Re: nfs client lockdep warning

[Peter Zijlstra <a.p.zijlstra@chello.nl>]

On Fri, 2009-01-09 at 17:23 +1100, Herbert Xu wrote:
> On Fri, Jan 09, 2009 at 06:58:17AM +0100, Peter Zijlstra wrote:
> >
> > > > > > Jan  8 15:29:17 ying3 kernel:        [<c054bb26>] copy_from_user+0x36/0x90
> > > > > > Jan  8 15:29:17 ying3 kernel:        [<c0737800>] memcpy_fromiovecend+0x60/0x90
> > > > > > Jan  8 15:29:17 ying3 kernel:        [<c075ce9f>] ip_generic_getfrag+0x9f/0xb0
> > > > > > Jan  8 15:29:17 ying3 kernel:        [<c075cb6b>] ip_append_data+0x6bb/0x950
> > > > > > Jan  8 15:29:17 ying3 kernel:        [<c077cb0d>] udp_sendmsg+0x2bd/0x680
> > > > > > Jan  8 15:29:17 ying3 kernel:        [<c07835f9>] inet_sendmsg+0x39/0x70
> > > > > > Jan  8 15:29:17 ying3 kernel:        [<c072f0e6>] sock_sendmsg+0xd6/0x100
> > > > > > Jan  8 15:29:17 ying3 kernel:        [<c072f485>] kernel_sendmsg+0x35/0x50
> > > > > > Jan  8 15:29:17 ying3 kernel:        [<c07da698>] xs_send_kvec+0xa8/0xb0
> > > > > > Jan  8 15:29:17 ying3 kernel:        [<c07da711>] xs_sendpages+0x71/0x210
> > > > > > Jan  8 15:29:17 ying3 kernel:        [<c07dab30>] xs_udp_send_request+0x40/0x120
> > > > > > Jan  8 15:29:17 ying3 kernel:        [<c07d91ff>] xprt_transmit+0x6f/0x280
> > > > > > Jan  8 15:29:17 ying3 kernel:        [<c07d6c1f>] call_transmit+0x17f/0x250
> > > > > > Jan  8 15:29:17 ying3 kernel:        [<c07dda13>] __rpc_execute+0x73/0x250
> > 
> > Is there anything pinning that user-space page, if not its free to
> > fault, which would mean its a real warning... 
> 
> It's kernel memory, not user memory.  See xs_udp_send_request in
> net/sunrpc.

Ahh, so something like this ought to fix it I think...

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
---
diff --git a/mm/memory.c b/mm/memory.c
index 7b9db65..a2ed52e 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -3079,6 +3079,9 @@ void print_vma_addr(char *prefix, unsigned long ip)
 #ifdef CONFIG_PROVE_LOCKING
 void might_fault(void)
 {
+	if (get_fs() == KERNEL_DS)
+		return;
+
 	might_sleep();
 	/*
 	 * it would be nicer only to annotate paths which are not under