uma_zalloc_arg complaining about non-sleepable locks

[Marius Strobl]

On Tue, Jan 26, 2010 at 09:46:44AM -0500, John Baldwin wrote:
> On Tuesday 26 January 2010 2:33:37 am Peter Jeremy wrote:
> > I have just upgraded to 8-STABLE/amd64 from about 18 hours ago and am
> > now getting regular (the following pair of messages about every
> > minute) compaints as follows:
> > 
> > kernel: uma_zalloc_arg: zone "mbuf" with the following non-sleepable locks held:
> > kernel: exclusive sleep mutex sp_lock (sp_lock) r = 0 (0xffffff000460bb00) locked @ /usr/src/sys/rpc/svc.c:1098
> > kernel: KDB: stack backtrace:
> > kernel: db_trace_self_wrapper() at db_trace_self_wrapper+0x2a
> > kernel: _witness_debugger() at _witness_debugger+0x2c
> > kernel: witness_warn() at witness_warn+0x2c2
> > kernel: uma_zalloc_arg() at uma_zalloc_arg+0x29d
> > kernel: nfs_realign() at nfs_realign+0x5f
> > kernel: fha_assign() at fha_assign+0x2d8
> > kernel: svc_run_internal() at svc_run_internal+0x1ee
> > kernel: svc_thread_start() at svc_thread_start+0xb
> > kernel: fork_exit() at fork_exit+0x112
> > kernel: fork_trampoline() at fork_trampoline+0xe
> > kernel: --- trap 0xc, rip = 0x80069e04c, rsp = 0x7fffffffe6d8, rbp = 0x5 ---
> > kernel: uma_zalloc_arg: zone "mbuf" with the following non-sleepable locks held:
> > kernel: exclusive sleep mutex sp_lock (sp_lock) r = 0 (0xffffff000460bb00) locked @ /usr/src/sys/rpc/svc.c:1098
> > kernel: KDB: stack backtrace:
> > kernel: db_trace_self_wrapper() at db_trace_self_wrapper+0x2a
> > kernel: _witness_debugger() at _witness_debugger+0x2c
> > kernel: witness_warn() at witness_warn+0x2c2
> > kernel: uma_zalloc_arg() at uma_zalloc_arg+0x29d
> > kernel: nfs_realign() at nfs_realign+0x5f
> > kernel: fha_assign() at fha_assign+0x2d8
> > kernel: svc_run_internal() at svc_run_internal+0x1ee
> > kernel: svc_thread_start() at svc_thread_start+0xb
> > kernel: fork_exit() at fork_exit+0x112
> > kernel: fork_trampoline() at fork_trampoline+0xe
> > kernel: --- trap 0xc, rip = 0x80069e04c, rsp = 0x7fffffffe6d8, rbp = 0x5 ---
> > 
> > It looks like NFS is missing some lock/unlock pairs.  Has anyone else
> > seen this?  And does anyone have a fix?
> 
> I suspect this was caused by the recent alignment fixes to NFS.  I've cc'd
> Marius.
> 

Could you please give the following patch a try?
http://people.freebsd.org/~marius/fha_extract_info_realign2.diff

It would also be great if one of the NFS gurus could have a look at
the whole issue. Unfortunately, I hadn't received a reply regarding
the original patch.

Marius