Re: Reboots when /proc is mounted

[Sad Clouds <cryintothebluesky%googlemail.com@localhost>]

On Saturday 23 January 2010 19:27:26 Sad Clouds wrote:
> dell$ uname -a
> NetBSD dell 5.0_STABLE NetBSD 5.0_STABLE (GENERIC) #1: Sat Jan 23 15:05:16
>  GMT 2010  root@p3smp:/opt/obj.amd64/sys/arch/amd64/compile/GENERIC amd64
> 
> On this machine when I try building lang/mono and /proc is mounted, the
> machine gets rebooted. It looks like this package triggers a bug, because
>  when I umount /proc and try building the same package, everything works
>  fine.
> 
> I think I've seen this problem before some time ago. I'll try to get a
>  crash dump and get a back trace when I have some time.
> 
> Has anyone seen this before and knows where the problem is?
> 

OK, below is the back trace. I think this is the same bug as kern/40549 which 
has been open for a year now.

Reader / writer lock error: rw_vector_enter: locking against myself

lock address: 0xffff800052ca8710
current cpu: 1
current lwp: 0xffff80004bcb6000
owner/count: 0xffff80004bcb6000 flags: 0x0000000000000004

panic: lock error
fatal breakpoint trap in supervisor mode
trap type 1 code 0 rip ffffffff80507095 cs 8 rflags 246 cr2 7f7ffd9b910c cpl 0
rsp ffff80004bcdd400
Stopped in pid 19567.1 (gtar) at netbsd:breakpoint+0x5 leave

bt

breakpoint() at netbsd:breakpoint+0x5
panic() at netbsd:panic+0x246
lockdebug_abort() at netbsd:lockdebug_abort+0x3a
rw_vector_enter() at netbsd:rw_vector_enter+0x25b
vlockmgr() at netbsd:vlockmgr+0xb8
VOP_LOCK() at netbsd:VOP_LOCK+0x64
vn_lock() at netbsd:vn_lock+0x69
getcwd_common() at netbsd:getcwd_common+0x81
procfs_readlink() at netbsd:procfs_readlink+0x34e
layer_bypass() at netbsd:layer_bypass+0xfe
VOP_READLINK() at netbsd:VOP_READLINK+0x69
namei() at netbsd:namei+0x257
do_sys_stat() at netbsd:do_sys_stat+0x44
sys__stat30() at netbsd:sys__stat30+0x28
syscall() at netbsd:syscall+0xab