Re: OOPS in sysctl.c

[Holger Schurig <hs4233@mail.mn-solutions.de>]

On Wednesday 06 February 2008 20:03:53 you wrote:
> Please, post .config and "ls -lR /proc/sys" output _before_
> suspend/resume.

Just a note: I don't use vanilla v2.6.24, but v2.6.24 plus a 
bunch of patches on top of it. I minimized them and for v2.6.24, 
this is the minimal number of patches needed to get the device 
up & running and also supending & resuming:

$ quilt series
+ rt3000-core.patch
+ rt3000-defconfig.patch
+ rt3000-smc91x-mac.patch
+ mn-keyb.patch
+ mn-smc91x.patch
+ mn-suspend-ignore-console.patch
= ARM-pxa-fix-PXA27x-resume.patch

rt3000-core: this is the RT3000/RT4000 machine specific code, 
mostly in arch/arm/mach-pxa

rt3000-smc91x-mac.patch: code that put's a random MAC into the 
smc91x chip

mn-keyb.pm: my keyboard driver

mn-smc91x.patch: port and IRQ definitions for the SMC91111 
ethernet chip

mn-suspend-ignore-console.patch: a patch for kernel/power/main.c 
that puts #if 0...#endif around all calls to 
pm_prepare_console(), pm_restore_console(), suspend_console() 
and resume_console(). Without this patch the device would hang 
while suspending.

ARM-pxa-fix-PXA27x-resume.patch: the same as "git show 
dd01b2fc79a567ae03d0c96ddf61eb4de729d36d" from linux-git. 
Without this patch, the device would hang while resuming.

I'm using the first 6 patches since 2.6.13-rc2, just with little 
updates to the changing API or changing Makefile/Kconfig files. 

When I apply exactly the same 6 patches to v2.6.24-7284-g9ef9dc6, 
then the OOPS doesn't happen. Therefore I assume that the 
culprit is not in one of these patches. It might just be 
important :-)