ahc/ahd locking

[Peter Avalos]

On Sun, Jan 13, 2008 at 05:40:13PM -0800, Matthew Dillon wrote:
> 
>     That seems more softupdatesish.  I haven't seen that particular panic
>     in a long time but it's possible that something got corrupted on-disk
>     during your testing.  Softupdates is ultra sensitive to disk corruption.
> 
>     If cam/ahc/ahd were losing track of an IO it would most like panic
>     with 'biodone: bp %p already done!' or something like that.
> 

So I got a crashdump on this last one:

panic: softdep_deallocate_dependencies: dangling deps
mp_lock = 00000000; cpuid = 0
boot() called on cpu#0
Uptime: 52m47s

dumping to dev #da/0x20001, blockno 378927

(kgdb) bt
#0  dumpsys () at thread.h:83
#1  0xc01a3081 in boot (howto=260) at /usr/src/sys/kern/kern_shutdown.c:375
#2  0xc01a3344 in panic (fmt=Variable "fmt" is not available.
) at /usr/src/sys/kern/kern_shutdown.c:800
#3  0xc028931c in softdep_deallocate_dependencies (bp=0x0) at /usr/src/sys/vfs/ufs/ffs_softdep.c:5141
#4  0xc01dc7f1 in brelse (bp=0xc3e85a5c) at buf2.h:204
#5  0xc01dd0bb in bqrelse (bp=0xc3e85a5c) at /usr/src/sys/kern/vfs_bio.c:1406
#6  0xc0287b05 in ffs_balloc (ap=0xe8bffad4) at /usr/src/sys/vfs/ufs/ffs_balloc.c:290
#7  0xc01f4f3f in vop_balloc (ops=0xd359eb70, vp=0xe8d0fea8, startoffset=Unhandled dwarf expression opcode 0x93
) at /usr/src/sys/kern/vfs_vopops.c:731
#8  0xc02913c8 in ffs_write (ap=0xe8bffba0) at ufs_readwrite.c:331
#9  0xc01f592a in vop_write (ops=0xd359eb70, vp=0xe8d0fea8, uio=0xe8bffca0, ioflag=8323075, cred=0xe892ed18) at /usr/src/sys/kern/vfs_vopops.c:351
#10 0xc01f4957 in vn_write (fp=0xd704c510, uio=0xe8bffca0, cred=0xe892ed18, flags=0) at /usr/src/sys/kern/vfs_vnops.c:708
#11 0xc01c0c11 in kern_pwritev (fd=3, auio=0xe8bffca0, flags=0, res=0xe8bffcf0) at file2.h:72
#12 0xc01c1064 in sys_write (uap=0xe8bffcf0) at /usr/src/sys/kern/sys_generic.c:327
#13 0xc031703e in syscall2 (frame=0xe8bffd40) at /usr/src/sys/platform/pc32/i386/trap.c:1345
#14 0xc0304025 in Xint0x80_syscall () at /usr/src/sys/platform/pc32/i386/exception.s:872
#15 0x280b3d48 in ?? ()

You think this is the ahd driver doing something wrong because of the
locking patch?

--Peter
Attachment:
pgp00009.pgp
-------------- next part --------------
A non-text attachment was scrubbed...
Name: pgp00009.pgp
Type: application/octet-stream
Size: 189 bytes
Desc: "Description: PGP signature"
URL: <http://lists.dragonflybsd.org/pipermail/kernel/attachments/20080113/e352c329/attachment-0020.obj>