci.h and compilation aborts as hcd.o is not built. Has anyone ever got a USB host controller driver and USB stack compiled without having PCI? Are the any patch suggestions? Thank you so much, Markus Klama. From ashish_ibm at rediffmail.com Sat Mar 5 19:03:50 2005 From: ashish_ibm at rediffmail.com (ashish anand) Date: 6 Mar 2005 00:03:50 -0000 Subject: [uClinux-dev] polling through kernel-thread or timer routine...? Message-ID: <20050306000350.26477.qmail@webmail6.rediffmail.com> Hello, if i want to activate a device polling routine , typically I have two choice either i can run as kernel thread or through timer. which approach should be better in terms of system responsiveness given that, 1> polling routine is not brief and has good amount of processing 2> my requirement is not very strict about polling frequency. Best Regards, Ashish -------------- next part -------------- An HTML attachment was scrubbed... URL: From gerg at snapgear.com Sun Mar 6 19:44:53 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 07 Mar 2005 10:44:53 +1000 Subject: [uClinux-dev] Re: [patch 1/2] delete unused file include_asm_m68knommu_io_hw_swap.h In-Reply-To: <20050306104415.E114D1F202@trashy.coderock.org> References: <20050306104415.E114D1F202@trashy.coderock.org> Message-ID: <422BA405.8080201@snapgear.com> Hi Domen, domen at coderock.org wrote: > Remove nowhere referenced file. (egrep "filename\." didn't find anything) > > Signed-off-by: Domen Puncer > --- > > > kj/include/asm-m68knommu/io_hw_swap.h | 113 ---------------------------------- > 1 files changed, 113 deletions(-) I already have that in my patch queue. Thanks Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Sun Mar 6 19:46:38 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 07 Mar 2005 10:46:38 +1000 Subject: [uClinux-dev] Re: [patch 2/2] delete unused file include_asm_m68knommu_semp3.h In-Reply-To: <20050306104418.C6E301F203@trashy.coderock.org> References: <20050306104418.C6E301F203@trashy.coderock.org> Message-ID: <422BA46E.3010802@snapgear.com> Hi Domon, domen at coderock.org wrote: > Remove nowhere referenced file. (egrep "filename\." didn't find anything) > > Signed-off-by: Domen Puncer > --- > > > kj/include/asm-m68knommu/semp3.h | 60 --------------------------------------- > 1 files changed, 60 deletions(-) Thanks, I queue that to submit to Linus. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Sun Mar 6 20:35:40 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 07 Mar 2005 11:35:40 +1000 Subject: [uClinux-dev] Re: [PATCH] vc547x vendors/ti/vc547x dir In-Reply-To: References: Message-ID: <422BAFEC.30003@snapgear.com> Hi Harald, Harald K?the wrote: > here is the vendor/Ti/vc547x directory with an example configuration for jffs2. Applied, thanks. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From serena at flex-p.com Sun Mar 6 20:44:54 2005 From: serena at flex-p.com (Serena Kuan) Date: Mon, 7 Mar 2005 09:44:54 +0800 Subject: [uClinux-dev] USB ehci driver not working on Coyote Message-ID: <006c01c522b7$45cb7b50$2101a8c0@BCCHUAH> Hi, I have trying out the IXDP425 NAS-USB2 using Snapgear- 3.0 source distribution, IXP400 v1.3 software and arm- linux-tools-20030927.tar file, on the COYOTE board. As in the example, a NEC-based adapter is being used too, i.e. Belkin F5U220. I am facing a problem as the usb ehci is not working. Here is the error message during startup: Using /lib/modules/2.4.22- uc0/kernel/drivers/usb/host/ehci-hcd.o hcd.c: Found HC with no IRQ. Check BIOS/PCI 00:0f.2 setup! When I checked on the devices status, here is the message from /proc/pci and seems like there is NO IRC being sent to the USB2.0 controller. Bus 0, device 15, function 0: USB Controller: NEC Corporation USB (rev 65). IRQ 28. Master Capable. Latency=128. Min Gnt=1.Max Lat=42. Non-prefetchable 32 bit memory at 0x4bfff000 [0x4bffffff]. Bus 0, device 15, function 1: USB Controller: NEC Corporation USB (#2) (rev 65). IRQ 28. Master Capable. Latency=128. Min Gnt=1.Max Lat=42. Non-prefetchable 32 bit memory at 0x4bffe000 [0x4bffefff]. Bus 0, device 15, function 2: USB Controller: NEC Corporation USB 2.0 (rev 2). Master Capable. Latency=128. Min Gnt=16.Max Lat=34. Non-prefetchable 32 bit memory at 0x4bffdf00 [0x4bffdfff]. Lastly, from /proc/bus/usb/devices, its shown that the ehci driver is not in used, but I assumed that this is due to the IRQ not being assigned to the device in the first place. T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=12 MxCh= 2 B: Alloc= 0/900 us ( 0%), #Int= 0, #Iso= 0 D: Ver= 1.10 Cls=09(hub ) Sub=00 Prot=00 MxPS= 8 #Cfgs= 1 P: Vendor=0000 ProdID=0000 Rev= 0.00 S: Product=USB OHCI Root Hub S: SerialNumber=c28c2000 C:* #Ifs= 1 Cfg#= 1 Atr=40 MxPwr= 0mA I: If#= 0 Alt= 0 #EPs= 1 Cls=09(hub ) Sub=00 Prot=00 Driver=hub E: Ad=81(I) Atr=03(Int.) MxPS= 2 Ivl=255ms T: Bus=01 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=12 MxCh= 3 B: Alloc= 0/900 us ( 0%), #Int= 0, #Iso= 0 D: Ver= 1.10 Cls=09(hub ) Sub=00 Prot=00 MxPS= 8 #Cfgs= 1 P: Vendor=0000 ProdID=0000 Rev= 0.00 S: Product=USB OHCI Root Hub S: SerialNumber=c28bc000 C:* #Ifs= 1 Cfg#= 1 Atr=40 MxPwr= 0mA I: If#= 0 Alt= 0 #EPs= 1 Cls=09(hub ) Sub=00 Prot=00 Driver=hub E: Ad=81(I) Atr=03(Int.) MxPS= 2 Ivl=255ms T: Bus=01 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#= 2 Spd=12 MxCh= 0 D: Ver= 2.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS=64 #Cfgs= 1 P: Vendor=04b4 ProdID=6830 Rev= 0.01 S: Manufacturer=Cypress Semiconductor S: Product=USB2.0 Storage Device S: SerialNumber=DEF106207355 C:* #Ifs= 1 Cfg#= 1 Atr=c0 MxPwr= 0mA I: If#= 0 Alt= 0 #EPs= 2 Cls=08(stor.) Sub=06 Prot=50 Driver=usb-storage E: Ad=02(O) Atr=02(Bulk) MxPS= 64 Ivl=0ms E: Ad=88(I) Atr=02(Bulk) MxPS= 64 Ivl=0ms Please advise. Thank you. -------------- next part -------------- An HTML attachment was scrubbed... URL: From gerg at snapgear.com Sun Mar 6 23:58:11 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 07 Mar 2005 14:58:11 +1000 Subject: [uClinux-dev] USB ehci driver not working on Coyote In-Reply-To: <006c01c522b7$45cb7b50$2101a8c0@BCCHUAH> References: <006c01c522b7$45cb7b50$2101a8c0@BCCHUAH> Message-ID: <422BDF63.7010607@snapgear.com> Hi Serena, Serena Kuan wrote: > I have trying out the IXDP425 NAS-USB2 using Snapgear- > 3.0 source distribution, That is very old now. I would suggest using the 3.2 package if you can. > IXP400 v1.3 software and arm- > linux-tools-20030927.tar file, on the COYOTE board. > > As in the example, a NEC-based adapter is being used > too, i.e. Belkin F5U220. > > I am facing a problem as the usb ehci is not working. > Here is the error message during startup: > Using /lib/modules/2.4.22- > uc0/kernel/drivers/usb/host/ehci-hcd.o > hcd.c: Found HC with no IRQ. Check BIOS/PCI 00:0f.2 > setup! > When I checked on the devices status, here is the > message from /proc/pci and seems like there is NO IRC > being sent to the USB2.0 controller. > Bus 0, device 15, function 0: > USB Controller: NEC Corporation USB (rev 65). > IRQ 28. ^^^^^^^^^ There is the interrupt being assigned... I don't recall the exact Linux version in the 3.0 package right now, but I am pretty sure I recall it being buggy on USB2 devices on non-x86 hardware. Regards greg > Master Capable. Latency=128. Min Gnt=1.Max > Lat=42. > Non-prefetchable 32 bit memory at 0x4bfff000 > [0x4bffffff]. > Bus 0, device 15, function 1: > USB Controller: NEC Corporation USB (#2) (rev 65). > IRQ 28. > Master Capable. Latency=128. Min Gnt=1.Max > Lat=42. > Non-prefetchable 32 bit memory at 0x4bffe000 > [0x4bffefff]. > Bus 0, device 15, function 2: > USB Controller: NEC Corporation USB 2.0 (rev 2). > Master Capable. Latency=128. Min Gnt=16.Max > Lat=34. > Non-prefetchable 32 bit memory at 0x4bffdf00 > [0x4bffdfff]. > > Lastly, from /proc/bus/usb/devices, its shown that the > ehci driver is not in used, but I assumed that this is due > to the IRQ not being assigned to the device in the first > place. > T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 > Spd=12 MxCh= 2 > B: Alloc= 0/900 us ( 0%), #Int= 0, #Iso= 0 > D: Ver= 1.10 Cls=09(hub ) Sub=00 Prot=00 MxPS= 8 > #Cfgs= 1 > P: Vendor=0000 ProdID=0000 Rev= 0.00 > S: Product=USB OHCI Root Hub > S: SerialNumber=c28c2000 > C:* #Ifs= 1 Cfg#= 1 Atr=40 MxPwr= 0mA > I: If#= 0 Alt= 0 #EPs= 1 Cls=09(hub ) Sub=00 Prot=00 > Driver=hub > E: Ad=81(I) Atr=03(Int.) MxPS= 2 Ivl=255ms > > T: Bus=01 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 > Spd=12 MxCh= 3 > B: Alloc= 0/900 us ( 0%), #Int= 0, #Iso= 0 > D: Ver= 1.10 Cls=09(hub ) Sub=00 Prot=00 MxPS= 8 > #Cfgs= 1 > P: Vendor=0000 ProdID=0000 Rev= 0.00 > S: Product=USB OHCI Root Hub > S: SerialNumber=c28bc000 > C:* #Ifs= 1 Cfg#= 1 Atr=40 MxPwr= 0mA > I: If#= 0 Alt= 0 #EPs= 1 Cls=09(hub ) Sub=00 Prot=00 > Driver=hub > E: Ad=81(I) Atr=03(Int.) MxPS= 2 Ivl=255ms > > T: Bus=01 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#= 2 > Spd=12 MxCh= 0 > D: Ver= 2.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS=64 > #Cfgs= 1 > P: Vendor=04b4 ProdID=6830 Rev= 0.01 > S: Manufacturer=Cypress Semiconductor > S: Product=USB2.0 Storage Device > S: SerialNumber=DEF106207355 > C:* #Ifs= 1 Cfg#= 1 Atr=c0 MxPwr= 0mA > I: If#= 0 Alt= 0 #EPs= 2 Cls=08(stor.) Sub=06 Prot=50 > Driver=usb-storage > E: Ad=02(O) Atr=02(Bulk) MxPS= 64 Ivl=0ms > E: Ad=88(I) Atr=02(Bulk) MxPS= 64 Ivl=0ms > > Please advise. Thank you. > > > ------------------------------------------------------------------------ > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Mon Mar 7 00:17:33 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 07 Mar 2005 15:17:33 +1000 Subject: [uClinux-dev] Re: [PATCH] vc547x ethernet fixes In-Reply-To: References: Message-ID: <422BE3ED.4080203@snapgear.com> H Harald, Harald K?the wrote: > this patch is against uClinux dist 20041215. > It is for the ethernet driver of the C5471 > > Here's a short list of changes: > - reworked most of the register/bit definitions according to manual > - added option to enable/disable the use of the ESM (Ethernet State machine) > - added option to give all TX descriptors of a packet to HW at once which > fixes TX underrun problems, only implemented if ESM is not used so far. Looks good, applied. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From robert.e.ranslam at intel.com Mon Mar 7 02:01:23 2005 From: robert.e.ranslam at intel.com (Ranslam, Robert E) Date: Sun, 6 Mar 2005 23:01:23 -0800 Subject: [uClinux-dev] USB ehci driver not working on Coyote Message-ID: <8D5F43CD1194B04CAD2AC656844E8F2B04A5F317@orsmsx409> Yes...Very old. Get SG3.2 I do not recall seeing this problem but then I was using an IXDP425. MiniPCI issue perhaps? If you want to see the product this spawned see http://www.nslu2-linux.org/ >-----Original Message----- >From: uclinux-dev-bounces at uClinux.org >[mailto:uclinux-dev-bounces at uClinux.org] On Behalf Of Greg Ungerer >Sent: Sunday, March 06, 2005 8:58 PM >To: Serena Kuan >Cc: uclinux-dev at uClinux.org >Subject: Re: [uClinux-dev] USB ehci driver not working on Coyote > >Hi Serena, > >Serena Kuan wrote: >> I have trying out the IXDP425 NAS-USB2 using Snapgear- 3.0 source >> distribution, > >That is very old now. I would suggest using the 3.2 package if you can. > > >> IXP400 v1.3 software and arm- >> linux-tools-20030927.tar file, on the COYOTE board. >> >> As in the example, a NEC-based adapter is being used too, >i.e. Belkin >> F5U220. >> >> I am facing a problem as the usb ehci is not working. >> Here is the error message during startup: >> Using /lib/modules/2.4.22- >> uc0/kernel/drivers/usb/host/ehci-hcd.o >> hcd.c: Found HC with no IRQ. Check BIOS/PCI 00:0f.2 setup! >> When I checked on the devices status, here is the message from >> /proc/pci and seems like there is NO IRC being sent to the USB2.0 >> controller. >> Bus 0, device 15, function 0: >> USB Controller: NEC Corporation USB (rev 65). >> IRQ 28. >^^^^^^^^^ >There is the interrupt being assigned... > >I don't recall the exact Linux version in the 3.0 package >right now, but I am pretty sure I recall it being buggy on >USB2 devices on >non-x86 hardware. > >Regards >greg > > > > >> Master Capable. Latency=128. Min Gnt=1.Max Lat=42. >> Non-prefetchable 32 bit memory at 0x4bfff000 [0x4bffffff]. >> Bus 0, device 15, function 1: >> USB Controller: NEC Corporation USB (#2) (rev 65). >> IRQ 28. >> Master Capable. Latency=128. Min Gnt=1.Max Lat=42. >> Non-prefetchable 32 bit memory at 0x4bffe000 [0x4bffefff]. >> Bus 0, device 15, function 2: >> USB Controller: NEC Corporation USB 2.0 (rev 2). >> Master Capable. Latency=128. Min Gnt=16.Max Lat=34. >> Non-prefetchable 32 bit memory at 0x4bffdf00 [0x4bffdfff]. >> >> Lastly, from /proc/bus/usb/devices, its shown that the ehci >driver is >> not in used, but I assumed that this is due to the IRQ not being >> assigned to the device in the first place. >> T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 >> Spd=12 MxCh= 2 >> B: Alloc= 0/900 us ( 0%), #Int= 0, #Iso= 0 >> D: Ver= 1.10 Cls=09(hub ) Sub=00 Prot=00 MxPS= 8 #Cfgs= 1 >> P: Vendor=0000 ProdID=0000 Rev= 0.00 >> S: Product=USB OHCI Root Hub >> S: SerialNumber=c28c2000 >> C:* #Ifs= 1 Cfg#= 1 Atr=40 MxPwr= 0mA >> I: If#= 0 Alt= 0 #EPs= 1 Cls=09(hub ) Sub=00 Prot=00 Driver=hub >> E: Ad=81(I) Atr=03(Int.) MxPS= 2 Ivl=255ms >> >> T: Bus=01 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 >> Spd=12 MxCh= 3 >> B: Alloc= 0/900 us ( 0%), #Int= 0, #Iso= 0 >> D: Ver= 1.10 Cls=09(hub ) Sub=00 Prot=00 MxPS= 8 #Cfgs= 1 >> P: Vendor=0000 ProdID=0000 Rev= 0.00 >> S: Product=USB OHCI Root Hub >> S: SerialNumber=c28bc000 >> C:* #Ifs= 1 Cfg#= 1 Atr=40 MxPwr= 0mA >> I: If#= 0 Alt= 0 #EPs= 1 Cls=09(hub ) Sub=00 Prot=00 Driver=hub >> E: Ad=81(I) Atr=03(Int.) MxPS= 2 Ivl=255ms >> >> T: Bus=01 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#= 2 >> Spd=12 MxCh= 0 >> D: Ver= 2.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS=64 #Cfgs= 1 >> P: Vendor=04b4 ProdID=6830 Rev= 0.01 >> S: Manufacturer=Cypress Semiconductor >> S: Product=USB2.0 Storage Device >> S: SerialNumber=DEF106207355 >> C:* #Ifs= 1 Cfg#= 1 Atr=c0 MxPwr= 0mA >> I: If#= 0 Alt= 0 #EPs= 2 Cls=08(stor.) Sub=06 Prot=50 >> Driver=usb-storage >> E: Ad=02(O) Atr=02(Bulk) MxPS= 64 Ivl=0ms >> E: Ad=88(I) Atr=02(Bulk) MxPS= 64 Ivl=0ms >> >> Please advise. Thank you. >> >> >> >---------------------------------------------------------------------- >> -- >> >> _______________________________________________ >> uClinux-dev mailing list >> uClinux-dev at uClinux.org >> http://mailman.uclinux.org/mailman/listinfo/uclinux-dev >> This message was resent by uclinux-dev at uclinux.org > >-- >--------------------------------------------------------------- >--------- >Greg Ungerer -- Chief Software Dude EMAIL: >gerg at snapgear.com >SnapGear -- a CyberGuard Company PHONE: +61 7 >3435 2888 >825 Stanley St, FAX: +61 7 >3891 3630 >Woolloongabba, QLD, 4102, Australia WEB: >http://www.SnapGear.com >_______________________________________________ >uClinux-dev mailing list >uClinux-dev at uClinux.org >http://mailman.uclinux.org/mailman/listinfo/uclinux-dev >This message was resent by uclinux-dev at uclinux.org > From lars.friedrich at wago.com Mon Mar 7 02:15:47 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Mon, 7 Mar 2005 08:15:47 +0100 Subject: [uClinux-dev] polling through kernel-thread or timer routine...? Message-ID: > which approach should be better in terms of system responsiveness given that, > 1> polling routine is not brief and has good amount of processing > 2> my requirement is not very strict about polling frequency. It shouldn't matter regarding system responsiveness, both are not preemptible by the scheduler and therefore increase the latency quite similiar. The timer function will be handled by the timer irq handler though, so it will execute after the clock tick for sure, no matter how busy the system is. According to your requirements it doesn't really matter that much when the job is done so there is no need to give it a timer routine priority. Also it's easier to suspend a kernel thread in the middle of the processing, so you can do some processing, schedule it away to decrease latency and when it returns do the 2nd half. Best regards, Lars Friedrich From vanitha at agilis.stengg.com Mon Mar 7 11:03:31 2005 From: vanitha at agilis.stengg.com (Vanitha Ramaswami) Date: Mon, 7 Mar 2005 16:03:31 -0000 Subject: [uClinux-dev] How to run PPPoE over the WAN interface(/dev/hssi0) ? Message-ID: <01C5232F.35B72FC0.vanitha@agilis.stengg.com> Hi All, I need to run PPPoE over the WAN interface (/dev/hssi0). Its a high speed serial interface. Is there a way for me to specify my WAN interface to the PPPoE client ? I am able to run PPP over the High speed serial interface. To do that, I have written an user space application d then opens a pair of pseudo-terminals and runs PPP over the slave side of those ptys.. But in PPPoE should i modify the pppoe client code to send/receive packets over the High speed serial driver..? Thanks Vanitha [This e-mail is confidential and may be priviledged. If you are not the intended recipient, please kindly notify us immediately and delete the message from your system; please do not copy or use it for any purpose, nor disclose its contents to any other person. Thank you.] ---ST Electronics Group--- From nunolambuca at contar-lda.pt Mon Mar 7 05:35:20 2005 From: nunolambuca at contar-lda.pt (=?iso-8859-1?Q?Nuno_Lambu=E7a?=) Date: Mon, 7 Mar 2005 10:35:20 -0000 Subject: [uClinux-dev] polling through kernel-thread or timer routine...? In-Reply-To: Message-ID: I need some advice also about threads. I have a polling thread, and this polling is time sensitive. I send a question through serial port and expect an answer for a short period of time. What happens is that runnig other apps can make timeout and the answer is there, but not fetched by OS. Any ideas on this? Any way I pass the control just after my complete polling? -----Mensagem original----- De: uclinux-dev-bounces at uClinux.org [mailto:uclinux-dev-bounces at uClinux.org] Em nome de Friedrich, Lars Enviada: segunda-feira, 7 de Mar?o de 2005 7:16 Para: uclinux-dev at uClinux.org Assunto: RE: [uClinux-dev] polling through kernel-thread or timer routine...? > which approach should be better in terms of system responsiveness given that, > 1> polling routine is not brief and has good amount of processing > 2> my requirement is not very strict about polling frequency. It shouldn't matter regarding system responsiveness, both are not preemptible by the scheduler and therefore increase the latency quite similiar. The timer function will be handled by the timer irq handler though, so it will execute after the clock tick for sure, no matter how busy the system is. According to your requirements it doesn't really matter that much when the job is done so there is no need to give it a timer routine priority. Also it's easier to suspend a kernel thread in the middle of the processing, so you can do some processing, schedule it away to decrease latency and when it returns do the 2nd half. Best regards, Lars Friedrich _______________________________________________ uClinux-dev mailing list uClinux-dev at uClinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev at uclinux.org From mkr at kron-systems.de Mon Mar 7 09:42:46 2005 From: mkr at kron-systems.de (Manuel Kron) Date: Mon, 7 Mar 2005 15:42:46 +0100 Subject: [uClinux-dev] MTD/JFFS2/ create/read/write Message-ID: <005c01c52323$efd83a40$3401a8c0@linus> Hi, I have a Coldfire 5272 with uClinux kernel 2.4 on 2MB Flash and 16MB RAM. At the moment the filesystem is "read-only" but I want to read/write files for datalogging and store some configurations. What I have to do in order to write a file on the Flash? When I use MTD: what are the correct kernel options? Regards, Manuel -------------- next part -------------- An HTML attachment was scrubbed... URL: From anders at ontimenet.com Mon Mar 7 09:55:58 2005 From: anders at ontimenet.com (=?iso-8859-1?Q?Anders_=D6hlander?=) Date: Mon, 7 Mar 2005 15:55:58 +0100 Subject: [uClinux-dev] Problem to build userland for uClinux with kernel 2.6 GCC arm-uclinux- version 3.4.0. Message-ID: <20050307145600.59A6D19E8E@uclinux.org> Hi all I have problem to build the romfs with the 2004-12-15 dist. I am using kernel 2.6 and the GCC arm-uclinux- version 3.4.0. I have no problem to build the kernel but the user part is failing. Do you have any suggestions ? make[2]: Entering directory `/home/anders/uClinux-dist/user/init' arm-uclinux-gcc -Os -g -Dlinux -D__linux__ -Dunix -D__uCl inux__ -DEMBED -nostdinc -I/home/anders/uClinux-dist/include -I/home/anders/uCli nux-dist/include/include -fno-builtin -nostartfiles -D__PIC__ -fpic -msingle-pic -base -Wl,-elf2flt /home/anders/uClinux-dist/lib/crt0.o /home/anders/uClinux-dis t/lib/crti.o /home/anders/uClinux-dist/lib/crtn.o -L/home/anders/uClinux-dist/li b -o init simpleinit.o -lcrypt -lc /usr/local/arm-uclinux-tools/lib/gcc/arm-uclinux/3.4.0/../../../../arm-uclin ux/b in/ld.real: ERROR: simpleinit.o uses hardware FP, whereas init.elf2flt uses soft ware FP File format not recognized: failed to merge target specific data of file simplei nit.o collect2: ld returned 1 exit status make[2]: *** [init] Fel 1 make[2]: Leaving directory `/home/anders/uClinux-dist/user/init' make[1]: *** [all] Fel 2 make[1]: Leaving directory `/home/anders/uClinux-dist/user' make: *** [subdirs] Fel 1 [anders at dhcppc6 uClinux-dist]$ BR /Anders ?hlander -------------- next part -------------- An HTML attachment was scrubbed... URL: From mrossell at ya.com Mon Mar 7 10:26:04 2005 From: mrossell at ya.com (Marc Rossell) Date: Mon, 7 Mar 2005 16:26:04 +0100 Subject: [uClinux-dev] MTD/JFFS2/ create/read/write References: <005c01c52323$efd83a40$3401a8c0@linus> Message-ID: <006901c52329$faa5d350$8301a8c0@pentium3> Hi Manuel, Just look at: http://www.enseirb.fr/~kadionik/embedded/uclinux/mtd/howto_mtd.html Regards, ---------------------- | Marc Rossell | ---------------------- Eng. Superior Inform?tic www.englobatec.org Telf. 937620907-665029452 Fax. 937620907 Englobatec, Sinergia S.L. ----- Original Message ----- From: Manuel Kron To: uclinux-dev at uclinux.org Sent: Monday, March 07, 2005 3:42 PM Subject: [uClinux-dev] MTD/JFFS2/ create/read/write Hi, I have a Coldfire 5272 with uClinux kernel 2.4 on 2MB Flash and 16MB RAM. At the moment the filesystem is "read-only" but I want to read/write files for datalogging and store some configurations. What I have to do in order to write a file on the Flash? When I use MTD: what are the correct kernel options? Regards, Manuel ------------------------------------------------------------------------------ _______________________________________________ uClinux-dev mailing list uClinux-dev at uClinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev at uclinux.org -------------- next part -------------- An HTML attachment was scrubbed... URL: From mbeaulieu at eocycle.com Mon Mar 7 10:29:39 2005 From: mbeaulieu at eocycle.com (Martin Beaulieu) Date: Mon, 7 Mar 2005 10:29:39 -0500 Subject: [uClinux-dev] Booting romfs with blkmem in ROM References: <000e01c51f67$42675d10$8000a8c0@MartinLogiciel> Message-ID: <01e201c5232a$7ae00ef0$8000a8c0@MartinLogiciel> Hi, I use a 5407 on my own board. I boot kernel in rom memory. UClinux kernel boot correctly. But when i try to load the romfs, i have some problems. I use the blkmem utility. I have the following message : Blkmem 1 disk images: 0: 18F2000-1931FFF [VIRTUAL 18F2000-1931FFF] (RO) VFS: Mounted root (romfs filesystem) readonly. After, i have a message that ramfs.img can't load in ram1 or ram0. Finally, i have the path and i try to load my application but the software freeze. What is my problem ? I need help thanks Martin -------------- next part -------------- An HTML attachment was scrubbed... URL: From fabrice.poulin at medianesysteme.com Mon Mar 7 11:50:05 2005 From: fabrice.poulin at medianesysteme.com (Fabrice Poulin) Date: Mon, 7 Mar 2005 17:50:05 +0100 Subject: [uClinux-dev] (no subject) Message-ID: hi ! Does someone can tell me where can find a example code for external interrupts INT[6:1] ? I have Coldfire board with 5272 with a version of uClinux. Thank Fabrice From lucy.wang at philips.com Tue Mar 1 21:22:42 2005 From: lucy.wang at philips.com (lucy.wang at philips.com) Date: Wed, 2 Mar 2005 10:22:42 +0800 Subject: [uClinux-dev] [PATCH]linux-2.6.5 patch for Philips LPC22xx Message-ID: Hi All, Attached is the linux-2.6.5 patch for Philips LPC22xx platform. In the next e-mail, I'll post the uClinux patch and a readme file for Philips LPC22xx. Best regards, Lucy -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: linux-2.6.5-lpc22xx.patch.gz Type: application/octet-stream Size: 189215 bytes Desc: not available URL: From swenggsri at yahoo.com Mon Mar 7 13:29:56 2005 From: swenggsri at yahoo.com (Sriram Chadalavada) Date: Mon, 7 Mar 2005 10:29:56 -0800 (PST) Subject: [uClinux-dev] 2.6 kernel image size HUGE!!!! In-Reply-To: Message-ID: <20050307182956.48332.qmail@web21601.mail.yahoo.com> Hi everyone, I am trying to download a 2.6.x kernel (part of latest 20041215 distribution) on a Arcturus uCdimm that has 8MB RAM and 2 MB ROM. Unfortunately, when I try compiling the uClinux distribution with this kernel, the image size is turning to be a whopping 272MB!!!!! total 272M -rw-r--r-- 1 root root 270M Feb 21 18:57 image.bin -rwxr-xr-x 1 root root 112K Feb 21 18:57 linux.data -rwxr-xr-x 1 root root 269M Feb 21 18:57 linux.text -rw-r--r-- 1 root root 791K Feb 21 18:57 romfs.img linux.text seems to be the main culprit... When I compile with the 2.4.x kernel, the sizes were: total 2.8M -rw-r--r-- 1 root root 1.4M Mar 5 21:43 image.bin -rwxr-xr-x 1 root root 168K Mar 5 21:43 linux.data -rwxr-xr-x 1 root root 687K Mar 5 21:43 linux.text -rw-r--r-- 1 root root 540K Mar 5 21:43 romfs.img That's almost 10 times smaller. Some postings on the mailing list did talk about this problem but I was not able to find a solution. What is the typical (default for uCdimm) image size with the 2.6.x kernel? Is it always larger than than 1MB? Please let me know if you need any other information to solve this problem. Thanks, Sriram __________________________________ Celebrate Yahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of the Web http://birthday.yahoo.com/netrospective/ From hansgnu at yahoo.com Mon Mar 7 15:20:55 2005 From: hansgnu at yahoo.com (Joao Antonio P Rabelo HansGNU) Date: Mon, 7 Mar 2005 12:20:55 -0800 (PST) Subject: [uClinux-dev] User Programs Message-ID: <20050307202055.41566.qmail@web53802.mail.yahoo.com> Hello, I try to modify uClinux to add more user's programs, but i must modify "/uclinux/user/Makefile" and other conf/makefiles. How can I do it? Anyone can help me to do it? thanks, Joao Antonio ===== .''`. Jo?o Ant?nio de Paiva Rabelo ::' : IRC - irc.freenode.net/6667 - #Debian-BR #MinaSLivre `.`'` UIN - 61461402 JABBER - hansgnu at jabber.org `- Nunca deixe de perguntar e use DEBIAN !!! __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com From mkr at kron-systems.de Mon Mar 7 16:22:54 2005 From: mkr at kron-systems.de (Manuel Kron) Date: Mon, 7 Mar 2005 22:22:54 +0100 Subject: [uClinux-dev] Writing a file to filesystem Message-ID: <010301c5235b$d5f13cc0$3401a8c0@linus> Hi all, I have a simple question: How can I write a file or some blocks to the Flash? I have a Coldfire 5272 with uClinux kernel 2.4 on 2MB Flash and 16MB RAM. Do you have a idea? Thanks and Regards, Manuel -------------- next part -------------- An HTML attachment was scrubbed... URL: From ctusar at videon-central.com Mon Mar 7 16:43:25 2005 From: ctusar at videon-central.com (Cory Tusar) Date: Mon, 07 Mar 2005 16:43:25 -0500 Subject: [uClinux-dev] User Programs In-Reply-To: <20050307202055.41566.qmail@web53802.mail.yahoo.com> References: <20050307202055.41566.qmail@web53802.mail.yahoo.com> Message-ID: <422CCAFD.9090508@videon-central.com> Joao Antonio P Rabelo HansGNU wrote: > Hello, > > I try to modify uClinux to add more user's programs, > but i must modify "/uclinux/user/Makefile" and other > conf/makefiles. How can I do it? Anyone can help me to > do it? Start by reading uClinux-dist/Documentation/Adding-User-Apps-HOWTO Follow that with uClinux-dist/config/scripts/README.uClinux Follow that with uClinux-dist/config/scripts/README.Menuconfig Follow that with uCLinux-dist/linux-2.4.x/Documentation/kbuild/* -Cory -- Cory T. Tusar Embedded Systems Engineer Videon Central, Inc. 2171 Sandy Drive State College, PA 16801 (814) 235-1111 x316 (814) 235-1118 fax "Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it." --Brian W. Kernighan From davidm at snapgear.com Mon Mar 7 16:47:13 2005 From: davidm at snapgear.com (David McCullough) Date: Tue, 8 Mar 2005 07:47:13 +1000 Subject: [uClinux-dev] Booting romfs with blkmem in ROM In-Reply-To: <01e201c5232a$7ae00ef0$8000a8c0@MartinLogiciel> References: <000e01c51f67$42675d10$8000a8c0@MartinLogiciel> <01e201c5232a$7ae00ef0$8000a8c0@MartinLogiciel> Message-ID: <20050307214713.GA6577@beast> Jivin Martin Beaulieu lays it down ... > Hi, > I use a 5407 on my own board. I boot kernel in rom memory. UClinux kernel boot correctly. > > But when i try to load the romfs, i have some problems. I use the blkmem utility. > > I have the following message : > Blkmem 1 disk images: > 0: 18F2000-1931FFF [VIRTUAL 18F2000-1931FFF] (RO) > VFS: Mounted root (romfs filesystem) readonly. > > After, i have a message that ramfs.img can't load in ram1 or ram0. > > Finally, i have the path and i try to load my application but the software freeze. > > What is my problem ? Hard to say, why not send in the all the messages so we can see what is actually happening ? Do you have RAM disk support in your kernel ? Which kernel is it exactly, 2.0, 2.4 or 2.6 ? Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From davidm at snapgear.com Mon Mar 7 16:53:43 2005 From: davidm at snapgear.com (David McCullough) Date: Tue, 8 Mar 2005 07:53:43 +1000 Subject: [uClinux-dev] Problem to build userland for uClinux with kernel 2.6 GCC arm-uclinux- version 3.4.0. In-Reply-To: <20050307145600.59A6D19E8E@uclinux.org> References: <20050307145600.59A6D19E8E@uclinux.org> Message-ID: <20050307215343.GC6577@beast> Jivin Anders ?hlander lays it down ... > Hi all > > I have problem to build the romfs with the 2004-12-15 dist. I am using > kernel 2.6 and the GCC arm-uclinux- version 3.4.0. > > I have no problem to build the kernel but the user part is failing. > > > > Do you have any suggestions ? Looks like you may need a -msoft-float for this compiler. Add it to the vendors/config/armnommu/config.arch or perhaps to the CPU flags in: vendors///config.arch Cheers, Davidm > make[2]: Entering directory `/home/anders/uClinux-dist/user/init' > > arm-uclinux-gcc -Os -g -Dlinux -D__linux__ -Dunix > -D__uCl > > inux__ -DEMBED -nostdinc -I/home/anders/uClinux-dist/include > -I/home/anders/uCli > > nux-dist/include/include -fno-builtin -nostartfiles -D__PIC__ -fpic > -msingle-pic > > -base -Wl,-elf2flt /home/anders/uClinux-dist/lib/crt0.o > /home/anders/uClinux-dis > > t/lib/crti.o /home/anders/uClinux-dist/lib/crtn.o > -L/home/anders/uClinux-dist/li > > b -o init simpleinit.o -lcrypt -lc > > /usr/local/arm-uclinux-tools/lib/gcc/arm-uclinux/3.4.0/../../../../arm-uclin > ux/b > > in/ld.real: ERROR: simpleinit.o uses hardware FP, whereas init.elf2flt uses > soft > > ware FP > > File format not recognized: failed to merge target specific data of file > simplei > > nit.o > > collect2: ld returned 1 exit status > > make[2]: *** [init] Fel 1 > > make[2]: Leaving directory `/home/anders/uClinux-dist/user/init' > > make[1]: *** [all] Fel 2 > > make[1]: Leaving directory `/home/anders/uClinux-dist/user' > > make: *** [subdirs] Fel 1 > > [anders at dhcppc6 uClinux-dist]$ > > > > > > BR > > /Anders ?hlander > > > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From ben at thingmagic.com Mon Mar 7 19:57:49 2005 From: ben at thingmagic.com (Ben Giddings) Date: Mon, 07 Mar 2005 19:57:49 -0500 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like)system In-Reply-To: <20050303231108.32192.qmail@web51910.mail.yahoo.com> References: <20050303231108.32192.qmail@web51910.mail.yahoo.com> Message-ID: <422CF88D.5090404@thingmagic.com> Doug Kehn wrote: > I followed the steps as given in the ADI-COYOTE-HOWTO > for uClinux20041215 (Snapgear 3.2 has the same HOWTO > and Snapgear 3.2 is I believe equivalent to > uClinux20041215). I used the following packages when > following the HOWTO: > > arm-linux-tools-20031127.tar.gz > uClinux20041215.tar.bz2 > snapgear-3.1.x-ixp400-v1.4.2-support.tar.gz > ixp400AccessLibrary-1_4.zip Hmm, so this one doesn't require the shar file? I haven't been able to get this to work yet. When I try to compile a straight IXDP425 build, it dies when trying to build the Intel USB modules. Ben From ben at thingmagic.com Mon Mar 7 20:00:33 2005 From: ben at thingmagic.com (Ben Giddings) Date: Mon, 07 Mar 2005 20:00:33 -0500 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like)system In-Reply-To: <422805B4.4060900@snapgear.com> References: <20050303150346.30406.qmail@web51904.mail.yahoo.com> <42277797.6010806@thingmagic.com> <422805B4.4060900@snapgear.com> Message-ID: <422CF931.3010004@thingmagic.com> Greg Ungerer wrote: >> Hmm, when I tried to apply the patch, it wouldn't apply: >> >> patching file modules/ixp425/ixp400/ixp400_xscale_sw/Makefile >> Reversed (or previously applied) patch detected! Assume -R? [n] >> >> I looked at modules/ixp425/ixp400/ixp400_xscale_sw/Makefile and it >> looks like the patches are already there. > > > Had you already applied Rob's patch first? > You would only need to apply one patch or the other... No. That was using the 'vanilla' ixp400AccessLibrary-1_4.zip, the snapgear3.2 distro, and the modules-20041130.patch. I'll try again if this configuration should work, but it didn't work for me. Ben From ben at thingmagic.com Mon Mar 7 20:01:11 2005 From: ben at thingmagic.com (Ben Giddings) Date: Mon, 07 Mar 2005 20:01:11 -0500 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like) system In-Reply-To: <4226649D.9050200@snapgear.com> References: <4224A62A.3040909@thingmagic.com> <4225163D.4010601@snapgear.com> <42264B00.3060904@thingmagic.com> <4226649D.9050200@snapgear.com> Message-ID: <422CF957.9080204@thingmagic.com> Greg Ungerer wrote: >> As someone else suggested, I'm going to try the "IXP400 Access Library >> patch" from the snapgear site. Is it an alternative to the patches at >> the ixp4xx-osdg site? When you're building, are you using Intel's >> library, and if so what method are you using to integrate their patches? > > > The snapgear patch was primarily designed to add support for the IXDG425 > (montejade) board to the AccessLibrary. Not so much to replace Rob's > patch on ixp4xx-osdg. > > As you have found out though the snapgear AccessLibrary patch also > requires the Intel DSR library (the ixdg425 has a bunch of voice > ports on it). Ok, I read another message that says that it isn't actually required, that you can just ignore the messages saying it is. I'll try again with it tomorrow. Ben From serena at flex-p.com Mon Mar 7 20:41:52 2005 From: serena at flex-p.com (Serena Kuan) Date: Tue, 8 Mar 2005 09:41:52 +0800 Subject: [uClinux-dev] USB ehci driver not working on Coyote References: <8D5F43CD1194B04CAD2AC656844E8F2B04A5F317@orsmsx409> Message-ID: <002e01c52380$0317bac0$2101a8c0@BCCHUAH> I was thinking of using SG3.16 instead, as I think there isn't any IXDP425 support for SG3.2 yet. Would that be alright? ----- Original Message ----- From: "Ranslam, Robert E" To: "Greg Ungerer" ; "Serena Kuan" Cc: Sent: Monday, March 07, 2005 3:01 PM Subject: RE: [uClinux-dev] USB ehci driver not working on Coyote Yes...Very old. Get SG3.2 I do not recall seeing this problem but then I was using an IXDP425. MiniPCI issue perhaps? If you want to see the product this spawned see http://www.nslu2-linux.org/ >-----Original Message----- >From: uclinux-dev-bounces at uClinux.org >[mailto:uclinux-dev-bounces at uClinux.org] On Behalf Of Greg Ungerer >Sent: Sunday, March 06, 2005 8:58 PM >To: Serena Kuan >Cc: uclinux-dev at uClinux.org >Subject: Re: [uClinux-dev] USB ehci driver not working on Coyote > >Hi Serena, > >Serena Kuan wrote: >> I have trying out the IXDP425 NAS-USB2 using Snapgear- 3.0 source >> distribution, > >That is very old now. I would suggest using the 3.2 package if you can. > > >> IXP400 v1.3 software and arm- >> linux-tools-20030927.tar file, on the COYOTE board. >> >> As in the example, a NEC-based adapter is being used too, >i.e. Belkin >> F5U220. >> >> I am facing a problem as the usb ehci is not working. >> Here is the error message during startup: >> Using /lib/modules/2.4.22- >> uc0/kernel/drivers/usb/host/ehci-hcd.o >> hcd.c: Found HC with no IRQ. Check BIOS/PCI 00:0f.2 setup! >> When I checked on the devices status, here is the message from >> /proc/pci and seems like there is NO IRC being sent to the USB2.0 >> controller. >> Bus 0, device 15, function 0: >> USB Controller: NEC Corporation USB (rev 65). >> IRQ 28. >^^^^^^^^^ >There is the interrupt being assigned... > >I don't recall the exact Linux version in the 3.0 package >right now, but I am pretty sure I recall it being buggy on >USB2 devices on >non-x86 hardware. > >Regards >greg > >--------------------------------------------------------------- >--------- >Greg Ungerer -- Chief Software Dude EMAIL: >gerg at snapgear.com >SnapGear -- a CyberGuard Company PHONE: +61 7 >3435 2888 >825 Stanley St, FAX: +61 7 >3891 3630 >Woolloongabba, QLD, 4102, Australia WEB: >http://www.SnapGear.com >_______________________________________________ >uClinux-dev mailing list >uClinux-dev at uClinux.org >http://mailman.uclinux.org/mailman/listinfo/uclinux-dev >This message was resent by uclinux-dev at uclinux.org > From devi at tenet.res.in Tue Mar 8 12:15:32 2005 From: devi at tenet.res.in (devi) Date: Tue, 8 Mar 2005 14:15:32 -0300 Subject: [uClinux-dev] keyboard Message-ID: <20050308171532.10820.qmail@mail.tenet.res.in> hi, When I type a character l and s for the command ls and press the enter key my cursor comes to the nextline(carriage return + line feed) but the command is not getting executed.Any idea Regards Devi From gerg at snapgear.com Tue Mar 8 01:37:21 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Tue, 08 Mar 2005 16:37:21 +1000 Subject: [uClinux-dev] USB ehci driver not working on Coyote In-Reply-To: <002e01c52380$0317bac0$2101a8c0@BCCHUAH> References: <8D5F43CD1194B04CAD2AC656844E8F2B04A5F317@orsmsx409> <002e01c52380$0317bac0$2101a8c0@BCCHUAH> Message-ID: <422D4821.8050506@snapgear.com> Hi Serena, Serena Kuan wrote: > I was thinking of using SG3.16 instead, as I think there isn't any IXDP425 > support for SG3.2 yet. Would that be alright? The IXDP425 support in SG3.2 is the same as SG3.16. Regards Greg > ----- Original Message ----- > From: "Ranslam, Robert E" > To: "Greg Ungerer" ; "Serena Kuan" > Cc: > Sent: Monday, March 07, 2005 3:01 PM > Subject: RE: [uClinux-dev] USB ehci driver not working on Coyote > > > Yes...Very old. Get SG3.2 > > I do not recall seeing this problem but then I was using an IXDP425. > MiniPCI issue perhaps? > > If you want to see the product this spawned see > http://www.nslu2-linux.org/ > > > > >>-----Original Message----- >>From: uclinux-dev-bounces at uClinux.org >>[mailto:uclinux-dev-bounces at uClinux.org] On Behalf Of Greg Ungerer >>Sent: Sunday, March 06, 2005 8:58 PM >>To: Serena Kuan >>Cc: uclinux-dev at uClinux.org >>Subject: Re: [uClinux-dev] USB ehci driver not working on Coyote >> >>Hi Serena, >> >>Serena Kuan wrote: >> >>>I have trying out the IXDP425 NAS-USB2 using Snapgear- 3.0 source >>>distribution, >> >>That is very old now. I would suggest using the 3.2 package if you can. >> >> >> >>>IXP400 v1.3 software and arm- >>>linux-tools-20030927.tar file, on the COYOTE board. >>> >>>As in the example, a NEC-based adapter is being used too, >> >>i.e. Belkin >> >>>F5U220. >>> >>>I am facing a problem as the usb ehci is not working. >>>Here is the error message during startup: >>>Using /lib/modules/2.4.22- >>>uc0/kernel/drivers/usb/host/ehci-hcd.o >>>hcd.c: Found HC with no IRQ. Check BIOS/PCI 00:0f.2 setup! >>>When I checked on the devices status, here is the message from >>>/proc/pci and seems like there is NO IRC being sent to the USB2.0 >>>controller. >>>Bus 0, device 15, function 0: >>>USB Controller: NEC Corporation USB (rev 65). >>>IRQ 28. >> >>^^^^^^^^^ >>There is the interrupt being assigned... >> >>I don't recall the exact Linux version in the 3.0 package >>right now, but I am pretty sure I recall it being buggy on >>USB2 devices on >>non-x86 hardware. >> >>Regards >>greg >> >>--------------------------------------------------------------- >>--------- >>Greg Ungerer -- Chief Software Dude EMAIL: >>gerg at snapgear.com >>SnapGear -- a CyberGuard Company PHONE: +61 7 >>3435 2888 >>825 Stanley St, FAX: +61 7 >>3891 3630 >>Woolloongabba, QLD, 4102, Australia WEB: >>http://www.SnapGear.com >>_______________________________________________ >>uClinux-dev mailing list >>uClinux-dev at uClinux.org >>http://mailman.uclinux.org/mailman/listinfo/uclinux-dev >>This message was resent by uclinux-dev at uclinux.org >> > > > > -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Tue Mar 8 01:55:16 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Tue, 08 Mar 2005 16:55:16 +1000 Subject: [uClinux-dev] 2.6 kernel image size HUGE!!!! In-Reply-To: <20050307182956.48332.qmail@web21601.mail.yahoo.com> References: <20050307182956.48332.qmail@web21601.mail.yahoo.com> Message-ID: <422D4C54.4020400@snapgear.com> Hi Sriram, Sriram Chadalavada wrote: > I am trying to download a 2.6.x kernel (part of > latest 20041215 distribution) on a Arcturus uCdimm > that has 8MB RAM and 2 MB ROM. > > Unfortunately, when I try compiling the uClinux > distribution with this kernel, the image size is > turning to be a whopping 272MB!!!!! The problem turns out to be that the final objcopy to create the linux.text file is not removing the .init section, as it should be. > total 272M > -rw-r--r-- 1 root root 270M Feb 21 18:57 image.bin > -rwxr-xr-x 1 root root 112K Feb 21 18:57 linux.data > -rwxr-xr-x 1 root root 269M Feb 21 18:57 linux.text > -rw-r--r-- 1 root root 791K Feb 21 18:57 romfs.img > > linux.text seems to be the main culprit... > > When I compile with the 2.4.x kernel, the sizes were: > > total 2.8M > -rw-r--r-- 1 root root 1.4M Mar 5 21:43 image.bin > -rwxr-xr-x 1 root root 168K Mar 5 21:43 linux.data > -rwxr-xr-x 1 root root 687K Mar 5 21:43 linux.text > -rw-r--r-- 1 root root 540K Mar 5 21:43 romfs.img > > That's almost 10 times smaller. Some postings on the > mailing list did talk about this problem but I was not > able to find a solution. > > What is the typical (default for uCdimm) image size > with the 2.6.x kernel? Is it always larger than than > 1MB? Please let me know if you need any other > information to solve this problem. The image size should be only marginally larger than the 2.4 image. Try the attached patch taht fixes the kernel text objcopy. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: ucdimm.patch URL: From lars.friedrich at wago.com Tue Mar 8 01:56:27 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Tue, 8 Mar 2005 07:56:27 +0100 Subject: [uClinux-dev] polling through kernel-thread or timer routine...? Message-ID: > I need some advice also about threads. > I have a polling thread, and this polling is time sensitive. I send a > question through serial port and expect an answer for a short > period of time. What happens is that runnig other apps can make timeout > and the answer is there, but not fetched by OS. This sounds like you are talking about user space. I suspect a problem in your application, I doubt the kernel ignores the serial port. Without a source code snippet showing the important parts there is not much to say about it though. Best regards, Lars Friedrich From lars.friedrich at wago.com Tue Mar 8 02:05:43 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Tue, 8 Mar 2005 08:05:43 +0100 Subject: [uClinux-dev] Writing a file to filesystem Message-ID: > I have a simple question: How can I write a file or > some blocks to the Flash? cp file /dev/mtdblockX Please be more specific. Simple questions get simple answers. As you already got the pointer to mtd/jffs2 info I suspect this is all you wanted to know. Best regards, Lars Friedrich From gerg at snapgear.com Tue Mar 8 02:12:28 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Tue, 08 Mar 2005 17:12:28 +1000 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like) system In-Reply-To: <422CF957.9080204@thingmagic.com> References: <4224A62A.3040909@thingmagic.com> <4225163D.4010601@snapgear.com> <42264B00.3060904@thingmagic.com> <4226649D.9050200@snapgear.com> <422CF957.9080204@thingmagic.com> Message-ID: <422D505C.8010208@snapgear.com> Hi Ben, Ben Giddings wrote: > Greg Ungerer wrote: > >>> As someone else suggested, I'm going to try the "IXP400 Access Library >>> patch" from the snapgear site. Is it an alternative to the patches at >>> the ixp4xx-osdg site? When you're building, are you using Intel's >>> library, and if so what method are you using to integrate their patches? >> >> >> >> The snapgear patch was primarily designed to add support for the IXDG425 >> (montejade) board to the AccessLibrary. Not so much to replace Rob's >> patch on ixp4xx-osdg. >> >> As you have found out though the snapgear AccessLibrary patch also >> requires the Intel DSR library (the ixdg425 has a bunch of voice >> ports on it). > > > Ok, I read another message that says that it isn't actually required, > that you can just ignore the messages saying it is. I'll try again with > it tomorrow. Yes, try that. You will get a bunch of warnings about some DSR files not being present - but ignore those. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From lars.friedrich at wago.com Tue Mar 8 02:24:47 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Tue, 8 Mar 2005 08:24:47 +0100 Subject: [uClinux-dev] keyboard Message-ID: > When I type a character l and s for the command ls and press > the enter key my cursor comes to the nextline(carriage return > + line feed) but the command is not getting executed.Any idea I assume that you do something like printk(key); Not sure why you expect that something else will happen. How shall the system know about the key? You need to translate the keycode to a keysymbol for the tty terminal driver and then pass the keysymbol to the input buffer of the tty driver. Then other applications can fetch the data from the tty and act accordingly. Examine an existing keyboard driver for something like keysym = key_map[keycode]; This is where the mapping from keycode to keysymbol is done. Then you might want to scan an existing keyboard driver for everything tty related in order to figure out how the passing is done. Best regards, Lars Friedrich From H.Haeberle at pilz.de Tue Mar 8 02:36:08 2005 From: H.Haeberle at pilz.de (Heinz Haeberle) Date: Tue, 8 Mar 2005 08:36:08 +0100 Subject: [uClinux-dev] is there a patch for 2.6.11 coming? In-Reply-To: <20050307170007.E2E0C19F7A@uclinux.org> Message-ID: I am planing on moving to the newest kernel - at least for a try. I am using 2.6.6 right now. To send patches I obviously need to do so with the latest kernel sources to make it possible to apply the patches into the CVS - if somebody is in the mood for doing so. Which kernel should I choose for starting? The 2.6.10-uc0 or the 2.6.11 vanilla? Heinz From lars.friedrich at wago.com Tue Mar 8 02:53:47 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Tue, 8 Mar 2005 08:53:47 +0100 Subject: [uClinux-dev] [PATCH]uClinux patch for Philips LPC22xx Message-ID: > Attached is the uClinux patch for Philips LPC22xx, including the > vendors/Philips/LPC22xx directory and a new config.in file. Is it just me or are there plenty of non-ascii characters in the comments? ?????????????? is not really understandable for most. uClinux-2.4.x/arch/sh/vmlinux.lds.rom.S: - + (_mem_start - _rom_store) + _ctr; + + (_mem_start - _rom_store); + . = ALIGN(16); IIRC ALIGN(16) will align to the _next_ boundary if the address is already aligned. Although I can understand your intention to get rid of the _ctr symbol, I'm not really convinced the solution is better. Or are there other reasons for the change? Care to elaborate your motivation? Best regards, Lars Friedrich From saiful at modular-corp.com Tue Mar 8 03:21:46 2005 From: saiful at modular-corp.com (saiful) Date: Tue, 8 Mar 2005 16:21:46 +0800 Subject: [uClinux-dev] 'GLIBC_2.3' not found Message-ID: <200503081621.AA2123104382@modular-corp.com> Hi, For some reasons, selecting some functions in Make Menuconfig produced this warning message while compiling: /snapgear/glibc/build/ncurses/build-tic /home/uclinux/snapgear/glibc/build/libc.so.6:versi on 'GLIBC_2.3' not found (required by /usr/lib/libncurses.so.5) when I did ldd: ldd libc.so.6 /libc.so.6: /lib/ld-linux.so.2: version 'GLIBC_2.1.1' not found /libc.so.6: /lib/ld-linux.so.2: version 'GLIBC_2.2.3' not found /libc.so.6: /lib/ld-linux.so.2: version 'GLIBC_2.2' not found I am not sure why libc.so.6 looked for other version of glibc and not 'GLIBC-2.3'? FYI, I am using Redhat 9 and supposed to have GLIBC_2.3. I don't understand why this can happen. Doing 'ldconfig' did not solve my problem actually Saiful From prsbarc1 at yahoo.co.in Tue Mar 8 04:08:52 2005 From: prsbarc1 at yahoo.co.in (Rakesh Ranjan) Date: Tue, 8 Mar 2005 09:08:52 +0000 (GMT) Subject: [uClinux-dev] can skb_reserve be used safely for passing proprietary header..? Message-ID: <20050308090852.93111.qmail@web8506.mail.in.yahoo.com> my requirement is as follow:- I have some routing UDP applications running in userspace while sending their standard packets they need to prepend the standard packets with a 16 byte proprietary header and send it to switch ingress/egrees processing interface , packet after getting processed from there would be forwarded to appropiate ethernet ports. I want to integrate application with ip stack . for that I registered my own udp_sendmsg () routine to handle 16 byte proprietary header. my question is that , would it be possible to pass this 16 byte header to lower layer , by allocating additional 16 bytes using skb_reserve() and pass the standard packet as such and further lower layer calling proprietary routine through hard_start_xmit() will take care of prepending 16 bytes of header from skb headroom .... Regards, Rakesh Yahoo! India Matrimony: Find your life partneronline. -------------- next part -------------- An HTML attachment was scrubbed... URL: From mrossell at ya.com Tue Mar 8 04:13:05 2005 From: mrossell at ya.com (Marc Rossell) Date: Tue, 8 Mar 2005 10:13:05 +0100 Subject: [uClinux-dev] MTD/JFFS2/ create/read/write References: <005c01c52323$efd83a40$3401a8c0@linus> <006901c52329$faa5d350$8301a8c0@pentium3> <00ce01c52338$346dd0c0$3401a8c0@linus> Message-ID: <004601c523bf$0a5d2790$8301a8c0@pentium3> Hi Manuel, you need the MTD_UTILS if you want for example to erase a MTD partition or to create a JFFSX filesystem in Flash memory. I don't know if your idea is to create a file system in that memory, but you will need least 5 or 6 contiguous sectors to be able to create one. With your previous configuration this isn't possible. You only can access in RAW format to sectors 9 and 10 if you creata a MTD partition with them. (cp file /dev/mtdX, mv file /dev/mtdX, cp /dev/mtdX file, vi /dev/mtdX...). Regards, Marc ----- Original Message ----- From: Manuel Kron To: Marc Rossell Sent: Monday, March 07, 2005 6:07 PM Subject: Re: [uClinux-dev] MTD/JFFS2/ create/read/write Hi Marc, thanks for the tip. Now my flash has 11 sectors: 0 16k Bootloader (colio) 1 8k Firmware config 2 8k 3 224k 4 256k uClinux + Firmware (romfs) 5 256k uClinux + Firmware (romfs) 6 256k uClinux + Firmware (romfs) 7 256k uClinux + Firmware (romfs) 8 256k uClinux + Firmware (romfs) 9 256k This is a free sector 10 256k This is a free sector What I want is to configure the last sectors for data writing (application). My options on the linux kernel 2.4: Memory Technology Devices (MTD) ---> <*> Memory Technology Device (MTD) support Under /dev/ I found -> /dev/mtd0..6 ->/dev/mtdblock0..6 When add the mtd-utils on the kernel menuconfig, my new kernel isn't able to boot! --- MTD utils [*] mtd-utils Now: what's are the options in the menuconfig in order to enable the mt-utils. (do I need the mtd-utils?) Do you have a idea? regards, Manuel ----- Original Message ----- From: Marc Rossell To: uclinux-dev at uclinux.org Sent: Monday, March 07, 2005 4:26 PM Subject: Re: [uClinux-dev] MTD/JFFS2/ create/read/write Hi Manuel, Just look at: http://www.enseirb.fr/~kadionik/embedded/uclinux/mtd/howto_mtd.html Regards, ---------------------- | Marc Rossell | ---------------------- Eng. Superior Inform?tic www.englobatec.org Telf. 937620907-665029452 Fax. 937620907 Englobatec, Sinergia S.L. ----- Original Message ----- From: Manuel Kron To: uclinux-dev at uclinux.org Sent: Monday, March 07, 2005 3:42 PM Subject: [uClinux-dev] MTD/JFFS2/ create/read/write Hi, I have a Coldfire 5272 with uClinux kernel 2.4 on 2MB Flash and 16MB RAM. At the moment the filesystem is "read-only" but I want to read/write files for datalogging and store some configurations. What I have to do in order to write a file on the Flash? When I use MTD: what are the correct kernel options? Regards, Manuel -------------------------------------------------------------------------- _______________________________________________ uClinux-dev mailing list uClinux-dev at uClinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev at uclinux.org ---------------------------------------------------------------------------- _______________________________________________ uClinux-dev mailing list uClinux-dev at uClinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev at uclinux.org -------------- next part -------------- An HTML attachment was scrubbed... URL: From rdkehn at yahoo.com Tue Mar 8 06:54:36 2005 From: rdkehn at yahoo.com (Doug Kehn) Date: Tue, 8 Mar 2005 03:54:36 -0800 (PST) Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like)system In-Reply-To: 6667 Message-ID: <20050308115436.11503.qmail@web51906.mail.yahoo.com> --- Ben Giddings wrote: > > Hmm, so this one doesn't require the shar file? > Correct. No shar file required. > I haven't been able to get this to work yet. When I > try to compile a > straight IXDP425 build, it dies when trying to build > the Intel USB modules. > Hmm. Did you start with a fresh install? Do you encounter the same problem when attempting to compile a straight Coyote? What is the compile error? Regards, ...doug __________________________________ Celebrate Yahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of the Web http://birthday.yahoo.com/netrospective/ From gerg at snapgear.com Tue Mar 8 08:00:44 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Tue, 08 Mar 2005 23:00:44 +1000 Subject: [uClinux-dev] is there a patch for 2.6.11 coming? In-Reply-To: References: Message-ID: <422DA1FC.2090100@snapgear.com> Hi Heinz, Heinz Haeberle wrote: > I am planing on moving to the newest kernel - at least for a try. I am > using 2.6.6 right now. > To send patches I obviously need to do so with the latest kernel sources to > make it possible > to apply the patches into the CVS - if somebody is in the mood for doing > so. There is no CVS repository for the uCLinux 2.6 patches... > Which kernel should I choose for starting? The 2.6.10-uc0 or the 2.6.11 > vanilla? I would suggest using linux-2.6.10-uc0 right now. Non-mmu support in 2.6.11 is rather broken right now. I am working on some fixes :-) Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From danders at amltd.com Tue Mar 8 09:56:11 2005 From: danders at amltd.com (David Anders) Date: Tue, 08 Mar 2005 08:56:11 -0600 Subject: [uClinux-dev] Possible GPL violation in JuiceBox distributed by emsoft Message-ID: <422DBD0B.1000100@amltd.com> Fellow developers: after reading this post http://www.ucdot.org/articles/05/01/24/1244222.shtml , i purchased a juicebox to see if i could make some customizations to the unit. i was unable to find posted source code, and no offer anywhere to be found for gettting source. i finally email the sales department at Emsoft regarding getting source code. shortly there after Martin Siu post this message on the uclinux-dev mailing list - http://mailman.uclinux.org/pipermail/uclinux-dev/2005-February/031206.html , but he failed to mention that the SD and nand drivers that were being using in the JuiceBox were not distributed with this source. after emailing him asking him why this code was missing, his response was: "The nand flash (matrix), and sd driver were not included with the original uclinux distribution. They are proprietary source code developed entirely by EmSoft, written from scratch, and therefore do not bind to the GPL license. Please let me know if you wish to use them, as I may provide the driver binaries to you as kernel loadable modules." this did not sound correct. i pulled a copy of the romfs from the onboard rom via jtag, and mounted it as a loop device. i found no kernel modules. i also got a serial interface built and verified boot log per the howto here - http://www.linux-hacker.net/cgi-bin/UltraBoard/UltraBoard.pl?Action=ShowPost&Board=MJB&Post=12&Idle=0&Sort=0&Order=Descend&Page=0&Session= i then sent Martin the following message: "I have double checked with several authorities regarding your assertion that you sd/mmc and nand code is not bound to the GPL. Clearly under section II of the GPL license states that your code does fall under the GPL if you are linking the drivers into a monolithic kernel. I've verified via jtag on 7 units that no kernel modules are present in the firmware loaded on the juicebox. as such, the functionality of the sd/mmc and nand driver can not be removed from the working kernel distributed in the juicebox firmware.For this reason the kernel drivers are not offered as a separate work as described in Section II of the GPL and must therefore be distributed under the terms and conditions of the GPL. I hope that this request from one of your customers is met by prompt action. I realise that this request may not be something you had anticipated. I hope you have seen that it is fully explained in the licensing of the GPL sources that you have used to build your product. I will take no further action on this until March 15th,2005 at which point I will begin to actively involve the Free Software Foundation and others by informing them of the manner in which this product is in violation of the GPL license." To which Martin responded again, that their code was not covered under GPL. I've carried this as far as i can since i am not a primary copyright holder to any of the linux/uclinux kernel code. if anyone has any recommendations or could assist in resolving this impass, it would be greatly appreciated. thanks David Anders From mkr at kron-systems.de Tue Mar 8 10:12:07 2005 From: mkr at kron-systems.de (Manuel Kron) Date: Tue, 8 Mar 2005 16:12:07 +0100 Subject: [uClinux-dev] Can't allocate major number 31 for MTD Message-ID: <011e01c523f1$34104860$3401a8c0@linus> Hi, I try to enable MTD on my kernel. After compile and boot I found following error on bootlog: Can't allocate major number 31 for Memory Technology Devices. Here are my kernel options: Memory Technology Devices (MTD) ---> ------------------------------------------ <*> Memory Technology Device (MTD) support [ ] Debugging <*> MTD partitioning suppor < > MTD concatenating support < > RedBoot partition table parsing < > Command line partition table parsing --- User Modules And Translation Layer <*> Direct char device access to MTD devices <*> Caching block device access to MTD device < > FTL (Flash Translation Layer) support < > NFTL (NAND Flash Translation Layer) suppor < > INFTL (Inverse NAND Flash Translation Layer) support RAM/ROM/Flash chip drivers ---> Mapping drivers for chip access ---> Self-contained MTD device drivers --- NAND Flash Device Drivers ---> RAM/ROM/Flash chip drivers ---> <*> Detect flash chips by Common Flash Interface (CFI) probe < > Detect JEDEC JESD21c compatible flash chips [ ] Flash chip driver advanced configuration options < > Support for Intel/Sharp flash chips <*> Support for AMD/Fujitsu flash chips < > Support for ST (Advanced Architecture) flash chips < > Support for RAM chips in bus mapping < > Support for ROM chips in bus mapping < > Support for absent chips in bus mappin [ ] Older (theoretically obsoleted now) drivers for non-CFI chips Mapping drivers for chip access ---> <*> CFI Flash device mapped on COBRA5272 Block devices ---> ... <*> RAM disk support (4096) Default RAM disk size [ ] Initial RAM disk (initrd) support <*> ROM disk memory block device (blkmem) (AMD) FLASH type (2MB) FLASH size (16bit) FLASH bit width ------------------------------ Any idea? Manuel ------- I can't find mtdblock in devices: /> cat /proc/devices Character devices: 1 mem 2 pty 3 ttyp 4 ttyS 5 cua 10 misc 90 mtd 162 raw Block devices: 1 ramdisk 31 Blkmem /> -------------- next part -------------- An HTML attachment was scrubbed... URL: From H.Haeberle at pilz.de Tue Mar 8 10:39:11 2005 From: H.Haeberle at pilz.de (Heinz Haeberle) Date: Tue, 8 Mar 2005 16:39:11 +0100 Subject: [uClinux-dev] init/Kconfig SHM error In-Reply-To: <20050307170007.E2E0C19F7A@uclinux.org> Message-ID: I have build the 2.6.10-uc0 kernel for mcf5249C3 While doing so I figured the config option SHMEM was set and there wasn't even a way to reset it I guess the error is in init/Kconfig I have added the following lines to config SHMEM .... + depends on MMU .... and config TINY_SHMEM default !SHMEM + depends on MMU The result will be that there is no SHMEM at all for this target and IMHO this is ok From mkr at kron-systems.de Tue Mar 8 11:08:33 2005 From: mkr at kron-systems.de (Manuel Kron) Date: Tue, 8 Mar 2005 17:08:33 +0100 Subject: [uClinux-dev] MTD mount -t ext2 Message-ID: <017c01c523f9$13d3be80$3401a8c0@linus> Hello, How can I create a simple ext2 Filesystem-Image and mount it on a mtd device? Regards, Manuel -------------- next part -------------- An HTML attachment was scrubbed... URL: From paulr at rcom-software.com Tue Mar 8 11:52:51 2005 From: paulr at rcom-software.com (Paul Romero) Date: Tue, 08 Mar 2005 08:52:51 -0800 Subject: [uClinux-dev] M5259C3 UC Linux Source Code Message-ID: <422DD863.BC3B6619@rcom-software.com> Dear User Group: Where can one find a version of the UC Linux source code that satisfies the following criteria ? (i.e. Version numbers would be helpful.) 1) Ported to the version of the M5249C3 board with 2 MB of Flash memory and no MMU. (i.e. The 2MB of SDRAM work.) 2) All of UC Linux--including the Kernel--runs in the 8 MB of SDRAM. 3) Compatible with Colilo version 0.1.3. (i.e. The tar ball is colilo-20040221.tar.gz.) Any comments on compatibility issues would be appreciated. -- Paul Romero RCOM Communications Software Phone/Fax: (510)339-2628 E-Mail: paulr at rcom-software.com From paulr at rcom-software.com Tue Mar 8 11:56:45 2005 From: paulr at rcom-software.com (Paul Romero) Date: Tue, 08 Mar 2005 08:56:45 -0800 Subject: [uClinux-dev] M5249C3 UC Linux Source Code - NOT M5259C3 ! Message-ID: <422DD94D.383AD37B@rcom-software.com> Dear User Group: Where can one find a version of the UC Linux source code that satisfies the following criteria ? (i.e. Version numbers would be helpful.) 1) Ported to the version of the M5249C3 board with 2 MB of Flash memory and no MMU. (i.e. The 2MB of SDRAM work.) 2) All of UC Linux--including the Kernel--runs in the 8 MB of SDRAM. 3) Compatible with Colilo version 0.1.3. (i.e. The tar ball is colilo-20040221.tar.gz.) Any comments on compatibility issues would be appreciated. -- Paul Romero RCOM Communications Software Phone/Fax: (510)339-2628 E-Mail: paulr at rcom-software.com From yachou at essi.fr Tue Mar 8 12:11:00 2005 From: yachou at essi.fr (yachou at essi.fr) Date: Tue, 8 Mar 2005 18:11:00 +0100 (CET) Subject: [uClinux-dev] (no subject) Message-ID: <58659.193.251.187.216.1110301860.squirrel@webmail.essi.fr> Hi, I have a project which relates to the study of management memory in reduced Linux, that's mean without MMU in an embedded system: processor ARM9 16Mo of RAM and various peripherals (USB...). That made 2mounths that I work on this known as project and the problem which I often did meet is that the applications carried out by the ARM9 make memory allocation and all must hold in 16Mo, therefore it is necessary to optimize the memory allocation (malloc) in order to minimize fragmentation . can you help me while sending to me the source code of dealing with this problem. thank you. Cheers. --------------------------------------------------------------- Samih YACHOU Master STIC Syst?mes Embarqu?s From yachou at essi.fr Tue Mar 8 12:12:14 2005 From: yachou at essi.fr (yachou at essi.fr) Date: Tue, 8 Mar 2005 18:12:14 +0100 (CET) Subject: [uClinux-dev] fragmenation problem in Uclinux Message-ID: <58696.193.251.187.216.1110301934.squirrel@webmail.essi.fr> Hi, I have a project which relates to the study of management memory in reduced Linux, that's mean without MMU in an embedded system: processor ARM9 16Mo of RAM and various peripherals (USB...). That made 2mounths that I work on this known as project and the problem which I often did meet is that the applications carried out by the ARM9 make memory allocation and all must hold in 16Mo, therefore it is necessary to optimize the memory allocation (malloc) in order to minimize fragmentation . can you help me while sending to me the source code of dealing with this problem. thank you. Cheers. ----------------------------------------------------------------- Samih YACHOU Master STIC Syst?mes Embarqu?s From ben at thingmagic.com Tue Mar 8 14:11:20 2005 From: ben at thingmagic.com (Ben Giddings) Date: Tue, 08 Mar 2005 14:11:20 -0500 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like) system In-Reply-To: <422D505C.8010208@snapgear.com> References: <4224A62A.3040909@thingmagic.com> <4225163D.4010601@snapgear.com> <42264B00.3060904@thingmagic.com> <4226649D.9050200@snapgear.com> <422CF957.9080204@thingmagic.com> <422D505C.8010208@snapgear.com> Message-ID: <422DF8D8.1020506@thingmagic.com> Greg Ungerer wrote: >> Ok, I read another message that says that it isn't actually required, >> that you can just ignore the messages saying it is. I'll try again >> with it tomorrow. > > > Yes, try that. You will get a bunch of warnings about some DSR files > not being present - but ignore those. Ok, the problem with that is that the ixp400AccessLibrary-1_4.zip file I has doesn't seem to be what the modules-20041130.patch patch expects. The Patch wants to modify modules/ixp425/ixp400/ixp400_xscale_sw/Makefile as follows: # # Top level xscale_sw Makefile # +# File Version: $Revision: 1.291 $ +# # -- Intel Copyright Notice -- # # Copyright 2002-2003 Intel Corporation All Rights Reserved. @@ -45,26 +47,25 @@ else ifeq ($(WIND_HOST_TYPE), sun4-solaris2) IX_HOST_OS := solaris + else + # Otherwise, try OSTYPE + IX_HOST_OS := $(OSTYPE) endif endif But here's the Makefile as it came out of ixp400AccessLibrary-1_4.zip: # # Top level xscale_sw Makefile # # File Version: $Revision: 1.291 $ # # -- Intel Copyright Notice -- # --- # In a Tornado environment, WIND_HOST_TYPE should be set ifeq ($(WIND_HOST_TYPE), x86-win32) IX_HOST_OS := windows else ifeq ($(WIND_HOST_TYPE), sun4-solaris2) IX_HOST_OS := solaris else # Otherwise, try OSTYPE IX_HOST_OS := $(OSTYPE) endif endif In other words, the Makefile already has the changes the patch is trying to apply. That's the case for a whole lot of other files here too. It prompted me to see if I wanted to reverse the patch, but of course I chose not to. I finished the patching, ignoring all the DSP patches, and not reversing all the other ones. I did a 'make xconfig', and am currently trying to build a default-configured IXDP425 kernel/ramdisk. Wish me luck! Ben From kapheine at gmail.com Tue Mar 8 14:32:40 2005 From: kapheine at gmail.com (Zachary Landau) Date: Tue, 8 Mar 2005 14:32:40 -0500 Subject: [uClinux-dev] Can't allocate major number 31 for MTD In-Reply-To: <011e01c523f1$34104860$3401a8c0@linus> References: <011e01c523f1$34104860$3401a8c0@linus> Message-ID: > I try to enable MTD on my kernel. After compile and boot I found following > error on bootlog: > > Can't allocate major number 31 for Memory Technology Devices. > > /> cat /proc/devices > Character devices: > 1 mem > 2 pty > 3 ttyp > 4 ttyS > 5 cua > 10 misc > 90 mtd > 162 raw > > Block devices: > 1 ramdisk > 31 Blkmem Blkmem uses the same device number as mtdblock. Disable blkmem or change the device number of one of them (and on the files in /dev). -- Zachary P. Landau From ben at thingmagic.com Tue Mar 8 16:45:43 2005 From: ben at thingmagic.com (Ben Giddings) Date: Tue, 08 Mar 2005 16:45:43 -0500 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like) system In-Reply-To: <422DF8D8.1020506@thingmagic.com> References: <4224A62A.3040909@thingmagic.com> <4225163D.4010601@snapgear.com> <42264B00.3060904@thingmagic.com> <4226649D.9050200@snapgear.com> <422CF957.9080204@thingmagic.com> <422D505C.8010208@snapgear.com> <422DF8D8.1020506@thingmagic.com> Message-ID: <422E1D07.3020607@thingmagic.com> I'm making it a bit farther now. I managed to get the linux and user parts of the system to start compiling, but when it hit the Intel stuff, it had problems. Apparently some of the changes to the Makefile were already there, but others weren't, so I had to change the LINUX_CROSS_COMPILE setting, as well as the LDFLAGS: ixp400/ixp400_xscale_sw/Makefile # BEN: No hardhats here! #LINUX_CROSS_COMPILE := $(HARDHAT_BASE)/devkit/arm/xscale_be/bin/xscale_be- LINUX_CROSS_COMPILE := arm-linux- and # LDFLAGS := -r LDFLAGS := -r -EB That got me a bit further. I think later, I'll have to go out and reconstruct what that Makefile should look like, but at the moment it's working better. The problem is now it's having a problem copying files: cp /snapgear-3.2/snapgear/modules/ixp425/ixp400/ixp400_xscale_sw/src/dsr/dsr.o ixp400_dsr.o cp: cannot stat `/snapgear-3.2/snapgear/modules/ixp425/ixp400/ixp400_xscale_sw/src/dsr/dsr.o': No such file or directory I think that's because I didn't include the Intel DSP files. I think I'll have to change the codelets(?) that it tries to build so it avoids those. Ben From ben at thingmagic.com Tue Mar 8 17:44:10 2005 From: ben at thingmagic.com (Ben Giddings) Date: Tue, 08 Mar 2005 17:44:10 -0500 Subject: [uClinux-dev] Possible GPL violation in JuiceBox distributed by emsoft In-Reply-To: <422DBD0B.1000100@amltd.com> References: <422DBD0B.1000100@amltd.com> Message-ID: <422E2ABA.6040003@thingmagic.com> David Anders wrote: > I've carried this as far as i can since i am not a primary copyright > holder to any of the linux/uclinux kernel code. if anyone has any > recommendations or could assist in resolving this impass, it would be > greatly appreciated. I assume you've read the FSF's GPL violations page: http://www.gnu.org/licenses/gpl-violation.html The tough thing will be tracking down who owns the copyrights to the parts of the system involved. I think if you're talking NAND devices, that's MTD related so maybe David Woodhouse should be contacted? http://www.linux-mtd.infradead.org/ from the quote, it sounds like they may have misconceptions about the GPL: "They are proprietary source code developed entirely by EmSoft, written from scratch, and therefore do not bind to the GPL license." I noticed he didn't mention anything about linking, a very significant part of the GPL deals with linking to GPLed code. Linus has essentially said that as far as he's concerned, kernel modules can be proprietary, but parts of a monolithic kernel must be licensed under the GPL. Please tread lightly when following this up. It seems as though you've been fair so far. It may just be a misunderstanding, and hopefully it can be cleared up without any harsh words or threats. Ben From ben at thingmagic.com Tue Mar 8 18:34:15 2005 From: ben at thingmagic.com (Ben Giddings) Date: Tue, 08 Mar 2005 18:34:15 -0500 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like) system In-Reply-To: <422E1D07.3020607@thingmagic.com> References: <4224A62A.3040909@thingmagic.com> <4225163D.4010601@snapgear.com> <42264B00.3060904@thingmagic.com> <4226649D.9050200@snapgear.com> <422CF957.9080204@thingmagic.com> <422D505C.8010208@snapgear.com> <422DF8D8.1020506@thingmagic.com> <422E1D07.3020607@thingmagic.com> Message-ID: <422E3677.7010703@thingmagic.com> Ben Giddings wrote: > cp > /snapgear-3.2/snapgear/modules/ixp425/ixp400/ixp400_xscale_sw/src/dsr/dsr.o > ixp400_dsr.o > cp: cannot stat > `/snapgear-3.2/snapgear/modules/ixp425/ixp400/ixp400_xscale_sw/src/dsr/dsr.o': > No such file or directory Ok, I fixed this by commenting out "dsr" stuff from ixp400/Makefile: in ixp400/Makefile: comment out IXP400_MODULES_INSTALL := $(IXP400_MODULES) ixp400_dsr.o #IXP400_MODULES_INSTALL := $(IXP400_MODULES) ixp400_dsr.o And ixp400_dsr.o: cp $(IX_XSCALE_SW)/src/dsr/dsr.o ixp400_dsr.o #ixp400_dsr.o: # cp $(IX_XSCALE_SW)/src/dsr/dsr.o ixp400_dsr.o Now it builds a Ramdisk and kernel cleanly. Can anybody send me (or send me links to) the patched versions of: modules/ixp425/ixp400/ixp400_xscale_sw/Makefile linux-2.4.x/include/linux/miscdevice.h Patch says that the ones I get straight out of the ixp400AccessLibrary-1_4.zip are already patched, but if I look closer it looks like some changes are there, others aren't. I could go through the patch file, line by line, and see what needs to be fixed, but it would be easier to have the correct versions. I'm also going to have to look at these ones to see if they'pre properly patched, but since they're Intel's software, I think I'll have to do them by hand. modules/ixp425/ixp400/ixp400_xscale_sw/src/include/IxOsServicesComponents.h modules/ixp425/ixp400/ixp400_xscale_sw/src/include/IxQMgr.h modules/ixp425/ixp400/ixp400_xscale_sw/src/qmgr/IxQMgrDispatcher.c modules/ixp425/ixp400/ixp400_xscale_sw/src/qmgr/IxQMgrDispatcher_p.h modules/ixp425/ixp400/ixp400_xscale_sw/src/qmgr/IxQMgrSymbols.c Ben From gerg at snapgear.com Tue Mar 8 19:28:43 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Wed, 09 Mar 2005 10:28:43 +1000 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like) system In-Reply-To: <422E1D07.3020607@thingmagic.com> References: <4224A62A.3040909@thingmagic.com> <4225163D.4010601@snapgear.com> <42264B00.3060904@thingmagic.com> <4226649D.9050200@snapgear.com> <422CF957.9080204@thingmagic.com> <422D505C.8010208@snapgear.com> <422DF8D8.1020506@thingmagic.com> <422E1D07.3020607@thingmagic.com> Message-ID: <422E433B.9090104@snapgear.com> Hi Ben, Ben Giddings wrote: > I'm making it a bit farther now. I managed to get the linux and user > parts of the system to start compiling, but when it hit the Intel stuff, > it had problems. Apparently some of the changes to the Makefile were > already there, but others weren't, so I had to change the > LINUX_CROSS_COMPILE setting, as well as the LDFLAGS: > > ixp400/ixp400_xscale_sw/Makefile > > # BEN: No hardhats here! > #LINUX_CROSS_COMPILE := $(HARDHAT_BASE)/devkit/arm/xscale_be/bin/xscale_be- > LINUX_CROSS_COMPILE := arm-linux- > > and > > # LDFLAGS := -r > LDFLAGS := -r -EB > > That got me a bit further. I think later, I'll have to go out and > reconstruct what that Makefile should look like, but at the moment it's > working better. The problem is now it's having a problem copying files: I would be nice if when you get it all too work we generate a patch specifically for use without the DSP library... > cp > /snapgear-3.2/snapgear/modules/ixp425/ixp400/ixp400_xscale_sw/src/dsr/dsr.o > ixp400_dsr.o > cp: cannot stat > `/snapgear-3.2/snapgear/modules/ixp425/ixp400/ixp400_xscale_sw/src/dsr/dsr.o': > No such file or directory > > I think that's because I didn't include the Intel DSP files. I think > I'll have to change the codelets(?) that it tries to build so it avoids > those. Yep, that is part of the DSP library code. You won't need it. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From lucy.wang at philips.com Tue Mar 8 20:05:24 2005 From: lucy.wang at philips.com (lucy.wang at philips.com) Date: Wed, 9 Mar 2005 09:05:24 +0800 Subject: [uClinux-dev] [PATCH]uClinux patch for Philips LPC22xx Message-ID: Hi, Lars, Honestly speaking, I can't understand what you mean. I didn't make any change to linux-2.4.x in the Philips LPC22xx patches. Where do you read such non-ascii characters? Where do you see about the IIRC ALIGN(16)? Best regards, Lucy "Friedrich, Lars" 2005-03-08 15:53 To: Lucy Wang/PEK/SC/PHILIPS at PHILIPS cc: Subject: RE: [uClinux-dev] [PATCH]uClinux patch for Philips LPC22xx Classification: > Attached is the uClinux patch for Philips LPC22xx, including the > vendors/Philips/LPC22xx directory and a new config.in file. Is it just me or are there plenty of non-ascii characters in the comments? ?????????????? is not really understandable for most. uClinux-2.4.x/arch/sh/vmlinux.lds.rom.S: - + (_mem_start - _rom_store) + _ctr; + + (_mem_start - _rom_store); + . = ALIGN(16); IIRC ALIGN(16) will align to the _next_ boundary if the address is already aligned. Although I can understand your intention to get rid of the _ctr symbol, I'm not really convinced the solution is better. Or are there other reasons for the change? Care to elaborate your motivation? Best regards, Lars Friedrich -------------- next part -------------- An HTML attachment was scrubbed... URL: From liuyoulin at mindray.com.cn Tue Mar 8 20:19:04 2005 From: liuyoulin at mindray.com.cn (Leo Liu) Date: Wed, 9 Mar 2005 09:19:04 +0800 Subject: [uClinux-dev] Any scripts or hints for building a toolchain with uClibc for Motorola MCF547x? Message-ID: Hi,all! Did anyone have a toolchain with uClibc for Motorola mcf5475 board? I started with the buildroot in uClibc website but got many errors during compilation So I thought someone here maybe had done that ok and could gave some scripts or hints to me Thanks! Best regards! Leo Liu From rtseng at emsoftltd.com Wed Mar 9 01:32:24 2005 From: rtseng at emsoftltd.com (Richard Tseng) Date: Wed, 9 Mar 2005 14:32:24 +0800 Subject: [uClinux-dev] Possible GPL violation in JuiceBox distributed byemsoft References: <422DBD0B.1000100@amltd.com> Message-ID: <2efa01c52471$c37a8c60$4d01a8c0@NJB> Please see below, Richard Tseng Emsoft Ltd ----- Original Message ----- From: "David Anders" To: Sent: Tuesday, March 08, 2005 10:56 PM Subject: [uClinux-dev] Possible GPL violation in JuiceBox distributed byemsoft > Fellow developers: > > after reading this post > http://www.ucdot.org/articles/05/01/24/1244222.shtml , As mentioned in this post, we understood GPL requirements, and we were preparing the source distribution when ucdot.org post the Juicebox story. We made the source available shortly thereafter. > i purchased a > juicebox to see if i could make some customizations to the unit. i was > unable to find posted source code, and no offer anywhere to be found for > gettting source. i finally email the sales department at Emsoft > regarding getting source code. shortly there after Martin Siu post this > message on the uclinux-dev mailing list - > http://mailman.uclinux.org/pipermail/uclinux-dev/2005-February/031206.html The posting of uclinux source was planned in discussions with uclinux maintainers, and not as a result of your email to sales department of Emsoft. > , but he failed to mention that the SD and nand drivers that were being > using in the JuiceBox were not distributed with this source. after > emailing him asking him why this code was missing, his response was: > > "The nand flash (matrix), and sd driver were not included with the > original uclinux distribution. They are proprietary source code > developed entirely by EmSoft, written from scratch, and therefore do > not bind to the GPL license. Please let me know if you wish to > use them, as I may provide the driver binaries to you as kernel > loadable modules." this did not sound correct. There is no Nand flash in Juicebox. The video player which uses Matrix flash cartridges was not developed by Emsoft. As for our staff's response to you, it was erroneous. The SD drivers were developed from scratch based on proprietary information from SDA. It was written as loadable modules and not a monolithic part of uclinux. > i pulled a copy of the romfs from the > onboard rom via jtag, and mounted it as a loop device. i found no kernel > modules. i also got a serial interface built and verified boot log per > the howto here - > http://www.linux-hacker.net/cgi-bin/UltraBoard/UltraBoard.pl?Action=ShowPost &Board=MJB&Post=12&Idle=0&Sort=0&Order=Descend&Page=0&Session= > > i then sent Martin the following message: > > "I have double checked with several authorities > regarding your assertion that you sd/mmc and nand code > is not bound to the GPL. Clearly under section II of > the GPL license states that your code does fall under > the GPL if you are linking the drivers into a > monolithic kernel. I've verified via jtag on 7 units > that no kernel modules are present in the firmware > loaded on the juicebox. as such, the functionality of > the sd/mmc and nand driver can not be removed from the > working kernel distributed in the juicebox > firmware.For this reason the kernel drivers are not > offered as a separate work as described in Section II > of the GPL and must therefore be distributed under the > terms and conditions of the GPL. I hope that this > request from one of your customers is met by prompt > action. I realise that this request may not be > something you had anticipated. I hope you have seen > that it is fully explained in the licensing of the GPL > sources that you have used to build your product. I > will take no further action on this until March > 15th,2005 at which point I will begin to actively > involve the Free Software Foundation and others by > informing them of the manner in which this product is > in violation of the GPL license." > > To which Martin responded again, that their code was not covered under GPL. Once again, the SD drivers were written as separate loadable modules. We do not plan to distribute their source code, which is in line with Linus' opinion that kernel modules can be proprietary. > > > I've carried this as far as i can since i am not a primary copyright > holder to any of the linux/uclinux kernel code. if anyone has any > recommendations or could assist in resolving this impass, it would be > greatly appreciated. Emsoft is the copyright holder of the SD driver code for Juicebox. Emsoft is a developer of original solutions running on Linux. We have made public our enhancements to GPL software in the past - see http://www.emsoftltd.com/products/listing/nanogtk_content.htm . We have published source code for uclinux in Juicebox. We will continue to uphold GPL principles in the future. > > thanks > David Anders > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org From s892229 at mail.yzu.edu.tw Wed Mar 9 02:09:57 2005 From: s892229 at mail.yzu.edu.tw (=?big5?B?s6+rs7ZX?=) Date: Wed, 9 Mar 2005 15:09:57 +0800 Subject: [uClinux-dev] [KVM] porting KVM on uClinux 2.4 Message-ID: <002f01c52477$00d4ee70$2d58718c@allen> Hello everyone: I try to port KVM on uClinux. I use uClinux-dist(not the newest version) and it can choose 2.0 or 2.4 kernel. during compiling, I also meet the locale support problem. I follow Erik's suggestion to replace the uClibc version from 0.9.11 to 0.9.26 even 0.9.27. In uClibc choosing list,I open the uClibc locale support and others,such as pthread support,etc.However,when I make all of system, It shows that preverify.elf2flt: In function `__uClibc_init': preverify.elf2flt(.text+0x12354): undefined reference to `__pthread_initialize_minimal' collect2: ld returned 1 exit status make[3]: *** [preverify] Error 1 make[3]: Leaving directory `/root/uClinux-dist/user/kvm/tools/preverifier/build/linux' make[2]: *** [all] Error 1 make[2]: Leaving directory `/root/uClinux-dist/user/kvm' make[1]: *** [all] Error 2 make[1]: Leaving directory `/root/uClinux-dist/user' make: *** [subdirs] Error 1 Actually,when it compiles the boa and busybox,there are some problem to appear.but I close it to compile and it will appear the above message.I try to slove the problem.I choose the General position independent code(PIC) of General libary seeting in uClibc. [*] Generate Position Independent Code (PIC) [*] Enable support for shared libraries (NEW) [*] Compile native shared library loader (NEW) [ ] Only load shared libraries which can share their text segment (NEW) [*] Native shared library loader 'ldd' support (NEW) [*] Enable shared library loader cache (NEW) [ ] Enable shared library loader preload file support (NEW) OK, when I compile again, I got anther wrong meg. /usr/local/bin/arm-elf-ld.real: cannot open /usr/local/lib/gcc-lib/arm-elf/2.95.3/fpic/msingle-pic-base/crtbeginS.o: No such file or directory I try to copy the file and anther crtendS.o to the above path. I got many errors... please help me Thanks and regards! Allen -------------- next part -------------- An HTML attachment was scrubbed... URL: From lars.friedrich at wago.com Wed Mar 9 02:16:59 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Wed, 9 Mar 2005 08:16:59 +0100 Subject: [uClinux-dev] MTD mount -t ext2 Message-ID: > How can I create a simple ext2 Filesystem-Image This is not really uClinux related. man dd man mkfs man mke2fs google creating filesystem image linux > and mount it on a mtd device? I don't think you mean 'mount' here, I assume you mean 'write'. You don't want to do this. It's ugly but acceptable when doing this for read-only partitions, but evil[TM] for read-write partitions. Anyway, the answer would be: cp imagefile /dev/mtdwhatever Best regards, Lars Friedrich From lars.friedrich at wago.com Wed Mar 9 02:37:20 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Wed, 9 Mar 2005 08:37:20 +0100 Subject: [uClinux-dev] Possible GPL violation in JuiceBox distributedby emsoft Message-ID: > Linus has essentially said that as far as he's concerned, > kernel modules can be proprietary, Actually he said that a kernel module is not guaranteed to be GPL free but that it neither falls automatically under the GPL, the decision has to be made as the case arises. Best regards, Lars Friedrich From lars.friedrich at wago.com Wed Mar 9 03:00:03 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Wed, 9 Mar 2005 09:00:03 +0100 Subject: [uClinux-dev] [PATCH]uClinux patch for Philips LPC22xx Message-ID: > Honestly speaking, I can't understand what you mean. > I didn't make any change to linux-2.4.x in the Philips LPC22xx patches. > Where do you see about the IIRC ALIGN (16)? Urks.. messed this up with a different patch I had been reviewing at the same time. You are officially declared innocent ;-) > Where do you read such non-ascii characters? uClinux-Philips-LPC22xx.tar i.e: ---- startup.S ;?????????????????????? IMPORT main ;C?????????????? IMPORT TargetResetInit ;???????????????? CODE32 AREA vectors,CODE,READONLY ENTRY ;?????????? Reset [...] ;???????????????????????????????????????? LDR R0, =PINSEL0 LDR R1, =0x80000005 ;ext2 for ethernet, uart0 STR R1, [R0] [...] ---- target.c #include "config.h" /********************************************************************************************************* ** ????????: TargetResetInit ** ????????: ????main?????????????????????????????????????????????? ********************************************************************************************************/ void TargetResetInit(void) { MEMMAP = 0x3; //remap to external memory /* ?????????????????? */ PLLCON = 1; [...] ---- config.h #include "lpc2294.h" /* ????????, Fosc??Fcclk??Fcco??Fpclk????????*/ #define Fosc 10000000 //????????,10MHz #define Fcclk (Fosc * 4) //??????????40MHz #define Fcco (Fcclk * 4) //CCO????????????Fcclk??2??4??8??16??????????156MHz~320MHz #define Fpclk (Fcclk / 4) * 1 //VPB????????????(Fcclk / 4) The linux patch file looks okay, just the uClinux one seems a bit messed up in the comments. Best regards, Lars Friedrich From lucy.wang at philips.com Wed Mar 9 03:03:36 2005 From: lucy.wang at philips.com (lucy.wang at philips.com) Date: Wed, 9 Mar 2005 16:03:36 +0800 Subject: [uClinux-dev] [PATCH]uClinux patch for Philips LPC22xx Message-ID: Thanks. That's some Chinese comments. I'll modify it. Thank you. Actually that's not in the uClinux-dist src code but a simple bootloader reference code for customers. Best regards, Lucy "Friedrich, Lars" 2005-03-09 16:00 To: Lucy Wang/PEK/SC/PHILIPS at PHILIPS cc: Subject: RE: [uClinux-dev] [PATCH]uClinux patch for Philips LPC22xx Classification: > Honestly speaking, I can't understand what you mean. > I didn't make any change to linux-2.4.x in the Philips LPC22xx patches. > Where do you see about the IIRC ALIGN (16)? Urks.. messed this up with a different patch I had been reviewing at the same time. You are officially declared innocent ;-) > Where do you read such non-ascii characters? uClinux-Philips-LPC22xx.tar i.e: ---- startup.S ;?????????????????????? IMPORT main ;C?????????????? IMPORT TargetResetInit ;???????????????? CODE32 AREA vectors,CODE,READONLY ENTRY ;?????????? Reset [...] ;???????????????????????????????????????? LDR R0, =PINSEL0 LDR R1, =0x80000005 ;ext2 for ethernet, uart0 STR R1, [R0] [...] ---- target.c #include "config.h" /********************************************************************************************************* ** ????????: TargetResetInit ** ????????: ????main?????????????????????????????????????????????? ********************************************************************************************************/ void TargetResetInit(void) { MEMMAP = 0x3; //remap to external memory /* ?????????????????? */ PLLCON = 1; [...] ---- config.h #include "lpc2294.h" /* ????????, Fosc??Fcclk??Fcco??Fpclk????????*/ #define Fosc 10000000 //????????,10MHz #define Fcclk (Fosc * 4) //??????????40MHz #define Fcco (Fcclk * 4) //CCO????????????Fcclk??2??4??8??16??????????156MHz~320MHz #define Fpclk (Fcclk / 4) * 1 //VPB????????????(Fcclk / 4) The linux patch file looks okay, just the uClinux one seems a bit messed up in the comments. Best regards, Lars Friedrich -------------- next part -------------- An HTML attachment was scrubbed... URL: From lars.friedrich at wago.com Wed Mar 9 03:33:50 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Wed, 9 Mar 2005 09:33:50 +0100 Subject: [uClinux-dev] Possible GPL violation in JuiceBox distributedbyemsoft Message-ID: > Once again, the SD drivers were written as separate loadable > modules. We do not plan to distribute their source code, > which is in line with Linus' opinion that kernel modules can > be proprietary. The use of the word 'can' here is quite weasel like. They can be. They can be not, too. There is no such opinion from Linus that it's just up to the writer of the kernel module if it falls under the GPL or not - unless you are able to quote it. As you have stated the driver was written from scratch specially for the linux kernel, it rather matches Linus' opinion that it _is_ derived work from the kernel and falls under the GPL. [kernel modules] "Basically: - anything that was written with Linux in mind (whether it then _also_ works on other operating systems or not) is clearly partially a derived work." -- Linus Torvald, LKML, Wed, 3 Dec 2003 Subject: Linux GPL and binary module exception clause? "Does that mean that any kernel module is automatically not a derived work? HELL NO! It has nothing to do with modules per se[...]" -- Linus Torvald, LKML, Wed, 3 Dec 2003 Subject: Linux GPL and binary module exception clause? I find "HELL NO" as quite clear reaction. Please elaborate your reasons why the driver is not derived work despite the "it's a kernel module" argument. Best regards, Lars Friedrich From lucy.wang at philips.com Wed Mar 9 03:21:36 2005 From: lucy.wang at philips.com (lucy.wang at philips.com) Date: Wed, 9 Mar 2005 16:21:36 +0800 Subject: [uClinux-dev] [PATCH]uClinux patch for Philips LPC22xx Message-ID: Here is the new patch. - Modify the Chinese comments to English. - Delete some unused files and modify the file structure a little. "Friedrich, Lars" 2005-03-09 16:00 To: Lucy Wang/PEK/SC/PHILIPS at PHILIPS cc: Subject: RE: [uClinux-dev] [PATCH]uClinux patch for Philips LPC22xx Classification: > Honestly speaking, I can't understand what you mean. > I didn't make any change to linux-2.4.x in the Philips LPC22xx patches. > Where do you see about the IIRC ALIGN (16)? Urks.. messed this up with a different patch I had been reviewing at the same time. You are officially declared innocent ;-) > Where do you read such non-ascii characters? uClinux-Philips-LPC22xx.tar i.e: ---- startup.S ;?????????????????????? IMPORT main ;C?????????????? IMPORT TargetResetInit ;???????????????? CODE32 AREA vectors,CODE,READONLY ENTRY ;?????????? Reset [...] ;???????????????????????????????????????? LDR R0, =PINSEL0 LDR R1, =0x80000005 ;ext2 for ethernet, uart0 STR R1, [R0] [...] ---- target.c #include "config.h" /********************************************************************************************************* ** ????????: TargetResetInit ** ????????: ????main?????????????????????????????????????????????? ********************************************************************************************************/ void TargetResetInit(void) { MEMMAP = 0x3; //remap to external memory /* ?????????????????? */ PLLCON = 1; [...] ---- config.h #include "lpc2294.h" /* ????????, Fosc??Fcclk??Fcco??Fpclk????????*/ #define Fosc 10000000 //????????,10MHz #define Fcclk (Fosc * 4) //??????????40MHz #define Fcco (Fcclk * 4) //CCO????????????Fcclk??2??4??8??16??????????156MHz~320MHz #define Fpclk (Fcclk / 4) * 1 //VPB????????????(Fcclk / 4) The linux patch file looks okay, just the uClinux one seems a bit messed up in the comments. Best regards, Lars Friedrich -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: uClinux-Philips-LPC22xx.tar.gz Type: application/octet-stream Size: 26954 bytes Desc: not available URL: From rtseng at emsoftltd.com Wed Mar 9 05:47:16 2005 From: rtseng at emsoftltd.com (Richard Tseng) Date: Wed, 9 Mar 2005 18:47:16 +0800 Subject: [uClinux-dev] Possible GPL violation in JuiceBox distributedbyemsoft References: Message-ID: <2f8e01c52495$5e506020$4d01a8c0@NJB> We have written the SD driver from scratch based on SDA specifications, then ported it to uclinux. We did refer to Linux header files such as kernel.h, init.h, module.h for this port. Are you saying all Linux loadable modules MUST fall under GPL since they must refer to some Linux header files ? Richard ----- Original Message ----- From: "Friedrich, Lars" To: "Richard Tseng" Cc: "David Anders" ; Sent: Wednesday, March 09, 2005 4:33 PM Subject: RE: [uClinux-dev] Possible GPL violation in JuiceBox distributedbyemsoft > > Once again, the SD drivers were written as separate loadable > > modules. We do not plan to distribute their source code, > > which is in line with Linus' opinion that kernel modules can > > be proprietary. > > The use of the word 'can' here is quite weasel like. They can be. > They can be not, too. There is no such opinion from Linus that > it's just up to the writer of the kernel module if it falls under > the GPL or not - unless you are able to quote it. > > As you have stated the driver was written from scratch > specially for the linux kernel, it rather matches Linus' > opinion that it _is_ derived work from the kernel and > falls under the GPL. > > [kernel modules] > "Basically: > - anything that was written with Linux in mind (whether it then _also_ > works on other operating systems or not) is clearly partially a derived > work." -- Linus Torvald, LKML, Wed, 3 Dec 2003 > Subject: Linux GPL and binary module exception clause? > > "Does that mean that any kernel module is automatically not a derived work? > HELL NO! It has nothing to do with modules per se[...]" > -- Linus Torvald, LKML, Wed, 3 Dec 2003 > Subject: Linux GPL and binary module exception clause? > > I find "HELL NO" as quite clear reaction. Please elaborate your reasons > why the driver is not derived work despite the "it's a kernel module" > argument. > > Best regards, > Lars Friedrich From lanza at ivreasistemi.it Wed Mar 9 05:51:38 2005 From: lanza at ivreasistemi.it (Enrico Lanza) Date: Wed, 9 Mar 2005 11:51:38 +0100 Subject: [uClinux-dev] Enabling SSL with Boa on Samsung S3C4510B Message-ID: Hi all, I'm trying to enable SSL with BOA on Samsung S3C4510B. I download the openssl-0.9.7b.tar.gz from www.openssl.org The building operations goes well, but when Boa starts I get the follow error: Unhandled fault: external abort on linefetch (F4) at 0x00000001 fault-common.c(97): start_code=0x400040, start_stack=0x4fffa8) I increased the stack size using the compiler option FLTFLAGS += -s 0x20000 but the error still occur. Please, somebody can help me? Thanks Enrico -------------- next part -------------- An HTML attachment was scrubbed... URL: From eauth at softsys.co.at Wed Mar 9 06:26:34 2005 From: eauth at softsys.co.at (Erwin Authried) Date: Wed, 09 Mar 2005 12:26:34 +0100 Subject: [uClinux-dev] Possible GPL violation in JuiceBox distributedbyemsoft In-Reply-To: <2f8e01c52495$5e506020$4d01a8c0@NJB> References: <2f8e01c52495$5e506020$4d01a8c0@NJB> Message-ID: <1110367594.29254.182.camel@justakiss> Am Mit, den 09.03.2005 schrieb Richard Tseng um 11:47: > We have written the SD driver from scratch based on SDA specifications, then > ported it to uclinux. We did refer to Linux header files such as kernel.h, > init.h, module.h for this port. Are you saying all Linux loadable modules > MUST fall under GPL since they must refer to some Linux header files ? > Richard Hi, there is an article on linuxdevices covering this topic: http://www.linuxdevices.com/articles/AT5041108431.html Regards, Erwin From lars.friedrich at wago.com Wed Mar 9 07:50:48 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Wed, 9 Mar 2005 13:50:48 +0100 Subject: [uClinux-dev] Possible GPL violation in JuiceBox distributedbyemsoft Message-ID: > We have written the SD driver from scratch based on SDA > specifications, for which OS? > then ported it to uclinux. We did refer to > Linux header files such as kernel.h, init.h, module.h for > this port. Are you saying all Linux loadable modules > MUST fall under GPL since they must refer to some Linux > header files ? No. I'm saying that "it's a loadable module" is not an argument why it does not fall under the GPL. Also I'm saying that Linus never said/wrote that there would be a general binary-only loadable module GPL exception. Of course I accept references to prove me wrong in this regard. This is what I'm saying. Not more, not less. Best regards, Lars Friedrich From mkr at kron-systems.de Wed Mar 9 08:02:36 2005 From: mkr at kron-systems.de (Manuel Kron) Date: Wed, 9 Mar 2005 14:02:36 +0100 Subject: [uClinux-dev] Using 128kb Flash as a file system Message-ID: <00a501c524a8$46cb9560$3401a8c0@linus> Dear all, my ambitions: I want to use 128k Flash as a file system on my Coldfire 5272 board. I have a 2MB Flash(AMD) with 11 Sectors and the least 2 sectors are free. What I done: 1. Flash procedure on host ---------------------------- # Create e2fs.img with a size of 128kByte: cd /home/mkr/flash dd if=/dev/zero of=e2fs.img bs=1024 count=128 /sbin/mke2fs -N 64 -m 50 -F e2fs.img # Test the ext2 image: mkdir -p mnt [sudo] mount -t ext2 e2fs.img mnt -o loop echo "Hello from the file system" > mnt/hello.txt [sudo] umount mnt rmdir mnt 2. Copying the flash file into a free sector ------------------------------------------ Free Sector: starts here e.g.: 0xFFFC 0000 (256k) (I use the CFFlasher from freescale in order to load the image into the flash) 3. Mounting the Flash ---------------------- /> mount -t ext2 /dev/mtdblock6 /mnt mount: Mounting /dev/mtdblock6 on /mnt failed: Invalid argument pid 29: failed 256 /> VFS: Can't find ext2 filesystem on dev mtdblock(31,6). What's my mistake? Best Regards, Manuel -------------- next part -------------- An HTML attachment was scrubbed... URL: From jmcnamara at kimble.ie Wed Mar 9 08:45:22 2005 From: jmcnamara at kimble.ie (Jim McNamara) Date: Wed, 9 Mar 2005 13:45:22 -0000 Subject: [uClinux-dev] Possible GPL violation in JuiceBoxdistributedbyemsoft In-Reply-To: Message-ID: > > We have written the SD driver from scratch based on SDA > > specifications, > > for which OS? > > > then ported it to uclinux. We did refer to > > Linux header files such as kernel.h, init.h, module.h for > > this port. Are you saying all Linux loadable modules > > MUST fall under GPL since they must refer to some Linux > > header files ? > > No. > I'm saying that "it's a loadable module" is not an argument why > it does not fall under the GPL. So, a standalone driver, which is subsequently ported to uClinux as a loadable module doesn't fall under the GPL, but a driver that is originally designed to be a loadable module of uClinux does, regardless of whether that driver works with another OS or not?? > Also I'm saying that Linus never said/wrote that there would > be a general binary-only loadable module GPL exception. Of course > I accept references to prove me wrong in this regard. > > This is what I'm saying. Not more, not less. > > Best regards, > Lars Friedrich > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org > From eauth at softsys.co.at Wed Mar 9 08:55:05 2005 From: eauth at softsys.co.at (Erwin Authried) Date: Wed, 09 Mar 2005 14:55:05 +0100 Subject: [uClinux-dev] Using 128kb Flash as a file system In-Reply-To: <00a501c524a8$46cb9560$3401a8c0@linus> References: <00a501c524a8$46cb9560$3401a8c0@linus> Message-ID: <1110376505.7108.16.camel@kabel> On Wed, 2005-03-09 at 14:02, Manuel Kron wrote: > Dear all, > my ambitions: I want to use 128k Flash as a file system on my Coldfire > 5272 > board. > I have a 2MB Flash(AMD) with 11 Sectors and the least 2 sectors are > free. > > What I done: > 1. Flash procedure on host > ---------------------------- > # Create e2fs.img with a size of 128kByte: > cd /home/mkr/flash > dd if=/dev/zero of=e2fs.img bs=1024 count=128 > /sbin/mke2fs -N 64 -m 50 -F e2fs.img > # Test the ext2 image: > mkdir -p mnt > [sudo] mount -t ext2 e2fs.img mnt -o loop > echo "Hello from the file system" > mnt/hello.txt > [sudo] umount mnt > rmdir mnt > > 2. Copying the flash file into a free sector > ------------------------------------------ > Free Sector: starts here e.g.: 0xFFFC 0000 (256k) > (I use the CFFlasher from freescale in order to load the image into > the > flash) > > 3. Mounting the Flash > ---------------------- > /> mount -t ext2 /dev/mtdblock6 /mnt > mount: Mounting /dev/mtdblock6 on /mnt failed: Invalid argument > pid 29: failed 256 > /> > VFS: Can't find ext2 filesystem on dev mtdblock(31,6). > > What's my mistake? > > Best Regards, > Manuel Hi, despite of the problem with mounting that you still have, are you aware that you can use the ext2 filesystem only for read/only access if you use it in this way? If that's what you plan to do, I'm wondering why you want to use ext2 at all. romfs or cramfs would be a better choice for that. Regards, Erwin From palmer at sedsystems.ca Wed Mar 9 09:33:25 2005 From: palmer at sedsystems.ca (Russ Palmer) Date: Wed, 09 Mar 2005 08:33:25 -0600 Subject: [uClinux-dev] Installing uClinux on ConnectME In-Reply-To: <421CA7A3.7070502@sedsystems.ca> References: <421C9FD4.5030001@sasktel.net> <421CA1E0.4030501@sedsystems.ca> <421CA7A3.7070502@sedsystems.ca> Message-ID: <422F0935.5070803@sedsystems.ca> An HTML attachment was scrubbed... URL: From cmays at wavetrix.com Wed Mar 9 14:32:40 2005 From: cmays at wavetrix.com (Chapman Mays) Date: Wed, 9 Mar 2005 13:32:40 -0600 Subject: [uClinux-dev] Installing uClinux on ConnectME In-Reply-To: <20050304053307.2EF1DE966@mail2.atmark-techno.com> Message-ID: <000401c524de$c4155840$7700a8c0@TracyDesk> Yashi, You are partially correct. In chapter 12 of the user's guide, it does talk about how to re-load hermit and then an OS image. What is missing is how you get the XK bootloader menu in the first place which I guess is really what I'm asking. Is there a better way to get to the XK menu other than what I have described? So now that I know how to get this working on some of my existing ConnectME devices, is it possible to get it working on the newer ones (the so-called "-C" modules)? from what I understand, the bootloader is now part of the downloaded image with the NetOS environment and there doesn't seem to be an equivalent of the XK menu. Thanks, Chapman -----Original Message----- From: Yasushi SHOJI [mailto:yashi at atmark-techno.com] Sent: Thursday, March 03, 2005 11:33 PM To: Chapman Mays Cc: uclinux-dev at uClinux.org Subject: Re: [uClinux-dev] Installing uClinux on ConnectME At Tue, 22 Feb 2005 17:46:01 -0600, Chapman Mays wrote: > > So, finally my questions: Has anyone described and/or documented the process > for doing this? If I take the process that I used on my development board > will that work? haven't we documented on user's manual? check chapter 12 in user's guide. -- yashi From jsujjavanich at syntech-fuelmaster.com Wed Mar 9 17:41:58 2005 From: jsujjavanich at syntech-fuelmaster.com (Jate Sujjavanich) Date: Wed, 9 Mar 2005 17:41:58 -0500 Subject: [uClinux-dev] uclinux 2.6.10 fails on linking Message-ID: I am getting the following error when compiling uClinux 2.6.10 with GCC 3.4: LD init/built-in.o LD vmlinux arch/m68knommu/platform/5235/built-in.o(.text+0x2a): In function `config_BSP': arch/m68knommu/platform/5235/config.c:75: undefined reference to `coldfire_pit_init' arch/m68knommu/platform/5235/built-in.o(.text+0x36):arch/m68knommu/platf orm/5235/config.c:76: undefined reference to `coldfire_pit_tick' arch/m68knommu/platform/5235/built-in.o(.text+0x42):arch/m68knommu/platf orm/5235/config.c:77: undefined reference to `coldfire_pit_offset' mm/built-in.o(.text+0xc4e2): In function `shmem_getpage': mm/shmem.c:921: undefined reference to `swapin_readahead' mm/built-in.o(.text+0xc95a): In function `shmem_populate': mm/shmem.c:1200: undefined reference to `install_page' mm/built-in.o(.text+0xc9dc):mm/shmem.c:1206: undefined reference to `install_file_pte' mm/built-in.o(.text+0xca44): In function `shmem_lock': mm/shmem.c:1244: undefined reference to `user_shm_lock' mm/built-in.o(.text+0xca72):mm/shmem.c:1249: undefined reference to `user_shm_unlock' make[1]: *** [vmlinux] Error 1 make[1]: Leaving directory `/home/jate/uClinux-2.6/linux-2.6.x' make: *** [linux] Error 1 I am not sure how to display the command line causing this error. - Jate S. -------------- next part -------------- An HTML attachment was scrubbed... URL: From jsujjavanich at syntech-fuelmaster.com Wed Mar 9 17:43:48 2005 From: jsujjavanich at syntech-fuelmaster.com (Jate Sujjavanich) Date: Wed, 9 Mar 2005 17:43:48 -0500 Subject: [uClinux-dev] uclinux 2.6.10 fails on linking Message-ID: I forgot to say that I am porting 2.6 over to the MCF5235. ________________________________ From: uclinux-dev-bounces at uClinux.org [mailto:uclinux-dev-bounces at uClinux.org] On Behalf Of Jate Sujjavanich Sent: Wednesday, March 09, 2005 5:42 PM To: uClinux development list Subject: [uClinux-dev] uclinux 2.6.10 fails on linking I am getting the following error when compiling uClinux 2.6.10 with GCC 3.4: LD init/built-in.o LD vmlinux arch/m68knommu/platform/5235/built-in.o(.text+0x2a): In function `config_BSP': arch/m68knommu/platform/5235/config.c:75: undefined reference to `coldfire_pit_init' arch/m68knommu/platform/5235/built-in.o(.text+0x36):arch/m68knommu/platf orm/5235/config.c:76: undefined reference to `coldfire_pit_tick' arch/m68knommu/platform/5235/built-in.o(.text+0x42):arch/m68knommu/platf orm/5235/config.c:77: undefined reference to `coldfire_pit_offset' mm/built-in.o(.text+0xc4e2): In function `shmem_getpage': mm/shmem.c:921: undefined reference to `swapin_readahead' mm/built-in.o(.text+0xc95a): In function `shmem_populate': mm/shmem.c:1200: undefined reference to `install_page' mm/built-in.o(.text+0xc9dc):mm/shmem.c:1206: undefined reference to `install_file_pte' mm/built-in.o(.text+0xca44): In function `shmem_lock': mm/shmem.c:1244: undefined reference to `user_shm_lock' mm/built-in.o(.text+0xca72):mm/shmem.c:1249: undefined reference to `user_shm_unlock' make[1]: *** [vmlinux] Error 1 make[1]: Leaving directory `/home/jate/uClinux-2.6/linux-2.6.x' make: *** [linux] Error 1 I am not sure how to display the command line causing this error. - Jate S. From iain at engtech.ca Wed Mar 9 17:50:20 2005 From: iain at engtech.ca (Iain Galloway) Date: Wed, 09 Mar 2005 18:50:20 -0400 Subject: [uClinux-dev] ada gcc-gnat and elf toolchain. In-Reply-To: References: Message-ID: <1110408620.27029.36.camel@localhost.localdomain> We build and sell a MCF5272 board. I had an inquiry asking if gcc-gnat the ada compiler could work with the elf toolchain. It looks like gcc-gnat definitely has problems with m68k-gcc-elf at rev 2.95 Has anyone tried getting gcc-gnat going with gcc3 ? http://uclinux.org/pub/uClinux/m68k-elf-tools/gcc-3/ Thanks in advance. -- Iain Galloway, P.Eng. Engineering Technologies Canada Ltd. www.engtech.ca 902-628-1705 From jwilliams at itee.uq.edu.au Wed Mar 9 20:51:18 2005 From: jwilliams at itee.uq.edu.au (John Williams) Date: Thu, 10 Mar 2005 11:51:18 +1000 Subject: [uClinux-dev] Using 128kb Flash as a file system In-Reply-To: <1110376505.7108.16.camel@kabel> References: <00a501c524a8$46cb9560$3401a8c0@linus> <1110376505.7108.16.camel@kabel> Message-ID: <422FA816.3020103@itee.uq.edu.au> Erwin Authried wrote: > On Wed, 2005-03-09 at 14:02, Manuel Kron wrote: > >>Dear all, >>my ambitions: I want to use 128k Flash as a file system on my Coldfire >>5272 >>board. > despite of the problem with mounting that you still have, are you aware > that you can use the ext2 filesystem only for read/only access if you > use it in this way? If that's what you plan to do, I'm wondering why you > want to use ext2 at all. romfs or cramfs would be a better choice for > that. Indeed, for such a small flash partition, perhaps flatfsd would be a better way to go, rather than a full-blown filesystem Manuel - there's an article at ucdot.org about using flatfsd, might be of interest to you. Regards, John From mike at steroidmicros.com Wed Mar 9 20:59:30 2005 From: mike at steroidmicros.com (mike) Date: Wed, 9 Mar 2005 17:59:30 -0800 Subject: [uClinux-dev] m68k-bdm-elf-gdb Cygwin Message-ID: <000601c52514$cd1b9e00$6202a8c0@flanders> Hello, Has anyone successfully used the m68k-bdm-elf-gdb from Cygwin with gdbserver TCP/IP on uClinux? (Network or Serial) I have tried the one from http://www.uclinux.org/pub/uClinux/m68k-elf-tools/gcc-3/ (GDB 6.1) and am getting the following message (Software breakpoints not implemented for this target) when trying to do anything (i.e. set a breakpoint, step, etc). After searching the archives I see that this has come up before (http://mailman.uclinux.org/pipermail/uclinux-dev/2004-February/024150.html) , but no resolution has been posted. The error message is the same under both a Linux Host (Red Hat 9) and Cygwin for GDB 6.1. If I use GDB 5.0 as part of the older toolchain (http://www.uclinux.org/pub/uClinux/m68k-elf-tools/) on a Linux (Red Hat 9) host then everything seems to work OK (there is no pre-built Cygwin chain for testing this version). Does anyone know why 6.1 version of gdb doesn't work with uClinux gdbserver? Does the gdbserver and gdb host need to be the same version? If so what version is the gdbserver included with the uClinux build? And what host versions are compatible. I really need to be able to use gdb 6.0 or above on the host machine and on windows. Thanks, Mike Lavender From wnlee at cad.csie.ncku.edu.tw Wed Mar 9 22:02:23 2005 From: wnlee at cad.csie.ncku.edu.tw (Wayne Lee) Date: Thu, 10 Mar 2005 11:02:23 +0800 Subject: [uClinux-dev] almost boot Linux, but fail at "kernel BUG at kernel/exit.c:934!" after doing "/bin/expand" In-Reply-To: <200503011040.j21AeSfx008702@cad.csie.ncku.edu.tw> Message-ID: <200503100305.j2A354fw023141@cad.csie.ncku.edu.tw> Hi, David and community, I found the reason and fix it. For arm926ej-s cpu, I use the following parameters in arch/armnommu/Makefile. arch-$(CONFIG_CPU_32v5) :=-D__LINUX_ARM_ARCH__=5 $(call check_gcc,-march=armv5te) This toolchain leads to this kernel bug because it compile incorrect for __xchg() which is defined in "include/asm-armnommu/system.h". Hence, the kernel determines the process status wrong. After we use C function __xchg() instead of accelerated inline assembly __xchg(), the kernel is booted. Bravo! Wayne -----Original Message----- From: uclinux-dev-bounces at uClinux.org [mailto:uclinux-dev-bounces at uClinux.org] On Behalf Of Wayne Lee Sent: Tuesday, March 01, 2005 6:38 PM To: uClinux development list Subject: [uClinux-dev] almost boot Linux,but fail at "kernel BUG at kernel/exit.c:934!" afterdoing "/bin/expand" Hi, folks, I try to port uClinux 2.6.5 on an arm926 board. (without chip's MMU capability) However, it really disappointed me that uClinux 2.6.5 failed to boot after issuing this command: "/bin/expand /etc/ramfs.img /dev/ram1" The error message is: kernel BUG at kernel/exit.c:934! It is too difficult to figure out what is the kernel bug. Could anyone kindly give me some hints? I have ever adopted the same packages to port uClinux 2.6.5 on arm940t board. In this time, the main different is that arm926 cpu is configured, and "armv5te" compile option is used. The packages I adopted: uClinux-dist-20040408.tar.gz linux-2.6.5.tar.gz linux-2.6.5-hsc2.patch.gz (hyok.choi's patch) The toolchain I adopted: arm-elf-tools-20040427.sh The boot messages: Linux version 2.6.5 (root at wayne_linux) (gcc version 2.95.3 20010315 (release)(ColdFire patches - 20010318 from http://fiddes.net/coldfire/)(uClinux XIP and shared lib patches from http://www.snapgear.com/)) #116 Tue Mar 1 17:37:10 CST 2005 CPU: ARM926EJ-Sid(wb) [41069263] revision 3 (ARMv5TEJ) CPU: D VIPT write-back cache CPU: I cache: 32768 bytes, associativity 4, 32 byte lines, 256 sets CPU: D cache: 32768 bytes, associativity 4, 32 byte lines, 256 sets Machine: QPT200 board On node 0 totalpages: 16384 DMA zone: 0 pages, LIFO batch:1 Normal zone: 16384 pages, LIFO batch:4 HighMem zone: 0 pages, LIFO batch:1 Built 1 zonelists Kernel command line: root=/dev/ram0 initrd=0x02000000,800K keepinitrd PID hash table entries: 512 (order 9: 4096 bytes) Memory: 64MB = 64MB total Memory: 63248KB available (608K code, 81K data, 44K init) Calibrating delay loop... 0.88 BogoMIPS Dentry cache hash table entries: 8192 (order: 3, 32768 bytes) Inode-cache hash table entries: 4096 (order: 2, 16384 bytes) Mount-cache hash table entries: 512 (order: 0, 4096 bytes) checking if image is initramfs...it isn't (ungzip failed); looks like an initrd POSIX conformance testing by UNIFIX Linux NoNET1.0 for Linux 2.6 ikconfig 0.7 with /proc/config* Initializing QPT200 Internal UART ttyS0 at I/O 0x80000100 (irq = 17) is a QPT200 Internal UART RAMDISK driver initialized: 16 RAM disks of 8192K size 1024 blocksize loop: loaded (max 8 devices) RAMDISK: romfs filesystem found at block 0 RAMDISK: Loading 772 blocks [1 disk] into ram disk... |/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/ -\|/-\|/-\|done. VFS: Mounted root (romfs filesystem) readonly. Freeing init memory: 44K Shell invoked to run file: /etc/rc Command: hostname QPT-200 Command: /bin/expand /etc/ramfs.img /dev/ram1 init/expand.c argv[0] is [/bin/expand] argv[1] is [/etc/ramfs.img] argv[2] is [/dev/ram1] read(fdi,&len,4); fdi=[0] len=[34422656] len = ntohl(len); lseek(fdo, 0, SEEK_SET); fdo=[3] close fdi[0] close fdo[3] free(buf) exit(0); kernel BUG at kernel/exit.c:934! _______________________________________________ uClinux-dev mailing list uClinux-dev at uClinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev at uclinux.org -- No virus found in this incoming message. Checked by AVG Anti-Virus. Version: 7.0.308 / Virus Database: 266.7.1 - Release Date: 2005/3/9 -- No virus found in this outgoing message. Checked by AVG Anti-Virus. Version: 7.0.308 / Virus Database: 266.7.1 - Release Date: 2005/3/9 From wnlee at cad.csie.ncku.edu.tw Wed Mar 9 22:14:48 2005 From: wnlee at cad.csie.ncku.edu.tw (Wayne Lee) Date: Thu, 10 Mar 2005 11:14:48 +0800 Subject: [uClinux-dev] where should we store ttb for MMU? Message-ID: <200503100317.j2A3HOfw023264@cad.csie.ncku.edu.tw> Hi, folks, In fact, I am now porting an uClinux no VM port on arm926ej-s CPU. In order to turn-on D-cache, I use flat-mapping & section-mapped MMU and then turn on D-cache. The method is referred from Samsung s5c7375 port. I store TTB in 0x4000. However, while kernel is booting, the TTB is overwritten and data abort occurred. I am wonder where should we place the ttb in uClinux 2.6? Do we need to modify the vmlinux.lds? Please kindly give me your suggestions. Wayne /* * Kernel startup entry point. */ __INIT .type stext, #function ENTRY(stext) mov r12, r0 mov r0, #PSR_F_BIT | PSR_I_BIT | MODE_SVC @ make sure svc mode msr cpsr_c, r0 @ and all irqs disabled /* * ARM920T need MMU enable for D-cache. * we just make the identical table before the MMU enabling. */ #ifdef CONFIG_CPU_MXU_ENABLE @ /* Write domain id (cp15_r3) */ @ mvn r0, #0 /* Domains 0, 1 = client */ @ mcr p15, 0, r0, c3, c0, 0 /* load domain access register */ @ /* Set control register v4 */ MOV r0, #0 MCR p15, 0, r0, c7, c7, 0 @ invalidate I,D caches MCR p15, 0, r0, c7, c10, 4 @ drain WB MCR p15, 0, r0, c8, c7, 0 @ invalidate I & D TLBs MRC p15, 0, r0, c1, c0, 0 @ ctrl register BIC r0, r0, #0x000f @ ............wcam BIC r0, r0, #0x1100 @ ...i...s........ MCR p15, 0, r0, c1, c0, 0 @ ctrl register @ mov r0, #0 @ mcr p15, 0, r0, c13, c0, 0 /* zero PID */ @ mcr p15, 0, r0, c7, c7, 0 /* invalidate I,D caches */ @ mcr p15, 0, r0, c7, c10, 4 /* drain write buffer */ @ mcr p15, 0, r0, c8, c7, 0 /* invalidate I,D TLBs */ @ mrc p15, 0, r0, c1, c0, 0 /* get control register */ @ bic r0, r0, #0x000f @ ............wcam @ bic r0, r0, #0x1100 @ ...i...s........ @ @bic r0, r0, #0x0001 /* disable MMU */ @ mcr p15, 0, r0, c1, c0, 0 /* write control register */ init_ttb: /* * We must reserve DRAM_BASE ~ DRAM_BASE+0x8000. * And Kernel TEXTADDR must be after DRAM_BASE + 0x8000 */ #if !defined(CONFIG_DRAM_BASE) || !defined(CONFIG_DRAM_SIZE) #define CONFIG_DRAM_BASE 0x00000000 #define CONFIG_DRAM_SIZE 0x04000000 #endif @ ldr r0,=(CONFIG_DRAM_BASE + 0x4000) /* set TTB (Must be on 16KB boundary) */ @ ldr r0,=(CONFIG_DRAM_BASE + CONFIG_DRAM_SIZE/2 + 0x4000) /* set TTB (Must be on 16KB boundary) */ @ ldr r0,=(CONFIG_DRAM_BASE + CONFIG_DRAM_SIZE - 0x4000) /* set TTB (Must be on 16KB boundary) */ ldr r0,=(CONFIG_DRAM_BASE + CONFIG_DRAM_SIZE) /* set TTB (Must be on 16KB boundary) */ mcr p15, 0, r0, c2, c0, 0 /* write to TTB(CP15,c2) */ /* * * Create translation table for flat mapping * Top 12 bits of VA is pointer into table * Create 4096 entries from 000xxxxx to fffxxxxx * * Set Section descriptor. 4K entry totla and * each entry describe 1MB. */ mov r1, #0x1000 @ 4K counter subs r1,r1,#1 mov r2,#0b000000000000 @ set access permissions (AP) for full access SVC/USR (11:10) @ mov r2,#0b110000000000 @ set access permissions (AP) for full access SVC/USR (11:10) orr r2,r2,#0b000111100000 @ set for domain 15 (8:5) orr r2,r2,#0b000000010000 @ must be 1 (4) orr r2,r2,#0b000000000000 @ set non cachable non bufferable (CB) (3:2) orr r2,r2,#0b000000000010 @ set for 1Mb section (1:0) init_ttb_1: /* create 4K entry */ orr r3,r2,r1,lsl#20 @ use loop counter to create individual table entries str r3,[r0,r1,lsl#2] @ str r3 at TTB base + loopcount*4 subs r1,r1,#1 @ decrement loop counter bpl init_ttb_1 /* resetting the SDRAM area to cacheable*/ /* DRAM_BASE ~ DRAM_BASE+DRAM_SIZE */ /* 64MB -> 64 entry -> 0x40 */ mov r1, #(CONFIG_DRAM_SIZE >> 20 ) @ 64 entry =ase addr subs r1, r1, #1 mov r2,#0b110000000000 @ set access permissions (AP) for full access SVC/USR (11:10) orr r2,r2,#0b000111100000 @ set for domain 15 (8:5) orr r2,r2,#0b000000010000 @ must be 1 (4) orr r2,r2,#0b000000001100 @ set cachable but unbufferable (CB) (3:2) @ orr r2,r2,#0b000000001000 @ set cachable but unbufferable (CB) (3:2) orr r2,r2,#0b000000000010 @ set for 1Mb section (1:0) init_ttb_2: /* create 64 entry (for 64MB SDRAM) */ add r4, r1, #(CONFIG_DRAM_BASE >> 20 ) @ drambase >> 20 bit . orr r3,r2,r4,lsl#20 @ use loop counter to create individual table entries str r3,[r0,r4,lsl#2] @ str r3 at TTB base + loopcount*4 subs r1,r1,#1 @ decrement loop counter bpl init_ttb_2 @init_domains : we define all domains are manager. so no access permission check is occured. mvn r0, #0 @ trick. 0 -1 == 0xFFFFFFFF mcr p15, 0, r0, c3, c0, 0 @ write to CP15 register 5 /* * set global core configurations */ mrc p15, 0, r0, c1, c0, 0 @ read CP15 register 1 into r0 orr r0, r0, #(0x1 <<12) @ enable I Cache orr r0, r0, #(0x1 <<2) @ enable D Cache orr r0, r0, #(0b11 <<30) @ enable asynchronous clocking mode orr r0, r0, #0x1 @ enable MMU mcr p15, 0, r0, c1, c0, 0 @ write cp15 register 1 /* * set global core configurations */ mrc p15, 0, r0, c1, c0, 0 @ read CP15 register 1 into r0 @#ifndef CONFIG_CPU_ICACHE_DISABLE @ orr r0, r0, #(0x1 <<12) @ enable I Cache @#endif @#ifndef CONFIG_CPU_DCACHE_DISABLE @ orr r0, r0, #(0x1 <<2) @ enable D Cache @#endif @#ifdef CONFIG_ALIGNMENT_TRAP @ orr r0, r0, #2 @ ...........A. @#endif @ orr r0, r0, #(0b11 <<30) @ enable asynchronous clocking mode @ orr r0, r0, #0x1 @ enable MMU mcr p15, 0, r0, c1, c0, 0 @ write cp15 register 1 nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop @ mrc p15, 0, r0, c1, c0, 0 @ read CP15 register 1 into r0 @#ifndef CONFIG_CPU_ICACHE_DISABLE @ orr r0, r0, #(0x1 <<12) @ enable I Cache @#endif @#ifndef CONFIG_CPU_DCACHE_DISABLE @ orr r0, r0, #(0x1 <<2) @ enable D Cache @#endif @ mcr p15, 0, r0, c1, c0, 0 @ write CP15 register 1 #endif /* * set global core configurations */ mrc p15, 0, r0, c1, c0, 0 @ read CP15 register 1 into r0 #ifndef CONFIG_CPU_ICACHE_DISABLE orr r0, r0, #(0x1 <<12) @ enable I Cache #endif mcr p15, 0, r0, c1, c0, 0 @ write cp15 register 1 -- No virus found in this outgoing message. Checked by AVG Anti-Virus. Version: 7.0.308 / Virus Database: 266.7.1 - Release Date: 2005/3/9 From wnlee at cad.csie.ncku.edu.tw Wed Mar 9 22:25:10 2005 From: wnlee at cad.csie.ncku.edu.tw (Wayne Lee) Date: Thu, 10 Mar 2005 11:25:10 +0800 Subject: [uClinux-dev] Does all busybox utilities support uClinux? Message-ID: <200503100327.j2A3Rlfw023364@cad.csie.ncku.edu.tw> Hi, all, When we make menuconfig for busybox in uClinux-dist-20041215, only a few utilities can be configured for uClinux. Doest it mean that only the appeared utilities support uClinux? How about other many busybox utilities? If all busybox utilities support uClinux, it will be very wonderful. Regards, Wayne -- No virus found in this outgoing message. Checked by AVG Anti-Virus. Version: 7.0.308 / Virus Database: 266.7.1 - Release Date: 2005/3/9 From wnlee at cad.csie.ncku.edu.tw Wed Mar 9 22:29:48 2005 From: wnlee at cad.csie.ncku.edu.tw (Wayne Lee) Date: Thu, 10 Mar 2005 11:29:48 +0800 Subject: [uClinux-dev] where should we store ttb for MMU? In-Reply-To: <200503100317.j2A3HOfw023264@cad.csie.ncku.edu.tw> Message-ID: <200503100332.j2A3WOfw023424@cad.csie.ncku.edu.tw> /* * Kernel startup entry point. */ __INIT .type stext, #function ENTRY(stext) mov r12, r0 mov r0, #PSR_F_BIT | PSR_I_BIT | MODE_SVC @ make sure svc mode msr cpsr_c, r0 @ and all irqs disabled /* * ARM920T need MMU enable for D-cache. * we just make the identical table before the MMU enabling. */ #ifdef CONFIG_CPU_MXU_ENABLE @ /* Write domain id (cp15_r3) */ @ mvn r0, #0 /* Domains 0, 1 = client */ @ mcr p15, 0, r0, c3, c0, 0 /* load domain access register */ @ /* Set control register v4 */ MOV r0, #0 MCR p15, 0, r0, c7, c7, 0 @ invalidate I,D caches MCR p15, 0, r0, c7, c10, 4 @ drain WB MCR p15, 0, r0, c8, c7, 0 @ invalidate I & D TLBs MRC p15, 0, r0, c1, c0, 0 @ ctrl register BIC r0, r0, #0x000f @ ............wcam BIC r0, r0, #0x1100 @ ...i...s........ MCR p15, 0, r0, c1, c0, 0 @ ctrl register @ mov r0, #0 @ mcr p15, 0, r0, c13, c0, 0 /* zero PID */ @ mcr p15, 0, r0, c7, c7, 0 /* invalidate I,D caches */ @ mcr p15, 0, r0, c7, c10, 4 /* drain write buffer */ @ mcr p15, 0, r0, c8, c7, 0 /* invalidate I,D TLBs */ @ mrc p15, 0, r0, c1, c0, 0 /* get control register */ @ bic r0, r0, #0x000f @ ............wcam @ bic r0, r0, #0x1100 @ ...i...s........ @ @bic r0, r0, #0x0001 /* disable MMU */ @ mcr p15, 0, r0, c1, c0, 0 /* write control register */ init_ttb: /* * We must reserve DRAM_BASE ~ DRAM_BASE+0x8000. * And Kernel TEXTADDR must be after DRAM_BASE + 0x8000 */ #if !defined(CONFIG_DRAM_BASE) || !defined(CONFIG_DRAM_SIZE) #define CONFIG_DRAM_BASE 0x00000000 #define CONFIG_DRAM_SIZE 0x04000000 #endif ldr r0,=(CONFIG_DRAM_BASE + 0x4000) /* set TTB (Must be on 16KB boundary) */ @ ldr r0,=(CONFIG_DRAM_BASE + CONFIG_DRAM_SIZE/2 + 0x4000) /* set TTB (Must be on 16KB boundary) */ @ ldr r0,=(CONFIG_DRAM_BASE + CONFIG_DRAM_SIZE - 0x4000) /* set TTB (Must be on 16KB boundary) */ @ ldr r0,=(CONFIG_DRAM_BASE + CONFIG_DRAM_SIZE) /* set TTB (Must be on 16KB boundary) */ mcr p15, 0, r0, c2, c0, 0 /* write to TTB(CP15,c2) */ /* * * Create translation table for flat mapping * Top 12 bits of VA is pointer into table * Create 4096 entries from 000xxxxx to fffxxxxx * * Set Section descriptor. 4K entry totla and * each entry describe 1MB. */ mov r1, #0x1000 @ 4K counter subs r1,r1,#1 mov r2,#0b000000000000 @ set access permissions (AP) for full access SVC/USR (11:10) @ mov r2,#0b110000000000 @ set access permissions (AP) for full access SVC/USR (11:10) orr r2,r2,#0b000111100000 @ set for domain 15 (8:5) orr r2,r2,#0b000000010000 @ must be 1 (4) orr r2,r2,#0b000000000000 @ set non cachable non bufferable (CB) (3:2) orr r2,r2,#0b000000000010 @ set for 1Mb section (1:0) init_ttb_1: /* create 4K entry */ orr r3,r2,r1,lsl#20 @ use loop counter to create individual table entries str r3,[r0,r1,lsl#2] @ str r3 at TTB base + loopcount*4 subs r1,r1,#1 @ decrement loop counter bpl init_ttb_1 /* resetting the SDRAM area to cacheable*/ /* DRAM_BASE ~ DRAM_BASE+DRAM_SIZE */ /* 64MB -> 64 entry -> 0x40 */ mov r1, #(CONFIG_DRAM_SIZE >> 20 ) @ 64 entry =ase addr subs r1, r1, #1 mov r2,#0b110000000000 @ set access permissions (AP) for full access SVC/USR (11:10) orr r2,r2,#0b000111100000 @ set for domain 15 (8:5) orr r2,r2,#0b000000010000 @ must be 1 (4) orr r2,r2,#0b000000001100 @ set cachable but unbufferable (CB) (3:2) @ orr r2,r2,#0b000000001000 @ set cachable but unbufferable (CB) (3:2) orr r2,r2,#0b000000000010 @ set for 1Mb section (1:0) init_ttb_2: /* create 64 entry (for 64MB SDRAM) */ add r4, r1, #(CONFIG_DRAM_BASE >> 20 ) @ drambase >> 20 bit . orr r3,r2,r4,lsl#20 @ use loop counter to create individual table entries str r3,[r0,r4,lsl#2] @ str r3 at TTB base + loopcount*4 subs r1,r1,#1 @ decrement loop counter bpl init_ttb_2 @init_domains : we define all domains are manager. so no access permission check is occured. mvn r0, #0 @ trick. 0 -1 == 0xFFFFFFFF mcr p15, 0, r0, c3, c0, 0 @ write to CP15 register 5 /* * set global core configurations */ mrc p15, 0, r0, c1, c0, 0 @ read CP15 register 1 into r0 orr r0, r0, #(0x1 <<12) @ enable I Cache orr r0, r0, #(0x1 <<2) @ enable D Cache orr r0, r0, #(0b11 <<30) @ enable asynchronous clocking mode orr r0, r0, #0x1 @ enable MMU mcr p15, 0, r0, c1, c0, 0 @ write cp15 register 1 /* * set global core configurations */ mrc p15, 0, r0, c1, c0, 0 @ read CP15 register 1 into r0 @#ifndef CONFIG_CPU_ICACHE_DISABLE @ orr r0, r0, #(0x1 <<12) @ enable I Cache @#endif @#ifndef CONFIG_CPU_DCACHE_DISABLE @ orr r0, r0, #(0x1 <<2) @ enable D Cache @#endif @#ifdef CONFIG_ALIGNMENT_TRAP @ orr r0, r0, #2 @ ...........A. @#endif @ orr r0, r0, #(0b11 <<30) @ enable asynchronous clocking mode @ orr r0, r0, #0x1 @ enable MMU mcr p15, 0, r0, c1, c0, 0 @ write cp15 register 1 nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop @ mrc p15, 0, r0, c1, c0, 0 @ read CP15 register 1 into r0 @#ifndef CONFIG_CPU_ICACHE_DISABLE @ orr r0, r0, #(0x1 <<12) @ enable I Cache @#endif @#ifndef CONFIG_CPU_DCACHE_DISABLE @ orr r0, r0, #(0x1 <<2) @ enable D Cache @#endif @ mcr p15, 0, r0, c1, c0, 0 @ write CP15 register 1 #endif /* * set global core configurations */ mrc p15, 0, r0, c1, c0, 0 @ read CP15 register 1 into r0 #ifndef CONFIG_CPU_ICACHE_DISABLE orr r0, r0, #(0x1 <<12) @ enable I Cache #endif mcr p15, 0, r0, c1, c0, 0 @ write cp15 register 1 -----Original Message----- From: uclinux-dev-bounces at uClinux.org [mailto:uclinux-dev-bounces at uClinux.org] On Behalf Of Wayne Lee Sent: Thursday, March 10, 2005 11:15 AM To: uClinux development list Subject: [uClinux-dev] where should we store ttb for MMU? Hi, folks, In fact, I am now porting an uClinux no VM port on arm926ej-s CPU. In order to turn-on D-cache, I use flat-mapping & section-mapped MMU and then turn on D-cache. The method is referred from Samsung s5c7375 port. I store TTB in 0x4000. However, while kernel is booting, the TTB is overwritten and data abort occurred. I am wonder where should we place the ttb in uClinux 2.6? Do we need to modify the vmlinux.lds? Please kindly give me your suggestions. Wayne /* * Kernel startup entry point. */ __INIT .type stext, #function ENTRY(stext) mov r12, r0 mov r0, #PSR_F_BIT | PSR_I_BIT | MODE_SVC @ make sure svc mode msr cpsr_c, r0 @ and all irqs disabled /* * ARM920T need MMU enable for D-cache. * we just make the identical table before the MMU enabling. */ #ifdef CONFIG_CPU_MXU_ENABLE @ /* Write domain id (cp15_r3) */ @ mvn r0, #0 /* Domains 0, 1 = client */ @ mcr p15, 0, r0, c3, c0, 0 /* load domain access register */ @ /* Set control register v4 */ MOV r0, #0 MCR p15, 0, r0, c7, c7, 0 @ invalidate I,D caches MCR p15, 0, r0, c7, c10, 4 @ drain WB MCR p15, 0, r0, c8, c7, 0 @ invalidate I & D TLBs MRC p15, 0, r0, c1, c0, 0 @ ctrl register BIC r0, r0, #0x000f @ ............wcam BIC r0, r0, #0x1100 @ ...i...s........ MCR p15, 0, r0, c1, c0, 0 @ ctrl register @ mov r0, #0 @ mcr p15, 0, r0, c13, c0, 0 /* zero PID */ @ mcr p15, 0, r0, c7, c7, 0 /* invalidate I,D caches */ @ mcr p15, 0, r0, c7, c10, 4 /* drain write buffer */ @ mcr p15, 0, r0, c8, c7, 0 /* invalidate I,D TLBs */ @ mrc p15, 0, r0, c1, c0, 0 /* get control register */ @ bic r0, r0, #0x000f @ ............wcam @ bic r0, r0, #0x1100 @ ...i...s........ @ @bic r0, r0, #0x0001 /* disable MMU */ @ mcr p15, 0, r0, c1, c0, 0 /* write control register */ init_ttb: /* * We must reserve DRAM_BASE ~ DRAM_BASE+0x8000. * And Kernel TEXTADDR must be after DRAM_BASE + 0x8000 */ #if !defined(CONFIG_DRAM_BASE) || !defined(CONFIG_DRAM_SIZE) #define CONFIG_DRAM_BASE 0x00000000 #define CONFIG_DRAM_SIZE 0x04000000 #endif @ ldr r0,=(CONFIG_DRAM_BASE + 0x4000) /* set TTB (Must be on 16KB boundary) */ @ ldr r0,=(CONFIG_DRAM_BASE + CONFIG_DRAM_SIZE/2 + 0x4000) /* set TTB (Must be on 16KB boundary) */ @ ldr r0,=(CONFIG_DRAM_BASE + CONFIG_DRAM_SIZE - 0x4000) /* set TTB (Must be on 16KB boundary) */ ldr r0,=(CONFIG_DRAM_BASE + CONFIG_DRAM_SIZE) /* set TTB (Must be on 16KB boundary) */ mcr p15, 0, r0, c2, c0, 0 /* write to TTB(CP15,c2) */ /* * * Create translation table for flat mapping * Top 12 bits of VA is pointer into table * Create 4096 entries from 000xxxxx to fffxxxxx * * Set Section descriptor. 4K entry totla and * each entry describe 1MB. */ mov r1, #0x1000 @ 4K counter subs r1,r1,#1 mov r2,#0b000000000000 @ set access permissions (AP) for full access SVC/USR (11:10) @ mov r2,#0b110000000000 @ set access permissions (AP) for full access SVC/USR (11:10) orr r2,r2,#0b000111100000 @ set for domain 15 (8:5) orr r2,r2,#0b000000010000 @ must be 1 (4) orr r2,r2,#0b000000000000 @ set non cachable non bufferable (CB) (3:2) orr r2,r2,#0b000000000010 @ set for 1Mb section (1:0) init_ttb_1: /* create 4K entry */ orr r3,r2,r1,lsl#20 @ use loop counter to create individual table entries str r3,[r0,r1,lsl#2] @ str r3 at TTB base + loopcount*4 subs r1,r1,#1 @ decrement loop counter bpl init_ttb_1 /* resetting the SDRAM area to cacheable*/ /* DRAM_BASE ~ DRAM_BASE+DRAM_SIZE */ /* 64MB -> 64 entry -> 0x40 */ mov r1, #(CONFIG_DRAM_SIZE >> 20 ) @ 64 entry =ase addr subs r1, r1, #1 mov r2,#0b110000000000 @ set access permissions (AP) for full access SVC/USR (11:10) orr r2,r2,#0b000111100000 @ set for domain 15 (8:5) orr r2,r2,#0b000000010000 @ must be 1 (4) orr r2,r2,#0b000000001100 @ set cachable but unbufferable (CB) (3:2) @ orr r2,r2,#0b000000001000 @ set cachable but unbufferable (CB) (3:2) orr r2,r2,#0b000000000010 @ set for 1Mb section (1:0) init_ttb_2: /* create 64 entry (for 64MB SDRAM) */ add r4, r1, #(CONFIG_DRAM_BASE >> 20 ) @ drambase >> 20 bit . orr r3,r2,r4,lsl#20 @ use loop counter to create individual table entries str r3,[r0,r4,lsl#2] @ str r3 at TTB base + loopcount*4 subs r1,r1,#1 @ decrement loop counter bpl init_ttb_2 @init_domains : we define all domains are manager. so no access permission check is occured. mvn r0, #0 @ trick. 0 -1 == 0xFFFFFFFF mcr p15, 0, r0, c3, c0, 0 @ write to CP15 register 5 /* * set global core configurations */ mrc p15, 0, r0, c1, c0, 0 @ read CP15 register 1 into r0 orr r0, r0, #(0x1 <<12) @ enable I Cache orr r0, r0, #(0x1 <<2) @ enable D Cache orr r0, r0, #(0b11 <<30) @ enable asynchronous clocking mode orr r0, r0, #0x1 @ enable MMU mcr p15, 0, r0, c1, c0, 0 @ write cp15 register 1 /* * set global core configurations */ mrc p15, 0, r0, c1, c0, 0 @ read CP15 register 1 into r0 @#ifndef CONFIG_CPU_ICACHE_DISABLE @ orr r0, r0, #(0x1 <<12) @ enable I Cache @#endif @#ifndef CONFIG_CPU_DCACHE_DISABLE @ orr r0, r0, #(0x1 <<2) @ enable D Cache @#endif @#ifdef CONFIG_ALIGNMENT_TRAP @ orr r0, r0, #2 @ ...........A. @#endif @ orr r0, r0, #(0b11 <<30) @ enable asynchronous clocking mode @ orr r0, r0, #0x1 @ enable MMU mcr p15, 0, r0, c1, c0, 0 @ write cp15 register 1 nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop nop @ mrc p15, 0, r0, c1, c0, 0 @ read CP15 register 1 into r0 @#ifndef CONFIG_CPU_ICACHE_DISABLE @ orr r0, r0, #(0x1 <<12) @ enable I Cache @#endif @#ifndef CONFIG_CPU_DCACHE_DISABLE @ orr r0, r0, #(0x1 <<2) @ enable D Cache @#endif @ mcr p15, 0, r0, c1, c0, 0 @ write CP15 register 1 #endif /* * set global core configurations */ mrc p15, 0, r0, c1, c0, 0 @ read CP15 register 1 into r0 #ifndef CONFIG_CPU_ICACHE_DISABLE orr r0, r0, #(0x1 <<12) @ enable I Cache #endif mcr p15, 0, r0, c1, c0, 0 @ write cp15 register 1 -- No virus found in this outgoing message. Checked by AVG Anti-Virus. Version: 7.0.308 / Virus Database: 266.7.1 - Release Date: 2005/3/9 _______________________________________________ uClinux-dev mailing list uClinux-dev at uClinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev at uclinux.org -- No virus found in this incoming message. Checked by AVG Anti-Virus. Version: 7.0.308 / Virus Database: 266.7.1 - Release Date: 2005/3/9 -- No virus found in this outgoing message. Checked by AVG Anti-Virus. Version: 7.0.308 / Virus Database: 266.7.1 - Release Date: 2005/3/9 From davidm at snapgear.com Wed Mar 9 22:48:25 2005 From: davidm at snapgear.com (David McCullough) Date: Thu, 10 Mar 2005 13:48:25 +1000 Subject: [uClinux-dev] Does all busybox utilities support uClinux? In-Reply-To: <200503100327.j2A3Rlfw023364@cad.csie.ncku.edu.tw> References: <200503100327.j2A3Rlfw023364@cad.csie.ncku.edu.tw> Message-ID: <20050310034825.GA1171@beast> Jivin Wayne Lee lays it down ... > > > Hi, all, > > When we make menuconfig for busybox in uClinux-dist-20041215, only a > few utilities can be configured for uClinux. > Doest it mean that only the appeared utilities support uClinux? > How about other many busybox utilities? > > If all busybox utilities support uClinux, it will be very wonderful. I thought that uClinux-dist-20041215 would have listed all busybox utilities (well almost all of them). Make sure you check in the busybox config page, it should be huge. Not all of them are on by default, but you should be able to turn which ever ones you want on, Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From wnlee at cad.csie.ncku.edu.tw Wed Mar 9 23:15:19 2005 From: wnlee at cad.csie.ncku.edu.tw (Wayne Lee) Date: Thu, 10 Mar 2005 12:15:19 +0800 Subject: [uClinux-dev] Does all busybox utilities support uClinux? In-Reply-To: <20050310034825.GA1171@beast> Message-ID: <200503100418.j2A4Hufw023930@cad.csie.ncku.edu.tw> Hi, David, You are right. :) I copy busybox from uClinux-dist-20041215/user to uClinux-20040408/user and it does not take effect. After using uClinux-dist-20041215, more utilities are supported than older uClinux-dist. Regards, Wayne -----Original Message----- From: David McCullough [mailto:davidm at snapgear.com] Sent: Thursday, March 10, 2005 11:48 AM To: Wayne Lee Cc: uClinux development list Subject: Re: [uClinux-dev] Does all busybox utilities support uClinux? Jivin Wayne Lee lays it down ... > > > Hi, all, > > When we make menuconfig for busybox in uClinux-dist-20041215, only a > few utilities can be configured for uClinux. > Doest it mean that only the appeared utilities support uClinux? > How about other many busybox utilities? > > If all busybox utilities support uClinux, it will be very wonderful. I thought that uClinux-dist-20041215 would have listed all busybox utilities (well almost all of them). Make sure you check in the busybox config page, it should be huge. Not all of them are on by default, but you should be able to turn which ever ones you want on, Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org -- No virus found in this incoming message. Checked by AVG Anti-Virus. Version: 7.0.308 / Virus Database: 266.7.1 - Release Date: 2005/3/9 -- No virus found in this outgoing message. Checked by AVG Anti-Virus. Version: 7.0.308 / Virus Database: 266.7.1 - Release Date: 2005/3/9 From davidm at snapgear.com Wed Mar 9 23:31:41 2005 From: davidm at snapgear.com (David McCullough) Date: Thu, 10 Mar 2005 14:31:41 +1000 Subject: [uClinux-dev] Does all busybox utilities support uClinux? In-Reply-To: <200503100418.j2A4Hufw023930@cad.csie.ncku.edu.tw> References: <20050310034825.GA1171@beast> <200503100418.j2A4Hufw023930@cad.csie.ncku.edu.tw> Message-ID: <20050310043141.GA6538@beast> Jivin Wayne Lee lays it down ... > > Hi, David, > > You are right. :) > I copy busybox from uClinux-dist-20041215/user to uClinux-20040408/user and > it does not take effect. > After using uClinux-dist-20041215, more utilities are supported than older > uClinux-dist. It may be possible to add the config/config.in busybox section from the uClinux-dist-20041215 to uClinux-20040408 if you really need to stay on the old version, Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From kavlak at rediffmail.com Thu Mar 10 00:37:19 2005 From: kavlak at rediffmail.com (kavitha lakshmi) Date: 10 Mar 2005 05:37:19 -0000 Subject: [uClinux-dev] Multicast problem in uclinux 2.0.38 Message-ID: <20050310053719.21247.qmail@webmail17.rediffmail.com> Hi all, I am trying to multicast from a arm7tdmi based board using uClinux 2.0.38 but not successful.I have given the configuration details,could someone help how to enable configure to get multicast work. #ifconfig eth0 192.168.7.4 broadcast 192.168.7.255 netmask 255.255.255.0 #route add -net 192.168.7.0 netmask 255.255.255.0 eth0 #route add -net 235.23.5.0 netmask 255.255.255.0 eth0 note:ping works Also tried multicast and allmulti options for the interface, but did'nt work -------------- next part -------------- An HTML attachment was scrubbed... URL: From swenggsri at yahoo.com Thu Mar 10 00:44:41 2005 From: swenggsri at yahoo.com (Sriram Chadalavada) Date: Wed, 9 Mar 2005 21:44:41 -0800 (PST) Subject: [uClinux-dev] 2.6 kernel image size HUGE!!!! In-Reply-To: 6667 Message-ID: <20050310054441.55034.qmail@web21603.mail.yahoo.com> Thanks Greg. That fixed the problem. Cheers, Sriram --- Greg Ungerer wrote: > Hi Sriram, > > Sriram Chadalavada wrote: > > I am trying to download a 2.6.x kernel (part of > > latest 20041215 distribution) on a Arcturus uCdimm > > that has 8MB RAM and 2 MB ROM. > > > > Unfortunately, when I try compiling the uClinux > > distribution with this kernel, the image size is > > turning to be a whopping 272MB!!!!! > > The problem turns out to be that the final objcopy > to > create the linux.text file is not removing the .init > section, as it should be. > > > > total 272M > > -rw-r--r-- 1 root root 270M Feb 21 18:57 > image.bin > > -rwxr-xr-x 1 root root 112K Feb 21 18:57 > linux.data > > -rwxr-xr-x 1 root root 269M Feb 21 18:57 > linux.text > > -rw-r--r-- 1 root root 791K Feb 21 18:57 > romfs.img > > > > linux.text seems to be the main culprit... > > > > When I compile with the 2.4.x kernel, the sizes > were: > > > > total 2.8M > > -rw-r--r-- 1 root root 1.4M Mar 5 21:43 > image.bin > > -rwxr-xr-x 1 root root 168K Mar 5 21:43 > linux.data > > -rwxr-xr-x 1 root root 687K Mar 5 21:43 > linux.text > > -rw-r--r-- 1 root root 540K Mar 5 21:43 > romfs.img > > > > That's almost 10 times smaller. Some postings on > the > > mailing list did talk about this problem but I was > not > > able to find a solution. > > > > What is the typical (default for uCdimm) image > size > > with the 2.6.x kernel? Is it always larger than > than > > 1MB? Please let me know if you need any other > > information to solve this problem. > > The image size should be only marginally larger than > the 2.4 image. > > Try the attached patch taht fixes the kernel text > objcopy. > > Regards > Greg > > > > ------------------------------------------------------------------------ > Greg Ungerer -- Chief Software Dude EMAIL: > gerg at snapgear.com > SnapGear -- a CyberGuard Company PHONE: > +61 7 3435 2888 > 825 Stanley St, FAX: > +61 7 3891 3630 > Woolloongabba, QLD, 4102, Australia WEB: > http://www.SnapGear.com > > Index: vendors/Arcturus/uCdimm/Makefile > =================================================================== > RCS file: > /cvs/sw/new-wave/vendors/Arcturus/uCdimm/Makefile,v > retrieving revision 1.3 > diff -u -r1.3 Makefile > --- vendors/Arcturus/uCdimm/Makefile 20 Jun 2004 > 23:56:14 -0000 1.3 > +++ vendors/Arcturus/uCdimm/Makefile 8 Mar 2005 > 06:51:58 -0000 > @@ -82,7 +82,7 @@ > $(ROOTDIR)/$(LINUXDIR)/linux > $(IMAGEDIR)/linux.data > $(CROSS_COMPILE)objcopy -O binary > --remove-section=.ramvec \ > --remove-section=.bss --remove-section=.data \ > - --remove-section=.eram \ > + --remove-section=.init --remove-section=.eram \ > > --set-section-flags=.romvec=CONTENTS,ALLOC,LOAD,READONLY,CODE > \ > $(ROOTDIR)/$(LINUXDIR)/linux > $(IMAGEDIR)/linux.text > cat $(IMAGEDIR)/linux.text $(IMAGEDIR)/linux.data > $(ROMFSIMG) > $(IMAGE) > __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/ From lars.friedrich at wago.com Thu Mar 10 02:33:28 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Thu, 10 Mar 2005 08:33:28 +0100 Subject: [uClinux-dev] Possible GPL violation inJuiceBoxdistributedbyemsoft Message-ID: > > No. > > I'm saying that "it's a loadable module" is not an argument > > why it does not fall under the GPL. > So, a standalone driver, which is subsequently ported to uClinux > as a loadable module doesn't fall under the GPL, I would use the term "is likely not to fall under..". > but a driver that is originally designed to be a loadable module > of uClinux does, regardless of whether that > driver works with another OS or not?? Well, it's impossible that a driver works with another OS. Maybe there is an architecture independant library which can be used by different drivers for different OSs. But yes, a driver that was originally developed with the linux kernel in mind is more likely to be derived from the kernel. Look at it this way: You can write a driver which can be linked or loaded - with the same source code. Now if the source code is completely the same, why would the source code in one case fall under the GPL and in the other way not, just depending on how the source code is used? This does not make sense. The source code is either derived work or is not, it can't matter how you compile the source code to determine if it is derived or not. People use things like "if it's a patch, it's derived" to simplify the legal situation and some people start to think that this is what counts - how source code is used. But this is of course nonsense. The abstraction "if it's a patch, it's derived" does not mean the source code of a patch is derived because it patches the kernel, it means:"The source code plays so much with the kernel internals that it must be derived work." Similiar with loadable modules, just the other way around. You just can't say "it's a loadable module and therefore not derived". The question is how much it plays (or relies on) kernel internals. You would never dare to say that Tolkien's Lord of the Rings is copyrighted when you read it on paper but not when you read it as pdf online. The content decides the legal situation, not the form. Similiar the content of the loadable module decides if it's derived from the kernel or not and not just the fact in which form the source code is usually used. Best regards, Lars Friedrich From swenggsri at yahoo.com Thu Mar 10 04:01:13 2005 From: swenggsri at yahoo.com (Sriram Chadalavada) Date: Thu, 10 Mar 2005 01:01:13 -0800 (PST) Subject: [uClinux-dev] 2.6 kernel not booting on uCdimm Message-ID: <20050310090113.2941.qmail@web21605.mail.yahoo.com> Hi everyone, Thanks Greg for letting me know about the 2.6 Makefile patch. I urgently need to get the 2.6 kernel working on uCdimm to test a program. My uCdimm boots normally (from ROM in all cases) when a 2.4.24 kernel (20041215 distribution) is programmed onto it. But when I program 2.6 kernel images (~ 1.6MB) onto the uCdimm flash and reboot, only the following is displayed on the serial console, after which the uCdimm hangs. ======================================================= uCbootstrap v1.5.9 (c) Copyright 1999-2000 Rt-Control Inc. (c) Copyright 2000-2001 Lineo, Inc. All Rights Reserved. AUTOBOOT not "yes" or greater than 0. Ignored FLASH type 00c0 [AT49BV1614] DP|002000 DP|004000 DP|006000 DP|008000 DP|00a000 DP|00c000 DP|00e000 D-|010000 D-|018000 D-|020000 D-|030000 D-|040000 D-|050000 D-|060000 D-|070000 D-|080000 D-|090000 D-|0a0000 D-|0b0000 D-|0c0000 D-|0d0000 D-|0e0000 D-|0f0000 D-|100000 D-|110000 D-|120000 D-|130000 D-|140000 D-|150000 D-|160000 D-|170000 D-|180000 D-|190000 D-|1a0000 D-|1b0000 D-|1c0000 D-|1d0000 D-|1e0000 D-|1f0000 D-|200000 B$ go go! 0x10c10000 ======================================================= I tried with 2.6.5 kernel of the distribution AND a uCLinux patched 2.6.10 kernel. I also tried on a different uCdimm. Same result. The kernel configuration for the 2.6.10 kernel is attached. I would greatly appreciate if someone can point out what is going on here. Please let me know if you need more information. Thanks in advance, Sriram __________________________________ Do you Yahoo!? Yahoo! Sports - Sign up for Fantasy Baseball. http://baseball.fantasysports.yahoo.com/ -------------- next part -------------- A non-text attachment was scrubbed... Name: config-snapshot Type: application/octet-stream Size: 2373 bytes Desc: config-snapshot URL: From lara_john79 at yahoo.com Thu Mar 10 07:00:14 2005 From: lara_john79 at yahoo.com (Lara John) Date: Thu, 10 Mar 2005 04:00:14 -0800 (PST) Subject: [uClinux-dev] NEC USB UHCI related problem on MS 7751R CP 01 Message-ID: <20050310120014.78893.qmail@web53204.mail.yahoo.com> Dear all, I am trying to port embedded linux kernel (ver. 2.4.26) on MS7751RCP01. This board has got NEC's usb 2.0 UHCI host controller. I am facing followoing problems. 1) This controller is detected by the kernel on boot time, but it is probing the wrong IRQ. (The IRQ to IRL is managed by the FPGA), the kernel should allocate IRQ 9 but the messages says it is IRQ 10 The /proc/bus/usb/devices says: T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=12 MxCh= 1 B: Alloc= 0/900 us ( 0%), #Int= 0, #Iso= 0 D: Ver= 1.10 Cls=09(hub ) Sub=00 Prot=00 MxPS= 8 #Cfgs= 1 P: Vendor=0000 ProdID=0000 Rev= 0.00 S: Product=USB OHCI Root Hub S: SerialNumber=fd001000 C:* #Ifs= 1 Cfg#= 1 Atr=40 MxPwr= 0mA I: If#= 0 Alt= 0 #EPs= 1 Cls=09(hub ) Sub=00 Prot=00 Driver=hub E: Ad=81(I) Atr=03(Int.) MxPS= 2 Ivl=255ms T: Bus=01 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=12 MxCh= 1 B: Alloc= 0/900 us ( 0%), #Int= 0, #Iso= 0 D: Ver= 1.10 Cls=09(hub ) Sub=00 Prot=00 MxPS= 8 #Cfgs= 1 P: Vendor=0000 ProdID=0000 Rev= 0.00 S: Product=USB OHCI Root Hub S: SerialNumber=fd000000 C:* #Ifs= 1 Cfg#= 1 Atr=40 MxPwr= 0mA I: If#= 0 Alt= 0 #EPs= 1 Cls=09(hub ) Sub=00 Prot=00 Driver=hub E: Ad=81(I) Atr=03(Int.) MxPS= 2 Ivl=255ms 2) I am trying to mount SD Card (MMC) on USB with usb reader. I have selected the build in option for SCSI Storage and USB storage related. If i boot the kernel without this device, it goes fine, but as soon as i attatch the USB reader with SD card, it hangs (no panic.. seems that it is in some ageless loop :( ) The last message it gave is as follows: hub.c: new USB device 00:03.0-1, assigned address 2 Please help me resolving this issue. If the solution related to it is already resolved in newer releases please let me know. Thanks in advance. Best regards, Lara John PS: If any information is missing please let me know. __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/ From jsujjavanich at syntech-fuelmaster.com Thu Mar 10 08:35:02 2005 From: jsujjavanich at syntech-fuelmaster.com (Jate Sujjavanich) Date: Thu, 10 Mar 2005 08:35:02 -0500 Subject: [uClinux-dev] m68k-bdm-elf-gdb Cygwin Message-ID: The bug can be repaired with a patch to GDB. I received that error under Cygwin and Linux. The patch is on the following mailing list archive page: http://mailman.uclinux.org/pipermail/uclinux-dev/2004-March/024885.html It works under Linux for sure, but is untested on Windows. The archived post is buried in the search engine. - Jate S. -----Original Message----- From: uclinux-dev-bounces at uClinux.org [mailto:uclinux-dev-bounces at uClinux.org] On Behalf Of mike Sent: Wednesday, March 09, 2005 9:00 PM To: 'uClinux development list' Subject: [uClinux-dev] m68k-bdm-elf-gdb Cygwin Hello, Has anyone successfully used the m68k-bdm-elf-gdb from Cygwin with gdbserver TCP/IP on uClinux? (Network or Serial) I have tried the one from http://www.uclinux.org/pub/uClinux/m68k-elf-tools/gcc-3/ (GDB 6.1) and am getting the following message (Software breakpoints not implemented for this target) when trying to do anything (i.e. set a breakpoint, step, etc). After searching the archives I see that this has come up before (http://mailman.uclinux.org/pipermail/uclinux-dev/2004-February/024150.h tml) , but no resolution has been posted. The error message is the same under both a Linux Host (Red Hat 9) and Cygwin for GDB 6.1. If I use GDB 5.0 as part of the older toolchain (http://www.uclinux.org/pub/uClinux/m68k-elf-tools/) on a Linux (Red Hat 9) host then everything seems to work OK (there is no pre-built Cygwin chain for testing this version). Does anyone know why 6.1 version of gdb doesn't work with uClinux gdbserver? Does the gdbserver and gdb host need to be the same version? If so what version is the gdbserver included with the uClinux build? And what host versions are compatible. I really need to be able to use gdb 6.0 or above on the host machine and on windows. Thanks, Mike Lavender _______________________________________________ uClinux-dev mailing list uClinux-dev at uClinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev at uclinux.org From paul-rimmer at ronin-tech.com Thu Mar 10 09:45:08 2005 From: paul-rimmer at ronin-tech.com (paul-rimmer at ronin-tech.com) Date: Thu, 10 Mar 2005 07:45:08 -0700 (MST) Subject: [uClinux-dev] m68k-bdm-elf-gdb Cygwin In-Reply-To: References: Message-ID: <4231.216.123.208.8.1110465908.squirrel@ronin-tech.com> Anyone know if this allows debugging of threaded applications? Cheers, Paul > The bug can be repaired with a patch to GDB. I received that error under > Cygwin and Linux. The patch is on the following mailing list archive > page: > > http://mailman.uclinux.org/pipermail/uclinux-dev/2004-March/024885.html > > It works under Linux for sure, but is untested on Windows. The archived > post is buried in the search engine. > > - Jate S. > > -----Original Message----- > From: uclinux-dev-bounces at uClinux.org > [mailto:uclinux-dev-bounces at uClinux.org] On Behalf Of mike > Sent: Wednesday, March 09, 2005 9:00 PM > To: 'uClinux development list' > Subject: [uClinux-dev] m68k-bdm-elf-gdb Cygwin > > Hello, > > Has anyone successfully used the m68k-bdm-elf-gdb from Cygwin with > gdbserver TCP/IP on uClinux? (Network or Serial) > > I have tried the one from > http://www.uclinux.org/pub/uClinux/m68k-elf-tools/gcc-3/ (GDB 6.1) and > am getting the following message (Software breakpoints not implemented > for this > target) when trying to do anything (i.e. set a breakpoint, step, etc). > After searching the archives I see that this has come up before > (http://mailman.uclinux.org/pipermail/uclinux-dev/2004-February/024150.h > tml) > , but no resolution has been posted. > > The error message is the same under both a Linux Host (Red Hat 9) > and Cygwin for GDB 6.1. If I use GDB 5.0 as part of the older toolchain > (http://www.uclinux.org/pub/uClinux/m68k-elf-tools/) on a Linux (Red Hat > 9) host then everything seems to work OK (there is no pre-built Cygwin > chain for testing this version). > > Does anyone know why 6.1 version of gdb doesn't work with uClinux > gdbserver? Does the gdbserver and gdb host need to be the same version? > If so what version is the gdbserver included with the uClinux build? And > what host versions are compatible. I really need to be able to use gdb > 6.0 or above on the host machine and on windows. > > Thanks, > > Mike Lavender From mbeaulieu at eocycle.com Thu Mar 10 10:58:03 2005 From: mbeaulieu at eocycle.com (Martin Beaulieu) Date: Thu, 10 Mar 2005 10:58:03 -0500 Subject: [uClinux-dev] insmod modules Message-ID: <027301c52589$f1d766a0$8000a8c0@MartinLogiciel> Hi, I have some problems when i try to load a module in my kernel, my kernel freeze. I have the linux-2.4.x that boot in rom on a 5407. I use the blkmem utility. What is the problem ? Thanks Martin -------------- next part -------------- An HTML attachment was scrubbed... URL: From zaidan_alaoui at menara.ma Thu Mar 10 11:01:19 2005 From: zaidan_alaoui at menara.ma (Zaidan ALAOUI) Date: Thu, 10 Mar 2005 16:01:19 -0000 Subject: [uClinux-dev] CTRL-E specification for ADSL 2+ modem Message-ID: <1B2047299D24EF4BA0457A57818EE9BC02E04EEF@EXCLU2.menara.local> Hello , I am trying to write a driver for a ADSL 2+ Modem based on the ST20196 or the MTC 20166 chip. However, I haven't arrived to get the CTL-E interface specification used to contol this modem. Please, if someone have this spec or any modem driver source code that uses this interface. Please, send to me. Thanks in advance, best regrads, Zaidan From m.sicheri at ctsgroup.it Thu Mar 10 11:22:45 2005 From: m.sicheri at ctsgroup.it (Marco SICHERI [CTS]) Date: Thu, 10 Mar 2005 17:22:45 +0100 Subject: [uClinux-dev] insmod modules References: <027301c52589$f1d766a0$8000a8c0@MartinLogiciel> Message-ID: <012201c5258d$64e2aa80$110110ac@Msicheri> What mean 'kernel freeze'? What do you mean for kernel freeze? Ossequi, (Best Regards/Obrigado) Marco SICHERI CTS electronics - R&D Department phone: +39 0125 235637, +39 0125 235630 e-mail: m.sicheri at ctsgroup.it http://www.ctsgroup.it ----- Original Message ----- From: Martin Beaulieu To: uClinux development list Sent: Thursday, March 10, 2005 4:58 PM Subject: [uClinux-dev] insmod modules Hi, I have some problems when i try to load a module in my kernel, my kernel freeze. I have the linux-2.4.x that boot in rom on a 5407. I use the blkmem utility. What is the problem ? Thanks Martin ------------------------------------------------------------------------------ _______________________________________________ uClinux-dev mailing list uClinux-dev at uClinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev at uclinux.org -------------- next part -------------- An HTML attachment was scrubbed... URL: From mbeaulieu at eocycle.com Thu Mar 10 11:17:22 2005 From: mbeaulieu at eocycle.com (Martin Beaulieu) Date: Thu, 10 Mar 2005 11:17:22 -0500 Subject: [uClinux-dev] insmod modules References: <20050310160440.4430C14DA29@mail07.powweb.com> Message-ID: <028601c5258c$a52e47d0$8000a8c0@MartinLogiciel> Hi, this is the last line : VFS: Mounted root (romfs filesystem) readonly. Sash command shell (version 1.1.1) /> insmod uccrt insmod: /lib/modules/2.4.22-uc0: uccrt is my own application. It is a application for RTAI. This application work with the uClinux-dist and RTAI in ram on the 5407 eval board. I try to load uClinux and RTAI in rom on a 5407 (on my own board). I use the uClinux-dist-20030909 with linux-2.4.x. I use the blkmem utility that i modify for the arena[0].address point to my flash. Thanks a lot Martin ----- Original Message ----- From: "Carlos Manuel Duclos Vergara" To: "Martin Beaulieu" Sent: Thursday, March 10, 2005 11:04 AM Subject: Re: [uClinux-dev] insmod modules > Hi, > >>I have some problems when i try to load a module in my kernel, my kernel >>freeze. >>I have the linux-2.4.x that boot in rom on a 5407. I use the blkmem >>utility. >>What is the problem ? > > a little bit of information about your problem would be really helpful. > Can you describe us what are you doing, what module are you trying to > load, what kernel version....... > > and obviously, try to paste the error messages here > > cheers! > -- > Carlos Manuel Duclos Vergara > http://www.elbazaar.cl > > From ben at thingmagic.com Wed Mar 9 17:51:19 2005 From: ben at thingmagic.com (Ben Giddings) Date: Wed, 09 Mar 2005 17:51:19 -0500 Subject: [uClinux-dev] Problems with Snapgear 3.2 for Xscale (Coyote-like) system In-Reply-To: <422E433B.9090104@snapgear.com> References: <4224A62A.3040909@thingmagic.com> <4225163D.4010601@snapgear.com> <42264B00.3060904@thingmagic.com> <4226649D.9050200@snapgear.com> <422CF957.9080204@thingmagic.com> <422D505C.8010208@snapgear.com> <422DF8D8.1020506@thingmagic.com> <422E1D07.3020607@thingmagic.com> <422E433B.9090104@snapgear.com> Message-ID: <422F7DE7.6040300@thingmagic.com> Greg Ungerer wrote: > I would be nice if when you get it all too work we generate a patch > specifically for use without the DSP library... I've got a patch here that works for me. I don't know if it does everything you want, and since I'm using Linux I don't care much about the vxworks stuff, but I'll attach it and you can take a look. Ben -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: modules-no_dsp-2005-03-09.patch URL: From carlos at embedded.cl Thu Mar 10 12:59:43 2005 From: carlos at embedded.cl (Carlos Manuel Duclos Vergara) Date: Thu, 10 Mar 2005 17:59:43 -0000 Subject: [uClinux-dev] insmod modules Message-ID: <20050310175946.44CCE14DA4A@mail07.powweb.com> An embedded and charset-unspecified text was scrubbed... Name: not available URL: From carlos at embedded.cl Thu Mar 10 14:05:18 2005 From: carlos at embedded.cl (Carlos Manuel Duclos Vergara) Date: Thu, 10 Mar 2005 19:05:18 -0000 Subject: [uClinux-dev] insmod modules Message-ID: <20050310190525.0158614DA18@mail07.powweb.com> An embedded and charset-unspecified text was scrubbed... Name: not available URL: From swenggsri at yahoo.com Thu Mar 10 14:28:17 2005 From: swenggsri at yahoo.com (Sriram Chadalavada) Date: Thu, 10 Mar 2005 11:28:17 -0800 (PST) Subject: [uClinux-dev] uclinux 2.6.10 fails on linking Message-ID: <20050310192817.48961.qmail@web21603.mail.yahoo.com> Hi Jate, I had a similar problem with the 2.6.10 kernel on uCdimm (Motorola MC68VZ328). I commented out the statement: obj-$(CONFIG_SHMEM) += shmem.o in the linux-2.6.x/mm/Makefile(for some reason, the .config always includes CONFIG_SHMEM). That fixed the 'shmem' linker errors. - Sriram __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/ From carlos at embedded.cl Thu Mar 10 15:15:42 2005 From: carlos at embedded.cl (Carlos Manuel Duclos Vergara) Date: Thu, 10 Mar 2005 20:15:42 -0000 Subject: [uClinux-dev] insmod modules Message-ID: <20050310201545.7EB7814DA29@mail07.powweb.com> An embedded and charset-unspecified text was scrubbed... Name: not available URL: From mbeaulieu at eocycle.com Thu Mar 10 16:20:50 2005 From: mbeaulieu at eocycle.com (Martin Beaulieu) Date: Thu, 10 Mar 2005 16:20:50 -0500 Subject: Fw: [uClinux-dev] insmod modules Message-ID: <037701c525b7$09a665b0$8000a8c0@MartinLogiciel> > Hi, > > I load my kernel in flash not in ram. > > When i check the sdram with my bootloader, all is OK. And i have the same > configuration in uClinux. I thinks that the problem is not my sdram > configuration. > > P.S. In last test, i don't call module with the good directory :( When i > include directory, i have the following message > > /> insmod /lib/modules/rtai/uccrt.o > BINFMT_FLAT: ROM mapping of file (we hope) > BINFMT_FLAT: Allocated data+bss+stack (25011 bytes): d0004 > p=d7ffc > start_thread(regs=0xddfcc, entry=0x18f2128, start_stack=0xd7f7c) > Using /lib/ > thanks for all > > Martin > > ----- Original Message ----- > From: "Carlos Manuel Duclos Vergara" > To: "Martin Beaulieu" > Cc: > Sent: Thursday, March 10, 2005 3:15 PM > Subject: Re: [uClinux-dev] insmod modules > > >> >I am not sur that the config of my sdram controller is good but it work. >> >> If I was you I would started looking into this kind of problems >> >>> make some test and active binfmt_flat trace. The result is : >>>/> insmod uccrt >>>BINFMT_FLAT: ROM mapping of file (we hope) >>>BINFMT_FLAT: Allocated data+bss+stack (24367 bytes): d8004 >>>p=dfffc >>>start_thread(regs=0xcffcc, entry=0x18f2128, start_stack=0xdff90) >>>insmod: /lib/modules/2.4.22-uc0: Nof sruceh efi_lef odr sdieretct oray r >> >> typical sdram timing problem, read as: >> >> "No such file or directory ...." >> >>>I have the same problem >> >> see??, there is a problem with your sdram controller initialization. Take >> the datasheet of your sdram chips and write a small program to test your >> ram. Then when you get the parameters ok, modify colilo (or your >> bootloader) and rewrite the sdram initialization. Usually you need to do >> several steps to get the sdram controller working (including writing data >> using the address lines) >> >> -- >> Carlos Manuel Duclos Vergara >> http://www.elbazaar.cl >> >> > From manjursk at flex-p.com Thu Mar 10 19:22:06 2005 From: manjursk at flex-p.com (Manjur S K) Date: Fri, 11 Mar 2005 08:22:06 +0800 Subject: [uClinux-dev] problem with NEC USB UHCI driver Message-ID: <003f01c525d0$5f61aff0$4601a8c0@flexpusera63> Hi All Good Morning, IXDP425 NAS-USB2 using Snapgear- 3.0 source distribution, IXP400 v1.3 software and arm- linux-tools-20030927.tar file, on the COYOTE board. I am facing a problem as the usb ehci is not working. Here is the error message during startup: Using /lib/modules/2.4.22- uc0/kernel/drivers/usb/host/ehci-hcd.o hcd.c: Found HC with no IRQ. Check BIOS/PCI 00:0f.2 setup! Is there any way where i can free the interrupt (where the interrupt is assigned for some devices but not being used) Tottally how to get the free interrupt Thank u in advance Regards Manjur -------------------------------------------------| Manjur S Kolhar Software Engineer Flex-p Industries SDN BHD UMNO Menara Penang Malaysia H/P:006-0164370430 ------------------------------------------------| -------------- next part -------------- An HTML attachment was scrubbed... URL: From gerg at snapgear.com Thu Mar 10 19:48:26 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Fri, 11 Mar 2005 10:48:26 +1000 Subject: [uClinux-dev] uclinux 2.6.10 fails on linking In-Reply-To: References: Message-ID: <4230EADA.4080109@snapgear.com> Hi Jate, Jate Sujjavanich wrote: > I am getting the following error when compiling uClinux 2.6.10 with GCC 3.4: > > LD init/built-in.o > LD vmlinux > arch/m68knommu/platform/5235/built-in.o(.text+0x2a): In function > `config_BSP': > arch/m68knommu/platform/5235/config.c:75: undefined reference to > `coldfire_pit_init' > arch/m68knommu/platform/5235/built-in.o(.text+0x36):arch/m68knommu/platform/5235/config.c:76: > undefined reference to `coldfire_pit_tick' > arch/m68knommu/platform/5235/built-in.o(.text+0x42):arch/m68knommu/platform/5235/config.c:77: > undefined reference to `coldfire_pit_offset' That looks like you need to change ~/linux-2.6.x/arch/m68knommu/platform/5307/Makefile. For the 5235 you obviously the pit timer code, and whether to include thios code or not is done in this Makefile. Look at the other examples that include "pit.o" and do similar. > mm/built-in.o(.text+0xc4e2): In function `shmem_getpage': > mm/shmem.c:921: undefined reference to `swapin_readahead' > mm/built-in.o(.text+0xc95a): In function `shmem_populate': > mm/shmem.c:1200: undefined reference to `install_page' > mm/built-in.o(.text+0xc9dc):mm/shmem.c:1206: undefined reference to > `install_file_pte' > mm/built-in.o(.text+0xca44): In function `shmem_lock': > mm/shmem.c:1244: undefined reference to `user_shm_lock' > mm/built-in.o(.text+0xca72):mm/shmem.c:1249: undefined reference to > `user_shm_unlock' > make[1]: *** [vmlinux] Error 1 > make[1]: Leaving directory `/home/jate/uClinux-2.6/linux-2.6.x' > make: *** [linux] Error 1 > I am not sure how to display the command line causing this error. See Heinz Haeberle's recent patch on this list do default SHMEM to disabled. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From rgetz at blackfin.uclinux.org Fri Mar 11 15:50:11 2005 From: rgetz at blackfin.uclinux.org (rgetz at blackfin.uclinux.org) Date: Fri, 11 Mar 2005 12:50:11 -0800 Subject: [uClinux-dev] debugging flat files Message-ID: <200503111250.11548.rgetz@blackfin.uclinux.org> Are there any tools for dumping flat files (like objdump -d?) Thanks -Robin From gerg at snapgear.com Thu Mar 10 23:44:31 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Fri, 11 Mar 2005 14:44:31 +1000 Subject: [uClinux-dev] NEC USB UHCI related problem on MS 7751R CP 01 In-Reply-To: <20050310120014.78893.qmail@web53204.mail.yahoo.com> References: <20050310120014.78893.qmail@web53204.mail.yahoo.com> Message-ID: <4231222F.1040900@snapgear.com> Hi Lara John, Lara John wrote: > I am trying to port embedded linux kernel (ver. > 2.4.26) on MS7751RCP01. > > This board has got NEC's usb 2.0 UHCI host controller. > > I am facing followoing problems. > > 1) This controller is detected by the kernel on boot > time, but it is probing the wrong IRQ. (The IRQ to IRL > is managed by the FPGA), the kernel should allocate > IRQ 9 but the messages says it is IRQ 10 The allocation of IRQ's is machine specific, you need to find the code for your machine in your kernel version. This being wrong is a possible (even likely) cause of your system hanging when you have a USB device on the bus. Regards Greg > The /proc/bus/usb/devices says: > T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 > Spd=12 MxCh= 1 > B: Alloc= 0/900 us ( 0%), #Int= 0, #Iso= 0 > D: Ver= 1.10 Cls=09(hub ) Sub=00 Prot=00 MxPS= 8 > #Cfgs= 1 > P: Vendor=0000 ProdID=0000 Rev= 0.00 > S: Product=USB OHCI Root Hub > S: SerialNumber=fd001000 > C:* #Ifs= 1 Cfg#= 1 Atr=40 MxPwr= 0mA > I: If#= 0 Alt= 0 #EPs= 1 Cls=09(hub ) Sub=00 Prot=00 > Driver=hub > E: Ad=81(I) Atr=03(Int.) MxPS= 2 Ivl=255ms > T: Bus=01 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 > Spd=12 MxCh= 1 > B: Alloc= 0/900 us ( 0%), #Int= 0, #Iso= 0 > D: Ver= 1.10 Cls=09(hub ) Sub=00 Prot=00 MxPS= 8 > #Cfgs= 1 > P: Vendor=0000 ProdID=0000 Rev= 0.00 > S: Product=USB OHCI Root Hub > S: SerialNumber=fd000000 > C:* #Ifs= 1 Cfg#= 1 Atr=40 MxPwr= 0mA > I: If#= 0 Alt= 0 #EPs= 1 Cls=09(hub ) Sub=00 Prot=00 > Driver=hub > E: Ad=81(I) Atr=03(Int.) MxPS= 2 Ivl=255ms > > 2) I am trying to mount SD Card (MMC) on USB with usb > reader. I have selected the build in option for SCSI > Storage and USB storage related. If i boot the kernel > without this device, it goes fine, but as soon as i > attatch the USB reader with SD card, it hangs (no > panic.. seems that it is in some ageless loop :( ) > > The last message it gave is as follows: > hub.c: new USB device 00:03.0-1, assigned address 2 > > > Please help me resolving this issue. If the solution > related to it is already resolved in newer releases > please let me know. > > Thanks in advance. > > Best regards, > Lara John > > PS: If any information is missing please let me know. > > > > __________________________________ > Do you Yahoo!? > Yahoo! Small Business - Try our new resources site! > http://smallbusiness.yahoo.com/resources/ > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org > -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Thu Mar 10 23:45:21 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Fri, 11 Mar 2005 14:45:21 +1000 Subject: [uClinux-dev] debugging flat files In-Reply-To: <200503111250.11548.rgetz@blackfin.uclinux.org> References: <200503111250.11548.rgetz@blackfin.uclinux.org> Message-ID: <42312261.3060202@snapgear.com> Hi Robin, rgetz at blackfin.uclinux.org wrote: > Are there any tools for dumping flat files (like objdump -d?) Usually I just the dump ELF file that the flat file came from. Rgeards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From wnlee at cad.csie.ncku.edu.tw Thu Mar 10 23:53:43 2005 From: wnlee at cad.csie.ncku.edu.tw (Wayne Lee) Date: Fri, 11 Mar 2005 12:53:43 +0800 Subject: [uClinux-dev] cramfs as root filesystem in uClinux Message-ID: <200503110456.j2B4uOfx016858@cad.csie.ncku.edu.tw> Hi, folks, Since cramfs has typical 50% compression ratio, I would like to use cramfs as root filesystem instead of romfs. Does someone else know the correct settings to use cramfs as root filesystem? What is the "root=" should we specified in kernel command line? Thanks. Wayne Produce cramfs image: $ mkcramfs -b 4096 romfs/ images/cramfs.img warning: gids truncated to 8 bits (this may be a security concern) $ ls -l images/romfs.img -rw-rw-r-- 1 wayne wayne 902144 Mar 11 12:23 images/romfs.img Kernel support for cramfs # # Miscellaneous filesystems # # CONFIG_ADFS_FS is not set # CONFIG_AFFS_FS is not set # CONFIG_HFS_FS is not set # CONFIG_HFSPLUS_FS is not set # CONFIG_BEFS_FS is not set # CONFIG_BFS_FS is not set # CONFIG_EFS_FS is not set CONFIG_CRAMFS=y # CONFIG_VXFS_FS is not set # CONFIG_HPFS_FS is not set # CONFIG_QNX4FS_FS is not set # CONFIG_SYSV_FS is not set # CONFIG_UFS_FS is not set From rgetz at blackfin.uclinux.org Fri Mar 11 16:11:00 2005 From: rgetz at blackfin.uclinux.org (rgetz at blackfin.uclinux.org) Date: Fri, 11 Mar 2005 13:11:00 -0800 Subject: [uClinux-dev] debugging flat files Message-ID: <200503111311.00486.rgetz@blackfin.uclinux.org> Greg wrote: > Usually I just the dump ELF file that the flat file came from. Yeah - we are just trying to figure out some issues in elf2flt - our architecture needs to have certain things memory aligned, and elf2flt seems to be messing up some C++ applications. (C is OK, and works well) Thanks -Robin From gerg at snapgear.com Fri Mar 11 00:16:52 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Fri, 11 Mar 2005 15:16:52 +1000 Subject: [uClinux-dev] [KVM] porting KVM on uClinux 2.4 In-Reply-To: <002f01c52477$00d4ee70$2d58718c@allen> References: <002f01c52477$00d4ee70$2d58718c@allen> Message-ID: <423129C4.7080901@snapgear.com> Hi, ??? wrote: > I try to port KVM on uClinux. I use uClinux-dist(not the newest version) > and it can choose 2.0 or 2.4 kernel. during compiling, I also meet the > locale support problem. I follow Erik's suggestion to replace the uClibc > version from 0.9.11 to 0.9.26 even 0.9.27. Why not use a newer version of uClinux-dist? You will get the newer uClibc for one thing. Regards Greg > In uClibc choosing list,I open the uClibc locale support and others,such > as pthread support,etc.However,when I make all of system, It shows that > preverify.elf2flt: In function `__uClibc_init': > preverify.elf2flt(.text+0x12354): undefined reference to > `__pthread_initialize_minimal' > collect2: ld returned 1 exit status > make[3]: *** [preverify] Error 1 > make[3]: Leaving directory > `/root/uClinux-dist/user/kvm/tools/preverifier/build/linux' > make[2]: *** [all] Error 1 > make[2]: Leaving directory `/root/uClinux-dist/user/kvm' > make[1]: *** [all] Error 2 > make[1]: Leaving directory `/root/uClinux-dist/user' > make: *** [subdirs] Error 1 > > Actually,when it compiles the boa and busybox,there are some problem to > appear.but I close it to compile and it will appear the above message.I > try to slove the problem.I choose the General position independent > code(PIC) of General libary seeting in uClibc. > [*] Generate Position Independent Code > (PIC) > [*] Enable support for shared libraries > (NEW) > [*] Compile native shared library loader > (NEW) > [ ] Only load shared libraries which can share their text > segment (NEW) > [*] Native shared library loader 'ldd' support > (NEW) > [*] Enable shared library loader cache > (NEW) > [ ] Enable shared library loader preload file support (NEW) > > OK, when I compile again, I got anther wrong meg. > > /usr/local/bin/arm-elf-ld.real: cannot open > /usr/local/lib/gcc-lib/arm-elf/2.95.3/fpic/msingle-pic-base/crtbeginS.o: > No such file or directory > > I try to copy the file and anther crtendS.o to the above path. I got > many errors... > > please help me > > Thanks and regards! > > Allen > > > > ------------------------------------------------------------------------ > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Fri Mar 11 00:21:09 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Fri, 11 Mar 2005 15:21:09 +1000 Subject: [uClinux-dev] where should we store ttb for MMU? In-Reply-To: <200503100317.j2A3HOfw023264@cad.csie.ncku.edu.tw> References: <200503100317.j2A3HOfw023264@cad.csie.ncku.edu.tw> Message-ID: <42312AC5.9060206@snapgear.com> Hi Wayne, Wayne Lee wrote: > In fact, I am now porting an uClinux no VM port on arm926ej-s CPU. > In order to turn-on D-cache, I use flat-mapping & section-mapped MMU and > then turn on D-cache. > The method is referred from Samsung s5c7375 port. > I store TTB in 0x4000. > However, while kernel is booting, the TTB is overwritten and data abort > occurred. Who overwrites it? What is the kernel trying to do when overwritting this area? Regards Greg > I am wonder where should we place the ttb in uClinux 2.6? > Do we need to modify the vmlinux.lds? > Please kindly give me your suggestions. > > Wayne > > /* > * Kernel startup entry point. > */ > __INIT > .type stext, #function > ENTRY(stext) > mov r12, r0 > mov r0, #PSR_F_BIT | PSR_I_BIT | MODE_SVC @ make sure svc mode > msr cpsr_c, r0 @ and all irqs disabled > > > > > /* > * ARM920T need MMU enable for D-cache. > * we just make the identical table before the MMU enabling. > */ > #ifdef CONFIG_CPU_MXU_ENABLE > @ /* Write domain id (cp15_r3) */ > @ mvn r0, #0 /* Domains 0, 1 = client */ > @ mcr p15, 0, r0, c3, c0, 0 /* load domain access register */ > @ /* Set control register v4 */ > > MOV r0, #0 > MCR p15, 0, r0, c7, c7, 0 @ invalidate I,D caches > MCR p15, 0, r0, c7, c10, 4 @ drain WB > MCR p15, 0, r0, c8, c7, 0 @ invalidate I & D TLBs > MRC p15, 0, r0, c1, c0, 0 @ ctrl register > BIC r0, r0, #0x000f @ ............wcam > BIC r0, r0, #0x1100 @ ...i...s........ > MCR p15, 0, r0, c1, c0, 0 @ ctrl register > > @ mov r0, #0 > @ mcr p15, 0, r0, c13, c0, 0 /* zero PID */ > @ mcr p15, 0, r0, c7, c7, 0 /* invalidate I,D caches */ > @ mcr p15, 0, r0, c7, c10, 4 /* drain write buffer */ > @ mcr p15, 0, r0, c8, c7, 0 /* invalidate I,D TLBs */ > @ mrc p15, 0, r0, c1, c0, 0 /* get control register */ > @ bic r0, r0, #0x000f @ ............wcam > @ bic r0, r0, #0x1100 @ ...i...s........ > @ @bic r0, r0, #0x0001 /* disable MMU */ > @ mcr p15, 0, r0, c1, c0, 0 /* write control register */ > > > init_ttb: > /* > * We must reserve DRAM_BASE ~ DRAM_BASE+0x8000. > * And Kernel TEXTADDR must be after DRAM_BASE + 0x8000 > */ > #if !defined(CONFIG_DRAM_BASE) || !defined(CONFIG_DRAM_SIZE) > #define CONFIG_DRAM_BASE 0x00000000 > #define CONFIG_DRAM_SIZE 0x04000000 > #endif > @ ldr r0,=(CONFIG_DRAM_BASE + 0x4000) /* set TTB (Must be on 16KB > boundary) */ > @ ldr r0,=(CONFIG_DRAM_BASE + CONFIG_DRAM_SIZE/2 + 0x4000) /* set > TTB (Must be on 16KB boundary) */ > @ ldr r0,=(CONFIG_DRAM_BASE + CONFIG_DRAM_SIZE - 0x4000) /* set > TTB (Must be on 16KB boundary) */ > ldr r0,=(CONFIG_DRAM_BASE + CONFIG_DRAM_SIZE) /* set TTB (Must > be on 16KB boundary) */ > mcr p15, 0, r0, c2, c0, 0 /* write to TTB(CP15,c2) */ > > /* > * > * Create translation table for flat mapping > * Top 12 bits of VA is pointer into table > * Create 4096 entries from 000xxxxx to fffxxxxx > * > * Set Section descriptor. 4K entry totla and > * each entry describe 1MB. > */ > mov r1, #0x1000 @ 4K counter > subs r1,r1,#1 > mov r2,#0b000000000000 @ set access permissions (AP) > for full access SVC/USR (11:10) > @ mov r2,#0b110000000000 @ set access permissions (AP) > for full access SVC/USR (11:10) > orr r2,r2,#0b000111100000 @ set for domain 15 (8:5) > orr r2,r2,#0b000000010000 @ must be 1 (4) > orr r2,r2,#0b000000000000 @ set non cachable non > bufferable (CB) (3:2) > orr r2,r2,#0b000000000010 @ set for 1Mb section (1:0) > init_ttb_1: /* create 4K entry */ > orr r3,r2,r1,lsl#20 @ use loop counter to create > individual table entries > str r3,[r0,r1,lsl#2] @ str r3 at TTB base + > loopcount*4 > subs r1,r1,#1 @ decrement loop counter > bpl init_ttb_1 > > /* resetting the SDRAM area to cacheable*/ > /* DRAM_BASE ~ DRAM_BASE+DRAM_SIZE */ > /* 64MB -> 64 entry -> 0x40 */ > mov r1, #(CONFIG_DRAM_SIZE >> 20 ) @ 64 entry =ase addr > subs r1, r1, #1 > > mov r2,#0b110000000000 @ set access permissions (AP) > for full access SVC/USR (11:10) > orr r2,r2,#0b000111100000 @ set for domain 15 (8:5) > orr r2,r2,#0b000000010000 @ must be 1 (4) > orr r2,r2,#0b000000001100 @ set cachable but unbufferable > (CB) (3:2) > @ orr r2,r2,#0b000000001000 @ set cachable but unbufferable > (CB) (3:2) > orr r2,r2,#0b000000000010 @ set for 1Mb section (1:0) > > init_ttb_2: /* create 64 entry (for 64MB SDRAM) */ > add r4, r1, #(CONFIG_DRAM_BASE >> 20 ) @ drambase >> 20 bit . > orr r3,r2,r4,lsl#20 @ use loop counter to create > individual table entries > str r3,[r0,r4,lsl#2] @ str r3 at TTB base + > loopcount*4 > subs r1,r1,#1 @ decrement loop counter > bpl init_ttb_2 > @init_domains : we define all domains are manager. so no access > permission check is occured. > mvn r0, #0 @ trick. 0 -1 == 0xFFFFFFFF > mcr p15, 0, r0, c3, c0, 0 @ write to CP15 register 5 > > /* > * set global core configurations > */ > mrc p15, 0, r0, c1, c0, 0 @ read CP15 register 1 into r0 > > orr r0, r0, #(0x1 <<12) @ enable I Cache > orr r0, r0, #(0x1 <<2) @ enable D Cache > orr r0, r0, #(0b11 <<30) @ enable asynchronous clocking > mode > orr r0, r0, #0x1 @ enable MMU > > mcr p15, 0, r0, c1, c0, 0 @ write cp15 register 1 > > /* > * set global core configurations > */ > mrc p15, 0, r0, c1, c0, 0 @ read CP15 register 1 into r0 > > > @#ifndef CONFIG_CPU_ICACHE_DISABLE > @ orr r0, r0, #(0x1 <<12) @ enable I Cache > @#endif > @#ifndef CONFIG_CPU_DCACHE_DISABLE > @ orr r0, r0, #(0x1 <<2) @ enable D Cache > @#endif > @#ifdef CONFIG_ALIGNMENT_TRAP > @ orr r0, r0, #2 @ ...........A. > @#endif > @ orr r0, r0, #(0b11 <<30) @ enable asynchronous clocking > mode > @ orr r0, r0, #0x1 @ enable MMU > > mcr p15, 0, r0, c1, c0, 0 @ write cp15 register 1 > nop > nop > nop > nop > nop > nop > nop > nop > nop > nop > nop > nop > nop > nop > nop > nop > nop > nop > nop > nop > > @ mrc p15, 0, r0, c1, c0, 0 @ read CP15 register 1 into r0 > @#ifndef CONFIG_CPU_ICACHE_DISABLE > @ orr r0, r0, #(0x1 <<12) @ enable I Cache > @#endif > @#ifndef CONFIG_CPU_DCACHE_DISABLE > @ orr r0, r0, #(0x1 <<2) @ enable D Cache > @#endif > @ mcr p15, 0, r0, c1, c0, 0 @ write CP15 register 1 > > > #endif > > > /* > * set global core configurations > */ > mrc p15, 0, r0, c1, c0, 0 @ read CP15 register 1 into r0 > > #ifndef CONFIG_CPU_ICACHE_DISABLE > orr r0, r0, #(0x1 <<12) @ enable I Cache > #endif > mcr p15, 0, r0, c1, c0, 0 @ write cp15 register 1 > > -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From davidm at snapgear.com Fri Mar 11 00:45:59 2005 From: davidm at snapgear.com (David McCullough) Date: Fri, 11 Mar 2005 15:45:59 +1000 Subject: [uClinux-dev] cramfs as root filesystem in uClinux In-Reply-To: <200503110456.j2B4uOfx016858@cad.csie.ncku.edu.tw> References: <200503110456.j2B4uOfx016858@cad.csie.ncku.edu.tw> Message-ID: <20050311054559.GA18679@beast> Jivin Wayne Lee lays it down ... > > Hi, folks, > > Since cramfs has typical 50% compression ratio, I would like to use cramfs > as root filesystem instead of romfs. > > Does someone else know the correct settings to use cramfs as root > filesystem? > What is the "root=" should we specified in kernel command line? Within uClinux cramfs will need to special root= line if you put the cramfs into the same device as your romfs, just do the same thing. Plenty of targets in the uClinux-dist use cramfs, do a: grep -l mkcramfs vendors/*/*/Makefile Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From davidm at snapgear.com Fri Mar 11 00:48:13 2005 From: davidm at snapgear.com (David McCullough) Date: Fri, 11 Mar 2005 15:48:13 +1000 Subject: [uClinux-dev] debugging flat files In-Reply-To: <200503111311.00486.rgetz@blackfin.uclinux.org> References: <200503111311.00486.rgetz@blackfin.uclinux.org> Message-ID: <20050311054813.GB18679@beast> Jivin rgetz at blackfin.uclinux.org lays it down ... > > Greg wrote: > > Usually I just the dump ELF file that the flat file came from. > > Yeah - we are just trying to figure out some issues in elf2flt - our > architecture needs to have certain things memory aligned, and elf2flt seems > to be messing up some C++ applications. (C is OK, and works well) Sounds more like problems in your linker script todo with CTORS/DTORS or possible exceptions. Looking at the elf file and the output from elf2flt -v should help, though there is a lot of info to wade through. FLTFLAGS += -v Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From wangjinfeng at gmail.com Fri Mar 11 01:39:28 2005 From: wangjinfeng at gmail.com (Wang JinFeng) Date: Fri, 11 Mar 2005 14:39:28 +0800 Subject: [uClinux-dev] undefined reference to Message-ID: kenel 2.4.20 uc0 wanna to make a new kernel to support nfs chose nfs support in config but get kernel compile down with debug message armksyms.c: At top level: armksyms.c:188: `__memzero' undeclared here (not in a function) armksyms.c:188: initializer element is not constant armksyms.c:188: (near initialization for `__ksymtab___memzero.value') make[2]: *** [armksyms.o] Error 1 make[2]: Leaving directory `/home/redhat/uClinux-dist/linux-2.4.x/arch/armnommu/kernel' make[1]: *** [_dir_arch/armnommu/kernel] Error 2 make[1]: Leaving directory `/home/redhat/uClinux-dist/linux-2.4.x' make: *** [linux] Error 1 need your help,thanks armksyms.c attached -- Company : R&D, ARIM Address : NO.72 Xisihuan South Road,Beijing,China 100071 Ph. (86) 010-63812255?3804 Fax (86) 010-63851580 Email : wangjinfeng at gmail.com -------------- next part -------------- A non-text attachment was scrubbed... Name: Administrator Type: application/octet-stream Size: 6889 bytes Desc: not available URL: From wnlee at cad.csie.ncku.edu.tw Fri Mar 11 01:47:37 2005 From: wnlee at cad.csie.ncku.edu.tw (Wayne Lee) Date: Fri, 11 Mar 2005 14:47:37 +0800 Subject: [uClinux-dev] cramfs as root filesystem in uClinux In-Reply-To: <20050311054559.GA18679@beast> Message-ID: <200503110650.j2B6oIfx018444@cad.csie.ncku.edu.tw> Thanks. Davidm :> I successfully mount cramfs.img as root filesystem with original kernel command line. The key is the mkcramfs utility. Before seeing your suggestion, I use Fedora Core 3 built-in mkcramfs utility. After using $(ROOTDIR)/user/cramfs/mkcramfs.c to compile mkcramfs, correct cramfs.img is built. Regards, Wayne -----Original Message----- From: David McCullough [mailto:davidm at snapgear.com] Sent: Friday, March 11, 2005 1:46 PM To: Wayne Lee Cc: uClinux development list Subject: Re: [uClinux-dev] cramfs as root filesystem in uClinux Jivin Wayne Lee lays it down ... > > Hi, folks, > > Since cramfs has typical 50% compression ratio, I would like to use cramfs > as root filesystem instead of romfs. > > Does someone else know the correct settings to use cramfs as root > filesystem? > What is the "root=" should we specified in kernel command line? Within uClinux cramfs will need to special root= line if you put the cramfs into the same device as your romfs, just do the same thing. Plenty of targets in the uClinux-dist use cramfs, do a: grep -l mkcramfs vendors/*/*/Makefile Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From gerg at snapgear.com Fri Mar 11 01:55:32 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Fri, 11 Mar 2005 16:55:32 +1000 Subject: [uClinux-dev] 2.6 kernel not booting on uCdimm In-Reply-To: <20050310090113.2941.qmail@web21605.mail.yahoo.com> References: <20050310090113.2941.qmail@web21605.mail.yahoo.com> Message-ID: <423140E4.3020000@snapgear.com> Hi Sriram, Sriram Chadalavada wrote: > My uCdimm boots normally (from ROM in all cases) when > a 2.4.24 kernel (20041215 distribution) is programmed > onto it. The 2.6 code is significantly different in most repsects. I don't use a uCdimm myself, so it is not a platform I test. Can anybody report recent 2.6 uClinux kernels working on uCdimm? > But when I program 2.6 kernel images (~ 1.6MB) onto > the uCdimm flash and reboot, only the following is > displayed on the serial console, after which the > uCdimm hangs. > > ======================================================= > B$ go > go! 0x10c10000 > ======================================================= > > I tried with 2.6.5 kernel of the distribution AND a > uCLinux patched 2.6.10 kernel. I also tried on a > different uCdimm. Same result. The kernel > configuration for the 2.6.10 kernel is attached. > > I would greatly appreciate if someone can point out > what is going on here. Please let me know if you need > more information. I would check through the startup code first, comparing with the 2.4 kernel - that should be similar at least. After that you may need to try sprinkling UART trace throughout the early code to find where it is failing. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Fri Mar 11 02:14:28 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Fri, 11 Mar 2005 17:14:28 +1000 Subject: [uClinux-dev] cramfs as root filesystem in uClinux In-Reply-To: <200503110650.j2B6oIfx018444@cad.csie.ncku.edu.tw> References: <200503110650.j2B6oIfx018444@cad.csie.ncku.edu.tw> Message-ID: <42314554.3030902@snapgear.com> Hi Wayne, Wayne Lee wrote: > Thanks. Davidm :> > > I successfully mount cramfs.img as root filesystem with original kernel > command line. > > The key is the mkcramfs utility. > Before seeing your suggestion, I use Fedora Core 3 built-in mkcramfs > utility. > After using $(ROOTDIR)/user/cramfs/mkcramfs.c to compile mkcramfs, correct > cramfs.img is built. Probably an endianess problem. The mkcramfs.c and CRAMfs code in the uClinux kernel have been modified to work on either little or big endian targets. Regards Greg > -----Original Message----- > From: David McCullough [mailto:davidm at snapgear.com] > Sent: Friday, March 11, 2005 1:46 PM > To: Wayne Lee > Cc: uClinux development list > Subject: Re: [uClinux-dev] cramfs as root filesystem in uClinux > > > Jivin Wayne Lee lays it down ... > >>Hi, folks, >> >>Since cramfs has typical 50% compression ratio, I would like to use cramfs >>as root filesystem instead of romfs. >> >>Does someone else know the correct settings to use cramfs as root >>filesystem? >>What is the "root=" should we specified in kernel command line? > > > Within uClinux cramfs will need to special root= line if you put the > cramfs into the same device as your romfs, just do the same thing. > > Plenty of targets in the uClinux-dist use cramfs, do a: > > grep -l mkcramfs vendors/*/*/Makefile > > Cheers, > Davidm > -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From manish.mishra at st.com Fri Mar 11 06:15:35 2005 From: manish.mishra at st.com (Manish Kumar MISHRA) Date: Fri, 11 Mar 2005 16:45:35 +0530 Subject: [uClinux-dev] problem compiling application having vfork and thread Message-ID: <001e01c5262b$a618f5b0$4d04b40a@dlh.st.com> I am trying to compile one application on uclinux platform (2.4 version) and this application has both vfork and posix threads. I am getting the following error after compilation upnpd.elf2flt(.text+0x1eed8): In function `__fork': : undefined reference to `__libc_fork' collect2: ld returned 1 exit status make[2]: *** [upnpd] Error 1 The code gets compiled if i remove either fork or thread from the code. It seems fork and threads cannot coexist in uclinux platform. Has anybody encountered such error during compilation? Please let me know if some fix is avaialbe for this problem? Thanx in advance, Manish From giovanni.casoli at atengineering.it Fri Mar 11 07:18:52 2005 From: giovanni.casoli at atengineering.it (Giovanni Casoli) Date: Fri, 11 Mar 2005 13:18:52 +0100 Subject: [uClinux-dev] Broken early mask Message-ID: <42318CAC.8010201@atengineering.it> Hi all, does someone know the meaning of the "broken early mask of 5272 silicon" mentioned in linux-2.[46].x/include/asm-m68knommu/system.h, before the various HARD_RESET_NOW() defines? A link to a FAQ or some other relevant document would also be fine. Thanks Giovanni Casoli From gerg at snapgear.com Fri Mar 11 09:26:03 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Sat, 12 Mar 2005 00:26:03 +1000 Subject: [uClinux-dev] Broken early mask In-Reply-To: <42318CAC.8010201@atengineering.it> References: <42318CAC.8010201@atengineering.it> Message-ID: <4231AA7B.4020002@snapgear.com> Hi Giovanni, Giovanni Casoli wrote: > does someone know the meaning of the "broken early mask of 5272 > silicon" mentioned in linux-2.[46].x/include/asm-m68knommu/system.h, > before the various HARD_RESET_NOW() defines? A link to a FAQ or some > other relevant document would also be fine. Your best chance is to look for the 5272 Errata. Check the Freescale web site. IIRC the problem was related to the address range that CS0 was mapped to on power up was very small (something like 4k). Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From mbeaulieu at eocycle.com Fri Mar 11 09:27:39 2005 From: mbeaulieu at eocycle.com (Martin Beaulieu) Date: Fri, 11 Mar 2005 09:27:39 -0500 Subject: [uClinux-dev] insmod modules References: <20050310201545.7EB7814DA29@mail07.powweb.com> Message-ID: <03d801c52646$7eebdc80$8000a8c0@MartinLogiciel> Hi, I made some change on my sdram controller initialization. After timing change, i can load a basic module with printk("Hello World");. But when i try to load rtai, i have the following message : /lib/modules/rtai> insmod rtai.o BINFMT_FLAT: ROM mapping of file (we hope) BINFMT_FLAT: Allocated data+bss+stack (24992 bytes): d0004 p=d7ffc start_thread(regs=0xddfcc, entry=0x18f2128, start_stack=0xd7f90) Using rtai.o?*** LINE 1111 *** FORMAT=4 Current process id is 0 BAD KERNEL TRAP: 00000000 PC: [<00000002>] SR: 2011 SP: 000ddf98 a2: 00000000 d0: 001d8000 d1: 001d8000 d2: 0000005a d3: 6ae9cced d4: 00000000 d5: 00006ca8 a0: 000f012a a1: 000dc000 Process (pid: 0, stackpage=000dd000) Frame format=4 Stack from 000ddfd4: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 Call Trace: Code: ffff ffff ffff 0180 dc18 0180 de40 0180 de40 0180 Kernel panic: Attempted to kill the idle task! In idle task - not syncing What is the problem ? Thanks for all Martin ----- Original Message ----- From: "Carlos Manuel Duclos Vergara" To: "Martin Beaulieu" Cc: Sent: Thursday, March 10, 2005 3:15 PM Subject: Re: [uClinux-dev] insmod modules > >I am not sur that the config of my sdram controller is good but it work. > > If I was you I would started looking into this kind of problems > >> make some test and active binfmt_flat trace. The result is : >>/> insmod uccrt >>BINFMT_FLAT: ROM mapping of file (we hope) >>BINFMT_FLAT: Allocated data+bss+stack (24367 bytes): d8004 >>p=dfffc >>start_thread(regs=0xcffcc, entry=0x18f2128, start_stack=0xdff90) >>insmod: /lib/modules/2.4.22-uc0: Nof sruceh efi_lef odr sdieretct oray r > > typical sdram timing problem, read as: > > "No such file or directory ...." > >>I have the same problem > > see??, there is a problem with your sdram controller initialization. Take > the datasheet of your sdram chips and write a small program to test your > ram. Then when you get the parameters ok, modify colilo (or your > bootloader) and rewrite the sdram initialization. Usually you need to do > several steps to get the sdram controller working (including writing data > using the address lines) > > -- > Carlos Manuel Duclos Vergara > http://www.elbazaar.cl > > From carlos at embedded.cl Fri Mar 11 10:54:02 2005 From: carlos at embedded.cl (Carlos Manuel Duclos Vergara) Date: Fri, 11 Mar 2005 15:54:02 -0000 Subject: [uClinux-dev] insmod modules Message-ID: <20050311155404.7809D14E5D4@mail07.powweb.com> An embedded and charset-unspecified text was scrubbed... Name: not available URL: From mbeaulieu at eocycle.com Fri Mar 11 11:25:20 2005 From: mbeaulieu at eocycle.com (Martin Beaulieu) Date: Fri, 11 Mar 2005 11:25:20 -0500 Subject: [uClinux-dev] insmod modules References: <20050311155404.7809D14E5D4@mail07.powweb.com> Message-ID: <03f701c52656$ec42f060$8000a8c0@MartinLogiciel> see below Martin ----- Original Message ----- From: "Carlos Manuel Duclos Vergara" To: "Martin Beaulieu" Cc: "uClinux development list" Sent: Friday, March 11, 2005 10:54 AM Subject: Re: [uClinux-dev] insmod modules > Hi, > > what sdram chips are you using? mt48lc4m16a2 of Micron > can you paste the initialization code you used? void mcf5407_sdramc_init1 (MCF5407_IMM *imm) { /******************************************************* * This routine is the first half of the SDRAM Controller * initialization. It sets up all of the configuration * information. We need time for eight refreshes * between this program and the second half of the init. *******************************************************/ unsigned junk = 0x9696; // 0xA5A5; /******************************************************** * DRAM Controller Refresh Register calculation: * * # of bus clocks = (RC + 1) * 16 * (703.83 bus clocks / 16) - 1 = RC * RC = 42.99 = 42 = 0x2A ********************************************************/ printk("Boot %d\n", __LINE__); /* Initialize DRAM Registers: DCR, DACR, DMR */ #if 1 MCF5407_WR_DRAMC_DCR(imm,0x8230); printk("Boot %d\n", __LINE__); MCF5407_WR_DRAMC_DACR0(imm,0x00002230); //34 printk("Boot %d\n", __LINE__); MCF5407_WR_DRAMC_DMR0(imm, 0x00FC0001); //007C printk("Boot %d\n", __LINE__); printk("Boot %d\n", __LINE__); MCF5407_WR_DRAMC_DACR1(imm,0x00802230); MCF5407_WR_DRAMC_DMR1(imm, 0x00FC0001); //007C /* Set IP (bit 3) in DACR */ MCF5407_WR_DRAMC_DACR0(imm,0x00002238); MCF5407_WR_DRAMC_DACR1(imm,0x00802238); printk("Boot %d\n", __LINE__); /* Write to each bank to initiate precharge */ *(uint16 *)0x00000000 = junk; *(uint16 *)0x00800000 = junk; /* Set RE (bit 15) in DACR */ MCF5407_WR_DRAMC_DACR0(imm,0x0000A230); MCF5407_WR_DRAMC_DACR1(imm,0x0080A230); #endif printk("Boot %d\n", __LINE__); } /****************************************************************/ void mcf5407_sdramc_init2 (MCF5407_IMM *imm) { /******************************************************** * This second half of the SDRAM Controller initialization * finishes the configuration. ********************************************************/ unsigned junk = 0x00; printk("Boot %d -P-\n", __LINE__); /* Finish the configuration by issuing the IMRS. */ #if 1 MCF5407_WR_DRAMC_DACR0(imm,0x0000A270); *(uint16 *)0x00001800 = junk; MCF5407_WR_DRAMC_DACR1(imm,0x0080A270); *(uint16 *)0x00801800 = junk; #endif // printf("Boot %d\n", __LINE__); /* Write to the SDRAM Mode Register (see UM pg. 11-34) */ printk("Boot %d\n", __LINE__); } > > sdram initialization is different from dram initialization, it is not > simply setting some parameters, there is a whole procedure to initialize > the state machine inside a sdram > > cheers > -- > Carlos Manuel Duclos Vergara > http://www.elbazaar.cl > > From mcnernbm at notes.udayton.edu Fri Mar 11 12:18:30 2005 From: mcnernbm at notes.udayton.edu (mcnernbm at notes.udayton.edu) Date: Fri, 11 Mar 2005 12:18:30 -0500 Subject: [uClinux-dev] memec micoblaze v2p7 Message-ID: Does anyone have a working hardware and software uclinx project for the memec virtex II pro board with the microblaze processor? Thanks Brett McNerney -------------- next part -------------- An HTML attachment was scrubbed... URL: From swenggsri at yahoo.com Fri Mar 11 12:35:54 2005 From: swenggsri at yahoo.com (Sriram Chadalavada) Date: Fri, 11 Mar 2005 09:35:54 -0800 (PST) Subject: [uClinux-dev] 2.6 kernel not booting on uCdimm In-Reply-To: 6667 Message-ID: <20050311173555.77993.qmail@web21602.mail.yahoo.com> Hi Greg, Thanks for the reply. > The 2.6 code is significantly different in most > repsects. I don't use a uCdimm myself, so it is not > a platform I test. > > Can anybody report recent 2.6 uClinux kernels > working > on uCdimm? My interest is primarily is in getting the uCdimm serial port driver to work. I am bringing up a KISS AX.25 network interface using a ported program. But when I try to ping to this interface from a linux box, the kernel hangs. Kernel panic Interrupt handler - not syncing. This however does not happen when I perform the test on an x86 box. The mkiss driver (drivers/net/hamradio/mkiss.c) seems to be the same on my linux box and the 2.4.24 kernel on uCdimm. The difference in behavior I assumed may have resulted from the uCdimm serial port driver. This driver code has been updated for 2.6. If I know what improvements/bug-fixes have been made to the uCdimm serial port driver, perhaps I can modify the 2.4.24 driver code itself rather than going all the way to move to 2.6. I would appreciate if you can tell me who to contact about that. > I would check through the startup code first, > comparing > with the 2.4 kernel - that should be similar at > least. > > After that you may need to try sprinkling UART trace > throughout the early code to find where it is > failing. > Thanks. I will try that out in the meantime. Cheers, Sriram __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/ From carlos at embedded.cl Fri Mar 11 13:16:17 2005 From: carlos at embedded.cl (Carlos Manuel Duclos Vergara) Date: Fri, 11 Mar 2005 18:16:17 -0000 Subject: [uClinux-dev] insmod modules Message-ID: <20050311181619.B056D14EBE4@mail07.powweb.com> An embedded and charset-unspecified text was scrubbed... Name: not available URL: From kapheine at gmail.com Fri Mar 11 14:50:57 2005 From: kapheine at gmail.com (Zachary Landau) Date: Fri, 11 Mar 2005 14:50:57 -0500 Subject: [uClinux-dev] 2.4 qspi driver modifications - some concerns In-Reply-To: <001c01c50f46$ca2d2210$07321e0a@telex.de> References: <200502091311.21071.ngustavson@emacinc.com> <001c01c50f46$ca2d2210$07321e0a@telex.de> Message-ID: > > First, let me say this is a very nice clean driver. > > And I do like it's organization. > > > > However, I do have some concerns/suggestions. > > > > First off, I don't see any mention of CONFIG_M5282 in the new > > driver, which > > was definitely supported in the original driver. > > This code appears to be a massive re-write of the original,not just a > > modification) did all the processor support carry over? > > The driver is based on an older version of the mcf_qspi driver. At this > time there was no support for mcf5282. I think there are only a few > lines > of code to add. Attached is a patch (that should be applied on top of the uclinux-dist-20041215-new-spi-driver.patch that adds 5282 support back in. I just copied the required code from the current mcf_qspi.c driver. -- Zachary P. Landau -------------- next part -------------- A non-text attachment was scrubbed... Name: uclinux-dist-20041215-new-spi-5282.patch Type: text/x-patch Size: 2466 bytes Desc: not available URL: From Herbert.Yeung at uts.edu.au Fri Mar 11 16:50:59 2005 From: Herbert.Yeung at uts.edu.au (Herbert) Date: Sat, 12 Mar 2005 08:50:59 +1100 (EST) Subject: [uClinux-dev] Re: Force compiling libfl.a for Cobra5272 Message-ID: Hi, I am having trouble trying to compile flex (specifically the flex library libfl.a) for uClinux kernel 2.4.x using the uClibc library. I am getting the following error: make[3]: *** No rule to make target `../MISC/alloca.c', needed by `alloca.c'. Stop. I have had a look at the archives and it seems that this does not work on some chipsets, as in it has not been ported correctly yet. I was wondering though whether anyone had it working on the Motorola Cobra5272. I have tried to manually add alloca.c and libfl.a compiles but the kernel does not build correctly. Does anyone know of anyway to get both libfl.a compiled as well as the uClinux kernel 2.4.x working??? Any help would be of great assistance, Herb From gwilla at essex.ac.uk Fri Mar 11 18:46:54 2005 From: gwilla at essex.ac.uk (Greg) Date: Fri, 11 Mar 2005 23:46:54 -0000 Subject: [uClinux-dev] libupnp problem Message-ID: <20050311234650.603F02BC030@smtp.nildram.co.uk> Hi, I am having difficulty getting libupnp (which comes with uClinux-dist-20041215.tar.gz) to compile for the Arcturus uCdimm. Has anyone had it working? Or have any idea what needs to be changed to make it compile properly? Any help very much appreciated! Thanks, Greg -------------- next part -------------- An HTML attachment was scrubbed... URL: From devi at tenet.res.in Sat Mar 12 11:20:47 2005 From: devi at tenet.res.in (devi) Date: Sat, 12 Mar 2005 13:20:47 -0300 Subject: [uClinux-dev] framebuffer Message-ID: <20050312162047.13345.qmail@mail.tenet.res.in> hi, How to configure the framebuffer console.Since I need the virtual console for the keyboard. Can anyone help me to configure it. Regards Devi From maverick2004 at 21cn.com Sat Mar 12 02:47:52 2005 From: maverick2004 at 21cn.com (Maverick Gao) Date: Sat, 12 Mar 2005 15:47:52 +0800 Subject: [uClinux-dev] uClinux memory problem Message-ID: <000f01c526d7$d0881310$4400a8c0@gaojunbin> hi, all I'm working with an uClinux 2.4.20 platform (with a flash disk) now. it is my test code below: int main() { FILE * f; static char buffer[102400]; int i; static char filename[20]; for (i = 1;i < 21;++ i) { sprintf(filename,"/sd/sda/%d",i); // sd/sda is a flash disk if (NULL == (f = fopen(filename,"wb"))) { printf("err open file %s\n",filename); break; } fwrite(buffer,1,102400,f); fwrite(buffer,1,102400,f); fwrite(buffer,1,102400,f); fwrite(buffer,1,102400,f); fwrite(buffer,1,102400,f); fwrite(buffer,1,102400,f); fwrite(buffer,1,102400,f); fwrite(buffer,1,102400,f); fclose(f); printf("write file %s ok\n",filename); system("free"); } return 0; } and output: write file /sd/sda/1 ok total: used: free: shared: buffers: cached: Mem: 5951488 2691072 3260416 0 90112 880640 Swap: 0 0 0 write file /sd/sda/2 ok total: used: free: shared: buffers: cached: Mem: 5951488 3690496 2260992 0 106496 1699840 Swap: 0 0 0 write file /sd/sda/3 ok total: used: free: shared: buffers: cached: Mem: 5951488 4689920 1261568 0 114688 2519040 Swap: 0 0 0 write file /sd/sda/4 ok total: used: free: shared: buffers: cached: Mem: 5951488 5693440 258048 0 131072 3338240 Swap: 0 0 0 write file /sd/sda/5 ok total: used: free: shared: buffers: cached: Mem: 5951488 5767168 184320 0 139264 3371008 Swap: 0 0 0 write file /sd/sda/6 ok total: used: free: shared: buffers: cached: Mem: 5951488 5693440 258048 0 155648 3272704 Swap: 0 0 0 write file /sd/sda/7 ok total: used: free: shared: buffers: cached: Mem: 5951488 5734400 217088 0 163840 3305472 Swap: 0 0 0 write file /sd/sda/8 ok Allocation of length 76909 from process 20 failed Buffer memory: 176kB Cache memory: 3132kB Free pages: 456kB ( 0kB HighMem) Zone:DMA freepages: 0kB min: 0kB low: 0kB high: 0kB Zone:Normal freepages: 456kB min: 80kB low: 160kB high: 240kB Zone:HighMem freepages: 0kB min: 0kB low: 0kB high: 0kB ( Active: 52, inactive: 775, free: 114 ) = 0kB) 20*4kB 13*8kB 9*16kB 2*32kB 1*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB = 4 56kB) = 0kB) Unable to allocate RAM for process text/data, errno 12 why, memory fragment or filesystem cache ? how can I do to avoid it ? Best Regards. Maverick Gao -------------- next part -------------- An HTML attachment was scrubbed... URL: From bob_news2004 at 163.com Sun Mar 13 06:03:00 2005 From: bob_news2004 at 163.com (bob_news2004) Date: Sun, 13 Mar 2005 19:03:00 +0800 Subject: [uClinux-dev] uClinux memory problem References: <000f01c526d7$d0881310$4400a8c0@gaojunbin> Message-ID: <004001c527bc$390f3500$520b1fac@bobzhang01> try ----- Original Message ----- From: Maverick Gao To: uclinux-dev at uClinux.org Sent: Saturday, March 12, 2005 3:47 PM Subject: [uClinux-dev] uClinux memory problem hi, all I'm working with an uClinux 2.4.20 platform (with a flash disk) now. it is my test code below: int main() { FILE * f; static char buffer[102400]; int i; static char filename[20]; for (i = 1;i < 21;++ i) { sprintf(filename,"/sd/sda/%d",i); // sd/sda is a flash disk if (NULL == (f = fopen(filename,"wb"))) { printf("err open file %s\n",filename); break; } fwrite(buffer,1,102400,f); fwrite(buffer,1,102400,f); fwrite(buffer,1,102400,f); fwrite(buffer,1,102400,f); fwrite(buffer,1,102400,f); fwrite(buffer,1,102400,f); fwrite(buffer,1,102400,f); fwrite(buffer,1,102400,f); fclose(f); printf("write file %s ok\n",filename); system("free"); } return 0; } and output: write file /sd/sda/1 ok total: used: free: shared: buffers: cached: Mem: 5951488 2691072 3260416 0 90112 880640 Swap: 0 0 0 write file /sd/sda/2 ok total: used: free: shared: buffers: cached: Mem: 5951488 3690496 2260992 0 106496 1699840 Swap: 0 0 0 write file /sd/sda/3 ok total: used: free: shared: buffers: cached: Mem: 5951488 4689920 1261568 0 114688 2519040 Swap: 0 0 0 write file /sd/sda/4 ok total: used: free: shared: buffers: cached: Mem: 5951488 5693440 258048 0 131072 3338240 Swap: 0 0 0 write file /sd/sda/5 ok total: used: free: shared: buffers: cached: Mem: 5951488 5767168 184320 0 139264 3371008 Swap: 0 0 0 write file /sd/sda/6 ok total: used: free: shared: buffers: cached: Mem: 5951488 5693440 258048 0 155648 3272704 Swap: 0 0 0 write file /sd/sda/7 ok total: used: free: shared: buffers: cached: Mem: 5951488 5734400 217088 0 163840 3305472 Swap: 0 0 0 write file /sd/sda/8 ok Allocation of length 76909 from process 20 failed Buffer memory: 176kB Cache memory: 3132kB Free pages: 456kB ( 0kB HighMem) Zone:DMA freepages: 0kB min: 0kB low: 0kB high: 0kB Zone:Normal freepages: 456kB min: 80kB low: 160kB high: 240kB Zone:HighMem freepages: 0kB min: 0kB low: 0kB high: 0kB ( Active: 52, inactive: 775, free: 114 ) = 0kB) 20*4kB 13*8kB 9*16kB 2*32kB 1*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB = 4 56kB) = 0kB) Unable to allocate RAM for process text/data, errno 12 why, memory fragment or filesystem cache ? how can I do to avoid it ? Best Regards. Maverick Gao ------------------------------------------------------------------------------ _______________________________________________ uClinux-dev mailing list uClinux-dev at uClinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev at uclinux.org -------------- next part -------------- An HTML attachment was scrubbed... URL: From saiful at modular-corp.com Sun Mar 13 20:26:05 2005 From: saiful at modular-corp.com (saiful) Date: Mon, 14 Mar 2005 09:26:05 +0800 Subject: [uClinux-dev] compiling error Message-ID: <200503140926.AA46597072@modular-corp.com> Hi, For some reasons, selecting some functions in Make Menuconfig produced this warning message while compiling: /snapgear/glibc/build/ncurses/build-tic /home/uclinux/snapgear/glibc/build/libc.so.6:version 'GLIBC_2.3' not found (required by /usr/lib/libncurses.so.5) when I did ldd: ldd libc.so.6 /libc.so.6: /lib/ld-linux.so.2: version 'GLIBC_2.1.1' not found /libc.so.6: /lib/ld-linux.so.2: version 'GLIBC_2.2.3' not found /libc.so.6: /lib/ld-linux.so.2: version 'GLIBC_2.2' not found I wonder why /libc.so.6 looked for other version of glibc and not 'GLIBC-2.3'? and why ../../snapgear/glibc/build/libc.so.6 requires version 'GLIBC_2.3' ? FYI, I am using Redhat 9 and supposed to have GLIBC_2.3. I don't understand why this can happen. Doing 'ldconfig' did not solve my problem actually Saiful From swenggsri at yahoo.com Sun Mar 13 22:14:13 2005 From: swenggsri at yahoo.com (Sriram Chadalavada) Date: Sun, 13 Mar 2005 19:14:13 -0800 (PST) Subject: [uClinux-dev] 2.6 kernel not booting on uCdimm In-Reply-To: 6667 Message-ID: <20050314031413.64284.qmail@web21603.mail.yahoo.com> Hi Greg, The last two days were a good education on the startup of uClinux on ucdimm(MC68VZ328) :). Please let me know if there is a Changelog for the 68328serial.c (MC68VZ328 serial driver). As I mentioned earlier, my immediate concern is to get the kiss AX.25 packets tx/rx working without a kernel panic. And I want to know if there have been any bug-fixes/critical improvements in 2.6 version of the driver from the 2.4.24 version, so that I can make changes to the 2.4.24 code itself. > I would check through the startup code first, > comparing > with the 2.4 kernel - that should be similar at > least. > > After that you may need to try sprinkling UART trace > throughout the early code to find where it is > failing. These were my observations: 1. The uClinux patches(suffixed with uc0) for kernels >2.6.3 are getting rid of the startup files in the arch/m68knommu/platform/68VZ328/ucdimm directory (crt0_*.S). I found this behavior in both 2.6.5 and 2.6.10. I don't know if the startup mechanism has changed for these kernels because I could not find these files elsewhere in the source tree. Is this ok? Hence, I worked mostly with 2.6.3 uClinux patch and compared it with my working 2.4.24 kernel. 2. Where are the rom.ld, himem.ld and ram.ld files located in 2.6.x? Again, these were neither supplied by the vanilla kernel nor by the corresponding uclinux patch. Please let me know if the use of these files has been superseded by some other method in 2.6. Because, even without those files in that directory, the start address of the kernel was the same as that for 2.4.24 kernel = 0x10c10400. Disassembly of section .text(2.6.3): 10c10400 <_start>: 10c10400: 3e3c 0050 movew #80,%d7 3. I ran the above 2.6.3 kernel on two uCdimms. Kernel uCBootloader version Result --------------------------------------------- 2.6.3 1.5.9(boots from 0x10c10400) NO difference 2.6.3 1.7.6 Same as above 2.4.24 1.5.9 Successful boot 2.4.24 1.7.6 Same as above (boots from 0x10c10000??Maybe not) I got an email from another ucdimm-2.6 user who encountered exactly the same problems as I did. Please let me know if the behavior described above(lack of startup files) is causing the boot problems and if there are patches available to rectify the same. Thanks, Sriram __________________________________ Do you Yahoo!? Yahoo! Mail - Easier than ever with enhanced search. Learn more. http://info.mail.yahoo.com/mail_250 From mcnernbm at notes.udayton.edu Sun Mar 13 22:47:55 2005 From: mcnernbm at notes.udayton.edu (mcnernbm at notes.udayton.edu) Date: Sun, 13 Mar 2005 22:47:55 -0500 Subject: [uClinux-dev] dhrystone Message-ID: When attepmting to compile the source files for the dhrystone benchmarks supplied in the uclinux-distribution under EDK 6.3 I recive the following error. Any one know how to fix this so I can get the benchmarks to run on a microbalze and the powerpc on a virtex II pro with no OS so I can compare some different microblaze and ppc configurations. /xygdrive/c/DOCUME~1/mac1321/LOCALS~1/Temp/ccOH335B.o: In function `main': /xygdrive/c/DOCUME~1/mac1321/LOCALS~1/Temp/ccOH335B.o(.text+0x1b4): undefined reference to `times' /xygdrive/c/DOCUME~1/mac1321/LOCALS~1/Temp/ccOH335B.o(.text+0x3e8): undefined reference to `times' the code is directly from the uclinux-dist with no changes made yet. When times is commented out it runs but gives no results. Thanks Brett -------------- next part -------------- An HTML attachment was scrubbed... URL: From davidm at snapgear.com Sun Mar 13 23:01:26 2005 From: davidm at snapgear.com (David McCullough) Date: Mon, 14 Mar 2005 14:01:26 +1000 Subject: [uClinux-dev] dhrystone In-Reply-To: References: Message-ID: <20050314040126.GA8284@beast> Jivin mcnernbm at notes.udayton.edu lays it down ... > When attepmting to compile the source files for the dhrystone benchmarks > supplied in the uclinux-distribution under EDK 6.3 I recive the following > error. Any one know how to fix this so I can get the benchmarks to run on > a microbalze and the powerpc on a virtex II pro with no OS so I can > compare some different microblaze and ppc configurations. > > /xygdrive/c/DOCUME~1/mac1321/LOCALS~1/Temp/ccOH335B.o: In function `main': > /xygdrive/c/DOCUME~1/mac1321/LOCALS~1/Temp/ccOH335B.o(.text+0x1b4): > undefined reference to `times' > /xygdrive/c/DOCUME~1/mac1321/LOCALS~1/Temp/ccOH335B.o(.text+0x3e8): > undefined reference to `times' > > the code is directly from the uclinux-dist with no changes made yet. When > times is commented out it runs but gives no results. times is a system call (man times). It should be there in your libc. It built fine for me here on a SuperH, I suspect it will for most archs. Double check your link line I guess, Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From gerg at snapgear.com Sun Mar 13 23:59:37 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 14 Mar 2005 14:59:37 +1000 Subject: [uClinux-dev] 2.6 kernel not booting on uCdimm In-Reply-To: <20050314031413.64284.qmail@web21603.mail.yahoo.com> References: <20050314031413.64284.qmail@web21603.mail.yahoo.com> Message-ID: <42351A39.6080201@snapgear.com> Hi Sriram, Sriram Chadalavada wrote: > The last two days were a good education on the > startup of uClinux on ucdimm(MC68VZ328) :). > > Please let me know if there is a Changelog for the > 68328serial.c (MC68VZ328 serial driver). As I I have my local CVS log, but there has really been very little work done on this driver in all of the 2.6 kernels. Georges Menie did some cleanups in 2.5 series for this. I don't think many people are using 2.6 on 68328 platforms. (And I say that because I get very few patches for fixes or changes to code specific to them). > These were my observations: > > 1. The uClinux patches(suffixed with uc0) for kernels > >>2.6.3 are getting rid of the startup files in the > > arch/m68knommu/platform/68VZ328/ucdimm directory > (crt0_*.S). I found this behavior in both 2.6.5 and > 2.6.10. I don't know if the startup mechanism has > changed for these kernels because I could not find > these files elsewhere in the source tree. Is this ok? > > Hence, I worked mostly with 2.6.3 uClinux patch and > compared it with my working 2.4.24 kernel. In recent 2.6 kernels I have been rationalizing the start up code. Instead of it being specific to each board, I have made it common to all boards with the same CPU family (ie same for all ColdFire CPU's, same for all 68328 CPU's, etc). > 2. Where are the rom.ld, himem.ld and ram.ld files > located in 2.6.x? Again, these were neither supplied > by the vanilla kernel nor by the corresponding uclinux > patch. Please let me know if the use of these files > has been superseded by some other method in 2.6. > Because, even without those files in that directory, > the start address of the kernel was the same as that > for 2.4.24 kernel = 0x10c10400. There is a single linker script for all m68knommu targets: ~/arch/m68knommu/kernel/vmlinux.lds.S. All of the functionality of each of the older files should be catered for in this script. > Disassembly of section .text(2.6.3): > > 10c10400 <_start>: > 10c10400: 3e3c 0050 movew #80,%d7 > > 3. I ran the above 2.6.3 kernel on two uCdimms. > > Kernel uCBootloader version Result > --------------------------------------------- > 2.6.3 1.5.9(boots from 0x10c10400) NO difference > > 2.6.3 1.7.6 Same as above > > 2.4.24 1.5.9 Successful boot > > 2.4.24 1.7.6 Same as above > > (boots from 0x10c10000??Maybe not) > > I got an email from another ucdimm-2.6 user who > encountered exactly the same problems as I did. Please > let me know if the behavior described above(lack of > startup files) is causing the boot problems and if > there are patches available to rectify the same. It shoulnd't be. But then I cannot directly test uCdimm, I don't have one. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Mon Mar 14 00:05:54 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 14 Mar 2005 15:05:54 +1000 Subject: [uClinux-dev] init/Kconfig SHM error In-Reply-To: References: Message-ID: <42351BB2.40807@snapgear.com> Hi Heinz, Heinz Haeberle wrote: > I have build the 2.6.10-uc0 kernel for mcf5249C3 > While doing so I figured the config option SHMEM was set and there wasn't > even a way to reset it > I guess the error is in init/Kconfig > > I have added the following lines to > > config SHMEM > .... > + depends on MMU > .... > > and > > config TINY_SHMEM > default !SHMEM > + depends on MMU > > > The result will be that there is no SHMEM at all for this target and IMHO > this is ok In 2.6.11 you should be able to use TINY_SHMEM now. At least in my test builds, and the 2.6.11-uc0 patches, it seems to at least compile :-) Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From devi at tenet.res.in Mon Mar 14 11:43:49 2005 From: devi at tenet.res.in (devi) Date: Mon, 14 Mar 2005 13:43:49 -0300 Subject: [uClinux-dev] console Message-ID: <20050314164349.30352.qmail@mail.tenet.res.in> hi, how do I attach a console to my /dev/fb0.I think I need to use fbcon.c.I am getting a print statement Console: switching to frame buffer device while booting.But I don't get any text on my frame buffer device when I type the keyboard. Regards Devi From gerg at snapgear.com Mon Mar 14 00:45:13 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 14 Mar 2005 15:45:13 +1000 Subject: [uClinux-dev] linux-2.6.11-uc0 patch set released Message-ID: <423524E9.2050803@snapgear.com> Hi All, An update of the uClinux (MMU-less) fixups against 2.6.11. Most new changes center around the recent nommu changes to keep the mm list as a vma list. Still a bunch of old changes I need to push up stream in this patch too. http://www.uclinux.org/pub/uClinux/uClinux-2.6.x/linux-2.6.11-uc0.patch.gz Change log: . import of linux-2.6.11 . change vma list setup for nommu . fix MAGIC_ROMPTR nommu support . remove unused semp3.h Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Mon Mar 14 00:58:19 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 14 Mar 2005 15:58:19 +1000 Subject: [uClinux-dev] [PATCH]uClinux patch for Philips LPC22xx In-Reply-To: References: Message-ID: <423527FB.7090703@snapgear.com> Hi Lucy, lucy.wang at philips.com wrote: > Here is the new patch. > - Modify the Chinese comments to English. > - Delete some unused files and modify the file structure a little. I have merged the vendors/Philips directory into the uClinux-dist tree. The uClinux-dist does not contain boot laoders, so I cannot include that. Also the config.in is auto-generated when you first do a "make xconfig" based on the contents of the vendors directory, so no need to send changes to or that whole file. Regards Greg > > > > *"Friedrich, Lars" * > > 2005-03-09 16:00 > > > To: Lucy Wang/PEK/SC/PHILIPS at PHILIPS > cc: > Subject: RE: [uClinux-dev] [PATCH]uClinux patch for > Philips LPC22xx > > Classification: > > > > > > Honestly speaking, I can't understand what you mean. > > I didn't make any change to linux-2.4.x in the Philips LPC22xx patches. > > Where do you see about the IIRC ALIGN (16)? > > Urks.. messed this up with a different patch I had been reviewing at the > same time. You are officially declared innocent ;-) > > > Where do you read such non-ascii characters? > > uClinux-Philips-LPC22xx.tar > i.e: > ---- startup.S > ;?????????????????????? > IMPORT main ;C?????????????? > IMPORT TargetResetInit ;???????????????? > > CODE32 > > AREA vectors,CODE,READONLY > ENTRY > ;?????????? > Reset > [...] > ;???????????????????????????????????????? > > LDR R0, =PINSEL0 > LDR R1, =0x80000005 > ;ext2 for ethernet, uart0 > STR R1, [R0] > > [...] > ---- target.c > #include "config.h" > /********************************************************************************************************* > ** ????????: TargetResetInit > ** ????????: ????main?????????????????????????????????????????????? > ********************************************************************************************************/ > void TargetResetInit(void) > { > MEMMAP = 0x3; //remap to external memory > > /* ?????????????????? */ > PLLCON = 1; > [...] > ---- config.h > #include "lpc2294.h" > > /* ????????, Fosc??Fcclk??Fcco??Fpclk????????*/ > #define Fosc 10000000 //????????,10MHz > #define Fcclk (Fosc * 4) //??????????40MHz > #define Fcco (Fcclk * 4) > //CCO????????????Fcclk??2??4??8??16??????????156MHz~320MHz > #define Fpclk (Fcclk / 4) * 1 > //VPB????????????(Fcclk / 4) > > > > The linux patch file looks okay, just the uClinux one seems a bit messed > up in the comments. > > Best regards, > Lars Friedrich > > > > ------------------------------------------------------------------------ > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From lucy.wang at philips.com Mon Mar 14 01:03:38 2005 From: lucy.wang at philips.com (lucy.wang at philips.com) Date: Mon, 14 Mar 2005 14:03:38 +0800 Subject: [uClinux-dev] [PATCH]uClinux patch for Philips LPC22xx Message-ID: Thank you, Greg! Greg Ungerer 2005-03-14 13:58 To: Lucy Wang/PEK/SC/PHILIPS at PHILIPS cc: "Friedrich, Lars" uclinux-dev at uClinux.org Subject: Re: [uClinux-dev] [PATCH]uClinux patch for Philips LPC22xx Classification: Hi Lucy, lucy.wang at philips.com wrote: > Here is the new patch. > - Modify the Chinese comments to English. > - Delete some unused files and modify the file structure a little. I have merged the vendors/Philips directory into the uClinux-dist tree. The uClinux-dist does not contain boot laoders, so I cannot include that. Also the config.in is auto-generated when you first do a "make xconfig" based on the contents of the vendors directory, so no need to send changes to or that whole file. Regards Greg > > > > *"Friedrich, Lars" * > > 2005-03-09 16:00 > > > To: Lucy Wang/PEK/SC/PHILIPS at PHILIPS > cc: > Subject: RE: [uClinux-dev] [PATCH]uClinux patch for > Philips LPC22xx > > Classification: > > > > > > Honestly speaking, I can't understand what you mean. > > I didn't make any change to linux-2.4.x in the Philips LPC22xx patches. > > Where do you see about the IIRC ALIGN (16)? > > Urks.. messed this up with a different patch I had been reviewing at the > same time. You are officially declared innocent ;-) > > > Where do you read such non-ascii characters? > > uClinux-Philips-LPC22xx.tar > i.e: > ---- startup.S > ;?????????????????????? > IMPORT main ;C?????????????? > IMPORT TargetResetInit ;???????????????? > > CODE32 > > AREA vectors,CODE,READONLY > ENTRY > ;?????????? > Reset > [...] > ;???????????????????????????????????????? > > LDR R0, =PINSEL0 > LDR R1, =0x80000005 > ;ext2 for ethernet, uart0 > STR R1, [R0] > > [...] > ---- target.c > #include "config.h" > /********************************************************************************************************* > ** ????????: TargetResetInit > ** ????????: ????main?????????????????????????????????????????????? > ********************************************************************************************************/ > void TargetResetInit(void) > { > MEMMAP = 0x3; //remap to external memory > > /* ?????????????????? */ > PLLCON = 1; > [...] > ---- config.h > #include "lpc2294.h" > > /* ????????, Fosc??Fcclk??Fcco??Fpclk????????*/ > #define Fosc 10000000 //????????,10MHz > #define Fcclk (Fosc * 4) //??????????40MHz > #define Fcco (Fcclk * 4) > //CCO????????????Fcclk??2??4??8??16??????????156MHz~320MHz > #define Fpclk (Fcclk / 4) * 1 > //VPB????????????(Fcclk / 4) > > > > The linux patch file looks okay, just the uClinux one seems a bit messed > up in the comments. > > Best regards, > Lars Friedrich > > > > ------------------------------------------------------------------------ > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com -------------- next part -------------- An HTML attachment was scrubbed... URL: From mailtomax at 163.com Mon Mar 14 01:47:09 2005 From: mailtomax at 163.com (=?gb2312?B?1tzX1M6w?=) Date: Mon, 14 Mar 2005 14:47:09 +0800 Subject: [uClinux-dev] help me! Message-ID: <20050314064446.73E0E1A0D0@uclinux.org> hi all: I download a uclibc version uClibc-0.9.27 and arm-elf-tools-.sh form uclibc.org, then I compile a simple code,but it take some errors, it makes me crazy, could you help me? thanks thanks thanks! arm-elf-gcc -Os -fno-builtin -nostartfiles -Wl,-elf2flt /usr/local/arm-elf/lib/crt0.o libtest.o -lgcc -lpthread -lc -L/usr/local/arm-elf/lib -L. -o libtest arm-elf-gcc libtest.o -lgcc -lpthread -lc -L/usr/local/arm-elf/lib -L. -o libtest -fno-builtin /usr/local/lib/gcc-lib/arm-elf/2.95.3/libgcc.a(__main.o): In function `__do_global_ctors': __main.o(.text+0xc4): undefined reference to `__CTOR_LIST__' /usr/local/lib/gcc-lib/arm-elf/2.95.3/libgcc.a(__main.o)(.data+0x0): undefined reference to `__DTOR_LIST__' collect2: ld returned 1 exit status ????????mailtomax at 163.com ??????????2005-03-14 From mohuaidong at gmail.com Mon Mar 14 02:11:09 2005 From: mohuaidong at gmail.com (Huaidong Mo) Date: Mon, 14 Mar 2005 15:11:09 +0800 Subject: [uClinux-dev] pppd on uclinux,need help Message-ID: <6df4decc05031323115c1ccced@mail.gmail.com> HI: I need to use pppd to dial. First, I want to use tip to test my "/dev/ttyS0". I wonder whether I can dial up to network only with tip. when I input ATDT#777,there are no echo information.Is this correct? If dial succeed,can tip give me information like "input username:" "input passwprd:"? Now ,on my platform,I only get strange info "? N ?TC".Can anybody tell me,if all go well,what should I get? my tip programme come from uClinux-dist-20040408. From manish.mishra at st.com Mon Mar 14 03:10:05 2005 From: manish.mishra at st.com (Manish Kumar MISHRA) Date: Mon, 14 Mar 2005 13:40:05 +0530 Subject: [uClinux-dev] problem compiling application having vfork and thread Message-ID: <003001c5286d$3fb770a0$4d04b40a@dlh.st.com> Hi, I am trying to compile one application on uclinux platform (2.4 version) and this application has both vfork and posix threads. I am getting the following error after compilation upnpd.elf2flt(.text+0x1eed8): In function `__fork': : undefined reference to `__libc_fork' collect2: ld returned 1 exit status make[2]: *** [upnpd] Error 1 The code gets compiled if i remove either fork or thread from the code. It seems fork and threads cannot coexist in uclinux platform. Has anybody encountered such error during compilation? Please let me know if some fix is avaialbe for this problem? Thanx in advance, Manish From mkr at kron-systems.de Mon Mar 14 03:34:57 2005 From: mkr at kron-systems.de (Manuel Kron) Date: Mon, 14 Mar 2005 09:34:57 +0100 Subject: [uClinux-dev] problem compiling application having vfork and thread References: <003001c5286d$3fb770a0$4d04b40a@dlh.st.com> Message-ID: <001701c52870$b64a0680$3401a8c0@linus> Hi Manish, Here is a small example: /*************************************************************************** / /* * Thread demo * Compile: m68k-elf-gcc t3.c -lpthread -o demo -Wl, -elf2flt -m5200 -lc */ /*************************************************************************** / #include #include #include int counter = 0; /*************************************************************************** / /* * The producer thead */ void *ThreadLoop(void *arg) { while(1) { counter++; sleep(1); } } /*************************************************************************** / /* MAIN */ int main(void) { int iThread, mirror; pthread_t mythread; iThread = pthread_create(&mythread, NULL, ThreadLoop, NULL); if (iThread!=0) { printf("Error creating thread\n"); exit(-1); } printf("Counter (incremented by Thread)="); mirror = counter; while(1) { if(mirror != counter) { printf("%04d\b\b\b\b",counter); fflush(stdout); mirror = counter; } } return(0); } Regards, Manuel Hi, I am trying to compile one application on uclinux platform (2.4 version) and this application has both vfork and posix threads. I am getting the following error after compilation upnpd.elf2flt(.text+0x1eed8): In function `__fork': : undefined reference to `__libc_fork' collect2: ld returned 1 exit status make[2]: *** [upnpd] Error 1 The code gets compiled if i remove either fork or thread from the code. It seems fork and threads cannot coexist in uclinux platform. Has anybody encountered such error during compilation? Please let me know if some fix is avaialbe for this problem? Thanx in advance, Manish _______________________________________________ uClinux-dev mailing list uClinux-dev at uClinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev at uclinux.org From fabio.giovagnini at aurion-tech.com Mon Mar 14 05:46:02 2005 From: fabio.giovagnini at aurion-tech.com (Fabio Giovagnini) Date: Mon, 14 Mar 2005 11:46:02 +0100 Subject: [uClinux-dev] help me! In-Reply-To: <20050314064446.73E0E1A0D0@uclinux.org> References: <20050314064446.73E0E1A0D0@uclinux.org> Message-ID: <200503141146.03360.fabio.giovagnini@aurion-tech.com> you have to check the crt0.S and crt1.c toy are using related to linker script file toy are invoking with ld. Best Regards Fabio Giovagnini Alle 07:47, luned? 14 marzo 2005, ??? ha scritto: > hi all: > I download a uclibc version uClibc-0.9.27 and arm-elf-tools-.sh form > uclibc.org, then I compile a simple code,but it take some errors, it makes > me crazy, could you help me? thanks thanks thanks! > > arm-elf-gcc -Os -fno-builtin -nostartfiles -Wl,-elf2flt > /usr/local/arm-elf/lib/crt0.o libtest.o -lgcc -lpthread -lc > -L/usr/local/arm-elf/lib -L. -o libtest > > arm-elf-gcc libtest.o -lgcc -lpthread -lc -L/usr/local/arm-elf/lib -L. -o > libtest -fno-builtin > /usr/local/lib/gcc-lib/arm-elf/2.95.3/libgcc.a(__main.o): In function > `__do_global_ctors': > > __main.o(.text+0xc4): undefined reference to `__CTOR_LIST__' > /usr/local/lib/gcc-lib/arm-elf/2.95.3/libgcc.a(__main.o)(.data+0x0): > undefined reference to `__DTOR_LIST__' collect2: ld returned 1 exit status > > > > ????????mailtomax at 163.com > ??????????2005-03-14 -- Fabio Giovagnini Aurion s.r.l. www.aurion-tech.com From fabrice.poulin at medianesysteme.com Mon Mar 14 06:02:45 2005 From: fabrice.poulin at medianesysteme.com (Fabrice Poulin) Date: Mon, 14 Mar 2005 12:02:45 +0100 Subject: [uClinux-dev] problem with /dev/port Message-ID: Hi, I work on the board COBRA with coldfire 5272 and I would use access to IO of the board. But I have this problem, I don't have the file "/dev/port" ! So, how access to write and read on the IO pin of this board. Thank you Fabrice From wangjinfeng at gmail.com Mon Mar 14 01:28:57 2005 From: wangjinfeng at gmail.com (Wang JinFeng) Date: Mon, 14 Mar 2005 14:28:57 +0800 Subject: [uClinux-dev] uclinux mount redhat linux err Message-ID: redhat linux 8.0(192.168.11.247) /etc/export: /opt (ro) uclinux-2.4.20(192.168.11.248) command: mount -o nolock -t nfs 192.168.11.247:/opt /tmp get err messages following: >nfs warning: mount version older than kernel >NFS: mount program didn't pass remote address! >mount: Mounting 192.168.11.247:/opt on /mnt failed: Unknown error 22 >pid 29: failed 256 Anyone may give me some advices?3x -- Company : R&D, ARIM Address : NO.72 Xisihuan South Road,Beijing,China 100071 Ph. (86) 010-63812255?3804 Fax (86) 010-63851580 Email : wangjinfeng at gmail.com From ngustavson at emacinc.com Mon Mar 14 08:53:18 2005 From: ngustavson at emacinc.com (NZG) Date: Mon, 14 Mar 2005 07:53:18 -0600 Subject: [uClinux-dev] problem with /dev/port In-Reply-To: References: Message-ID: <200503140753.19038.ngustavson@emacinc.com> I don't know if /dev/port makes sense in the Coldfire world since everything is memory mapped. If your wanting user space access, you probably need to use /dev/mem. (Although I've never tried this) In kernel space, register access is just done with pointers. NZG. On Monday 14 March 2005 05:02, Fabrice Poulin wrote: > Hi, > > I work on the board COBRA with coldfire 5272 and I would use access to IO > of the board. > But I have this problem, I don't have the file "/dev/port" ! > So, how access to write and read on the IO pin of this board. > > Thank you > > Fabrice > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org From tss at Wiseone.zzn.com Mon Mar 14 09:50:28 2005 From: tss at Wiseone.zzn.com (tss ooty) Date: Mon, 14 Mar 2005 20:20:28 +0530 Subject: [uClinux-dev] Multicasting from uCLinux... Message-ID: <0FF7D363AB02B7348BA211A044D22525@tss.Wiseone.zzn.com> Hi ucDevelopers, We are trying to enable multicasting on a linux box over here. We have the wiscore board with Conexant chip, ARM7TDMI core. We are using the precompiled uCLinux kernel and TCP/IP stack along with the board. The board runs uCLinux-2.0.38 with Busybox- 0.60.5. We are not able to multicast to a host with the address 236.23.5.5 NOTE: 1. The network is able to receive packets (ethereal captures using promiscous mode but not in non-promiscous mode), but the application program fails to receive it i.e. the socket does not recv at all (ex: a small multicast receiver program does not receive it). 2. The same code compiled and executed on host will run perfectly! ------------------ The cmds. we stepped are: # ifcofig eth0 192.168.7.4 netmask 255.255.255.0 broadcast 192.168.7.255 # route add -net 192.168.7.0 netmask 255.255.255.0 eth0 # route add -net 236.23.5.0 netmask 255.255.255.0 eth0 Are we missing anything? Any thoughts/opinions/information on this matter would be appreciated. Thank you. #Sar Get your own Free E-mail at Witch Wisdom http://www.witchwisdom.com ___________________________________________________________ Get your own Web-based E-mail Service at http://www.zzn.com From swenggsri at yahoo.com Mon Mar 14 10:31:31 2005 From: swenggsri at yahoo.com (Sriram Chadalavada) Date: Mon, 14 Mar 2005 07:31:31 -0800 (PST) Subject: [uClinux-dev] 2.6 kernel not booting on uCdimm In-Reply-To: 6667 Message-ID: <20050314153131.30243.qmail@web21604.mail.yahoo.com> Hi Greg, Thanks for the quick reply. > I have my local CVS log, but there has really been > very little > work done on this driver in all of the 2.6 kernels. > Georges Menie > did some cleanups in 2.5 series for this. > Please email me your log and if possible, Menie's changes for 2.5. > > 2. Where are the rom.ld, himem.ld and ram.ld files > > located in 2.6.x? Again, these were neither > supplied > > by the vanilla kernel nor by the corresponding > uclinux > > patch. Please let me know if the use of these > files > > has been superseded by some other method in 2.6. > > Because, even without those files in that > directory, > > the start address of the kernel was the same as > that > > for 2.4.24 kernel = 0x10c10400. > > There is a single linker script for all m68knommu > targets: > ~/arch/m68knommu/kernel/vmlinux.lds.S. All of the > functionality of each of the older files should be > catered > for in this script. Thanks for clarifying that. One question though. How can I use the assembly 'putc' in this script? Where are the assembly instructions (loading the stack ptr, copying data segment from ROM to RAM .....) corresponding to _start/_stext? Cheers, Sriram __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/ From mbeaulieu at eocycle.com Mon Mar 14 10:51:44 2005 From: mbeaulieu at eocycle.com (Martin Beaulieu) Date: Mon, 14 Mar 2005 10:51:44 -0500 Subject: [uClinux-dev] insmod modules References: <20050311181619.B056D14EBE4@mail07.powweb.com> Message-ID: <009b01c528ad$bd4c61c0$8000a8c0@MartinLogiciel> Hi, I check all of my sdram configuration and all setting is OK. I follow the procedur for initialized. When i try to load a module, the kernel freeze :( Blkmem copyright 1998,1999 D. Jeff Dionne Blkmem copyright 1998 Kenneth Albanowski Blkmem 1 disk images: 0: 18F2000-1944BFF [VIRTUAL 18F2000-1944BFF] (RO) RAMDISK driver initialized: 16 RAM disks of 4096K size 1024 blocksize VFS: Mounted root (romfs filesystem) readonly. Sash command shell (version 1.1.1) /> cd lib /lib> cd modules /lib/modules> cd rtai /lib/modules/rtai> insmod rtai.o Using rtai.o The trace "Using rtai.o" is in insmod ??? I can run a application correctly but i can load a module, what is the problem ? I need help Martin ----- Original Message ----- From: "Carlos Manuel Duclos Vergara" To: "Martin Beaulieu" Cc: "uClinux development list" Sent: Friday, March 11, 2005 1:16 PM Subject: Re: [uClinux-dev] insmod modules > If you don't tell us the sdram chips and the configuration of your board > it is very unlikely that we know if it is ok or not > it is very important to follow the procedure indicated in your sdram > chip's datasheet to make them work. > > are you sure that the compiler is generating the right code for: > *(uint16 *)0x00000000 = junk; > > it must write something directly to that memory address, to be sure use > volatile the keyword: > *(volatile uint16 *)0x00000000 = junk; > > -- > Carlos Manuel Duclos Vergara > http://www.elbazaar.cl > > From c.holst at gmx.de Mon Mar 14 11:39:25 2005 From: c.holst at gmx.de (Christoph Holst) Date: Mon, 14 Mar 2005 17:39:25 +0100 Subject: [uClinux-dev] Searching for Evaluation Board Message-ID: <20050314163939.DF4FB19E61@uclinux.org> Hi there! I'm an embedded Linux newbie and would like to start playing with a cheap Evaluation Board, preferably with ARM architecture and onboard LCD and Audio drivers. Is there any good company selling those boards in Europe? Regards, Chris -------------- next part -------------- An HTML attachment was scrubbed... URL: From carlos at embedded.cl Mon Mar 14 11:58:37 2005 From: carlos at embedded.cl (Carlos Manuel Duclos Vergara) Date: Mon, 14 Mar 2005 16:58:37 -0000 Subject: [uClinux-dev] insmod modules Message-ID: <20050314165839.3B63C14DA15@mail07.powweb.com> An embedded and charset-unspecified text was scrubbed... Name: not available URL: From mbeaulieu at eocycle.com Mon Mar 14 12:12:44 2005 From: mbeaulieu at eocycle.com (Martin Beaulieu) Date: Mon, 14 Mar 2005 12:12:44 -0500 Subject: [uClinux-dev] insmod modules References: <20050314165839.3B63C14DA15@mail07.powweb.com> Message-ID: <00d501c528b9$0abdaa30$8000a8c0@MartinLogiciel> Hi > did you check the "Load modules in kernel memory" opion when you build > insmod? Yes, it's the same result. > which insmod are you using? there are two insmod in the uclinux > distribution I use the version on BusyBox, where is the other version ? Thanks Martin From mike at steroidmicros.com Mon Mar 14 15:52:20 2005 From: mike at steroidmicros.com (mike) Date: Mon, 14 Mar 2005 12:52:20 -0800 Subject: [uClinux-dev] Interrupts and uCLinux Message-ID: <000b01c528d7$b831d380$6202a8c0@flanders> Hello, I am using a ColdFire5282 chip and would like to handle interrupts from the General Purpose Timer pins and the IRQ pins in uCLinux. from what I understand this can be accomplished from kernel space within a device driver, but the latency for the interrupts will be an issue (i.e. can take milli-seconds). Also, interrupt level/priority may also become an issue. So the question I would like to pose is... Is it possible to do any sort of interrupt handling from within uClinux for GPT events like input capture and output compare at a reasonably fast rate (i.e. in the KHz range)? Or must this be tackled with the real time extensions like RTAI? Or others? Any opinions are greatly appreciated, Thanks, Mike Lavender From DaveM at larsondavis.com Mon Mar 14 18:04:14 2005 From: DaveM at larsondavis.com (Dave Memmott) Date: Mon, 14 Mar 2005 16:04:14 -0700 Subject: [uClinux-dev] return value for execve( ) Message-ID: Hi, I would like to get the return value of the program call by execve( ) in the example below. Unfortunately execve only returns a error value if ran unsuccessfully and NOT if the program argument ran unsuccessfully. I though about using a script to capture the return value of e2fsck( ) and write it to a file but the file system is umount at this point. char * argv[4] = {"/bin/e2fsck", "-p", "/dev/flash0", NULL}; execve("/bin/e2fsck",argv, envp_init); Any ideas how to get the return value of e2fsck( )? Dave M. From davidm at snapgear.com Mon Mar 14 18:30:30 2005 From: davidm at snapgear.com (David McCullough) Date: Tue, 15 Mar 2005 09:30:30 +1000 Subject: [uClinux-dev] return value for execve( ) In-Reply-To: References: Message-ID: <20050314233030.GB32166@beast> Jivin Dave Memmott lays it down ... > Hi, > > I would like to get the return value of the program call by execve( ) in the example below. Unfortunately execve only returns a error value if ran unsuccessfully and NOT if the program argument ran unsuccessfully. I though about using a script to capture the return value of e2fsck( ) and write it to a file but the file system is umount at this point. > > char * argv[4] = {"/bin/e2fsck", "-p", "/dev/flash0", NULL}; > execve("/bin/e2fsck",argv, envp_init); > > Any ideas how to get the return value of e2fsck( )? It's pretty easy, lots of examples in the uClinux-dist if you go looking. Lots of examples on the net as well. Here is a non-complete overview of what to do, you should read the "man" pages for wait(2) to understand this better. #include #include pid_t pid, wpid; int status, rc; char * argv[4] = {"/bin/e2fsck", "-p", "/dev/flash0", NULL}; ... pid = vfork(); if (pid == (pid_t) -1) { perror("vfork"); return; } if (pid == (pid_t) 0) { execve("/bin/e2fsck",argv, envp_init); perror("execve"); _exit(1); } wpid = wait(&status) if (wpid != pid) { /* deal with this however suits your needs */ /* it could be an error or another child exiting */ return } rc = WEXITSTATUS(status); Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From gerg at snapgear.com Mon Mar 14 19:57:39 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Tue, 15 Mar 2005 10:57:39 +1000 Subject: [uClinux-dev] Interrupts and uCLinux In-Reply-To: <000b01c528d7$b831d380$6202a8c0@flanders> References: <000b01c528d7$b831d380$6202a8c0@flanders> Message-ID: <42363303.7090005@snapgear.com> Hi Mike, mike wrote: > I am using a ColdFire5282 chip and would like to handle interrupts from > the General Purpose Timer pins and the IRQ pins in uCLinux. from what I > understand this can be accomplished from kernel space within a device > driver, but the latency for the interrupts will be an issue (i.e. can take > milli-seconds). Also, interrupt level/priority may also become an issue. > > So the question I would like to pose is... Is it possible to do any sort > of interrupt handling from within uClinux for GPT events like input capture > and output compare at a reasonably fast rate (i.e. in the KHz range)? Or > must this be tackled with the real time extensions like RTAI? Or others? Well, you do it. That is the kernel can handle a few thousand interrupts a second. But the jitter and latency is probably not going to be all that good overall. If your requirement is that your interrupt _must_ be handled with some small period of time then you need to do something more. Either with real time extensions, or more simple techniques like using level7/NMI interrupts. (This may well be enough for you if your only requirement is that an interrupt must always be serviced quickly). Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From saiful at modular-corp.com Mon Mar 14 22:53:43 2005 From: saiful at modular-corp.com (saiful) Date: Tue, 15 Mar 2005 11:53:43 +0800 Subject: [uClinux-dev] root password for Soekris Message-ID: <200503151153.AA320340904@modular-corp.com> Hi, I boot up Soekris from my IDE flash. As stated in Documentation the root password for the Soekris login is 'default' but I still cannot login. Is there a new root password for Soekris. Saiful From yashi at atmark-techno.com Tue Mar 15 00:03:16 2005 From: yashi at atmark-techno.com (Yasushi SHOJI) Date: Tue, 15 Mar 2005 14:03:16 +0900 Subject: [uClinux-dev] renaming uClinux-dist Message-ID: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> Hi all, We, Atmark Techno, has been distributing uClinux-dist as a development source code base for our product, including ARM CPU with MMU, in addition to MicroBlaze, a soft core CPU without MMU. Many of our customers give us feedback saying that all of our product is uClinux based, since the tarball is name after uClinux-dist. Would you, as a maintainer and as a member of community, be OK if we rename the tarball to something like "Atmark Techno Dist", so that our customer don't get wrong impression that all of our product is based on uClinux? We will, of course, put words in documents saying that this source code is based on uClinux-dist and can be downloaded from such and such. # Don't get me wrong. uClinux _is_ great. it's just that some people # think the way we don't think ;p What do you all think? -- yashi From davidm at snapgear.com Tue Mar 15 00:49:33 2005 From: davidm at snapgear.com (David McCullough) Date: Tue, 15 Mar 2005 15:49:33 +1000 Subject: [uClinux-dev] root password for Soekris In-Reply-To: <200503151153.AA320340904@modular-corp.com> References: <200503151153.AA320340904@modular-corp.com> Message-ID: <20050315054933.GA6807@beast> Jivin saiful lays it down ... > Hi, > > I boot up Soekris from my IDE flash. As stated in Documentation the > root password for the Soekris login is 'default' but I still cannot > login. Is there a new root password for Soekris. Are you using uClibc or glibc ? It possible something is wrong with the config. Check that the Makefile is adding a /etc/passwd or /etc/config/passwd (if flatfsd is enabled). You might want to try making your own passwd file to see if that helps. Is it telling you that the password is incorrect or is it some other error ? Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From saiful at modular-corp.com Tue Mar 15 01:35:42 2005 From: saiful at modular-corp.com (saiful) Date: Tue, 15 Mar 2005 14:35:42 +0800 Subject: [uClinux-dev] root password for Soekris Message-ID: <200503151435.AA80478976@modular-corp.com> Hi David, I am using glibc. The errror is "login incorrect". If I edit the passwd file, what is the password entry for "root" if i just want make return key as my passowrd ? Saiful ---------- Original Message ---------------------------------- From: David McCullough Date: Tue, 15 Mar 2005 15:49:33 +1000 > >Jivin saiful lays it down ... >> Hi, >> >> I boot up Soekris from my IDE flash. As stated in Documentation the >> root password for the Soekris login is 'default' but I still cannot >> login. Is there a new root password for Soekris. > >Are you using uClibc or glibc ? > >It possible something is wrong with the config. Check that the Makefile >is adding a /etc/passwd or /etc/config/passwd (if flatfsd is enabled). > >You might want to try making your own passwd file to see if that helps. > >Is it telling you that the password is incorrect or is it some other >error ? > >Cheers, >Davidm > >-- >David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com >Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org > From lars.friedrich at wago.com Tue Mar 15 01:51:26 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Tue, 15 Mar 2005 07:51:26 +0100 Subject: [uClinux-dev] insmod modules Message-ID: > /lib> cd modules > /lib/modules> cd rtai > /lib/modules/rtai> insmod rtai.o > Using rtai.o > > The trace "Using rtai.o" is in insmod ??? > > I can run a application correctly but i can load a module, > what is the problem ? You've written before you could load a basic module. Why do you think the module loading is broken then? How do you know it's not RTAI that is broken for your architecture? Best regards, Lars Friedrich From pwilshire at cox.net Tue Mar 15 01:53:46 2005 From: pwilshire at cox.net (Phil Wilshire) Date: Tue, 15 Mar 2005 01:53:46 -0500 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> Message-ID: <4236867A.4090004@cox.net> Hi Why not call your version the ATlinux-dist. Or should we come up with a new name for the whole distribution since it is not just MMULess any more. Any Thoughts .... Phil Wilshrie Yasushi SHOJI wrote: >Hi all, > >We, Atmark Techno, has been distributing uClinux-dist as a development >source code base for our product, including ARM CPU with MMU, in >addition to MicroBlaze, a soft core CPU without MMU. > >Many of our customers give us feedback saying that all of our product >is uClinux based, since the tarball is name after uClinux-dist. > >Would you, as a maintainer and as a member of community, be OK if we >rename the tarball to something like "Atmark Techno Dist", so that our >customer don't get wrong impression that all of our product is based >on uClinux? > >We will, of course, put words in documents saying that this source >code is based on uClinux-dist and can be downloaded from such and >such. > ># Don't get me wrong. uClinux _is_ great. it's just that some people ># think the way we don't think ;p > >What do you all think? >-- > yashi > >_______________________________________________ >uClinux-dev mailing list >uClinux-dev at uClinux.org >http://mailman.uclinux.org/mailman/listinfo/uclinux-dev >This message was resent by uclinux-dev at uclinux.org > > > From devi at tenet.res.in Tue Mar 15 13:17:08 2005 From: devi at tenet.res.in (devi) Date: Tue, 15 Mar 2005 23:47:08 +0530 Subject: [uClinux-dev] console Message-ID: <20050315181708.17942.qmail@mail.tenet.res.in> hi, In console.c take_over_console() exits from the loop.That's why I am getting Console:switching to frame buffer device. How to avoid exiting from the loop.So that it process con_init in fbcon.c Regards Devi From wnlee at cad.csie.ncku.edu.tw Tue Mar 15 02:18:53 2005 From: wnlee at cad.csie.ncku.edu.tw (Wayne Lee) Date: Tue, 15 Mar 2005 15:18:53 +0800 Subject: [uClinux-dev] where should we store ttb for MMU? In-Reply-To: <42312AC5.9060206@snapgear.com> Message-ID: <200503150721.j2F7LoPI009571@cad.csie.ncku.edu.tw> Hi, Greg and community, I know that CONFIG_DRAM_BASE+0x100 is reserved for boot parameters. CONFIG_DRAM_BASE+0x200 is vector table (__real_stubs_start). I do an experiment to fill 0xFFFF_FFFF from 0x0 to 0x7FFF using ICE. After kernel is booted with I-Cache turn on only, 0x4000~0x7FFF is no longer be 0xFFFF_FFFF. The memory region 0x4000~0x7FFF is just the memory region I store flat-mapped TTB for MMU. Hence, the CPU always generates data abort while booting with MMU enabled. Are there any memory region is reserved for TTB? I know uClinux aims to run Linux on no-MMU chips. Hence, uClinux may _not_ reserve a protected memory region for TTB while a CPU with MMU is adopted. Please kindly give suggestions, thanks. The booting messages with I-Cache, store TTB at 0x4000 and enable MMU/D-Cache. Linux version 2.6.5-lwn1 (wayne at fedora-linux) (gcc version 2.95.3 20010315 (release)(ColdFire patches - 20010318 from http://fiddes.net/coldfire/)(uClinux XIP and shared lib patches from http://www.snapgear.com/)) #38 Wed Mar 9 12:36:26 CST 2005 CPU: ARM926EJ-Sid(wb) [41069263] revision 3 (ARMv5TEJ) CPU: D VIPT write-back cache CPU: I cache: 32768 bytes, associativity 4, 32 byte lines, 256 sets CPU: D cache: 32768 bytes, associativity 4, 32 byte lines, 256 sets Machine: QL201 On node 0 totalpages: 16384 DMA zone: 0 pages, LIFO batch:1 Normal zone: 16384 pages, LIFO batch:4 HighMem zone: 0 pages, LIFO batch:1 Built 1 zonelists Kernel command line: root=/dev/ram0 initrd=0x02000000,1000K keepinitrd PID hash table entries: 512 (order 9: 4096 bytes) Memory: 64MB = 64MB total Memory: 62496KB available (1100K code, 119K data, 60K init) Calibrating delay loop... 15.92 BogoMIPS Dentry cache hash table entries: 8192 (order: 3, 32768 bytes) Inode-cache hash table entries: 4096 (order: 2, 16384 bytes) Mount-cache hash table entries: 512 (order: 0, 4096 bytes) checking if image is initramfs...it isn't (ungzip failed); looks like an initrd POSIX conformance testing by UNIFIX NET: Registered protocol family 16 ikconfig 0.7 with /proc/config* ttyS0 at I/O 0x80000100 (irq = 17) is a builtin QL201 UART RAMDISK driver initialized: 16 RAM disks of 8192K size 1024 blocksize loop: loaded (max 8 devices) NET: Registered protocol family 2 The booting messages with only I Cache. (Successfully booted) ... RAMDISK driver initialized: 16 RAM disks of 8192K size 1024 blocksize loop: loaded (max 8 devices) NET: Registered protocol family 2 IP: routing cache hash table of 512 buckets, 4Kbytes TCP: Hash tables configured (established 4096 bind 4096) IPv4 over IPv4 tunneling driver GRE over IPv4 tunneling driver NET: Registered protocol family 17 ... -----Original Message----- From: Greg Ungerer [mailto:gerg at snapgear.com] Sent: Friday, March 11, 2005 1:21 PM To: Wayne Lee Cc: uClinux development list Subject: Re: [uClinux-dev] where should we store ttb for MMU? Hi Wayne, Wayne Lee wrote: > In fact, I am now porting an uClinux no VM port on arm926ej-s CPU. > In order to turn-on D-cache, I use flat-mapping & section-mapped MMU and > then turn on D-cache. > The method is referred from Samsung s5c7375 port. > I store TTB in 0x4000. > However, while kernel is booting, the TTB is overwritten and data abort > occurred. Who overwrites it? What is the kernel trying to do when overwritting this area? Regards Greg > I am wonder where should we place the ttb in uClinux 2.6? > Do we need to modify the vmlinux.lds? > Please kindly give me your suggestions. > > Wayne > From yashi at atmark-techno.com Tue Mar 15 03:13:37 2005 From: yashi at atmark-techno.com (Yasushi SHOJI) Date: Tue, 15 Mar 2005 17:13:37 +0900 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <4236867A.4090004@cox.net> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> Message-ID: <20050315081337.0A198633A@mail2.atmark-techno.com> At Tue, 15 Mar 2005 01:53:46 -0500, Phil Wilshire wrote: > > Why not call your version the ATlinux-dist. Than's our plan. We just thought, at first, that it's rude to call the dist with different name even if it has a few patche in. > Or should we come up with a new name for the whole distribution since it > is not just > MMULess any more. well, that'll do too. BTW, I enjoyed a session at FOSDEM about Bitbake. http://developer.berlios.de/projects/bitbake/ it doesn't support uclinux yet but it seems very nice. the biggest drawback is that it requires more than 1GB of ram on development system. :O (OK, I haven't tested it yet. so I apologize if this is a FUD) Mickey, who gave us very nice introduction about Bitbake at FOSDEM, told us that it'd be easy to support uClinux. and there is already a few sub-project to reduce the memory consumption. regards, -- yashi From huangl at parawireless.com Tue Mar 15 03:19:11 2005 From: huangl at parawireless.com (Liang Huang) Date: Tue, 15 Mar 2005 16:19:11 +0800 Subject: [uClinux-dev] about pthread_self() Message-ID: <42369A7F.8080505@parawireless.com> A newbie's question: sometimes when I call pthread_self(), the return value is 0. It is supposed to return the thread ID of the current thread. Could anyone give me some hints? Best Regards, Liang From davidm at snapgear.com Tue Mar 15 03:20:15 2005 From: davidm at snapgear.com (David McCullough) Date: Tue, 15 Mar 2005 18:20:15 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050315081337.0A198633A@mail2.atmark-techno.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <20050315081337.0A198633A@mail2.atmark-techno.com> Message-ID: <20050315082015.GA11932@beast> Jivin Yasushi SHOJI lays it down ... > At Tue, 15 Mar 2005 01:53:46 -0500, > Phil Wilshire wrote: > > > > Why not call your version the ATlinux-dist. > > Than's our plan. We just thought, at first, that it's rude to call > the dist with different name even if it has a few patche in. My guess is that you are the first to ask, but most certainly not the first to do it :-) We did the same thing with SGLinux, which is basically a rename of the uClinux-dist to get away from the !MMU syndrome. Somehow uClinux == !MMU, even though that has never been the case AFAIK. So you could just use SGLinux ;-), but I don't think anyone will have a problem with you changing the name of the archive or even shipping a modded version, plenty of others have already. Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From gerg at snapgear.com Tue Mar 15 06:17:11 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Tue, 15 Mar 2005 21:17:11 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <4236867A.4090004@cox.net> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> Message-ID: <4236C437.6060404@snapgear.com> Hi All, Phil Wilshire wrote: > Hi > Why not call your version the ATlinux-dist. > > Or should we come up with a new name for the whole distribution since it > is not just > MMULess any more. I get that comment a bit. And I have spoken to many people who are confused by the fact that uClinux-dist (heck, even just the uClinux kernel sources) work with VM processors. Do we need a new name? Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Tue Mar 15 06:22:35 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Tue, 15 Mar 2005 21:22:35 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> Message-ID: <4236C57B.706@snapgear.com> Hi Yashi, Yasushi SHOJI wrote: > We, Atmark Techno, has been distributing uClinux-dist as a development > source code base for our product, including ARM CPU with MMU, in > addition to MicroBlaze, a soft core CPU without MMU. > > Many of our customers give us feedback saying that all of our product > is uClinux based, since the tarball is name after uClinux-dist. > > Would you, as a maintainer and as a member of community, be OK if we > rename the tarball to something like "Atmark Techno Dist", so that our > customer don't get wrong impression that all of our product is based > on uClinux? I certainly don't have a problem with it. As Dave points out, a few others have essentially done this in the past. Regards Greg > We will, of course, put words in documents saying that this source > code is based on uClinux-dist and can be downloaded from such and > such. > > # Don't get me wrong. uClinux _is_ great. it's just that some people > # think the way we don't think ;p > > What do you all think? > -- > yashi > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org > -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From fabrice.poulin at medianesysteme.com Tue Mar 15 09:33:10 2005 From: fabrice.poulin at medianesysteme.com (Fabrice Poulin) Date: Tue, 15 Mar 2005 15:33:10 +0100 Subject: [uClinux-dev] problem to write/read to IO port Message-ID: Hi, I work with the coldfire 5272 but I already asked the question about acess to IO port. In fact, there isn't device /dev/port to write or read with uClinux on the 5272, so how I do to write or read on the Port A for exemple? If someboby has an example code, I would like this...? Thank Fabrice From wnlee at cad.csie.ncku.edu.tw Tue Mar 15 10:06:16 2005 From: wnlee at cad.csie.ncku.edu.tw (Wayne Lee) Date: Tue, 15 Mar 2005 23:06:16 +0800 Subject: [uClinux-dev] Is smc91x.c the best choice for smsc91c111 chip in uClinux 2.6? Message-ID: <200503151509.j2FF94KE003732@cad.csie.ncku.edu.tw> Hi, all, I found smc91x.c was in mainstream Linux kernel since 2.6.8. Smc91x.c seems popular in ARM Linux field. Does smc91x.c be the most suitable driver for smsc91c111 chip in uClinux 2.6? Does anyone ever support smsc91c111 chip in uClinux 2.6? Thanks. Wayne -------------- next part -------------- An HTML attachment was scrubbed... URL: From geoffrey at pager.net Tue Mar 15 10:31:43 2005 From: geoffrey at pager.net (Geoffrey Wossum) Date: Tue, 15 Mar 2005 09:31:43 -0600 Subject: OT: BitBake Memory Usage (was Re: [uClinux-dev] renaming uClinux-dist) In-Reply-To: <20050315081337.0A198633A@mail2.atmark-techno.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <20050315081337.0A198633A@mail2.atmark-techno.com> Message-ID: <1110900703.4328.4.camel@localhost> On Tue, 2005-03-15 at 17:13 +0900, Yasushi SHOJI wrote: > it doesn't support uclinux yet but it seems very nice. the biggest > drawback is that it requires more than 1GB of ram on development > system. :O (OK, I haven't tested it yet. so I apologize if this is a > FUD) Unfortunately, bitbake really does require a boatload of memory. I had to upgrade my laptop from 512 MB to 1 GB, because it just couldn't do a build with 512 MB (even with Psyco disabled). It also probably didn't help that I run KDE and Evolution... --- Geoffrey Wossum Long Range Systems - http://www.pager.net From dhowells at redhat.com Tue Mar 15 10:39:42 2005 From: dhowells at redhat.com (David Howells) Date: Tue, 15 Mar 2005 15:39:42 +0000 Subject: [uClinux-dev] return value for execve( ) In-Reply-To: References: Message-ID: <15012.1110901182@redhat.com> Dave Memmott wrote: > I would like to get the return value of the program call by execve( ) in the > example below. Unfortunately execve only returns a error value if ran > unsuccessfully and NOT if the program argument ran unsuccessfully. I though > about using a script to capture the return value of e2fsck( ) and write it > to a file but the file system is umount at this point. > > char * argv[4] = {"/bin/e2fsck", "-p", "/dev/flash0", NULL}; > execve("/bin/e2fsck",argv, envp_init); > > Any ideas how to get the return value of e2fsck( )? Erm... execve() is a no-return thing. The process that calls execve() has the program it was running ripped out and replaced with the one specified as the first argument, provided various checks the kernel makes succeed. If those checks fail, obviously execve() will return to the caller with an error. Once the kernel decides that all checks bear out, it will delete the processes VM; after that it can only kill the process entirely. This it does by sending a SIGKILL to the process. What do you mean by the "return value of the program"? The exit code? If so, you should vfork() and wait(). You could also use system() if you don't want to do all the steps yourself. David From wnlee at cad.csie.ncku.edu.tw Tue Mar 15 10:52:39 2005 From: wnlee at cad.csie.ncku.edu.tw (Wayne Lee) Date: Tue, 15 Mar 2005 23:52:39 +0800 Subject: [uClinux-dev] root password for Soekris In-Reply-To: <200503151435.AA80478976@modular-corp.com> Message-ID: <200503151555.j2FFtQKE004196@cad.csie.ncku.edu.tw> Try this /etc/passwd, passwd is "12345" root:ab80DAu2C3JN6:0:0::/:/bin/sh You can also choose the way into shell without login. Wayne -----Original Message----- From: uclinux-dev-bounces at uClinux.org [mailto:uclinux-dev-bounces at uClinux.org] On Behalf Of saiful Sent: Tuesday, March 15, 2005 2:36 PM To: saiful; David McCullough Cc: uclinux-dev at uClinux.org Subject: Re: [uClinux-dev] root password for Soekris Hi David, I am using glibc. The errror is "login incorrect". If I edit the passwd file, what is the password entry for "root" if i just want make return key as my passowrd ? Saiful ---------- Original Message ---------------------------------- From: David McCullough Date: Tue, 15 Mar 2005 15:49:33 +1000 > >Jivin saiful lays it down ... >> Hi, >> >> I boot up Soekris from my IDE flash. As stated in Documentation the >> root password for the Soekris login is 'default' but I still cannot >> login. Is there a new root password for Soekris. > >Are you using uClibc or glibc ? > >It possible something is wrong with the config. Check that the Makefile >is adding a /etc/passwd or /etc/config/passwd (if flatfsd is enabled). > >You might want to try making your own passwd file to see if that helps. > >Is it telling you that the password is incorrect or is it some other >error ? > >Cheers, >Davidm > >-- >David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com >Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org > _______________________________________________ uClinux-dev mailing list uClinux-dev at uClinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev at uclinux.org From scjody at steamballoon.com Tue Mar 15 11:17:44 2005 From: scjody at steamballoon.com (Jody McIntyre) Date: Tue, 15 Mar 2005 11:17:44 -0500 Subject: [uClinux-dev] [RFC] Proposal for getting LTP to run on uClinux Message-ID: <20050315161744.GS1111@conscoop.ottawa.on.ca> Proposal for getting LTP to run on uClinux Steamballoon Incorporated Sponsored by Analog Devices 1. Issues with the LTP on uClinux Since the LTP was written for a desktop/server linux platform, there are some issues with running it on a uClinux-based platform. The lack of fork(), incomplete mmap() capabilities, and limited resources on the devices means that all calls to fork() must be changed to vfork() and the functionality of each test must be fixed to either exec() a separate child or the same program with an appropriate argument. 2. Analysis of problematic cases 2.1 fork() There are four cases where fork() calls are used: - Either the parent does a wait()/waitpid() or the child does an exec() immediately after the fork(). - The functionality of the child is in a well defined function that can easily be called from an exec()ed copy of the program with a special argument. - The test needs to be completely refactored to work with vfork(). - The test is inappropriate for uClinux since it tests fork() itself. Also, the LTP allows multiple tests to be run simultaneously by parsing the -c parameter and forking that many tests. If the tests are meant to be run simultaneously then a method to exec() the children will be needed or only one test will run at a time. Since we are targeting devices with limited resources, this may not be an issue. We have identified 210 tests that use the fork() system call. 2.1.1 Parent does a wait / child does an exec There are 119 tests where fork() is used and the parent does a wait()/waitpid() or the child does an exec(). All these tests can just be converted to using vfork(). 2.1.2 Child/parent code needs to be split across exec There are 65 tests that are fairly straightforward to convert to vfork()/exec(). The main challenge is in passing variables across exec() using arguments or other methods, since parent variables will no longer be available to the child. Functions to support this will be added to the test library. Then, the tests will be converted to use them. This will mostly be cut and paste work. An example of a test like this is kernel/syscalls/fcntl/fcntl11.c . 2.1.3 Test needs to be completely refactored 14 tests will need to be completely refactored to work with vfork(). 2.1.4 Inappropriate test The remaining 12 tests are tests of the fork() system call itself. These are clearly inappropriate for a uClinux environment. 2.2 mmap() Most of the mmap() test cases will require changes to the options to the mmap() call. Also, any mmap() call that expects a certain VM address to be returned will fail. Tests which require this behavior are not appropriate for uClinux. Any mmap() call that is expecting the mapping to be backed by disk will fail unless the test is done read-only on a romfs file system. Of the 57 tests that have an mmap() call, 32 of them just require the removal of MAP_PRIVATE on the options. 25 of the tests are not appropriate for uClinux. We recommend putting the options for mmap() in a #define and using the -DUCLINUX compile time flag (see below) to choose the appropriate options. 2.3 Build changes Changes will need to be made to the build system to differentiate a normal (desktop/server) LTP build from a uClinux build. There are some tests that are not appropriate for uClinux. We recommend adding a uClinux target to the Makefile so that building for uClinux could be done using make uclinux instead of make. This would also add the -DUCLINUX build flag which would be used in the source to select fork()/vfork() behavior, etc.. 2.4 Run changes The LTP will need to be modified to be either runnable from NFS or through rcp/rsh (as with gcc regression testing.) We recommend running over NFS since it would require the fewest changes. NFS gives the appearance of a normal filesystem and LTP can be run by mounting the directory where it was built and running it from there. We tested this using i386 systems as the "uClinux host" and "development server" and LTP runs correctly. 2.5 Library changes 2.5.1 vfork()/exec() Since many testcases will need to be able to re-exec() themselves and choose the child code path, we will implement a generic self_exec wrapper that will pass the correct arguments to the child process. We propose adding the -C option for this purpose. Each test on startup will call into the library to store the startup information and to branch to the child code path if necessary. Making this as easy to use as possible increases the chances that future tests will be written in such a way as to work in uClinux. 2.5.2 TEST_XXXX Re-implement TEST_XXXX functions for uClinux. This means converting fork() to vfork()/exec(). This will use a similar mechanism to that described in section 2.5.1. This will affect the -c and -i options of the tests. 3.0 Patches to LTP As work is accepted by ADI, we will be sending patches to the LTP maintainers. >From correspondence with them, they would be willing to have the updates integrated in the March release, as long as the patches don't affect their currently supported platforms. This may not be a viable time-frame for ADI but it does show their willingness to cover uClinux as well. Care must be taken to maintain the same testing semantics on modified test cases. 4.0 Testing As the tests are modified their output will be compared to the original version to make sure there are no differences. Once all the cases are done, the full output of runltp will be compared with both sets. The resulting files will be compared using diff to make sure the same results are observed using both. 5.0 Sample Change A sample change is attached as appendix A. This is a vfork()/exec() conversion as described in section 2.1.2. Since the test library has not yet been modified, all the necessary changes have been made directly to the test. Also, vfork()/exec() will be used even on non-uClinux systems. This should not be done in the final version. Appendix A: Sample Change: Attached separately. -------------- next part -------------- This is a vfork()/exec() conversion as described in section 2.1.2 of our proposal. Since the test library has not yet been modified, all the necessary changes have been made directly to the test. Also, vfork()/exec() will be used even on non-uClinux systems. This should not be done in the final version. Index: fcntl/fcntl11.c =================================================================== --- fcntl.orig/fcntl11.c 2005-01-04 16:04:18.000000000 -0500 +++ fcntl/fcntl11.c 2005-02-21 13:20:30.000000000 -0500 @@ -32,6 +32,7 @@ * * HISTORY * 07/2001 Ported by Wayne Boyer + * 02/2005 Modified to work on uClinux as a demonstration. * * RESTRICTIONS * None @@ -40,6 +41,7 @@ #include #include #include +#include #include #include #include @@ -298,18 +300,52 @@ catch_child() cleanup(); } +static int child = 0; +static char *child_arg; +static option_t user_options[] = { + { "C:", &child, &child_arg }, /* -C option */ + { NULL, NULL, NULL } +}; + +static void user_opts_help() +{ + printf(" -C ARG Run the child process with arguments ARG " + "(for internal use)\n"); +} + int main(int ac, char **av) { struct flock tl; int lc; /* loop counter */ char *msg; /* message returned from parse_opts */ + char *start_cwd; /* Store cwd, for exec after vfork */ /* parse standard options */ - if ((msg = parse_opts(ac, av, (option_t *)NULL, NULL)) != (char *)NULL){ + if ((msg = parse_opts(ac, av, user_options, &user_opts_help)) != + (char *)NULL){ tst_brkm(TBROK, cleanup, "OPTION PARSING ERROR - %s", msg); } + start_cwd = getcwd(NULL, 0); + if (child) { + char *child_dir; + + child_dir = strtok(child_arg, ","); + parent_pipe[0] = strtol(strtok(NULL, ","), NULL, 10); + parent_pipe[1] = strtol(strtok(NULL, ","), NULL, 10); + child_pipe[0] = strtol(strtok(NULL, ","), NULL, 10); + child_pipe[1] = strtol(strtok(NULL, ","), NULL, 10); + fd = strtol(strtok(NULL, ","), NULL, 10); + + if (chdir(child_dir) < 0) { + tst_resm(TFAIL, "chdir failed"); + exit(1); + } + + do_child(); + } + setup(); /* global setup */ /* Check for looping state if -i option is given */ @@ -317,8 +353,23 @@ int main(int ac, char **av) /* reset Tst_count in case we are looping */ Tst_count = 0; - if ((child_pid = fork()) == 0) { /* parent */ - do_child(); + if ((child_pid = vfork()) == 0) { /* child */ + char *tmp_cwd; + char arg[1000]; + + tmp_cwd = getcwd(NULL, 0); + chdir(start_cwd); + + sprintf(arg, "%s,%d,%d,%d,%d,%d", tmp_cwd, + parent_pipe[0], parent_pipe[1], + child_pipe[0], child_pipe[1], fd); + + if (execlp(av[0], av[0], "-C", arg, + (char *) NULL) < 0) { + perror("exec"); + tst_resm(TFAIL, "exec failed"); + } + exit(1); } else if (child_pid < 0) { tst_resm(TFAIL, "Fork failed"); cleanup(); From mike at steroidmicros.com Tue Mar 15 12:54:16 2005 From: mike at steroidmicros.com (mike) Date: Tue, 15 Mar 2005 09:54:16 -0800 Subject: [uClinux-dev] problem to write/read to IO port In-Reply-To: Message-ID: <006b01c52988$03635620$6202a8c0@flanders> Hello, I am not sure about the 5272, but on the 5282 you can just write to the IO using the memory map described in the mcf5282.h (or mcf5272.h) file. The only thing about the 5282 is that by default, access to the IO space is restricted from user mode. You can get around this by only accessing the IO in kernel/supervisor mode from within a device driver. Another way... this restriction can be overridden by writing to the GPACR0 (5282), which can be done in your bootloader or in the crt0.s file for your linux kernel. Mike >>-----Original Message----- >>From: uclinux-dev-bounces at uClinux.org [mailto:uclinux-dev- >>bounces at uClinux.org] On Behalf Of Fabrice Poulin >>Sent: March 15, 2005 6:33 AM >>To: uclinux-dev at uclinux.org >>Subject: [uClinux-dev] problem to write/read to IO port >> >> >>Hi, >> >>I work with the coldfire 5272 but I already asked the question about acess >>to IO port. >>In fact, there isn't device /dev/port to write or read with uClinux on the >>5272, so how I do to write or read on the Port A for exemple? >>If someboby has an example code, I would like this...? >> >>Thank >> >>Fabrice >> >>_______________________________________________ >>uClinux-dev mailing list >>uClinux-dev at uClinux.org >>http://mailman.uclinux.org/mailman/listinfo/uclinux-dev >>This message was resent by uclinux-dev at uclinux.org From mbeaulieu at eocycle.com Tue Mar 15 13:22:46 2005 From: mbeaulieu at eocycle.com (Martin Beaulieu) Date: Tue, 15 Mar 2005 13:22:46 -0500 Subject: [uClinux-dev] insmod modules References: <20050314165839.3B63C14DA15@mail07.powweb.com> Message-ID: <004e01c5298c$0123f0a0$8000a8c0@MartinLogiciel> Hi, I change my memory size and all work correctly. I have corruption of memory. I make a patch with all of my modification. I send the patch this week. I have a configuration of uClinux and RTAI that works on rom. Thanks for all Martin ----- Original Message ----- From: "Carlos Manuel Duclos Vergara" To: "Martin Beaulieu" Cc: "uClinux development list" Sent: Monday, March 14, 2005 11:58 AM Subject: Re: [uClinux-dev] insmod modules > Hi, > >>I check all of my sdram configuration and all setting is OK. I follow the >>procedur for initialized. > > ok, assuming it is working ok > >>When i try to load a module, the kernel freeze :( >>Blkmem copyright 1998,1999 D. Jeff Dionne >>Blkmem copyright 1998 Kenneth Albanowski >>Blkmem 1 disk images: >>0: 18F2000-1944BFF [VIRTUAL 18F2000-1944BFF] (RO) >>RAMDISK driver initialized: 16 RAM disks of 4096K size 1024 blocksize >>VFS: Mounted root (romfs filesystem) readonly. >>Sash command shell (version 1.1.1) >>/> cd lib >>/lib> cd modules >>/lib/modules> cd rtai >>/lib/modules/rtai> insmod rtai.o >>Using rtai.o > > did you check the "Load modules in kernel memory" opion when you build > insmod? > which insmod are you using? there are two insmod in the uclinux > distribution > > Cheers! > -- > Carlos Manuel Duclos Vergara > http://www.elbazaar.cl > > From jsujjavanich at syntech-fuelmaster.com Tue Mar 15 15:22:44 2005 From: jsujjavanich at syntech-fuelmaster.com (Jate Sujjavanich) Date: Tue, 15 Mar 2005 15:22:44 -0500 Subject: [uClinux-dev] kernel task switch - bad frame format Message-ID: Kernel: 2.6.10 for the MCF5235 GCC: 3.4 I am debugging my initial port over to the MCF5235. I get a panic in what I believe is a task switch during start_kernel()/kmem_cache_init()/enable_cpucache()/do_tune_cpucache(...) /kfree(ccold)/cache_flusharry(...). In "static inline task_struct *get_current", there is a die_if_kernel("bad frame format") call. I would like to know if I should be receiving any console messages yet. I think the kernel console output task has not begun yet. Jate Sujjavanich -------------- next part -------------- An HTML attachment was scrubbed... URL: From jsujjavanich at syntech-fuelmaster.com Tue Mar 15 16:17:54 2005 From: jsujjavanich at syntech-fuelmaster.com (Jate Sujjavanich) Date: Tue, 15 Mar 2005 16:17:54 -0500 Subject: [uClinux-dev] kernel task switch - bad frame format Message-ID: The problem was that I did not have 'coldfire serial support' enabled. :P - Jate ________________________________ From: uclinux-dev-bounces at uClinux.org [mailto:uclinux-dev-bounces at uClinux.org] On Behalf Of Jate Sujjavanich Sent: Tuesday, March 15, 2005 3:23 PM To: uClinux development list Subject: [uClinux-dev] kernel task switch - bad frame format Kernel: 2.6.10 for the MCF5235 GCC: 3.4 I am debugging my initial port over to the MCF5235. I get a panic in what I believe is a task switch during start_kernel()/kmem_cache_init()/enable_cpucache()/do_tune_cpucache(...) /kfree(ccold)/cache_flusharry(...). In "static inline task_struct *get_current", there is a die_if_kernel("bad frame format") call. I would like to know if I should be receiving any console messages yet. I think the kernel console output task has not begun yet. Jate Sujjavanich From dapalmer at soe.ucsd.edu Tue Mar 15 16:18:52 2005 From: dapalmer at soe.ucsd.edu (dapalmer) Date: Tue, 15 Mar 2005 13:18:52 -0800 Subject: [uClinux-dev] Driver for Intel Ethernet Phy Message-ID: <006301c529a4$96c8e210$8d936e89@citpc11> All: We have teamed a TMS320VC5471 with an Intel LXT971A Dual-Speed Fast Ethernet 10/100 PHY Transceiver. Does anyone have a ucLinux driver for this combo? The DSP side of the 5471 looks like a C54xx. Best regards, Doug Douglas A. Palmer, PhD UCSD - Calit2 Email: dapalmer at ucsd.edu -------------- next part -------------- An HTML attachment was scrubbed... URL: From xwang at edgewaternetworks.com Tue Mar 15 16:34:39 2005 From: xwang at edgewaternetworks.com (Xiaobo) Date: Tue, 15 Mar 2005 13:34:39 -0800 Subject: [uClinux-dev] Telnet issue In-Reply-To: <20050315170006.CF6C41A05A@uclinux.org> Message-ID: <200503152136.j2FLalQ8084113@edgewaternetworks.com> we have a device running snapgear 3.1.0. After scanning it with nessus, it reported that the telnet is vulnerable (see link below). I suspect this is a false report since the problem is old and likely patched in snapgear distribution. Can somebody confirm this or a patch is needed to fix this problem for snapgear-3.1.0? http://www.securityfocus.com/bid/3064 thanks in advance Eric From jwilliams at itee.uq.edu.au Tue Mar 15 18:25:43 2005 From: jwilliams at itee.uq.edu.au (John Williams) Date: Wed, 16 Mar 2005 09:25:43 +1000 Subject: [uClinux-dev] [RFC] Proposal for getting LTP to run on uClinux In-Reply-To: <20050315161744.GS1111@conscoop.ottawa.on.ca> References: <20050315161744.GS1111@conscoop.ottawa.on.ca> Message-ID: <42376EF7.4090909@itee.uq.edu.au> Hi Jody, Jody McIntyre wrote: > Proposal for getting LTP to run on uClinux > Steamballoon Incorporated > Sponsored by Analog Devices [snip] I think this is an outstanding effort, and will be a very valuable contribution to uClinux as a whole. I don't have any specific comments on your approach, it all sounds reasonable and the willingness of the LTP maintainers to consider your integration approach is very encouraging. I have previously experimented with the kcov kernel converage tests from LTP, and have demonstrated the kcov kernel module under uClinux. I'll be happy to share this experience if it is of use in your efforts. Regards, John -- Dr John Williams, Research Fellow, Embedded Systems Group / Reconfigurable Computing School of ITEE, The University of Queensland, Brisbane, Australia (p) +61 7 33652185 (f) +61 7 33654999 (m) +61 403969243 From jsujjavanich at syntech-fuelmaster.com Tue Mar 15 18:53:05 2005 From: jsujjavanich at syntech-fuelmaster.com (Jate Sujjavanich) Date: Tue, 15 Mar 2005 18:53:05 -0500 Subject: [uClinux-dev] 2.6.x MTD, mounting root fs romfs BUG Message-ID: My configuartion is: MTD, MTD_CMDLINE_PARTS, MTD_CHAR, MTD_BLOCK_RO, MTD_RAM, MTD_ROM, MTD_UCLINUX ------------------------- I am attempting to mount romfs using MTD on linux 2.6.10, and I receive the following error: uclinux[mtd]: RAM probe address=0x111278 size=0xe9000 Creating 1 MTD partitions on "RAM": 0x00000000-0x000e9000 : "ROMfs" mtd: Giving out device 0 to ROMfs uclinux[mtd]: set ROMfs to be root filesystem kernel BUG at drivers/mtd/mtdcore.c:225! Kernel panic - not syncing: BUG! ------------------------- I made the following modifications to the drivers/mtd/maps/uclinux.c code. (see the defined(CONFIG_M5235EVB)) int __init uclinux_mtd_init(void) { struct mtd_info *mtd; struct map_info *mapp; extern char _ebss; #if defined(CONFIG_M5235EVB) mapp = &uclinux_ram_map; mapp->phys = (unsigned long) &_ebss; mapp->size = PAGE_ALIGN(*((unsigned long *)((&_ebss) + 8))); mapp->bankwidth = 4; mapp->map_priv_2 = (unsigned long) &_ebss; #else mapp = &uclinux_ram_map; mapp->phys = (unsigned long) &_ebss; mapp->size = PAGE_ALIGN(*((unsigned long *)((&_ebss) + 8))); mapp->bankwidth = 4; #endif printk("uclinux[mtd]: RAM probe address=0x%x size=0x%x\n", (int) mapp->map_priv_2, (int) mapp->size); mapp->virt = ioremap_nocache(mapp->phys, mapp->size); if (mapp->virt == 0) { printk("uclinux[mtd]: ioremap_nocache() failed\n"); return(-EIO); } ------------------------- My ROMFS is at that address (_ebss) by crt0_ram.S. It is bugging out on put_mtd_device(). Can anyone provide assistance? Jate Sujjavanich -------------- next part -------------- An HTML attachment was scrubbed... URL: From vanitha at agilis.stengg.com Wed Mar 16 04:16:48 2005 From: vanitha at agilis.stengg.com (Vanitha Ramaswami) Date: Wed, 16 Mar 2005 09:16:48 -0000 Subject: [uClinux-dev] signal safe mutex Message-ID: <01C52A08.E21A3340.vanitha@agilis.stengg.com> Hi All, I need to share a variable within a signal handler and from a process created using pthread. I think i cannot use the pthread_mutex_lock since its not safe to use from a signal handler. can someone suggest someother way of doing this. Thanks Vanitha [This e-mail is confidential and may be priviledged. If you are not the intended recipient, please kindly notify us immediately and delete the message from your system; please do not copy or use it for any purpose, nor disclose its contents to any other person. Thank you.] ---ST Electronics Group--- From gerg at snapgear.com Tue Mar 15 20:52:29 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Wed, 16 Mar 2005 11:52:29 +1000 Subject: [uClinux-dev] Telnet issue In-Reply-To: <200503152136.j2FLalQ8084113@edgewaternetworks.com> References: <200503152136.j2FLalQ8084113@edgewaternetworks.com> Message-ID: <4237915D.9040300@snapgear.com> Hi Xiaobo, Xiaobo wrote: > we have a device running snapgear 3.1.0. After scanning it with nessus, it > reported that the telnet is vulnerable (see link below). I suspect this is > a false report since the problem is old and likely patched in snapgear > distribution. Can somebody confirm this or a patch is needed to fix this > problem for snapgear-3.1.0? > > http://www.securityfocus.com/bid/3064 This doesn't really surprise me too much. The telnetd in the tree (and this means uClinux-dist too) is pretty old. Probably not been updated in 5 years (seriously :-(. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Tue Mar 15 20:54:27 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Wed, 16 Mar 2005 11:54:27 +1000 Subject: [uClinux-dev] [RFC] Proposal for getting LTP to run on uClinux In-Reply-To: <20050315161744.GS1111@conscoop.ottawa.on.ca> References: <20050315161744.GS1111@conscoop.ottawa.on.ca> Message-ID: <423791D3.3070509@snapgear.com> Hi Jody, Jody McIntyre wrote: > Proposal for getting LTP to run on uClinux > Steamballoon Incorporated > Sponsored by Analog Devices This sounds like a mighty fine idea! I take it you will be specifically target uClinux on blackfin? Do you plan to run it on any other architectures as part of your work? Regards Greg > 1. Issues with the LTP on uClinux > > Since the LTP was written for a desktop/server linux platform, there are some > issues with running it on a uClinux-based platform. The lack of fork(), > incomplete mmap() capabilities, and limited resources on the devices means that > all calls to fork() must be changed to vfork() and the functionality of each > test must be fixed to either exec() a separate child or the same program with > an appropriate argument. > > 2. Analysis of problematic cases > > 2.1 fork() > > There are four cases where fork() calls are used: > > - Either the parent does a wait()/waitpid() or the child does an exec() > immediately after the fork(). > > - The functionality of the child is in a well defined function that can easily > be called from an exec()ed copy of the program with a special argument. > > - The test needs to be completely refactored to work with vfork(). > > - The test is inappropriate for uClinux since it tests fork() itself. > > Also, the LTP allows multiple tests to be run simultaneously by parsing the -c > parameter and forking that many tests. If the tests are meant to be run > simultaneously then a method to exec() the children will be needed or only one > test will run at a time. Since we are targeting devices with limited > resources, this may not be an issue. > > We have identified 210 tests that use the fork() system call. > > 2.1.1 Parent does a wait / child does an exec > > There are 119 tests where fork() is used and the parent does a wait()/waitpid() > or the child does an exec(). All these tests can just be converted to using > vfork(). > > 2.1.2 Child/parent code needs to be split across exec > > There are 65 tests that are fairly straightforward to convert to > vfork()/exec(). The main challenge is in passing variables across exec() using > arguments or other methods, since parent variables will no longer be available > to the child. Functions to support this will be added to the test library. > Then, the tests will be converted to use them. This will mostly be cut and > paste work. An example of a test like this is kernel/syscalls/fcntl/fcntl11.c . > > 2.1.3 Test needs to be completely refactored > > 14 tests will need to be completely refactored to work with vfork(). > > 2.1.4 Inappropriate test > > The remaining 12 tests are tests of the fork() system call itself. These are > clearly inappropriate for a uClinux environment. > > 2.2 mmap() > > Most of the mmap() test cases will require changes to the options to the mmap() > call. Also, any mmap() call that expects a certain VM address to be returned > will fail. Tests which require this behavior are not appropriate for uClinux. > Any mmap() call that is expecting the mapping to be backed by disk will fail > unless the test is done read-only on a romfs file system. > > Of the 57 tests that have an mmap() call, 32 of them just require the removal > of MAP_PRIVATE on the options. 25 of the tests are not appropriate for > uClinux. We recommend putting the options for mmap() in a #define and using the > -DUCLINUX compile time flag (see below) to choose the appropriate options. > > 2.3 Build changes > > Changes will need to be made to the build system to differentiate a normal > (desktop/server) LTP build from a uClinux build. There are some tests that are > not appropriate for uClinux. We recommend adding a uClinux target to the > Makefile so that building for uClinux could be done using make uclinux instead > of make. > > This would also add the -DUCLINUX build flag which would be used in the source > to select fork()/vfork() behavior, etc.. > > 2.4 Run changes > > The LTP will need to be modified to be either runnable from NFS or through > rcp/rsh (as with gcc regression testing.) We recommend running over NFS since > it would require the fewest changes. NFS gives the appearance of a normal > filesystem and LTP can be run by mounting the directory where it was built and > running it from there. We tested this using i386 systems as the "uClinux > host" and "development server" and LTP runs correctly. > > 2.5 Library changes > > 2.5.1 vfork()/exec() > > Since many testcases will need to be able to re-exec() themselves and choose > the child code path, we will implement a generic self_exec wrapper that will > pass the correct arguments to the child process. We propose adding the -C > option for this purpose. Each test on startup will call into the library to > store the startup information and to branch to the child code path if > necessary. > > Making this as easy to use as possible increases the chances that future tests > will be written in such a way as to work in uClinux. > > 2.5.2 TEST_XXXX > > Re-implement TEST_XXXX functions for uClinux. This means converting fork() to > vfork()/exec(). This will use a similar mechanism to that described in section > 2.5.1. This will affect the -c and -i options of the tests. > > 3.0 Patches to LTP > > As work is accepted by ADI, we will be sending patches to the LTP maintainers. >>From correspondence with them, they would be willing to have the updates > integrated in the March release, as long as the patches don't affect their > currently supported platforms. This may not be a viable time-frame for ADI but > it does show their willingness to cover uClinux as well. Care must be taken to > maintain the same testing semantics on modified test cases. > > 4.0 Testing > > As the tests are modified their output will be compared to the original version > to make sure there are no differences. Once all the cases are done, the full > output of runltp will be compared with both sets. The resulting files will be > compared using diff to make sure the same results are observed using both. > > 5.0 Sample Change > > A sample change is attached as appendix A. This is a vfork()/exec() conversion > as described in section 2.1.2. Since the test library has not yet been > modified, all the necessary changes have been made directly to the test. Also, > vfork()/exec() will be used even on non-uClinux systems. This should not be > done in the final version. > > Appendix A: Sample Change: Attached separately. > > > ------------------------------------------------------------------------ > > This is a vfork()/exec() conversion as described in section 2.1.2 of our > proposal. Since the test library has not yet been modified, all the necessary > changes have been made directly to the test. Also, vfork()/exec() will be > used even on non-uClinux systems. This should not be done in the final > version. > > Index: fcntl/fcntl11.c > =================================================================== > --- fcntl.orig/fcntl11.c 2005-01-04 16:04:18.000000000 -0500 > +++ fcntl/fcntl11.c 2005-02-21 13:20:30.000000000 -0500 > @@ -32,6 +32,7 @@ > * > * HISTORY > * 07/2001 Ported by Wayne Boyer > + * 02/2005 Modified to work on uClinux as a demonstration. > * > * RESTRICTIONS > * None > @@ -40,6 +41,7 @@ > #include > #include > #include > +#include > #include > #include > #include > @@ -298,18 +300,52 @@ catch_child() > cleanup(); > } > > +static int child = 0; > +static char *child_arg; > +static option_t user_options[] = { > + { "C:", &child, &child_arg }, /* -C option */ > + { NULL, NULL, NULL } > +}; > + > +static void user_opts_help() > +{ > + printf(" -C ARG Run the child process with arguments ARG " > + "(for internal use)\n"); > +} > + > int main(int ac, char **av) > { > struct flock tl; > > int lc; /* loop counter */ > char *msg; /* message returned from parse_opts */ > + char *start_cwd; /* Store cwd, for exec after vfork */ > > /* parse standard options */ > - if ((msg = parse_opts(ac, av, (option_t *)NULL, NULL)) != (char *)NULL){ > + if ((msg = parse_opts(ac, av, user_options, &user_opts_help)) != > + (char *)NULL){ > tst_brkm(TBROK, cleanup, "OPTION PARSING ERROR - %s", msg); > } > > + start_cwd = getcwd(NULL, 0); > + if (child) { > + char *child_dir; > + > + child_dir = strtok(child_arg, ","); > + parent_pipe[0] = strtol(strtok(NULL, ","), NULL, 10); > + parent_pipe[1] = strtol(strtok(NULL, ","), NULL, 10); > + child_pipe[0] = strtol(strtok(NULL, ","), NULL, 10); > + child_pipe[1] = strtol(strtok(NULL, ","), NULL, 10); > + fd = strtol(strtok(NULL, ","), NULL, 10); > + > + if (chdir(child_dir) < 0) { > + tst_resm(TFAIL, "chdir failed"); > + exit(1); > + } > + > + do_child(); > + } > + > setup(); /* global setup */ > > /* Check for looping state if -i option is given */ > @@ -317,8 +353,23 @@ int main(int ac, char **av) > /* reset Tst_count in case we are looping */ > Tst_count = 0; > > - if ((child_pid = fork()) == 0) { /* parent */ > - do_child(); > + if ((child_pid = vfork()) == 0) { /* child */ > + char *tmp_cwd; > + char arg[1000]; > + > + tmp_cwd = getcwd(NULL, 0); > + chdir(start_cwd); > + > + sprintf(arg, "%s,%d,%d,%d,%d,%d", tmp_cwd, > + parent_pipe[0], parent_pipe[1], > + child_pipe[0], child_pipe[1], fd); > + > + if (execlp(av[0], av[0], "-C", arg, > + (char *) NULL) < 0) { > + perror("exec"); > + tst_resm(TFAIL, "exec failed"); > + } > + exit(1); > } else if (child_pid < 0) { > tst_resm(TFAIL, "Fork failed"); > cleanup(); > > > ------------------------------------------------------------------------ > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Tue Mar 15 20:56:05 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Wed, 16 Mar 2005 11:56:05 +1000 Subject: [uClinux-dev] Driver for Intel Ethernet Phy In-Reply-To: <006301c529a4$96c8e210$8d936e89@citpc11> References: <006301c529a4$96c8e210$8d936e89@citpc11> Message-ID: <42379235.1090702@snapgear.com> Hi Douglas, dapalmer wrote: > We have teamed a TMS320VC5471 with an Intel LXT971A Dual-Speed Fast > Ethernet 10/100 PHY Transceiver. Does anyone have a ucLinux driver for > this combo? The DSP side of the 5471 looks like a C54xx. I can't say I have tried this combination. But it is not so much a new driver you need. You need to modify the current FEC driver (~/drivers/net/fec.c) to detect and support this PHY. That is a much easier task, the FEC driver already supports multiple PHY types. It is easy to add new ones. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From jonc at neb.rr.com Tue Mar 15 20:57:19 2005 From: jonc at neb.rr.com (jonc) Date: Tue, 15 Mar 2005 19:57:19 -0600 Subject: [uClinux-dev] problem to write/read to IO port In-Reply-To: References: Message-ID: <200503151957.20146.jonc@neb.rr.com> Hi Fabrice, Checkout: ??????"mcf5272 gpio sample code ? " ????????????????mailman.uclinux.org/pipermail/uclinux-dev/2003-April/016491.html There is an attachment for code that will work on a mcf5272. Mike is right about other ColdFires such as the mcf5282 that have Peripheral Access Control Registers which must be set first before peripherals can be accessed from user space. Once you get the hang of read and writing from user space, then writing a device driver will be easier. Jon C. On Tuesday 15 March 2005 8:33 am, Fabrice Poulin wrote: > Hi, > > I work with the coldfire 5272 but I already asked the question about acess > to IO port. > In fact, there isn't device /dev/port to write or read with uClinux on the > 5272, so how I do to write or read on the Port A for exemple? > If someboby has an example code, I would like this...? > > Thank > > Fabrice > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org From gerg at snapgear.com Tue Mar 15 21:04:43 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Wed, 16 Mar 2005 12:04:43 +1000 Subject: [uClinux-dev] 2.6.x MTD, mounting root fs romfs BUG In-Reply-To: References: Message-ID: <4237943B.2060609@snapgear.com> Hi Jate, Jate Sujjavanich wrote: > My configuartion is: > > MTD, MTD_CMDLINE_PARTS, MTD_CHAR, MTD_BLOCK_RO, MTD_RAM, MTD_ROM, > MTD_UCLINUX > > ------------------------- > I am attempting to mount romfs using MTD on linux 2.6.10, and I receive Are you using un-modified linux-2.6.10 or linux-2.6.10-uc0? > the following error: > > uclinux[mtd]: RAM probe address=0x111278 size=0xe9000 > Creating 1 MTD partitions on "RAM": > 0x00000000-0x000e9000 : "ROMfs" > mtd: Giving out device 0 to ROMfs > uclinux[mtd]: set ROMfs to be root filesystem > kernel BUG at drivers/mtd/mtdcore.c:225! > Kernel panic - not syncing: BUG! There is a couple of fixups for uclinux.c in linux-2.6.10-uc0. For one thing it should not be calling mtd_put_device() like it does in the original uclinux.c in 2.6.10. Maybe your problem here. Regards Greg > ------------------------- > I made the following modifications to the drivers/mtd/maps/uclinux.c > code. (see the defined(CONFIG_M5235EVB)) > > int __init uclinux_mtd_init(void) > { > struct mtd_info *mtd; > struct map_info *mapp; > extern char _ebss; > > #if defined(CONFIG_M5235EVB) > mapp = &uclinux_ram_map; > mapp->phys = (unsigned long) &_ebss; > mapp->size = PAGE_ALIGN(*((unsigned long *)((&_ebss) + 8))); > mapp->bankwidth = 4; > mapp->map_priv_2 = (unsigned long) &_ebss; > #else > mapp = &uclinux_ram_map; > mapp->phys = (unsigned long) &_ebss; > mapp->size = PAGE_ALIGN(*((unsigned long *)((&_ebss) + 8))); > mapp->bankwidth = 4; > #endif > printk("uclinux[mtd]: RAM probe address=0x%x size=0x%x\n", > (int) mapp->map_priv_2, (int) mapp->size); > > mapp->virt = ioremap_nocache(mapp->phys, mapp->size); > > if (mapp->virt == 0) { > printk("uclinux[mtd]: ioremap_nocache() failed\n"); > return(-EIO); > } > > ------------------------- > My ROMFS is at that address (_ebss) by crt0_ram.S. It is bugging out on > put_mtd_device(). Can anyone provide assistance? > > > Jate Sujjavanich > > > > ------------------------------------------------------------------------ > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From bernie at develer.com Tue Mar 15 22:21:28 2005 From: bernie at develer.com (Bernardo Innocenti) Date: Wed, 16 Mar 2005 04:21:28 +0100 Subject: [uClinux-dev] PATCH: Missing headers in elf2flt/flthdr.c Message-ID: <4237A638.6080705@develer.com> Hello, I've committed this patch as obvious: --- flthdr.c.orig 2005-03-16 04:12:42.000000000 +0100 +++ flthdr.c 2005-03-16 04:13:07.000000000 +0100 @@ -10,6 +10,8 @@ #include /* Userland pieces of the ANSI C standard I/O package */ #include /* Userland prototypes of the Unix std system calls */ #include +#include /* exit() */ +#include /* strcat(), strcpy() */ /* macros for conversion between host and (internet) network byte order */ #include /* Consts and structs defined by the internet system */ -- // Bernardo Innocenti - Develer S.r.l., R&D dept. \X/ http://www.develer.com/ From bernie at develer.com Tue Mar 15 22:24:21 2005 From: bernie at develer.com (Bernardo Innocenti) Date: Wed, 16 Mar 2005 04:24:21 +0100 Subject: [uClinux-dev] Re: Toolchain problem with Amd64 In-Reply-To: <422447AB.6050304@carroll.de> References: <422447AB.6050304@carroll.de> Message-ID: <4237A6E5.3050806@develer.com> Sorry for the very long delay... I was busy at the time I received your mail and I forgot to get back to it. Jens Carroll wrote: > first of all thanx for the great work you have done with the newest > GCC-3.x toolchain (uclinux-tools-20050221). > However I did find a little problem with the elf2flt tools. I have > created the arm-toolchain with a Amd64 machine and it seems that the > allignment has been taken from the cross compile machine (64bit) rather > than the expected 32-bit for the arm-family. > > I have attached a simple hello-world binary which you could open with a > hex-editor to see the wrong allignment. > I have build the same toolchain with an Amd thunderbird and everything > compiles and works just dandy. > > If you find the problem and create a patch, I am happy to test it in > case you haven't a Amd64 around. Oops. I've seen where the bug lies. It's in flat.h: the types of the fields in struct flat_hdr were specified as "unsigned long" where they should have been "uint32_t" etc. The following patch should fix the problem, but I can't test on amd64. Please apply it to your local copy of elf2flt and report if you still see any difference in the output. I will apply it to the master elf2flt copy after your confirmation and if David agrees. --- flat.h.orig 2005-03-16 04:10:20.000000000 +0100 +++ flat.h 2005-03-16 04:18:15.000000000 +0100 @@ -11,7 +11,10 @@ #define _LINUX_FLAT_H #ifdef __KERNEL__ -#include +# include +# include /* uint32_t, int32_t */ +#else +# include /* uint32_t, int32_t */ #endif #define FLAT_VERSION 0x00000004L @@ -29,25 +32,25 @@ struct flat_hdr { char magic[4]; - unsigned long rev; /* version (as above) */ - unsigned long entry; /* Offset of first executable instruction + uint32_t rev; /* version (as above) */ + uint32_t entry; /* Offset of first executable instruction with text segment from beginning of file */ - unsigned long data_start; /* Offset of data segment from beginning of + uint32_t data_start; /* Offset of data segment from beginning of file */ - unsigned long data_end; /* Offset of end of data segment + uint32_t data_end; /* Offset of end of data segment from beginning of file */ - unsigned long bss_end; /* Offset of end of bss segment from beginning + uint32_t bss_end; /* Offset of end of bss segment from beginning of file */ /* (It is assumed that data_end through bss_end forms the bss segment.) */ - unsigned long stack_size; /* Size of stack, in bytes */ - unsigned long reloc_start; /* Offset of relocation records from + uint32_t stack_size; /* Size of stack, in bytes */ + uint32_t reloc_start; /* Offset of relocation records from beginning of file */ - unsigned long reloc_count; /* Number of relocation records */ - unsigned long flags; - unsigned long build_date; /* When the program/library was built */ - unsigned long filler[5]; /* Reservered, set to zero */ + uint32_t reloc_count; /* Number of relocation records */ + uint32_t flags; + uint32_t build_date; /* When the program/library was built */ + uint32_t filler[5]; /* Reservered, set to zero */ }; #define FLAT_FLAG_RAM 0x0001 /* load program entirely into RAM */ @@ -75,22 +78,22 @@ struct flat_hdr { #define OLD_FLAT_RELOC_TYPE_BSS 2 typedef union { - unsigned long value; + uint32_t value; struct { # if defined(mc68000) && !defined(CONFIG_COLDFIRE) - signed long offset : 30; - unsigned long type : 2; -# define OLD_FLAT_FLAG_RAM 0x1 /* load program entirely into RAM */ + int32_t offset : 30; + uint32_t type : 2; +# define OLD_FLAT_FLAG_RAM 0x1 /* load program entirely into RAM */ # elif defined(__BIG_ENDIAN_BITFIELD) - unsigned long type : 2; - signed long offset : 30; -# define OLD_FLAT_FLAG_RAM 0x1 /* load program entirely into RAM */ + uint32_t type : 2; + int32_t offset : 30; +# define OLD_FLAT_FLAG_RAM 0x1 /* load program entirely into RAM */ # elif defined(__LITTLE_ENDIAN_BITFIELD) - signed long offset : 30; - unsigned long type : 2; -# define OLD_FLAT_FLAG_RAM 0x1 /* load program entirely into RAM */ + int32_t offset : 30; + uint32_t type : 2; +# define OLD_FLAT_FLAG_RAM 0x1 /* load program entirely into RAM */ # else -# error "Unknown bitfield order for flat files." +# error "Unknown bitfield order for flat files." # endif } reloc; } flat_v2_reloc_t; -- // Bernardo Innocenti - Develer S.r.l., R&D dept. \X/ http://www.develer.com/ From yashi at atmark-techno.com Tue Mar 15 22:53:06 2005 From: yashi at atmark-techno.com (Yasushi SHOJI) Date: Wed, 16 Mar 2005 12:53:06 +0900 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <4236C437.6060404@snapgear.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> Message-ID: <20050316035306.B2F226373@mail2.atmark-techno.com> At Tue, 15 Mar 2005 21:17:11 +1000, Greg Ungerer wrote: > > Hi All, > > Phil Wilshire wrote: > > Hi > > Why not call your version the ATlinux-dist. > > > > Or should we come up with a new name for the whole distribution since it > > is not just > > MMULess any more. > > I get that comment a bit. And I have spoken to many people who > are confused by the fact that uClinux-dist (heck, even just > the uClinux kernel sources) work with VM processors. > > Do we need a new name? That'd help _me_ ;) not sure what we should name it though. I think the question is what do we want the dist to be. if we know the answer, we can name the dist after it. some easy one would be embedded-linux-dist ;) -- yashi I've been thinking this for a while. embedded-debian is also thinking about _how_ to adopt the debian way into embedded system. bitbake is one solution for that. uClinux is, for sure, another solution. ipkg system from handhelds.org is doing great job, too. but, most projects have pros and cons. ie. for uClinux-dist what happens if we contiune to add great free/open software to the dist? how big the tarball is acceptable? would bitbake way be better if it can redule the required memmory usage? how about the debian or even gentoo way? is anyone talking to other project? can we help each other to dominate the world of embedded system with linux? or is embedded linux too broad topic? sure there isn't a silver bullet, and variety of choice is a good thing. but it'd be nice if we can colabolate to reduce the duplicated work. From marny at rambler.ru Wed Mar 16 02:33:17 2005 From: marny at rambler.ru (Igor Marnat) Date: Wed, 16 Mar 2005 10:33:17 +0300 Subject: [uClinux-dev] Driver for Intel Ethernet Phy In-Reply-To: <006301c529a4$96c8e210$8d936e89@citpc11> References: <006301c529a4$96c8e210$8d936e89@citpc11> Message-ID: <200503161033.18340.marny@rambler.ru> On 16 March 2005 00:18, dapalmer wrote: > We have teamed a TMS320VC5471 with an Intel LXT971A Dual-Speed Fast > Ethernet 10/100 PHY Transceiver. Does anyone have a ucLinux driver for > this combo? The DSP side of the 5471 looks like a C54xx. You can just review http://cvs.uclinux.org/cgi-bin/cvsweb.cgi/uClinux-2.4.x/drivers/net/eth_c5471hw.c and add support for your new device there. Regards, Igor Marnat From mcalamelli at gmail.com Wed Mar 16 08:41:59 2005 From: mcalamelli at gmail.com (Massimiliano Calamelli) Date: Wed, 16 Mar 2005 14:41:59 +0100 Subject: [uClinux-dev] Kernel 2.6.X and CVS Message-ID: <2c2ac6c20503160541767c3907@mail.gmail.com> There's a CVS repository for 2.6.x uclinux source tree? thanks Massimiliano From jsujjavanich at syntech-fuelmaster.com Wed Mar 16 09:33:48 2005 From: jsujjavanich at syntech-fuelmaster.com (Jate Sujjavanich) Date: Wed, 16 Mar 2005 09:33:48 -0500 Subject: [uClinux-dev] linux-2.6.10-uc0, 11-uc0 romptr error Message-ID: This is for both the M5282EVB and my M5235EVB port for linux 2.6.10-uc0 and .11-uc0. I have upgraded to -uc0, and I get the following error on make: mm/nommu.c: In function `do_mmap_pgoff': mm/nommu.c:615: error: structure has no member named `romptr' mm/nommu.c:616: error: structure has no member named `romptr' make[2]: *** [mm/nommu.o] Error 1 Any hints? Jate Sujjavanich -------------- next part -------------- An HTML attachment was scrubbed... URL: From wnlee at cad.csie.ncku.edu.tw Wed Mar 16 09:40:46 2005 From: wnlee at cad.csie.ncku.edu.tw (Wayne Lee) Date: Wed, 16 Mar 2005 22:40:46 +0800 Subject: [uClinux-dev] RE: Is smc91x.c the best choice for smsc91c111 chip in uClinux 2.6? In-Reply-To: Message-ID: <200503161443.j2GEhaKE026942@cad.csie.ncku.edu.tw> Hi Heinz, It is nice to hear about your works. However, there is _not_ your smc91c111 2.6 patch in uClinux-dev. Do you post it on somewhere else? The google search result about your posts, http://www.google.com.tw/search?hl=zh-TW&q=Heinz+Haeberle+site%3Amailman.ucl inux.org%2Fpipermail%2Fuclinux-dev%2F&meta= Thanks. Wayne -----Original Message----- From: Heinz Haeberle [mailto:H.Haeberle at pilz.de] Sent: Wednesday, March 16, 2005 9:11 PM To: uclinux-dev at uClinux.org Cc: wnlee at cad.csie.ncku.edu.tw Subject: Is smc91x.c the best choice for smsc91c111 chip in uClinux 2.6? Since I was starting with 2.6.6 without a 91c111 driver at all I have ported it over from the 2.4 kernel. Guess I have posted it already. But if there is a maintream driver for the smsc 91x chips I will probably go with this in the future. Will see how it looks like. I use a 5249C3 like board Heinz From swenggsri at yahoo.com Wed Mar 16 09:58:46 2005 From: swenggsri at yahoo.com (Sriram Chadalavada) Date: Wed, 16 Mar 2005 06:58:46 -0800 (PST) Subject: [uClinux-dev] URGENT - kernel panic with uCdimm serial driver - KISS AX.25 interface Message-ID: <20050316145846.59850.qmail@web21604.mail.yahoo.com> Hi everyone, I am trying to get a KISS AX.25 network interface working with uClinux 2.4.24 on a uCdimm (Motorola MC68VZ328). I have ported an application 'kissattach' that runs as a daemon to bring up a KISS network interface. THe idea is to connect a hamradio to one of the serial ports (in this case port 1) and get TCP/IP running over the hamradio link. # ifconfig ax0 ax0 Link encap:AMPR AX.25 HWaddr KC0POH-9 inet addr:192.168.0.9 Bcast:192.168.0.255 Mask:255.255.255.0 UP BROADCAST RUNNING MTU:512 Metric:1 -- -- ---------------------- ------------------------------ However, when I try to ping this interface from an x86 box or receive a ping reply, the kernel panics (see attachment). The problem does not occur when this test is carried on a normal x86 box (same kernel). Since MKISS driver(linux-2.4.x/drivers/net/hamradio/mkiss.c) and kernel AX.25/network infrastructure seem pretty much same for both architectures, the difference seems to me to be in the way the uCdimm serial driver works. I would really appreciate if experts with experience in uCdimm serial driver, kernel network stack development can help me solve this problem or provide suggestions on how to proceed. The communication system of our project(www.engr.ku.edu/~kutesat) is held up due to the problem and any assistance will be duly credited. Thanks in advance, Sriram __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/ -------------- next part -------------- A non-text attachment was scrubbed... Name: ax25-kernel-panic Type: application/octet-stream Size: 1284 bytes Desc: ax25-kernel-panic URL: From rgetz at blackfin.uclinux.org Wed Mar 16 13:27:46 2005 From: rgetz at blackfin.uclinux.org (Robin Getz) Date: Wed, 16 Mar 2005 13:27:46 -0500 Subject: [uClinux-dev] [RFC] Proposal for getting LTP to run on uClinux Message-ID: <6.1.1.1.0.20050316131246.01dde578@ptg1.spd.analog.com> Greg asked: >Jody McIntyre wrote: > > Proposal for getting LTP to run on uClinux Steamballoon Incorporated > > Sponsored by Analog Devices > >This sounds like a mighty fine idea! >I take it you will be specifically target uClinux on blackfin? While the main development targets will be the Blackfin processor (based on the work that we have done at http://blackfin.uclinux.org ) the overall goal is to get LTP working on any uClinux/uClibc/nommu based platform. >Do you plan to run it on any other architectures as part of your work? Yes - Jody and the nice folks at Steamballoon will be working with things on two uClinux/uClibc/nommu platforms - the BF533 based Blackfin processor, and - a uCdimm based product from Arcturus Networks http://www.arcturusnetworks.com/ucdimm.shtml. This is to rule out bugs in the Blackfin port causing issues in the tests. Any uClinux/uClibc/nommu based system that supports some type of NFS mount or NFS boot should be able to run the tests. When this is done, we will be hosting a tinderbox (http://www.mozilla.org/tinderbox.html), which runs this test continually on our cvs kernel - to find/fix issues before they make it into a release. This is something that the OSDL (http://tinderbox.osdl.org/) is doing on the mainline kernel - we thought in order to prove stability, we should do it on our hardware. I would be interested in people's thoughts on the idea. -Robin From jsujjavanich at syntech-fuelmaster.com Wed Mar 16 17:35:25 2005 From: jsujjavanich at syntech-fuelmaster.com (Jate Sujjavanich) Date: Wed, 16 Mar 2005 17:35:25 -0500 Subject: [uClinux-dev] 2.6.10-uc0 fec.c, inconsistent operand constraints - resend Message-ID: I am working on activating the FEC for the Coldfire 5235 under linux 2.6.10-uc0. I turned on the mii option, and now it breaks. First off - do I need this? If so, what could the problem be? include/asm/bitops.h: In function `mii_check_link': include/asm/bitops.h:197: warning: read-write constraint does not allow a register include/asm/bitops.h:197: warning: read-write constraint does not allow a register include/asm/bitops.h:176: warning: read-write constraint does not allow a register include/asm/bitops.h:176: warning: read-write constraint does not allow a register include/asm/bitops.h:197: error: inconsistent operand constraints in an `asm' make[3]: *** [drivers/net/mii.o] Error 1 Jate Sujjavanich From H.Haeberle at pilz.de Wed Mar 16 08:10:59 2005 From: H.Haeberle at pilz.de (Heinz Haeberle) Date: Wed, 16 Mar 2005 14:10:59 +0100 Subject: [uClinux-dev] Is smc91x.c the best choice for smsc91c111 chip in uClinux 2.6? In-Reply-To: <20050315170008.B543D1A083@uclinux.org> Message-ID: Since I was starting with 2.6.6 without a 91c111 driver at all I have ported it over from the 2.4 kernel. Guess I have posted it already. But if there is a maintream driver for the smsc 91x chips I will probably go with this in the future. Will see how it looks like. I use a 5249C3 like board Heinz From swenggsri at yahoo.com Wed Mar 16 01:38:41 2005 From: swenggsri at yahoo.com (Sriram Chadalavada) Date: Tue, 15 Mar 2005 22:38:41 -0800 (PST) Subject: [uClinux-dev] URGENT - kernel panic with uCdimm serial driver - KISS AX.25 interface Message-ID: <20050316063841.68118.qmail@web21602.mail.yahoo.com> Hi everyone, I am trying to get a KISS AX.25 network interface working with uClinux 2.4.24 on a uCdimm (Motorola MC68VZ328). I have ported an application 'kissattach' that runs as a daemon to bring up a KISS network interface. THe idea is to connect a hamradio to one of the serial ports (in this case port 1) and get TCP/IP running over the hamradio link. # ifconfig ax0 ax0 Link encap:AMPR AX.25 HWaddr KC0POH-9 inet addr:192.168.0.9 Bcast:192.168.0.255 Mask:255.255.255.0 UP BROADCAST RUNNING MTU:512 Metric:1 -- -- ---------------------- ------------------------------ However, when I try to ping this interface from an x86 box or receive a ping reply, the kernel panics (see attachment). The problem does not occur when this test is carried on a normal x86 box (same kernel). Since MKISS driver (linux-2.4.x/drivers/net/hamradio/mkiss.c) and kernel AX.25/network infrastructure seem pretty much same for both architectures, the difference seems to me to be in the way the uCdimm serial driver works. I would really appreciate if experts with experience in uCdimm serial driver, kernel network stack development can help me solve this problem or provide suggestions on how to proceed. The communication system of our project(www.engr.ku.edu/~kutesat) is held up due to the problem and any assistance will be duly credited. Thanks in advance, Sriram __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/ -------------- next part -------------- A non-text attachment was scrubbed... Name: ax25-kernel-panic Type: application/octet-stream Size: 1284 bytes Desc: ax25-kernel-panic URL: From fabrice.poulin at medianesysteme.com Wed Mar 16 05:21:49 2005 From: fabrice.poulin at medianesysteme.com (Fabrice Poulin) Date: Wed, 16 Mar 2005 11:21:49 +0100 Subject: [uClinux-dev] problem with outb Message-ID: Hi, I would like to use "outb" but I have always this error: undefined reference to 'outb' collect2: ld returned 1 exit status However, I use the option -o2 during my compilation, So I don't understand... Somebody has an idea? Thank Fabrice From gerg at snapgear.com Wed Mar 16 06:00:55 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Wed, 16 Mar 2005 21:00:55 +1000 Subject: [uClinux-dev] Driver for Intel Ethernet Phy In-Reply-To: <42379235.1090702@snapgear.com> References: <006301c529a4$96c8e210$8d936e89@citpc11> <42379235.1090702@snapgear.com> Message-ID: <423811E7.9080801@snapgear.com> Hi Douglas, Oops, sorry, I thought you where referring to the 5271... Regards Greg Greg Ungerer wrote: > Hi Douglas, > > dapalmer wrote: > >> We have teamed a TMS320VC5471 with an Intel LXT971A Dual-Speed Fast >> Ethernet 10/100 PHY Transceiver. Does anyone have a ucLinux driver >> for this combo? The DSP side of the 5471 looks like a C54xx. > > > I can't say I have tried this combination. But it is not so much a > new driver you need. You need to modify the current FEC driver > (~/drivers/net/fec.c) to detect and support this PHY. That is a > much easier task, the FEC driver already supports multiple PHY > types. It is easy to add new ones. > > Regards > Greg > > > > ------------------------------------------------------------------------ > Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com > SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 > 825 Stanley St, FAX: +61 7 3891 3630 > Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com > -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From jsujjavanich at syntech-fuelmaster.com Wed Mar 16 14:20:33 2005 From: jsujjavanich at syntech-fuelmaster.com (Jate Sujjavanich) Date: Wed, 16 Mar 2005 14:20:33 -0500 Subject: [uClinux-dev] 2.6.10-uc0 fec.c, inconsistent operand constraints Message-ID: I am working on activating the FEC for the Coldfire 5235 under linux 2.6.10-uc0. I turned on the mii option, and now it breaks. First off - do I need this? If so, what could the problem be? include/asm/bitops.h: In function `mii_check_link': include/asm/bitops.h:197: warning: read-write constraint does not allow a register include/asm/bitops.h:197: warning: read-write constraint does not allow a register include/asm/bitops.h:176: warning: read-write constraint does not allow a register include/asm/bitops.h:176: warning: read-write constraint does not allow a register include/asm/bitops.h:197: error: inconsistent operand constraints in an `asm' make[3]: *** [drivers/net/mii.o] Error 1 Jate Sujjavanich -------------- next part -------------- An HTML attachment was scrubbed... URL: From saiful at modular-corp.com Tue Mar 15 21:25:31 2005 From: saiful at modular-corp.com (saiful) Date: Wed, 16 Mar 2005 10:25:31 +0800 Subject: [uClinux-dev] root password for Soekris Message-ID: <200503161025.AA3126460752@modular-corp.com> Hi Wayne, Thanks for your suggestion. I did that but still the root login password '12345' is incorrect. I can go the shell without login but I need to use the tinylogin facility. I wonder why /etc/securetty and /etc/shadow were not generated by Soekris even both options were selected during make menuconfig. Saiful Saiful ---------- Original Message ---------------------------------- From: "Wayne Lee" Date: Tue, 15 Mar 2005 23:52:39 +0800 > >Try this /etc/passwd, passwd is "12345" >root:ab80DAu2C3JN6:0:0::/:/bin/sh > >You can also choose the way into shell without login. > >Wayne > >-----Original Message----- >From: uclinux-dev-bounces at uClinux.org >[mailto:uclinux-dev-bounces at uClinux.org] On Behalf Of saiful >Sent: Tuesday, March 15, 2005 2:36 PM >To: saiful; David McCullough >Cc: uclinux-dev at uClinux.org >Subject: Re: [uClinux-dev] root password for Soekris > >Hi David, > >I am using glibc. The errror is "login incorrect". > >If I edit the passwd file, what is the password entry for "root" if i just >want make return key as my passowrd ? > >Saiful > > >---------- Original Message ---------------------------------- >From: David McCullough >Date: Tue, 15 Mar 2005 15:49:33 +1000 > >> >>Jivin saiful lays it down ... >>> Hi, >>> >>> I boot up Soekris from my IDE flash. As stated in Documentation the >>> root password for the Soekris login is 'default' but I still cannot >>> login. Is there a new root password for Soekris. >> >>Are you using uClibc or glibc ? >> >>It possible something is wrong with the config. Check that the Makefile >>is adding a /etc/passwd or /etc/config/passwd (if flatfsd is enabled). >> >>You might want to try making your own passwd file to see if that helps. >> >>Is it telling you that the password is incorrect or is it some other >>error ? >> >>Cheers, >>Davidm >> >>-- >>David McCullough, davidm at snapgear.com Ph:+61 7 34352815 >http://www.SnapGear.com >>Custom Embedded Solutions + Security Fx:+61 7 38913630 >http://www.uCdot.org >> >_______________________________________________ >uClinux-dev mailing list >uClinux-dev at uClinux.org >http://mailman.uclinux.org/mailman/listinfo/uclinux-dev >This message was resent by uclinux-dev at uclinux.org > > > From H.Haeberle at pilz.de Wed Mar 16 09:45:24 2005 From: H.Haeberle at pilz.de (Heinz Haeberle) Date: Wed, 16 Mar 2005 15:45:24 +0100 Subject: [uClinux-dev] RE: Is smc91x.c the best choice for smsc91c111 chip in uClinux 2.6? In-Reply-To: <200503161443.j2GEhaKE026942@cad.csie.ncku.edu.tw> Message-ID: > It is nice to hear about your works. > > However, there is _not_ your smc91c111 2.6 patch in uClinux-dev. > Do you post it on somewhere else? Here it is (See attached file: smc91111.h)(See attached file: smc91111.c) -------------- next part -------------- A non-text attachment was scrubbed... Name: smc91111.h Type: application/octet-stream Size: 20509 bytes Desc: not available URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: smc91111.c Type: application/octet-stream Size: 106517 bytes Desc: not available URL: From dhowells at redhat.com Wed Mar 16 11:14:07 2005 From: dhowells at redhat.com (David Howells) Date: Wed, 16 Mar 2005 16:14:07 +0000 Subject: [uClinux-dev] linux-2.6.10-uc0, 11-uc0 romptr error In-Reply-To: References: Message-ID: <14708.1110989647@redhat.com> Jate Sujjavanich wrote: > > This is for both the M5282EVB and my M5235EVB port for linux 2.6.10-uc0 and > .11-uc0. I have upgraded to -uc0, and I get the following error on make: > > mm/nommu.c: In function `do_mmap_pgoff': > mm/nommu.c:615: error: structure has no member named `romptr' > mm/nommu.c:616: error: structure has no member named `romptr' > make[2]: *** [mm/nommu.o] Error 1 > > Any hints? Yes. romptr should no longer be necessary. With a patch currently residing in Andrew Morton's 2.6.11-mm4 kernel, it should now be sufficient to do the following steps: (1) Create a backing_dev_info struct for ROMFS mappings that specify the capabilities as: BDI_CAP_NO_ACCT_DIRTY | BDI_CAP_NO_WRITEBACK | BDI_CAP_MAP_COPY | BDI_CAP_MAP_DIRECT | BDI_CAP_READ_MAP | BDI_CAP_EXEC_MAP This specifies that pages on this inode shouldn't be accounted dirty and shouldn't be written back. Data can be copied from romfs if need be, but can be mapped directly for a read-only or executable mapping, assuming the underlying ROM permits reading for execution. Writable mappings will be copied, and the copy mapped. (2) Provide a file->f_op->get_unmapped_area() method to indicate the address at which a mapping should be based. Note that the address returned by this method does not need to be page aligned, even though the file offset must be. (3) Provide a file->f_op->mmap() method just to say that the file is mappable. It doesn't have to actually do anything other than return 0. (4) Provide a file->f_op->read() method so that mmap can read the file when taking a copy. (5) Have the binfmt or whatever make a PROT_READ|PROT_EXEC, MAP_PRIVATE mapping for an executable's text segment as normal. The R/O segments will be mapped directly off of the ROM if possible, or else read into a bit of memory, which will then be shared. (6) Have the binfmt or whatever make a PROT_READ|PROT_WRITE, MAP_PRIVATE mapping for an executable's data segment as normal. The R/W segments will be read into a chunk of memory and kept private. David From scjody at steamballoon.com Wed Mar 16 10:25:28 2005 From: scjody at steamballoon.com (Jody McIntyre) Date: Wed, 16 Mar 2005 10:25:28 -0500 Subject: [uClinux-dev] [RFC] Proposal for getting LTP to run on uClinux In-Reply-To: <423791D3.3070509@snapgear.com> References: <20050315161744.GS1111@conscoop.ottawa.on.ca> <423791D3.3070509@snapgear.com> Message-ID: <20050316152528.GX1111@conscoop.ottawa.on.ca> On Wed, Mar 16, 2005 at 11:54:27AM +1000, Greg Ungerer wrote: > > This sounds like a mighty fine idea! > I take it you will be specifically target uClinux on blackfin? That is correct. > Do you plan to run it on any other architectures as part of > your work? Other than making sure the tests results are unchanged on non-uClinux (i386), no, but I won't be doing anything blackfin-specific, so the work should be useful on other architectures. Jody > Regards > Greg From gerg at snapgear.com Wed Mar 16 18:59:07 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Thu, 17 Mar 2005 09:59:07 +1000 Subject: [uClinux-dev] Kernel 2.6.X and CVS In-Reply-To: <2c2ac6c20503160541767c3907@mail.gmail.com> References: <2c2ac6c20503160541767c3907@mail.gmail.com> Message-ID: <4238C84B.7000904@snapgear.com> Hi Massimiliano, Massimiliano Calamelli wrote: > There's a CVS repository for 2.6.x uclinux source tree? No CVS for uClinux-2.6.x. Only the -uc patch sets. (This simplifies my keeping 2.6 up to date, so I don't have any plan to start a CVS for it either). Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Wed Mar 16 19:05:25 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Thu, 17 Mar 2005 10:05:25 +1000 Subject: [uClinux-dev] linux-2.6.10-uc0, 11-uc0 romptr error In-Reply-To: References: Message-ID: <4238C9C5.204@snapgear.com> Hi Jate, Jate Sujjavanich wrote: > This is for both the M5282EVB and my M5235EVB port for linux 2.6.10-uc0 > and .11-uc0. I have upgraded to -uc0, and I get the following error on make: > > mm/nommu.c: In function `do_mmap_pgoff': > mm/nommu.c:615: error: structure has no member named `romptr' > mm/nommu.c:616: error: structure has no member named `romptr' > make[2]: *** [mm/nommu.o] Error 1 > > Any hints? Do you have the kernel option CONFIG_MAGIC_ROM_PTR enabled? The "romptr" field should be conditionaly compiled in for the block_device_operations structure in ~/linux-2.6.x/include/linux/fs.h. It should look like: struct block_device_operations { int (*open) (struct inode *, struct file *); int (*release) (struct inode *, struct file *); int (*ioctl) (struct inode *, struct file *, unsigned, unsigned long); long (*compat_ioctl) (struct file *, unsigned, unsigned long); int (*media_changed) (struct gendisk *); int (*revalidate_disk) (struct gendisk *); #ifdef CONFIG_MAGIC_ROM_PTR int (*romptr) (struct block_device *, struct vm_area_struct *); #endif /* CONFIG_MAGIC_ROM_PTR */ struct module *owner; }; Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Wed Mar 16 19:40:15 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Thu, 17 Mar 2005 10:40:15 +1000 Subject: [uClinux-dev] 2.6.10-uc0 fec.c, inconsistent operand constraints - resend In-Reply-To: References: Message-ID: <4238D1EF.8040701@snapgear.com> Hi Jate, Jate Sujjavanich wrote: > I am working on activating the FEC for the Coldfire 5235 under linux > 2.6.10-uc0. I turned on the mii option, and now it breaks. First off - > do I need this? If so, what could the problem be? You don't need the generic mii option enable for the FEC driver. Regards Greg > include/asm/bitops.h: In function `mii_check_link': > include/asm/bitops.h:197: warning: read-write constraint does not allow > a register > include/asm/bitops.h:197: warning: read-write constraint does not allow > a register > include/asm/bitops.h:176: warning: read-write constraint does not allow > a register > include/asm/bitops.h:176: warning: read-write constraint does not allow > a register > include/asm/bitops.h:197: error: inconsistent operand constraints in an > `asm' > make[3]: *** [drivers/net/mii.o] Error 1 > > > Jate Sujjavanich > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org > -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Wed Mar 16 20:27:43 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Thu, 17 Mar 2005 11:27:43 +1000 Subject: [uClinux-dev] problem with outb In-Reply-To: References: Message-ID: <4238DD0F.9090002@snapgear.com> Hi Fabrice, Fabrice Poulin wrote: > I would like to use "outb" but I have always this error: > > undefined reference to 'outb' > collect2: ld returned 1 exit status > > However, I use the option -o2 during my compilation, So I don't > understand... > Somebody has an idea? WHat is your target architecture? Normally this is defined in your architecture io.h. Did you include ? Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From vanitha at agilis.stengg.com Thu Mar 17 04:44:40 2005 From: vanitha at agilis.stengg.com (Vanitha Ramaswami) Date: Thu, 17 Mar 2005 09:44:40 -0000 Subject: [uClinux-dev] Perl Compilation problem Message-ID: <01C52AD5.F1379700.vanitha@agilis.stengg.com> Hi All, Have anyone got compiled PERL successfully in snapgear distribution for IXP425. I am getting the error #bin/sh: ./miniperl: cannot execute binary file. Thanks Vanitha [This e-mail is confidential and may be priviledged. If you are not the intended recipient, please kindly notify us immediately and delete the message from your system; please do not copy or use it for any purpose, nor disclose its contents to any other person. Thank you.] ---ST Electronics Group--- From vanitha at agilis.stengg.com Thu Mar 17 04:51:02 2005 From: vanitha at agilis.stengg.com (Vanitha Ramaswami) Date: Thu, 17 Mar 2005 09:51:02 -0000 Subject: [uClinux-dev] RE: Perl Compilation problem Message-ID: <01C52AD6.D549D200.vanitha@agilis.stengg.com> Hi All, I applied the command that Saverio has mentioned in one of the earlier posts and i am getting this error now? perl -ni~ -we 'print unless /: Hi All, > > Have anyone got compiled PERL successfully in snapgear distribution for IXP425. I am getting the error > #bin/sh: ./miniperl: cannot execute binary file. > > > Thanks > Vanitha [This e-mail is confidential and may be priviledged. If you are not the intended recipient, please kindly notify us immediately and delete the message from your system; please do not copy or use it for any purpose, nor disclose its contents to any other person. Thank you.] ---ST Electronics Group--- From NLiu at smmi2.medison.com Wed Mar 16 21:48:54 2005 From: NLiu at smmi2.medison.com (Ning Liu) Date: Thu, 17 Mar 2005 10:48:54 +0800 Subject: [uClinux-dev] Problem about insmod a module into uClinux Message-ID: Hi all : I have a problem about insmod a module into uClinux of Samsung 44b0. In kernel configuration i have choosed [*]Enable loadable module support [ ]Set version...... [ ]Kernel module loader In BusyBox [*]insmod [*]insmod:load in kernel memory [*]lsmod [*]rmmod But when i load a module aaa.o,it report: /> chmod 777 /ramdisk/*.o /> insmod /ramdisk/aaa.o Using /ramdisk/aaa.o Unhandled fault: external abort on linefetch (F4) at 0x00000001 fault-common.c(97): start_code=0xc380040, start_stack=0xc3bff78) pid 29: failed 7 /> lsmod Module Size Used by Not tainted aaa 136 0 (uninitialized) /> chmod 777 /lib /lib: Read-only file system And my aaa is so simple #define MODULE #include int init_module(void) { // printk ( "<1>Hello , World \n"); return 0; } void cleanup_module(void) { // printk( "<1>Goodbye cruel world\n"); } Can any one help me? Thanks in advance! Best Regards. Ninig Liu -------------- next part -------------- An HTML attachment was scrubbed... URL: From gerg at snapgear.com Thu Mar 17 01:56:36 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Thu, 17 Mar 2005 16:56:36 +1000 Subject: [uClinux-dev] where should we store ttb for MMU? In-Reply-To: <200503150721.j2F7LoPI009571@cad.csie.ncku.edu.tw> References: <200503150721.j2F7LoPI009571@cad.csie.ncku.edu.tw> Message-ID: <42392A24.5070001@snapgear.com> Hi Wayne, Wayne Lee wrote: > I know that CONFIG_DRAM_BASE+0x100 is reserved for boot parameters. > CONFIG_DRAM_BASE+0x200 is vector table (__real_stubs_start). > > I do an experiment to fill 0xFFFF_FFFF from 0x0 to 0x7FFF using ICE. > After kernel is booted with I-Cache turn on only, 0x4000~0x7FFF is no longer > be 0xFFFF_FFFF. > The memory region 0x4000~0x7FFF is just the memory region I store > flat-mapped TTB for MMU. > Hence, the CPU always generates data abort while booting with MMU enabled. > > Are there any memory region is reserved for TTB? > I know uClinux aims to run Linux on no-MMU chips. > Hence, uClinux may _not_ reserve a protected memory region for TTB while a > CPU with MMU is adopted. I would expect that you should use the same region as MMU linux does (I take it that is 0x4000-0x7fff?). What is the nature of the data abort? What address (PC) does it occur? Regards Greg > Please kindly give suggestions, thanks. > > The booting messages with I-Cache, store TTB at 0x4000 and enable > MMU/D-Cache. > > Linux version 2.6.5-lwn1 (wayne at fedora-linux) (gcc version 2.95.3 20010315 > (release)(ColdFire patches - 20010318 from > http://fiddes.net/coldfire/)(uClinux XIP and shared lib patches from > http://www.snapgear.com/)) #38 Wed Mar 9 12:36:26 CST 2005 > CPU: ARM926EJ-Sid(wb) [41069263] revision 3 (ARMv5TEJ) > CPU: D VIPT write-back cache > CPU: I cache: 32768 bytes, associativity 4, 32 byte lines, 256 sets > CPU: D cache: 32768 bytes, associativity 4, 32 byte lines, 256 sets > Machine: QL201 > On node 0 totalpages: 16384 > DMA zone: 0 pages, LIFO batch:1 > Normal zone: 16384 pages, LIFO batch:4 > HighMem zone: 0 pages, LIFO batch:1 > Built 1 zonelists > Kernel command line: root=/dev/ram0 initrd=0x02000000,1000K keepinitrd > PID hash table entries: 512 (order 9: 4096 bytes) > Memory: 64MB = 64MB total > Memory: 62496KB available (1100K code, 119K data, 60K init) > Calibrating delay loop... 15.92 BogoMIPS > Dentry cache hash table entries: 8192 (order: 3, 32768 bytes) > Inode-cache hash table entries: 4096 (order: 2, 16384 bytes) > Mount-cache hash table entries: 512 (order: 0, 4096 bytes) > checking if image is initramfs...it isn't (ungzip failed); looks like an > initrd > POSIX conformance testing by UNIFIX > NET: Registered protocol family 16 > ikconfig 0.7 with /proc/config* > ttyS0 at I/O 0x80000100 (irq = 17) is a builtin QL201 UART > RAMDISK driver initialized: 16 RAM disks of 8192K size 1024 blocksize > loop: loaded (max 8 devices) > NET: Registered protocol family 2 > > > The booting messages with only I Cache. (Successfully booted) > ... > RAMDISK driver initialized: 16 RAM disks of 8192K size 1024 blocksize > loop: loaded (max 8 devices) > NET: Registered protocol family 2 > IP: routing cache hash table of 512 buckets, 4Kbytes > TCP: Hash tables configured (established 4096 bind 4096) > IPv4 over IPv4 tunneling driver > GRE over IPv4 tunneling driver > NET: Registered protocol family 17 > ... > > -----Original Message----- > From: Greg Ungerer [mailto:gerg at snapgear.com] > Sent: Friday, March 11, 2005 1:21 PM > To: Wayne Lee > Cc: uClinux development list > Subject: Re: [uClinux-dev] where should we store ttb for MMU? > > Hi Wayne, > > Wayne Lee wrote: > >>In fact, I am now porting an uClinux no VM port on arm926ej-s CPU. >>In order to turn-on D-cache, I use flat-mapping & section-mapped MMU and >>then turn on D-cache. >>The method is referred from Samsung s5c7375 port. >>I store TTB in 0x4000. >>However, while kernel is booting, the TTB is overwritten and data abort >>occurred. > > > Who overwrites it? > What is the kernel trying to do when overwritting this area? > > Regards > Greg > > > > > >>I am wonder where should we place the ttb in uClinux 2.6? >>Do we need to modify the vmlinux.lds? >>Please kindly give me your suggestions. >> >>Wayne >> > > > > -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Thu Mar 17 02:03:48 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Thu, 17 Mar 2005 17:03:48 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050316035306.B2F226373@mail2.atmark-techno.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> <20050316035306.B2F226373@mail2.atmark-techno.com> Message-ID: <42392BD4.3030201@snapgear.com> Hi Yashi, Yasushi SHOJI wrote: > At Tue, 15 Mar 2005 21:17:11 +1000, > Greg Ungerer wrote: > That'd help _me_ ;) > > not sure what we should name it though. I think the question is what > do we want the dist to be. > > if we know the answer, we can name the dist after it. > > some easy one would be embedded-linux-dist ;) That better captures what the current package is all about. Anyone else care to have their 2cents worth? Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From Andrey.Butok at iss.org.ua Thu Mar 17 02:11:34 2005 From: Andrey.Butok at iss.org.ua (Andrey Butok) Date: Thu, 17 Mar 2005 09:11:34 +0200 Subject: [uClinux-dev] next version of uClinux-dist? References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com><20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> Message-ID: <007001c52ac0$914801f0$f103a8c0@kesllaptop3> Hello Can anybody please tell me when the next version of uClinux distribution will be available (after 20041215) ? What is the usual frequency of uClinux-dist-xx release? Best regards Andrey From gerg at snapgear.com Thu Mar 17 02:19:09 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Thu, 17 Mar 2005 17:19:09 +1000 Subject: [uClinux-dev] 2.6 kernel not booting on uCdimm In-Reply-To: <20050314153131.30243.qmail@web21604.mail.yahoo.com> References: <20050314153131.30243.qmail@web21604.mail.yahoo.com> Message-ID: <42392F6D.5000107@snapgear.com> Hi Sriram, Sriram Chadalavada wrote: >>I have my local CVS log, but there has really been >>very little >>work done on this driver in all of the 2.6 kernels. >>Georges Menie >>did some cleanups in 2.5 series for this. > > Please email me your log and if possible, Menie's > changes for 2.5. Here the log I have for the only significant change: revision 1.13 date: 2003-07-04 01:31:05 +0000; author: gerg; state: Exp; lines: +7 -7 Here is a patch which clean up the 68328 serial driver in the 2.5 kernel, removing the compilation warnings. Patch submitted by Georges Menie . >>There is a single linker script for all m68knommu >>targets: >>~/arch/m68knommu/kernel/vmlinux.lds.S. All of the >>functionality of each of the older files should be >>catered >>for in this script. > > > Thanks for clarifying that. One question though. How > can I use the assembly 'putc' in this script? Where > are the assembly instructions (loading the stack ptr, > copying data segment from ROM to RAM .....) > corresponding to _start/_stext? vmlinux.lds.S is not an assembler file, it is a linker script. It is passed through c-preprocessor to insert kernel level definitions (to get relevent addresses, etc). But it is the linker script used by ld. If you want to put in debug trace you need to start with the start up assembler, you probably want something like: ~/arch/m68knommu/platform/68328/head-ram.S Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From lars.friedrich at wago.com Thu Mar 17 02:30:52 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Thu, 17 Mar 2005 08:30:52 +0100 Subject: [uClinux-dev] renaming uClinux-dist Message-ID: > > if we know the answer, we can name the dist after it. > > some easy one would be embedded-linux-dist ;) > That better captures what the current package is all about. > Anyone else care to have their 2cents worth? Is there really much difference between micro-controller-linux-dist and embedded-linux-dist? One describes the kind of hardware most likely used and the other the area of usage and both is inaccurate, just in different ways. Embedded-linux will run fine on non-embedded systems as uClinux will run fine on MMU systems. What will be the benefit of renaming? + _maybe_ less confusing What are the disadvantages? - All old references break Websites, books, mailing list archives, domain names like uClinux.org .. Then how do you want to name it? eLinux? This has been used in the past already. Just my 2 cents. Best regards, Lars Friedrich From ysato at users.sourceforge.jp Thu Mar 17 03:10:25 2005 From: ysato at users.sourceforge.jp (Yoshinori Sato) Date: Thu, 17 Mar 2005 17:10:25 +0900 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> Message-ID: At Tue, 15 Mar 2005 14:03:16 +0900, Yasushi SHOJI wrote: > > Hi all, > > We, Atmark Techno, has been distributing uClinux-dist as a development > source code base for our product, including ARM CPU with MMU, in > addition to MicroBlaze, a soft core CPU without MMU. > > Many of our customers give us feedback saying that all of our product > is uClinux based, since the tarball is name after uClinux-dist. > > Would you, as a maintainer and as a member of community, be OK if we > rename the tarball to something like "Atmark Techno Dist", so that our > customer don't get wrong impression that all of our product is based > on uClinux? > > We will, of course, put words in documents saying that this source > code is based on uClinux-dist and can be downloaded from such and > such. > > # Don't get me wrong. uClinux _is_ great. it's just that some people > # think the way we don't think ;p > > What do you all think? I do not stick to many names. "uClinux" points at a thing of uclinux.org and does not have description not to be kernel? -- Yoshinori Sato From H.Haeberle at pilz.de Thu Mar 17 04:02:02 2005 From: H.Haeberle at pilz.de (Heinz Haeberle) Date: Thu, 17 Mar 2005 10:02:02 +0100 Subject: [uClinux-dev] RE:linux-2.6.10-uc0, 11-uc0 romptr error In-Reply-To: <20050316170009.7C6421A1D4@uclinux.org> Message-ID: go to the kernel konfiguration (make xconfig) and search in: filesystem, romfilesystem ROM filesystem support execute user programs directly from rom <- disable this From lara_john79 at yahoo.com Thu Mar 17 04:04:57 2005 From: lara_john79 at yahoo.com (Lara John) Date: Thu, 17 Mar 2005 01:04:57 -0800 (PST) Subject: [uClinux-dev] Mounting root file system on USB disk Message-ID: <20050317090457.37209.qmail@web53201.mail.yahoo.com> Hi, Thanks a ton for the direction. I got the IRQ problem resolved. and I can use the Compact Flash attatchd to USB with USB reader. At present i am looking for a patch/suggestion which would help me mounting my root file system on compact flash. The kernel and bootloader are in FLASH and can boot fine. But in development phase i want to use root file system Compact Flash. I appreciate the help i got from the list. Thanks again. Best Regards, Lara --- Greg Ungerer wrote: > Hi Lara John, > > Lara John wrote: > > I am trying to port embedded linux kernel (ver. > > 2.4.26) on MS7751RCP01. > > > > This board has got NEC's usb 2.0 UHCI host > controller. > > > > I am facing followoing problems. > > > > 1) This controller is detected by the kernel on > boot > > time, but it is probing the wrong IRQ. (The IRQ to > IRL > > is managed by the FPGA), the kernel should > allocate > > IRQ 9 but the messages says it is IRQ 10 > > The allocation of IRQ's is machine specific, you > need to > find the code for your machine in your kernel > version. > > This being wrong is a possible (even likely) cause > of your system hanging when you have a USB device > on the bus. > > Regards > Greg > > > > > > The /proc/bus/usb/devices says: > > T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 > > Spd=12 MxCh= 1 > > B: Alloc= 0/900 us ( 0%), #Int= 0, #Iso= 0 > > D: Ver= 1.10 Cls=09(hub ) Sub=00 Prot=00 MxPS= 8 > > #Cfgs= 1 > > P: Vendor=0000 ProdID=0000 Rev= 0.00 > > S: Product=USB OHCI Root Hub > > S: SerialNumber=fd001000 > > C:* #Ifs= 1 Cfg#= 1 Atr=40 MxPwr= 0mA > > I: If#= 0 Alt= 0 #EPs= 1 Cls=09(hub ) Sub=00 > Prot=00 > > Driver=hub > > E: Ad=81(I) Atr=03(Int.) MxPS= 2 Ivl=255ms > > T: Bus=01 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 > > Spd=12 MxCh= 1 > > B: Alloc= 0/900 us ( 0%), #Int= 0, #Iso= 0 > > D: Ver= 1.10 Cls=09(hub ) Sub=00 Prot=00 MxPS= 8 > > #Cfgs= 1 > > P: Vendor=0000 ProdID=0000 Rev= 0.00 > > S: Product=USB OHCI Root Hub > > S: SerialNumber=fd000000 > > C:* #Ifs= 1 Cfg#= 1 Atr=40 MxPwr= 0mA > > I: If#= 0 Alt= 0 #EPs= 1 Cls=09(hub ) Sub=00 > Prot=00 > > Driver=hub > > E: Ad=81(I) Atr=03(Int.) MxPS= 2 Ivl=255ms > > > > 2) I am trying to mount SD Card (MMC) on USB with > usb > > reader. I have selected the build in option for > SCSI > > Storage and USB storage related. If i boot the > kernel > > without this device, it goes fine, but as soon as > i > > attatch the USB reader with SD card, it hangs (no > > panic.. seems that it is in some ageless loop :( ) > > > > The last message it gave is as follows: > > hub.c: new USB device 00:03.0-1, assigned address > 2 > > > > > > Please help me resolving this issue. If the > solution > > related to it is already resolved in newer > releases > > please let me know. > > > > Thanks in advance. > > > > Best regards, > > Lara John > > > > PS: If any information is missing please let me > know. > > > > > > > > __________________________________ > > Do you Yahoo!? > > Yahoo! Small Business - Try our new resources > site! > > http://smallbusiness.yahoo.com/resources/ > > _______________________________________________ > > uClinux-dev mailing list > > uClinux-dev at uClinux.org > > > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > > This message was resent by uclinux-dev at uclinux.org > > > > -- > ------------------------------------------------------------------------ > Greg Ungerer -- Chief Software Dude EMAIL: > gerg at snapgear.com > SnapGear -- a CyberGuard Company PHONE: > +61 7 3435 2888 > 825 Stanley St, FAX: > +61 7 3891 3630 > Woolloongabba, QLD, 4102, Australia WEB: > http://www.SnapGear.com > __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com From dhowells at redhat.com Thu Mar 17 04:08:50 2005 From: dhowells at redhat.com (David Howells) Date: Thu, 17 Mar 2005 09:08:50 +0000 Subject: [uClinux-dev] Is smc91x.c the best choice for smsc91c111 chip in uClinux 2.6? In-Reply-To: References: Message-ID: <19833.1111050530@redhat.com> Heinz Haeberle wrote: > Since I was starting with 2.6.6 without a 91c111 driver at all I have ported > it over from the 2.4 kernel. Guess I have posted it already. But if there > is a maintream driver for the smsc 91x chips I will probably go with this in > the future. Will see how it looks like. There is in 2.6.10/11: drivers/net/smc91x.[ch] To use it, you need to put something like this in your arch: static struct resource my_smc91c111_resources[] = { [0] = { .start = SMC_BASE_ADDRESS, .end = SMC_BASE_END, .flags = IORESOURCE_MEM, }, [1] = { .start = SMC_IRQ, .end = SMC_IRQ, .flags = IORESOURCE_IRQ, }, }; static struct platform_device my_smc91c111_device = { .name = "smc91x", .id = 0, .num_resources = ARRAY_SIZE(my_smc91c111_resources), .resource = my_smc91c111_resources, }; static int __init unit_device_init(void) { platform_device_register(&my_smc91c111_device); return 0; } This registers the device and its configuration which then auto-configures the driver. David From jwilliams at itee.uq.edu.au Thu Mar 17 04:46:14 2005 From: jwilliams at itee.uq.edu.au (John Williams) Date: Thu, 17 Mar 2005 19:46:14 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <42392BD4.3030201@snapgear.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> <20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> Message-ID: <423951E6.5040607@itee.uq.edu.au> Hi Greg, Greg Ungerer wrote: > Yasushi SHOJI wrote: > >> At Tue, 15 Mar 2005 21:17:11 +1000, >> Greg Ungerer wrote: >> That'd help _me_ ;) >> >> not sure what we should name it though. I think the question is what >> do we want the dist to be. >> >> if we know the answer, we can name the dist after it. >> >> some easy one would be embedded-linux-dist ;) > > That better captures what the current package is all about. > Anyone else care to have their 2cents worth? This is a good discussion to be having - really I think the bigger question is "where is uclinux-dist going?" Some random observations: It's pretty agile. dist-ifying an existing app is getting easier and easier - witness some apps now that are based on generic GNU configure setups. It's system-oriented (e.g creates complete embedded systems - kernel, apps, libs, modules, ...) It's Free. I'm not very familiar with other embedded distributions, but I do believe e.g. that several thousand dollars for montavista linux is excessive. Some things I'd like to see in uClinux-dist: Real-time CVS tracking. The model of periodic releases places pressure on Dave and Greg (maybe! :) and is a bit inflexible. I see this a lot on the Microblaze platform where every single system can be different. The ability to create new vendors targets and immediately publish them through CVS would be very valuable. It's getting so that I'm also considering a custom uClinux-dist for microblaze. Higher-level abstractions and tools in vendor/board management. There's a lot of commonality between vendors and boards/platforms in the dist. Currently the logic is all coded into the hundreds of vendors Makefiles. This improved a little bit recently with the common config area, but a more data-driven approach to this would be really valuable. As for names, maybe minidist, microdist, little-linux, ... I'll think some more, see if any great ideas are forthcoming. A note, I recently came across the Linux Mark Institute, who handle matters relating to the use of the word "linux" in other trademarks. http://www.linuxmark.org/ . Anyone registering a trademark that contains the word "linux" is required/asked/morally obliged to obtain a license to do so. Anyway that's enough for now. cheers, John From roberto.farina at cefriel.it Thu Mar 17 06:46:14 2005 From: roberto.farina at cefriel.it (Roberto Farina) Date: Thu, 17 Mar 2005 11:46:14 +0000 Subject: [uClinux-dev] Problem about insmod a module into uClinux In-Reply-To: References: Message-ID: <42396E06.80802@cefriel.it> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 [ ]Kernel module loader I think this is your problem; try enabling this choice. best regards Roberto Ning Liu wrote: > > Hi all : > > I have a problem about insmod a module into uClinux of Samsung > 44b0. > > In kernel configuration i have choosed > [*]Enable loadable module support > [ ]Set version...... > [ ]Kernel module loader > In BusyBox > [*]insmod > [*]insmod:load in kernel memory > [*]lsmod > [*]rmmod > But when i load a module aaa.o,it report: > /> chmod 777 /ramdisk/*.o > /> insmod /ramdisk/aaa.o > Using /ramdisk/aaa.o > Unhandled fault: external abort on linefetch (F4) at > 0x00000001 > fault-common.c(97): start_code=0xc380040, > start_stack=0xc3bff78) > pid 29: failed 7 > /> lsmod > Module Size Used by Not tainted > aaa 136 0 (uninitialized) > /> chmod 777 /lib > /lib: Read-only file system > And my aaa is so simple > #define MODULE > > #include > int init_module(void) > { > // printk ( "<1>Hello , World \n"); > return 0; > } > void cleanup_module(void) > { > // printk( "<1>Goodbye cruel world\n"); > } > Can any one help me? Thanks in advance! > > Best Regards. > Ninig Liu > > > ------------------------------------------------------------------------ > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org - -- Roberto Farina CEFRIEL ? Politecnico di Milano Via Fucini, 2 ? 20133 Milano (Italy) p. +39 02 23954 345 - f. +39 02 23954 254 e. roberto.farina at cefriel.it - w. www.cefriel.it -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.7 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFCOW4GzF5ONZxALKMRAqLvAKC1yyYZUwjVUgrhh7ri9eDXyU4u2gCg5dzT 8yJIXF6l8d9N9ZpAxy5Ukn8= =mpBU -----END PGP SIGNATURE----- From wnlee at cad.csie.ncku.edu.tw Thu Mar 17 05:51:53 2005 From: wnlee at cad.csie.ncku.edu.tw (Wayne Lee) Date: Thu, 17 Mar 2005 18:51:53 +0800 Subject: [uClinux-dev] Is smc91x.c the best choice for smsc91c111 chip inuClinux 2.6? In-Reply-To: <19833.1111050530@redhat.com> Message-ID: <200503171054.j2HAsnKE016896@cad.csie.ncku.edu.tw> Hi, Howells, I now access smsc91c111 by a 16-bit read/write register (port IO), this port IO register is located at 0xc000600 (32bit register), for example. Should the SMC_BASE_ADDRESS be 0xC000_0600 and SMC_BASE_END be 0xC000_0603? static struct resource my_smc91c111_resources[] = { [0] = { .start = SMC_BASE_ADDRESS, .end = SMC_BASE_END, .flags = IORESOURCE_MEM, }, [1] = { .start = SMC_IRQ, .end = SMC_IRQ, .flags = IORESOURCE_IRQ, }, }; Thanks. Wayne -----Original Message----- From: uclinux-dev-bounces at uClinux.org [mailto:uclinux-dev-bounces at uClinux.org] On Behalf Of David Howells Sent: Thursday, March 17, 2005 5:09 PM To: Heinz Haeberle Cc: uclinux-dev at uClinux.org Subject: Re: [uClinux-dev] Is smc91x.c the best choice for smsc91c111 chip inuClinux 2.6? Heinz Haeberle wrote: > Since I was starting with 2.6.6 without a 91c111 driver at all I have ported > it over from the 2.4 kernel. Guess I have posted it already. But if there > is a maintream driver for the smsc 91x chips I will probably go with this in > the future. Will see how it looks like. There is in 2.6.10/11: drivers/net/smc91x.[ch] To use it, you need to put something like this in your arch: static struct resource my_smc91c111_resources[] = { [0] = { .start = SMC_BASE_ADDRESS, .end = SMC_BASE_END, .flags = IORESOURCE_MEM, }, [1] = { .start = SMC_IRQ, .end = SMC_IRQ, .flags = IORESOURCE_IRQ, }, }; static struct platform_device my_smc91c111_device = { .name = "smc91x", .id = 0, .num_resources = ARRAY_SIZE(my_smc91c111_resources), .resource = my_smc91c111_resources, }; static int __init unit_device_init(void) { platform_device_register(&my_smc91c111_device); return 0; } This registers the device and its configuration which then auto-configures the driver. David _______________________________________________ uClinux-dev mailing list uClinux-dev at uClinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev at uclinux.org From aravindforn at yahoo.co.in Thu Mar 17 06:03:42 2005 From: aravindforn at yahoo.co.in (Aravind babu) Date: Thu, 17 Mar 2005 11:03:42 +0000 (GMT) Subject: [uClinux-dev] Memory leak with mgetty-1.1.28 Message-ID: <20050317110342.12015.qmail@web8402.mail.in.yahoo.com> Hi all, I am using uclinux 2.4.20 on MIPS.By default in our box,mgetty will start.we are using mgetty-1.1.28 If i put box ideally for 5 days without doing anything memory is going down by 20MB.If i disabled mgetty , all the memory lost is coming back.How to resolve this memory leak? Also i am not dialing in or dialing out.Just the mgetty process is running. Give me any ideas to resolve this. Thanks in advance, Aravind. __________________________________ Do you Yahoo!? Make Yahoo! your home page http://www.yahoo.com/r/hs From dhowells at redhat.com Thu Mar 17 06:08:04 2005 From: dhowells at redhat.com (David Howells) Date: Thu, 17 Mar 2005 11:08:04 +0000 Subject: [uClinux-dev] Is smc91x.c the best choice for smsc91c111 chip inuClinux 2.6? In-Reply-To: <200503171054.j2HAsnKE016896@cad.csie.ncku.edu.tw> References: <200503171054.j2HAsnKE016896@cad.csie.ncku.edu.tw> Message-ID: <6917.1111057684@redhat.com> Wayne Lee wrote: > I now access smsc91c111 by a 16-bit read/write register (port IO), this port > IO register is located at 0xc000600 (32bit register), for example. > > Should the SMC_BASE_ADDRESS be 0xC000_0600 and SMC_BASE_END be 0xC000_0603? Well, the compiler will probably complain at the '_' characters in the middle of the numbers. Apart from that, it may be reasonable. I've given it a 256-byte window because that's what the CPU bus controller on this board I've been playing with assigns to that device. David From devi at tenet.res.in Thu Mar 17 17:59:41 2005 From: devi at tenet.res.in (devi) Date: Fri, 18 Mar 2005 04:29:41 +0530 Subject: [uClinux-dev] console Message-ID: <20050317225941.17572.qmail@mail.tenet.res.in> hi, how to register vc/0 in console.c.In the con_init const char *display_desc = NULL; if (conswitchp) display_desc = conswitchp->con_startup(); if(!displaydesc) fg_console=0;return; since there is no conswitchp and display_desc.so it returns. It is not registering the console_driver.Now I have the serial console and no virtual console. So how to register the virtual console.Because in the take_over_console() it searches for the virtual console data. if (!vc_cons[i].d || !vc_cons[i].d->vc_sw) continue; Since there is no virtual console it exits from the for loop. Regards Devi From rdkehn at yahoo.com Thu Mar 17 07:37:08 2005 From: rdkehn at yahoo.com (Doug Kehn) Date: Thu, 17 Mar 2005 04:37:08 -0800 (PST) Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: 6667 Message-ID: <20050317123708.24195.qmail@web51903.mail.yahoo.com> --- "Friedrich, Lars" wrote: > > What are the disadvantages? > - All old references break > Websites, books, mailing list archives, domain names > like uClinux.org .. > This is an excellent point. Maybe instead of a name change, consideration could be given to updating the www.uclinux.org website. Thinking about it is the confusion over what uClinux really is coming the "What is uClinux?" (www.uclinux.org/description) page? --- CUT --- What is uClinux? uClinux is a derivative of Linux 2.0 kernel intended for microcontrollers without Memory Management Units (MMUs). --- CUT --- I could see how newcomers to uClinux might get confused. Maybe I added 3-cents here. ...doug __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/ From davidm at snapgear.com Thu Mar 17 08:29:02 2005 From: davidm at snapgear.com (David McCullough) Date: Thu, 17 Mar 2005 23:29:02 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050317123708.24195.qmail@web51903.mail.yahoo.com> References: <20050317123708.24195.qmail@web51903.mail.yahoo.com> Message-ID: <20050317132902.GA31588@beast> Jivin Doug Kehn lays it down ... > --- "Friedrich, Lars" wrote: > > > > What are the disadvantages? > > - All old references break > > Websites, books, mailing list archives, domain names > > like uClinux.org .. > > > This is an excellent point. > > Maybe instead of a name change, consideration could be > given to updating the www.uclinux.org website. > Thinking about it is the confusion over what uClinux > really is coming the "What is uClinux?" > (www.uclinux.org/description) page? > > --- CUT --- > What is uClinux? > uClinux is a derivative of Linux 2.0 kernel intended > for microcontrollers without Memory Management Units > (MMUs). > --- CUT --- > > I could see how newcomers to uClinux might get > confused. I agree with Lars and Doug on this one. I think uClinux-dist is (by definition of uC) the correct name, unfortunately most common usage in the linux kernel space means uC == !MMU. I think uClinux-dist has a reputation/brand name recognition that should not be overlooked. And of course it is easier to keep the same name. So if uClinux-dist is kept, the best thing to do is educate the population, fixing up the aging descriptions on uClinux.org would be a good start. I guess that a 5c worth from me (we dropped 1's and 2's ;-) Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From davidm at snapgear.com Thu Mar 17 08:52:17 2005 From: davidm at snapgear.com (David McCullough) Date: Thu, 17 Mar 2005 23:52:17 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <423951E6.5040607@itee.uq.edu.au> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> <20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> <423951E6.5040607@itee.uq.edu.au> Message-ID: <20050317135217.GB31588@beast> Jivin John Williams lays it down ... > Hi Greg, > > Greg Ungerer wrote: > > >Yasushi SHOJI wrote: > > > >>At Tue, 15 Mar 2005 21:17:11 +1000, > >>Greg Ungerer wrote: > >>That'd help _me_ ;) > >> > >>not sure what we should name it though. I think the question is what > >>do we want the dist to be. > >> > >>if we know the answer, we can name the dist after it. > >> > >>some easy one would be embedded-linux-dist ;) > > > >That better captures what the current package is all about. > >Anyone else care to have their 2cents worth? > > This is a good discussion to be having - really I think the bigger > question is "where is uclinux-dist going?" Good question ;-) It seems to be going just about everywhere at times. > Some random observations: > > It's pretty agile. dist-ifying an existing app is getting easier and > easier - witness some apps now that are based on generic GNU configure > setups. > > It's system-oriented (e.g creates complete embedded systems - kernel, > apps, libs, modules, ...) > > It's Free. I'm not very familiar with other embedded distributions, but > I do believe e.g. that several thousand dollars for montavista linux is > excessive. I have to agree there. One thing I like is that at the end of the day it is just scripts and Makefiles, so you can make it do anything, and fairly quickly. The more fancy the IDE, the more chance you just cannot do something simple :-( > Some things I'd like to see in uClinux-dist: > > Real-time CVS tracking. The model of periodic releases places pressure > on Dave and Greg (maybe! :) and is a bit inflexible. I see this a lot Actually it's the opposite. We have chosen the periodic release (and no mods to the dist portion) so that we can reduce the amount of tree syncing we have to do. For example, the 2.4 kernel causes me much pain. It goes something like: * tag cvs.uclinux.org kernel * import cvs.uclinux.org kernel to our tree * merge changes and fix conflicts * build and test our kernel * generate diffs against cvs.uclinux.org kernel and apply them to cvs.uclinux.org * import new linux kernel version to cvs.uclinux.org * merge changes and fix conflicts * compile, test and fix cvs.uclinux.org version * import cvs.uclinux.org kernel to our cvs * merge changes and fix conflicts * compile, test and fix our kernel * generate diff against cvs.uclinux.org kernel and apply to cvs.uclinux.org * generate patches etc * send out release mail * do a short happy dance :-) :-) :) now I could cut out some steps above but then the merging portion gets much more prone to errors and nasty conflicts, and that area is bad enough already :-) Since it's unlikely we would ever run directly off the public cvs internally, we want to avoid all this work on the rest of the dist files. So by asking no one to mod the public dist files, we can merge in a new dist with no conflicts or change syncing required, much easier :-) > on the Microblaze platform where every single system can be different. > The ability to create new vendors targets and immediately publish them > through CVS would be very valuable. It's getting so that I'm also > considering a custom uClinux-dist for microblaze. Ok, perhaps we could create a patch system to operate between releases where a patch area on cvs.uclinux.org provides patches that are applied to a checked out dist in sequence to get to the latest. This would not cause us any pain at all and would allow you or others to make changes available immediately. If a patch appears after a cvs update, the the make system would see it, apply it and mark it as applied. The theory is all patches would be applied at the next release, the patch directory would become empty and thus a cvs update would just say "... already contains the changes ....". Patch names should start with a unique increasing sequence number :-) It may even be possible to introduce a "patch pending" or experimental area in this. Does it sound like this has any merit ? More importantly does it sound workable ? > Higher-level abstractions and tools in vendor/board management. There's > a lot of commonality between vendors and boards/platforms in the dist. > Currently the logic is all coded into the hundreds of vendors Makefiles. > This improved a little bit recently with the common config area, but > a more data-driven approach to this would be really valuable. This has improved even more from the release with all the dev nodes in a common file and most of the config.arch setting moved to a common file as well. Still all/more ideas always welcomed. > As for names, maybe minidist, microdist, little-linux, ... > > I'll think some more, see if any great ideas are forthcoming. A note, I > recently came across the Linux Mark Institute, who handle matters > relating to the use of the word "linux" in other trademarks. > http://www.linuxmark.org/ . Anyone registering a trademark that > contains the word "linux" is required/asked/morally obliged to obtain a > license to do so. Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From yashi at atmark-techno.com Thu Mar 17 09:13:22 2005 From: yashi at atmark-techno.com (Yasushi SHOJI) Date: Thu, 17 Mar 2005 23:13:22 +0900 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050317132902.GA31588@beast> References: <20050317123708.24195.qmail@web51903.mail.yahoo.com> <20050317132902.GA31588@beast> Message-ID: <20050317141322.9A3A4E8A6@mail2.atmark-techno.com> At Thu, 17 Mar 2005 23:29:02 +1000, David McCullough wrote: > > > Jivin Doug Kehn lays it down ... > > --- "Friedrich, Lars" wrote: > > > > > > What are the disadvantages? > > > - All old references break > > > Websites, books, mailing list archives, domain names > > > like uClinux.org .. > > > > > This is an excellent point. > > > > Maybe instead of a name change, consideration could be > > given to updating the www.uclinux.org website. > > Thinking about it is the confusion over what uClinux > > really is coming the "What is uClinux?" > > (www.uclinux.org/description) page? > > > > --- CUT --- > > What is uClinux? > > uClinux is a derivative of Linux 2.0 kernel intended > > for microcontrollers without Memory Management Units > > (MMUs). > > --- CUT --- > > > > I could see how newcomers to uClinux might get > > confused. > > I agree with Lars and Doug on this one. > > I think uClinux-dist is (by definition of uC) the correct name, > unfortunately most common usage in the linux kernel space means uC == !MMU. well that's the whole point. _Unfortunately_, we have created the brand-image of uC as *no MMU*. > I think uClinux-dist has a reputation/brand name recognition that should > not be overlooked. And of course it is easier to keep the same name. yup it's easy to keep the same name but ... > So if uClinux-dist is kept, the best thing to do is educate the > population, fixing up the aging descriptions on uClinux.org would be a > good start. it's tough to educate the population, even after we got well-known brand. my two yen ;) -- yashi From dhowells at redhat.com Thu Mar 17 09:28:25 2005 From: dhowells at redhat.com (David Howells) Date: Thu, 17 Mar 2005 14:28:25 +0000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050317135217.GB31588@beast> References: <20050317135217.GB31588@beast> <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> <20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> <423951E6.5040607@itee.uq.edu.au> Message-ID: <17013.1111069705@redhat.com> David McCullough wrote: > For example, the 2.4 kernel causes me much pain. It goes something > like: Can you get some of the stuff back into the vanilla 2.4 kernel? And likewise for 2.6. David From yashi at atmark-techno.com Thu Mar 17 09:33:21 2005 From: yashi at atmark-techno.com (Yasushi SHOJI) Date: Thu, 17 Mar 2005 23:33:21 +0900 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050317135217.GB31588@beast> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> <20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> <423951E6.5040607@itee.uq.edu.au> <20050317135217.GB31588@beast> Message-ID: <20050317143321.6A8CAED30@mail2.atmark-techno.com> At Thu, 17 Mar 2005 23:52:17 +1000, David McCullough wrote: > Jivin John Williams lays it down ... > > on the Microblaze platform where every single system can be different. > > The ability to create new vendors targets and immediately publish them > > through CVS would be very valuable. It's getting so that I'm also > > considering a custom uClinux-dist for microblaze. > > Ok, perhaps we could create a patch system to operate between releases > where a patch area on cvs.uclinux.org provides patches that are applied > to a checked out dist in sequence to get to the latest. that'd be very nice. Quilt should do the job. https://savannah.nongnu.org/projects/quilt -- yashi From alandd at consultpros.com Thu Mar 17 11:51:16 2005 From: alandd at consultpros.com (Alan Dayley) Date: Thu, 17 Mar 2005 09:51:16 -0700 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <42392BD4.3030201@snapgear.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> Message-ID: <200503170951.16616.alandd@consultpros.com> On Thursday 17 March 2005 12:03 am, Greg Ungerer wrote: > > > > some easy one would be embedded-linux-dist ;) > > That better captures what the current package is all about. > Anyone else care to have their 2cents worth? From a humble lurker... Renaming to anything would be a mistake for two reasons: 1. The history, books, links everywhere, etc. that create the community and user support all say uCLinux. Renaming will break that and will take time and effort to rebuild it. Building a kernel is very easy compared with building a community. 2. It won't matter. For years after the name change, everyone will still call it uCLinux. Everyone will still refer to it as "the ditro formerly known as uCLinux." 3 years from now, it will still be uCLinux to most that seek it out. I'd rather see documentation be updated and education be pushed than the name changed. I'll fade back to the shadows now... Alan From jsujjavanich at syntech-fuelmaster.com Thu Mar 17 11:59:56 2005 From: jsujjavanich at syntech-fuelmaster.com (Jate Sujjavanich) Date: Thu, 17 Mar 2005 11:59:56 -0500 Subject: [uClinux-dev] fec, linux 2.6.11-uc0, MCF5235 Message-ID: I have been porting linux 2.6.11-uc0 to the MCF5235. It boots successfully, but I am having a problem detecting the onboard fast ethernet controller. There is a SIOCGIFHWADDR error when dhcp tries to use the address, and there is no eth0 device. There are no kernel messages about network device detection (fec: *). I've added defines for the 5235 to fec.c and fec.h. The KS8721BL is specified in the driver. Under network devices, I have selected NETDEVICES & NET_ETHERNET under network device support. When/how does the kernel perform fec detection? What could be the source of my trouble? - Jate Sujjavanich -------------- next part -------------- An HTML attachment was scrubbed... URL: From anders at ontimenet.com Thu Mar 17 14:01:19 2005 From: anders at ontimenet.com (=?iso-8859-1?Q?Anders_=D6hlander?=) Date: Thu, 17 Mar 2005 20:01:19 +0100 Subject: [uClinux-dev] WEB server for uClinux. Message-ID: <20050317190111.887AC1A11D@uclinux.org> Hi all I am looking on to implement a WEB server on my uClinux device. Is there anyone that has some experience on that ? Which one is best ? As I understand it is the httpd and the boa server that are available, is that true ? Are there more to chose of ? Has anyone used some host tools for generate WEB pages that is functioning with uClinux server ? Thanks for your input. BR /Anders ?hlander -------------- next part -------------- An HTML attachment was scrubbed... URL: From ben at thingmagic.com Thu Mar 17 14:25:56 2005 From: ben at thingmagic.com (Ben Giddings) Date: Thu, 17 Mar 2005 14:25:56 -0500 Subject: [uClinux-dev] dl-runtime error with appWeb on an XScale (glibc, linux-2.4.x, snapgear 3.2) Message-ID: <4239D9C4.1080304@thingmagic.com> I've managed to get the snapgear 3.2 distro to do what I need, it now builds a full fledged distribution in the format I need. The problem now is appWeb. I used to use boa, but some of the features of appWeb looked promising, so I tried it instead. It builds without complaints, but when I try to run it: # /bin/appWeb --help Inconsistency detected by ld.so: dl-runtime.c: 72: fixup: Assertion `((reloc->r_info) & 0xff) == 22' failed! I suspect the problem might be related to appWeb using C++ where everything else on the system (as far as I know) uses C. That's just a guess though, I've never seen this type of error before. AppWeb installs a "libminiStdc++" in /etc/appWeb/bin/libminiStdc++.so, which seems to be a strange location to me. AppWeb also puts some other strange files on the system, that would make sense on a desktop system, but don't really fit with an embedded device: /etc/rc.d /etc/rc.d/init.d /etc/rc.d/init.d/appWeb /etc/rc.d/rc5.d /etc/rc.d/rc5.d/S81appWeb /etc/rc.d/rc0.d /etc/rc.d/rc0.d/K15appWeb /etc/rc.d/rc1.d /etc/rc.d/rc1.d/K15appWeb /etc/rc.d/rc2.d /etc/rc.d/rc2.d/K15appWeb /etc/rc.d/rc3.d /etc/rc.d/rc3.d/K15appWeb /etc/rc.d/rc4.d /etc/rc.d/rc4.d/K15appWeb /etc/rc.d/rc6.d /etc/rc.d/rc6.d/K15appWeb Since I'm really not using runlevels, this seems a bit excessive. I can easily trim this down, but I'm curious about whether other people are using appWeb on small, embedded devices, and what experiences they have had. Ben From eauth at softsys.co.at Thu Mar 17 15:28:13 2005 From: eauth at softsys.co.at (Erwin Authried) Date: Thu, 17 Mar 2005 21:28:13 +0100 Subject: [uClinux-dev] WEB server for uClinux. In-Reply-To: <20050317190111.887AC1A11D@uclinux.org> References: <20050317190111.887AC1A11D@uclinux.org> Message-ID: <1111091293.1543.15.camel@justakiss> Am Don, den 17.03.2005 schrieb Anders ?hlander um 20:01: > Hi all > > I am looking on to implement a WEB server on my uClinux device. > > Is there anyone that has some experience on that ? > > Which one is best ? > > > > As I understand it is the httpd and the boa server that are available, > is that true ? > > Are there more to chose of ? > > take a look at the uClinux directory -> Software/Embedded Webservers http://uclinux.home.at/ Regards, Erwin From mike at steroidmicros.com Thu Mar 17 15:30:22 2005 From: mike at steroidmicros.com (mike) Date: Thu, 17 Mar 2005 12:30:22 -0800 Subject: [uClinux-dev] RTAI and uClinux Message-ID: <001b01c52b30$26ac5560$6202a8c0@flanders> Hello, Does anyone have a recent (uClinux-2.4.28) rthal patch for the uClinux kernel. I have tried to use the "patch-rthal-2.4.17-uc0-m68knommu" from the RTAI-24.1.13 release but it does not apply cleanly. I can try to apply it by hand but figured there must be someone who has done this already. Thanks, Mike Lavender From anders at ontimenet.com Thu Mar 17 15:40:28 2005 From: anders at ontimenet.com (=?iso-8859-1?Q?Anders_=D6hlander?=) Date: Thu, 17 Mar 2005 21:40:28 +0100 Subject: SV: [uClinux-dev] WEB server for uClinux. In-Reply-To: <4239E16B.9000208@thingmagic.com> Message-ID: <20050317204014.358F91A127@uclinux.org> Hi Ben Thanks for your answer. How do you get the information from the device to the browser ? Do you write c-code that sends out HTML or is that any nicer way of doing that ? Is there any way to use a tool to generate the WEB-site structure ? Like pagemaker or something ? Can you use some security on the BOA server, like ssl or https ? BR /Anders -----Ursprungligt meddelande----- Fr?n: Ben Giddings [mailto:ben at thingmagic.com] Skickat: den 17 mars 2005 20:59 Till: Anders ?hlander ?mne: Re: [uClinux-dev] WEB server for uClinux. Anders ?hlander wrote: > I am looking on to implement a WEB server on my uClinux device. > > Is there anyone that has some experience on that ? > > Which one is best ? I've been successfully using boa on an Intel Xscale embedded device running glibc/linux 2.4.22 (the snapgear 3.0 distribution). It has worked just fine. Boa supports web pages and CGIs, and although it is pretty basic, it met my needs. If the content you serve up is simply static HTML pages, images, and CGIs, boa should work fine. I'm now trying to get appWeb working (see my message right after yours). It has lots of features that boa doesn't have, but I'm having some trouble getting it working. Ben From mbeaulieu at eocycle.com Thu Mar 17 15:38:34 2005 From: mbeaulieu at eocycle.com (Martin Beaulieu) Date: Thu, 17 Mar 2005 15:38:34 -0500 Subject: [uClinux-dev] RTAI and uClinux References: <001b01c52b30$26ac5560$6202a8c0@flanders> Message-ID: <009301c52b31$4b300fc0$8000a8c0@MartinLogiciel> Hi Mike, I make a patch for rtai-24.1.12 for uClinux. It's for motorolla board. Maybe it work in rtai-24.1.13 :) bye Martin Beaulieu D?veloppeur Logiciel ?ocycle Technologie 418-833-0926 poste 224 ----- Original Message ----- From: "mike" To: "'uClinux development list'" Sent: Thursday, March 17, 2005 3:30 PM Subject: [uClinux-dev] RTAI and uClinux > Hello, > > Does anyone have a recent (uClinux-2.4.28) rthal patch for the uClinux > kernel. I have tried to use the "patch-rthal-2.4.17-uc0-m68knommu" from > the > RTAI-24.1.13 release but it does not apply cleanly. I can try to apply it > by hand but figured there must be someone who has done this already. > > Thanks, > > Mike Lavender > > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org > > -------------- next part -------------- A non-text attachment was scrubbed... Name: rtai-24.1.12_MCF5407.patch Type: application/octet-stream Size: 55172 bytes Desc: not available URL: From info at hostinthebox.net Thu Mar 17 15:40:54 2005 From: info at hostinthebox.net (dan) Date: Thu, 17 Mar 2005 13:40:54 -0700 Subject: [uClinux-dev] WEB server for uClinux. In-Reply-To: <20050317190111.887AC1A11D@uclinux.org> References: <20050317190111.887AC1A11D@uclinux.org> Message-ID: <4239EB56.6030202@hostinthebox.net> Anders ?hlander wrote: > Hi all > > I am looking on to implement a WEB server on my uClinux device. > > Is there anyone that has some experience on that ? > > Which one is best ? > > > > As I understand it is the httpd and the boa server that are available, > is that true ? > > Are there more to chose of ? > > > > Has anyone used some host tools for generate WEB pages that is > functioning with uClinux server ? > > > > Thanks for your input. > > > > BR > > /Anders ?hlander Long time lurker, first time poster. Hello. I've been playing around with ttylinux for a long time now. It's not quite an embedded system, but rather, a minimalistic Linux installation. My goal in the near future is to try to port the entire distribution to some smaller architecture. Most of my work for the past few weeks has been into the iPaq, because I believe it's quite capable. The only major hurdle I've come across is finding a suitable bootloader for the device to the point where the device can boot stand-alone, and not have to rely on any additional outside components (tftpboot, netboot, etc etc). I am saying this with absolutely no disrespect to the uClinux guys, by talking abotu another distribution. I believe the uClinux target platform is far more minimalistic than a PDA or other handheld, which will eventually be my target platform. Keep up the good work guys, I love to watch the development happen. Anyway... ttylinux comes with a small web server called simply thttpd (http://www.acme.com/software/thttpd/). I know that many people on this list have had experience with other tiny webservers, but I've found this to be extremely small, but at the same time, very functional if all you're after are the basics. The Apache Foundation also distributes a less-than-mature tiny webserver, but the name escapes me at the moment. Perhaps someone else can refresh my memory here? Hope that helps -dant From ben at thingmagic.com Thu Mar 17 15:49:47 2005 From: ben at thingmagic.com (Ben Giddings) Date: Thu, 17 Mar 2005 15:49:47 -0500 Subject: SV: [uClinux-dev] WEB server for uClinux. In-Reply-To: References: Message-ID: <4239ED6B.80708@thingmagic.com> Anders ?hlander wrote: > Hi Ben > Thanks for your answer. > > How do you get the information from the device to the browser ? > Do you write c-code that sends out HTML or is that any nicer way of doing > that ? Do you mean for device status and/or settings changes? We do that with CGIs written in C that contain a lot of printf() calls. It's far from the optimal solution, but the CGIs are small, so that's good. > Is there any way to use a tool to generate the WEB-site structure ? Like > pagemaker or something ? I don't know about pagemaker, but I know a lot of editors (GoLive, DreamWeaver, etc) can generate templates. The problem is, most template-using languages (PHP, Perl, Ruby, Python, etc.) are fairly large, and unsuitable for our application. > Can you use some security on the BOA server, like ssl or https ? I don't think you can. The two above limitations (lack of encryption, and no template system) is why we are currently looking at using appWeb. It is more flexible when it comes to those kinds of features, however it is also larger, and I'm not yet sure whether it is small enough for our needs. (That is, assuming we can even get it working) Ben From info at hostinthebox.net Thu Mar 17 16:11:28 2005 From: info at hostinthebox.net (dan) Date: Thu, 17 Mar 2005 14:11:28 -0700 Subject: [uClinux-dev] General Bootloader Question(s) Message-ID: <4239F280.6070909@hostinthebox.net> Hello, all - I have been practicing with the uClinux distribution for quite some time here - building toolchains, creating a sane building env, etc etc. At this point, I think I'm about ready to take on the world. However, what still baffles me, is how people are able to make bootloaders for specific devices. I would imagine that this varies depending on the architecture and platform used, but where would one start creating a bootloader? Are bootloaders even required on most embedded platforms? Embedded computers are something that I am, ashamed to say, still a bit new at. If said devices do use a bootloader, I suppose it would benefit me greatly if there was an opensource "all-in-one", if you will - a kit of some sort that helps you create a bootloader, if required. Again, assumuing that these devices actually need bootloaders. I've been learning a lot off this list for the past couple weeks, and would like to thank you all for just talking about your projects. It's interesting to read it all. Thanks -dant From Herbert.Yeung at uts.edu.au Thu Mar 17 16:52:46 2005 From: Herbert.Yeung at uts.edu.au (Herbert) Date: Fri, 18 Mar 2005 08:52:46 +1100 (EST) Subject: [uClinux-dev] Re: Getting flex/libfl.a to work on Cobra 5272 In-Reply-To: <20050315170008.B543D1A083@uclinux.org> Message-ID: Has anyone had any luck in the past in compiling libfl.a library using kernel 2.4.x and uClibc? Under "Force library compilation" of make xconfig there is an option to force compile the flex library. However, when this option is selected there is a complaint from make of the following error: "../MISC/alloca.c" missing. Thankyou for your assistance, Herb From davidm at snapgear.com Thu Mar 17 17:49:45 2005 From: davidm at snapgear.com (David McCullough) Date: Fri, 18 Mar 2005 08:49:45 +1000 Subject: [uClinux-dev] WEB server for uClinux. In-Reply-To: <4239EB56.6030202@hostinthebox.net> References: <20050317190111.887AC1A11D@uclinux.org> <4239EB56.6030202@hostinthebox.net> Message-ID: <20050317224945.GA9057@beast> Jivin dan lays it down ... > Anders ?hlander wrote: > >Hi all > > > >I am looking on to implement a WEB server on my uClinux device. > > > >Is there anyone that has some experience on that ? > > > >Which one is best ? Depends on your needs, we are using fnord from the uCLinux-dist now, we used to use boa from the dist. > >As I understand it is the httpd and the boa server that are available, > >is that true ? > > > >Are there more to chose of ? The uClinux-dist httpd servers are: fnord boa appWeb thttpd mini_httpd httpd and perhaps some others I have missed ;-) Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From jsujjavanich at syntech-fuelmaster.com Thu Mar 17 17:54:56 2005 From: jsujjavanich at syntech-fuelmaster.com (Jate Sujjavanich) Date: Thu, 17 Mar 2005 17:54:56 -0500 Subject: [uClinux-dev] fec, linux 2.6.11-uc0, MCF5235 Message-ID: It turns out that the Kconfig file under drivers/net was not set up properly. The FEC choice was dependent on M528x, M527x, etc., but not for 5235. I updated this, and fec.c now compiles. However, now booting causes the Coldfire to stop responding over BDM. The kernel messages get to RAMDISK Initialized. The crash occurs within "do_init_calls()". I am currently trying to change the kernel debug level. Can I do this by adding debug=0 to the compiled-in kernel string? - Jate ________________________________ From: uclinux-dev-bounces at uClinux.org [mailto:uclinux-dev-bounces at uClinux.org] On Behalf Of Jate Sujjavanich Sent: Thursday, March 17, 2005 12:00 PM To: uClinux development list Subject: [uClinux-dev] fec, linux 2.6.11-uc0, MCF5235 I have been porting linux 2.6.11-uc0 to the MCF5235. It boots successfully, but I am having a problem detecting the onboard fast ethernet controller. There is a SIOCGIFHWADDR error when dhcp tries to use the address, and there is no eth0 device. There are no kernel messages about network device detection (fec: *). I've added defines for the 5235 to fec.c and fec.h. The KS8721BL is specified in the driver. Under network devices, I have selected NETDEVICES & NET_ETHERNET under network device support. When/how does the kernel perform fec detection? What could be the source of my trouble? - Jate Sujjavanich From davidm at snapgear.com Thu Mar 17 17:55:26 2005 From: davidm at snapgear.com (David McCullough) Date: Fri, 18 Mar 2005 08:55:26 +1000 Subject: [uClinux-dev] General Bootloader Question(s) In-Reply-To: <4239F280.6070909@hostinthebox.net> References: <4239F280.6070909@hostinthebox.net> Message-ID: <20050317225526.GB9057@beast> Jivin dan lays it down ... > Hello, all - > > I have been practicing with the uClinux distribution for quite some time > here - building toolchains, creating a sane building env, etc etc. At > this point, I think I'm about ready to take on the world. > > However, what still baffles me, is how people are able to make > bootloaders for specific devices. I would imagine that this varies > depending on the architecture and platform used, but where would one > start creating a bootloader? Are bootloaders even required on most > embedded platforms? Embedded computers are something that I am, ashamed > to say, still a bit new at. Every embedded platform has a boot loader. They are usually a combo of assembler and some C. The initialise all the hardware needed to run anything (RAM, chip selects and so on). You need one, no question. uBoot seems to be pretty popular with uClinux systems these days. > If said devices do use a bootloader, I suppose it would benefit me > greatly if there was an opensource "all-in-one", if you will - a kit of > some sort that helps you create a bootloader, if required. Again, > assumuing that these devices actually need bootloaders. Look at uboot http://uboot.sourceforge.net/ If you are using Coldfire there is colilo. If you hunt around you should find some more. Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From info at hostinthebox.net Thu Mar 17 18:18:24 2005 From: info at hostinthebox.net (dan) Date: Thu, 17 Mar 2005 16:18:24 -0700 Subject: [uClinux-dev] General Bootloader Question(s) In-Reply-To: <20050317225526.GB9057@beast> References: <4239F280.6070909@hostinthebox.net> <20050317225526.GB9057@beast> Message-ID: <423A1040.2040505@hostinthebox.net> David McCullough wrote: > Jivin dan lays it down ... > >>Hello, all - >> >>I have been practicing with the uClinux distribution for quite some time >>here - building toolchains, creating a sane building env, etc etc. At >>this point, I think I'm about ready to take on the world. >> >>However, what still baffles me, is how people are able to make >>bootloaders for specific devices. I would imagine that this varies >>depending on the architecture and platform used, but where would one >>start creating a bootloader? Are bootloaders even required on most >>embedded platforms? Embedded computers are something that I am, ashamed >>to say, still a bit new at. > > > Every embedded platform has a boot loader. They are usually a combo of > assembler and some C. The initialise all the hardware needed to run > anything (RAM, chip selects and so on). > > You need one, no question. uBoot seems to be pretty popular with > uClinux systems these days. > > >>If said devices do use a bootloader, I suppose it would benefit me >>greatly if there was an opensource "all-in-one", if you will - a kit of >>some sort that helps you create a bootloader, if required. Again, >>assumuing that these devices actually need bootloaders. > > > Look at uboot > > http://uboot.sourceforge.net/ > > If you are using Coldfire there is colilo. If you hunt around you > should find some more. > > Cheers, > Davidm > David - I appreciate your reply, and I have a better understanding now. I remember talking with you back and forth via email about porting uClinux to my Palm IIIxe. I hope no one here thinks I'm crazy for once again looking into some options for placing uClinux on PDAs again. Totally not my area, but I am still very interested in that. So I'm going to try to give it another stab. Now that we know the device needs a bootloader, how might one go about placing the bootloader on the actual device? I would imagine that this isn't as simple as a one-line answer, as most devices differ so much, as far as the data input methods are concerned. I'm just extremely curious at this point, and am looking for a starting point to investigate this. Thanks again -dant From gerg at snapgear.com Thu Mar 17 19:18:12 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Fri, 18 Mar 2005 10:18:12 +1000 Subject: [uClinux-dev] fec, linux 2.6.11-uc0, MCF5235 In-Reply-To: References: Message-ID: <423A1E44.8000607@snapgear.com> Hi Jate, Jate Sujjavanich wrote: > I have been porting linux 2.6.11-uc0 to the MCF5235. It boots > successfully, but I am having a problem detecting the onboard fast > ethernet controller. There is a SIOCGIFHWADDR error when dhcp tries to > use the address, and there is no eth0 device. There are no kernel > messages about network device detection (fec: *). > > I've added defines for the 5235 to fec.c and fec.h. The KS8721BL is > specified in the driver. Under network devices, I have selected > NETDEVICES & NET_ETHERNET under network device support. > > When/how does the kernel perform fec detection? What could be the source > of my trouble? The FEC driver is just inited like any other driver, through the "module_init()" machanism. Start tarcing the code from fec_enet_module_init(). You should quickly be able to determine if that init routine is even getting called. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Thu Mar 17 20:04:44 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Fri, 18 Mar 2005 11:04:44 +1000 Subject: [uClinux-dev] General Bootloader Question(s) In-Reply-To: <20050317225526.GB9057@beast> References: <4239F280.6070909@hostinthebox.net> <20050317225526.GB9057@beast> Message-ID: <423A292C.6020302@snapgear.com> David McCullough wrote: > Jivin dan lays it down ... > >>Hello, all - >> >>I have been practicing with the uClinux distribution for quite some time >>here - building toolchains, creating a sane building env, etc etc. At >>this point, I think I'm about ready to take on the world. >> >>However, what still baffles me, is how people are able to make >>bootloaders for specific devices. I would imagine that this varies >>depending on the architecture and platform used, but where would one >>start creating a bootloader? Are bootloaders even required on most >>embedded platforms? Embedded computers are something that I am, ashamed >>to say, still a bit new at. > > > Every embedded platform has a boot loader. Well, no, they don't :-) You can modifiy the linux kernel startup code to init your hardware, and avoid using a boot loader entirely. I have done this on a few occassions. The "ROM" based start up code for the M5249C3 platform in the uClinux-2.4.x code is an example of exactly this. Having a boot loader can proivde you with some advantages though. For one you usually have some type of recovery mechanism if you load in a bad kernel. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From davidm at snapgear.com Thu Mar 17 20:34:01 2005 From: davidm at snapgear.com (David McCullough) Date: Fri, 18 Mar 2005 11:34:01 +1000 Subject: [uClinux-dev] General Bootloader Question(s) In-Reply-To: <423A292C.6020302@snapgear.com> References: <4239F280.6070909@hostinthebox.net> <20050317225526.GB9057@beast> <423A292C.6020302@snapgear.com> Message-ID: <20050318013401.GA32181@beast> Jivin Greg Ungerer lays it down ... ... > >Every embedded platform has a boot loader. > > Well, no, they don't :-) :-) -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From NLiu at smmi2.medison.com Thu Mar 17 22:16:20 2005 From: NLiu at smmi2.medison.com (Ning Liu) Date: Fri, 18 Mar 2005 11:16:20 +0800 Subject: [uClinux-dev] Problem about insmod a module into uClinux In-Reply-To: <42396E06.80802@cefriel.it> Message-ID: Thanks for your answer! I had added this choice and tested it.I can load a module from /lib/modules,like that /> insmod mymodule Using /lib/modules/2.4.24-uc0/kernel/mymodule/mymodule.o My Module init /> lsmod Module Size Used by Not tainted mymodule 232 0 (unused) /> rmmod mymodule My Module fini /> lsmod Module Size Used by Not tainted /> but I can't load it from nfs /var/tmp/uClinux-dist/modules/mymodule> ls -l -rw-r--r-- 1 1012 1012 449 Mar 17 2005 Makefile -rw-r--r-- 1 1012 1012 717 Aug 29 2003 mymodule.c -rwxrwxrwx 1 0 0 10296 Mar 18 2005 mymodule.o /var/tmp/uClinux-dist/modules/mymodule> rmmod mymodule My Module fini /var/tmp/uClinux-dist/modules/mymodule> insmod ./mymodule.o Using ./mymodule.o Unhandled fault: external abort on linefetch (F4) at 0x00000001 fault-common.c(97): start_code=0xc380040, start_stack=0xc3bff7c) pid 28: failed 7 My host system's fs is ext3 and the develop board uses ext2. I'm not sure whether it has any effect. Roberto Farina 2005-03-17 ?? 07:46 Please respond to roberto.farina To: Ning Liu cc: "'uclinux-dev at uClinux.org'" Subject: Re: [uClinux-dev] Problem about insmod a module into uClinux -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 [ ]Kernel module loader I think this is your problem; try enabling this choice. best regards Roberto Ning Liu wrote: > > Hi all : > > I have a problem about insmod a module into uClinux of Samsung > 44b0. > > In kernel configuration i have choosed > [*]Enable loadable module support > [ ]Set version...... > [ ]Kernel module loader > In BusyBox > [*]insmod > [*]insmod:load in kernel memory > [*]lsmod > [*]rmmod > But when i load a module aaa.o,it report: > /> chmod 777 /ramdisk/*.o > /> insmod /ramdisk/aaa.o > Using /ramdisk/aaa.o > Unhandled fault: external abort on linefetch (F4) at > 0x00000001 > fault-common.c(97): start_code=0xc380040, > start_stack=0xc3bff78) > pid 29: failed 7 > /> lsmod > Module Size Used by Not tainted > aaa 136 0 (uninitialized) > /> chmod 777 /lib > /lib: Read-only file system > And my aaa is so simple > #define MODULE > > #include > int init_module(void) > { > // printk ( "<1>Hello , World \n"); > return 0; > } > void cleanup_module(void) > { > // printk( "<1>Goodbye cruel world\n"); > } > Can any one help me? Thanks in advance! > > Best Regards. > Ninig Liu > > > ------------------------------------------------------------------------ > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org - -- Roberto Farina CEFRIEL ?Politecnico di Milano Via Fucini, 2 ?20133 Milano (Italy) p. +39 02 23954 345 - f. +39 02 23954 254 e. roberto.farina at cefriel.it - w. www.cefriel.it -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.7 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFCOW4GzF5ONZxALKMRAqLvAKC1yyYZUwjVUgrhh7ri9eDXyU4u2gCg5dzT 8yJIXF6l8d9N9ZpAxy5Ukn8= =mpBU -----END PGP SIGNATURE----- -------------- next part -------------- An HTML attachment was scrubbed... URL: From NLiu at smmi2.medison.com Thu Mar 17 23:39:22 2005 From: NLiu at smmi2.medison.com (Ning Liu) Date: Fri, 18 Mar 2005 12:39:22 +0800 Subject: [uClinux-dev] Problem about insmod a module into uClinux In-Reply-To: Message-ID: "Ning Liu" Sent by: uclinux-dev-bounces at uClinux.org 2005-03-18 ?? 11:16 To: roberto.farina at cefriel.it, gerg at snapgear.com cc: "'uclinux-dev at uClinux.org'" Subject: Re: [uClinux-dev] Problem about insmod a module into uClinux I found a strange thing is that the mymodule.o file in the kernel /lib/modules/2.4.24-uc0/kernel/mymodule/mymodule.o is look like this -rw-r--r-- 1 0 0 988 Jan 1 00:18 mymodule.o but in the host dir uClinux-dist/modules/mymodule it is larger -rwxr-xr-x 1 0 0 10296 Jan 1 00:20 mymodule.o the fist file can be loaded in where ever be copy ,but the second can't. Thanks for your answer! I had added this choice and tested it.I can load a module from /lib/modules,like that /> insmod mymodule Using /lib/modules/2.4.24-uc0/kernel/mymodule/mymodule.o My Module init /> lsmod Module Size Used by Not tainted mymodule 232 0 (unused) /> rmmod mymodule My Module fini /> lsmod Module Size Used by Not tainted /> but I can't load it from nfs /var/tmp/uClinux-dist/modules/mymodule> ls -l -rw-r--r-- 1 1012 1012 449 Mar 17 2005 Makefile -rw-r--r-- 1 1012 1012 717 Aug 29 2003 mymodule.c -rwxrwxrwx 1 0 0 10296 Mar 18 2005 mymodule.o /var/tmp/uClinux-dist/modules/mymodule> rmmod mymodule My Module fini /var/tmp/uClinux-dist/modules/mymodule> insmod ./mymodule.o Using ./mymodule.o Unhandled fault: external abort on linefetch (F4) at 0x00000001 fault-common.c(97): start_code=0xc380040, start_stack=0xc3bff7c) pid 28: failed 7 My host system's fs is ext3 and the develop board uses ext2. I'm not sure whether it has any effect. Roberto Farina 2005-03-17 ?? 07:46 Please respond to roberto.farina To: Ning Liu cc: "'uclinux-dev at uClinux.org'" Subject: Re: [uClinux-dev] Problem about insmod a module into uClinux -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 [ ]Kernel module loader I think this is your problem; try enabling this choice. best regards Roberto Ning Liu wrote: > > Hi all : > > I have a problem about insmod a module into uClinux of Samsung > 44b0. > > In kernel configuration i have choosed > [*]Enable loadable module support > [ ]Set version...... > [ ]Kernel module loader > In BusyBox > [*]insmod > [*]insmod:load in kernel memory > [*]lsmod > [*]rmmod > But when i load a module aaa.o,it report: > /> chmod 777 /ramdisk/*.o > /> insmod /ramdisk/aaa.o > Using /ramdisk/aaa.o > Unhandled fault: external abort on linefetch (F4) at > 0x00000001 > fault-common.c(97): start_code=0xc380040, > start_stack=0xc3bff78) > pid 29: failed 7 > /> lsmod > Module Size Used by Not tainted > aaa 136 0 (uninitialized) > /> chmod 777 /lib > /lib: Read-only file system > And my aaa is so simple > #define MODULE > > #include > int init_module(void) > { > // printk ( "<1>Hello , World \n"); > return 0; > } > void cleanup_module(void) > { > // printk( "<1>Goodbye cruel world\n"); > } > Can any one help me? Thanks in advance! > > Best Regards. > Ninig Liu > > > ------------------------------------------------------------------------ > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org - -- Roberto Farina CEFRIEL ?Politecnico di Milano Via Fucini, 2 ?20133 Milano (Italy) p. +39 02 23954 345 - f. +39 02 23954 254 e. roberto.farina at cefriel.it - w. www.cefriel.it -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.7 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFCOW4GzF5ONZxALKMRAqLvAKC1yyYZUwjVUgrhh7ri9eDXyU4u2gCg5dzT 8yJIXF6l8d9N9ZpAxy5Ukn8= =mpBU -----END PGP SIGNATURE----- _______________________________________________ uClinux-dev mailing list uClinux-dev at uClinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev at uclinux.org -------------- next part -------------- An HTML attachment was scrubbed... URL: From gerg at snapgear.com Fri Mar 18 00:35:00 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Fri, 18 Mar 2005 15:35:00 +1000 Subject: [uClinux-dev] next version of uClinux-dist? In-Reply-To: <007001c52ac0$914801f0$f103a8c0@kesllaptop3> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com><20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> <007001c52ac0$914801f0$f103a8c0@kesllaptop3> Message-ID: <423A6884.2040100@snapgear.com> Hi Andrey, Andrey Butok wrote: > Can anybody please tell me when the next version of uClinux distribution > will be available (after 20041215) ? I don't have a fixed timeframe for the next release. I would like to start a test release soon (next few weeks). > What is the usual frequency of uClinux-dist-xx release? I guess we are running at about 3 to 4 months between releases. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Fri Mar 18 01:10:04 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Fri, 18 Mar 2005 16:10:04 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <423951E6.5040607@itee.uq.edu.au> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> <20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> <423951E6.5040607@itee.uq.edu.au> Message-ID: <423A70BC.3060203@snapgear.com> Hi John, All, John Williams wrote: > Real-time CVS tracking. The model of periodic releases places pressure > on Dave and Greg (maybe! :) and is a bit inflexible. I see this a lot > on the Microblaze platform where every single system can be different. > The ability to create new vendors targets and immediately publish them > through CVS would be very valuable. It's getting so that I'm also > considering a custom uClinux-dist for microblaze. Probably the thing that holds me up most from releasing more often (and beleive me, I would really like to do that), is the amount of time it takes me just to do a spread of minimal testing. Making sure that a bunch of targets actually build and compile. You probably wouldn't beleive ho many things get accidentially broken very quickly :-( Tar'ing up a big glob of source doesn't take long in itself :-) The reason I do at least minimal build testing is that many people used to complain at me when their favorite target didn't at least build right in the current dist. Would people be more forgiving of a sometimes broken uClinux-dist, but have it released much more often? Some time back (circa 2000/2001) I tried keeping a "stable" branch and a testing branch, where the testing branch was released more frequently (I thing from memory I did it about once a month). Reality is though that the stable branch just didn't get any attention, and the testing branch never quite converged to a point where everything worked well to make it a stable branch). Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Fri Mar 18 01:11:32 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Fri, 18 Mar 2005 16:11:32 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050317123708.24195.qmail@web51903.mail.yahoo.com> References: <20050317123708.24195.qmail@web51903.mail.yahoo.com> Message-ID: <423A7114.5050405@snapgear.com> Hi Doug, Doug Kehn wrote: > --- "Friedrich, Lars" wrote: >>What are the disadvantages? >>- All old references break >>Websites, books, mailing list archives, domain names >>like uClinux.org .. >> > > This is an excellent point. > > Maybe instead of a name change, consideration could be > given to updating the www.uclinux.org website. Agreed. www.uclinux.org needs much attention. A lot if it is way out of date. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Fri Mar 18 01:14:09 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Fri, 18 Mar 2005 16:14:09 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050317141322.9A3A4E8A6@mail2.atmark-techno.com> References: <20050317123708.24195.qmail@web51903.mail.yahoo.com> <20050317132902.GA31588@beast> <20050317141322.9A3A4E8A6@mail2.atmark-techno.com> Message-ID: <423A71B1.3060600@snapgear.com> Hi All, Yasushi SHOJI wrote: > At Thu, 17 Mar 2005 23:29:02 +1000, > David McCullough wrote: >>I think uClinux-dist is (by definition of uC) the correct name, >>unfortunately most common usage in the linux kernel space means uC == !MMU. > > > well that's the whole point. _Unfortunately_, we have created the > brand-image of uC as *no MMU*. I don't think any amount of education will solve this. Beleive me, I have been telling a lot of people for a long time now :-) Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Fri Mar 18 01:16:44 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Fri, 18 Mar 2005 16:16:44 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <17013.1111069705@redhat.com> References: <20050317135217.GB31588@beast> <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> <20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> <423951E6.5040607@itee.uq.edu.au> <17013.1111069705@redhat.com> Message-ID: <423A724C.7030104@snapgear.com> Hi David, David Howells wrote: > David McCullough wrote: > > >>For example, the 2.4 kernel causes me much pain. It goes something >>like: > > > Can you get some of the stuff back into the vanilla 2.4 kernel? Won't happen with 2.4, it is essentially in maintaince mode. > And likewise > for 2.6. Everything that should be in 2.6 pretty much already is. I make sure of that. I send updates to Linus regularly. Merging recent uClinux specific changes into 2.6 is almost trivial, and that is what the 2.6 -uc patch series is all about. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From davidm at snapgear.com Fri Mar 18 01:20:38 2005 From: davidm at snapgear.com (David McCullough) Date: Fri, 18 Mar 2005 16:20:38 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <423A724C.7030104@snapgear.com> References: <20050317135217.GB31588@beast> <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> <20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> <423951E6.5040607@itee.uq.edu.au> <17013.1111069705@redhat.com> <423A724C.7030104@snapgear.com> Message-ID: <20050318062038.GB12070@beast> Jivin Greg Ungerer lays it down ... > Hi David, > > David Howells wrote: > >David McCullough wrote: > > > > > >>For example, the 2.4 kernel causes me much pain. It goes something > >>like: > > > > > >Can you get some of the stuff back into the vanilla 2.4 kernel? > > Won't happen with 2.4, it is essentially in maintaince mode. The actual 2.4 kernel merge isn't the bulk of the work at thi spoint, it's just the large number of kernel trees that get touched from my point of view :-) Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From yashi at atmark-techno.com Fri Mar 18 01:23:07 2005 From: yashi at atmark-techno.com (Yasushi SHOJI) Date: Fri, 18 Mar 2005 15:23:07 +0900 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050318062038.GB12070@beast> References: <20050317135217.GB31588@beast> <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> <20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> <423951E6.5040607@itee.uq.edu.au> <17013.1111069705@redhat.com> <423A724C.7030104@snapgear.com> <20050318062038.GB12070@beast> Message-ID: <20050318062307.2E112ED44@mail2.atmark-techno.com> At Fri, 18 Mar 2005 16:20:38 +1000, David McCullough wrote: > > Jivin Greg Ungerer lays it down ... > > Hi David, > > > > David Howells wrote: > > >David McCullough wrote: > > > > > > > > >>For example, the 2.4 kernel causes me much pain. It goes something > > >>like: > > > > > > > > >Can you get some of the stuff back into the vanilla 2.4 kernel? > > > > Won't happen with 2.4, it is essentially in maintaince mode. > > The actual 2.4 kernel merge isn't the bulk of the work at thi spoint, it's > just the large number of kernel trees that get touched from my point of > view :-) oops. I still have pending patches for ppc arch merge. would it be acceptable? -- yashi From davidm at snapgear.com Fri Mar 18 01:32:05 2005 From: davidm at snapgear.com (David McCullough) Date: Fri, 18 Mar 2005 16:32:05 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050318062307.2E112ED44@mail2.atmark-techno.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> <20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> <423951E6.5040607@itee.uq.edu.au> <17013.1111069705@redhat.com> <423A724C.7030104@snapgear.com> <20050318062038.GB12070@beast> <20050318062307.2E112ED44@mail2.atmark-techno.com> Message-ID: <20050318063205.GA645@beast> Jivin Yasushi SHOJI lays it down ... > At Fri, 18 Mar 2005 16:20:38 +1000, > David McCullough wrote: > > > > Jivin Greg Ungerer lays it down ... > > > Hi David, > > > > > > David Howells wrote: > > > >David McCullough wrote: > > > > > > > > > > > >>For example, the 2.4 kernel causes me much pain. It goes something > > > >>like: > > > > > > > > > > > >Can you get some of the stuff back into the vanilla 2.4 kernel? > > > > > > Won't happen with 2.4, it is essentially in maintaince mode. > > > > The actual 2.4 kernel merge isn't the bulk of the work at thi spoint, it's > > just the large number of kernel trees that get touched from my point of > > view :-) > > oops. I still have pending patches for ppc arch merge. would it be > acceptable? Most likely :-) Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From bennyc at rtunet.com Fri Mar 18 01:56:19 2005 From: bennyc at rtunet.com (Benny Chen) Date: Fri, 18 Mar 2005 16:56:19 +1000 Subject: [uClinux-dev] Unable to Register IRQ number of external GPIO interrupt line. GPIOINTR Message-ID: <200503180656.j2I6uMFp007831@serv0.rtunet.com> Hi there, I am writing this RS485 specific device driver for an uart connected externally to the Cirrus EP9312 target (ARM9). However, I have problem registering the IRQ_NR of the GPIOINTR line in my device driver. A normal call to request_irq(59, my_handler, SA_INTERRUPT, ?mydriver?,NULL) in my driver returns a ?EINVAL error, not allowing me to register this interrupt number. The interrupt number 59 is taken from the VIC Interrupt Source of the 9312 User Guide for GPIOINTR line. I have set up the VIC2IntEnable register of the Vectored Interrupt Controller to enable this interrupt line and had read the same register to verified that the register is configured as per expected. Do anyone know why the Kernel do not know that this irq number is valid? Is there anything else that I have not done to enable this irq line? Thank you so much for your help. Cheers, Benny -- No virus found in this outgoing message. Checked by AVG Anti-Virus. Version: 7.0.308 / Virus Database: 266.7.3 - Release Date: 15/03/2005 -------------- next part -------------- An HTML attachment was scrubbed... URL: From paul.mcgougan at braintree.com.au Fri Mar 18 03:06:37 2005 From: paul.mcgougan at braintree.com.au (Paul McGougan) Date: Fri, 18 Mar 2005 18:06:37 +1000 Subject: [uClinux-dev] Adding a new Platform to a 2.6 kernel build Message-ID: There is a good document in the "Documentation" branch of the dist which I have previously used for adding a new platform to a 2.4 kernel, but are there any similar instructions for doing the same under a 2.6 kernel? The 2.4 instructions don't appear to be applicable to a 2.6 build. Regards, Paul McGougan Technology Development Manager Braintree Communications Pty Ltd -- This information together with any attachments is for the use of the intended recipient(s) only and may contain confidential and/or privileged information and is subject to copyright. If you have received this email in error please inform the sender as quickly as possible and delete this email and any copies of this information from your computer system network. If you are not the intended recipient of this email, you must not copy, distribute or take any action(s) that relies on this information. Any form of disclosure, modification, distribution and/or publication of this email is strictly prohibited. From maverick2004 at 21cn.com Fri Mar 18 04:03:12 2005 From: maverick2004 at 21cn.com (Maverick Gao) Date: Fri, 18 Mar 2005 17:03:12 +0800 Subject: [uClinux-dev] where to download fsck Message-ID: <002001c52b99$8c3150c0$4400a8c0@gaojunbin> hi all, anyone can tell me where to download a tool to fsck fat and fat32 under uClinux. Best Regard. Maverick Gao -------------- next part -------------- An HTML attachment was scrubbed... URL: From gerg at snapgear.com Fri Mar 18 06:16:15 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Fri, 18 Mar 2005 21:16:15 +1000 Subject: [uClinux-dev] Adding a new Platform to a 2.6 kernel build In-Reply-To: References: Message-ID: <423AB87F.906@snapgear.com> Hi Paul, Paul McGougan wrote: > There is a good document in the "Documentation" branch of the dist which I > have previously used for adding a new platform to a 2.4 kernel, but are > there any similar instructions for doing the same under a 2.6 kernel? I don't know of any. > The 2.4 instructions don't appear to be applicable to a 2.6 build. The general idea is the same, but the details of the kernel changes are a little different. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From deltha at analog.ro Fri Mar 18 07:04:30 2005 From: deltha at analog.ro (Rujinschi Remus) Date: Fri, 18 Mar 2005 14:04:30 +0200 Subject: [uClinux-dev] nommu for m68k shared library currently supported Message-ID: <87165196515.20050318140430@analog.ro> Hello , I would like to know if there's shared library currently supported for M68K family also if there's any console/uart ircda jffs2 drivers avaliable. Thanks in advance, Rujinschi Remus mailto:deltha at analog.ro -- This message was scanned for spam and viruses by BitDefender. For more information please visit http://linux.bitdefender.com/ From rdkehn at yahoo.com Fri Mar 18 08:10:19 2005 From: rdkehn at yahoo.com (Doug Kehn) Date: Fri, 18 Mar 2005 05:10:19 -0800 (PST) Subject: [uClinux-dev] Defining Variables in Header Files Message-ID: <20050318131019.78898.qmail@web51901.mail.yahoo.com> I've been noticing that more and more source out there (e.g. ISC DHCP 3.0.2) is defining variables in header files. The header file is included in more than one source file which results in "multiple definition of `...'" compiler error in uClinux. However, the same source compiles fine on x86. Has anyone else noticed this? Thanks, ...doug __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com From rdkehn at yahoo.com Fri Mar 18 11:58:22 2005 From: rdkehn at yahoo.com (Doug Kehn) Date: Fri, 18 Mar 2005 08:58:22 -0800 (PST) Subject: [uClinux-dev] Defining Variables in Header Files Message-ID: <20050318165822.31716.qmail@web51905.mail.yahoo.com> -- Doug Kehn wrote: > I've been noticing that more and more source out > there (e.g. ISC DHCP 3.0.2) is defining variables > in header files. The header file is included in > more than one source file which results in > "multiple definition of `...'" compiler error in > uClinux. However, the same source compiles fine > on x86. Has anyone else noticed this? The reason for the compile errors with "multiple definition ..." in uClinux is due to the -fno-common compiler option. (BTW, I'm compiling for the IXP425 with the arm-linux toolchain.) If I add this compiler option to the Makefile(s) when building the same source on my host system (x86), I get the same "multiple definition ..." compiler errors. I prefer the uClinux way (i.e. use the -fno-common compiler option). Maybe I'm old-school here but I prefer to define variables in C files and declare them (if need be) with extern in H files. This little nuance makes porting annoying. Regards, ...doug __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/ From basicmark at yahoo.com Fri Mar 18 13:12:21 2005 From: basicmark at yahoo.com (Mark Underwood) Date: Fri, 18 Mar 2005 18:12:21 +0000 (GMT) Subject: [uClinux-dev] Unable to Register IRQ number of external GPIO interrupt line. GPIOINTR In-Reply-To: 6667 Message-ID: <20050318181221.97584.qmail@web30310.mail.mud.yahoo.com> Hi Benny, Just I quick thought, if I remember correctly there is a #define MAX_IRQ or some such in each platforms irqs.h, is this bigger then 59? Best Regards, Mark --- Benny Chen wrote: > Hi there, > I am writing this RS485 specific device driver for > an uart connected > externally to the Cirrus EP9312 target (ARM9). > However, > I have problem registering the IRQ_NR of the > GPIOINTR line in my device > driver. > > A normal call to request_irq(59, my_handler, > SA_INTERRUPT, ?mydriver?,NULL) > in my driver returns a ?EINVAL error, not allowing > me to register this > interrupt number. > > The interrupt number 59 is taken from the VIC > Interrupt Source of the 9312 > User Guide for GPIOINTR line. > > I have set up the VIC2IntEnable register of the > Vectored Interrupt > Controller to enable this interrupt line and had > read the same register to > verified that the register is configured as per > expected. > > Do anyone know why the Kernel do not know that this > irq number is valid? Is > there anything else that I have not done to enable > this irq line? > > Thank you so much for your help. > > Cheers, > Benny > > -- > No virus found in this outgoing message. > Checked by AVG Anti-Virus. > Version: 7.0.308 / Virus Database: 266.7.3 - Release > Date: 15/03/2005 > > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org Send instant messages to your online friends http://uk.messenger.yahoo.com From mdurrant at uclinux.org Fri Mar 18 14:49:11 2005 From: mdurrant at uclinux.org (Michael Durrant) Date: Fri, 18 Mar 2005 14:49:11 -0500 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <423A7114.5050405@snapgear.com> References: <20050317123708.24195.qmail@web51903.mail.yahoo.com> <423A7114.5050405@snapgear.com> Message-ID: <423B30B7.8020306@uclinux.org> Greg Ungerer wrote: > Hi Doug, > > Doug Kehn wrote: > >> --- "Friedrich, Lars" wrote: >> >>> What are the disadvantages? >>> - All old references break >>> Websites, books, mailing list archives, domain names >>> like uClinux.org .. >>> >> >> This is an excellent point. >> >> Maybe instead of a name change, consideration could be >> given to updating the www.uclinux.org website. > > > Agreed. www.uclinux.org needs much attention. > A lot if it is way out of date. I am agreed on the clean up of the site. Volunteers are welcome to contact me directly as well as forwarding recommended text changes on existing pages etc etc... email: mike at uclinux.org For me the uClinux brand issue is important as we have excellent recognition with the name. It has been with the public community since 1998. As for the MMU versus NO MMU perception we need to start the education at the web site end that uClinux is a operating system solution for the embedded space regardless of whether the CPU has a full, partial or no MMU. The selection of c library, kernel and user applications that the distribution provides is clearly beyond the exclusive realm of target support for mmu-less processors. The distribution that Greg, David and community have evolved is clearly an embedded distribution and fits well within the original goals we have for the uClinux "Embedded Linux / Microcontroller Project" Michael Durrant mike at uclinux.org From thritter_list at gmx.at Fri Mar 18 16:31:43 2005 From: thritter_list at gmx.at (Thomas Ritter) Date: Fri, 18 Mar 2005 22:31:43 +0100 (MET) Subject: [uClinux-dev] armnommu support Message-ID: <7041.1111181503@www68.gmx.net> The latest version for armnommu I found is linux-2.6.9 (Nov 2004) from http://opensrc.sec.samsung.com Did I miss a switch in the armnommu maintainance ? Since 2.6.10 there are no armnommu patches on www.uclinux.org available. regards, Thomas -- "Happy ProMail" bis 24. M?rz: http://www.gmx.net/de/go/promail Zum 6. Geburtstag gibt's GMX ProMail jetzt 66 Tage kostenlos! From robert at schwebel.de Fri Mar 18 19:24:57 2005 From: robert at schwebel.de (Robert Schwebel) Date: Sat, 19 Mar 2005 01:24:57 +0100 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050315081337.0A198633A@mail2.atmark-techno.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <20050315081337.0A198633A@mail2.atmark-techno.com> Message-ID: <20050319002457.GI11709@pengutronix.de> On Tue, Mar 15, 2005 at 05:13:37PM +0900, Yasushi SHOJI wrote: > BTW, I enjoyed a session at FOSDEM about Bitbake. You may also want to have a look at PTXdist: http://www.pengutronix.de/software/ptxdist_en.html It mostly follows the Kconfig+Makefiles way, but in contrast to the uCLinux dist you don't have to download multi-ten megabyte images before you can start. It catches all the necessary archives from the network when they are needed. Robert -- Dipl.-Ing. Robert Schwebel | http://www.pengutronix.de Pengutronix - Linux Solutions for Science and Industry Handelsregister: Amtsgericht Hildesheim, HRA 2686 Hannoversche Str. 2, 31134 Hildesheim, Germany Phone: +49-5121-206917-0 | Fax: +49-5121-206917-9 From robert at schwebel.de Fri Mar 18 19:37:21 2005 From: robert at schwebel.de (Robert Schwebel) Date: Sat, 19 Mar 2005 01:37:21 +0100 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <423951E6.5040607@itee.uq.edu.au> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> <20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> <423951E6.5040607@itee.uq.edu.au> Message-ID: <20050319003721.GJ11709@pengutronix.de> Hi, On Thu, Mar 17, 2005 at 07:46:14PM +1000, John Williams wrote: > This is a good discussion to be having - really I think the bigger > question is "where is uclinux-dist going?" Some thoughts from the PTXdist competitor ;) When I started to search for a build system some years ago I also had a look at the uCLinux build system and in fact copied a lot of it's ideas into PTXdist. But there are several things which havn't been acceptable to what I had in mind. > It's pretty agile. dist-ifying an existing app is getting easier and > easier - witness some apps now that are based on generic GNU configure > setups. How is the process of feeding uCLinux modifications back to upstream solved today? Last time I looked I had the impression that it does more follow a "we port it when we need it" approach, with the consequence that it worked then for people who use the uCLinux dist. Is there some kind of a "we feed patches upstream" policy in uCLinux-dist? We have a patch repository for PTXdist which is applied on the fly to the original sources when they are being built. So there is a clear separation between patches and upstream sources; patches are reviewed on a regular base and feeded to the upstream maintainer. > It's system-oriented (e.g creates complete embedded systems - kernel, > apps, libs, modules, ...) > > It's Free. I'm not very familiar with other embedded distributions, but > I do believe e.g. that several thousand dollars for montavista linux is > excessive. Dito with PTXdist. > Some things I'd like to see in uClinux-dist: > > Real-time CVS tracking. The model of periodic releases places pressure > on Dave and Greg (maybe! :) and is a bit inflexible. I see this a lot > on the Microblaze platform where every single system can be different. > The ability to create new vendors targets and immediately publish them > through CVS would be very valuable. It's getting so that I'm also > considering a custom uClinux-dist for microblaze. Hmmm, if you have real customer projects you at least need some stability points, and you need control over it. Having a RCS system is cool (we have svn for PTXdist), but that doesn't mean that lots of people should get write access. Building systems is a complex and fragile thing and should be supervised by a maintainer who knows what he does. Robert -- Dipl.-Ing. Robert Schwebel | http://www.pengutronix.de Pengutronix - Linux Solutions for Science and Industry Handelsregister: Amtsgericht Hildesheim, HRA 2686 Hannoversche Str. 2, 31134 Hildesheim, Germany Phone: +49-5121-206917-0 | Fax: +49-5121-206917-9 From mike at steroidmicros.com Fri Mar 18 21:36:37 2005 From: mike at steroidmicros.com (Mike Lavender) Date: Fri, 18 Mar 2005 18:36:37 -0800 Subject: [uClinux-dev] Sending email Message-ID: <002501c52c2c$7cf52170$6202a8c0@flanders> Hello, Does anybody have example code for sending an email from within a C program on uClinux. I have compiled in the mail program and it works from the command line, but I am not sure how to invoke it from a C program and get the message body in. I have tried using "popen" but it doesn't seem to work (possibly do to a fork issue?). Is this how people are sending emails is by invoking a command line from a C program like in the Linux desktop world with sendmail? Or are people adding smtp code directly to their application? Thank you, Mike Lavender From hyok.choi at samsung.com Fri Mar 18 23:02:20 2005 From: hyok.choi at samsung.com (Hyok S. Choi) Date: Sat, 19 Mar 2005 13:02:20 +0900 Subject: [uClinux-dev] armnommu support In-Reply-To: <7041.1111181503@www68.gmx.net> Message-ID: <0IDK00BENZ9Y1D@mmp2.samsung.com> Dear armnommu users, The armnommu support is under development phase to fully share the arm dir. It has taken some time because of some issue of armulator. Please wait a while, and it will be provided. :-) Regards, Hyok HYOK S. CHOI Engineer (System/Kernel) Digital Media R&D Center Samsung Electronics Co.,Ltd. tel: +82-31-200-8594/5723 fax: +82-31-200-3427 e-mail: hyok.choi at samsung.com [Linux 2.6 armnommu Kernel Maintainer] http://opensrc.sec.samsung.com > -----Original Message----- > From: uclinux-dev-bounces at uclinux.org > [mailto:uclinux-dev-bounces at uclinux.org] On Behalf Of Thomas Ritter > Sent: Saturday, March 19, 2005 6:32 AM > To: uclinux-dev at uclinux.org > Subject: [uClinux-dev] armnommu support > > The latest version for armnommu I found is linux-2.6.9 (Nov > 2004) from http://opensrc.sec.samsung.com > > Did I miss a switch in the armnommu maintainance ? > Since 2.6.10 there are no armnommu patches on www.uclinux.org > available. > > regards, > Thomas > > -- > "Happy ProMail" bis 24. Mdrz: > http://www.gmx.net/de/go/promail Zum 6. Geburtstag gibt's GMX > ProMail jetzt 66 Tage kostenlos! > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uclinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org > > From srinivas.vutukuri at gmail.com Fri Mar 18 23:24:19 2005 From: srinivas.vutukuri at gmail.com (srinivas naga vutukuri) Date: Sat, 19 Mar 2005 09:54:19 +0530 Subject: [uClinux-dev] Sending email Mike Lavender Message-ID: Hi Mike, just check, whether usage of system will work out... system ("mailprogramme name"); best regards, srinivas. From bennyc at rtunet.com Fri Mar 18 23:45:39 2005 From: bennyc at rtunet.com (Benny Chen) Date: Sat, 19 Mar 2005 14:45:39 +1000 Subject: [uClinux-dev] Unable to Register IRQ number of external GPIO interrupt line. GPIOINTR In-Reply-To: <20050318181221.97584.qmail@web30310.mail.mud.yahoo.com> Message-ID: <200503190445.j2J4jgFZ010774@serv0.rtunet.com> Hi Mark, Not quite the NR_IRQS constant as it is set to 128. However, while looking for that I found the code that initialises the target IRQ and found that in the irqs.h the GPIOINR bits has been masked and set to invalid. After changing that my driver is now happy as irq 59 is allocated as seen in /proc/interrupts. Thank you for your help. Cheers, Benny -----Original Message----- From: Mark Underwood [mailto:basicmark at yahoo.com] Sent: Saturday, 19 March 2005 4:12 AM To: Benny Chen Cc: uclinux-dev at uclinux.org Subject: Re: [uClinux-dev] Unable to Register IRQ number of external GPIO interrupt line. GPIOINTR Hi Benny, Just I quick thought, if I remember correctly there is a #define MAX_IRQ or some such in each platforms irqs.h, is this bigger then 59? Best Regards, Mark --- Benny Chen wrote: > Hi there, > I am writing this RS485 specific device driver for > an uart connected > externally to the Cirrus EP9312 target (ARM9). > However, > I have problem registering the IRQ_NR of the > GPIOINTR line in my device > driver. > > A normal call to request_irq(59, my_handler, > SA_INTERRUPT, ?mydriver?,NULL) > in my driver returns a ?EINVAL error, not allowing > me to register this > interrupt number. > > The interrupt number 59 is taken from the VIC > Interrupt Source of the 9312 > User Guide for GPIOINTR line. > > I have set up the VIC2IntEnable register of the > Vectored Interrupt > Controller to enable this interrupt line and had > read the same register to > verified that the register is configured as per > expected. > > Do anyone know why the Kernel do not know that this > irq number is valid? Is > there anything else that I have not done to enable > this irq line? > > Thank you so much for your help. > > Cheers, > Benny > > -- > No virus found in this outgoing message. > Checked by AVG Anti-Virus. > Version: 7.0.308 / Virus Database: 266.7.3 - Release > Date: 15/03/2005 > > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uClinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org Send instant messages to your online friends http://uk.messenger.yahoo.com -- No virus found in this incoming message. Checked by AVG Anti-Virus. Version: 7.0.308 / Virus Database: 266.7.4 - Release Date: 18/03/2005 -- No virus found in this outgoing message. Checked by AVG Anti-Virus. Version: 7.0.308 / Virus Database: 266.7.4 - Release Date: 18/03/2005 From yashi at atmark-techno.com Sat Mar 19 02:06:27 2005 From: yashi at atmark-techno.com (Yasushi SHOJI) Date: Sat, 19 Mar 2005 16:06:27 +0900 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050319002457.GI11709@pengutronix.de> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <20050315081337.0A198633A@mail2.atmark-techno.com> <20050319002457.GI11709@pengutronix.de> Message-ID: <20050319070627.50E1B6341@mail2.atmark-techno.com> Hi Robert, At Sat, 19 Mar 2005 01:24:57 +0100, Robert Schwebel wrote: > > On Tue, Mar 15, 2005 at 05:13:37PM +0900, Yasushi SHOJI wrote: > > BTW, I enjoyed a session at FOSDEM about Bitbake. > > You may also want to have a look at PTXdist: > > http://www.pengutronix.de/software/ptxdist_en.html ah, bummer. I should have stopped by and had some chat while I was at embedded workd 2005. maybe next year ;) > It mostly follows the Kconfig+Makefiles way, but in contrast to the > uCLinux dist you don't have to download multi-ten megabyte images before > you can start. It catches all the necessary archives from the network > when they are needed. ok so it's basically like bitbake but has same script interface like uclinux-dist? -- yashi From yashi at atmark-techno.com Sat Mar 19 02:46:28 2005 From: yashi at atmark-techno.com (Yasushi SHOJI) Date: Sat, 19 Mar 2005 16:46:28 +0900 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050319003721.GJ11709@pengutronix.de> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> <20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> <423951E6.5040607@itee.uq.edu.au> <20050319003721.GJ11709@pengutronix.de> Message-ID: <20050319074628.C3F226341@mail2.atmark-techno.com> At Sat, 19 Mar 2005 01:37:21 +0100, Robert Schwebel wrote: > > How is the process of feeding uCLinux modifications back to upstream > solved today? Last time I looked I had the impression that it does more > follow a "we port it when we need it" approach, with the consequence > that it worked then for people who use the uCLinux dist. Is there some > kind of a "we feed patches upstream" policy in uCLinux-dist? except kernels, I don't think we have it. > We have a patch repository for PTXdist which is applied on the fly to > the original sources when they are being built. So there is a clear > separation between patches and upstream sources; patches are reviewed on > a regular base and feeded to the upstream maintainer. I'm pretty sure this will put much more burden to the dist maintainer if community doesn't help them. and from my experience, not everybody is good at managing patches even with tools like a sane version control system (vcs) or patch management system (pms). > > Some things I'd like to see in uClinux-dist: > > > > Real-time CVS tracking. The model of periodic releases places pressure > > on Dave and Greg (maybe! :) and is a bit inflexible. I see this a lot > > on the Microblaze platform where every single system can be different. > > The ability to create new vendors targets and immediately publish them > > through CVS would be very valuable. It's getting so that I'm also > > considering a custom uClinux-dist for microblaze. > > Hmmm, if you have real customer projects you at least need some > stability points, and you need control over it. Having a RCS system is > cool (we have svn for PTXdist), but that doesn't mean that lots of > people should get write access. Building systems is a complex and > fragile thing and should be supervised by a maintainer who knows what he > does. IMHO, you don't have to have a write access to the upstream repo to do that. but as John is saying, it's very convenient if we can retrieve patches others sent in. # or, vcs with chage propagation capability suite to the dist # community? plus if we, who has real customers (of your definition), send stabilization patch back to upstream, we can grow a helthy upstream repo. ;) # sure sometime we need special patch to do wacky thing for our # custmer, and that wacky thing might not be acceptable for upstream. regards, -- yashi From yashi at atmark-techno.com Sat Mar 19 03:56:57 2005 From: yashi at atmark-techno.com (Yasushi SHOJI) Date: Sat, 19 Mar 2005 17:56:57 +0900 Subject: [uClinux-dev] WEB server for uClinux. In-Reply-To: <20050317190111.887AC1A11D@uclinux.org> References: <20050317190111.887AC1A11D@uclinux.org> Message-ID: <20050319085657.EE92D6373@mail2.atmark-techno.com> At Thu, 17 Mar 2005 20:01:19 +0100, Anders ?hlander wrote: > > I am looking on to implement a WEB server on my uClinux device. > > Is there anyone that has some experience on that ? > > Which one is best ? depends. I've been using thttpd. but it has a bug in post method. # pipe won't work well with vfork. -- yashi From robert at schwebel.de Sat Mar 19 05:57:25 2005 From: robert at schwebel.de (Robert Schwebel) Date: Sat, 19 Mar 2005 11:57:25 +0100 Subject: [uClinux-dev] armnommu support In-Reply-To: <0IDK00BENZ9Y1D@mmp2.samsung.com> References: <7041.1111181503@www68.gmx.net> <0IDK00BENZ9Y1D@mmp2.samsung.com> Message-ID: <20050319105724.GN11709@pengutronix.de> Hyok, On Sat, Mar 19, 2005 at 01:02:20PM +0900, Hyok S. Choi wrote: > The armnommu support is under development phase to fully share the arm dir. > It has taken some time because of some issue of armulator. > > Please wait a while, and it will be provided. :-) Do you develop on a publically available repository? SVN or something like that? Robert -- Dipl.-Ing. Robert Schwebel | http://www.pengutronix.de Pengutronix - Linux Solutions for Science and Industry Handelsregister: Amtsgericht Hildesheim, HRA 2686 Hannoversche Str. 2, 31134 Hildesheim, Germany Phone: +49-5121-206917-0 | Fax: +49-5121-206917-9 From robert at schwebel.de Sat Mar 19 06:06:25 2005 From: robert at schwebel.de (Robert Schwebel) Date: Sat, 19 Mar 2005 12:06:25 +0100 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050319070627.50E1B6341@mail2.atmark-techno.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <20050315081337.0A198633A@mail2.atmark-techno.com> <20050319002457.GI11709@pengutronix.de> <20050319070627.50E1B6341@mail2.atmark-techno.com> Message-ID: <20050319110625.GO11709@pengutronix.de> On Sat, Mar 19, 2005 at 04:06:27PM +0900, Yasushi SHOJI wrote: > ah, bummer. I should have stopped by and had some chat while I was at > embedded workd 2005. maybe next year ;) Definitely :-) > ok so it's basically like bitbake but has same script interface like > uclinux-dist? What do you mean with script interface? We use the kernel config system Kconfig for "make menuconfig" like stuff, just like the uCLinux distro. PTXdist's backend is one huge logical Makefile, consisting separate Makefiles for all the included software packages, so you can make inter-packet dependencies with normal make mechanisms. Ontop of that we have support for building toolchains with crosstool; PTXdist adds a menuconfig interface to crosstool, so you can easily select which versions shall be used. We have standard targets, called projects, and it is possible to use in-tree and off-tree (private) projects, for example if you do your private customer project which isn't supposed to go into the standard tarball. Then, lots of automatic compile and run tests which help to notice if something has been broken. Just support for nommu targets is not really tested very well yet :-) Robert -- Dipl.-Ing. Robert Schwebel | http://www.pengutronix.de Pengutronix - Linux Solutions for Science and Industry Handelsregister: Amtsgericht Hildesheim, HRA 2686 Hannoversche Str. 2, 31134 Hildesheim, Germany Phone: +49-5121-206917-0 | Fax: +49-5121-206917-9 From robert at schwebel.de Sat Mar 19 06:21:55 2005 From: robert at schwebel.de (Robert Schwebel) Date: Sat, 19 Mar 2005 12:21:55 +0100 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050319074628.C3F226341@mail2.atmark-techno.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> <20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> <423951E6.5040607@itee.uq.edu.au> <20050319003721.GJ11709@pengutronix.de> <20050319074628.C3F226341@mail2.atmark-techno.com> Message-ID: <20050319112155.GP11709@pengutronix.de> On Sat, Mar 19, 2005 at 04:46:28PM +0900, Yasushi SHOJI wrote: > > How is the process of feeding uCLinux modifications back to upstream > > solved today? Last time I looked I had the impression that it does more > > follow a "we port it when we need it" approach, with the consequence > > that it worked then for people who use the uCLinux dist. Is there some > > kind of a "we feed patches upstream" policy in uCLinux-dist? > > except kernels, I don't think we have it. I think this is a very important topic. One of the uggliest problems these communities have is lack of feedback. Especially commercial users are really happy to get stuff from the community but if it works "somehow", they don't care about The Right Solution (TM) for a problem and they don't care about feeding back things to upstream. As a commecial Embedded Linux company, we have to do something to fullfill our customers needs to have something that "just works". Being active in the Linux community for >12 years now, we know quite well that long term Open Source will only fulfill it's image of being better, more stable and superior to closed source stuff when there is an active community process with *feedback*. Our patch system mainly follows what Dan Kegel has invented for crosstool: a repository with patches, sorted by upstream project name. PTXdist automatically checks if there are patches and applies them. This way you can always see what's needed and you can easily feed things upstream. Here's how it looks like: http://www.pengutronix.de/software/ptxdist/patches-cvs/ > I'm pretty sure this will put much more burden to the dist maintainer > if community doesn't help them. and from my experience, not everybody > is good at managing patches even with tools like a sane version > control system (vcs) or patch management system (pms). I care about contributors following some rules about patches; we recently introduced that for new patches this header has to be added: http://www.pengutronix.de/software/ptxdist/patches-cvs/header which adds all the necessary information about what the patch tries to fix, if it is already fixed in upstream versions etc. But you are right, more help from the community would surely be helpful. > IMHO, you don't have to have a write access to the upstream repo to do > that. but as John is saying, it's very convenient if we can retrieve > patches others sent in. Ack. Usually patches fromt the community are added pretty fast, taken that they are not too intrusive, because that needs deeper investigation. At the very last, stability is the main concern, because our trees are constantly used in critical customer applications. One of the most important things is IMHO testing. We have compile-tests in PTXdist for quite some time now, which look like this: Automatic Compilation Test -------------------------- start: Sa M?r 19 11:05:03 CET 2005 config...: i586-generic-glibc date.....: Sa M?r 19 11:05:04 CET 2005 user.....: robert at himalia buildtime: 0h20m4s result...: 0 config...: visbox date.....: Sa M?r 19 11:25:17 CET 2005 user.....: robert at himalia buildtime: 0h28m31s result...: 0 This helps you to find out quickly if something is broken. It does not help you if things are broken on runtime, that's why we started to develop our Embedded Linux Test Framework (ELTF) which is an object oriented remote test system with tests being written in Python and being scripted with XML; you can connect a real hardware to a test box and play send/expect games with all the interfaces. That's IMHO the next step to automatically run test suites on real hardware. Robert -- Dipl.-Ing. Robert Schwebel | http://www.pengutronix.de Pengutronix - Linux Solutions for Science and Industry Handelsregister: Amtsgericht Hildesheim, HRA 2686 Hannoversche Str. 2, 31134 Hildesheim, Germany Phone: +49-5121-206917-0 | Fax: +49-5121-206917-9 From jhml at gmx.net Sat Mar 19 13:47:59 2005 From: jhml at gmx.net (Johann Hanne) Date: Sat, 19 Mar 2005 19:47:59 +0100 Subject: [uClinux-dev] Illegal instruction with setjmp() on ARM Message-ID: <200503191948.00559.jhml@gmx.net> Hi, I'm currently trying to build a shell for an embedded ARM box. The two alternatives are "sh" from uClinux-dist und "msh" from busybox. It seems to me that "msh" is based on "sh" (guessed from a quick look at the source code), so it isn't surprising that I have the same problem with both: # ./sh Illegal instruction # I've tracked down the problem by putting debug printf's into the sources: The "Illegal instruction" error is caused by the first setjmp() call in sh1.c, line 243: setjmp(m1); /* Bruce Evans' fix */ The same call exists in busybox, msh.c, line 1107: setjmp(failpt = m1); /* Bruce Evans' fix */ Google gave me two problems with setjmp() on ARM: - http://www.uclinux.org/pub/uClinux/archive/7205.html This is related to a bug in elf2flt, the fix has been applied in March 2002 by Greg Ungerer - http://www.uclibc.org/lists/uclibc/2003-February/005726.html This is related to a bug in uClibc, the fix has been applied in February 2003 by David McCullough The problem is that I'm using the gcc 2.95.3 toolchain from http://www.uclinux.org/pub/uClinux/arm-elf-tools/tools-20030314/ which has both fixed included. So there seems to be some other bug, which I'm unable to find... The board has an Arm940sid based CPU (Conexant) CX82100. Help, please? Cheers, Johann From rgetz at blackfin.uclinux.org Sat Mar 19 22:53:31 2005 From: rgetz at blackfin.uclinux.org (Robin Getz) Date: Sat, 19 Mar 2005 22:53:31 -0500 Subject: [uClinux-dev] renaming uClinux-dist Message-ID: <6.1.1.1.0.20050319222019.01e44e50@ptg1.spd.analog.com> Yasushi SHOJI wrote: >One of the most important things is IMHO testing. I could not agree more. There is alot of difference between a linux prompt, and having a working system. >We have compile-tests I would be interested to understand what test suites you are running? On blackfin.uclinux.org we are doing: - the built-in gcc dejagnu tests (http://gcc.gnu.org/install/test.html) - on the simulator - on the hardware, running on a compiled uClinux kernel - plum-hall (http://www.plumhall.com/stec.html) - on the simulator The testing we are doing runs 24 hours a day, 7 days a week, building on multiple distributions - and if any differences are see either between last runs, or distributions, it emails our development team (soon to be setting up tinderbox for other to look at). We are currently pulling from our CVS, but soon to be pulling from the FSF, to make sure that other people - working on other ports at the FSF - don't cause any issues/problems for our work. We attempting to do the same with U-boot (with a bunch of custom build expect scripts), and kernel (with parts of LTP and custom scripts). I would be very interested in your Embedded Linux Test Framework (ELTF) that you describe. Are there any more details on that? -Robin From leelawrenceyh at yahoo.com.cn Sun Mar 20 00:34:35 2005 From: leelawrenceyh at yahoo.com.cn (lawrence lee) Date: Sun, 20 Mar 2005 13:34:35 +0800 (CST) Subject: [uClinux-dev] Error when compile Perl on Uclinux Message-ID: <20050320053435.19994.qmail@web15010.mail.cnb.yahoo.com> Hi, I plan to compile the Perl5.8.6 on uclinux using toolchain. But I got compile error like: /usr/bin/ar cr ../../lib/auto /Dynaloader/Dynaloader.a Dynaloader.o&&... /usr/bin/ar: ../../lib/auto/Dynaloadr/Dynaloader.a : No such file or diretory ../../lib/auto/Dynaloadr/ is a valid directory. Can somebody help me? --------------------------------- Do You Yahoo!? 150??MP3???????????? ??????????????????? 1G??1000??????????? -------------- next part -------------- An HTML attachment was scrubbed... URL: From ulf.soderberg at gmail.com Sun Mar 20 05:46:55 2005 From: ulf.soderberg at gmail.com (=?ISO-8859-1?Q?Ulf_S=F6derberg?=) Date: Sun, 20 Mar 2005 11:46:55 +0100 Subject: [uClinux-dev] USB Gadget Device support Message-ID: <423D549F.8050200@gmail.com> Hi, I was wondering if anyone has had any success in using the USB Gadget drivers on a Intel IXP4xx target running the 2.6.11 kernel? from the driver sources (pxa2xx_udh.c etc.) it would seem that the 2.6 version of the driver is a bit older than the 2.4 version. After a bit of Googling on the subject, I found evidence that someone really had it working with the 2.4 kernel, but I haven't found any comments on USB Gadget + IXP4xx + Linux 2.6. I'm using the Snapgear distribution. -- Ulf S?derberg http://osgifun.blogspot.com From basicmark at yahoo.com Sun Mar 20 06:32:45 2005 From: basicmark at yahoo.com (Mark Underwood) Date: Sun, 20 Mar 2005 11:32:45 +0000 (GMT) Subject: [uClinux-dev] Searching for Evaluation Board In-Reply-To: 6667 Message-ID: <20050320113245.1746.qmail@web30307.mail.mud.yahoo.com> Hi Christoph I now have a Helio, ?14.51 from ebay! I have already started to take it apart and found that the flash and RAM is on separate board that is easy accessible via a removable back cover. This means that the CPU bus is available via that connector. My hope is that there are more address lines and CS lines then are currently being used so I can create a replacement board with more RAM, flash and maybe Ethernet & FPGA. I have download he current version of Linux and root filesystem that currently runs on it with the hope of a reasonably simple port to the 2.6.x kernel. The Helio also has two connectors on the bottom, one serial port and one unknown. Once I have a working 2.6.x kernel I plan to add removable mass-storage in the form of a SD/MMC SPI interface (if I can access one of the SPI ports). Best Regards, Mark --- Christoph Holst wrote: > Hi Mark > > Thanks for the links. I guess my budget likes PDAs > much more than evaluation > boards. The helio definitely seems to be a nice toy > for my embLinux plans. I > will look around for some used ones. > > Good luck for your search! > > Christoph > > -----Original Message----- > From: Mark Underwood [mailto:basicmark at yahoo.com] > Sent: Dienstag, 15. M?rz 2005 16:23 > To: Christoph Holst > Subject: Re: [uClinux-dev] Searching for Evaluation > Board > > Hi Chris, > > I am also looking for a cheap Evaluation board, but > haven?t found much in my price range with the > features > I need. The best resource for any cash strapped > hacker > seems to be consumer electronics. You can buy a new > Palm PDA for less then ?40 or if you want to add you > own hardware then maybe a handspring, again around > ?40 > from ebay. If you need more processing power and > have > a bigger budget them how about a Sharp Zaurus or > IPAQ. > I?m currently leaning towards a Vtech Helio. > http://www.linuxdevices.com/articles/AT6258429117.html > It has a 75MHz MIPS and has linux 2.4.0 running on > it > and PicoGui. It has a RAM module which I?m hoping to > use as an expansion port and I have seen on ebay for > ?10! > If you more of a hardware developer then you might > want to look at: > http://www.olimex.com/dev/ > > Best Regards, > > Mark > > --- Christoph Holst wrote: > > Hi there! > > > > > > > > I'm an embedded Linux newbie and would like to > start > > playing with a cheap > > Evaluation Board, preferably with ARM architecture > > and onboard LCD and Audio > > drivers. Is there any good company selling those > > boards in Europe? > > > > > > > > Regards, > > > > Chris > > > > > > > > > _______________________________________________ > > uClinux-dev mailing list > > uClinux-dev at uClinux.org > > > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > > This message was resent by uclinux-dev at uclinux.org > > Send instant messages to your online friends > http://uk.messenger.yahoo.com > > > Send instant messages to your online friends http://uk.messenger.yahoo.com From robert at schwebel.de Sun Mar 20 07:42:35 2005 From: robert at schwebel.de (Robert Schwebel) Date: Sun, 20 Mar 2005 13:42:35 +0100 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <6.1.1.1.0.20050319222019.01e44e50@ptg1.spd.analog.com> References: <6.1.1.1.0.20050319222019.01e44e50@ptg1.spd.analog.com> Message-ID: <20050320124235.GV11709@pengutronix.de> On Sat, Mar 19, 2005 at 10:53:31PM -0500, Robin Getz wrote: > Yasushi SHOJI wrote: > >One of the most important things is IMHO testing. I'm not Yasushi Shoji, but I wrote it anyway ;) > I could not agree more. There is alot of difference between a linux > prompt, and having a working system. Ack. Systems often work quite fast until the prompt, but show subtile bugs later. > I would be interested to understand what test suites you are running? We are in a very early stage wrt. testing, so we don't have any suites which are run regularly yet. The ELTF code was written in Jan/Feb, and it will only be our infrastructure for running real test suites. At least the basic LTP stuff would be interesting, plus test for known issues like softfloat on ARM. > On blackfin.uclinux.org we are doing: > - the built-in gcc dejagnu tests (http://gcc.gnu.org/install/test.html) > - on the simulator > - on the hardware, running on a compiled uClinux kernel > - plum-hall (http://www.plumhall.com/stec.html) > - on the simulator Is that one open source? It doesn't look so. > I would be very interested in your Embedded Linux Test Framework > (ELTF) that you describe. Are there any more details on that? Not yet; I'm in fact not very decided yet how to deal with that code. We normally do all our work as Open Source, but as hardware vendors tend to taking what they get from the community without giving much back, I'm not sure if we shouldn't just leave that beast inhouse and use it to our advantage, probably sharing it with some well-known Open Source maintainers. No offensen intended in your direction - I have no experience yet with Analog, but we maintain quite some ARM boards and two of the machine architectures in the official kernel and our experiences are not the best when it comes to vendors being interested in things that are more than "It just works to the commandline, so Linux must be supported". If you are interested, please contact me privately. Robert -- Dipl.-Ing. Robert Schwebel | http://www.pengutronix.de Pengutronix - Linux Solutions for Science and Industry Handelsregister: Amtsgericht Hildesheim, HRA 2686 Hannoversche Str. 2, 31134 Hildesheim, Germany Phone: +49-5121-206917-0 | Fax: +49-5121-206917-9 From robert at schwebel.de Sun Mar 20 07:45:24 2005 From: robert at schwebel.de (Robert Schwebel) Date: Sun, 20 Mar 2005 13:45:24 +0100 Subject: [uClinux-dev] Error when compile Perl on Uclinux In-Reply-To: <20050320053435.19994.qmail@web15010.mail.cnb.yahoo.com> References: <20050320053435.19994.qmail@web15010.mail.cnb.yahoo.com> Message-ID: <20050320124524.GW11709@pengutronix.de> On Sun, Mar 20, 2005 at 01:34:35PM +0800, lawrence lee wrote: > I plan to compile the Perl5.8.6 on uclinux using toolchain. But I got compile > error like: > /usr/bin/ar cr ../../lib/auto /Dynaloader/Dynaloader.a Dynaloader.o&&... > /usr/bin/ar: ../../lib/auto/Dynaloadr/Dynaloader.a : No such file or diretory > ../../lib/auto/Dynaloadr/ is a valid directory. Can somebody help me? I doubt that it will ever be possible to cleanly cross compile Perl, at least not in this universe. After some flamewar with the Parrot maintainers I even doubt that it will be possible with Perl 6, because theit policy allows to build that stuff natively on a 20 years old VAX, but not cross on most of today's embedded hardware. If you search something that cross compiles cleanly and additionally offers a well structured object oriented paradigm, use Python. *running away from the upcoming flamewar* Robert -- Dipl.-Ing. Robert Schwebel | http://www.pengutronix.de Pengutronix - Linux Solutions for Science and Industry Handelsregister: Amtsgericht Hildesheim, HRA 2686 Hannoversche Str. 2, 31134 Hildesheim, Germany Phone: +49-5121-206917-0 | Fax: +49-5121-206917-9 From robert at schwebel.de Sun Mar 20 07:47:46 2005 From: robert at schwebel.de (Robert Schwebel) Date: Sun, 20 Mar 2005 13:47:46 +0100 Subject: [uClinux-dev] USB Gadget Device support In-Reply-To: <423D549F.8050200@gmail.com> References: <423D549F.8050200@gmail.com> Message-ID: <20050320124746.GX11709@pengutronix.de> On Sun, Mar 20, 2005 at 11:46:55AM +0100, Ulf S?derberg wrote: > I was wondering if anyone has had any success in using the USB Gadget > drivers on a Intel IXP4xx target running the 2.6.11 kernel? From the > driver sources (pxa2xx_udh.c etc.) it would seem that the 2.6 version > of the driver is a bit older than the 2.4 version. After a bit of > Googling on the subject, I found evidence that someone really had it > working with the 2.4 kernel, but I haven't found any comments on USB > Gadget + IXP4xx + Linux 2.6. We have been heavily involved in the development of the PXA gadget driver, and it worked back in the 2.6.6 days. I havn't used it lately, but as David Brownell has just posted some positive test results on linux-usb-dev, it should work, at least in vanilla kernels. Robert -- Dipl.-Ing. Robert Schwebel | http://www.pengutronix.de Pengutronix - Linux Solutions for Science and Industry Handelsregister: Amtsgericht Hildesheim, HRA 2686 Hannoversche Str. 2, 31134 Hildesheim, Germany Phone: +49-5121-206917-0 | Fax: +49-5121-206917-9 From jhml at gmx.net Sun Mar 20 10:26:18 2005 From: jhml at gmx.net (Johann Hanne) Date: Sun, 20 Mar 2005 16:26:18 +0100 Subject: [uClinux-dev] [patch] 1/4: add CnxtBsp_Mac_MutEx_Chk() function required by Conexant Emac driver Message-ID: <200503201626.23119.jhml@gmx.net> This function is required for the Conexant Emac driver. The kernel will not compile without it when the Emac driver is enabled. -------------- next part -------------- A non-text attachment was scrubbed... Name: add_mac_mutex_chk.diff Type: text/x-diff Size: 1095 bytes Desc: not available URL: From jhml at gmx.net Sun Mar 20 10:43:46 2005 From: jhml at gmx.net (Johann Hanne) Date: Sun, 20 Mar 2005 16:43:46 +0100 Subject: [uClinux-dev] [patch] 2/4: disable machine specific linker script Message-ID: <200503201643.46293.jhml@gmx.net> This patch disables the cx821xx specific linker script in favor of the generic armnommu linker script given in the Makefile which works fine for me. This is done the same way for the other armnommu machines, so it should be ok. -------------- next part -------------- A non-text attachment was scrubbed... Name: linkerscript.diff Type: text/x-diff Size: 431 bytes Desc: not available URL: From jhml at gmx.net Sun Mar 20 10:44:14 2005 From: jhml at gmx.net (Johann Hanne) Date: Sun, 20 Mar 2005 16:44:14 +0100 Subject: [uClinux-dev] [patch] 3/4: Enable Conexant Emac TX DMA for Tiburon board Message-ID: <200503201644.14959.jhml@gmx.net> This patch enables Conexant Emac TX DMA for the Tiburon board. It's also enabled in the original Tiburon firmware, so it should be ok. -------------- next part -------------- A non-text attachment was scrubbed... Name: cnxt_emac_enable_tx_dma.diff Type: text/x-diff Size: 573 bytes Desc: not available URL: From jhml at gmx.net Sun Mar 20 10:44:45 2005 From: jhml at gmx.net (Johann Hanne) Date: Sun, 20 Mar 2005 16:44:45 +0100 Subject: [uClinux-dev] [patch] 4/4: use static MAC addresses for Tiburon board Message-ID: <200503201644.45433.jhml@gmx.net> This patch defines static MAC addresses for the ethernet ports of the Tiburon board rather than generating random ones. The original firmware does the same thing, and I think we should behave the same way for this board. The proper way would be to give each board a different MAC address, so you can use more than one of these boards in the same ethernet segment. However, this is a board design issue, and the engineers seem to have ignored the issue. And IMHO assigning the same MAC address on each boot is still better than using random MAC addresses. -------------- next part -------------- A non-text attachment was scrubbed... Name: fixed_mac.diff Type: text/x-diff Size: 478 bytes Desc: not available URL: From jhml at gmx.net Sun Mar 20 10:57:58 2005 From: jhml at gmx.net (Johann Hanne) Date: Sun, 20 Mar 2005 16:57:58 +0100 Subject: [uClinux-dev] Information on the patch series Message-ID: <200503201657.59010.jhml@gmx.net> Hi, I'm still working on including support for the Actiontec Dual PC Modem in the official uclinux 2.4 tree, for more information see http://bettina-attack.de/jonny/view.php/projects/actiontec_dpcm_stuff/ and http://bettina-attack.de/jonny/view.php/projects/actiontec_dpcm_stuff/kernel_patches/ I've checked out the CVS tree today and noticed that only 4 little bits are missing. After applying them the kernel indeed works for the DPCM: --- jonny at dh3mb jonny $ telnet actionhack Trying 44.130.186.69... Connected to actionhack. Escape character is '^]'. uClinux login: root Password: BusyBox v1.00 (2005.03.04-16:45+0000) Built-in shell (msh) Enter 'help' for a list of built-in commands. # free total used free shared buffers Mem: 6004 1112 4892 0 20 Swap: 0 0 0 Total: 6004 1112 4892 # cat /proc/version Linux version 2.4.29-uc0 (jonny at dh3mb) (gcc version 2.95.4 20010319 (prerelease)) #1 Sun Mar 20 14:12:59 CET 2005 # cat /proc/cpuinfo Processor : ARM/CNXT Arm940sid(wb) rev 2 (v4l) BogoMIPS : 83.76 Hardware : CNXT CX821XX Revision : 0000 Serial : 0000000000000000 # --- The 4 patches are absolutetly non-intrusive, so I hope that someone can apply them. Would be a dream to see that 2.4.29-uc1 supports the DPCM out of the box! Cheers, Johann From pauli at bne.snapgear.com Sun Mar 20 18:40:56 2005 From: pauli at bne.snapgear.com (Pauli) Date: Mon, 21 Mar 2005 09:40:56 +1000 Subject: [uClinux-dev] nommu for m68k shared library currently supported In-Reply-To: Your message of "Fri, 18 Mar 2005 14:04:30 +0200." <87165196515.20050318140430@analog.ro> References: <87165196515.20050318140430@analog.ro> Message-ID: <20050320234057.0CBC858BB4@pauli2.sw.moreton.com.au> Hi, Rujinschi Remus wrote: > I would like to know if there's shared library currently supported for > M68K family also if there's any console/uart ircda jffs2 drivers > avaliable. There is shared library support for the M68k family. It isn't anything like shared libraries in the MMU world so don't start by making assumptions. lib/libc/ contains details on how to build stuff shared. Plus there is an FAQ on the web site which spells things out pretty well. Pauli -- Dr Paul Dale, Software Grunt, SnapGear - A CyberGuard Company Leaders in embedded Linux security http://www.SnapGear.com/ From gerg at snapgear.com Sun Mar 20 18:57:16 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 21 Mar 2005 09:57:16 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050319002457.GI11709@pengutronix.de> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <20050315081337.0A198633A@mail2.atmark-techno.com> <20050319002457.GI11709@pengutronix.de> Message-ID: <423E0DDC.6080809@snapgear.com> Hi Robert, Robert Schwebel wrote: > On Tue, Mar 15, 2005 at 05:13:37PM +0900, Yasushi SHOJI wrote: >>BTW, I enjoyed a session at FOSDEM about Bitbake. > > You may also want to have a look at PTXdist: > > http://www.pengutronix.de/software/ptxdist_en.html > > It mostly follows the Kconfig+Makefiles way, but in contrast to the > uCLinux dist you don't have to download multi-ten megabyte images before > you can start. It catches all the necessary archives from the network > when they are needed. from time to time people mention the size of the uClinux-dist as being a problem. I just don't see it myself. The current dist runs close to 200MB. Sure that is large, doesn't take all that long to download over a xDSL link though. But to compare, linux-2.6.11 archive is 46MB... So the uClinux-dist is around 4 times larger (comparing gzip archives here). Or worse still if you want a Fedora Core 3 install then you are looking at 4 large downloads, 3 of which are over 600MB... Once I have the uClinux-dist file I can move it around easily on the LAN, put it on a CD, I don't need a connection to the network any more. It also makes it very easy for companies using uClinux-dist to forfill their GPL obligations. They can put the uClinux-dist (and their) changes onto a CD or their web site. Just keeping patches to other source is not good enough for this. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Sun Mar 20 19:07:04 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 21 Mar 2005 10:07:04 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050319003721.GJ11709@pengutronix.de> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> <20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> <423951E6.5040607@itee.uq.edu.au> <20050319003721.GJ11709@pengutronix.de> Message-ID: <423E1028.8010608@snapgear.com> Hi Robert, Robert Schwebel wrote: > On Thu, Mar 17, 2005 at 07:46:14PM +1000, John Williams wrote: >>This is a good discussion to be having - really I think the bigger >>question is "where is uclinux-dist going?" > > Some thoughts from the PTXdist competitor ;) When I started to search > for a build system some years ago I also had a look at the uCLinux build > system and in fact copied a lot of it's ideas into PTXdist. But there > are several things which havn't been acceptable to what I had in mind. > >>It's pretty agile. dist-ifying an existing app is getting easier and >>easier - witness some apps now that are based on generic GNU configure >>setups. > > How is the process of feeding uCLinux modifications back to upstream > solved today? Last time I looked I had the impression that it does more > follow a "we port it when we need it" approach, with the consequence > that it worked then for people who use the uCLinux dist. Is there some > kind of a "we feed patches upstream" policy in uCLinux-dist? The uClinux-dist just tries to be a collection of sources, it doesn't itself try to manage the patch process. So the job of sending patches for modified sources upstream is the job of whoever ported the code. For one thing there is way too many packages in the dist that the maintainer of the dist could actually do it. We are up to something 150 packages, there is no way 1 or 2 people can co-ordinate patches for each with their rightful maintainers. And really the uClinux-dist maintainer shouldn't have too, whoever ported it, and needs it, should do their duty and send their changes upstream :-) That is whet I do for my changes to the 2.6 kernel for example. The other lesser problem is that a number of the packages in uClinux-dist don't have a home. Or their home is in uClinux-dist. > We have a patch repository for PTXdist which is applied on the fly to > the original sources when they are being built. So there is a clear > separation between patches and upstream sources; patches are reviewed on > a regular base and feeded to the upstream maintainer. Not to trivialize the point, but it is not difficult to get the source for any package and diff it against what is in the uClinux-dist. There is your patch (admittedly it doesn't have any user context - why the changes where made for example). But how often do you really reveiew the patches? For any large number of out-standing packages this becomes quite a difficult thing to do. >>Some things I'd like to see in uClinux-dist: >> >>Real-time CVS tracking. The model of periodic releases places pressure >>on Dave and Greg (maybe! :) and is a bit inflexible. I see this a lot >>on the Microblaze platform where every single system can be different. >>The ability to create new vendors targets and immediately publish them >>through CVS would be very valuable. It's getting so that I'm also >>considering a custom uClinux-dist for microblaze. > > > Hmmm, if you have real customer projects you at least need some > stability points, and you need control over it. Having a RCS system is > cool (we have svn for PTXdist), but that doesn't mean that lots of > people should get write access. Building systems is a complex and > fragile thing and should be supervised by a maintainer who knows what he > does. How do you handle the fact that the internet is a little more dynamic though? Web sites come and go, source packages locations change or disappear over time? With all the sources in the uClinux-dist package it is a frozen moment in time. I can get the uClinux-dist (or as they where uClinux-coldfire) packages from 1999 and still build them completely as they are. Now as to dynamic CVS acces to the dist. That is a concious decision on my part not to do it. That is to reduce the amount of time I (and a small set of others) have to maintain the public CVS. My prefernce would still be not to have a public CVS of the dist at all :-) Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Sun Mar 20 19:14:40 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 21 Mar 2005 10:14:40 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050319112155.GP11709@pengutronix.de> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> <20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> <423951E6.5040607@itee.uq.edu.au> <20050319003721.GJ11709@pengutronix.de> <20050319074628.C3F226341@mail2.atmark-techno.com> <20050319112155.GP11709@pengutronix.de> Message-ID: <423E11F0.6080405@snapgear.com> Robert Schwebel wrote: > On Sat, Mar 19, 2005 at 04:46:28PM +0900, Yasushi SHOJI wrote: > >>>How is the process of feeding uCLinux modifications back to upstream >>>solved today? Last time I looked I had the impression that it does more >>>follow a "we port it when we need it" approach, with the consequence >>>that it worked then for people who use the uCLinux dist. Is there some >>>kind of a "we feed patches upstream" policy in uCLinux-dist? >> >>except kernels, I don't think we have it. We have it in the sense that the person who modifies the code needs to do their duty and send their changes to the appropriate upstream maintainer. Not to push this task onto someone else :-) > I think this is a very important topic. One of the uggliest problems > these communities have is lack of feedback. Especially commercial users > are really happy to get stuff from the community but if it works > "somehow", they don't care about The Right Solution (TM) for a problem > and they don't care about feeding back things to upstream. > > As a commecial Embedded Linux company, we have to do something to > fullfill our customers needs to have something that "just works". Being > active in the Linux community for >12 years now, we know quite well that > long term Open Source will only fulfill it's image of being better, more > stable and superior to closed source stuff when there is an active > community process with *feedback*. This is ofcourse true. But I am not sure I see that pushing the problem of sending patches upstream onto the maintainer of a "distribution" is the right way to solve the problem. That path leads to an over-worked and frustrated set of maintainers, trying to co-ordinate lots of things they don't care that much about themselves ;-) Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Sun Mar 20 19:26:13 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 21 Mar 2005 10:26:13 +1000 Subject: [uClinux-dev] USB Gadget Device support In-Reply-To: <423D549F.8050200@gmail.com> References: <423D549F.8050200@gmail.com> Message-ID: <423E14A5.7030306@snapgear.com> Hi Ulf, Ulf S?derberg wrote: > I was wondering if anyone has had any success in using the USB Gadget > drivers on a Intel IXP4xx target running the 2.6.11 kernel? from the > driver sources (pxa2xx_udh.c etc.) it would seem that the 2.6 version of > the driver is a bit older than the 2.4 version. After a bit of Googling > on the subject, I found evidence that someone really had it working with > the 2.4 kernel, but I haven't found any comments on USB Gadget + IXP4xx > + Linux 2.6. > > I'm using the Snapgear distribution. I had it working the 2.4 kernels, on the IXP425. But I haven't tried it on 2.6... Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From davidm at snapgear.com Sun Mar 20 20:04:53 2005 From: davidm at snapgear.com (David McCullough) Date: Mon, 21 Mar 2005 11:04:53 +1000 Subject: [uClinux-dev] Defining Variables in Header Files In-Reply-To: <20050318165822.31716.qmail@web51905.mail.yahoo.com> References: <20050318165822.31716.qmail@web51905.mail.yahoo.com> Message-ID: <20050321010453.GA3090@beast> Jivin Doug Kehn lays it down ... > -- Doug Kehn wrote: > > > I've been noticing that more and more source out > > there (e.g. ISC DHCP 3.0.2) is defining variables > > in header files. The header file is included in > > more than one source file which results in > > "multiple definition of `...'" compiler error in > > uClinux. However, the same source compiles fine > > on x86. Has anyone else noticed this? > > The reason for the compile errors with "multiple > definition ..." in uClinux is due to the -fno-common > compiler option. (BTW, I'm compiling for the IXP425 > with the arm-linux toolchain.) > > If I add this compiler option to the Makefile(s) when > building the same source on my host system (x86), I > get the same "multiple definition ..." compiler > errors. > > I prefer the uClinux way (i.e. use the -fno-common > compiler option). Maybe I'm old-school here but I > prefer to define variables in C files and declare them > (if need be) with extern in H files. Our opinion exactly (most likely old-school :-). Basically some of the arm tool chains had/have a linker bug with common section allocations that occasionally caused application crashes and generally "weird" behaviour. After looking at it we decided that coding in a way that relies on common section allocation is just a really bad idea and we decided to make it illegal within the dist. All the packages in the dist were cleaned up at that time to deal with the change. There were quite a few, but not all that many. I actually think a few potential bugs were found in the cleanup as well :-) > This little nuance makes porting annoying. It does, but if we do it and send the changes to the package maintainers then maybe it will stop. It would be nice to hear what others think, but for me, I can't believe it even works, I had always assumed your would get "multiple defined" errors :-) Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From gerg at snapgear.com Sun Mar 20 20:11:47 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 21 Mar 2005 11:11:47 +1000 Subject: [uClinux-dev] [patch] 1/4: add CnxtBsp_Mac_MutEx_Chk() function required by Conexant Emac driver In-Reply-To: <200503201626.23119.jhml@gmx.net> References: <200503201626.23119.jhml@gmx.net> Message-ID: <423E1F53.6050703@snapgear.com> Hi Johann, Johann Hanne wrote: > This function is required for the Conexant Emac driver. The kernel will not > compile without it when the Emac driver is enabled. Applied. I take it your current set of patches replace those you emailed here a couple of weeks back? Thanks Greg > ------------------------------------------------------------------------ > > --- linux-2.4.27-uc1/arch/armnommu/mach-cx821xx/cnxtbsp.c 2005-02-12 13:26:11.000000000 +0100 > +++ linux-2.4.27-uc1.patched/arch/armnommu/mach-cx821xx/cnxtbsp.c 2005-02-27 14:18:31.000000000 +0100 > @@ -165,6 +165,30 @@ > *mac_addr_array = CnxtBsp_Get_Random_Byte(); > } > > +/****************************************************************************** > +| > +| Function: CnxtBsp_Mac_MutEx_Chk > +| > +| Description: Used to determine whether or not this device requires > +| mutually exclusive operation on the emac transmitters > +| > +| Returns: NONE > +| > +*******************************************************************************/ > + > +BOOLEAN CnxtBsp_Mac_MutEx_Chk( void ) > +{ > + switch( _sysGetDeviceType() ) > + { > + case 0: > + return TRUE; > + break; > + default: > + return FALSE; > + break; > + } > +} > + > #if CONFIG_CNXT_ADSL || CONFIG_CNXT_ADSL_MODULE > > /**************************************************************************** > > > ------------------------------------------------------------------------ > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uclinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Sun Mar 20 20:12:06 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 21 Mar 2005 11:12:06 +1000 Subject: [uClinux-dev] [patch] 2/4: disable machine specific linker script In-Reply-To: <200503201643.46293.jhml@gmx.net> References: <200503201643.46293.jhml@gmx.net> Message-ID: <423E1F66.1000202@snapgear.com> Hi Johann, Johann Hanne wrote: > This patch disables the cx821xx specific linker script in favor of the generic > armnommu linker script given in the Makefile which works fine for me. This is > done the same way for the other armnommu machines, so it should be ok. Applied. Thanks Greg > ------------------------------------------------------------------------ > > --- uc-origs/uClinux-2.4.27-uc1/arch/armnommu/Makefile 2004-12-12 21:48:45.000000000 +0100 > +++ uClinux-2.4.27-uc1/arch/armnommu/Makefile 2005-01-29 15:17:44.000000000 +0100 > @@ -148,7 +148,7 @@ > MACHINE = cx821xx > TEXTADDR = 0x800000 > CFLAGS += -DNO_MM -DMAGIC_ROM_PTR > -LINKFLAGS += -T arch/armnommu/mach-$(MACHINE)/ramlinux.lds > +#LINKFLAGS += -T arch/armnommu/mach-$(MACHINE)/ramlinux.lds > endif > > ifeq ($(CONFIG_ARCH_SPIPE),y) > > > ------------------------------------------------------------------------ > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uclinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Sun Mar 20 20:12:27 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 21 Mar 2005 11:12:27 +1000 Subject: [uClinux-dev] [patch] 3/4: Enable Conexant Emac TX DMA for Tiburon board In-Reply-To: <200503201644.14959.jhml@gmx.net> References: <200503201644.14959.jhml@gmx.net> Message-ID: <423E1F7B.7020804@snapgear.com> Ho Johann, Johann Hanne wrote: > This patch enables Conexant Emac TX DMA for the Tiburon board. It's also > enabled in the original Tiburon firmware, so it should be ok. Applied. Thanks Greg > ------------------------------------------------------------------------ > > diff -uNr linux.8/drivers/net/cnxt_emac/cnxtEmac.c linux/drivers/net/cnxt_emac/cnxtEmac.c > --- linux.8/drivers/net/cnxt_emac/cnxtEmac.c 2003-08-14 00:49:42.000000000 +0200 > +++ linux/drivers/net/cnxt_emac/cnxtEmac.c 2005-02-19 17:56:58.000000000 +0100 > @@ -83,8 +83,13 @@ > #define RX_SKB_DEBUG 0 > #define SETUP_FRAME_VERIFY 0 > #define RX_TASKLET 1 > +#ifdef CONFIG_BD_TIBURON > +#define TX_TASKLET 1 > +#define TX_DMA_INT 1 > +#else > #define TX_TASKLET 0 > #define TX_DMA_INT 0 > +#endif > #define RX_DMA_INT 1 > #define USE_TU_INTERRUPT 1 > #define USE_HASHTABLE 0 > > > ------------------------------------------------------------------------ > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uclinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Sun Mar 20 20:14:15 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 21 Mar 2005 11:14:15 +1000 Subject: [uClinux-dev] [patch] 4/4: use static MAC addresses for Tiburon board In-Reply-To: <200503201644.45433.jhml@gmx.net> References: <200503201644.45433.jhml@gmx.net> Message-ID: <423E1FE7.60101@snapgear.com> Hi Johann, Johann Hanne wrote: > This patch defines static MAC addresses for the ethernet ports of the Tiburon > board rather than generating random ones. The original firmware does the same > thing, and I think we should behave the same way for this board. > > The proper way would be to give each board a different MAC address, so you can > use more than one of these boards in the same ethernet segment. However, this > is a board design issue, and the engineers seem to have ignored the issue. > And IMHO assigning the same MAC address on each boot is still better than > using random MAC addresses. Random MAC address on each reboot - thats unusual ;-) Applied. Thanks Greg > ------------------------------------------------------------------------ > > --- linux-2.4.27-uc1/arch/armnommu/mach-cx821xx/cnxtbsp.c 2005-02-12 13:26:11.000000000 +0100 > +++ linux-2.4.27-uc1.patched/arch/armnommu/mach-cx821xx/cnxtbsp.c 2005-02-27 14:20:27.000000000 +0100 > @@ -162,7 +162,11 @@ > *mac_addr_array++ = 0xcd; > *mac_addr_array++ = 0x00; > *mac_addr_array++ = 0x02; > +#ifdef CONFIG_BD_TIBURON > + *mac_addr_array = 0xcc+emac_num; > +#else > *mac_addr_array = CnxtBsp_Get_Random_Byte(); > +#endif > } > > #if CONFIG_CNXT_ADSL || CONFIG_CNXT_ADSL_MODULE > > > ------------------------------------------------------------------------ > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uclinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From davidm at snapgear.com Sun Mar 20 20:15:45 2005 From: davidm at snapgear.com (David McCullough) Date: Mon, 21 Mar 2005 11:15:45 +1000 Subject: [uClinux-dev] Illegal instruction with setjmp() on ARM In-Reply-To: <200503191948.00559.jhml@gmx.net> References: <200503191948.00559.jhml@gmx.net> Message-ID: <20050321011545.GB3090@beast> Jivin Johann Hanne lays it down ... > Hi, > > I'm currently trying to build a shell for an embedded ARM box. The two > alternatives are "sh" from uClinux-dist und "msh" from busybox. It seems to > me that "msh" is based on "sh" (guessed from a quick look at the source > code), so it isn't surprising that I have the same problem with both: > > # ./sh > Illegal instruction > # > > I've tracked down the problem by putting debug printf's into the sources: The > "Illegal instruction" error is caused by the first setjmp() call in sh1.c, > line 243: > setjmp(m1); /* Bruce Evans' fix */ > The same call exists in busybox, msh.c, line 1107: > setjmp(failpt = m1); /* Bruce Evans' fix */ > > Google gave me two problems with setjmp() on ARM: > - http://www.uclinux.org/pub/uClinux/archive/7205.html > This is related to a bug in elf2flt, the fix has been applied in March 2002 > by Greg Ungerer > - http://www.uclibc.org/lists/uclibc/2003-February/005726.html > This is related to a bug in uClibc, the fix has been applied in February > 2003 by David McCullough > > The problem is that I'm using the gcc 2.95.3 toolchain from > http://www.uclinux.org/pub/uClinux/arm-elf-tools/tools-20030314/ which has > both fixed included. So there seems to be some other bug, which I'm unable to > find... > > The board has an Arm940sid based CPU (Conexant) CX82100. Sounds like you have a uCLibc without the correct floating point support in setjmp and friends. You should be able to find something on this in the archives. What sources (dist) are you using. Alternatively, just double check your .S files for setjmp/longjmp and sigsetjmp/siglongjmp and make sure they either always do the FP stuff or never do it (use #if 1/#if 0 instead of #if HAS_FPU or whatever :-), Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From davidm at snapgear.com Sun Mar 20 20:28:55 2005 From: davidm at snapgear.com (David McCullough) Date: Mon, 21 Mar 2005 11:28:55 +1000 Subject: [uClinux-dev] Information on the patch series In-Reply-To: <200503201657.59010.jhml@gmx.net> References: <200503201657.59010.jhml@gmx.net> Message-ID: <20050321012855.GA3903@beast> Jivin Johann Hanne lays it down ... > Hi, > > I'm still working on including support for the Actiontec Dual PC Modem in the > official uclinux 2.4 tree, for more information see > http://bettina-attack.de/jonny/view.php/projects/actiontec_dpcm_stuff/ > and > http://bettina-attack.de/jonny/view.php/projects/actiontec_dpcm_stuff/kernel_patches/ > > I've checked out the CVS tree today and noticed that only 4 little bits are > missing. After applying them the kernel indeed works for the DPCM: > --- > jonny at dh3mb jonny $ telnet actionhack > Trying 44.130.186.69... > Connected to actionhack. > Escape character is '^]'. > uClinux login: root > Password: > > > BusyBox v1.00 (2005.03.04-16:45+0000) Built-in shell (msh) > Enter 'help' for a list of built-in commands. > > # free > total used free shared buffers > Mem: 6004 1112 4892 0 20 > Swap: 0 0 0 > Total: 6004 1112 4892 > # cat /proc/version > Linux version 2.4.29-uc0 (jonny at dh3mb) (gcc version 2.95.4 20010319 > (prerelease)) #1 Sun Mar 20 14:12:59 CET 2005 > # cat /proc/cpuinfo > Processor : ARM/CNXT Arm940sid(wb) rev 2 (v4l) > BogoMIPS : 83.76 > Hardware : CNXT CX821XX > Revision : 0000 > Serial : 0000000000000000 > # > --- > > The 4 patches are absolutetly non-intrusive, so I hope that someone can apply > them. Would be a dream to see that 2.4.29-uc1 supports the DPCM out of the > box! Great :-) Tell us which 4 patches and we'll see if we can your dreams come true :-) Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From jwilliams at itee.uq.edu.au Sun Mar 20 21:27:38 2005 From: jwilliams at itee.uq.edu.au (John Williams) Date: Mon, 21 Mar 2005 12:27:38 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <423E0DDC.6080809@snapgear.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <20050315081337.0A198633A@mail2.atmark-techno.com> <20050319002457.GI11709@pengutronix.de> <423E0DDC.6080809@snapgear.com> Message-ID: <423E311A.3060208@itee.uq.edu.au> Hi Greg, Greg Ungerer wrote: > Hi Robert, > > Robert Schwebel wrote: > >> On Tue, Mar 15, 2005 at 05:13:37PM +0900, Yasushi SHOJI wrote: >> >>> BTW, I enjoyed a session at FOSDEM about Bitbake. >> >> >> You may also want to have a look at PTXdist: >> http://www.pengutronix.de/software/ptxdist_en.html >> >> It mostly follows the Kconfig+Makefiles way, but in contrast to the >> uCLinux dist you don't have to download multi-ten megabyte images before >> you can start. It catches all the necessary archives from the network >> when they are needed. > > > from time to time people mention the size of the uClinux-dist > as being a problem. I just don't see it myself. > > The current dist runs close to 200MB. Sure that is large, doesn't > take all that long to download over a xDSL link though. A dist cut without glibc would reduce this more than a little. Uncompressed glibc is 116meg. Are there many (any?) platforms regularly using this now? It seems like uClibc is by far the dominant player in NOMMU/uClinux-dist these days. Perhaps something to consider in response to the download size argument. John From Vikrant_Basotra at Satyam.com Sun Mar 20 23:56:36 2005 From: Vikrant_Basotra at Satyam.com (Vikrant_Basotra) Date: Mon, 21 Mar 2005 10:26:36 +0530 Subject: [uClinux-dev] Linux Porting problem on mpc8248 based board Message-ID: <3D9FDCA910DD4445896A0BE5ECF009D40306C850@bla.satyam.com> Hi All, I am trying to port Linux on the mpc8248 based board. Configurations of the board are as follows : Memory Region Size Address Range Port SizeSDRAM 32MByte 0000_0000 -- 01FF_FFFF 64 bit PPC Internal Memory 384KByte 0F00_0000 -- 0F05_FFFF PCI 1GByte 1000_0000 -- 4FFF_FFFF 32 bit Flash 32MByte FE00_0000 -- FFFF_FFFF 32 bit 1.SDRAM device size is 16MByte and 32 bit wide. We have used two SDRAMs for total size of 32MByte and 64bit wide. 2.Flash device size is 8MByte and 16 bit wide. We have used two Flashes for total size of 16MByte and 32bit wide Using uclibc as standard C library, and u-boot as boot loader Using the Linux kernel image of the EP8248 board, This is the out put which I am getting when I try to boot the board after porting the linux image on it.. U-Boot 1.0.0 (Mar 18 2005 - 18:36:05) MPC8260 Reset Status: External Soft, External Hard MPC8260 Clock Configuration - Bus-to-Core Mult 3.5x, VCO Div 2, 60x Bus Freq 30-100, Core Freq 100-400 - dfbrg 1, corecnf 0x1e, busdf 3, cpmdf 1, plldf 0, pllmf 5 - vco_out 396000000, scc_clk 99000000, brg_clk 24750000 - cpu_clk 346500000, cpm_clk 198000000, bus_clk 99000000 CPU: 8272 (HiP7 Rev 14, Mask A.0 1K50M) at 346.500 MHz Board: Embedded Planet EP8248, Revision 1.0 I2C: ready DRAM: 16 MB Top of RAM usable for U-Boot at: 01000000 Reserving 152k for U-Boot at: 00fd9000 Reserving 128k for malloc() at: 00fb8e00 Reserving 76 Bytes for Board Info at: 00fb8db4 Reserving 72 Bytes for Global Data at: 00fb8d6c New Stack Pointer is: 00fb8d48 FLASH: 16 MB In: serial Out: serial Err: serial Net: FCC1 ETHERNET Hit any key to stop autoboot: 5  4  3  2  1  0 ## Booting image at fff20d00 ... Image Name: Linux-2.4.27-1.3.1-432 Image Type: PowerPC Linux Kernel Image (gzip compressed) Data Size: 708524 Bytes = 691.9 kB Load Address: 00000000 Entry Point: 00000000 Verifying Checksum ... OK Uncompressing Kernel Image ... OK ## Current stack ends at 0x00FB8A98 => set upper limit to 0x00800000 No initrd ## Transferring control to Linux (at address 00000000) ... And after this nothing happens.... The control is coming uptill to the function (*kernel) (kbd, initrd_start, initrd_end, cmd_start, cmd_end); defined in u-boot/common/cmd_bootm.c 1. The values of the arguments are : initrd_start = 0 initrd_end = 0 : cmd_start = 7ffc00 : cmd_end= 7ffc1e 2. The values in the kbd strcture and hdr strcture are given below. kbd->bi_memstart: 0 kbd->bi_memsize: 1000000 kbd->bi_flashstart: ff800000 kbd->bi_flashoffset: 0 kbd->bi_sramstart: 0 kbd->bi_sramsize: 0 kbd->bi_immr_base: f0000000 kbd->bi_bootflags: 1 kbd->bi_ip_addr: c0a801de kbd->bi_enetaddr: 7ffbd8 kbd->bi_ethspeed: 2bd6 kbd->bi_intfreq: 15a kbd->bi_busfreq: 63 kbd->bi_cpmfreq: c6 kbd->bi_brgfreq: 18 kbd->bi_sccfreq: 63 kbd->bi_vco: 18c kbd->bi_baudrate: 1c200 hdr->ih_magic: 27051956 hdr->ih_hcrc: 0 hdr->ih_time: 423acbd3 hdr->ih_size: acfac hdr->ih_load: 0 hdr->ih_ep: 0 hdr->ih_dcrc: 6466e40d hdr->ih_os: 5 hdr->ih_arch: 7 hdr->ih_type: 2 hdr->ih_comp: 1 hdr->ih_name: Linux-2.4.27-1.3.1-432 2. In file arc/ppc/platforms/ep8248.h #define CPM_MAP_ADDR ((uint)0x0f000000) CPM_MAP_ADDR is given the value of CFG_IMMR which is defined in the u-boot. Can anybody tell me that where I am doing wrong... what I am missing to be done. Waiting for the help from the group. Regards, Vikrant ************************************************************************** This email (including any attachments) is intended for the sole use of the intended recipient/s and may contain material that is CONFIDENTIAL AND PRIVATE COMPANY INFORMATION. Any review or reliance by others or copying or distribution or forwarding of any or all of the contents in this message is STRICTLY PROHIBITED. If you are not the intended recipient, please contact the sender by email and delete all copies; your cooperation in this regard is appreciated. ************************************************************************** From gerg at snapgear.com Mon Mar 21 00:19:17 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 21 Mar 2005 15:19:17 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <423E311A.3060208@itee.uq.edu.au> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <20050315081337.0A198633A@mail2.atmark-techno.com> <20050319002457.GI11709@pengutronix.de> <423E0DDC.6080809@snapgear.com> <423E311A.3060208@itee.uq.edu.au> Message-ID: <423E5955.8010104@snapgear.com> Hi John, John Williams wrote: > Greg Ungerer wrote: >> Robert Schwebel wrote: >>> On Tue, Mar 15, 2005 at 05:13:37PM +0900, Yasushi SHOJI wrote: >>> It mostly follows the Kconfig+Makefiles way, but in contrast to the >>> uCLinux dist you don't have to download multi-ten megabyte images before >>> you can start. It catches all the necessary archives from the network >>> when they are needed. >> >> from time to time people mention the size of the uClinux-dist >> as being a problem. I just don't see it myself. >> >> The current dist runs close to 200MB. Sure that is large, doesn't >> take all that long to download over a xDSL link though. > > A dist cut without glibc would reduce this more than a little. > Uncompressed glibc is 116meg. Are there many (any?) platforms > regularly using this now? It seems like uClibc is by far the dominant > player in NOMMU/uClinux-dist these days. No doubt uClibc is domoniant (no glibc choice for non-MMU). Show of hands... who here uses glibc on VM platforms in the uClinux-dist? > Perhaps something to consider in response to the download size argument. Even if I take it out, it is only a short term fix. The dist will keep growing in size, and within a short time it will be larger than it is now - even without glibc. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Mon Mar 21 00:26:41 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Mon, 21 Mar 2005 15:26:41 +1000 Subject: [uClinux-dev] [PATCH]uClinux patch for Philips LPC22xx In-Reply-To: References: Message-ID: <423E5B11.8080202@snapgear.com> Hi Lucy, lucy.wang at philips.com wrote: > Attached is the uClinux patch for Philips LPC22xx, including the > vendors/Philips/LPC22xx directory and a new config.in file. > A readme file is also here. > Together with the linux-2.6.5 patch, you can begin the first trial of > uClinux+LPC22xx development. > Note that the device drivers, e.g. Ethernet driver(although I've > finished the rtl8019as driver), are not here. You patch looks to include all of Hyok's patch and then your changes on top. Would that be right? Can you generate a patch with just the changes from Hyok's armnommu support and your changes? My working 2.6 tree has Hyok's armnommu changes already in it. Applying you patch will mean a bit of cleanup work on my part. If you just have a patch for your changes that will make it a lot easier to apply right now. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From jwilliams at itee.uq.edu.au Mon Mar 21 00:41:16 2005 From: jwilliams at itee.uq.edu.au (John Williams) Date: Mon, 21 Mar 2005 15:41:16 +1000 Subject: [uClinux-dev] Defining Variables in Header Files In-Reply-To: <20050321010453.GA3090@beast> References: <20050318165822.31716.qmail@web51905.mail.yahoo.com> <20050321010453.GA3090@beast> Message-ID: <423E5E7C.5020902@itee.uq.edu.au> Hi David, David McCullough wrote: >>I prefer the uClinux way (i.e. use the -fno-common >>compiler option). Maybe I'm old-school here but I >>prefer to define variables in C files and declare them >>(if need be) with extern in H files. > > > It would be nice to hear what others think, but for me, I can't > believe it even works, I had always assumed your would get "multiple > defined" errors :-) Agree 100%. Current microblaze gcc doesn't even support -fno-common, so there's no way to build broken code like that. I think I sent in a few tidfy-ups for exacty that reason. John From eauth at softsys.co.at Mon Mar 21 02:11:18 2005 From: eauth at softsys.co.at (Erwin Authried) Date: Mon, 21 Mar 2005 08:11:18 +0100 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <423E1028.8010608@snapgear.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> <20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> <423951E6.5040607@itee.uq.edu.au> <20050319003721.GJ11709@pengutronix.de> <423E1028.8010608@snapgear.com> Message-ID: <1111389078.11799.21.camel@kabel> On Mon, 2005-03-21 at 01:07, Greg Ungerer wrote: ... > Not to trivialize the point, but it is not difficult to get the > source for any package and diff it against what is in the uClinux-dist. > There is your patch (admittedly it doesn't have any user context - > why the changes where made for example). > > But how often do you really reveiew the patches? > For any large number of out-standing packages this becomes quite > a difficult thing to do. > Hello Greg, what I'm missing in uClinux-dist is an easy way to track the changes that have been made to a source package. That's important when you want to switch to a later version of some package. At least for some of the packages, it is quite difficult to fetch the original source, either because there isn't any version information or the package isn't available on the internet anymore. > > > > > Hmmm, if you have real customer projects you at least need some > > stability points, and you need control over it. Having a RCS system is > > cool (we have svn for PTXdist), but that doesn't mean that lots of > > people should get write access. Building systems is a complex and > > fragile thing and should be supervised by a maintainer who knows what he > > does. > > How do you handle the fact that the internet is a little more dynamic > though? Web sites come and go, source packages locations change or > disappear over time? I have seen this to be a problem too for my Midori uClinux distirbution. This can be handled with a "fallback" archive of original sources. When the package cannot be fetched from the original place, it is automatically fetched from this location. Regards, Erwin From kanhu2004 at rediffmail.com Mon Mar 21 02:14:14 2005 From: kanhu2004 at rediffmail.com (kanhu rauta) Date: 21 Mar 2005 07:14:14 -0000 Subject: [uClinux-dev] Remote Logging Problem Message-ID: <20050321071414.28297.qmail@webmail18.rediffmail.com> Hi all, I am using BusyBox v0.60.4 with uClinux. I have selected the klogd and syslogd options of busybox on compilation. When I load this image on my gateway,both process has started successfully. />ps > PID TTY Uid Size State Command > 1 root 0 S /sbin/init > 2 root 0 S [keventd] > 3 root 0 S [ksoftirqd_CPU0] > 4 root 0 S [kswapd] > 5 root 0 S [bdflush] > 6 root 0 S [kupdated] > 51 root 0 S /bin/dhcpd LAN > 77 root 0 S chois -gserver 54273 > 80 root 0 S /bin/dnsmasq -r /etc/resolv.dnsmasq > 84 ttyS0 root 0 S /bin/sh > 85 root 0 S /bin/inetd > 86 root 0 S /bin/syslogd -n > 87 root 0 S /bin/klogd -n > 90 ttyS0 root 0 R ps I get the kernel log messages in the file /var/log/messages. I want to forward this kernel log to a remote machine,Then i exucute the follwing command />/bin/syslogd -R 192.168.5.4:514 & />ps > PID TTY Uid Size State Command > 1 root 0 S /sbin/init > 2 root 0 S [keventd] > 3 root 0 S [ksoftirqd_CPU0] > 4 root 0 S [kswapd] > 5 root 0 S [bdflush] > 6 root 0 S [kupdated] > 51 root 0 S /bin/dhcpd LAN > 77 root 0 S chois -gserver 54273 > 80 root 0 S /bin/dnsmasq -r /etc/resolv.dnsmasq > 84 ttyS0 root 0 S /bin/sh > 85 root 0 S /bin/inetd > 86 root 0 S /bin/syslogd -n > 87 root 0 S /bin/klogd -n > 89 ttyS0 root 0 S /bin/syslogd -R 192.168.5.4:514 > 93 ttyS0 root 0 R ps The machine 192.168.5.4 is reachable from my gateway(able to ping). The syslogd service on 192.168.5.4 is able to listen on remote machine.(Its /var/log/messages shows the log of 3-machines). But i am unable to get the Log of Gateway(192.168.5.1) on This machine(192.168.5.4). Sorry for my poor english, Please help me. With Regards Kanhu -------------- next part -------------- An HTML attachment was scrubbed... URL: From lars.friedrich at wago.com Mon Mar 21 02:29:40 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Mon, 21 Mar 2005 08:29:40 +0100 Subject: [uClinux-dev] renaming uClinux-dist Message-ID: > No doubt uClibc is domoniant (no glibc choice for non-MMU). > Show of hands... who here uses glibc on VM platforms in the > uClinux-dist? Well, I used uClibc to cross-check a bug in uC-libc. It was quite handy to just menuconfig another library and check if the bug is gone. Same could happen for uClibc vs. glibc. Maybe the source code should just be dumped, not the integration of glibc itself (in other words: make the glibc download optional, but don't remove all traces from the Makefiles etc.). > > Perhaps something to consider in response to the download > > size argument. > Even if I take it out, it is only a short term fix. The dist will > keep growing in size, and within a short time it will be larger > than it is now - even without glibc. Not sure if this argument counts. You wouldn't add 50MB of Paris Hilton's video to the dist, just because the size of the dist will grow anyway and it wouldn't matter much (and I think the video would be used more than glibc...). Best regards, Lars Friedrich (spring fever got me) From kcheong at gmail.com Mon Mar 21 06:03:19 2005 From: kcheong at gmail.com (Edward Chan) Date: Mon, 21 Mar 2005 19:03:19 +0800 Subject: [uClinux-dev] Question on uclinux memory allocation problem Message-ID: Dear all, I have bought a Samsung ARM7TDMI development board with 2MB NOR flash , 8 MB SDRAM, 16 MB Flash. Currently, I want to enable my C program to be run on the uClinux , but encounter insuffient memory error. In desktop, my program only occupy around 1MB, but under uclinux, it reported that my program required up to 8MB memory. I have already enabled Non power-of-2 kernel allocator (EXPERIMENTAL) in my uclinux kernel 2.4.12 and also compile with the kernel with XIP feature. I use the arm-elf-gcc with the following command : arm-elf-gcc -Wl, -elf2flt -O3 -D__PIC__ -fpic -msingle-pic-base -Wall -o test test.c and my program source only allocate memory using static array, e.g. char buffer[1024][1024]... So I don't know why my application will occupy so much memory in the uClinux platform. Would anyone help me to figure out the problem? Thanks for all of your kindess help! -- A newbie in uClinux, Edward Chan From kanhu2004 at rediffmail.com Mon Mar 21 06:21:08 2005 From: kanhu2004 at rediffmail.com (kanhu rauta) Date: 21 Mar 2005 11:21:08 -0000 Subject: [uClinux-dev] Remote Logging Problem with syslogd Message-ID: <20050321112108.20717.qmail@webmail17.rediffmail.com> Hi all, I am using BusyBox v0.60.4 with uClinux. I have selected the klogd and syslogd options of busybox on compilation. When I load this image on my gateway,both process has started successfully. />ps > PID TTY Uid Size State Command > 1 root 0 S /sbin/init > 2 root 0 S [keventd] > 3 root 0 S [ksoftirqd_CPU0] > 4 root 0 S [kswapd] > 5 root 0 S [bdflush] > 6 root 0 S [kupdated] > 51 root 0 S /bin/dhcpd LAN > 77 root 0 S chois -gserver 54273 > 80 root 0 S /bin/dnsmasq -r /etc/resolv.dnsmasq > 84 ttyS0 root 0 S /bin/sh > 85 root 0 S /bin/inetd > 86 root 0 S /bin/syslogd -n > 87 root 0 S /bin/klogd -n > 90 ttyS0 root 0 R ps I get the kernel log messages in the file /var/log/messages. I want to forward this kernel log to a remote machine,Then i exucute the follwing command />/bin/syslogd -R 192.168.5.4:514 & />ps > PID TTY Uid Size State Command > 1 root 0 S /sbin/init > 2 root 0 S [keventd] > 3 root 0 S [ksoftirqd_CPU0] > 4 root 0 S [kswapd] > 5 root 0 S [bdflush] > 6 root 0 S [kupdated] > 51 root 0 S /bin/dhcpd LAN > 77 root 0 S chois -gserver 54273 > 80 root 0 S /bin/dnsmasq -r /etc/resolv.dnsmasq > 84 ttyS0 root 0 S /bin/sh > 85 root 0 S /bin/inetd > 86 root 0 S /bin/syslogd -n > 87 root 0 S /bin/klogd -n > 89 ttyS0 root 0 S /bin/syslogd -R 192.168.5.4:514 > 93 ttyS0 root 0 R ps The machine 192.168.5.4 is reachable from my gateway(able to ping). The syslogd service on 192.168.5.4 is able to listen on remote machine.(Its /var/log/messages shows the log of 3-machines). But i am unable to get the Log of Gateway(192.168.5.1) on This machine(192.168.5.4). Sorry for my poor english, Please help me. With Regards Kanhu -------------- next part -------------- An HTML attachment was scrubbed... URL: From rdkehn at yahoo.com Mon Mar 21 07:04:42 2005 From: rdkehn at yahoo.com (Doug Kehn) Date: Mon, 21 Mar 2005 04:04:42 -0800 (PST) Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: 6667 Message-ID: <20050321120442.41089.qmail@web51908.mail.yahoo.com> --- Greg Ungerer wrote: > Show of hands... who here uses glibc on VM platforms > in the > uClinux-dist? > Yep. __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/ From jhml at gmx.net Mon Mar 21 12:35:15 2005 From: jhml at gmx.net (Johann Hanne) Date: Mon, 21 Mar 2005 18:35:15 +0100 Subject: [uClinux-dev] Illegal instruction with setjmp() on ARM In-Reply-To: <20050321011545.GB3090@beast> References: <200503191948.00559.jhml@gmx.net> <20050321011545.GB3090@beast> Message-ID: <200503211835.15930.jhml@gmx.net> Hi David, thanks for the hint. My uclibc was compiled with "HAS_FPU=y" (and the cx82100 doesn't have one). This was because I used the toolchain from http://www.uclinux.org/pub/uClinux/arm-elf-tools/tools-20030314/ and compiled it with the build-uclinux-tools.sh script, which keeps the default uclibc file. After patching the script to remove the HAS_FPU from the config file, my shell now runs fine. Cheers, Johann On Monday 21 March 2005 02:15, David McCullough wrote: > Jivin Johann Hanne lays it down ... > > > Hi, > > > > I'm currently trying to build a shell for an embedded ARM box. The two > > alternatives are "sh" from uClinux-dist und "msh" from busybox. It seems > > to me that "msh" is based on "sh" (guessed from a quick look at the > > source code), so it isn't surprising that I have the same problem with > > both: > > > > # ./sh > > Illegal instruction > > # > > > > I've tracked down the problem by putting debug printf's into the sources: > > The "Illegal instruction" error is caused by the first setjmp() call in > > sh1.c, line 243: > > setjmp(m1); /* Bruce Evans' fix */ > > The same call exists in busybox, msh.c, line 1107: > > setjmp(failpt = m1); /* Bruce Evans' fix */ > > > > Google gave me two problems with setjmp() on ARM: > > - http://www.uclinux.org/pub/uClinux/archive/7205.html > > This is related to a bug in elf2flt, the fix has been applied in March > > 2002 by Greg Ungerer > > - http://www.uclibc.org/lists/uclibc/2003-February/005726.html > > This is related to a bug in uClibc, the fix has been applied in > > February 2003 by David McCullough > > > > The problem is that I'm using the gcc 2.95.3 toolchain from > > http://www.uclinux.org/pub/uClinux/arm-elf-tools/tools-20030314/ which > > has both fixed included. So there seems to be some other bug, which I'm > > unable to find... > > > > The board has an Arm940sid based CPU (Conexant) CX82100. > > Sounds like you have a uCLibc without the correct floating point support > in setjmp and friends. You should be able to find something on this > in the archives. > > What sources (dist) are you using. > > Alternatively, just double check your .S files for setjmp/longjmp and > sigsetjmp/siglongjmp and make sure they either always do the FP stuff or > never do it (use #if 1/#if 0 instead of #if HAS_FPU or whatever :-), > > Cheers, > Davidm From jhml at gmx.net Mon Mar 21 13:19:05 2005 From: jhml at gmx.net (Johann Hanne) Date: Mon, 21 Mar 2005 19:19:05 +0100 Subject: [uClinux-dev] [patch] 1/4: add CnxtBsp_Mac_MutEx_Chk() function required by Conexant Emac driver In-Reply-To: <423E1F53.6050703@snapgear.com> References: <200503201626.23119.jhml@gmx.net> <423E1F53.6050703@snapgear.com> Message-ID: <200503211919.05113.jhml@gmx.net> Hi Greg, yes, these patches replace the three I posted some weeks ago. In fact they are the same, plus one new patch (the Emac TX DMA thing). Sorry, forgot to mention that. I re-posted them because I thought they got lost somehow and I had an additional patch anyway. Many thanks for applying the whole stuff. I've just checked out CVS and it now runs on the DPCM. Wow! :-)) Cheers, Johann On Monday 21 March 2005 02:11, Greg Ungerer wrote: > Hi Johann, > > Johann Hanne wrote: > > This function is required for the Conexant Emac driver. The kernel will > > not compile without it when the Emac driver is enabled. > > Applied. I take it your current set of patches replace those you > emailed here a couple of weeks back? > > Thanks > Greg > > > ------------------------------------------------------------------------ > > > > --- linux-2.4.27-uc1/arch/armnommu/mach-cx821xx/cnxtbsp.c 2005-02-12 > > 13:26:11.000000000 +0100 +++ > > linux-2.4.27-uc1.patched/arch/armnommu/mach-cx821xx/cnxtbsp.c 2005-02-27 > > 14:18:31.000000000 +0100 @@ -165,6 +165,30 @@ > > *mac_addr_array = CnxtBsp_Get_Random_Byte(); > > } > > > > +/*********************************************************************** > >******* +| > > +| Function: CnxtBsp_Mac_MutEx_Chk > > +| > > +| Description: Used to determine whether or not this device requires > > +| mutually exclusive operation on the emac > > transmitters +| > > +| Returns: NONE > > +| > > +************************************************************************ > >*******/ + > > +BOOLEAN CnxtBsp_Mac_MutEx_Chk( void ) > > +{ > > + switch( _sysGetDeviceType() ) > > + { > > + case 0: > > + return TRUE; > > + break; > > + default: > > + return FALSE; > > + break; > > + } > > +} > > + > > #if CONFIG_CNXT_ADSL || CONFIG_CNXT_ADSL_MODULE > > > > > > /************************************************************************ > >**** > > > > > > ------------------------------------------------------------------------ > > > > _______________________________________________ > > uClinux-dev mailing list > > uClinux-dev at uclinux.org > > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > > This message was resent by uclinux-dev at uclinux.org From rdkehn at yahoo.com Mon Mar 21 18:16:04 2005 From: rdkehn at yahoo.com (Doug Kehn) Date: Mon, 21 Mar 2005 15:16:04 -0800 (PST) Subject: [uClinux-dev] Memory Leak (Bridge Module Related) Message-ID: <20050321231605.97165.qmail@web51908.mail.yahoo.com> Board: ADI-Coyote Distro: uClinx-20041215 ixp400AccessLibrary-1_4.zip snapgear-3.1.x-ixp400-v1.4.2-support.tar.gz Kernel: Linux 2.4.27-uc1 Toolchain: arm-linux-tools-2003-1127 When I enable the bridge module memory is consumed until exhausted. At which point the kernel Oops. The rate at which memory is consumed is proportional to the network load. If I disable the bridge module (and just route) the leak is not present. In looking at slabinfo, the memory seems to be allocated to the size-64 object. For example, just before the Oops a cat of /proc/slabinfo yields: size-64 668854 668860 72 12620 12620 1 : 668854 836391 12620 0 0 I noticed that there is a 2.6.x tracing patch for the slab allocator. Does such a patch exist for the 2.4.x kernel? Or does anyone know of any tools/techniques for tracking slab related memory leaks? Thanks, ...doug __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/ From swenggsri at yahoo.com Mon Mar 21 18:40:05 2005 From: swenggsri at yahoo.com (Sriram Chadalavada) Date: Mon, 21 Mar 2005 15:40:05 -0800 (PST) Subject: [uClinux-dev] m68k 2.95-3 cross compiler build errors Message-ID: <20050321234005.39817.qmail@web21604.mail.yahoo.com> Hi everyone, I am trying to build the m68k 2.95-3 toolchain (with STRICT_ALIGNMENT turned on for uCdimm). Here's what I did ----------------- 1. Downloaded the gcc tarball and "full.patch" from http://www.uclinux.org/pub/uClinux/m68k-elf-tools/tools-20030314/. 2. Typed ./configure --host=i686-linux-gnu --target=m68k-elf-linux-gnu to set up the cross compiler configuration. 3. Changed the STRICT_ALIGNMENT macro value gcc/config/linux.h to 1. 4. Typed 'make'. 5. Obtained the following messages: config/svr4.h:418: warning: `MAX_OFILE_ALIGNMENT' redefined config/m68k/m68k.h:2132: warning: this is the location of the previous definition /tmp/ccYzw9t9.s: Assembler messages: /tmp/ccYzw9t9.s:287: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:288: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:289: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:290: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:291: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:292: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:293: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:295: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:297: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:298: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:299: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:300: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:301: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:302: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:303: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:304: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:305: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:306: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:307: Error: invalid character '.' in mnemonic /tmp/ccYzw9t9.s:308: Error: no such instruction: `unlk %a6' /tmp/ccYzw9t9.s:309: Error: no such instruction: `rts' These are m68k instructions, why are they not being recognized? Please let me know what is going on here. Did I miss something in the build process? THank you, Sriram __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/ From gerg at snapgear.com Mon Mar 21 19:52:33 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Tue, 22 Mar 2005 10:52:33 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: References: Message-ID: <423F6C51.6060800@snapgear.com> Hi Lars, Friedrich, Lars wrote: >>No doubt uClibc is domoniant (no glibc choice for non-MMU). >>Show of hands... who here uses glibc on VM platforms in the >>uClinux-dist? > > Well, I used uClibc to cross-check a bug in uC-libc. It was > quite handy to just menuconfig another library and check > if the bug is gone. Same could happen for uClibc vs. glibc. > Maybe the source code should just be dumped, not the > integration of glibc itself (in other words: make the glibc > download optional, but don't remove all traces from the > Makefiles etc.). I have done this cross checking against multiple libraries, it is very useful. >>>Perhaps something to consider in response to the download >>>size argument. >> >>Even if I take it out, it is only a short term fix. The dist will >>keep growing in size, and within a short time it will be larger >>than it is now - even without glibc. > > > Not sure if this argument counts. You wouldn't add 50MB of Paris > Hilton's video to the dist, just because the size of the dist will > grow anyway and it wouldn't matter much (and I think the video would > be used more than glibc...). Hmmm, well, yes :-) If glibc is used by a bunch of people though (and thus throwing the question out there) I would not want to take it out. Even if a Paris Hilton video would be used more often by some :-) To give it a reference, the glibc source probably adds about 15MB of the final uClinux-dist size. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From gerg at snapgear.com Mon Mar 21 20:00:19 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Tue, 22 Mar 2005 11:00:19 +1000 Subject: [uClinux-dev] [patch] 1/4: add CnxtBsp_Mac_MutEx_Chk() function required by Conexant Emac driver In-Reply-To: <200503211919.05113.jhml@gmx.net> References: <200503201626.23119.jhml@gmx.net> <423E1F53.6050703@snapgear.com> <200503211919.05113.jhml@gmx.net> Message-ID: <423F6E23.2040902@snapgear.com> Hi Johann, Johann Hanne wrote: > yes, these patches replace the three I posted some weeks ago. In fact they are > the same, plus one new patch (the Emac TX DMA thing). Sorry, forgot to > mention that. I re-posted them because I thought they got lost somehow and I > had an additional patch anyway. > > Many thanks for applying the whole stuff. I've just checked out CVS and it now > runs on the DPCM. Wow! :-)) No problem. I still had your last patches in my Inbox, just hadn't got to applying them yet. I still have a few patches in email from others I need to apply yet. When you send them small, one per email, I can reveiw and apply them much more quickly... Regards Greg > On Monday 21 March 2005 02:11, Greg Ungerer wrote: > >>Hi Johann, >> >>Johann Hanne wrote: >> >>>This function is required for the Conexant Emac driver. The kernel will >>>not compile without it when the Emac driver is enabled. >> >>Applied. I take it your current set of patches replace those you >>emailed here a couple of weeks back? >> >>Thanks >>Greg >> >> >>>------------------------------------------------------------------------ >>> >>>--- linux-2.4.27-uc1/arch/armnommu/mach-cx821xx/cnxtbsp.c 2005-02-12 >>>13:26:11.000000000 +0100 +++ >>>linux-2.4.27-uc1.patched/arch/armnommu/mach-cx821xx/cnxtbsp.c 2005-02-27 >>>14:18:31.000000000 +0100 @@ -165,6 +165,30 @@ >>> *mac_addr_array = CnxtBsp_Get_Random_Byte(); >>> } >>> >>>+/*********************************************************************** >>>******* +| >>>+| Function: CnxtBsp_Mac_MutEx_Chk >>>+| >>>+| Description: Used to determine whether or not this device requires >>>+| mutually exclusive operation on the emac >>>transmitters +| >>>+| Returns: NONE >>>+| >>>+************************************************************************ >>>*******/ + >>>+BOOLEAN CnxtBsp_Mac_MutEx_Chk( void ) >>>+{ >>>+ switch( _sysGetDeviceType() ) >>>+ { >>>+ case 0: >>>+ return TRUE; >>>+ break; >>>+ default: >>>+ return FALSE; >>>+ break; >>>+ } >>>+} >>>+ >>> #if CONFIG_CNXT_ADSL || CONFIG_CNXT_ADSL_MODULE >>> >>> >>>/************************************************************************ >>>**** >>> >>> >>>------------------------------------------------------------------------ >>> >>>_______________________________________________ >>>uClinux-dev mailing list >>>uClinux-dev at uclinux.org >>>http://mailman.uclinux.org/mailman/listinfo/uclinux-dev >>>This message was resent by uclinux-dev at uclinux.org > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uclinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org > -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From liubl at neusoft.com Mon Mar 21 20:04:55 2005 From: liubl at neusoft.com (liubaoliang) Date: Tue, 22 Mar 2005 09:04:55 +0800 Subject: [uClinux-dev] about the diffrent executing result between image.ram and image.rom Message-ID: <0IDQ00FFLAXKD7@smtp.neusoft.com> hi, all I had met a new problem.We have a EEROM,it's a IIC device.If I use the image.ram which is downloadded to RAM to operate the EEROM such as read and write,the executing result is ok.But if use the image.rom which is writed to FLASH ROM,sometime I found the result is fault,why? I think these two image use the same optimization level -O2,so the result should be same. Another reason maybe the compressed kernel have some problem during the compressing ? Best regards. ????????? Liu Baoliang 2005-03-22 Ph. (86) 024-83665554 Fax (86) 024-23782817 Email : liubl at neusoft.com From liubl at neusoft.com Mon Mar 21 23:42:55 2005 From: liubl at neusoft.com (liubaoliang) Date: Tue, 22 Mar 2005 12:42:55 +0800 Subject: [uClinux-dev] about the diffrent executing result between image.ram and image.rom Message-ID: <0IDQ00JDZL0XAB@smtp.neusoft.com> hi, all I had met a new problem.We have a EEROM,it's a IIC device.If I use the image.ram which is downloadded to RAM to operate the EEROM such as read and write,the executing result is ok.But if use the image.rom which is writed to FLASH ROM,sometime I found the result is fault,why? I think these two image use the same optimization level -O2,so the result should be same. Another reason maybe the compressed kernel have some problem during the compressing ? Best regards. ????????? Liu Baoliang 2005-03-22 Ph. (86) 024-83665554 Fax (86) 024-23782817 Email : liubl at neusoft.com Best regards. ????????? Liu Baoliang 2005-03-22 Company : Philips and Neusoft Medical Systems Co.,Ltd. Address : Neusoft Park Hun Nan Industrial Area New&High-Tech Development Zone Shenyang China 110179 Ph. (86) 024-83665554 Fax (86) 024-23782817 Email : liubl at neusoft.com From rootfs at gmail.com Tue Mar 22 01:33:23 2005 From: rootfs at gmail.com (Kartikeyan) Date: Tue, 22 Mar 2005 12:03:23 +0530 Subject: [uClinux-dev] uclinux newbie Message-ID: <423FBC33.4070400@gmail.com> Hello, I recently downloaded latest uclinux distribution and xcopilot simulator I am able to run uclinux on the simulator compiling I get a sash shell But i am not able to do telnet localhost 2000 # it seems nothing is listeninng to 2000 as suggested in http://www.cyberguard.info/snapgear/tb20020807.html by David McCullogh I have rh9 kernel versions 2.4.20-8 and 2.6.10 Can any one guide me what to do further and play with uclinux any urls to go further into simulator and learn etc Thankyou SK From devi at tenet.res.in Tue Mar 22 14:34:52 2005 From: devi at tenet.res.in (devi) Date: Wed, 23 Mar 2005 01:04:52 +0530 Subject: [uClinux-dev] console Message-ID: <20050322193452.31225.qmail@mail.tenet.res.in> hi, The characters that I write on the /dev/tty0 has to appear on the framebuffer console.How to do this.My framebuffer console shows a blank screen with linux logo on it with a blinking cursor.How do I test my frame buffer displays character.Any idea. Regards Devi From jwilliams at itee.uq.edu.au Tue Mar 22 04:29:48 2005 From: jwilliams at itee.uq.edu.au (John Williams) Date: Tue, 22 Mar 2005 19:29:48 +1000 Subject: [uClinux-dev] console In-Reply-To: <20050322193452.31225.qmail@mail.tenet.res.in> References: <20050322193452.31225.qmail@mail.tenet.res.in> Message-ID: <423FE58C.4020007@itee.uq.edu.au> hi Devi, devi wrote: > The characters that I write on the /dev/tty0 has to appear on the framebuffer console.How to do this.My framebuffer console shows a blank screen with linux logo on it with a blinking cursor.How do I test my frame buffer displays character.Any idea. I don't have an answer for you, but rather a request. It would be very valuable to me if you could explain the steps you took to get this far with your framebuffer - that is, getting the logo to display and so on. Most of the existing documentation assumes you just want to enable framebuffer on an already-supported board and platform - there are little or no details on how to do this for a new platform. any information you can share with the community in this regard would be greatly appreciated. Thanks, John From swenggsri at yahoo.com Tue Mar 22 05:06:16 2005 From: swenggsri at yahoo.com (Sriram Chadalavada) Date: Tue, 22 Mar 2005 02:06:16 -0800 (PST) Subject: [uClinux-dev] m68k toolchain STRICT ALIGNMENT turned on Message-ID: <20050322100616.36844.qmail@web21601.mail.yahoo.com> HI everyone, I am trying to overcome some memory alignment issues on my uCdimm(68000 core) that is causing the kernel to crash. The m68k toolchain (20030314) does not allow use of flags such as -m(no)strict-align (to overcome above problem) during compilation. Please let me know if anybody has worked on this and can provide me a working toolchain as soon as possible. Thank you in advance, Sriram __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/ From kanhu2004 at rediffmail.com Tue Mar 22 05:24:58 2005 From: kanhu2004 at rediffmail.com (kanhu rauta) Date: 22 Mar 2005 10:24:58 -0000 Subject: [uClinux-dev] syslogd -R option is not working Message-ID: <20050322102458.6806.qmail@webmail17.rediffmail.com> Hi all, I am using BusyBox v0.60.4 with uClinux. I have selected the klogd and syslogd options of busybox on compilation. When I load this image on my gateway,both process has started successfully. />ps PID TTY Uid Size State Command 1 root 0 S /sbin/init 2 root 0 S [keventd] 3 root 0 S [ksoftirqd_CPU0] 4 root 0 S [kswapd] 5 root 0 S [bdflush] 6 root 0 S [kupdated] 51 root 0 S /bin/dhcpd LAN 77 root 0 S chois -gserver 54273 80 root 0 S /bin/dnsmasq -r /etc/resolv.dnsmasq 84 ttyS0 root 0 S /bin/sh 85 root 0 S /bin/inetd 86 root 0 S /bin/syslogd -n 87 root 0 S /bin/klogd -n 90 ttyS0 root 0 R ps I get the kernel log messages in the file /var/log/messages. I want to forward this kernel log to a remote machine,Then i exucute the follwing command />/bin/syslogd -R 192.168.5.4:514 & />ps PID TTY Uid Size State Command 1 root 0 S /sbin/init 2 root 0 S [keventd] 3 root 0 S [ksoftirqd_CPU0] 4 root 0 S [kswapd] 5 root 0 S [bdflush] 6 root 0 S [kupdated] 51 root 0 S /bin/dhcpd LAN 77 root 0 S chois -gserver 54273 80 root 0 S /bin/dnsmasq -r /etc/resolv.dnsmasq 84 ttyS0 root 0 S /bin/sh 85 root 0 S /bin/inetd 86 root 0 S /bin/syslogd -n 87 root 0 S /bin/klogd -n 89 ttyS0 root 0 S /bin/syslogd -R 192.168.5.4:514 93 ttyS0 root 0 R ps The machine 192.168.5.4 is reachable from my gateway(able to ping). The syslogd service on 192.168.5.4 is able to listen on remote machine.(Its /var/log/messages shows the log of 3-machines). But i am unable to get the Log of Gateway(192.168.5.1) on This machine(192.168.5.4). Sorry for my poor english, Please help me. With Regards Kanhu Kanhu Charan Rauta Innomedia Techonologies Pvt. Ltd #3278,12Th Main, HAL 2nd stage. Banglore-560008. -------------- next part -------------- An HTML attachment was scrubbed... URL: From devi at tenet.res.in Tue Mar 22 16:28:14 2005 From: devi at tenet.res.in (devi) Date: Wed, 23 Mar 2005 02:58:14 +0530 Subject: [uClinux-dev] console Message-ID: <20050322212814.6968.qmail@mail.tenet.res.in> hi, Details to set the framebuffer device. The framebuffer driver for the device to be compiled (use skeletonnfb.c to write the driver)and call those framebuffer driver init from fbmem.c.So that the driver gets registered. Set the display parameters such as xres,yres,address of the framebuffer device and the display to use(like monochrome or color with how many bpps) in the frame buffer driver. Add a frame buffer device in vendors/Makefile. Execute the application like lissa.c to test whether the device works. To use it as a console enable the VT_CONSOLE. Regards Devi From marny at rambler.ru Tue Mar 22 05:50:53 2005 From: marny at rambler.ru (Igor Marnat) Date: Tue, 22 Mar 2005 13:50:53 +0300 Subject: [uClinux-dev] Remote Logging Problem with syslogd In-Reply-To: <20050321112108.20717.qmail@webmail17.rediffmail.com> References: <20050321112108.20717.qmail@webmail17.rediffmail.com> Message-ID: <200503221350.53909.marny@rambler.ru> On 21 March 2005 14:21, kanhu rauta wrote: > I get the kernel log messages in the file /var/log/messages. > I want to forward this kernel log to a remote machine,Then i exucute the > follwing command />/bin/syslogd -R 192.168.5.4:514 & > />ps > > > PID TTY Uid Size State Command ... > > 85 root 0 S /bin/inetd > > 86 root 0 S /bin/syslogd -n > > 87 root 0 S /bin/klogd -n > > 89 ttyS0 root 0 S /bin/syslogd -R 192.168.5.4:514 > > 93 ttyS0 root 0 R ps Perhaps, you have first to stop the first instance of "syslogd -n" which in the listing above has the PID of 86? If it doesn't help, you could define DEBUG_REMOTE in file syslogd.c, it might be helpful too to see the output. Regards, Igor From leelawrenceyh at yahoo.com.cn Tue Mar 22 07:13:07 2005 From: leelawrenceyh at yahoo.com.cn (lawrence lee) Date: Tue, 22 Mar 2005 20:13:07 +0800 (CST) Subject: [uClinux-dev] Does uclibc support iconv? Message-ID: <20050322121307.18641.qmail@web15002.mail.cnb.yahoo.com> Hi now I use uclibc-0.9.27.so. can you tell me if uclibc-0.9.27.so support the implementation of iconv_open? Because when I compiled my application, I found that errors: undefined reference to 'iconv_open' undefined referenct to 'dcgettext' undefined referenct to 'iconv_close' undefined referenct to 'iconv' undefined referenct to 'bindtextdomain' This functions' announcement is in /include/libintl.h and /include/iconv.h Thanks --------------------------------- Do You Yahoo!? 150??MP3???????????? ??????????????????? 1G??1000??????????? -------------- next part -------------- An HTML attachment was scrubbed... URL: From carlos at embedded.cl Tue Mar 22 07:56:49 2005 From: carlos at embedded.cl (Carlos Manuel Duclos Vergara) Date: Tue, 22 Mar 2005 12:56:49 -0000 Subject: [uClinux-dev] syslogd -R option is not working Message-ID: <20050322125653.3CBFC14DA37@mail07.powweb.com> An embedded and charset-unspecified text was scrubbed... Name: not available URL: From ulf.soderberg at gmail.com Tue Mar 22 10:39:47 2005 From: ulf.soderberg at gmail.com (=?ISO-8859-1?Q?Ulf_S=F6derberg?=) Date: Tue, 22 Mar 2005 16:39:47 +0100 Subject: [uClinux-dev] Wrong memory mapping for 2.6.11 and IPX425 Message-ID: <42403C43.8090000@gmail.com> We have been struggeling with the USB Gadget Device driver (pxa2xx_udc) for use with IXP425 and 2.6.11.3 kernel for a couple of days. Today we found that the following definition in the ixp4xx-regs.h file causes an overlap with the expansion bus: #define IXP425_PERIPHERAL_BASE_VIRT (0xffdb2000) By changing it to #define IXP425_PERIPHERAL_BASE_VIRT (0xffdb1000) the USB Gadget Device works ok (with a few minor fixes and additions that I will report later). -- Ulf S?derberg http://osgifun.blogspot.com From ngustavson at emacinc.com Tue Mar 22 13:46:03 2005 From: ngustavson at emacinc.com (NZG) Date: Tue, 22 Mar 2005 12:46:03 -0600 Subject: [uClinux-dev] Changing mcf serial consoles Message-ID: <200503221246.03373.ngustavson@emacinc.com> I need to put my serial console on the 3rd com port, rather than the first one on my Coldfire uClinux build. Currently the port number is hard coded, I can change the port it outputs on easily enough by changing mcfserial.c, down in the console section. But how do I change the port the console gets input on? This is not as obvious. thx, NZG. From devi at tenet.res.in Wed Mar 23 10:00:25 2005 From: devi at tenet.res.in (devi) Date: Wed, 23 Mar 2005 20:30:25 +0530 Subject: [uClinux-dev] console Message-ID: <20050323150025.27090.qmail@mail.tenet.res.in> hi, Details to set the framebuffer device. The framebuffer driver for the device to be compiled and call those framebuffer driver init from fbmem.c.So that the driver gets registered. Set the display parameters such as xres, yres, address of the framebuffer device and the display to use(like monochrome or color with how many bpps) in the frame buffer driver. Add a frame buffer device in vendors/Makefile. Execute the application like lissa.c to test whether the device works. To use it as a console enable the VT_CONSOLE. Regards Devi From CCarson at CBWAutomation.com Wed Mar 23 11:27:52 2005 From: CCarson at CBWAutomation.com (Carson, Corey) Date: Wed, 23 Mar 2005 09:27:52 -0700 Subject: [uClinux-dev] Dragonball VZ uCdimm LCD Message-ID: <203F609DF3BDD611843C000475A1967C06C8C6@mail.cbwautomation.com> Please Help!!! I have a Dragonball VZ and am running uClinux 2.4.x. I have an LCD screen that I bought at Arcturus (http://www.earthlcd.com/pdalcd.htm). All I want to do is display text on the LCD. Nothing fancy... Just a simple string. Someone please tell me there's documents explaining how to do this? I know I have the correct drivers setup. When I type lsmod, I get the following: Module Size Used by Not tainted Mc68328ts 7116 0 (unused) Mc68x328fb 4920 63 Fbcon-mfb 3068 0 [mc68x328fb] Lcd_module 10596 0 (unused) Could someone please post there C/C++ program showing me how to write text? Thanks, Corey From info at hostinthebox.net Wed Mar 23 17:33:29 2005 From: info at hostinthebox.net (dan) Date: Wed, 23 Mar 2005 15:33:29 -0700 Subject: [uClinux-dev] Sending email In-Reply-To: <002501c52c2c$7cf52170$6202a8c0@flanders> References: <002501c52c2c$7cf52170$6202a8c0@flanders> Message-ID: <4241EEB9.4030601@hostinthebox.net> Mike Lavender wrote: > Hello, > > Does anybody have example code for sending an email from within a C > program on uClinux. I have compiled in the mail program and it works from > the command line, but I am not sure how to invoke it from a C program and > get the message body in. I have tried using "popen" but it doesn't seem to > work (possibly do to a fork issue?). > > Is this how people are sending emails is by invoking a command line from > a C program like in the Linux desktop world with sendmail? Or are people > adding smtp code directly to their application? > > Thank you, > > Mike Lavender > Mike - I've been doing a lot of PHP lately, and have bene using raw sockets to connect directly to a remote mail server, and deliver messages following the SMTP (RFC821 - http://www.faqs.org/rfcs/rfc821.html) protocol. For one, raw sockets in this fashion have proven to be quicker, in my testing, than using php's mail() function. They also give me the ability to manipulate headers and such, as well. You might now be asking yourself why I'm talking about PHP. Well, I am stressing the idea of perhaps making raw sockets through your application that will connect to a remote SMTP server and deliver mail that way. `mail`'s source code should provide some insight as to how this is done from within C. Regretably, I do not know any C, but from a programming standpoint I would imagine that this would be quite trivial. Hope that helps -dant From devi at tenet.res.in Thu Mar 24 15:51:08 2005 From: devi at tenet.res.in (devi) Date: Fri, 25 Mar 2005 02:21:08 +0530 Subject: [uClinux-dev] console Message-ID: <20050324205108.23669.qmail@mail.tenet.res.in> hi, For the virtual console what are the device to be added in the vendors/Makefile.I added tty0,tty1,tty2 and tty3. Apart from this there is any such device to be added.How vc/0 is registered as tty0. Regards Devi From umesh at micro-web.co.kr Thu Mar 24 05:30:41 2005 From: umesh at micro-web.co.kr (umesh) Date: Thu, 24 Mar 2005 19:30:41 +0900 Subject: [uClinux-dev] memory allocation problem ! References: <0IDQ00FFLAXKD7@smtp.neusoft.com> Message-ID: <01bd01c5305c$a506fbf0$6501a8c0@umesh> Hi all, I am using SnapGear Embedded Linux Distribution having linux kernel version 2.4.20 & my borad is haivng 16 MB physical RAM. I'm facing a strange problem related to memory allocation via insmod . whenever I boot the system via flash & I try to insmod my mpeg4 driver which is big in size i.e.2.6 MB around, system gives a error : insmod: allocation of 768 pages failed! the mem_map is : init Mem: 129248 Slack: 3696 Shared: 0 sh Mem: 133408 Slack: 3066 Shared: 0 insmod Mem: 3704704 Slack: 72040 Shared: 0 Total Mem: 3967360 Slack: 78802 Shared: 0 Active: 0, inactive: 5, free: 1397 Free pages: 1397 (5588kB), %1 frag, %3 slack Free blks: 17 min=1 max=762 avg=82 Used blks: 18 min=1 max=1435 avg=149 But when I use NFS, I can successfully insmod the same mpeg4 driver without any memory allocation problem. Even I tried to make the romfs having only this driver module to ruduce the size of whole image, but getting the same error. Cud somebody help me? whats wrong here? Either NFS or romfs build , as image size is same except romfs and nfs code in the kernel, i think both will occupied the same amount of memory within the RAM after booting as I dont have any thing within romfs except driver module. does the driver within romfs which is somewhat big (2.6 mb) creating the prob ? why the system is unable to allocate the required memory as my board is having 11 MB RAM. Any suggesion is highly apreciated. Umesh micro-web From davidm at snapgear.com Thu Mar 24 07:03:30 2005 From: davidm at snapgear.com (David McCullough) Date: Thu, 24 Mar 2005 22:03:30 +1000 Subject: [uClinux-dev] Re: uclinux newbie telnet localhost 2000 not working In-Reply-To: <185c1c6005032401355d4a5862@mail.gmail.com> References: <185c1c6005032401355d4a5862@mail.gmail.com> Message-ID: <20050324120330.GB5300@beast> Jivin sunil s lays it down ... > Hello, > I recently downloaded uclinux and installed xcopilot and m68k tool chain. > Everything is successful. > I can run the uclinux under xcopilot. > > But in the Article > http://www.cyberguard.info/snapgear/tb20020807.html > it says how to telnet to xcopilot on port 2000 > I can't telnet telnet localhost 2000 after running xcopilot > rest all work sash shell comes up etc. You are running the telnet oin your dev machine, not under Xcopiot ? > Also i want to know can i include nfs support in the uclinux for > xcopilot target , and get binaries from the nfs server and run in the > simulator. Your would need to use ppp on the console of Xcopilot to do this. I would not recommend it. > Can i have a custom root file system in the simulator Easy, if you have built the source using the uClinux dist then: 1) you can copy binaries into romfs/... and then run "make image" to get a new rootfs image with you binaries, just restart xcopilot and you have it. 2) You can turn on applications using "make xconfig", rebuild and they will be in the filesystem ready to use. 3) likewise you can add your applications to the dist build framework, see the Adding -User_apps howto in the Documentation dir. Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From wnlee at cad.csie.ncku.edu.tw Thu Mar 24 08:46:20 2005 From: wnlee at cad.csie.ncku.edu.tw (Wayne Lee) Date: Thu, 24 Mar 2005 21:46:20 +0800 Subject: [uClinux-dev] Is smc91x.c the best choice for smsc91c111 chip inuClinux 2.6? In-Reply-To: <6917.1111057684@redhat.com> Message-ID: <200503241349.j2ODnTu8009802@cad.csie.ncku.edu.tw> Yap, I have back ported smc91x.[ch] from mainstream kernel 2.6.10 to uClinux. The smc91x.[ch] indeed works on uClinux and seems to be the best choice for 2.6.x uClinux. Thanks. Wayne -----Original Message----- From: dhowells at redhat.com [mailto:dhowells at redhat.com] Sent: Thursday, March 17, 2005 7:08 PM To: Wayne Lee Cc: 'Heinz Haeberle'; uclinux-dev at uClinux.org Subject: Re: [uClinux-dev] Is smc91x.c the best choice for smsc91c111 chip inuClinux 2.6? Wayne Lee wrote: > I now access smsc91c111 by a 16-bit read/write register (port IO), this port > IO register is located at 0xc000600 (32bit register), for example. > > Should the SMC_BASE_ADDRESS be 0xC000_0600 and SMC_BASE_END be 0xC000_0603? Well, the compiler will probably complain at the '_' characters in the middle of the numbers. Apart from that, it may be reasonable. I've given it a 256-byte window because that's what the CPU bus controller on this board I've been playing with assigns to that device. David From gerg at snapgear.com Thu Mar 24 09:43:28 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Fri, 25 Mar 2005 00:43:28 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <42409C82.5090701@itee.uq.edu.au> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <20050315081337.0A198633A@mail2.atmark-techno.com> <20050319002457.GI11709@pengutronix.de> <423E0DDC.6080809@snapgear.com> <423E311A.3060208@itee.uq.edu.au> <423E5955.8010104@snapgear.com> <20050322162414.2455DE9AB@mail2.atmark-techno.com> <42409C82.5090701@itee.uq.edu.au> Message-ID: <4242D210.5040908@snapgear.com> Hi John, John Williams wrote: > 1. We keep pushing. uClinux.org gets a revamp, and we push hard to > break the uC==NOMMU misconception. Just my opinion, but I don't think we will ever break that association now. And that in itself is not neccessarily a bad thing. It is the association that uClinux-dist only works on MMUless that is more annoying. > 3. How about two trees, and two dist archives? One is stable, exactly > as it currently is. The other is a daily/weekly/monthly tarball This is sortof what I do now, with the uClinux-dist-test cuts. Though not as often as daily/weekly/monthly. > generated directly from SnapGear's internal tree, where our patches > currently sit between dist cuts. The tarring and publishing can be > scripted, so no extra effort there. Patches on the dist go through the > same level review as they do now. On this same cycle, there is a > uClinux-dist-unstable CVS tree at cvs.uclinux.org, which tracks (or > simply "is") the unstable tree. When it's time for a new "stable" > dist-cut, everyone gets an opportunity to try SnapGear's test cut, which > in theory should match the current unstable tree anyway. Once the dust > settles, the new dist is released, and the unstable tree reverted back > to mirror the new "stable" tree, and the cycle repeats Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From katja.hahn at demetec.de Thu Mar 24 19:00:46 2005 From: katja.hahn at demetec.de (Katja Hahn) Date: Thu, 24 Mar 2005 16:00:46 -0800 Subject: [uClinux-dev] zombies after terminating threads Message-ID: <200503241600.46059.katja.hahn@demetec.de> Hi all, In my application I create 2 threads but after they terminated the join function returns "0". After checking the process status the threads appear as Zombies. Any Idea Why? I would be gratefull for any help? Kat My code looks like this: pthread_t createDataThreadID; pthread_t packDataThreadID; void * createData(void * in){ /* doing somthing */ pthread_exit(NULL); } void * packData(void * in){ //Thread to transmit data. /* doing something else */ pthread_exit(NULL); } void signals(unsigned char *data) { int errNo; unsigned long erg = 0; errNo = pthread_create( &createDataThreadID, 0, createData, NULL); if (errNo != 0) { logerror ("Error createDataThread"); } errNo = pthread_create (&packDataThreadID, 0, packData, (void*)packetType); if (errNo != 0) { logerror ("Error packDataThread"); } /* doing something else */ errNo = pthread_join(createDataThreadID, (void**)erg); printf("\ncreate data thread end status: %d\n", errNo ); errNo = pthread_join(packDataThreadID, (void**)erg); printf("\npack data thread end status: %d\n", errNo ); } From kapheine at gmail.com Tue Mar 22 13:17:51 2005 From: kapheine at gmail.com (Zachary Landau) Date: Tue, 22 Mar 2005 13:17:51 -0500 Subject: [uClinux-dev] Layered SPI Patches Message-ID: Hello, Has anyone found any issues with the layered SPI patch and the 5282 support patch on top of that? If not, is there anything keeping that from being checked into CVS? And if so, please mention the issues so we can try to resolve them. With the 5282 patch I posted last week, I've successfully used both the spi char driver and a custom made driver without any problems. -- Zachary P. Landau From yashi at atmark-techno.com Tue Mar 22 21:35:56 2005 From: yashi at atmark-techno.com (Yasushi SHOJI) Date: Wed, 23 Mar 2005 11:35:56 +0900 Subject: [uClinux-dev] dist size (was: renaming uClinux-dist) In-Reply-To: <42409C82.5090701@itee.uq.edu.au> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <20050315081337.0A198633A@mail2.atmark-techno.com> <20050319002457.GI11709@pengutronix.de> <423E0DDC.6080809@snapgear.com> <423E311A.3060208@itee.uq.edu.au> <423E5955.8010104@snapgear.com> <20050322162414.2455DE9AB@mail2.atmark-techno.com> <42409C82.5090701@itee.uq.edu.au> Message-ID: <20050323023556.ABF08E9AB@mail2.atmark-techno.com> At Wed, 23 Mar 2005 08:30:26 +1000, John Williams wrote: > > >>>Perhaps something to consider in response to the download size argument. > >> > >>Even if I take it out, it is only a short term fix. The dist will > >>keep growing in size, and within a short time it will be larger than > >>it is now - even without glibc. > > > > if we want the dist to be collection of source code, we can't avoid > > getting bigger and bigger. > > > > # don't get me wrong, I'm not against the idea of collection of > > # source. it's a feature. > > Dare I declare concensus here that - Paris Hilton videos notwithstanding > - tarball size is a non-issue? :) well I'm not saying it's a non-issue ;) > If physically hosting/distributing the thing becomes a concern, maybe > something like bittorrent could help out. or we can do how Debian is doing. apt works just fine with collection of package in CD _and_ collection on remote site. it's just that we are primarily focusing on source code collection instead of binary packages. does anyone know how gentoo's doing? I think they are handling source code as a first-class. or *bsd's ports? if we can have that kind of mechanism, we'll all be happy. ;) [very good summary ommitted.] regards, -- yashi From hyok.choi at samsung.com Thu Mar 24 04:56:17 2005 From: hyok.choi at samsung.com (Hyok S. Choi) Date: Thu, 24 Mar 2005 18:56:17 +0900 Subject: [uClinux-dev] [PATCH] the nommu support for ARM linux 2.6.10 Message-ID: <0IDU00M02OZXJG@mmp1.samsung.com> Hi all, I'm glad to release the first MPU and NO-MMU support patch for ARM linux 2.6.10 at : http://opensrc.sec.samsung.com/download/linux-2.6.10-hsc0.patch.gz Previously, it was released as "armnommu" architecture which was called uClinux/ARM and now is merged with "arm" architecture. You can select the memory management type "MMU" for linux and "MPU" / "NONE" for uClinux at the kernel configuration menu. Currently, the ATMEL, S5C7375, S3C24A0 platform is tested. I'm working on 2.6.11.5 patch, and it will be provided soon. It would be great if you give me some suggestions. Best Regards, Hyok ---- HYOK S. CHOI Digital Media R&D Center, Samsung Electronics Co.,Ltd. tel: +82-31-200-8594/5723 fax: +82-31-200-3427 e-mail: hyok.choi at samsung.com [Linux 2.6 MPU/noMMU Kernel Maintainer] http://opensrc.sec.samsung.com From rootfs at gmail.com Thu Mar 24 04:35:54 2005 From: rootfs at gmail.com (sunil s) Date: Thu, 24 Mar 2005 15:05:54 +0530 Subject: [uClinux-dev] uclinux newbie telnet localhost 2000 not working Message-ID: <185c1c6005032401355d4a5862@mail.gmail.com> Hello, I recently downloaded uclinux and installed xcopilot and m68k tool chain. Everything is successful. I can run the uclinux under xcopilot. But in the Article http://www.cyberguard.info/snapgear/tb20020807.html it says how to telnet to xcopilot on port 2000 I can't telnet telnet localhost 2000 after running xcopilot rest all work sash shell comes up etc. Also i want to know can i include nfs support in the uclinux for xcopilot target , and get binaries from the nfs server and run in the simulator. Can i have a custom root file system in the simulator Thankyou In advance S Kartikeyan From devi at tenet.res.in Wed Mar 23 15:23:52 2005 From: devi at tenet.res.in (devi) Date: Thu, 24 Mar 2005 01:53:52 +0530 Subject: [uClinux-dev] console Message-ID: <20050323202352.14347.qmail@mail.tenet.res.in> hi, How does a /dev/vc/0 related to /dev/tty0.In console.c vc/0 is registered.In vendors/makefile do I need to create vc/0 and tty0 and link them.Or only tty0 can be added. Regards Devi From kanhu2004 at rediffmail.com Tue Mar 22 03:36:38 2005 From: kanhu2004 at rediffmail.com (kanhu rauta) Date: 22 Mar 2005 08:36:38 -0000 Subject: [uClinux-dev] syslogd -R option is not working in busybox v0.60.4 Message-ID: <20050322083638.20997.qmail@webmail49.rediffmail.com> Hi all, I am using BusyBox v0.60.4 with uClinux. I have selected the klogd and syslogd options of busybox on compilation. When I load this image on my gateway,both process has started successfully. />ps PID TTY Uid Size State Command 1 root 0 S /sbin/init 2 root 0 S [keventd] 3 root 0 S [ksoftirqd_CPU0] 4 root 0 S [kswapd] 5 root 0 S [bdflush] 6 root 0 S [kupdated] 51 root 0 S /bin/dhcpd LAN 77 root 0 S chois -gserver 54273 80 root 0 S /bin/dnsmasq -r /etc/resolv.dnsmasq 84 ttyS0 root 0 S /bin/sh 85 root 0 S /bin/inetd 86 root 0 S /bin/syslogd -n 87 root 0 S /bin/klogd -n 90 ttyS0 root 0 R ps I get the kernel log messages in the file /var/log/messages. I want to forward this kernel log to a remote machine,Then i exucute the follwing command />/bin/syslogd -R 192.168.5.4:514 & />ps PID TTY Uid Size State Command 1 root 0 S /sbin/init 2 root 0 S [keventd] 3 root 0 S [ksoftirqd_CPU0] 4 root 0 S [kswapd] 5 root 0 S [bdflush] 6 root 0 S [kupdated] 51 root 0 S /bin/dhcpd LAN 77 root 0 S chois -gserver 54273 80 root 0 S /bin/dnsmasq -r /etc/resolv.dnsmasq 84 ttyS0 root 0 S /bin/sh 85 root 0 S /bin/inetd 86 root 0 S /bin/syslogd -n 87 root 0 S /bin/klogd -n 89 ttyS0 root 0 S /bin/syslogd -R 192.168.5.4:514 93 ttyS0 root 0 R ps The machine 192.168.5.4 is reachable from my gateway(able to ping). The syslogd service on 192.168.5.4 is able to listen on remote machine.(Its /var/log/messages shows the log of 3-machines). But i am unable to get the Log of Gateway(192.168.5.1) on This machine(192.168.5.4). Sorry for my poor english, Please help me. With Regards Kanhu Kanhu Charan Rauta Innomedia Techonologies Pvt. Ltd #3278,12Th Main, HAL 2nd stage. Banglore-560008. -------------- next part -------------- An HTML attachment was scrubbed... URL: From ulf.soderberg at gmail.com Mon Mar 21 03:33:19 2005 From: ulf.soderberg at gmail.com (=?ISO-8859-1?Q?Ulf_S=F6derberg?=) Date: Mon, 21 Mar 2005 09:33:19 +0100 Subject: [uClinux-dev] USB Gadget Device support In-Reply-To: <423E14A5.7030306@snapgear.com> References: <423D549F.8050200@gmail.com> <423E14A5.7030306@snapgear.com> Message-ID: <423E86CF.2080800@gmail.com> Greg Ungerer wrote: > Hi Ulf, > > Ulf S?derberg wrote: > >> I was wondering if anyone has had any success in using the USB Gadget >> drivers on a Intel IXP4xx target running the 2.6.11 kernel? from the >> driver sources (pxa2xx_udh.c etc.) it would seem that the 2.6 version >> of the driver is a bit older than the 2.4 version. After a bit of >> Googling on the subject, I found evidence that someone really had it >> working with the 2.4 kernel, but I haven't found any comments on USB >> Gadget + IXP4xx + Linux 2.6. >> >> I'm using the Snapgear distribution. > > > I had it working the 2.4 kernels, on the IXP425. > But I haven't tried it on 2.6... > > Regards > Greg > We had to change the sources a bit to get it to compile without errors. It seems that the probe function of pxa2xx_udh.c is never called in 2.6 on the Intel IXDPG425 (MoneJade) bord with Snapgear. /Ulf -- Ulf S?derberg http://osgifun.blogspot.com From msiu at emsoftltd.com Tue Mar 22 21:19:49 2005 From: msiu at emsoftltd.com (Martin Siu) Date: Wed, 23 Mar 2005 10:19:49 +0800 Subject: [uClinux-dev] shared lib on ARM nommu Message-ID: <002101c52f4e$cae7ca70$6f01a8c0@4010cdt> Hi all, I'm working on code size reduction in order to save up precious rom space. Are there any update to shared lib on ARM nommu lately? I understand that the m68k toolchain already has this support, but not on arm. I'm seeing a few approaches to tackle code size problems: 1) uClibc (tiny libs) 2) busybox like applets 3) compressed flat format binaries I'm already employing the above techniques. Any other suggestions? Thanks. Martin Siu Software Engineer EmSoft Ltd msiu at emsoftltd.com From info at hostinthebox.net Wed Mar 23 17:33:51 2005 From: info at hostinthebox.net (dan) Date: Wed, 23 Mar 2005 15:33:51 -0700 Subject: [uClinux-dev] Searching for a suitable platform for uClinux Message-ID: <4241EECF.7020007@hostinthebox.net> Hello, all - Long time reader, second time poster. I have been on a quest for most of this week so far to find some sort of SBC that also contains a fingerprint scanner. This device should be capable of booting, hopefully, uClinux (any other linux would suffice), perhaps run on direct DC power, and be relatively small (4"x4" or smaller). The device should have 16M of memory, and ultimately, a CF reader/writer. It doesn't matter which architecture this device is, but it should be fan-less and good on power consumption. This is kindof a general question, and might not be *directly* appropriate for this list, but I know that some people on the list might have exposure to this kind of device. If it is found that the platform is suitable for our purposes, there will most definately be a large order for said device. Anyway, I appreciate the time, all, and hope to hear back from some of you very soon. I've learned a lot about embedded Linux implementations just by reading this list daily, and taking some of the things mentioned and using that to further my development. I'd like to take a moment to also thank the uClinux team for their hard work, as well as the many people actively developing on it, to make it strong. Thanks again -dant From gaoyinghao at c-real.com.cn Tue Mar 22 20:35:02 2005 From: gaoyinghao at c-real.com.cn (gaoyinghao at c-real.com.cn) Date: Wed, 23 Mar 2005 09:35:02 +0800 Subject: [uClinux-dev] Mail Delivery (failure uclinux-dev@uclinux.org) Message-ID: <20050323013706.AD1B71A567@uclinux.org> An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: message.scr Type: audio/x-wav Size: 29568 bytes Desc: not available URL: From jwilliams at itee.uq.edu.au Tue Mar 22 17:30:26 2005 From: jwilliams at itee.uq.edu.au (John Williams) Date: Wed, 23 Mar 2005 08:30:26 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050322162414.2455DE9AB@mail2.atmark-techno.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <20050315081337.0A198633A@mail2.atmark-techno.com> <20050319002457.GI11709@pengutronix.de> <423E0DDC.6080809@snapgear.com> <423E311A.3060208@itee.uq.edu.au> <423E5955.8010104@snapgear.com> <20050322162414.2455DE9AB@mail2.atmark-techno.com> Message-ID: <42409C82.5090701@itee.uq.edu.au> Hi folks, Yasushi SHOJI wrote: > At Mon, 21 Mar 2005 15:19:17 +1000, > Greg Ungerer wrote: > >>>A dist cut without glibc would reduce this more than a little. >>>Uncompressed glibc is 116meg. Are there many (any?) platforms >>>regularly using this now? It seems like uClibc is by far the dominant >>>player in NOMMU/uClinux-dist these days. >> >>No doubt uClibc is domoniant (no glibc choice for non-MMU). >>Show of hands... who here uses glibc on VM platforms in the >>uClinux-dist? > > We do. > >>>Perhaps something to consider in response to the download size argument. >> >>Even if I take it out, it is only a short term fix. The dist will >>keep growing in size, and within a short time it will be larger than >>it is now - even without glibc. > > > if we want the dist to be collection of source code, we can't avoid > getting bigger and bigger. > > # don't get me wrong, I'm not against the idea of collection of > # source. it's a feature. Dare I declare concensus here that - Paris Hilton videos notwithstanding - tarball size is a non-issue? :) If physically hosting/distributing the thing becomes a concern, maybe something like bittorrent could help out. Reading back over this long and interesting thread, I see a few themes: 1. Image problem. The uClinux-dist and uClinux brands are worth keeping, but we need a concerted effort to spread the message that uC != NOMMU. We can probably all have 2 cents to say about text on the uClinux.org site - Michael D is ready for contributions to this effect. 2. As a technical infrastructure, uClinux-dist is nice to work with. As DaveM said, at the end of the day it's just scripts and Makefiles. We can do with it what we like (and regularly do ;) 3. Flexibility, in terms of release timing for the dist could be better. We all recognise SnapGear's effort in maintaining the dist and kernel trees, and I don't think anyone would be asking them to do more than they already are :) 4. Generating and pushing upstream patches for apps integrated into the dist, should be the responsibility of the individual who does the integration. This is recognition that youthful enthusiasm by others should not result in more hard work for the dist maintainers! :) Recent moves towards using vanilla GNU configure scripts in the dist reduce the impact of this so much. 5. Assuming the point about tarball size is settled, the monolithic nature of the dist is not a bug, it's a feature :) This is a philosophical difference from e.g. PTXdist - each has its merits, and there's certainly room for both in this world. It's unlikely that hundreds (thousands?) of uClinux-dist users are prepared to re-tool onto PTXdist, and vice versa. 6. A test harness (initially build-time, but ultimately runtime perhaps via LTP) would be an awesome addition to uClinux-dist. 7. There is no 7. So, I hope I've recognised the various views expressed by various people in the debate, and not misrepresented too badly :) Some ideas on the way forward, for each of the points above: 1. We keep pushing. uClinux.org gets a revamp, and we push hard to break the uC==NOMMU misconception. 2. We continue with efforts to standardise things in the vendors heirarchy. This can be incremental. 3. How about two trees, and two dist archives? One is stable, exactly as it currently is. The other is a daily/weekly/monthly tarball generated directly from SnapGear's internal tree, where our patches currently sit between dist cuts. The tarring and publishing can be scripted, so no extra effort there. Patches on the dist go through the same level review as they do now. On this same cycle, there is a uClinux-dist-unstable CVS tree at cvs.uclinux.org, which tracks (or simply "is") the unstable tree. When it's time for a new "stable" dist-cut, everyone gets an opportunity to try SnapGear's test cut, which in theory should match the current unstable tree anyway. Once the dust settles, the new dist is released, and the unstable tree reverted back to mirror the new "stable" tree, and the cycle repeats 4. Nothing required 5. ditto 6. This will have to evolve over time, esp. once we see how the recently announced uClinux/LTP project goes. OK that's enough from me - go on, rip it to shreds! :) John From yashi at atmark-techno.com Tue Mar 22 11:24:14 2005 From: yashi at atmark-techno.com (Yasushi SHOJI) Date: Wed, 23 Mar 2005 01:24:14 +0900 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <423E5955.8010104@snapgear.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <20050315081337.0A198633A@mail2.atmark-techno.com> <20050319002457.GI11709@pengutronix.de> <423E0DDC.6080809@snapgear.com> <423E311A.3060208@itee.uq.edu.au> <423E5955.8010104@snapgear.com> Message-ID: <20050322162414.2455DE9AB@mail2.atmark-techno.com> At Mon, 21 Mar 2005 15:19:17 +1000, Greg Ungerer wrote: > > > A dist cut without glibc would reduce this more than a little. > > Uncompressed glibc is 116meg. Are there many (any?) platforms > > regularly using this now? It seems like uClibc is by far the dominant > > player in NOMMU/uClinux-dist these days. > > No doubt uClibc is domoniant (no glibc choice for non-MMU). > Show of hands... who here uses glibc on VM platforms in the > uClinux-dist? We do. > > Perhaps something to consider in response to the download size argument. > > Even if I take it out, it is only a short term fix. The dist will > keep growing in size, and within a short time it will be larger than > it is now - even without glibc. if we want the dist to be collection of source code, we can't avoid getting bigger and bigger. # don't get me wrong, I'm not against the idea of collection of # source. it's a feature. -- yashi From msiu at emsoftltd.com Tue Mar 22 04:23:20 2005 From: msiu at emsoftltd.com (Martin Siu) Date: Tue, 22 Mar 2005 17:23:20 +0800 Subject: [uClinux-dev] shared lib on ARM nommu Message-ID: <00d501c52ec0$caf77260$6f01a8c0@4010cdt> Hi all, I'm working on code size reduction in order to save up precious rom space. Are there any update to shared lib on ARM nommu lately? I understand that the m68k toolchain already has this support, but not on arm. I'm seeing a few approaches to tackle code size problems: 1) uClibc (tiny libs) 2) busybox like applets 3) compressed flat format binaries I'm already employing the above techniques. Any other suggestions? Thanks very much. Martin Siu Software Engineer EmSoft Ltd (852) 2802-0283 msiu at emsoftltd.com From info at hostinthebox.net Wed Mar 23 17:29:42 2005 From: info at hostinthebox.net (dan) Date: Wed, 23 Mar 2005 15:29:42 -0700 Subject: [uClinux-dev] Remote Logging Problem with syslogd In-Reply-To: <20050321112108.20717.qmail@webmail17.rediffmail.com> References: <20050321112108.20717.qmail@webmail17.rediffmail.com> Message-ID: <4241EDD6.8000102@hostinthebox.net> kanhu rauta wrote: > Hi all, > > I am using BusyBox v0.60.4 with uClinux. > I have selected the klogd and syslogd options of busybox on compilation. > When I load this image on my gateway,both process has started successfully. > />ps > > > PID TTY Uid Size State Command > > 1 root 0 S /sbin/init > > 2 root 0 S [keventd] > > 3 root 0 S [ksoftirqd_CPU0] > > 4 root 0 S [kswapd] > > 5 root 0 S [bdflush] > > 6 root 0 S [kupdated] > > 51 root 0 S /bin/dhcpd LAN > > 77 root 0 S chois -gserver 54273 > > 80 root 0 S /bin/dnsmasq -r /etc/resolv.dnsmasq > > 84 ttyS0 root 0 S /bin/sh > > 85 root 0 S /bin/inetd > > 86 root 0 S /bin/syslogd -n > > 87 root 0 S /bin/klogd -n > > 90 ttyS0 root 0 R ps > > I get the kernel log messages in the file /var/log/messages. > I want to forward this kernel log to a remote machine,Then i exucute > the follwing command > />/bin/syslogd -R 192.168.5.4:514 & > />ps > > > PID TTY Uid Size State Command > > 1 root 0 S /sbin/init > > 2 root 0 S [keventd] > > 3 root 0 S [ksoftirqd_CPU0] > > 4 root 0 S [kswapd] > > 5 root 0 S [bdflush] > > 6 root 0 S [kupdated] > > 51 root 0 S /bin/dhcpd LAN > > 77 root 0 S chois -gserver 54273 > > 80 root 0 S /bin/dnsmasq -r /etc/resolv.dnsmasq > > 84 ttyS0 root 0 S /bin/sh > > 85 root 0 S /bin/inetd > > 86 root 0 S /bin/syslogd -n > > 87 root 0 S /bin/klogd -n > > 89 ttyS0 root 0 S /bin/syslogd -R 192.168.5.4:514 > > 93 ttyS0 root 0 R ps > > > The machine 192.168.5.4 is reachable from my gateway(able to ping). > The syslogd service on 192.168.5.4 is able to listen on remote > machine.(Its /var/log/messages shows the log of 3-machines). > But i am unable to get the Log of Gateway(192.168.5.1) on This > machine(192.168.5.4). > > Sorry for my poor english, > Please help me. > > With Regards > Kanhu > I don't recall a -R as a valid syslogd argument, however, "-r" is a valid argument. On the 192.168.5.4 machine, do you have a valid host lists to show which machines are allowed to record data to that host? Can you connect to the 192.168.5.4 machine any other way, via telnet or ssh perhaps? I don't think this is a uClinux-specific question, but definately one that people might be interested in. Hope those ideas give you something to run with -dant From budenszki.peter at radiant.hu Thu Mar 24 07:49:20 2005 From: budenszki.peter at radiant.hu (Budenszki Peter) Date: Thu, 24 Mar 2005 13:49:20 +0100 Subject: [uClinux-dev] MB86H21x experience Message-ID: <002b01c5306f$e6c4fde0$5102a8c0@peti> Hi, do anyone has experience with MB86H21x Fujitsu SMARTMPEG decoder? Or any working uCLinux port? regards, Peter From lucy.wang at philips.com Thu Mar 24 04:39:25 2005 From: lucy.wang at philips.com (lucy.wang at philips.com) Date: Thu, 24 Mar 2005 17:39:25 +0800 Subject: [uClinux-dev] [PATCH]uClinux patch for Philips LPC22xx Message-ID: Hi Greg, I'v clean up the linux-2.6.5 patch: - which includes only my changes from Hyok's armnommu support. I modified the uClinux patch a little meanwhile: - The directory name is changed from vendors/Philips/LPC2294 to vendors/Philips/LPC22xx Could you help to merge it as well? Thanks, Lucy Greg Ungerer 2005-03-21 13:26 To: Lucy Wang/PEK/SC/PHILIPS at PHILIPS cc: uclinux-dev at uClinux.org Subject: Re: [uClinux-dev] [PATCH]uClinux patch for Philips LPC22xx Classification: Hi Lucy, lucy.wang at philips.com wrote: > Attached is the uClinux patch for Philips LPC22xx, including the > vendors/Philips/LPC22xx directory and a new config.in file. > A readme file is also here. > Together with the linux-2.6.5 patch, you can begin the first trial of > uClinux+LPC22xx development. > Note that the device drivers, e.g. Ethernet driver(although I've > finished the rtl8019as driver), are not here. You patch looks to include all of Hyok's patch and then your changes on top. Would that be right? Can you generate a patch with just the changes from Hyok's armnommu support and your changes? My working 2.6 tree has Hyok's armnommu changes already in it. Applying you patch will mean a bit of cleanup work on my part. If you just have a patch for your changes that will make it a lot easier to apply right now. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: uClinux-dist-lpc22xx.tar.gz Type: application/octet-stream Size: 7919 bytes Desc: not available URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: linux-2.6.5-lpc22xx.patch.gz Type: application/octet-stream Size: 34901 bytes Desc: not available URL: From jwilliams at itee.uq.edu.au Tue Mar 22 17:35:52 2005 From: jwilliams at itee.uq.edu.au (John Williams) Date: Wed, 23 Mar 2005 08:35:52 +1000 Subject: [uClinux-dev] Re: [microblaze-uclinux] AC_DIVERT_HELP ncurses error for mutt in uClinux In-Reply-To: <20050323000713.14035.qmail@mail.tenet.res.in> References: <20050323000713.14035.qmail@mail.tenet.res.in> Message-ID: <42409DC8.5020303@itee.uq.edu.au> Quoting Stephan Kasal: Hi, On Wed, Mar 23, 2005 at 05:37:13AM +0530, Hema K wrote: >> ./configure: line 4573: `AC_DIVERT_HELP(Build-Tools file configure may not continue string "AC_DIVERT_HELP". When I download ncurses-5.4 from ftp.gnu.org, I don't see "AC_DIVERT_HELP" in ncurses-5.4/configure. Perhaps you run a misconfigured autoconf? Regards, Stepan Kasal From devi at tenet.res.in Thu Mar 24 13:15:24 2005 From: devi at tenet.res.in (devi) Date: Thu, 24 Mar 2005 23:45:24 +0530 Subject: [uClinux-dev] console Message-ID: <20050324181524.17313.qmail@mail.tenet.res.in> hi, For the virtual console what are the device to be added in the vendors/Makefile.I added tty0,tty1,tty2 and tty3. Apart from this there is any such device to be added. Regards Devi From jwilliams at itee.uq.edu.au Wed Mar 23 17:16:14 2005 From: jwilliams at itee.uq.edu.au (John Williams) Date: Thu, 24 Mar 2005 08:16:14 +1000 Subject: [uClinux-dev] Re: [microblaze-uclinux] AC_DIVERT_HELP ncurses error for mutt in uClinux In-Reply-To: <42409DC8.5020303@itee.uq.edu.au> References: <20050323000713.14035.qmail@mail.tenet.res.in> <42409DC8.5020303@itee.uq.edu.au> Message-ID: <4241EAAE.2060104@itee.uq.edu.au> Another (bounced) reply from Stepan: Hello, I noticed one more problem: On Wed, Mar 23, 2005 at 05:37:13AM +0530, Hema K wrote: >> ./configure --target=mb- --host=i686-pc-linux-gnu >> for configuring ncurses for microblaze-uclinux. this is wrong. Don't use --target here. [1] If you build a package on i686 and the result will run on microblaze, you need something like: ./configure --host=mb- --build=i686-pc-linux-gnu but the build environment can probably be autodetected: ./configure --host=mb- For explanation, read the node: Info: (autoconf.info)Specifying Names Footnotes: [1] --target is used only when compiling compilers or such, eg. gcc or binutils. With "normal programs", --target has no effect. Stepan From mohuaidong at gmail.com Mon Mar 21 23:45:59 2005 From: mohuaidong at gmail.com (Huaidong Mo) Date: Tue, 22 Mar 2005 12:45:59 +0800 Subject: [uClinux-dev] help me, I dispaired of pppd Message-ID: <6df4decc05032120455064e878@mail.gmail.com> Hi: I have troubled by pppd for several days. I need you help. I have read through ppp-howto. I also think it not too difficult.But I never succeed. Both my kernel and pppd come form uClinux-dist-20040408.tar.gz .I have create several configuration files as document said,such as "ppp-on","ppp-on-dial"; My cdma modem connects with ttyS0. when I send "ATE0" to ttyS0 (set to non-echo mode),I get error message: LCP: timeout sending Config-Requests! when I send "ATE1" to ttyS0(set to local echo mode),I get error message: Serial line is looped back! what are these message mean?How can I make sure that pppd have dialed the phone number "#777"? I'm sure all hardware work well. From alexandru at quickweb.ro Mon Mar 21 04:22:38 2005 From: alexandru at quickweb.ro (Alexandru Cristescu) Date: Mon, 21 Mar 2005 11:22:38 +0200 Subject: [uClinux-dev] libusb on 2.4.27-uc0 Message-ID: <1111396958.3287.62.camel@alex> Hi, I don't know if this list is the right place to send this, if not, my apologies and please direct me to where it should go. I'm trying to use libusb on a platform running the 2.4.27-uc0 kernel. So I got the latest source code for the libusb library and compiled them using arm-elf-gcc 2.95.3 with the -mapcs-32 -mtune=arm9tdmi -0s -W -Wall flags and then compiled the resulting source code into my application. Am I doing something wrong until here? I ran into problems when running the code on the platform tho... It hangs during an IOCTL_USB_REAPURB in one of the libusb functions... Anyone had any experience using libusb with uCLinux? Any idea what could be wrong? thanks in advance From rootfs at gmail.com Tue Mar 22 06:35:39 2005 From: rootfs at gmail.com (Kartikeyan) Date: Tue, 22 Mar 2005 17:05:39 +0530 Subject: [uClinux-dev] xcopilot simulator In-Reply-To: <423FBC33.4070400@gmail.com> References: <423FBC33.4070400@gmail.com> Message-ID: <4240030B.3080406@gmail.com> Kartikeyan wrote: > Hello, > I recently downloaded latest uclinux distribution and xcopilot simulator > I am able to run uclinux on the simulator compiling > I get a sash shell > But i am not able to do telnet localhost 2000 # it seems nothing is > listeninng to 2000 > as suggested in > http://www.cyberguard.info/snapgear/tb20020807.html > by David McCullogh > > I have rh9 kernel versions 2.4.20-8 and 2.6.10 > Can any one guide me what to do further and play with uclinux > any urls to go further into simulator and learn etc > > Thankyou > SK > From bennyc at rtunet.com Wed Mar 23 17:22:33 2005 From: bennyc at rtunet.com (Benny Chen) Date: Thu, 24 Mar 2005 08:22:33 +1000 Subject: [uClinux-dev] Cannot clear interrupt in ISR for GPIOINTR line: Cirrus 9312 Message-ID: <200503232222.j2NMMgVo023490@serv0.rtunet.com> Hi there, I have a problem trying to clear the interrupt for the GPIOINTR interrupt source in my interrupt service routine. Hence, my interrupt handler was repeatedly executed. Below is my hardware setup and an example of the commands I tried to use to clear the interrupt in my handler. This has cause me a few sleepless nights, hopefully someone can help. Thanks, Benny HARDWARE ------------------- Cirrus 9312 (ARM9) running SnapGear Linux SET UP ------------ The setup is we have an external uart with its INT line connected to one of the EGPIO port A lines. This line has been configured to be input, positive edge triggered interrupt line for the interrupt source GPIOINTR. I have also configured the GPIOINTR interrupt source in the VIC controller to be normal IRQ , non vectored. INTERRUPT HANDLER ---------------------------------- static void my_handler(int irq, void *dev_id, struct pt_regs *regs) { cli(); printk(KERN_ERR, ?HANDLER TRIGGERRED\n?); /* Read the UART ISR Reg */ readb(port + O_ISR); /* Disable Uart Interrupt */ writeb(0x00, port + O_IER); /* Attempt to clear the GPIOINT interrupt */ writel(0x04,vic2_base + O_AEOI); /* Read the VIC2 interrupt status reg, hoping that it will clear the interrupt */ readl(vic2_base + O_VIC2INTSTAT); /* Write to the VIC2VECTADDR reg, hoping to also clear the interrupt */ writel(0, vic2_base + O_VIC2VECTADDR); sti(); } -- No virus found in this outgoing message. Checked by AVG Anti-Virus. Version: 7.0.308 / Virus Database: 266.8.0 - Release Date: 21/03/2005 -------------- next part -------------- An HTML attachment was scrubbed... URL: From lekha at rtimes.com Mon Mar 21 01:32:04 2005 From: lekha at rtimes.com (Lekha) Date: Mon, 21 Mar 2005 00:32:04 -0600 Subject: [uClinux-dev] Coldfire 5272:getting stuck at sti()"privilege violation" Message-ID: <200503210032.AA281411754@rtimes.com> Hai, Iam trying to port uClinux-dist-20020220(linux2.4.x) to coldfire 5272 board,When booting ,as soon as my program reach the sti() function( that enables interrupt), my uCLinux system crashes and prints at console the string "privilege violation" along with the dump of the Address and Data register. At start up in the crt0_ram .s,Status register(SR) is made 0x2700 move.w #0x2700, %sr /* No interrupts */ (supervisor mode / interrupts disabled).So at sti() ,it must be still in the supervisor mode ,so why "privilege violation" ? Help please. Lekha From CCarson at CBWAutomation.com Mon Mar 21 08:52:39 2005 From: CCarson at CBWAutomation.com (Carson, Corey) Date: Mon, 21 Mar 2005 06:52:39 -0700 Subject: [uClinux-dev] Reading from the serial port Message-ID: <203F609DF3BDD611843C000475A1967C06C8BF@mail.cbwautomation.com> I'm running a Dragonball VZ microcontroller which has the uClinux2.4.x OS. I'm logging onto the microcontroller via serial port 1 (/dev/ttyS0) from a Red Hat machine where I'm using minicom. I'm using the following code, which is executed on the microcontroller, to read and write onto the other microcontroller serial port 2 (/dev/ttyS1). I'm able to write out from the microcontroller through the serial port 2 (/dev/ttyS1) but I can't read incoming data from the microcontroller serial port 2 (/dev/ttyS1). When I check my running microcontroller processes, it shows agetty running on serial port 2 (/dev/ttyS1). I'm not sure if that last bit was worth mentioning. Anyway, why am I not able to read but I am able to write. By the way, I know I'm writing because I'm seeing the word hello on the machine connected to serial port 2 (/dev/ttyS1). It appears the read command in my main function times out. Please someone help me. Here's the code I'm executing on the microcontroller: /* ** File: rs232.c ** ** Description: ** Provides an RS-232 interface that is very similar to the CVI provided ** interface library */ #include #include #include #include #include #include /*#define DEBUG*/ /* the maximum number of ports we are willing to open */ #define MAX_PORTS 4 /* this array hold information about each port we have opened */ struct PortInfo { int busy; char name[32]; int handle; } ports[MAX_PORTS]; /* ** Function: OpenCom ** ** Description: ** Opens a serial port with default parameters ** ** Arguments: ** portNo - handle used for further access ** deviceName - the name of the device to open ** ** Returns: ** -1 on failure */ int OpenCom(int portNo, const char deviceName[]) { return OpenComConfig(portNo, deviceName, 19200, 1, 8, 1, 0, 0); } /* ** Function: OpenComConfig ** ** Description: ** Opens a serial port with the specified parameters ** ** Arguments: ** portNo - handle used for further access ** deviceName - the name of the device to open ** baudRate - rate to open (19200 for example) ** parity - 0 for no parity ** dataBits - 7 or 8 ** stopBits - 1 or 2 ** iqSize - ignored ** oqSize - ignored ** ** Returns: ** -1 on failure ** ** Limitations: ** baudRate, parity, stopBits, iqSize, and oqSize are ignored */ int OpenComConfig(int port, const char deviceName[], int baudRate, int parity, int dataBits, int stopBits, int iqSize, int oqSize) { struct termios newtio; ports[port].busy = 1; if ((ports[port].handle = open(deviceName, O_RDWR, 0666)) == -1) { perror("open"); assert(0); } /* set the port to raw I/O */ newtio.c_cflag = CS8 | CLOCAL | CREAD; newtio.c_iflag = IGNPAR; newtio.c_oflag = 0; newtio.c_lflag = 0; newtio.c_cc[VINTR] = 0; newtio.c_cc[VQUIT] = 0; newtio.c_cc[VERASE] = 0; newtio.c_cc[VKILL] = 0; newtio.c_cc[VEOF] = 4; newtio.c_cc[VTIME] = 0; newtio.c_cc[VMIN] = 1; newtio.c_cc[VSWTC] = 0; newtio.c_cc[VSTART] = 0; newtio.c_cc[VSTOP] = 0; newtio.c_cc[VSUSP] = 0; newtio.c_cc[VEOL] = 0; newtio.c_cc[VREPRINT] = 0; newtio.c_cc[VDISCARD] = 0; newtio.c_cc[VWERASE] = 0; newtio.c_cc[VLNEXT] = 0; newtio.c_cc[VEOL2] = 0; cfsetospeed(&newtio, B19200); cfsetispeed(&newtio, B19200); tcsetattr(ports[port].handle, TCSANOW, &newtio); return 0; } /* ** Function: CloseCom ** ** Description: ** Closes a previously opened port ** ** Arguments: ** The port handle to close ** ** Returns: ** -1 on failure */ int CloseCom(int portNo) { if (ports[portNo].busy) { close(ports[portNo].handle); ports[portNo].busy = 0; return 0; } else { return -1; } } /* ** Function: ComRd ** ** Description: ** Reads the specified number of bytes from the port. ** Returns when these bytes have been read, or timeout occurs. ** ** Arguments: ** portNo - the handle ** buf - where to store the data ** maxCnt - the maximum number of bytes to read ** ** Returns: ** The actual number of bytes read */ int ComRd(int portNo, char buf[], int maxCnt) { int actualRead = 0; fd_set rfds; struct timeval tv; int retval; if (!ports[portNo].busy) { assert(0); } /* camp on the port until data appears or 5 seconds have passed */ FD_ZERO(&rfds); FD_SET(ports[portNo].handle, &rfds); tv.tv_sec = 5; tv.tv_usec = 0; retval = select(16, &rfds, NULL, NULL, &tv); if (retval) { actualRead = read(ports[portNo].handle, buf, maxCnt); puts(actualRead); } #ifdef DEBUG { unsigned int i; for (i = 0; i < actualRead; ++i) { if ((buf[i] > 0x20) && (buf[i] < 0x7f)) { printf("<'%c'", buf[i]); } else { printf("<%02X", buf[i]); } } } fflush(stdout); #endif /* DEBUG */ return actualRead; } /* ** Function: ComWrt ** ** Description: ** Writes out the specified bytes to the port ** ** Arguments: ** portNo - the handle of the port ** buf - the bytes to write ** maxCnt - how many to write ** ** Returns: ** The actual number of bytes written */ int ComWrt(int portNo, const char *buf, int maxCnt) { int written; if (!ports[portNo].busy) { assert(0); } #ifdef DEBUG { int i; for (i = 0; i < maxCnt; ++i) { if ((buf[i] > 0x20) && (buf[i] < 0x7f)) { printf(">'%c'", buf[i]); } else { printf(">%02X", buf[i]); } } } fflush(stdout); #endif /* DEBUG */ written = write(ports[portNo].handle, buf, maxCnt); return written; } int main() { int BytesWritten = 0; char serialbuff[200] = {0}; OpenCom(1, "/dev/ttyS1"); // Write to serial port. BytesWritten = ComWrt(1, "hello", 5); // Read from serial port. ComRd(1, serialbuff, 1); puts(serialbuff); // Close serial port. CloseCom(1); puts(" \n"); puts("done.\n"); } -------------- next part -------------- An HTML attachment was scrubbed... URL: From yashi at atmark-techno.com Thu Mar 24 07:37:21 2005 From: yashi at atmark-techno.com (Yasushi SHOJI) Date: Thu, 24 Mar 2005 21:37:21 +0900 Subject: [uClinux-dev] distclean on clean Message-ID: <20050324123721.472FFEDC0@mail2.atmark-techno.com> Hi all, some applications (ie. busybox) in user/ have distclean target. would it be ok to modify the user/Makefile to run distclean in those app dirs on make clean at toplevel? would it be acceptable? regards, -- yashi From rdkehn at yahoo.com Tue Mar 22 17:47:22 2005 From: rdkehn at yahoo.com (Doug Kehn) Date: Tue, 22 Mar 2005 14:47:22 -0800 (PST) Subject: [uClinux-dev] Memory Leak (Bridge Module Related) In-Reply-To: 6667 Message-ID: <20050322224722.8361.qmail@web51904.mail.yahoo.com> --- Doug Kehn wrote: > Board: ADI-Coyote > Distro: uClinx-20041215 > ixp400AccessLibrary-1_4.zip > > snapgear-3.1.x-ixp400-v1.4.2-support.tar.gz > Kernel: Linux 2.4.27-uc1 > Toolchain: arm-linux-tools-2003-1127 > > > When I enable the bridge module memory is consumed > until exhausted. At which point the kernel Oops. > The > rate at which memory is consumed is proportional to > the network load. If I disable the bridge module > (and > just route) the leak is not present. > > In looking at slabinfo, the memory seems to be > allocated to the size-64 object. For example, just > before the Oops a cat of /proc/slabinfo yields: > > size-64 668854 668860 72 12620 12620 1 : > 668854 836391 12620 0 0 > > I noticed that there is a 2.6.x tracing patch for > the > slab allocator. Does such a patch exist for the > 2.4.x > kernel? Or does anyone know of any tools/techniques > for tracking slab related memory leaks? > The leak was caused by a bug in the IXP425 ethernet driver (ixp425_eth.c) when the bridge module is used. As it turns out Intel has fixed this bug. The patch is available as part of ixp400 Access Library v1.5 (ixp400AccessLibrary-1_5.zip). If you are using ixp400 Access Library v1.4, like I am, it's best to apply the Intel patch manually. Regards, ...doug __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com From mike at steroidmicros.com Mon Mar 21 11:38:35 2005 From: mike at steroidmicros.com (Mike Lavender) Date: Mon, 21 Mar 2005 08:38:35 -0800 Subject: [uClinux-dev] Sending email Mike Lavender Message-ID: <006501c52e34$6e19f860$6202a8c0@flanders> Hi Srinivas, Thank you for the suggestion. I have been able to use the "system" call from within uClinux. The problem with the mail program is that it gets the message body from the stdin. I am not sure how I can get a handle to the child process' stdin so I can feed it the body of the message. Thanks, Mike >>-----Original Message----- >>From: uclinux-dev-bounces at uclinux.org [mailto:uclinux-dev- >>bounces at uclinux.org] On Behalf Of srinivas naga vutukuri >>Sent: March 18, 2005 8:24 PM >>To: uclinux-dev at uclinux.org >>Subject: [uClinux-dev] Sending email Mike Lavender >> >>Hi Mike, >> >>just check, whether usage of system will work out... >>system ("mailprogramme name"); >> >>best regards, >>srinivas. >>_______________________________________________ >>uClinux-dev mailing list >>uClinux-dev at uclinux.org >>http://mailman.uclinux.org/mailman/listinfo/uclinux-dev >>This message was resent by uclinux-dev at uclinux.org From hyok.choi at samsung.com Tue Mar 22 03:43:07 2005 From: hyok.choi at samsung.com (Hyok S. Choi) Date: Tue, 22 Mar 2005 17:43:07 +0900 Subject: [uClinux-dev] armnommu support In-Reply-To: <20050319105724.GN11709@pengutronix.de> Message-ID: <0IDQ008RMW9YF9@mmp1.samsung.com> Robert, After ARM arch directory merging, the repository will be shared by the arm repository. regards, Hyok > -----Original Message----- > From: Robert Schwebel [mailto:robert at pengutronix.de] On > Behalf Of Robert Schwebel > Sent: Saturday, March 19, 2005 7:57 PM > To: Hyok S. Choi > Cc: uclinux-dev at uclinux.org > Subject: Re: [uClinux-dev] armnommu support > > Hyok, > > On Sat, Mar 19, 2005 at 01:02:20PM +0900, Hyok S. Choi wrote: > > The armnommu support is under development phase to fully > share the arm dir. > > It has taken some time because of some issue of armulator. > > > > Please wait a while, and it will be provided. :-) > > Do you develop on a publically available repository? SVN or > something like that? > > Robert > -- > Dipl.-Ing. Robert Schwebel | http://www.pengutronix.de > Pengutronix - Linux Solutions for Science and Industry > Handelsregister: Amtsgericht Hildesheim, HRA 2686 > Hannoversche Str. 2, 31134 Hildesheim, Germany > Phone: +49-5121-206917-0 | Fax: +49-5121-206917-9 > > > From bbaldwin at indyme.com Thu Mar 24 21:21:08 2005 From: bbaldwin at indyme.com (Barry Baldwin) Date: Thu, 24 Mar 2005 18:21:08 -0800 Subject: [uClinux-dev] Coldfire Remote Debugging problem Message-ID: Hi All, I've been trying to debug an application remotely and I'm not having much luck. I'm using a 5272 with the 2.4 kernel and use u-boot as the boot-loader. I start the gdbserver by typing "gdbserver :3000 /bin/appname" on the coldfire and then "m68k-bdm-elf-gdb user/appname.gdb" on my development system then type "target remote 10.0.0.182:3000" in gdb then set a breakpoint by typing "b main" in gdb then when I type "cont" I get the message "Software breakpoints not implemented for this target" I've also tried debugging it using the bdm cable. When I try this method I set the breakpoint and hit continue but it never hits the breakpoint. I"ve tried using hbreak to set a hardware breakpoint but this had the same results. Never hits the breakpoint. I must be missing something obvious, can someone point out the error of my ways. Thanks very much in advance, Barry -------------- next part -------------- An HTML attachment was scrubbed... URL: From johnson at syncom.com.tw Fri Mar 25 01:49:02 2005 From: johnson at syncom.com.tw (johnson) Date: Fri, 25 Mar 2005 06:49:02 +0000 Subject: [uClinux-dev] How to make the arm-elf-gdb source view more comfortable? Message-ID: <1111733342.29087.9.camel@johnson.syncom.com.tw> Hi all,I was working on armnommu s3c4510b and run kernel 2.6.9 I use the arm-elf-gcc 2.9.3 to compile my program arm-elf-gcc -Wl,-elf2flt -o test test.c and use the insight 6.1 to debug my ap,I use another version of gdbserver I can debug the program but the source code view is so bad , It looks like the source will fly everywhere when I use step or next and may be just skip my breakpoint ,and go through is it the problem of binary flat code property ?. Or I should add some compile option to tell the gcc or gdb I use arm-elf-objdump -S test.gdb and the view seems to be well and I debug the program under pc ,it's more smooth and easy to trace the source ,why? can someone give me some hints Thanks a lot Any note is appreciated From asghar at innomedia.soft.net Fri Mar 25 05:15:53 2005 From: asghar at innomedia.soft.net (Asghar) Date: Fri, 25 Mar 2005 15:45:53 +0530 Subject: [uClinux-dev] Default gateway Ip address Message-ID: <4243E4D9.4020004@innomedia.soft.net> Hi all, Is there any way to get default gateway IP address through C function. Wating for any suggestion With thanks Asghar From p.t.schmidt at imail.de Fri Mar 25 07:59:03 2005 From: p.t.schmidt at imail.de (Thomas Schmidt) Date: Fri, 25 Mar 2005 13:59:03 +0100 Subject: AW: [uClinux-dev] MB86H21x experience In-Reply-To: <002b01c5306f$e6c4fde0$5102a8c0@peti> Message-ID: Hello Peter, I have a running port with serial interface driver and IDE driver. If you have any questions: p.t.schmidt at eplus-online.de Thomas -----Urspr?ngliche Nachricht----- Von: uclinux-dev-bounces at uclinux.org [mailto:uclinux-dev-bounces at uclinux.org]Im Auftrag von Budenszki Peter Gesendet: Donnerstag, 24. M?rz 2005 13:49 An: uclinux-dev at uclinux.org Betreff: [uClinux-dev] MB86H21x experience Hi, do anyone has experience with MB86H21x Fujitsu SMARTMPEG decoder? Or any working uCLinux port? regards, Peter _______________________________________________ uClinux-dev mailing list uClinux-dev at uclinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev at uclinux.org From jive74 at tiscali.it Fri Mar 25 08:47:17 2005 From: jive74 at tiscali.it (Giovanni Casoli) Date: Fri, 25 Mar 2005 14:47:17 +0100 Subject: [uClinux-dev] Default gateway Ip address In-Reply-To: <4243E4D9.4020004@innomedia.soft.net> References: <4243E4D9.4020004@innomedia.soft.net> Message-ID: <42441665.9050205@tiscali.it> Asghar wrote: > Hi all, > > Is there any way to get default gateway IP address through C function. > > Wating for any suggestion > > > > With thanks > Asghar Hi Asghar, the route command looks into /proc/net/route. A way would be that of reading this file. Regards Giovanni Casoli From dsl105854a at post.cybercity.dk Fri Mar 25 09:09:17 2005 From: dsl105854a at post.cybercity.dk (dsl105854a at post.cybercity.dk) Date: Fri, 25 Mar 2005 15:09:17 +0100 Subject: [uClinux-dev] Remote Logging Problem with syslogd In-Reply-To: <20050321112108.20717.qmail@webmail17.rediffmail.com> References: <20050321112108.20717.qmail@webmail17.rediffmail.com> Message-ID: <200503251509.17950.dsl105854a@post.cybercity.dk> Hi I'm currently using the syslog daemon in BusyBox v1.00. That one works fine. I start it by typing: /bin/syslogd -R192.168.5.4 & Also I have written this /etc/syslog.conf # Sample syslogd configuration file to # messages to a remote host forward all. *.* @192.168.5.4 which means that all messages are forwarded to 192.168.5.4 good luck Sune On Monday 21 March 2005 12:21, kanhu rauta wrote: > Hi all, > > I am using BusyBox v0.60.4 with uClinux. > I have selected the klogd and syslogd options of busybox on compilation. > When I load this image on my gateway,both process has started successfully. > />ps > > > PID TTY Uid Size State Command > > 1 root 0 S /sbin/init > > 2 root 0 S [keventd] > > 3 root 0 S [ksoftirqd_CPU0] > > 4 root 0 S [kswapd] > > 5 root 0 S [bdflush] > > 6 root 0 S [kupdated] > > 51 root 0 S /bin/dhcpd LAN > > 77 root 0 S chois -gserver 54273 > > 80 root 0 S /bin/dnsmasq -r /etc/resolv.dnsmasq > > 84 ttyS0 root 0 S /bin/sh > > 85 root 0 S /bin/inetd > > 86 root 0 S /bin/syslogd -n > > 87 root 0 S /bin/klogd -n > > 90 ttyS0 root 0 R ps > > I get the kernel log messages in the file /var/log/messages. > I want to forward this kernel log to a remote machine,Then i exucute the > follwing command />/bin/syslogd -R 192.168.5.4:514 & > />ps > > > PID TTY Uid Size State Command > > 1 root 0 S /sbin/init > > 2 root 0 S [keventd] > > 3 root 0 S [ksoftirqd_CPU0] > > 4 root 0 S [kswapd] > > 5 root 0 S [bdflush] > > 6 root 0 S [kupdated] > > 51 root 0 S /bin/dhcpd LAN > > 77 root 0 S chois -gserver 54273 > > 80 root 0 S /bin/dnsmasq -r /etc/resolv.dnsmasq > > 84 ttyS0 root 0 S /bin/sh > > 85 root 0 S /bin/inetd > > 86 root 0 S /bin/syslogd -n > > 87 root 0 S /bin/klogd -n > > 89 ttyS0 root 0 S /bin/syslogd -R 192.168.5.4:514 > > 93 ttyS0 root 0 R ps > > The machine 192.168.5.4 is reachable from my gateway(able to ping). > The syslogd service on 192.168.5.4 is able to listen on remote machine.(Its > /var/log/messages shows the log of 3-machines). But i am unable to get the > Log of Gateway(192.168.5.1) on This machine(192.168.5.4). > > Sorry for my poor english, > Please help me. > > With Regards > Kanhu From hyok.choi at samsung.com Fri Mar 25 19:47:58 2005 From: hyok.choi at samsung.com (Hyok S. Choi) Date: Sat, 26 Mar 2005 09:47:58 +0900 Subject: [uClinux-dev] [PATCH]uClinux patch for Philips LPC22xx In-Reply-To: Message-ID: <0IDX007FWOY2BX@mmp2.samsung.com> Hi Lucy, Okay, I'll look on the patch, but it could take some time to put in recent patch series. (like 2.6.10-hsc0 or 2.6.11-hsc0) I'd be happy if the patch could be made against 2.6.10-hsc0 if you have some time to look in. As I mentioned in recent patch announcement, from the 2.6.10-hsc0, the patch is merged into arm arch. Please feel free if I can help you with the patch releases. best regards, hyok HYOK S. CHOI Digital Media R&D Center, Samsung Electronics Co.,Ltd. tel: +82-31-200-8594/5723 fax: +82-31-200-3427 e-mail: hyok.choi at samsung.com [Linux 2.6 ARM MPU/noMMU Kernel Maintainer] ________________________________ From: uclinux-dev-bounces at uclinux.org [mailto:uclinux-dev-bounces at uclinux.org] On Behalf Of lucy.wang at philips.com Sent: Thursday, March 24, 2005 6:39 PM To: Greg Ungerer Cc: uclinux-dev at uClinux.org Subject: Re: [uClinux-dev] [PATCH]uClinux patch for Philips LPC22xx Hi Greg, I'v clean up the linux-2.6.5 patch: - which includes only my changes from Hyok's armnommu support. I modified the uClinux patch a little meanwhile: - The directory name is changed from vendors/Philips/LPC2294 to vendors/Philips/LPC22xx Could you help to merge it as well? Thanks, Lucy Greg Ungerer 2005-03-21 13:26 To: Lucy Wang/PEK/SC/PHILIPS at PHILIPS cc: uclinux-dev at uClinux.org Subject: Re: [uClinux-dev] [PATCH]uClinux patch for Philips LPC22xx Classification: Hi Lucy, lucy.wang at philips.com wrote: > Attached is the uClinux patch for Philips LPC22xx, including the > vendors/Philips/LPC22xx directory and a new config.in file. > A readme file is also here. > Together with the linux-2.6.5 patch, you can begin the first trial of > uClinux+LPC22xx development. > Note that the device drivers, e.g. Ethernet driver(although I've > finished the rtl8019as driver), are not here. You patch looks to include all of Hyok's patch and then your changes on top. Would that be right? Can you generate a patch with just the changes from Hyok's armnommu support and your changes? My working 2.6 tree has Hyok's armnommu changes already in it. Applying you patch will mean a bit of cleanup work on my part. If you just have a patch for your changes that will make it a lot easier to apply right now. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From hansgnu at yahoo.com Sun Mar 27 22:43:28 2005 From: hansgnu at yahoo.com (Joao Antonio P Rabelo HansGNU) Date: Sun, 27 Mar 2005 19:43:28 -0800 (PST) Subject: [uClinux-dev] uClinux In-Reply-To: 6667 Message-ID: <20050328034329.26997.qmail@web53802.mail.yahoo.com> Everbody, I have a WAP54G that one friend tried to change the firmeware, but he had problems. The WAP54G problems are I can not access the web control of the wap and I just can ping at 192.168.1.230. My wriless access point is a WAP54G v1.1. Any one know how can I do to port uclinux (firmware) again to this wap? I have another way to solve this problem? thanks, Joao Antonio .''`. Jo?o Ant?nio de Paiva Rabelo ::' : IRC - irc.freenode.net/6667 - #Debian-BR #MinaSLivre `.`'` UIN - 61461402 JABBER - hansgnu at jabber.org `- Nunca deixe de perguntar e use DEBIAN !!! __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/ From devi at tenet.res.in Mon Mar 28 12:21:35 2005 From: devi at tenet.res.in (devi) Date: Mon, 28 Mar 2005 22:51:35 +0530 Subject: [uClinux-dev] console Message-ID: <20050328172135.5785.qmail@mail.tenet.res.in> hi, The console uses devfs to register the driver. Why is that the console uses devfs to register the driver. Regards Devi From kunlunzhu at yahoo.com Mon Mar 28 03:52:11 2005 From: kunlunzhu at yahoo.com (kunlunzhu at yahoo.com) Date: Mon, 28 Mar 2005 00:52:11 -0800 (PST) Subject: [uClinux-dev] How to have a interrupt higher than kernel? In-Reply-To: 6667 Message-ID: <20050328085211.40068.qmail@web81404.mail.yahoo.com> Hi, All: When I use 115kbps baud rate with MCF5249 and boa server with uClinux, I encountered fifo overrun problem and lose receive data. I browser uClinux group and saw some one already pointed out the lagency problem because the interrupt is holdding too long by kernel. How difficult is it to add my own uart interrupt routine working outside of linux to handle all possible received UART rx interrupt? Could any one give me some hints to do so? Thank in advance. Kevin Zhu From eauth at softsys.co.at Mon Mar 28 03:56:35 2005 From: eauth at softsys.co.at (Erwin Authried) Date: Mon, 28 Mar 2005 10:56:35 +0200 Subject: [uClinux-dev] uClinux In-Reply-To: <20050328034329.26997.qmail@web53802.mail.yahoo.com> References: <20050328034329.26997.qmail@web53802.mail.yahoo.com> Message-ID: <1112000195.23484.10.camel@kabel> On Mon, 2005-03-28 at 05:43, Joao Antonio P Rabelo HansGNU wrote: > Everbody, > > I have a WAP54G that one friend tried to change the > firmeware, but he had problems. The WAP54G problems > are I can not access the web control of the wap and I > just can ping at 192.168.1.230. My wriless access > point is a WAP54G v1.1. Any one know how can I do to > port uclinux (firmware) again to this wap? I have > another way to solve this problem? > > thanks, > > Joao Antonio Hi, please google for wap54g + linux. There are at least 5 different firmware source distributions, some of them come with detailed instructions for setting up the mips toolchain and compiling. There are instructions how a "bricked" wap54g can be recovered as well. Regards, Erwin From srinivas.vutukuri at gmail.com Mon Mar 28 05:02:12 2005 From: srinivas.vutukuri at gmail.com (srinivas naga vutukuri) Date: Mon, 28 Mar 2005 15:32:12 +0530 Subject: [uClinux-dev] console Message-ID: Hi, Console is a tty device, and corresponding device file exists as /dev/console. So it needs to be registered with device file system routine. best regards, srinivas. From rootfs at gmail.com Mon Mar 28 06:25:06 2005 From: rootfs at gmail.com (sunil s) Date: Mon, 28 Mar 2005 16:55:06 +0530 Subject: [uClinux-dev] test mail Message-ID: <185c1c600503280325c511041@mail.gmail.com> I am not able to send mail to uclinux-dev though subscribed. Testing again SK From yogananda.ag at gmail.com Mon Mar 28 08:08:39 2005 From: yogananda.ag at gmail.com (Yogananda G) Date: Mon, 28 Mar 2005 18:38:39 +0530 Subject: [uClinux-dev] Question on uclinux memory allocation problem In-Reply-To: References: Message-ID: <34c1e45d0503280508251b716a@mail.gmail.com> Hi Edward, It seems that lot of your memory is been used up by the kernel and ramdisk. Hence you dont have enough memory to run your code on the board. Check for the free space availabe immediately after the bootup. Check for your stack size also. char buffer[1024][1024] and many such variables could be exceeding your stack limit. regards, yogs On Mon, 21 Mar 2005 19:03:19 +0800, Edward Chan wrote: > Dear all, > > I have bought a Samsung ARM7TDMI development board with 2MB NOR > flash , 8 MB SDRAM, 16 MB Flash. Currently, I want to enable my C > program to be run on the uClinux , but encounter insuffient memory > error. > > In desktop, my program only occupy around 1MB, but under uclinux, it > reported that my program required up to 8MB memory. I have already > enabled > Non power-of-2 kernel allocator (EXPERIMENTAL) > in my uclinux kernel 2.4.12 and also compile with the kernel with XIP feature. > > I use the arm-elf-gcc with the following command : > arm-elf-gcc -Wl, -elf2flt -O3 -D__PIC__ -fpic -msingle-pic-base -Wall > -o test test.c > > and my program source only allocate memory using static array, e.g. > char buffer[1024][1024]... > > So I don't know why my application will occupy so much memory in the > uClinux platform. Would anyone help me to figure out the problem? > Thanks for all of your kindess help! > > -- > A newbie in uClinux, > Edward Chan > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uclinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org > From yashi at atmark-techno.com Mon Mar 28 10:12:38 2005 From: yashi at atmark-techno.com (Yasushi SHOJI) Date: Tue, 29 Mar 2005 00:12:38 +0900 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <42409C82.5090701@itee.uq.edu.au> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <20050315081337.0A198633A@mail2.atmark-techno.com> <20050319002457.GI11709@pengutronix.de> <423E0DDC.6080809@snapgear.com> <423E311A.3060208@itee.uq.edu.au> <423E5955.8010104@snapgear.com> <20050322162414.2455DE9AB@mail2.atmark-techno.com> <42409C82.5090701@itee.uq.edu.au> Message-ID: <20050328151238.E1C46ED99@mail2.atmark-techno.com> Hi all, I'm back to Japan now. let's discuss little bit about this. At Wed, 23 Mar 2005 08:30:26 +1000, John Williams wrote: > > 1. Image problem. The uClinux-dist and uClinux brands are worth > keeping, but we need a concerted effort to spread the message that uC != > NOMMU. We can probably all have 2 cents to say about text on the > uClinux.org site - Michael D is ready for contributions to this effect. As Greg is mentioning, Greg and I are oppose to this idea. Specially if we keep !MMU kernel as uClinux. (Yes, I don't want to dump the brand of uClinux kernel) how long do we need and how hard should we put our effort break that association only for the dist? > 2. As a technical infrastructure, uClinux-dist is nice to work with. As > DaveM said, at the end of the day it's just scripts and Makefiles. We > can do with it what we like (and regularly do ;) Ok I agree. we'll do some work when we need. I don't think it's time yet. (of course, if anyone is willing to put some effort on this, I'm happy to see what comes up) > 3. Flexibility, in terms of release timing for the dist could be better. > We all recognise SnapGear's effort in maintaining the dist and kernel > trees, and I don't think anyone would be asking them to do more than > they already are :) I personally want to put some effort on this. would it be nice to have intermediate patch-sets between the official dist releases? What I'm thinking is that whenever a patch comes in to this ml and at least one of the dist maintainer says "applied", we keep that patch in a patch-set, and release it once in a while. would that be enough? in this way I don't think we put to much (if any) pressure to the official dist maintainer. once official (or -test, for that matter) is released, we sync it with it. since we keep most (if not all) of patches blessed by the maintainer, it should be easier to sync with the official. (ain't we? ;) > 5. Assuming the point about tarball size is settled, the monolithic > nature of the dist is not a bug, it's a feature :) This is a > philosophical difference from e.g. PTXdist - each has its merits, and > there's certainly room for both in this world. It's unlikely that > hundreds (thousands?) of uClinux-dist users are prepared to re-tool onto > PTXdist, and vice versa. so, do we all agree on this? I do. Let's think this issue again when the dist size exceeds a common cheap media, like a CD-RW stands for nowadays. > 6. A test harness (initially build-time, but ultimately runtime perhaps > via LTP) would be an awesome addition to uClinux-dist. I'm very much interested in this, but haven't had time to look close. regards, -- yashi From julien at alphanet.ch Mon Mar 28 17:10:23 2005 From: julien at alphanet.ch (Julien Linder) Date: Tue, 29 Mar 2005 00:10:23 +0200 Subject: [uClinux-dev] genromfs - giving dev access to anybody Message-ID: <424880CF.3090103@alphanet.ch> Hi all. I need free access to my /dev/ttyz5 device but I can't figure out how to do. Genromfs makes all devices rw for root only. Is there any additionnal parameter that can be given in the filename @ttyz5,c,3,22 ? Julien From bob at softmach.com Mon Mar 28 18:24:11 2005 From: bob at softmach.com (Bob Rubendunst) Date: Mon, 28 Mar 2005 17:24:11 -0600 Subject: [uClinux-dev] Re: uClinux-dev Digest, Vol 23, Issue 25 In-Reply-To: <20050325170008.6F0271A1BB@uclinux.org> References: <20050325170008.6F0271A1BB@uclinux.org> Message-ID: <6.1.2.0.0.20050328171750.043aae00@127.0.0.1> Corey, The agetty is reading the data before you can. Think of serial port input like joint bank checking - with all of it's problems if the check writers aren't cooperating. If you stop the agetty from running, you will probably get better results, but I can't say I looked over your code, either. At 11:00 AM 3/25/2005, you wrote: >Message: 22 >Date: Mon, 21 Mar 2005 06:52:39 -0700 >From: "Carson, Corey" >Subject: [uClinux-dev] Reading from the serial port >To: uclinux-dev at uclinux.org >Message-ID: > <203F609DF3BDD611843C000475A1967C06C8BF at mail.cbwautomation.com> >Content-Type: text/plain; charset="us-ascii" > >I'm running a Dragonball VZ microcontroller which has the uClinux2.4.x OS. >I'm logging onto the microcontroller via serial port 1 (/dev/ttyS0) from a >Red Hat machine where I'm using minicom. I'm using the following code, >which is executed on the microcontroller, to read and write onto the other >microcontroller serial port 2 (/dev/ttyS1). I'm able to write out from the >microcontroller through the serial port 2 (/dev/ttyS1) but I can't read >incoming data from the microcontroller serial port 2 (/dev/ttyS1). When I >check my running microcontroller processes, it shows agetty running on >serial port 2 (/dev/ttyS1). I'm not sure if that last bit was worth >mentioning. Anyway, why am I not able to read but I am able to write. By >the way, I know I'm writing because I'm seeing the word hello on the machine >connected to serial port 2 (/dev/ttyS1). > > > >It appears the read command in my main function times out. Please someone >help me. -Bob Rubendunst Soft Machines http://www.softmach.com "Everything serial from start to stop" From mike at steroidmicros.com Mon Mar 28 20:46:18 2005 From: mike at steroidmicros.com (Mike Lavender) Date: Mon, 28 Mar 2005 17:46:18 -0800 Subject: [uClinux-dev] 2.4 qspi driver modifications In-Reply-To: <200502091013.41500.ngustavson@emacinc.com> Message-ID: <000601c53401$1cc6f620$6202a8c0@flanders> Hello, I also need to stack an MMC driver on top of the QSPI. Do you have a copy of your driver that I can test on our hardware? Thanks, Mike Lavender >>-----Original Message----- >>From: uclinux-dev-bounces at uclinux.org [mailto:uclinux-dev- >>bounces at uclinux.org] On Behalf Of NZG >>Sent: February 9, 2005 8:14 AM >>To: uClinux development list >>Subject: Re: [uClinux-dev] 2.4 qspi driver modifications >> >>Please allow me to review the patch this morning, if you don't mind, >>before >>you commit it. >>I need to stack my qspi MMC driver(currently in development) on top of >>this >>driver and need to make sure that there is provision for it. >>If it does, then I'll say lets go for it. >>If not I will provide a recommended patch this afternoon. >> >>> Is there are consensus that these changes should go into the uClinux >>kernel ? >>> Anyone have any issues with it before I check it in :-) >> >>NZG. >>_______________________________________________ >>uClinux-dev mailing list >>uClinux-dev at uclinux.org >>http://mailman.uclinux.org/mailman/listinfo/uclinux-dev >>This message was resent by uclinux-dev at uclinux.org From lars.friedrich at wago.com Tue Mar 29 01:42:30 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Tue, 29 Mar 2005 08:42:30 +0200 Subject: [uClinux-dev] zombies after terminating threads Message-ID: > In my application I create 2 threads but after they > terminated the join function returns "0". After checking the > process status the threads appear as Zombies. > > Any Idea Why? > I would be gratefull for any help? This is most likely either a uClibc bug or the simple pthread library without proper pthread_join support (which again is nearly the same as the first case). If you get zombies, too, when you pthread_detach the threads, blame uClibc^1. If this should work though, you can emulate the pthread_join functionality with pthread_detach and pthread_cond_wait. Best regards, Lars Friedrich ^1 It's always helpful to also post here things like which uClinux-dist, which platform, which user space library and which version of the library is used. This eliminates some guessing, like if you are actually using uClibc. From lars.friedrich at wago.com Tue Mar 29 01:47:42 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Tue, 29 Mar 2005 08:47:42 +0200 Subject: [uClinux-dev] How to make the arm-elf-gdb source view morecomfortable? Message-ID: > Hi all,I was working on armnommu s3c4510b and run kernel 2.6.9 > I use the arm-elf-gcc 2.9.3 to compile my program > arm-elf-gcc -Wl,-elf2flt -o test test.c This is really your command line to compile the application? > I can debug the program but the source code view is so bad , > It looks like the source will fly everywhere when I use step or > next This is normal behaviour for compiler optimized code. According to the command line above no compiler optimized code should be generated though. Best regards, Lars Friedrich From devi at tenet.res.in Tue Mar 29 13:04:14 2005 From: devi at tenet.res.in (devi) Date: Tue, 29 Mar 2005 23:34:14 +0530 Subject: [uClinux-dev] keyboard Message-ID: <20050329180414.12926.qmail@mail.tenet.res.in> hi, When I access the keyboard which is connected to the board I am not able to acces the remote keyboard(which is connected to the PC).Any idea. Regards Devi From kadionik at enseirb.fr Tue Mar 29 03:46:06 2005 From: kadionik at enseirb.fr (Patrice Kadionik) Date: Tue, 29 Mar 2005 10:46:06 +0200 Subject: [uClinux-dev] Call For Presentation: LSM '05 conference on free software for embedded systems Message-ID: <424915CE.80600@enseirb.fr> Dear all, The LSM/RMLL '05 meeting (Libre Software Meeting - Rencontres Mondiales du Logiciel Libre 2005) is organized at Dijon, FRANCE during the 5th to 9th july 2005 at the University of Bourgogne. Like last year, a special Embedded Session is schedulded on free software (toolchains, embedded Linux, Real Time Linux...) for Embedded Systems. Like last year, you are welcome to participate. You'll find informations here http://rencontresmondiales.org/ and especially here: http://www.enseirb.fr/~kadionik/rmll2005/rmll2005.html A preliminary program has to be performed in 2 weeks. If you are interested to participate, you may simply email me the title of your presentation as soon as possible... Sincerely Yours; Pat. -- Patrice Kadionik. F6KQH / F4CUQ ----------- +---------------------------------------------------------------------+ +"Tout doit etre aussi simple que possible, pas seulement plus simple"+ +---------------------------------------------------------------------+ + Patrice Kadionik http://www.enseirb.fr/~kadionik + + ENSEIRB fax : +33 5.56.37.20.23 + + PO BOX 99 voice : +33 5.56.84.23.47 + + 33402 TALENCE Cedex http://www.enseirb.fr + + FRANCE mailto:kadionik at enseirb.fr + +---------------------------------------------------------------------+ From jmh at csm-instruments.com Tue Mar 29 04:34:29 2005 From: jmh at csm-instruments.com (Julien MARBACH) Date: Tue, 29 Mar 2005 11:34:29 +0200 Subject: [uClinux-dev] LUA (a small script interpreter) compilation in uClinux Message-ID: <42492125.40103@csm-instruments.com> I'm quite new to uClinux and i'm triying to add my first user-app to uClinux... Has anybody tried to compile lua (www.lua.org, a rather small script interpreter) in uClinux? I'm wondering what changes I have to make in the lua makefile to add it in the uClinux building process (I've alredy added the necessary entry in user/Makefile, config/config.in & config/Configure.help). here is a copy of the lua makefile : *** # makefile for Lua hierarchy # see INSTALL for installation instructions # see config for customization instructions LUA= . include $(LUA)/config # primary targets ("co" and "klean" are used for making the distribution) all clean co klean: dirs cd include; $(MAKE) $@ cd src; $(MAKE) $@ cd src/lib; $(MAKE) $@ cd src/luac; $(MAKE) $@ cd src/lua; $(MAKE) $@ # in case they were not created during unpacking dirs: bin lib bin lib: mkdir -p $@ # simple test to see Lua working test: all bin/lua test/hello.lua # remove debug information from binaries strip: $(STRIP) bin/* # official installation install: all strip mkdir -p $(INSTALL_BIN) $(INSTALL_INC) $(INSTALL_LIB) $(INSTALL_MAN) $(INSTALL_EXEC) bin/* $(INSTALL_BIN) $(INSTALL_DATA) include/*.h $(INSTALL_INC) $(INSTALL_DATA) lib/*.a $(INSTALL_LIB) $(INSTALL_DATA) doc/*.1 $(INSTALL_MAN) # shared libraries (for Linux) so: ld -o lib/liblua.so.$V -shared src/*.o ld -o lib/liblualib.so.$V -shared src/lib/*.o cd lib; ln -fs liblua.so.$V liblua.so; ln -fs liblualib.so.$V liblualib.so # binaries using shared libraries sobin: rm -f bin/* cd src/lua; $(MAKE) cd src/luac; $(MAKE) # install shared libraries soinstall: $(INSTALL_EXEC) lib/*.so.* $(INSTALL_LIB) cd $(INSTALL_LIB); ln -fs liblua.so.$V liblua.so; ln -fs liblualib.so.$V liblualib.so # clean shared libraries soclean: rm -f lib/*.so* bin/* # turn config into Lua code # uncomment the last sed expression if you want nil instead of empty strings lecho: @echo "-- $(LUA)/config for Lua $V" @echo "VERSION = '$(V)'" @make echo | grep = | sed -e 's/= /= "/' -e 's/$$/"/' #-e 's/""/nil/' @echo "-- EOF" newer: @find . -newer MANIFEST -type f # (end of Makefile) *** I believe that the "install" rule have to be changed to something like : romfs: $(ROMFSINST) bin/* $(INSTALL_BIN) include/*.h $(INSTALL_INC) lib/*.a $(INSTALL_LIB) WHat about the "soinstall" rule? Another question : do I have to expect that the build process will produce a large executable file/library? Any help would be most welcome. Salutations, Julien MARBACH From gerg at snapgear.com Tue Mar 29 09:16:26 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Wed, 30 Mar 2005 00:16:26 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <1111389078.11799.21.camel@kabel> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> <20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> <423951E6.5040607@itee.uq.edu.au> <20050319003721.GJ11709@pengutronix.de> <423E1028.8010608@snapgear.com> <1111389078.11799.21.camel@kabel> Message-ID: <4249633A.6040408@snapgear.com> Hi Erwin, Erwin Authried wrote: > On Mon, 2005-03-21 at 01:07, Greg Ungerer wrote: > ... >>Not to trivialize the point, but it is not difficult to get the >>source for any package and diff it against what is in the uClinux-dist. >>There is your patch (admittedly it doesn't have any user context - >>why the changes where made for example). >> >>But how often do you really reveiew the patches? >>For any large number of out-standing packages this becomes quite >>a difficult thing to do. > > what I'm missing in uClinux-dist is an easy way to track the changes > that have been made to a source package. That's important when you want > to switch to a later version of some package. At least for some of the > packages, it is quite difficult to fetch the original source, either > because there isn't any version information or the package isn't > available on the internet anymore. Good point. >>>Hmmm, if you have real customer projects you at least need some >>>stability points, and you need control over it. Having a RCS system is >>>cool (we have svn for PTXdist), but that doesn't mean that lots of >>>people should get write access. Building systems is a complex and >>>fragile thing and should be supervised by a maintainer who knows what he >>>does. >> >>How do you handle the fact that the internet is a little more dynamic >>though? Web sites come and go, source packages locations change or >>disappear over time? > > I have seen this to be a problem too for my Midori uClinux > distirbution. This can be handled with a "fallback" archive of original > sources. When the package cannot be fetched from the original place, it > is automatically fetched from this location. Ok, that sounds a good way to deal with it. Doesn't that mean though that you end up having to host all the packages that the distribution contains anyway? Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From eauth at softsys.co.at Tue Mar 29 11:16:59 2005 From: eauth at softsys.co.at (Erwin Authried) Date: Tue, 29 Mar 2005 18:16:59 +0200 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <4249633A.6040408@snapgear.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <4236C437.6060404@snapgear.com> <20050316035306.B2F226373@mail2.atmark-techno.com> <42392BD4.3030201@snapgear.com> <423951E6.5040607@itee.uq.edu.au> <20050319003721.GJ11709@pengutronix.de> <423E1028.8010608@snapgear.com> <1111389078.11799.21.camel@kabel> <4249633A.6040408@snapgear.com> Message-ID: <1112113019.2314.49.camel@justakiss> Am Die, den 29.03.2005 schrieb Greg Ungerer um 16:16: > Hi Erwin, > > Erwin Authried wrote: > > > > what I'm missing in uClinux-dist is an easy way to track the changes > > that have been made to a source package. That's important when you want > > to switch to a later version of some package. At least for some of the > > packages, it is quite difficult to fetch the original source, either > > because there isn't any version information or the package isn't > > available on the internet anymore. > > Good point. > > > >>>Hmmm, if you have real customer projects you at least need some > >>>stability points, and you need control over it. Having a RCS system is > >>>cool (we have svn for PTXdist), but that doesn't mean that lots of > >>>people should get write access. Building systems is a complex and > >>>fragile thing and should be supervised by a maintainer who knows what he > >>>does. > >> > >>How do you handle the fact that the internet is a little more dynamic > >>though? Web sites come and go, source packages locations change or > >>disappear over time? > > > > I have seen this to be a problem too for my Midori uClinux > > distirbution. This can be handled with a "fallback" archive of original > > sources. When the package cannot be fetched from the original place, it > > is automatically fetched from this location. > > Ok, that sounds a good way to deal with it. Doesn't that mean though > that you end up having to host all the packages that the distribution > contains anyway? Hello Greg, most of the time, only the location of the package changes. In this case, I'm doing an update to the package's build script to reflect the new location. Currently, only very few packages disappeared totally. And then, it's time to consider if the package doesn't require an update alltogether, because it may be rather outdated. If you see it this way, it's a automatic reminder from the build system to keep pace ;-) As you have already mentioned some time ago, there's no perfect distribution. There's no doubt that uClinux-dist is excellent and contains the largest number of packages, but I think we should keep an eye on other distributions like PTXdist, celinux, ETlinux, emdebian, Midori uClinux, ... (to name just a few) for inspiration to make a new distribution that combines the features of all of them. Regards, Erwin From jhml at gmx.net Tue Mar 29 15:53:56 2005 From: jhml at gmx.net (Johann Hanne) Date: Tue, 29 Mar 2005 22:53:56 +0200 Subject: [uClinux-dev] [patch] update cx821xx architecture Message-ID: <200503292254.09474.jhml@gmx.net> Hi, after searching for the reason of the ethernet tx timeouts I got on my DPCM for a while, I finally recognized that the uclinux tree within ftp://128.121.226.214/pub/dualpcmodem/actiontec_release_031017.tar.gz contains newer file for the cx821xx architecture than the current uClinux CVS. So I diffed the CVS tree against it and the results look ok, it only changes a few things, all within the arch directory: - major updates to the timer routines (these probably solve my ethernet problems) - lots of comment changes (version strings, updates of the year numbers) - DSL updates I've tested the patch against current CVS and it works, so it would be wonderful if you could apply it, Greg. :-) Thanks, Johann Changed Files: arch/armnommu/mach-cx821xx/LnxTools.c arch/armnommu/mach-cx821xx/Makefile arch/armnommu/mach-cx821xx/arch.c arch/armnommu/mach-cx821xx/cacheenb.S arch/armnommu/mach-cx821xx/cnxtadsl.c arch/armnommu/mach-cx821xx/cnxtbsp.c arch/armnommu/mach-cx821xx/cnxtirq.c arch/armnommu/mach-cx821xx/cnxtsio.c arch/armnommu/mach-cx821xx/cnxttimer.c arch/armnommu/mach-cx821xx/defeepromdata.h arch/armnommu/mach-cx821xx/dma.c arch/armnommu/mach-cx821xx/dmasrv.c arch/armnommu/mach-cx821xx/dsl.c arch/armnommu/mach-cx821xx/dslbase.c arch/armnommu/mach-cx821xx/dslisr.c arch/armnommu/mach-cx821xx/eeprom.c arch/armnommu/mach-cx821xx/eeprom.h arch/armnommu/mach-cx821xx/gpio.c arch/armnommu/mach-cx821xx/gpioisr.c arch/armnommu/mach-cx821xx/hstif.c arch/armnommu/mach-cx821xx/irq.c arch/armnommu/mach-cx821xx/iterate.c arch/armnommu/mach-cx821xx/leds.c arch/armnommu/mach-cx821xx/plldef.h arch/armnommu/mach-cx821xx/project.h arch/armnommu/mach-cx821xx/pwrmanag.c arch/armnommu/mach-cx821xx/sysalib.S arch/armnommu/mach-cx821xx/sysdev.S arch/armnommu/mach-cx821xx/syslib.c arch/armnommu/mach-cx821xx/time.c include/asm-armnommu/arch-cx821xx/Common.h include/asm-armnommu/arch-cx821xx/CommonData.h include/asm-armnommu/arch-cx821xx/ConfigLabels.h include/asm-armnommu/arch-cx821xx/OsDefs.h include/asm-armnommu/arch-cx821xx/OsTools.h include/asm-armnommu/arch-cx821xx/bspcfg.h include/asm-armnommu/arch-cx821xx/bsptypes.h include/asm-armnommu/arch-cx821xx/cardalpub.h include/asm-armnommu/arch-cx821xx/cardinfo.h include/asm-armnommu/arch-cx821xx/cnxt_module.h include/asm-armnommu/arch-cx821xx/cnxtadsl.h include/asm-armnommu/arch-cx821xx/cnxtbsp.h include/asm-armnommu/arch-cx821xx/cnxtirq.h include/asm-armnommu/arch-cx821xx/cnxtsio.h include/asm-armnommu/arch-cx821xx/cnxttimer.h include/asm-armnommu/arch-cx821xx/dma.h include/asm-armnommu/arch-cx821xx/dmasrv.h include/asm-armnommu/arch-cx821xx/dsl.h include/asm-armnommu/arch-cx821xx/dslbase.h include/asm-armnommu/arch-cx821xx/dslisr.h include/asm-armnommu/arch-cx821xx/gpio.h include/asm-armnommu/arch-cx821xx/hardware.h include/asm-armnommu/arch-cx821xx/hstif.h include/asm-armnommu/arch-cx821xx/hwio.h include/asm-armnommu/arch-cx821xx/hwiodmt.h include/asm-armnommu/arch-cx821xx/irq.h include/asm-armnommu/arch-cx821xx/irqs.h include/asm-armnommu/arch-cx821xx/iterate.h include/asm-armnommu/arch-cx821xx/pwrmanag.h include/asm-armnommu/arch-cx821xx/syslib.h include/asm-armnommu/arch-cx821xx/system.h include/asm-armnommu/arch-cx821xx/time.h include/asm-armnommu/arch-cx821xx/timex.h include/asm-armnommu/arch-cx821xx/usbdef.h New files: arch/armnommu/mach-cx821xx/cnxtflashio.c arch/armnommu/mach-cx821xx/gptimer.c arch/armnommu/mach-cx821xx/ostimer.c include/asm-armnommu/arch-cx821xx/gptimer.h include/asm-armnommu/arch-cx821xx/list.c include/asm-armnommu/arch-cx821xx/list.h include/asm-armnommu/arch-cx821xx/ostimer.h -------------- next part -------------- A non-text attachment was scrubbed... Name: cx821xx-update.diff.gz Type: application/x-gzip Size: 43093 bytes Desc: not available URL: From davidm at snapgear.com Tue Mar 29 18:33:51 2005 From: davidm at snapgear.com (David McCullough) Date: Wed, 30 Mar 2005 09:33:51 +1000 Subject: [uClinux-dev] distclean on clean In-Reply-To: <20050324123721.472FFEDC0@mail2.atmark-techno.com> References: <20050324123721.472FFEDC0@mail2.atmark-techno.com> Message-ID: <20050329233351.GA14340@beast> Jivin Yasushi SHOJI lays it down ... > Hi all, > > some applications (ie. busybox) in user/ have distclean target. would > it be ok to modify the user/Makefile to run distclean in those app > dirs on make clean at toplevel? would it be acceptable? We have a distclean target in the top level, hooking it into that would be fine. If not what is the motivation for the change (not that I am against it in any way, just like to know reasons :-) Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From davidm at snapgear.com Tue Mar 29 19:10:19 2005 From: davidm at snapgear.com (David McCullough) Date: Wed, 30 Mar 2005 10:10:19 +1000 Subject: [uClinux-dev] genromfs - giving dev access to anybody In-Reply-To: <424880CF.3090103@alphanet.ch> References: <424880CF.3090103@alphanet.ch> Message-ID: <20050330001019.GE14340@beast> Jivin Julien Linder lays it down ... > > Hi all. > > I need free access to my /dev/ttyz5 device but I can't > figure out how to do. Genromfs makes all devices rw for > root only. Is there any additionnal parameter that can > be given in the filename @ttyz5,c,3,22 ? I think genromfs honours file permissions. In the "make image" part of your vendors/*/*/Makefile, add a: chmod 666 $(ROMFSDIR)/dev/@ttyz5,c,3,22 I think that will do it, there is no other mechanism at this point if that fails. You may need to patch genromfs, Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From jwilliams at itee.uq.edu.au Tue Mar 29 19:18:00 2005 From: jwilliams at itee.uq.edu.au (John Williams) Date: Wed, 30 Mar 2005 10:18:00 +1000 Subject: [uClinux-dev] genromfs - giving dev access to anybody In-Reply-To: <20050330001019.GE14340@beast> References: <424880CF.3090103@alphanet.ch> <20050330001019.GE14340@beast> Message-ID: <4249F038.9070705@itee.uq.edu.au> Hi, David McCullough wrote: > Jivin Julien Linder lays it down ... > >>Hi all. >> >>I need free access to my /dev/ttyz5 device but I can't >>figure out how to do. Genromfs makes all devices rw for >>root only. Is there any additionnal parameter that can >>be given in the filename @ttyz5,c,3,22 ? > > > I think genromfs honours file permissions. > In the "make image" part of your vendors/*/*/Makefile, > add a: > > chmod 666 $(ROMFSDIR)/dev/@ttyz5,c,3,22 > > I think that will do it, there is no other mechanism > at this point if that fails. You may need to patch genromfs, I think I tried this during the telnetd /dev/pty permissions saga, and found that genromfs doesn't honour permissions on dev nodes :( Moving to cramfs worked, plus gave the added benefit of faster application load times :) Regards, John From johnson at syncom.com.tw Tue Mar 29 20:30:03 2005 From: johnson at syncom.com.tw (=?BIG5?Q?=AAL=AB=D8=A8?=}) Date: Wed, 30 Mar 2005 09:30:03 +0800 Subject: [uClinux-dev] How to make the arm-elf-gdb source view In-Reply-To: <20050329170007.ADC561A2AB@uclinux.org> References: <20050329170007.ADC561A2AB@uclinux.org> Message-ID: <20050330012023.M85782@webmail.syncom.com.tw> Hi folks First I should correct something the command I use to compile program is arm-elf-gcc -g -Wl,-elf2flt -o test test.c and I download test into the target and chmod +x test Then gdbserver :9000 test Process /bin/simple created; pid = 184 Listening on port 9000 On my PC I run my insight arm-elf-gdb and open the gdb console type the follow command target remote 192.168.1.4:9000 Remote debugging using 192.168.1.4:9000 0x00ce0050 in ?? () add-symbol-file test.gdb 0xce0050 care here,you may just minus the 0xce0050 0x10 and the debug simbol will match for it add-symbol-file test.gdb 0xce0040 and everything will look comfortable I find the problem and want to share with everyone you all may arm-elf-objdump -S test.gdb and when you target the remote machine see the first instruction code and you will find the first instruction code execute on the target is offset 0x10 to the test.gdb file So that's why we need to minus 0x10 From jasonlarabell at netzero.net Tue Mar 29 21:39:26 2005 From: jasonlarabell at netzero.net (Jason Larabell) Date: Tue, 29 Mar 2005 21:39:26 -0500 Subject: [uClinux-dev] uClinux or other on T5? Message-ID: <424A115E.1030109@netzero.net> Is it possible to use uClinux or perhaps any other distro (or whatever its called) on the palmOne Tungsten T5? If so how do I go about it? Does it coexist with the PalmOS? Can PalmOS applications be installed and used under it? Basically anything else that will help me understand using linux on it will also be appreciated. Thank you From devi at tenet.res.in Wed Mar 30 09:55:12 2005 From: devi at tenet.res.in (devi) Date: Wed, 30 Mar 2005 20:25:12 +0530 Subject: [uClinux-dev] booting sequence Message-ID: <20050330145512.10483.qmail@mail.tenet.res.in> hi, In the virtual console how do I get the booting sequence of uclinux.Any idea.My virtual console is with frame buffer support. After this print statement Console: switching to mono frame buffer device 40x15. I need to get the booting sequence on the virtual console.But I am still getting the booting sequence on the hyperterminal. Regards Devi From devi at tenet.res.in Wed Mar 30 10:02:48 2005 From: devi at tenet.res.in (devi) Date: Wed, 30 Mar 2005 20:32:48 +0530 Subject: [uClinux-dev] character display Message-ID: <20050330150248.11257.qmail@mail.tenet.res.in> hi, Is there any application(c program like lissa.c to check the graphical display) in uclinux to test the framebuffer to display character. regards Devi From davidm at snapgear.com Tue Mar 29 23:25:05 2005 From: davidm at snapgear.com (David McCullough) Date: Wed, 30 Mar 2005 14:25:05 +1000 Subject: [uClinux-dev] LUA (a small script interpreter) compilation in uClinux In-Reply-To: <42492125.40103@csm-instruments.com> References: <42492125.40103@csm-instruments.com> Message-ID: <20050330042505.GB3341@beast> Jivin Julien MARBACH lays it down ... > I'm quite new to uClinux and i'm triying to add my first user-app to > uClinux... > Has anybody tried to compile lua (www.lua.org, a rather small script > interpreter) in uClinux? I'm wondering what changes I have to make in > the lua makefile to add it in the uClinux building process (I've alredy > added the necessary entry in user/Makefile, config/config.in & > config/Configure.help). > > here is a copy of the lua makefile : > > *** > # makefile for Lua hierarchy > # see INSTALL for installation instructions > # see config for customization instructions > > LUA= . > > include $(LUA)/config > > # primary targets ("co" and "klean" are used for making the distribution) > all clean co klean: dirs > cd include; $(MAKE) $@ > cd src; $(MAKE) $@ > cd src/lib; $(MAKE) $@ > cd src/luac; $(MAKE) $@ > cd src/lua; $(MAKE) $@ > > # in case they were not created during unpacking > dirs: bin lib > > bin lib: > mkdir -p $@ > > # simple test to see Lua working > test: all > bin/lua test/hello.lua > > # remove debug information from binaries > strip: > $(STRIP) bin/* > > # official installation > install: all strip > mkdir -p $(INSTALL_BIN) $(INSTALL_INC) $(INSTALL_LIB) $(INSTALL_MAN) > $(INSTALL_EXEC) bin/* $(INSTALL_BIN) > $(INSTALL_DATA) include/*.h $(INSTALL_INC) > $(INSTALL_DATA) lib/*.a $(INSTALL_LIB) > $(INSTALL_DATA) doc/*.1 $(INSTALL_MAN) > > # shared libraries (for Linux) > so: > ld -o lib/liblua.so.$V -shared src/*.o > ld -o lib/liblualib.so.$V -shared src/lib/*.o > cd lib; ln -fs liblua.so.$V liblua.so; ln -fs liblualib.so.$V > liblualib.so > > # binaries using shared libraries > sobin: > rm -f bin/* > cd src/lua; $(MAKE) > cd src/luac; $(MAKE) > > # install shared libraries > soinstall: > $(INSTALL_EXEC) lib/*.so.* $(INSTALL_LIB) > cd $(INSTALL_LIB); ln -fs liblua.so.$V liblua.so; ln -fs > liblualib.so.$V liblualib.so > > # clean shared libraries > soclean: > rm -f lib/*.so* bin/* > > # turn config into Lua code > # uncomment the last sed expression if you want nil instead of empty strings > lecho: > @echo "-- $(LUA)/config for Lua $V" > @echo "VERSION = '$(V)'" > @make echo | grep = | sed -e 's/= /= "/' -e 's/$$/"/' #-e 's/""/nil/' > @echo "-- EOF" > > newer: > @find . -newer MANIFEST -type f > > # (end of Makefile) > > *** > > I believe that the "install" rule have to be changed to something like : > > romfs: > $(ROMFSINST) bin/* $(INSTALL_BIN) include/*.h $(INSTALL_INC) > lib/*.a $(INSTALL_LIB) ROMFSINST can only install a single applications at a time, so you need to look in bin/* and add a ROMFSINST action for each. You could do: $(ROMFSINST) bin/. $(INSTALL_BIN) $(ROMFSINST) include/. $(INSTALL_INC) $(ROMFSINST) lib/. $(INSTALL_LIB) but chances are you do not need the includes/libs on your uClinux platform, especially since it's unlikely you will have a compiler there ;-) So something like: $(ROMFSINST) bin/. /bin is probably enough. > WHat about the "soinstall" rule? No current non-mmu target uses .so's, so you can either leave this or look at some other Makefiles in the tree to see how it can be done. On MMU systems you will need to install the .so's into /lib. For now I would recommend disabling shared libs builds completely until you have the rest working. > Another question : do I have to expect that the build process will > produce a large executable file/library? Hopefully it produces something :-) Not sure I understand your issue here ? The size of the binary is the size you have to deal with. You do not need alll the dev libs/headers though, Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From posedge at gmail.com Wed Mar 30 00:15:35 2005 From: posedge at gmail.com (kartikeyan s) Date: Wed, 30 Mar 2005 10:45:35 +0530 Subject: [uClinux-dev] gethostbyname() on xcopilot simulator Message-ID: <4dbae2c205032921152c53670@mail.gmail.com> Hello, When I start xcopilot simulator uclinux i have following in the message list. NET4: Linux TCP/IP 1.0 for NET4.0 IP Protocols: ICMP, UDP, TCP But still when i run my program gethostbyname() returns NULL even for "localhost" and "CoPilot" No compilation errors on host. Can anyone explain why. Anyone has sucess with xcopilot in writing simple network programs in xcopilot please tell what settings i need to change From davidm at snapgear.com Wed Mar 30 00:41:29 2005 From: davidm at snapgear.com (David McCullough) Date: Wed, 30 Mar 2005 15:41:29 +1000 Subject: [uClinux-dev] [patch] update cx821xx architecture In-Reply-To: <200503292254.09474.jhml@gmx.net> References: <200503292254.09474.jhml@gmx.net> Message-ID: <20050330054129.GA6410@beast> Johann, Applied, if you get a chance to check it's all ok it would be appreciated :-) Thanks, Davidm Jivin Johann Hanne lays it down ... > Hi, > > after searching for the reason of the ethernet tx timeouts I got on my DPCM > for a while, I finally recognized that the uclinux tree within > ftp://128.121.226.214/pub/dualpcmodem/actiontec_release_031017.tar.gz > contains newer file for the cx821xx architecture than the current uClinux CVS. > So I diffed the CVS tree against it and the results look ok, it only changes > a few things, all within the arch directory: > - major updates to the timer routines (these probably solve my ethernet > problems) > - lots of comment changes (version strings, updates of the year numbers) > - DSL updates > > I've tested the patch against current CVS and it works, so it would be > wonderful if you could apply it, Greg. :-) > > Thanks, Johann > > Changed Files: > arch/armnommu/mach-cx821xx/LnxTools.c > arch/armnommu/mach-cx821xx/Makefile > arch/armnommu/mach-cx821xx/arch.c > arch/armnommu/mach-cx821xx/cacheenb.S > arch/armnommu/mach-cx821xx/cnxtadsl.c > arch/armnommu/mach-cx821xx/cnxtbsp.c > arch/armnommu/mach-cx821xx/cnxtirq.c > arch/armnommu/mach-cx821xx/cnxtsio.c > arch/armnommu/mach-cx821xx/cnxttimer.c > arch/armnommu/mach-cx821xx/defeepromdata.h > arch/armnommu/mach-cx821xx/dma.c > arch/armnommu/mach-cx821xx/dmasrv.c > arch/armnommu/mach-cx821xx/dsl.c > arch/armnommu/mach-cx821xx/dslbase.c > arch/armnommu/mach-cx821xx/dslisr.c > arch/armnommu/mach-cx821xx/eeprom.c > arch/armnommu/mach-cx821xx/eeprom.h > arch/armnommu/mach-cx821xx/gpio.c > arch/armnommu/mach-cx821xx/gpioisr.c > arch/armnommu/mach-cx821xx/hstif.c > arch/armnommu/mach-cx821xx/irq.c > arch/armnommu/mach-cx821xx/iterate.c > arch/armnommu/mach-cx821xx/leds.c > arch/armnommu/mach-cx821xx/plldef.h > arch/armnommu/mach-cx821xx/project.h > arch/armnommu/mach-cx821xx/pwrmanag.c > arch/armnommu/mach-cx821xx/sysalib.S > arch/armnommu/mach-cx821xx/sysdev.S > arch/armnommu/mach-cx821xx/syslib.c > arch/armnommu/mach-cx821xx/time.c > include/asm-armnommu/arch-cx821xx/Common.h > include/asm-armnommu/arch-cx821xx/CommonData.h > include/asm-armnommu/arch-cx821xx/ConfigLabels.h > include/asm-armnommu/arch-cx821xx/OsDefs.h > include/asm-armnommu/arch-cx821xx/OsTools.h > include/asm-armnommu/arch-cx821xx/bspcfg.h > include/asm-armnommu/arch-cx821xx/bsptypes.h > include/asm-armnommu/arch-cx821xx/cardalpub.h > include/asm-armnommu/arch-cx821xx/cardinfo.h > include/asm-armnommu/arch-cx821xx/cnxt_module.h > include/asm-armnommu/arch-cx821xx/cnxtadsl.h > include/asm-armnommu/arch-cx821xx/cnxtbsp.h > include/asm-armnommu/arch-cx821xx/cnxtirq.h > include/asm-armnommu/arch-cx821xx/cnxtsio.h > include/asm-armnommu/arch-cx821xx/cnxttimer.h > include/asm-armnommu/arch-cx821xx/dma.h > include/asm-armnommu/arch-cx821xx/dmasrv.h > include/asm-armnommu/arch-cx821xx/dsl.h > include/asm-armnommu/arch-cx821xx/dslbase.h > include/asm-armnommu/arch-cx821xx/dslisr.h > include/asm-armnommu/arch-cx821xx/gpio.h > include/asm-armnommu/arch-cx821xx/hardware.h > include/asm-armnommu/arch-cx821xx/hstif.h > include/asm-armnommu/arch-cx821xx/hwio.h > include/asm-armnommu/arch-cx821xx/hwiodmt.h > include/asm-armnommu/arch-cx821xx/irq.h > include/asm-armnommu/arch-cx821xx/irqs.h > include/asm-armnommu/arch-cx821xx/iterate.h > include/asm-armnommu/arch-cx821xx/pwrmanag.h > include/asm-armnommu/arch-cx821xx/syslib.h > include/asm-armnommu/arch-cx821xx/system.h > include/asm-armnommu/arch-cx821xx/time.h > include/asm-armnommu/arch-cx821xx/timex.h > include/asm-armnommu/arch-cx821xx/usbdef.h > > New files: > arch/armnommu/mach-cx821xx/cnxtflashio.c > arch/armnommu/mach-cx821xx/gptimer.c > arch/armnommu/mach-cx821xx/ostimer.c > include/asm-armnommu/arch-cx821xx/gptimer.h > include/asm-armnommu/arch-cx821xx/list.c > include/asm-armnommu/arch-cx821xx/list.h > include/asm-armnommu/arch-cx821xx/ostimer.h > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uclinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From hyok.choi at samsung.com Wed Mar 30 01:30:03 2005 From: hyok.choi at samsung.com (Hyok S. Choi) Date: Wed, 30 Mar 2005 15:30:03 +0900 Subject: [uClinux-dev] [PATCH] the nommu support for ARM linux 2.6.10 Message-ID: <0IE500K2ZJGAE8@mmp1.samsung.com> An updated MPU and noMMU support patch for ARM against linux 2.6.10 kernel is available at : http://opensrc.sec.samsung.com/download/linux-2.6.10-hsc1.patch.gz You can select the memory management type "MPU" or "NONE" in the arm kernel configuration menu, which was traditionally known as "armnommu" or uClinux/ARM by 2.6.9. (sure, you can choose "MMU" for vanila Linux :-) It's a different way from other uclinux arch. (i.e. m68knommu), which enables simultaneous support to use "singular address space" support even for MMU platforms. You can choose "MMU" or "NONE" for your mmu based arm platform with a few modification. i.e. virtual address --> physical address conversion. the 2.6.11.6-hsc0 patch will be available in this week, and some benchmark will be provided for both cases on a same h/w platform. and addtional MPU support API is pending for some services like memory protection, even for uClinux. any suggesstions welcomed. You can reach the project home at : http://opensrc.sec.samsung.com/ currently officially supported platforms are : s3c24a0, s5c7375, atmel, espd_s3c510b, P2001, s3c3410, s3cb0x thanks for contribution : Tobias Lorenz and Jiun-Shian Ho Regards, Hyok --- CHOI, HYOK-SUNG Engineer (Kernel/System Architecture) Digital Media R&D Center, Samsung Electronics Co.,Ltd. tel: +82-31-200-8594 fax: +82-31-200-3427 e-mail: hyok.choi at samsung.com [Linux 2.6 ARM MPU/noMMU Kernel Maintainer] http://opensrc.sec.samsung.com/ From kgeorgejoseph at rediffmail.com Wed Mar 30 01:33:44 2005 From: kgeorgejoseph at rediffmail.com (George Joseph) Date: 30 Mar 2005 06:33:44 -0000 Subject: [uClinux-dev] BootLoader port for Smasung S3C440BOX Message-ID: <20050330063344.32144.qmail@webmail30.rediffmail.com> Hello All, I am new to this group.I have some expertise on developing kernel drivers(from the Linux Device Drivers book)..........:-)...... I have a Samsung S3C440BOX board with 2MB flash,8MB SDRAM,1 USB (device) port,1 Ethernet RTL8019AS chip ,2 Serial(male) ports ,1 IDE interface and 4 Application buttons.The board is from a company called Anyway Technology Company in HK.I am interested in developing the bootloader for this board (I wish to learn the internal of OSs and Bootloaders). I do know that there are free bootloaders.I would like to customise the Loader and hence create a port for the board I have.Could any of you please tell me how I could get around doing this. I have a wiggler cable for downloading the binary to the FLASH and an application for the same.Are the above mentioned utilities sufficient for starting development.Any feedback on the above said would be highly appreciated. Thanks In Advance.George. -------------- next part -------------- An HTML attachment was scrubbed... URL: From hyok.choi at samsung.com Wed Mar 30 01:53:04 2005 From: hyok.choi at samsung.com (Hyok S. Choi) Date: Wed, 30 Mar 2005 15:53:04 +0900 Subject: [uClinux-dev] updating of uclinux.org? Message-ID: <0IE5001UCKIOM7@mmp2.samsung.com> Greetings! I have a question on the uclinux.org maintainance. Actually, much stuffs on the uclinux.org site is outdated. Do we need any maintain resources or self-maintained style re-structuring? (maybe wiki style stuffs are helpful, I think.) regards, Hyok --- CHOI, HYOK-SUNG Engineer (Kernel/System Architecture) Digital Media R&D Center, Samsung Electronics Co.,Ltd. tel: +82-31-200-8594 fax: +82-31-200-3427 e-mail: hyok.choi at samsung.com [Linux 2.6 ARM MPU/noMMU Kernel Maintainer] http://opensrc.sec.samsung.com/ From lars.friedrich at wago.com Wed Mar 30 01:56:18 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Wed, 30 Mar 2005 08:56:18 +0200 Subject: [uClinux-dev] gethostbyname() on xcopilot simulator Message-ID: > NET4: Linux TCP/IP 1.0 for NET4.0 > IP Protocols: ICMP, UDP, TCP > > But still when i run my program > gethostbyname() returns NULL even for "localhost" and "CoPilot" > No compilation errors on host. > > Can anyone explain why. You don't have a /etc/hosts file? Best regards, Lars Friedrich From elbadaouimourad at gmail.com Wed Mar 30 05:16:17 2005 From: elbadaouimourad at gmail.com (mourad el badaoui) Date: Wed, 30 Mar 2005 10:16:17 +0000 Subject: [uClinux-dev] compiling jamvm for arm/uclinux Message-ID: hi well, i am trying to compile jamvm for arm9/uclinux using arm-elf-tools, and i wonder know what are the arguments i must use for the compilation...... the configuration i am trying to use now is as follows: ########################################################## CC=arm-elf-gcc AR=arm-elf-ar LD=arm-elf-ld STRIP=arm-elf-strip RANLIB=arm-elf-ranlib ./configure --prefix=/root/arm-jamvm --build=i686-linux --host=arm-uclinux --with-classpath-lib-dir=/usr/local/classpath ########################################################## but i got an error related to "libdl": ############################### checking for dlopen in -ldl... no configure:error:libdl is missing ############################### i need your help thanks best regards -- Mourad EL BADAOUI From julien at alphanet.ch Wed Mar 30 05:20:25 2005 From: julien at alphanet.ch (Julien Linder) Date: Wed, 30 Mar 2005 12:20:25 +0200 Subject: [uClinux-dev] genromfs - giving dev access to anybody In-Reply-To: <424A44AA.6010701@alphanet.ch> References: <424880CF.3090103@alphanet.ch> <20050330001019.GE14340@beast> <4249F038.9070705@itee.uq.edu.au> <424A44AA.6010701@alphanet.ch> Message-ID: <424A7D69.3050703@alphanet.ch> John Williams wrote: > Hi, > > David McCullough wrote: > >> Jivin Julien Linder lays it down ... >> >>> Hi all. >>> >>> I need free access to my /dev/ttyz5 device but I can't >>> figure out how to do. Genromfs makes all devices rw for >>> root only. Is there any additionnal parameter that can >>> be given in the filename @ttyz5,c,3,22 ? >> >> >> >> >> I think genromfs honours file permissions. >> In the "make image" part of your vendors/*/*/Makefile, >> add a: >> >> chmod 666 $(ROMFSDIR)/dev/@ttyz5,c,3,22 >> >> I think that will do it, there is no other mechanism >> at this point if that fails. You may need to patch genromfs, > > > > I think I tried this during the telnetd /dev/pty permissions saga, and found that genromfs doesn't honour permissions on dev nodes :( > > Moving to cramfs worked, plus gave the added benefit of faster application load times :) > You are right : Genromfs doesn't seem to honour file permissions. Using Cramfs is the solution I took. Thank you for answering. Julien From id3 at jandei.com Wed Mar 30 05:48:08 2005 From: id3 at jandei.com (Juan Manuel ) Date: Wed, 30 Mar 2005 12:48:08 +0200 Subject: [uClinux-dev] Compiling problem with sendfile() + fnord Message-ID: <002d01c53515$f9480840$b400a8c0@jandei.org> Hi all, I would make that fnord uses sendfile() to improve the global performance of the webserver. Enabling this option it does not compile and shows an "undefined reference to 'sendfile' ". My configuration is: Platform: M5282 Distro: uclinux 20040218 Kernel: 2.4.24-uc0 Toolchain: m68k-elf-tools-20030314 The output is: ######################################################################## # UCLINUX_BUILD_USER=1 make user/fnord_only [ ! -d "user/fnord" ] || make ARCH=m68knommu CROSS_COMPILE=m68k-elf- -C user/fnord make[1]: Entering directory `/id/uclinux20040218/uClinux-distA-cfm/user/fnord' m68k-elf-gcc -m5307 -DCONFIG_COLDFIRE -Os -g -fomit-frame-pointer -m5307 -DCONFIG_COLDFIRE -Dlinux -D__linux__ -Dunix -D__uClinux__ -DEMBED -I/id/uclinux20040218/uClinux-distA-cfm/lib/libc/include -I/id/uclinux20040218/uClinux-distA-cfm/lib/libm -I/id/uclinux20040218/uClinux-distA-cfm/lib/libcrypt_old -I/id/uclinux20040218/uClinux-distA-cfm -fno-builtin -msep-data -I/id/uclinux20040218/uClinux-distA-cfm/linux-2.4.x/include -Wl,-elf2flt -Wl,-move-rodata -L/id/uclinux20040218/uClinux-distA-cfm/lib/libc/. -L/id/uclinux20040218/uClinux-distA-cfm/lib/libc/lib -L/id/uclinux20040218/uClinux-distA-cfm/lib/libm -L/id/uclinux20040218/uClinux-distA-cfm/lib/libnet -L/id/uclinux20040218/uClinux-distA-cfm/lib/libdes -L/id/uclinux20040218/uClinux-distA-cfm/lib/libaes -L/id/uclinux20040218/uClinux-distA-cfm/lib/libpcap -L/id/uclinux20040218/uClinux-distA-cfm/lib/libssl -L/id/uclinux20040218/uClinux-distA-cfm/lib/libcrypt_old -L/id/uclinux20040218/uClinux-distA-cfm/prop/libsnapgear++ -L/id/uclinux20040218/uClinux-distA-cfm/prop/libsnapgear -L/id/uclinux20040218/uClinux-distA-cfm/lib/zlib -o httpd httpd.o auth.o base64.o libowfat.a -lcrypt_old -lc httpd.elf2flt: In function `serve_static_data': /id/uclinux20040218/uClinux-distA-cfm/user/fnord/httpd.c:1235: undefined reference to `sendfile' collect2: ld returned 1 exit status make[1]: *** [httpd] Error 1 make[1]: Leaving directory `/id/uclinux20040218/uClinux-distA-cfm/user/fnord' make: *** [user/fnord_only] Error 2 ######################################################################## Also I made an app to test sendfile and it compiles fine with this makefile: ====================================================================== # # Makefile para test_sendfile # install_dir = /id/cp/borrar CROSS_COMPILE = /usr/local/bin/m68k-elf- AS = $(CROSS_COMPILE)as LD = $(CROSS_COMPILE)ld CC = $(CROSS_COMPILE)gcc AR = $(CROSS_COMPILE)ar ELF2FLT = $(CROSS_COMPILE)elf2flt FLTHDR = $(CROSS_COMPILE)flthdr CFLAGS += -Wall -O1 -g -m5307 LDLIBS += -Wl,-elf2flt -m5307 -lc -g test_sendfile: test_sendfile.o stack_size: $(FLTHDR) -s 16384 test_sendfile install_prog: cp test_sendfile /id/cp/borrar/test_sendfile install: clean test_sendfile install_prog clean: rm -f *.o *~ test_sendfile $(install_dir)/test_sendfile ======================================================================== but if I include the app in /user and compile it with "UCLINUX_BUILD_USER=1 make user/testsendfile_only" fails with "undefined reference to 'sendfile'". Of course both programs include . I had searched for any other user app that use sendfile into /user tree and there is no one that uses it. Maybe am I missing some tip regarding sendfile on uClinux (portability issues...)? Regards, ...................... Juan Manuel Ramil Departamento I+D id3 at jandei.com www.jandei.com 981137335 ...................... From CCarson at CBWAutomation.com Wed Mar 30 08:37:52 2005 From: CCarson at CBWAutomation.com (Carson, Corey) Date: Wed, 30 Mar 2005 06:37:52 -0700 Subject: [uClinux-dev] character display Message-ID: <203F609DF3BDD611843C000475A1967C06C8D3@mail.cbwautomation.com> I'm currently working on a font function that can be added to lissa. The way I do it is to map fonts as pixel information in an ascii file. A function opens the font ascii file, reads in the Boolean characters and writes pixels accordingly. At this point, my program is still in development however performance and aesthetics doesn't seem to be a problem. Corey -----Original Message----- From: uclinux-dev-bounces at uclinux.org [mailto:uclinux-dev-bounces at uclinux.org] On Behalf Of devi Sent: Wednesday, March 30, 2005 8:03 AM To: uclinux-dev at uclinux.org Subject: [uClinux-dev] character display hi, Is there any application(c program like lissa.c to check the graphical display) in uclinux to test the framebuffer to display character. regards Devi _______________________________________________ uClinux-dev mailing list uClinux-dev at uclinux.org http://mailman.uclinux.org/mailman/listinfo/uclinux-dev This message was resent by uclinux-dev at uclinux.org From pundrich at gmail.com Wed Mar 30 11:30:41 2005 From: pundrich at gmail.com (Gabriel Pundrich) Date: Wed, 30 Mar 2005 13:30:41 -0300 Subject: [uClinux-dev] Adding my apps Message-ID: <7c2bef0805033008301ff6237d@mail.gmail.com> Hi, I am trying to add a simple hello.c and I am getting problems, Could anybody help me? Ps: I am using a M5282EVB eval board with coldfire MCF5282 Thanks, Gabriel From ctusar at videon-central.com Wed Mar 30 11:40:05 2005 From: ctusar at videon-central.com (Cory Tusar) Date: Wed, 30 Mar 2005 11:40:05 -0500 Subject: [uClinux-dev] Adding my apps In-Reply-To: <7c2bef0805033008301ff6237d@mail.gmail.com> References: <7c2bef0805033008301ff6237d@mail.gmail.com> Message-ID: <424AD665.7010301@videon-central.com> Gabriel Pundrich wrote: > Hi, I am trying to add a simple hello.c and I am getting problems, > Could anybody help me? uClinux-dist/Documentation/Adding-User-Apps-HOWTO -- Cory T. Tusar Embedded Systems Engineer Videon Central, Inc. 2171 Sandy Drive State College, PA 16801 (814) 235-1111 x316 (814) 235-1118 fax "Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it." --Brian W. Kernighan From frnagel at gmail.com Wed Mar 30 11:38:07 2005 From: frnagel at gmail.com (Fernando Remus Nagel) Date: Wed, 30 Mar 2005 13:38:07 -0300 Subject: [uClinux-dev] Adding my apps In-Reply-To: <7c2bef0805033008301ff6237d@mail.gmail.com> References: <7c2bef0805033008301ff6237d@mail.gmail.com> Message-ID: <517442a00503300838258ccaf5@mail.gmail.com> Tu por aqui? On Wed, 30 Mar 2005 13:30:41 -0300, Gabriel Pundrich wrote: > Hi, I am trying to add a simple hello.c and I am getting problems, > Could anybody help me? > > Ps: I am using a M5282EVB eval board with coldfire MCF5282 > > Thanks, > Gabriel > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uclinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org > From eauth at softsys.co.at Wed Mar 30 11:39:51 2005 From: eauth at softsys.co.at (Erwin Authried) Date: Wed, 30 Mar 2005 18:39:51 +0200 Subject: [uClinux-dev] How to make the arm-elf-gdb source view In-Reply-To: <20050330012023.M85782@webmail.syncom.com.tw> References: <20050329170007.ADC561A2AB@uclinux.org> <20050330012023.M85782@webmail.syncom.com.tw> Message-ID: <1112200791.2312.61.camel@justakiss> Am Mit, den 30.03.2005 schrieb =?BIG5?Q?=AAL=AB=D8=A8?=} um 3:30: > Hi folks > First I should correct something > the command I use to compile program is > arm-elf-gcc -g -Wl,-elf2flt -o test test.c > > and I download test into the target > and chmod +x test > > Then gdbserver :9000 test > Process /bin/simple created; pid = 184 > Listening on port 9000 > > On my PC > I run my insight arm-elf-gdb > and open the gdb console > type the follow command > target remote 192.168.1.4:9000 > Remote debugging using 192.168.1.4:9000 > 0x00ce0050 in ?? () > > add-symbol-file test.gdb 0xce0050 > > care here,you may just minus the 0xce0050 0x10 > > and the debug simbol will match for it > add-symbol-file test.gdb 0xce0040 and everything will look comfortable > > I find the problem and want to share with everyone > > you all may arm-elf-objdump -S test.gdb > and when you target the remote machine see the first instruction code > and you will find the first instruction code execute on the target is > offset 0x10 to the test.gdb file > > So that's why we need to minus 0x10 > Hi, I wonder that this should really work, you didn't specify -s .data ... and -s .bss ... with add-symbol. Which version of gdbserver do you use? Regards, Erwin From frnagel at gmail.com Wed Mar 30 12:10:57 2005 From: frnagel at gmail.com (Fernando Remus Nagel) Date: Wed, 30 Mar 2005 14:10:57 -0300 Subject: [uClinux-dev] Adding my apps In-Reply-To: <7c2bef0805033008301ff6237d@mail.gmail.com> References: <7c2bef0805033008301ff6237d@mail.gmail.com> Message-ID: <517442a00503300910471c33e1@mail.gmail.com> Quais problemas? On Wed, 30 Mar 2005 13:30:41 -0300, Gabriel Pundrich wrote: > Hi, I am trying to add a simple hello.c and I am getting problems, > Could anybody help me? > > Ps: I am using a M5282EVB eval board with coldfire MCF5282 > > Thanks, > Gabriel > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uclinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org > From pundrich at gmail.com Wed Mar 30 12:11:09 2005 From: pundrich at gmail.com (Gabriel Pundrich) Date: Wed, 30 Mar 2005 14:11:09 -0300 Subject: [uClinux-dev] Adding my apps In-Reply-To: <517442a00503300838258ccaf5@mail.gmail.com> References: <7c2bef0805033008301ff6237d@mail.gmail.com> <517442a00503300838258ccaf5@mail.gmail.com> Message-ID: <7c2bef08050330091163fea8d2@mail.gmail.com> HI, I am using this how-to as guide since yesterday and I am having some problems like: 1- I created a directory in /xxx/uClinux-dist/foo as the guide says 2- I put the Makefile: >EXEC = foo >OBJS = foo.o > > all: $(EXEC) > > $(EXEC): $(OBJS) > $(CC) $(LDFLAGS) -o $@ $(OBJS) $(LDLIBS) > > romfs: > $(ROMFSINST) /bin/$(EXEC) > > clean: > -rm -f $(EXEC) *.elf *.gdb *.o 3- I edited the /xxx/uClinux-dist/Makefile and I put in the end of the file the line: dir_$(CONFIG_USER_FOO_FOO) += foo first question---> Its right where I put this line? 4- I edited the /xxx/uClinux-dist/config/config.in and write the line in the section of Misc: bool 'foo' CONFIG_USER_FOO_FOO 5- I enter in /xxx/uClinux-dist/foo and type "make" and the I got the files: debian:/xxx/uClinux-dist/foo# ls Makefile foo foo.c foo.o if I type the command file * see what I got: debian:/xxx/uClinux-dist/foo# file * Makefile: ASCII make commands text foo: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV), for GNU/Linux 2.2.0, dynamically linked (uses shared libs), not stripped foo.c: ASCII C program text foo.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped 6 - After this I type "make romfs" in the /xxx/uClinux-dist/foo directory and I got the following error: debian:/xxx/uClinux-dist/foo# make romfs /bin/foo make: /bin/foo: Command not found make: *** [romfs] Error 127 Thats all, I hope I wont get too extensive, What step do I doing wrong?? Thanks for the attention, Gabriel Pundrich On Wed, 30 Mar 2005 13:38:07 -0300, Fernando Remus Nagel wrote: > Tu por aqui? > > > On Wed, 30 Mar 2005 13:30:41 -0300, Gabriel Pundrich wrote: > > Hi, I am trying to add a simple hello.c and I am getting problems, > > Could anybody help me? > > > > Ps: I am using a M5282EVB eval board with coldfire MCF5282 > > > > Thanks, > > Gabriel > > _______________________________________________ > > uClinux-dev mailing list > > uClinux-dev at uclinux.org > > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > > This message was resent by uclinux-dev at uclinux.org > > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uclinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org > From r&d4 at dave-tech.it Wed Mar 30 12:19:24 2005 From: r&d4 at dave-tech.it (R&D4) Date: Wed, 30 Mar 2005 19:19:24 +0200 Subject: [uClinux-dev] BootLoader port for Smasung S3C440BOX In-Reply-To: <20050330170010.307AD1A2A6@uclinux.org> References: <20050330170010.307AD1A2A6@uclinux.org> Message-ID: <6.2.1.2.0.20050330190840.01cf42a0@192.168.2.1> Hi George! >Hello All, > >I am new to this group.I have some expertise on developing >kernel drivers(from the Linux Device Drivers book)..........:-)...... I >have a Samsung S3C440BOX board with 2MB flash,8MB SDRAM,1 USB (device) >port,1 Ethernet RTL8019AS chip ,2 Serial(male) ports ,1 IDE interface >and 4 Application buttons.The board is from a company called Anyway >Technology Company in HK.I am interested in developing the bootloader >for this board (I wish to learn the internal of OSs and >Bootloaders). I do know that there are free bootloaders.I would like to >customise the Loader and hence create a port for the board I have.Could >any of you please tell me how I could get around >doing this. Take a look at u-boot (http://sourceforge.net/projects/u-boot). It is a very flexible, universal and powerful bootloader (and it has been ported to lots of different platform!). I'm the developer of the port of u-boot on our S3C44B0X-based board with uClinux (called B2: 4MB flash, 16 MB SDRAM, SMSC91C96 ethernet, 2 serial ports and so on..). It should be easy to adapt my port to your board! >I have a wiggler cable for downloading the binary to the FLASH and an >application for the same.Are the above mentioned utilities sufficient >for starting development.Any feedback on the above said would be highly >appreciated. I was using wiggler too, an year ago.. it's a bit slow (at least in my configuration! :-) but enough develop a bootloader!! >Thanks In Advance.George. Best Regards, Amon DAVE Electronics System House - R&D Department web: http://www.dave-tech.it email: r&d4 at dave-tech.it From jhml at gmx.net Wed Mar 30 12:34:45 2005 From: jhml at gmx.net (Johann Hanne) Date: Wed, 30 Mar 2005 19:34:45 +0200 Subject: [uClinux-dev] [patch] update cx821xx architecture In-Reply-To: <20050330054129.GA6410@beast> References: <200503292254.09474.jhml@gmx.net> <20050330054129.GA6410@beast> Message-ID: <200503301934.57957.jhml@gmx.net> Hi David, I just checked out CVS and it works wonderfully. Thanks a lot! Looks like I can finally concentrate on userspace stuff. :-) Cheers, Johann On Wednesday 30 March 2005 07:41, David McCullough wrote: > Johann, > > Applied, if you get a chance to check it's all ok it would be > appreciated :-) > > Thanks, > Davidm > > > Jivin Johann Hanne lays it down ... > > > Hi, > > > > after searching for the reason of the ethernet tx timeouts I got on my > > DPCM for a while, I finally recognized that the uclinux tree within > > ftp://128.121.226.214/pub/dualpcmodem/actiontec_release_031017.tar.gz > > contains newer file for the cx821xx architecture than the current uClinux > > CVS. So I diffed the CVS tree against it and the results look ok, it only > > changes a few things, all within the arch directory: > > - major updates to the timer routines (these probably solve my ethernet > > problems) > > - lots of comment changes (version strings, updates of the year numbers) > > - DSL updates > > > > I've tested the patch against current CVS and it works, so it would be > > wonderful if you could apply it, Greg. :-) > > > > Thanks, Johann > > > > Changed Files: > > arch/armnommu/mach-cx821xx/LnxTools.c > > arch/armnommu/mach-cx821xx/Makefile > > arch/armnommu/mach-cx821xx/arch.c > > arch/armnommu/mach-cx821xx/cacheenb.S > > arch/armnommu/mach-cx821xx/cnxtadsl.c > > arch/armnommu/mach-cx821xx/cnxtbsp.c > > arch/armnommu/mach-cx821xx/cnxtirq.c > > arch/armnommu/mach-cx821xx/cnxtsio.c > > arch/armnommu/mach-cx821xx/cnxttimer.c > > arch/armnommu/mach-cx821xx/defeepromdata.h > > arch/armnommu/mach-cx821xx/dma.c > > arch/armnommu/mach-cx821xx/dmasrv.c > > arch/armnommu/mach-cx821xx/dsl.c > > arch/armnommu/mach-cx821xx/dslbase.c > > arch/armnommu/mach-cx821xx/dslisr.c > > arch/armnommu/mach-cx821xx/eeprom.c > > arch/armnommu/mach-cx821xx/eeprom.h > > arch/armnommu/mach-cx821xx/gpio.c > > arch/armnommu/mach-cx821xx/gpioisr.c > > arch/armnommu/mach-cx821xx/hstif.c > > arch/armnommu/mach-cx821xx/irq.c > > arch/armnommu/mach-cx821xx/iterate.c > > arch/armnommu/mach-cx821xx/leds.c > > arch/armnommu/mach-cx821xx/plldef.h > > arch/armnommu/mach-cx821xx/project.h > > arch/armnommu/mach-cx821xx/pwrmanag.c > > arch/armnommu/mach-cx821xx/sysalib.S > > arch/armnommu/mach-cx821xx/sysdev.S > > arch/armnommu/mach-cx821xx/syslib.c > > arch/armnommu/mach-cx821xx/time.c > > include/asm-armnommu/arch-cx821xx/Common.h > > include/asm-armnommu/arch-cx821xx/CommonData.h > > include/asm-armnommu/arch-cx821xx/ConfigLabels.h > > include/asm-armnommu/arch-cx821xx/OsDefs.h > > include/asm-armnommu/arch-cx821xx/OsTools.h > > include/asm-armnommu/arch-cx821xx/bspcfg.h > > include/asm-armnommu/arch-cx821xx/bsptypes.h > > include/asm-armnommu/arch-cx821xx/cardalpub.h > > include/asm-armnommu/arch-cx821xx/cardinfo.h > > include/asm-armnommu/arch-cx821xx/cnxt_module.h > > include/asm-armnommu/arch-cx821xx/cnxtadsl.h > > include/asm-armnommu/arch-cx821xx/cnxtbsp.h > > include/asm-armnommu/arch-cx821xx/cnxtirq.h > > include/asm-armnommu/arch-cx821xx/cnxtsio.h > > include/asm-armnommu/arch-cx821xx/cnxttimer.h > > include/asm-armnommu/arch-cx821xx/dma.h > > include/asm-armnommu/arch-cx821xx/dmasrv.h > > include/asm-armnommu/arch-cx821xx/dsl.h > > include/asm-armnommu/arch-cx821xx/dslbase.h > > include/asm-armnommu/arch-cx821xx/dslisr.h > > include/asm-armnommu/arch-cx821xx/gpio.h > > include/asm-armnommu/arch-cx821xx/hardware.h > > include/asm-armnommu/arch-cx821xx/hstif.h > > include/asm-armnommu/arch-cx821xx/hwio.h > > include/asm-armnommu/arch-cx821xx/hwiodmt.h > > include/asm-armnommu/arch-cx821xx/irq.h > > include/asm-armnommu/arch-cx821xx/irqs.h > > include/asm-armnommu/arch-cx821xx/iterate.h > > include/asm-armnommu/arch-cx821xx/pwrmanag.h > > include/asm-armnommu/arch-cx821xx/syslib.h > > include/asm-armnommu/arch-cx821xx/system.h > > include/asm-armnommu/arch-cx821xx/time.h > > include/asm-armnommu/arch-cx821xx/timex.h > > include/asm-armnommu/arch-cx821xx/usbdef.h > > > > New files: > > arch/armnommu/mach-cx821xx/cnxtflashio.c > > arch/armnommu/mach-cx821xx/gptimer.c > > arch/armnommu/mach-cx821xx/ostimer.c > > include/asm-armnommu/arch-cx821xx/gptimer.h > > include/asm-armnommu/arch-cx821xx/list.c > > include/asm-armnommu/arch-cx821xx/list.h > > include/asm-armnommu/arch-cx821xx/ostimer.h > > > > > > _______________________________________________ > > uClinux-dev mailing list > > uClinux-dev at uclinux.org > > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > > This message was resent by uclinux-dev at uclinux.org From pundrich at gmail.com Wed Mar 30 12:45:06 2005 From: pundrich at gmail.com (Gabriel Pundrich) Date: Wed, 30 Mar 2005 14:45:06 -0300 Subject: [uClinux-dev] Adding my apps In-Reply-To: <424AD665.7010301@videon-central.com> References: <7c2bef0805033008301ff6237d@mail.gmail.com> <424AD665.7010301@videon-central.com> Message-ID: <7c2bef080503300945576d3945@mail.gmail.com> Oh, sorry I found a BIG MISTAKE!! I didnt put my directory as subdirectory in user !! I will try this in this way now, Thanks! Gabriel On Wed, 30 Mar 2005 11:40:05 -0500, Cory Tusar wrote: > Gabriel Pundrich wrote: > > Hi, I am trying to add a simple hello.c and I am getting problems, > > Could anybody help me? > > uClinux-dist/Documentation/Adding-User-Apps-HOWTO > > -- > Cory T. Tusar > Embedded Systems Engineer > Videon Central, Inc. > 2171 Sandy Drive > State College, PA 16801 > (814) 235-1111 x316 > (814) 235-1118 fax > > "Debugging is twice as hard as writing the code in the first place. > Therefore, if you write the code as cleverly as possible, you are, > by definition, not smart enough to debug it." --Brian W. Kernighan > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uclinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org > From ctusar at videon-central.com Wed Mar 30 12:55:57 2005 From: ctusar at videon-central.com (Cory Tusar) Date: Wed, 30 Mar 2005 12:55:57 -0500 Subject: [uClinux-dev] Adding my apps In-Reply-To: <7c2bef08050330091163fea8d2@mail.gmail.com> References: <7c2bef0805033008301ff6237d@mail.gmail.com> <517442a00503300838258ccaf5@mail.gmail.com> <7c2bef08050330091163fea8d2@mail.gmail.com> Message-ID: <424AE82D.2070803@videon-central.com> Gabriel Pundrich wrote: > HI, > I am using this how-to as guide since yesterday and I am having some > problems like: > > 1- I created a directory in /xxx/uClinux-dist/foo as the guide says Try /xxx/uClinux-dist/user/foo > 2- I put the Makefile: > > >>EXEC = foo >>OBJS = foo.o >> >> all: $(EXEC) >> >> $(EXEC): $(OBJS) >> $(CC) $(LDFLAGS) -o $@ $(OBJS) $(LDLIBS) >> >> romfs: >> $(ROMFSINST) /bin/$(EXEC) >> >> clean: >> -rm -f $(EXEC) *.elf *.gdb *.o This should be /xxx/uClinux-dist/user/foo/Makefile > 3- I edited the /xxx/uClinux-dist/Makefile and I put in the end of > the file the line: > dir_$(CONFIG_USER_FOO_FOO) += foo > first question---> Its right where I put this line? No. Edit /xxx/uClinux-dist/user/Makefile > 4- I edited the /xxx/uClinux-dist/config/config.in and write the line > in the section of Misc: > bool 'foo' CONFIG_USER_FOO_FOO OK. > 5- I enter in /xxx/uClinux-dist/foo and type "make" and the I got the files: No. pushd /xxx/uClinux-dist && make menuconfig && make dep all && popd > debian:/xxx/uClinux-dist/foo# ls > Makefile foo foo.c foo.o > if I type the command file * see what I got: > debian:/xxx/uClinux-dist/foo# file * > Makefile: ASCII make commands text > foo: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV), > for GNU/Linux 2.2.0, dynamically linked (uses shared libs), not > stripped > foo.c: ASCII C program text > foo.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), > not stripped If done correctly, these should be built for your arch, not for x86. > 6 - After this I type "make romfs" in the /xxx/uClinux-dist/foo > directory and I got the following error: > debian:/xxx/uClinux-dist/foo# make romfs > /bin/foo > make: /bin/foo: Command not found > make: *** [romfs] Error 127 'romfs' is part of 'all'. > Thats all, I hope I wont get too extensive, > What step do I doing wrong?? Step 0 - uClinux-dist/README -- Cory T. Tusar Embedded Systems Engineer Videon Central, Inc. 2171 Sandy Drive State College, PA 16801 (814) 235-1111 x316 (814) 235-1118 fax "Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it." --Brian W. Kernighan From pundrich at gmail.com Wed Mar 30 13:02:46 2005 From: pundrich at gmail.com (Gabriel Pundrich) Date: Wed, 30 Mar 2005 15:02:46 -0300 Subject: [uClinux-dev] Adding my apps In-Reply-To: <424AE82D.2070803@videon-central.com> References: <7c2bef0805033008301ff6237d@mail.gmail.com> <517442a00503300838258ccaf5@mail.gmail.com> <7c2bef08050330091163fea8d2@mail.gmail.com> <424AE82D.2070803@videon-central.com> Message-ID: <7c2bef08050330100266253f9e@mail.gmail.com> Thanks! It works! Gabriel On Wed, 30 Mar 2005 12:55:57 -0500, Cory Tusar wrote: > Gabriel Pundrich wrote: > > > HI, > > I am using this how-to as guide since yesterday and I am having some > > problems like: > > > > 1- I created a directory in /xxx/uClinux-dist/foo as the guide says > > Try /xxx/uClinux-dist/user/foo > > > 2- I put the Makefile: > > > > > >>EXEC = foo > >>OBJS = foo.o > >> > >> all: $(EXEC) > >> > >> $(EXEC): $(OBJS) > >> $(CC) $(LDFLAGS) -o $@ $(OBJS) $(LDLIBS) > >> > >> romfs: > >> $(ROMFSINST) /bin/$(EXEC) > >> > >> clean: > >> -rm -f $(EXEC) *.elf *.gdb *.o > > This should be /xxx/uClinux-dist/user/foo/Makefile > > > 3- I edited the /xxx/uClinux-dist/Makefile and I put in the end of > > the file the line: > > dir_$(CONFIG_USER_FOO_FOO) += foo > > first question---> Its right where I put this line? > > No. Edit /xxx/uClinux-dist/user/Makefile > > > 4- I edited the /xxx/uClinux-dist/config/config.in and write the line > > in the section of Misc: > > bool 'foo' CONFIG_USER_FOO_FOO > > OK. > > > 5- I enter in /xxx/uClinux-dist/foo and type "make" and the I got the files: > > No. pushd /xxx/uClinux-dist && make menuconfig && make dep all && popd > > > debian:/xxx/uClinux-dist/foo# ls > > Makefile foo foo.c foo.o > > if I type the command file * see what I got: > > debian:/xxx/uClinux-dist/foo# file * > > Makefile: ASCII make commands text > > foo: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV), > > for GNU/Linux 2.2.0, dynamically linked (uses shared libs), not > > stripped > > foo.c: ASCII C program text > > foo.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), > > not stripped > > If done correctly, these should be built for your arch, not for x86. > > > 6 - After this I type "make romfs" in the /xxx/uClinux-dist/foo > > directory and I got the following error: > > debian:/xxx/uClinux-dist/foo# make romfs > > /bin/foo > > make: /bin/foo: Command not found > > make: *** [romfs] Error 127 > > 'romfs' is part of 'all'. > > > Thats all, I hope I wont get too extensive, > > What step do I doing wrong?? > > Step 0 - uClinux-dist/README > > -- > Cory T. Tusar > Embedded Systems Engineer > Videon Central, Inc. > 2171 Sandy Drive > State College, PA 16801 > (814) 235-1111 x316 > (814) 235-1118 fax > > "Debugging is twice as hard as writing the code in the first place. > Therefore, if you write the code as cleverly as possible, you are, > by definition, not smart enough to debug it." --Brian W. Kernighan > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uclinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org > From pundrich at gmail.com Wed Mar 30 15:29:30 2005 From: pundrich at gmail.com (Gabriel Pundrich) Date: Wed, 30 Mar 2005 17:29:30 -0300 Subject: [uClinux-dev] c++ Message-ID: <7c2bef0805033012297ce2ad81@mail.gmail.com> Hi, I just get my C app in uclinux with the Adding-User-Apps-HOWTO. Now I am trying to put a C++ user app in uclinux. Its a simples hello.cc like: #include using namespace std; int main() { cout << "hello World!\n"; return 0; } Anybody knows how to do it? Tnx! Gabriel Pundrich From davidm at snapgear.com Wed Mar 30 17:47:05 2005 From: davidm at snapgear.com (David McCullough) Date: Thu, 31 Mar 2005 08:47:05 +1000 Subject: [uClinux-dev] memory allocation problem ! In-Reply-To: <01bd01c5305c$a506fbf0$6501a8c0@umesh> References: <0IDQ00FFLAXKD7@smtp.neusoft.com> <01bd01c5305c$a506fbf0$6501a8c0@umesh> Message-ID: <20050330224705.GB3977@beast> Jivin umesh lays it down ... > > Hi all, > > I am using SnapGear Embedded Linux Distribution having linux kernel version > 2.4.20 > & my borad is haivng 16 MB physical RAM. > > I'm facing a strange problem related to memory allocation via insmod . > whenever I boot the system via flash & I try to insmod my mpeg4 driver > which is big in size i.e.2.6 MB around, system gives a error : > > insmod: allocation of 768 pages failed! > > the mem_map is : > > init Mem: 129248 Slack: 3696 Shared: 0 > sh Mem: 133408 Slack: 3066 Shared: 0 > insmod Mem: 3704704 Slack: 72040 Shared: 0 > Total Mem: 3967360 Slack: 78802 Shared: 0 > > Active: 0, inactive: 5, free: 1397 > Free pages: 1397 (5588kB), %1 frag, %3 slack > Free blks: 17 min=1 max=762 avg=82 > Used blks: 18 min=1 max=1435 avg=149 > > But when I use NFS, I can successfully insmod the same mpeg4 driver without > any memory allocation problem. > Even I tried to make the romfs having only this driver module to ruduce the > size of whole image, > but getting the same error. > Cud somebody help me? whats wrong here? > Either NFS or romfs build , as image size is same except romfs and nfs code > in the kernel, i think both will > occupied the same amount of memory within the RAM after booting as I dont > have any thing within romfs except driver module. > does the driver within romfs which is somewhat big (2.6 mb) creating the > prob ? > why the system is unable to allocate the required memory as my board is > having 11 MB RAM. > Any suggesion is highly apreciated. >from the above numbers, the largest contiguous block you have is ~3MiB. You can see where is says "Free blks: 17 min=1 max=762 avg=82", this means that the largest allocation we can fulfill is 762 blocks. In order to load your driver it needs to allocate 768 pages ! It cannot be done. Why ? And more so, why sometimes and not others ? Well it's all about fragmentation. The more apps you run etc the more likely you are to split large blocks of free memory into smaller ones. The best chance you have is to get /etc/rc to insmod the driver for you very early in the boot so that nothing else is running and your memory is less fragmented. You could run "cat /proc/meminfo" at the start of /etc/rc to see what is available. Another option is to enable /proc/mem_map so that you can get a better idea of who is using the memory and how it is fragmented. From there you may be able to work around the fragmentation, Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From gerg at snapgear.com Wed Mar 30 18:15:37 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Thu, 31 Mar 2005 09:15:37 +1000 Subject: [uClinux-dev] Coldfire 5272:getting stuck at sti()"privilege violation" In-Reply-To: <200503210032.AA281411754@rtimes.com> References: <200503210032.AA281411754@rtimes.com> Message-ID: <424B3319.1020000@snapgear.com> Hi Lekha, Lekha wrote: > Iam trying to port uClinux-dist-20020220(linux2.4.x) to coldfire 5272 board,When booting ,as soon as my program reach the Why are you using such an old code snapshot? The 5272 is well supported in all newer distributions, and is sure to have less bugs :-) Regards Greg > > sti() function( that enables interrupt), my uCLinux system crashes and prints at console the string "privilege violation" > > along with the dump of the Address and Data register. At start up in the crt0_ram .s,Status register(SR) is made 0x2700 > move.w #0x2700, %sr /* No interrupts */ > > (supervisor mode / interrupts disabled).So at sti() ,it must be still in the supervisor mode ,so why "privilege violation" ? > Help please. > > Lekha > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uclinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org > -- ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From davidm at snapgear.com Wed Mar 30 18:35:23 2005 From: davidm at snapgear.com (David McCullough) Date: Thu, 31 Mar 2005 09:35:23 +1000 Subject: [uClinux-dev] c++ In-Reply-To: <7c2bef0805033012297ce2ad81@mail.gmail.com> References: <7c2bef0805033012297ce2ad81@mail.gmail.com> Message-ID: <20050330233522.GD3977@beast> Jivin Gabriel Pundrich lays it down ... > Hi, > I just get my C app in uclinux with the Adding-User-Apps-HOWTO. > Now I am trying to put a C++ user app in uclinux. Its a simples hello.cc like: > > #include > using namespace std; > int main() > { > cout << "hello World!\n"; > return 0; > } > > Anybody knows how to do it? What does your Makefile look like ? Cheers, Davidm -- David McCullough, davidm at snapgear.com Ph:+61 7 34352815 http://www.SnapGear.com Custom Embedded Solutions + Security Fx:+61 7 38913630 http://www.uCdot.org From frnagel at gmail.com Wed Mar 30 20:15:10 2005 From: frnagel at gmail.com (Fernando Remus Nagel) Date: Wed, 30 Mar 2005 22:15:10 -0300 Subject: [uClinux-dev] c++ In-Reply-To: <7c2bef0805033012297ce2ad81@mail.gmail.com> References: <7c2bef0805033012297ce2ad81@mail.gmail.com> Message-ID: <517442a0050330171565a31ef2@mail.gmail.com> tu tem compilador c++ para esta arquitetura? On Wed, 30 Mar 2005 17:29:30 -0300, Gabriel Pundrich wrote: > Hi, > I just get my C app in uclinux with the Adding-User-Apps-HOWTO. > Now I am trying to put a C++ user app in uclinux. Its a simples hello.cc like: > > #include > using namespace std; > int main() > { > cout << "hello World!\n"; > return 0; > } > > Anybody knows how to do it? > > Tnx! > Gabriel Pundrich > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uclinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org > From girish.k at amdlbct.com Wed Mar 30 23:43:31 2005 From: girish.k at amdlbct.com (Girish.K) Date: Thu, 31 Mar 2005 10:13:31 +0530 Subject: [uClinux-dev] System Command Not working. Message-ID: <004701c535ac$30ca52a0$3500a8c0@bctserver.com> Dear All, I working on Coldfire MCF5249, using uClinux Kernel Version 2.0. My host platform is running on Redhat linux 9.0. When I install m68k-elf toolchain in the system and compile my application using the below Makefile. Where in w2wsdklibrary is a archive of our library software. EXEC =test CC = m68k-elf-gcc CFLAGS = -m5307 -DCONFIG_COLDFIRE -I.-Wunused -Wuninitialized -msep-data -Os -g -fomit-frame-pointer -Wl,-elf2flt INCLUDE = -I. LDLIBS = -lw2wsdklibrary SOURCE=test.c LDPATH= -L. DEBUG= all: $(EXEC) $(EXEC): $(CC) -O $(CFLAGS) -o $@ $(INCLUDE) $(SOURCE) $(LDPATH) $(LDLIBS)-lc clean: -rm -f $(EXEC) *.o *.elf *.gdb The application when run on the target platform with uClinux 2.0 kernel, its not able to execute "System" command, but if the compilation is done along with the distribution then the same code works for the system command. Please let me know what is the problem in the method of compilation or the code generation which is resulting in non execution of certain library Functions. Sample code can be as below. The code compiles on both the methods without error. int main() { system("ls "); } In anticipation Girish.K Confidentiality Notice: The information contained in this electronic message and any attachments to this message are intended for the exclusive use of the addressee(s) and may contain confidential or privileged information. If you are not the intended recipient, please notify the sender at AMDLBCT or sysadmin at amdlbct.com immediately and destroy all copies of this message and any attachments. -------------- next part -------------- An HTML attachment was scrubbed... URL: From gerg at snapgear.com Thu Mar 31 01:43:12 2005 From: gerg at snapgear.com (Greg Ungerer) Date: Thu, 31 Mar 2005 16:43:12 +1000 Subject: [uClinux-dev] renaming uClinux-dist In-Reply-To: <20050328151238.E1C46ED99@mail2.atmark-techno.com> References: <20050315050316.10CD5EBDF@mail2.atmark-techno.com> <4236867A.4090004@cox.net> <20050315081337.0A198633A@mail2.atmark-techno.com> <20050319002457.GI11709@pengutronix.de> <423E0DDC.6080809@snapgear.com> <423E311A.3060208@itee.uq.edu.au> <423E5955.8010104@snapgear.com> <20050322162414.2455DE9AB@mail2.atmark-techno.com> <42409C82.5090701@itee.uq.edu.au> <20050328151238.E1C46ED99@mail2.atmark-techno.com> Message-ID: <424B9C00.4020801@snapgear.com> Hi Yashi, Yasushi SHOJI wrote: > I'm back to Japan now. let's discuss little bit about this. > > At Wed, 23 Mar 2005 08:30:26 +1000, > John Williams wrote: >>1. Image problem. The uClinux-dist and uClinux brands are worth >>keeping, but we need a concerted effort to spread the message that uC != >>NOMMU. We can probably all have 2 cents to say about text on the >>uClinux.org site - Michael D is ready for contributions to this effect. > > > As Greg is mentioning, Greg and I are oppose to this idea. Specially > if we keep !MMU kernel as uClinux. (Yes, I don't want to dump the > brand of uClinux kernel) > > how long do we need and how hard should we put our effort break that > association only for the dist? And I really don't think we will ever break it. uClinux is just too well know as the MMUless stuff. The alternate plan (and one that snapgear has sort of followed) is the release the uClinux-dist under a different name. Duel branded if you like. Snapgear-dist is not immediately associated with uClinux - so you at least get past the MMUless conotations. >>3. Flexibility, in terms of release timing for the dist could be better. >> We all recognise SnapGear's effort in maintaining the dist and kernel >>trees, and I don't think anyone would be asking them to do more than >>they already are :) > > > I personally want to put some effort on this. would it be nice to > have intermediate patch-sets between the official dist releases? > > What I'm thinking is that whenever a patch comes in to this ml and at > least one of the dist maintainer says "applied", we keep that patch in > a patch-set, and release it once in a while. I am all for that - if I don't have to do the work of maintaining the patch :-) You would miss patches currently that are sent directly to Dave or myself. (Yep, this happens quite a bit). >>5. Assuming the point about tarball size is settled, the monolithic >>nature of the dist is not a bug, it's a feature :) This is a >>philosophical difference from e.g. PTXdist - each has its merits, and >>there's certainly room for both in this world. It's unlikely that >>hundreds (thousands?) of uClinux-dist users are prepared to re-tool onto >>PTXdist, and vice versa. > > > so, do we all agree on this? I do. Let's think this issue again when > the dist size exceeds a common cheap media, like a CD-RW stands for > nowadays. I have no issue with its size, and haven't seen any convincing arguments that it is really a problem:-) Erwin's point about being able to generate/download a patchset for each uClinux-dist component (kernel, lib, app, etc) is a valid one though. We should really have a way to get at patches from the original for each package. Regards Greg ------------------------------------------------------------------------ Greg Ungerer -- Chief Software Dude EMAIL: gerg at snapgear.com SnapGear -- a CyberGuard Company PHONE: +61 7 3435 2888 825 Stanley St, FAX: +61 7 3891 3630 Woolloongabba, QLD, 4102, Australia WEB: http://www.SnapGear.com From lars.friedrich at wago.com Thu Mar 31 01:50:43 2005 From: lars.friedrich at wago.com (Friedrich, Lars) Date: Thu, 31 Mar 2005 08:50:43 +0200 Subject: [uClinux-dev] System Command Not working. Message-ID: > CFLAGS = -m5307 -DCONFIG_COLDFIRE -I.-Wunused -Wuninitialized -msep-data -Os > -g -fomit-frame-pointer -Wl,-elf2flt - missing space between -I. and -Wunused - obsolete -I. entry - flawed usage of -g flag > $(CC) -O $(CFLAGS) -o $@ $(INCLUDE) $(SOURCE) $(LDPATH) $(LDLIBS)-lc - obsolete -O entry - -lc entry in wrong place > The application when run on the target platform with uClinux 2.0 kernel, > its not able to execute "System" command, This is not an error description, unless when you execute the program, you get a message that states "I'm not able to execute the 'system' command." > but if the compilation is done along with the distribution then the same code works > for the system command. So to sum it up: You have two different command lines to compile an application, one produces a correct result and the other one not. Please compare the two different command lines or if unable to, post them here. And you insist on not compiling it with the distribution why? > Please let me know what is the problem in the method of compilation or the code > generation which is resulting in non execution of certain library Functions. Wild guess: you link the wrong libc, because the library path is incomplete or wrong. Best regards, Lars Friedrich From kunlunzhu at yahoo.com Thu Mar 31 02:07:36 2005 From: kunlunzhu at yahoo.com (kunlunzhu at yahoo.com) Date: Wed, 30 Mar 2005 23:07:36 -0800 (PST) Subject: [uClinux-dev] why the overrun happens in coldfire 5249? In-Reply-To: 6667 Message-ID: <20050331070736.84280.qmail@web81408.mail.yahoo.com> Hi, All: I use UART and ethernet in my board. That's only two external interrupt in my board. And I got overrun in FIFO when I use UART and ethernet together. I check the ethernet driver, its interrupt level is 5. And mcfserial.c use interrupt level 6. I know there is another timer interrupt. It is set at level 5. If UART hold highest interrupt, FIFO of UART should never be overrun. Does any one could explain what could cause this? Your response is highly appreciated. Regards, Kevin From liubl at neusoft.com Thu Mar 31 02:59:44 2005 From: liubl at neusoft.com (liubaoliang) Date: Thu, 31 Mar 2005 15:59:44 +0800 Subject: [uClinux-dev] still the image.ram and image.rom problem Message-ID: <0IE700HH3I4TDM@smtp.neusoft.com> hi uclinux-dev? Excuse me ask the same question again. Our platform is s3c4510b and 2.4.0 version. In my application,we will read/write a EEROM which is a IIC device.If I use the image.ram anything is ok,but if I use the image.rom,sometime I found the result of read or write is fault. who can tell me why? M.r Huaidong Mo told me something about the diffrent hardware init,but I think it's not the key point,and I think the difrent kernel do the same work of hardware init.And I still suspect the compiling options or my source code has some problem. Best regards. ????????? Liu Baoliang 2005-03-31 Company : Philips and Neusoft Medical Systems Co.,Ltd. Address : Neusoft Park Hun Nan Industrial Area New&High-Tech Development Zone Shenyang China 110179 Ph. (86) 024-83665554 Fax (86) 024-23782817 Email : liubl at neusoft.com From eauth at softsys.co.at Thu Mar 31 07:20:27 2005 From: eauth at softsys.co.at (Erwin Authried) Date: Thu, 31 Mar 2005 14:20:27 +0200 Subject: [uClinux-dev] memory allocation problem ! In-Reply-To: <20050330224705.GB3977@beast> References: <0IDQ00FFLAXKD7@smtp.neusoft.com> <01bd01c5305c$a506fbf0$6501a8c0@umesh> <20050330224705.GB3977@beast> Message-ID: <1112271626.2314.78.camel@justakiss> Am Don, den 31.03.2005 schrieb David McCullough um 0:47: ... > Why ? And more so, why sometimes and not others ? Well it's all about > fragmentation. The more apps you run etc the more likely you are to > split large blocks of free memory into smaller ones. > > The best chance you have is to get /etc/rc to insmod the driver for you > very early in the boot so that nothing else is running and your memory > is less fragmented. > Compiling the driver into the kernel is the best option, if you have the choice to do that. Regards, Erwin From pwilshire at cox.net Thu Mar 31 09:00:05 2005 From: pwilshire at cox.net (Phil Wilshire) Date: Thu, 31 Mar 2005 09:00:05 -0500 Subject: [uClinux-dev] April San Jose uClinux Training. Message-ID: <424C0265.2000909@cox.net> Hello, This is a last call reminder for the upcoming uClinux Training in San Jose April 12th-14th 2005, We have only a few seats left in the class. The workshop provides a fast paced comprehensive and complete introduction to the uClinux distribution including many short cuts and tips. The topics are carefully selected to suit both beginners and experts. The target system selected is the Analog Devices Blackfin BF533. This device runs a linux 2.6 kernel and u-boot. Most of the class will apply to any hardware target using this distribution. The course covers the following topics: * Toolchain set up and use * Distribution set up and customization * U boot details * Jtag interface * Development host system Set up ( Including Cygwin ) * Kernel and User code debugging ( NEW item using kgdb over ethernet !) * Kernel Configuration and building * Root file system image generation * User code configuration * User Code basics * Network Code * Intro to Modules and Device Drivers * Interrupt handling * SPI and DMA interfaces * Webserver / CGI Integration to custom applications * Complete Web based Oscilloscope Demo application. More details are here: http://www.sysdcs.com/SDCSucbf.html Thanks Phil Wilshire From ngustavson at emacinc.com Thu Mar 31 09:26:43 2005 From: ngustavson at emacinc.com (NZG) Date: Thu, 31 Mar 2005 08:26:43 -0600 Subject: [uClinux-dev] Tinilogin getting started. Message-ID: <200503310826.43170.ngustavson@emacinc.com> Like many before me I'm getting ready to wade into the whole Tinilogin debacle. Changing the password that is stored in flash dynamically somehow, storing it, and reflashing the board. I've been browsing through the list archives trying to figure out the best way to get started with this, but there is a lot of information there, some of it conflicting, and a lot of it questions without answers. Has anyone written a README? Or somehow simplified the process that could give me a hand? thx, NZG From ngustavson at emacinc.com Thu Mar 31 09:44:13 2005 From: ngustavson at emacinc.com (NZG) Date: Thu, 31 Mar 2005 08:44:13 -0600 Subject: [uClinux-dev] Coldfire Mounting Internal RAM Message-ID: <200503310844.13468.ngustavson@emacinc.com> The new SoM-5282 provides a battery backed up line to the modules internal static RAM, therebye making it non-volatile. I would like to be able to mount this RAM as a disk and put a ext2 filesystem on it. Has anyone done any work in this area? I was thinking of grabbing Rubini's sbull driver from his Linux device drivers book and starting from there. Sound reasonable? thanks for any input, NZG. From kendrick.hamilton at sasktel.net Thu Mar 31 10:16:53 2005 From: kendrick.hamilton at sasktel.net (Kendrick Hamilton) Date: Thu, 31 Mar 2005 09:16:53 -0600 Subject: [uClinux-dev] Tinilogin getting started. In-Reply-To: <200503310826.43170.ngustavson@emacinc.com> References: <200503310826.43170.ngustavson@emacinc.com> Message-ID: <424C1465.7000901@sasktel.net> One way to go about it is set your root filesystem as a ramfs/initial ramfs. Provide a /etc/passwd and (if appropriate) /etc/shadow. Set the console to run a shell, not a login prompt. Power up the unit and run the passwd command then copy the password out of /etc/passwd or /etc/shadow. KGH NZG wrote: >Like many before me I'm getting ready to wade into the whole Tinilogin >debacle. Changing the password that is stored in flash dynamically somehow, >storing it, and reflashing the board. > >I've been browsing through the list archives trying to figure out the best way >to get started with this, but there is a lot of information there, some of it >conflicting, and a lot of it questions without answers. > >Has anyone written a README? >Or somehow simplified the process that could give me a hand? > >thx, >NZG >_______________________________________________ >uClinux-dev mailing list >uClinux-dev at uclinux.org >http://mailman.uclinux.org/mailman/listinfo/uclinux-dev >This message was resent by uclinux-dev at uclinux.org > > From jmh at csm-instruments.com Thu Mar 31 10:22:42 2005 From: jmh at csm-instruments.com (Julien MARBACH) Date: Thu, 31 Mar 2005 17:22:42 +0200 Subject: [uClinux-dev] about scripting languages & uClinux Message-ID: <424C15C2.1000100@csm-instruments.com> Hello, I'm looking for a scripting language to embed in my uClinux Board (DNP-5282). I'm currently focussing on Lua (www.lua.org) wich seems to be a pretty good choice. Has anyone used any scripting langages like lua, python, scheme, ruby, tcl, perl or whatever in a uClinux environement? Hich one seems the most adapted to a small factor uClinux environment? My embedded app will consist on a side in executing script files transmited to the ucLinux Board through http. The scripts will mostly consist in driving I/Os. On the other side the app will transmit data at a high rate (through http also). I don't know yet if I'm going to write the whole app in Lua (including a http server) or just use the lua interpreter to interpret small lua scripts transfered on the board through the boa server. The point is that I need a high tranfert rate and I don't know wich solution will be the fastest. What do you think? Thanks, Julien MARBACH From ngustavson at emacinc.com Thu Mar 31 10:10:23 2005 From: ngustavson at emacinc.com (NZG) Date: Thu, 31 Mar 2005 09:10:23 -0600 Subject: [uClinux-dev] Tinilogin getting started->changing root mount In-Reply-To: <424C1465.7000901@sasktel.net> References: <200503310826.43170.ngustavson@emacinc.com> <424C1465.7000901@sasktel.net> Message-ID: <200503310910.23426.ngustavson@emacinc.com> On Thursday 31 March 2005 09:16, Kendrick Hamilton wrote: > One way to go about it is set your root filesystem as a ramfs/initial > ramfs. Which leads me to my next question, how do you change where the root filesystem is mounted? On a standard debian system it's declared in /etc/fstab, but no such thing seems to exist in uC. Is this a build option? Where is it? thx, NZG. From kendrick.hamilton at sasktel.net Thu Mar 31 11:09:21 2005 From: kendrick.hamilton at sasktel.net (Kendrick Hamilton) Date: Thu, 31 Mar 2005 10:09:21 -0600 Subject: [uClinux-dev] Tinilogin getting started->changing root mount In-Reply-To: <200503310910.23426.ngustavson@emacinc.com> References: <200503310826.43170.ngustavson@emacinc.com> <424C1465.7000901@sasktel.net> <200503310910.23426.ngustavson@emacinc.com> Message-ID: <424C20B1.7030801@sasktel.net> Actually, the root filesystem is mounted by the kernel. Later on your debian system reads /etc/fstab (a file on the root filesystem so it needs to be mounted to be read) to change the mounts. The value of the root filesystem is compiled into the kernel but can be overriden with kernel command line parameters. Mine is: root=/dev/ram In uClinux-2.4.x for coldfire (at least my target), the kernel command line is specified when you do a make menuconfig/config/xconfig. Another thing you could boot up you system normally with the console set to run a shell. After it is running: 1. copy the contents of /etc to /var/etc_tmp 2. manually extract your ramfs.img to /dev/ram1 3. mount /dev/ram1 to /etc 4. copy the contents of /var/etc_tmp to /etc 5. change run passwd KGH NZG wrote: >On Thursday 31 March 2005 09:16, Kendrick Hamilton wrote: > > >>One way to go about it is set your root filesystem as a ramfs/initial >>ramfs. >> >> >Which leads me to my next question, how do you change where the root >filesystem is mounted? On a standard debian system it's declared >in /etc/fstab, but no such thing seems to exist in uC. Is this a build >option? Where is it? > >thx, >NZG. > >_______________________________________________ >uClinux-dev mailing list >uClinux-dev at uclinux.org >http://mailman.uclinux.org/mailman/listinfo/uclinux-dev >This message was resent by uclinux-dev at uclinux.org > > From eauth at softsys.co.at Thu Mar 31 11:24:24 2005 From: eauth at softsys.co.at (Erwin Authried) Date: Thu, 31 Mar 2005 18:24:24 +0200 Subject: [uClinux-dev] Coldfire Mounting Internal RAM In-Reply-To: <200503310844.13468.ngustavson@emacinc.com> References: <200503310844.13468.ngustavson@emacinc.com> Message-ID: <1112286264.2314.91.camel@justakiss> Am Don, den 31.03.2005 schrieb NZG um 16:44: > The new SoM-5282 provides a battery backed up line to the modules internal > static RAM, therebye making it non-volatile. > I would like to be able to mount this RAM as a disk and put a ext2 filesystem > on it. Has anyone done any work in this area? > I was thinking of grabbing Rubini's sbull driver from his Linux device drivers > book and starting from there. Sound reasonable? > > thanks for any input, > NZG. Hi, you can simply use MTD for that. There are already some nvram mappings that you can use as a template. Regards, Erwin From andrewmarshall at orange.net Thu Mar 31 12:21:11 2005 From: andrewmarshall at orange.net (Andrew Marshall) Date: Thu, 31 Mar 2005 18:21:11 +0100 Subject: [uClinux-dev] USB card reader media detection Message-ID: <004701c53616$0925a690$f601a8c0@SenatorPalmer> Hi all, I would very much appreciate any assistance anyone can offer on the following: I am currently running version 2.4.22 of the Kernel and am writing an application that needs to cope with the removal of card media form a USB flash card reader. Hotplug informs me when the card reader itself is added with my intention being to scan for partition entries at the newly created dev/scsi entry e.g /dev/scsi/host0/bus0/target0/lun0/part1 and mount any partitions detected. This works fine -- but I also need to detect when a memory card has been removed. I have found that the block device file (part1) corresponding to my flash card's partition is appropriately updated (i.e removed from /dev/scsi/host0/bus0/target0/lun0/ when the test card is removed) -- but only if the partition is *not* currently mounted! I found that once I have mounted the partition, the devfs 'part1' block device perpetually remains -- even when the card is removed, thus making removal detection a tad tricky. OK, it seems to be possible to unmount the partition, poll the /dev/scsi/host0/bus0/target0/lun0 entry and then remount once the poller has determined that the partition is still present -- but this takes time... And also seems a little extreme... Does anyone have any ideas on a fix or alternative way to determine removal of a flash card from a connected flash card reader. Any suggestions on this problem would be greatly appreciated. Many thanks! Andrew Marshall From ngustavson at emacinc.com Thu Mar 31 15:32:24 2005 From: ngustavson at emacinc.com (NZG) Date: Thu, 31 Mar 2005 14:32:24 -0600 Subject: [uClinux-dev] Coldfire Mounting Internal RAM->MTD documentation In-Reply-To: <1112286264.2314.91.camel@justakiss> References: <200503310844.13468.ngustavson@emacinc.com> <1112286264.2314.91.camel@justakiss> Message-ID: <200503311432.24372.ngustavson@emacinc.com> > you can simply use MTD for that. There are already some nvram mappings > that you can use as a template. > I am banging my head against the wall. It might be simple if I could find some decent documentation for MTD, but all I can find so far is some questions in the archive. Can anyone explain: 1. How are mtd's used? Are the mounted like blocks? Or only used with specific mtd-utils? 2. Can I put a standard filesystem, i.e. ext2,vfat on a mtd device? 3. When mtd is compiled into the kernel, it seems to load a "device table" at boot up, which appears to be mounting the devices somewhere. Where is this? I tried to use the "test driver" in RAM at the SRAM location. It booted up ok and proc reports cat /proc/mtd dev: size erasesize name mtd0: 00010000 00000400 "mtdram test device" I then used mknod to create a device mknod mtd0 b 31 0 which I can run fdisk on, but it reports 0 blocks. if I try to run mke2fs on it I get mtd0 is mounted; will not make a filesystem here! It is? where? 4. Is anyone having luck using this? The goal here is to create a writeable NV file system. Am I headed down the wrong road? thx, NZG On Thursday 31 March 2005 10:24, Erwin Authried wrote: > Am Don, den 31.03.2005 schrieb NZG um 16:44: > > The new SoM-5282 provides a battery backed up line to the modules > > internal static RAM, therebye making it non-volatile. > > I would like to be able to mount this RAM as a disk and put a ext2 > > filesystem on it. Has anyone done any work in this area? > > I was thinking of grabbing Rubini's sbull driver from his Linux device > > drivers book and starting from there. Sound reasonable? > > > > thanks for any input, > > NZG. > > Hi, > you can simply use MTD for that. There are already some nvram mappings > that you can use as a template. > > Regards, > Erwin > > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uclinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org From ngustavson at emacinc.com Thu Mar 31 17:01:24 2005 From: ngustavson at emacinc.com (NZG) Date: Thu, 31 Mar 2005 16:01:24 -0600 Subject: [uClinux-dev] Abandoning MTD for NVRAM block driver In-Reply-To: <200503311432.24372.ngustavson@emacinc.com> References: <200503310844.13468.ngustavson@emacinc.com> <1112286264.2314.91.camel@justakiss> <200503311432.24372.ngustavson@emacinc.com> Message-ID: <200503311601.24660.ngustavson@emacinc.com> I think I'm going to abandon the MTD approach, it's lack of documentation means it's not saving me any time, even if the code I need is buried in there somewhere. I think I'm going to go with a modified sbull approach, which is probably technically more correct as NVRAM doesn't suffer from MTD hardware problems, And I've got a book to document it. thx, NZG. On Thursday 31 March 2005 14:32, NZG wrote: > > you can simply use MTD for that. There are already some nvram mappings > > that you can use as a template. > > I am banging my head against the wall. It might be simple if I could find > some decent documentation for MTD, but all I can find so far is some > questions in the archive. > Can anyone explain: > 1. How are mtd's used? Are the mounted like blocks? Or only used with > specific mtd-utils? > 2. Can I put a standard filesystem, i.e. ext2,vfat on a mtd device? > 3. When mtd is compiled into the kernel, it seems to load a "device table" > at boot up, which appears to be mounting the devices somewhere. Where is > this? > > I tried to use the "test driver" in RAM at the SRAM location. It booted up > ok and proc reports > cat /proc/mtd > dev: size erasesize name > mtd0: 00010000 00000400 "mtdram test device" > I then used mknod to create a device > mknod mtd0 b 31 0 > which I can run fdisk on, but it reports 0 blocks. > if I try to run mke2fs on it I get > mtd0 is mounted; will not make a filesystem here! > It is? where? > > 4. Is anyone having luck using this? The goal here is to create a writeable > NV file system. Am I headed down the wrong road? > > thx, > NZG > > On Thursday 31 March 2005 10:24, Erwin Authried wrote: > > Am Don, den 31.03.2005 schrieb NZG um 16:44: > > > The new SoM-5282 provides a battery backed up line to the modules > > > internal static RAM, therebye making it non-volatile. > > > I would like to be able to mount this RAM as a disk and put a ext2 > > > filesystem on it. Has anyone done any work in this area? > > > I was thinking of grabbing Rubini's sbull driver from his Linux device > > > drivers book and starting from there. Sound reasonable? > > > > > > thanks for any input, > > > NZG. > > > > Hi, > > you can simply use MTD for that. There are already some nvram mappings > > that you can use as a template. > > > > Regards, > > Erwin > > > > > > _______________________________________________ > > uClinux-dev mailing list > > uClinux-dev at uclinux.org > > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > > This message was resent by uclinux-dev at uclinux.org > > _______________________________________________ > uClinux-dev mailing list > uClinux-dev at uclinux.org > http://mailman.uclinux.org/mailman/listinfo/uclinux-dev > This message was resent by uclinux-dev at uclinux.org From gregorb at paradise.net.nz Thu Mar 31 17:39:50 2005 From: gregorb at paradise.net.nz (Gregor Bruce) Date: Fri, 01 Apr 2005 10:39:50 +1200 Subject: [uClinux-dev] about scripting languages & uClinux In-Reply-To: <424C15C2.1000100@csm-instruments.com> References: <424C15C2.1000100@csm-instruments.com> Message-ID: <424C7C36.9000801@paradise.net.nz> Julien MARBACH wrote: > Hello, > I'm looking for a scripting language to embed in my uClinux Board > (DNP-5282). I'm currently focussing on Lua (www.lua.org) wich seems to > be a pretty good choice. Has anyone used any scripting langages like > lua, python, scheme, ruby, tcl, perl or whatever in a uClinux > environement? Hich one seems the most adapted to a small factor > uClinux environment? > Hi, I am working on implementing Tcl on a Coldfire 5275. This is the full Tcl version, not the tinytcl bundled with uClinux. So far I have got the latest alpha release, Tcl8.5a2 99% working. I leave out most of the Tcl lib stuff the binary executable is 626k. Gregor Bruce From Vikrant_Basotra at Satyam.com Thu Mar 31 18:07:09 2005 From: Vikrant_Basotra at Satyam.com (Vikrant_Basotra) Date: Fri, 1 Apr 2005 04:37:09 +0530 Subject: [uClinux-dev] Linux Bring up problem Message-ID: <3D9FDCA910DD4445896A0BE5ECF009D45098CF@bla.satyam.com> Hi All, I am trying to port the EP8248 Linux image (which works perfectly on the EP8248 board) on the VPN board (In house board) which is mpc8248 based board. I do not have any problem with the functioning of U-Boot but Linux crashes after printing "Uncompressing Kernel ... OK" and then it seems that nothing is happening. It would be great if u could help me resolve the issue. The following is how I have configured my board: 1. The value of the IMMR in HCRW is configured to 0xF0000000 (which is same as in the ep8248 image). The value of IMAP_ADDR in linux/arch/ppc/platforms/ep8248.h and u-boot/include/configs/ep8248.h is 0xF0000000 2. The memory map of EP8248 board (REFERENCE BOARD) is: ============================================================================ Memory Region Size Address Range ============================================================================ SDRAM 16MByte 0000_0000--07FF_FFFF Flash 8MByte FF80_0000--FFFF_FFFF 3. The memory map of the VPN Board (MY BOARD) is as follows: =========================================================================== Memory Region Size Address Range ======================================================================== SDRAM 32MByte 0000_0000 -- 01FF_FFFF Flash 32MByte FF00_0000 -- FFFF_FFFF from the u-boot the parameters passed to the kernel (pointer to a function which actually starts loading the kernel image) are : (*kernel) (kbd, initrd_start, initrd_end, cmd_start, cmd_end); kbd (ptr to board info strcture), initrd_start (=0, as in our case we dont have any ramdisk), initrd_end (=0), cmd_start (Start of command line string), cmd_end (End of command line string) 4. Board Info Structure defined in u-boot (/include/asm-ppc/u-boot.h)and Linux (include/asm-ppc/ep8248.h) are the same and is passed correctly from u-boot to Linux. By putting some Printk's I had cheacked that control is reaching to the function mapin_ram(void) in file arch/ppc/mm/pgtable.c Here it is supposed to map the 32 MB of virtual memory to the Physical memory. The control does not leave the for loop in which the memory mapping takes place. This is what the console log buffer has shown: ================================================================= md 0x00177e34 00177e34: 3c363e4d 656d6f72 79204241 54206d61 <6>Memory BAT ma 00177e44: 7070696e 673a2042 7070696e 673a2042 pping: Bpping: B 00177e54: 622c2042 4154333d 33324d62 2c207265 b, BAT3=32Mb, re 00177e64: 73696475 616c3a20 304d620a 3c343e4f sidual: 0Mb.<4>O 00177e74: 6f70733a 206b6572 6e656c20 61636365 ops: kernel acce 00177e84: 7373206f 66206261 64206172 65612c20 ss of bad area, 00177e94: 7369673a 2031310a 3c343e4e 49503a20 sig: 11.<4>NIP: 00177ea4: 30303030 30303030 20584552 3a203030 00000000 XER: 00 00177eb4: 30303030 3030204c 523a2030 30303030 000000 LR: 00000 00177ec4: 30303020 53503a20 30303030 30303030 000 SP: 00000000 00177ed4: 20524547 533a2063 30313530 30303020 REGS: c0150000 00177ee4: 54524150 3a203030 30302020 20204e6f TRAP: 0000 No 00177ef4: 74207461 696e7465 640a3c34 3e4d5352 t tainted.<4>MSR 00177f04: 3a203030 30303030 30302045 453a2030 : 00000000 EE: 0 00177f14: 2050523a 20302046 503a2030 204d453a PR: 0 FP: 0 ME: 00177f24: 20302049 522f4452 3a203030 0a3c343e 0 IR/DR: 00.<4> 00177f34: 5441534b 203d2063 30313466 3437305b TASK = c014f470[ 00177f44: 305d2027 73776170 70657227 204c6173 0] 'swapper' Las 00177f54: 74207379 7363616c 6c3a2030 200a3c34 t syscall: 0 .<4 00177f64: 3e6c6173 74206d61 74682030 30303030 >last math 00000 00177f74: 30303020 6c617374 20616c74 69766563 000 last altivec 00177f84: 20303030 30303030 300a3c34 3e43616c 00000000.<4>Cal 00177f94: 6c206261 636b7472 6163653a 20303030 l backtrace: 000 00177fa4: 30303030 34200a3c 343e4f6f 70733a20 00004 .<4>Oops: 00177fb4: 6b65726e 656c2061 63636573 73206f66 kernel access of 00177fc4: 20626164 20617265 612c2073 69673a20 bad area, sig: 00177fd4: 31310a3c 343e4e49 503a2030 30303030 11.<4>NIP: 00000 00177fe4: 30303020 5845523a 20303030 30303030 000 XER: 0000000 00177ff4: 30204c52 3a203030 30303030 30302053 0 LR: 00000000 S 00178004: 503a2030 30303030 30303020 52454753 P: 00000000 REGS 00178014: 3a206330 31353030 30302054 5241503a : c0150000 TRAP: 00178024: 20303030 30202020 204e6f74 20746169 0000 Not tai 00178034: 6e746564 0a3c343e 4d53523a 20303030 nted.<4>MSR: 000 00178044: 30303030 30204545 3a203020 50523a20 00000 EE: 0 PR: 00178054: 30204650 3a203020 4d453a20 30204952 0 FP: 0 ME: 0 IR 00178064: 2f44523a 2030300a 3c343e54 41534b20 /DR: 00.<4>TASK 00178074: 3d206330 31346634 37305b30 5d202773 = c014f470[0] 's 00178084: 77617070 65722720 4c617374 20737973 wapper' Last sys 00178094: 63616c6c 3a203020 0a3c343e 6c617374 call: 0 .<4>last 001780a4: 206d6174 68203030 30303030 3030206c math 00000000 l 001780b4: 61737420 616c7469 76656320 30303030 ast altivec 0000 001780c4: 30303030 0a3c343e 43616c6c 20626163 0000.<4>Call bac 001780d4: 6b747261 63653a20 30303030 30303034 ktrace: 00000004 001780e4: 200a3c34 3e4f6f70 733a206b 65726e65 .<4>Oops: kerne 001780f4: 6c206163 63657373 206f6620 62616420 l access of bad 00178104: 61726561 2c207369 673a2031 310a3c34 area, sig: 11.<4 00178114: 3e4e4950 3a203030 30303030 30312058 >NIP: 00000001 X 00178124: 45523a20 30303030 30303030 204c523a ER: 00000000 LR: 00178134: 20433031 35303044 30205350 3a203030 C01500D0 SP: 00 00178144: 30303030 30302052 4547533a 20633031 000000 REGS: c01 00178154: 35303030 30205452 41503a20 31303332 50000 TRAP: 1032 00178164: 20202020 4e6f7420 7461696e 7465640a Not tainted. 00178174: 3c343e4d 53523a20 66666666 66666666 <4>MSR: ffffffff 00178184: 2045453a 20312050 523a2031 2046503a EE: 1 PR: 1 FP: 00178194: 2031204d 453a2031 2049522f 44523a20 1 ME: 1 IR/DR: 001781a4: 31310a3c 343e5441 534b203d 20633031 11.<4>TASK = c01 001781b4: 34663437 305b305d 20277377 61707065 4f470[0] 'swappe 001781c4: 7227204c 61737420 73797363 616c6c3a r' Last syscall: 001781d4: 2030200a 3c343e6c 61737420 6d617468 0 .<4>last math 001781e4: 20303030 30303030 30206c61 73742061 00000000 last a 001781f4: 6c746976 65632030 30303030 3030300a ltivec 00000000. 00178204: 3c343eff ffffffc0 1502c0c0 15024361 <4>...........Ca 00178214: 6c6c2062 61636b74 72616365 3a203030 ll backtrace: 00 00178224: 30303030 3034200a 3c343e4f 6f70733a 000004 .<4>Oops: 00178234: 206b6572 6e656c20 61636365 7373206f kernel access o 00178244: 66206261 64206172 65612c20 7369673a f bad area, sig: 00178254: 2031310a 3c343e4e 49503a20 30303030 11.<4>NIP: 0000 00178264: 30303031 20584552 3a203030 30303030 0001 XER: 000000 00178274: 3030204c 523a2043 30313530 30443020 00 LR: C01500D0 00178284: 53503a20 46464646 46434330 20524547 SP: FFFFFCC0 REG 00178294: 533a2063 30313530 30303020 54524150 S: c0150000 TRAP 001782a4: 3a203130 33322020 20204e6f 74207461 : 1032 Not ta 001782b4: 696e7465 640a3c34 3e4d5352 3a206666 inted.<4>MSR: ff 001782c4: 66666666 66662045 453a2031 2050523a ffffff EE: 1 PR: 001782d4: 20312046 503a2031 204d453a 20312049 1 FP: 1 ME: 1 I 001782e4: 522f4452 3a203131 0a3c343e 5441534b R/DR: 11.<4>TASK 001782f4: 203d2063 30313466 3437305b 305d2027 = c014f470[0] ' 00178304: 73772720 4c617374 20737973 63616c6c sw' Last syscall 00178314: 3a202d31 30373337 31383034 30200a3c : -1073718040 .< 00178324: 343e6c61 7374206d 61746820 30303030 4>last math 0000 00178334: 30303030 206c6173 7420616c 74697665 0000 last altive 00178344: 63203030 30303030 30300a3c 343effff c 00000000.<4>.. 00178354: ffffc015 02c0c015 02000000 00000000 ................ 1.Is all this information sufficient to help you understand and pin point our mistake, Is there any thing other than the parameters which are passed to the kernel from the u-boot that I need to look into. 2. I have a query in the Linux source code. I would like to know why the value of "ioremap_base = 0xfe000000UL;" in linux/arch/ppc/mm/init.c line 356 has been hard coded and how was this value determined? What is the impact of this value? 3. In Reference board (EP8248, RAM size is 16MB) the PAGESIZE is defined as 4K, where as in our Board RAM Size is 32 MB. By putting Printk statement in the Pagefault function, I Came to know that the control is coming to this function and hence pagefault is occurring.Do I need to change the PAGESIZE to 8K Any pointers, ideas, comments would be of great help. i would be more than glad to provide you with any further information required to debug the problem. Best Regards , Vikrant ************************************************************************** This email (including any attachments) is intended for the sole use of the intended recipient/s and may contain material that is CONFIDENTIAL AND PRIVATE COMPANY INFORMATION. Any review or reliance by others or copying or distribution or forwarding of any or all of the contents in this message is STRICTLY PROHIBITED. If you are not the intended recipient, please contact the sender by email and delete all copies; your cooperation in this regard is appreciated. ************************************************************************** From liubl at neusoft.com Thu Mar 31 19:22:00 2005 From: liubl at neusoft.com (liubaoliang) Date: Fri, 01 Apr 2005 08:22:00 +0800 Subject: [uClinux-dev] how to realize a image.rom without compressed Message-ID: <0IE800HDCRLXDM@smtp.neusoft.com> hi uclinux-dev? Cause I found the diffrent exceuting result between the image.ram and image.rom,so I want get a image.ro whithout compressed ,how to do this? Best regards. ????????? Liu Baoliang 2005-04-01 Ph. (86) 024-83665554 Fax (86) 024-23782817 Email : liubl at neusoft.com From ben at thingmagic.com Thu Mar 31 19:36:52 2005 From: ben at thingmagic.com (Ben Giddings) Date: Thu, 31 Mar 2005 19:36:52 -0500 Subject: [uClinux-dev] How do I use MatrixSSL with uClinux Message-ID: <200503311936.53020.ben@thingmagic.com> Hi all, I'm running uClinux (actually "snapgear 3.2") on an Intel XScale system, and am trying to use MatrixSSL to provide SSL support to mbedthis appWeb. I replaced the version of appWeb that comes with the distribution with a newer one from the mbedthis site. This fixes an issue I was having with external CGIs, and it also provides the MatrixSSL module support, however it doesn't supply the MatrixSSL module itself. I found the matrixssl module, and tried compiling it with standard CC var settings (make CC=arm-linux-gcc ...) but it balks at -lcrypto. On my desktop system, /usr/lib/libcrypto.so is provided by openssl. If MatrixSSL is an alternative to OpenSSL it shouldn't require OpenSSL's libcrypto, right? The reason I'm trying to get MatrixSSL working is that it's supposed to be smaller (file-size) than OpenSSL. I understand it's difficult to get OpenSSL's library under 1MB, and that it adds a lot of code to apps to support it. Does anybody have any success stories in trimming down OpenSSL? Ben