[uClinux-dev] 5272, bootloader, and "Calibrating Delay Loops.."

Travis Griggs tgriggs at keyww.com
Tue Nov 6 11:03:07 EST 2001


Ian Leonard wrote:

> 
> On 2001.11.05 23:11 Travis Griggs wrote:
> 
>> I am thrilled to say that we got this working! Thank you, thank you, 
>> thank you to those who provided the helpful hints a couple of days 
>> ago. Especially Greg Ungerer. What you write the registers (long vs 
>> short vs char) as can make a big difference. :)
>>
>> I've attached the working version. We're a bit unclear to as to where 
>> to go with it from here. We don't have an ftp site of our own readily 
>> available. But we figured others might find it useful. We've added a 
>> README.5272 to the directory which documents what we added and some of 
>> our findings.
> 
> 
> Hi Travis,
> 
> Well done, you are a hero!
> 
> Unfortunately I can't make it work. This is my inexperience. If I
> download the image via ftp (dn -i colilo.bin). There appears to be
> a 0x400 byte offset to the start of the code. At first I thought this
> was an offset into the flash device. However my manual says
> it should be 0x100.
> 
> If I move it to the flash (upuser) and move JP13, nothing happens.
> 
> In fact I am generally confused and would appreciate your help.
> BTW, I have a Motorola evaluation board (M5272C3).

The blind leading the blind here. :) I am trying to recollect what the 
above (dn -i colilo.bin) does. If I remember right, this loads the code 
into RAM at 0x20000 (or there abouts). You want to be a little more 
drastic than that. You want to overwrite the dBug program that is in 
flash with colilo. To do that, you'll need a flash programmer. Maybe 
there's a way to do it with dBug, I dunno. This is the one piece I still 
have to use Windows for, because the flash programmer we have only works 
on windows. If anyone has any tips on what tools/methods/techniques are 
available for programming flash from a host Linux box, I would really 
appreciate it. As it is, we used the following process:

1) attach our BDM cable
2) reset the chip (from the programmer)
3) erase the module
4) flash the bootloader (colilo)
5) flash uClinux
6) power the board down, pull the BDM cable, and repower

4 & 5 are reversible. For our flash programmer, we HAD to use srecord 
formats, and they had to be S3 type records.

The bootloader lives at 0xffe00000. It must be there. The default 
location of the uClinux image should 0xffe20000. At one point, we were 
playing with running it (the bootloader) from 0xffe40000, which dBug 
defines as "user" flash. To make this work, I *think* the linker script 
would have to be modified a bit (and maybe a parameter in the C code). 
Also, the default address of uClinux would probably be somewhere beyond 
the bootloader (that would be a chance in main.c). We gave up, wondering 
if dBug and colilo were messing each other up. In retrospect, I don't 
think they were, and this indeed might actually work. I may have to give 
that a play. Going that route would allow you to have both dBug AND 
colilo + uClinux on there. Which it booted would be dependent on JP17. 
And you would be able to use dBug to flash stuff.

-- 
Travis Griggs
Member: 3rd Boolean State Software Collective
Key Technology
"It had better be a pretty good meeting to be better than no meeting at 
all"-- Boyd K. Packer

This message resent by the uclinux-dev at uclinux.org list server http://www.uClinux.org/



More information about the uClinux-dev mailing list