Jiri Olsa wrote:
> Jiri Kosina wrote:
>> On Mon, 3 Mar 2008, Jiri Olsa wrote:
>>
>>> ipcs.kernel.out.1 is the kernel log from the time strace ipcs was done.
>>> ipcs.kernel.out.2 is from the next run of ipcs that stuck,
>>> there's no overlay between those logs.
>>> When the kernel freeze I'm getting logs with the spinlock lockup.. and it's repeating.
>> Thanks. Have you tried turning lockdep off, if that still triggers?
>>
>> Aparently the first BUG leaves the spin lock locked, and all subsequent 
>> accessess to it lock the machine hard. We'll need to see what goes wrong 
>> inside lockdep.
>>
>> Could you please run
>>
>> 	addr2line -e vmlinux c0133034
>>
>> (passing the path to the vmlinux binary of the crashing kernel to the -e 
>> parameter).
>>
> 
> I've rebuilt the kernel meanwhile, so dont know how accurate this output is:
> 
> /home/jirka/xxx/kernel/linux-2.6-stable/include/asm/atomic_32.h:97
> 
> Anyway, I'll run the test again with your patch, and then we can run the addrline again...
> I'll try to turn the lockdep off as well.

I applied you diff and ran the test again..


your patch with LOCKDEP enabled:

ipcs.strace.1		- strace of ipcs from the first run
ipcs.kernel.out.1 	- kernel logs from 1st strace ipcs run
ipcs.kernel.out.2 	- kernel logs from 2nd strace ipcs run - system freezed


your patch with LOCKDEP disabled:

ipcs.kernel.out-NO_LOCKDEP.1 - kernel logs from 1st strace ipcs run - system freezed


maybe I overlooked smth, but following is the only way I found 
how to disable LOCKDEP, hope it is fine... let me know:

diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 80b7ba4..2bfd8dc 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -43,7 +43,7 @@ config GENERIC_CLOCKEVENTS_BROADCAST
 
 config LOCKDEP_SUPPORT
        bool
-       default y
+       default n
 
 config STACKTRACE_SUPPORT
        bool