cvs commit: src/sys/contrib/dev/acpica acconfig.h acenv.h acfreebsd.h acgcc.h acpi.h acpiosxf.h acpixf.h acutils.h aecommon.h dbfileio.c dmopcode.c exfldio.c exsystem.c hwregs.c osunixxf.c psparse.c rscreate.c rsdump.c utdebug.c uteval.c ...

[Nate Lawson]

Nate Lawson wrote:
> Jung-uk Kim wrote:
>> On Friday 23 March 2007 06:37 pm, Pawel Jakub Dawidek wrote:
>>> On Thu, Mar 22, 2007 at 05:58:28PM +0000, Jung-uk Kim wrote:
>>>> jkim        2007-03-22 17:58:28 UTC
>>>>
>>>>   FreeBSD src repository
>>>>
>>>>   Modified files:
>>>>     sys/contrib/dev/acpica acconfig.h acenv.h acfreebsd.h acgcc.h
>>>>                            acpi.h acpiosxf.h acpixf.h acutils.h
>>>>                            aecommon.h dbfileio.c dmopcode.c
>>>>                            exfldio.c exsystem.c hwregs.c
>>>>                            osunixxf.c psparse.c rscreate.c
>>>>                            rsdump.c utdebug.c uteval.c utglobal.c
>>>>   Log:
>>>>   Resolve conflicts of unchanged files that are off the vendor
>>>> branch.
>>> Can you verify that LOCK_PROFILING(9) still works after your
>>> import? After adding it to me kernel I got the panic below:
>> --- >8 --- SKIP --- >8 ---
>>
>> It works for me, i.e., I cannot reproduce the problem.
>>
>> # sysctl debug.lock.prof
>> debug.lock.prof.stats: No locking recorded
>> debug.lock.prof.collisions: 0
>> debug.lock.prof.hashsize: 4096
>> debug.lock.prof.rejected: 0
>> debug.lock.prof.maxrecords: 4096
>> debug.lock.prof.records: 0
>> debug.lock.prof.acquisitions: 0
>> debug.lock.prof.enable: 0
>> # sysctl debug.lock.prof.enable=1
>> debug.lock.prof.enable: 0 -> 1
>> # sysctl debug.lock.prof.records
>> debug.lock.prof.records: 690
>>
>> Do you have any other special option in the kernel config file?
> 
> I'm wondering if acpi-ca now allows recursive acquisitions of mutexes by
> AML code.  So seeing his acpidump may be helpful (acpidump -dt | gzip
> -c9 > pjd.asl.gz).
> 

Actually, first cvsup and see if my commit helped.

-- 
Nate