Re: ahc/ahd locking

[Peter Avalos <pavalos@xxxxxxxxxxxx>]

On Sun, Jan 13, 2008 at 07:25:38PM -0500, Peter Avalos wrote:
> 
> Indeed.  I'm running a test kernel now that's using lockmgr locks. This
> is the patchset:
> 
> http://www.theshell.com/~pavalos/wip/aic7xxx-locking5.patch
> 

And I just got a crash:

panic: softdep_deallocate_dependencies: dangling deps
mp_lock = 00000000; cpuid = 0
Trace beginning at frame 0xd9f4e89c
panic(d9f4e8c0,c4050f54,c4050f54,e912ff60,d9f4e8c0) at panic+0x14d
panic(c0363180,d9f4e8fc,c01dc7f1,c4050f54,21020164) at panic+0x14d
softdep_deallocate_dependencies(c4050f54) at softdep_deallocate_dependencies+0x19
brelse(c4050f54) at brelse+0x1ff
bqrelse(c4050f54) at bqrelse+0x145
biodone(c4050ff4,da130818,c4050f54) at biodone+0x463
dadone(c3c63fb8,e5f33e50,c039d31c,ff800000,d9f4ecb4) at dadone+0x2d8
camisr(d9f4ed84,c01841a3,0,0,43b1) at camisr+0x29b
swi_cambio(0,0) at swi_cambio+0xd
ithread_handler(43,0,0,0,0) at ithread_handler+0x123
lwkt_exit() at lwkt_exit
boot() called on cpu#0
Uptime: 4h48m49s

dumping to dev #da/0x20001, blockno 378927
dump panic: softdep_deallocate_dependencies: dangling deps
mp_lock = 00000000; cpuid = 0
boot() called on cpu#0
Uptime: 4h48m50s
Dump already in progress, bailing...

And so on...

So my question is there something wrong with the locking in ahd that is
causing cam (da) to have problems, or is this just a coincidence and the
real problem is in cam (da). Unfortunately, no crash dump available.

--Peter

Attachment: pgp00008.pgp
Description: PGP signature