[uClinux-dev] Problem "make"ing uClinux

Kendrick Hamilton hamilton at sedsystems.ca
Tue Mar 18 17:10:02 EST 2003


I noticed that you are using the 2.4 68360 kernel. Is somebody monitoring
this to  check for things like this?
You might want to try the 2.0.x kernel. I know that it works.
KGH


On Wed, 19 Mar 2003, John Williams wrote:

> Hi Andrew,
>
> Chizhik, Andrew wrote:
> > Looking at the lines mentioned here, there is no "sw_usp" there. For
> > example, line 125 in process.c reads:
> > printk("USP: %081x\n", rdusp());
>
> > I assume the problem lies in the call to rdusp(). What does that function
> > do? What do these errors mean?
>
> sw_usp is a location in kernel memory where the User Stack Pointer
> (hence the name "usp") is saved upon entry to the kernel.  This is
> because the 5307 has no concept of a user/supervisor stack pointer.
>
> To that end, rdusp() is probably a macro, defined in
> /include/asm-m68knommu somewhere.  Ah yes, there it is, an inline
> function defined in processor.h
>
>
> extern inline unsigned long rdusp(void) {
> #ifdef CONFIG_COLDFIRE
>          extern unsigned int     sw_usp;
>          return(sw_usp);
> #else
>          unsigned long usp;
>
>          __asm__ __volatile__("move %/usp,%0" : "=a" (usp));
>          return usp;
> #endif
> }
>
> It looks like the configuration for your platform is defining
> CONFIG_COLDFIRE, when perhaps it shouldn't.  I'm no m68k expert - does
> the 68360 have hardware user/supervisor stack pointers, or is it
> emulated in /platform/XXX/entry.S ?  In /platform/5307/entry.S you can
> see how the sw_usp is used on entry /exit to the kernel.
>
> Sorry this is not a solution, just some expanatory stuff.  The
> coldfire/m68k experts should have more for details for you!
>
> Regards,
>
> John
>
> _______________________________________________
> 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
>




More information about the uClinux-dev mailing list