Re: crash in tmpfs_itimes since vmlocking2 merge

[Antti Kantee <pooka%cs.hut.fi@localhost>]

On Wed Jan 23 2008 at 12:14:58 +0000, Chavdar Ivanov wrote:
> Indeed the mfs panic has gone. Now I get - with a Release build
> overnight on i386 -
> 
> syncing disks... done
> unmounting file systems...uvm_fault(0xca7f6ab8, 0, 1) -> 0xe
> kernel: supervisor trap page fault, code=0
> Stopped in pid 4779.1 (reboot) at    netbsd:fstrans_getstate+0x6:
> movl 0x920(%eax),%eax
> 
> db> bt
> fstrans_getstate(0,0,cc362a4c,c02c7786,c0a67424) at 
> netbsd:fstrans_getstate+0x6
> ffs_full_fsync(ca7f9c80,1,0,0,0) at netbsd:ffs_full_fsync+0xa7
> ffs_fsync(cc362b1c,cba8dc84,ffffffff,0,cba8d25c) at netbsd:ffs_fsync+0x4e
> VOP_FSYNC(ca7f9c80,ca7e4f00,1,0,0) at netbsd:VOP_FSYNC+0x41
> ffs_flushfiles(c15a1000,2,cba86060,c025d148,cba8dc84) at
> netbsd:ffs_flushfiles+0x8d
> ffs_unmount(c15a1000,80000,,0,0,cc362d00) at netbsd:ffs_unmount+0x48
> dounmount(c15a1000,80000,cba86060,0,cc362d00) at netbsd:dounmount+0xc0
> vfs_unmountall(cba86060,0,0,0,cc362c94) at netbsd:vfs_unmountall+0x68
> vfs_shutdown(ca7e4f00,0,cc362cdc,c02a14bd,0) at netbsd:vfs_shutdown+0x89
> cpu_reboot(0,0,0,0,c139f4fc) at netbsd:cpu_reboot+0x3b
> sys_reboot(cba86060,cc362d00,cc362d28,0,81e9000) at netbsd:sys_reboot+0x65
> syscall(cc362d48,b3,ab,1f,1f) at netbsd:syscall+0xca
> 
> [snip]
> 
> after selecting 'reboot' from the sysinst menu.

Uuuh, brilliant.  I knew there was a reason my mother warned me about
this stuff.

Can you exit to shell from sysinst and show what the mount table looks
like before reboot.  It seems that the device entry for the ffs causing
the panic is on a file system which is unmounted before ffs, but I can't
quite wrap my head around how that would happen.  /dev is on the root
mfs on the install, no?

-- 
Antti Kantee <pooka%iki.fi@localhost>                     Of course he runs 
NetBSD
http://www.iki.fi/pooka/                          http://www.NetBSD.org/
    "la qualité la plus indispensable du cuisinier est l'exactitude"