On Fri, 2012-10-12 at 19:56 -0700, Larry McVoy wrote:
> On Sat, Oct 13, 2012 at 02:52:51AM +0000, Myklebust, Trond wrote:
> > On Fri, 2012-10-12 at 19:31 -0700, Larry McVoy wrote:
> > > As I've said, debugging this is going to be hard, it means stopping my
> > > company. We can do one more crash, you guys can tell me what to do,
> > > but that's about it.
> >
> > Do you have a reproducer?
>
> Oh yeah, it happens every time. Run skype on a mac w/ your home dir NFS
> mounted. Skype tries to do some lock operation and the machine panics.
>
> We've done it 4 times in a row to be sure we weren't smoking crack. No
> crack. Just panics.
OK. I think I can see what is happening. We need to filter the return
values from nlm_lookup_file() when dealing with NLMv1 and NLMv3.
Can you please try the attached patch?
Cheers
Trond
PS: you are presumably running NFSv2 on your Macs. Odd that they should
default to that...
--
Trond Myklebust
Linux NFS client maintainer
NetApp
Trond.Myklebust@xxxxxxxxxx
www.netapp.com
From 1f0d4d1bb77fa0216dd2fb2041cc9b44db4697e0 Mon Sep 17 00:00:00 2001
From: Trond Myklebust <Trond.Myklebust@xxxxxxxxxx>
Date: Sat, 13 Oct 2012 00:30:28 -0400
Subject: [PATCH] NLM: nlm_lookup_file() may return NLMv4-specific error codes
If the filehandle is stale, or open access is denied for some reason,
nlm_fopen() may return one of the NLMv4-specific error codes nlm4_stale_fh
or nlm4_failed. These get passed right through nlm_lookup_file(),
and so when nlmsvc_retrieve_args() calls the latter, it needs to filter
the result through the cast_status() machinery.
Failure to do so, will trigger the BUG_ON() in encode_nlm_stat...
Signed-off-by: Trond Myklebust <Trond.Myklebust@xxxxxxxxxx>
---
fs/lockd/clntxdr.c | 2 +-
fs/lockd/svcproc.c | 3 ++-
2 files changed, 3 insertions(+), 2 deletions(-)
diff --git a/fs/lockd/clntxdr.c b/fs/lockd/clntxdr.c
index 180ac34..cf04cda 100644
--- a/fs/lockd/clntxdr.c
+++ b/fs/lockd/clntxdr.c
@@ -223,7 +223,7 @@ static void encode_nlm_stat(struct xdr_stream *xdr,
{
__be32 *p;
- BUG_ON(be32_to_cpu(stat) > NLM_LCK_DENIED_GRACE_PERIOD);
+ WARN_ON_ONCE(be32_to_cpu(stat) > NLM_LCK_DENIED_GRACE_PERIOD);
p = xdr_reserve_space(xdr, 4);
*p = stat;
}
diff --git a/fs/lockd/svcproc.c b/fs/lockd/svcproc.c
index d27aab1..d413af3 100644
--- a/fs/lockd/svcproc.c
+++ b/fs/lockd/svcproc.c
@@ -67,7 +67,8 @@ nlmsvc_retrieve_args(struct svc_rqst *rqstp, struct nlm_args *argp,
/* Obtain file pointer. Not used by FREE_ALL call. */
if (filp != NULL) {
- if ((error = nlm_lookup_file(rqstp, &file, &lock->fh)) != 0)
+ error = cast_status(nlm_lookup_file(rqstp, &file, &lock->fh));
+ if (error != 0)
goto no_locks;
*filp = file;
--
1.7.11.7
