Re: RELENG_5 panic: mtx_lock() with ACPI

Anish Mistry wrote:
> On Thursday 30 September 2004 03:06 pm, Poul-Henning Kamp wrote:
> 
>>In message <415C4F5A.2080701_at_root.org>, Nate Lawson writes:
>>
>>>I assume phk would be handling this since it's in the tty code.  If not,
>>>please let us know so one of us can address this before the release.
>>>
>>>
>>>>>>struct tty *
>>>>>>ttymalloc(struct tty *tp)
>>>>>>{
>>>>>>       static int once;
>>>>>>
>>>>>>       if (!once) {
>>>>>>               mtx_init(&tty_list_mutex, "ttylist", NULL, MTX_DEF);
>>>>>>               once++;
>>>>>>       }
>>>>>>
>>>>>>This code is not MP safe as multiple processors could attempt to call
>>>>>>mtx_init() twice.
>>
>>We have no calls to ttymalloc() which are not protected by Giant,
>>so this theory doesn't explain the problem.
> 
> Any progress, I'd hate to have to disable ACPI for the release.  Is there any 
> way I could provide more info?

I'll try to look at it on the bus tonight.  See if you can narrow down 
if one ACPI component is contributing to this, i.e. 
debug.acpi.disabled="cpu" or whatever (see acpi.4 for the full list). 
The sysctl run from the rc scripts to get entropy is what triggers the 
panic.

I have no idea how acpi can affect the tty mutex.

-- 
Nate