Re: panic with 5.99.55 starting postgres

[Marc Balmer <mbalmer%NetBSD.org@localhost>]

Am 26.08.11 20:35, schrieb Darren Reed:
> J. Hannken-Illjes wrote:
>> On Aug 25, 2011, at 11:18 PM, Darren Reed wrote:
>>
>>  
>>> starting postgres panic's my system every time with this panic:
>>>
>>> panic: kernel diagnostic assertion "uvm_page_locked_p(pg
>>> e " ../../../../arch/x86/x86/pmap.c, line 3214
>>>
>>>
>>> From the pmap.c that was used to compile the kernel is this
>>> extract, with **** being the line that panic's the system.
>>>
>>>       pg = PHYS_TO_VM_PAGE(pmap_pte2pa(opte));
>>>
>>>       KASSERTMSG(pg != NULL, ("pmap_remove_pte: unmanaged page marked "
>>>           "PG_PVLIST, va = %#" PRIxVADDR ", pa = %#" PRIxPADDR,
>>>           va, (paddr_t)pmap_pte2pa(opte)));
>>>
>>> ****    KASSERT(uvm_page_locked_p(pg));
>>>
>>>       /* Sync R/M bits. */
>>>
>>> Seen before?
>>> Known?
>>> Fixed? ;)
>>>
>>> And perhaps more importantly, does anyone know when and where
>>> this was introduced?
>>>
>>> What's the last safe version of NetBSD to use postgres with?
>>>     
>>
>> PR kern/45177 already addresses this one.  It is not a problem with
>> postgres but with the kernel failing O_DIRECT.
>>   
> 
> So I've worked around the problem by just commenting out O_DIRECT in
> <sys/fcntl.h> and recompiling.
>


That is not really a solution... Someone should fix the problem, maybe
the one who broke it in the first place..