From: J. Bruce Fields Date: Wed, 28 Sep 2011 01:42:29 +0000 (-0400) Subject: nfsd4: fix state lock usage in LOCKU X-Git-Url: https://git.karo-electronics.de/?a=commitdiff_plain;h=71c3bcd71393a9e67d5b77597a612537f89c30ed;p=linux-beck.git nfsd4: fix state lock usage in LOCKU In commit 5ec094c1096ab3bb795651855d53f18daa26afde "nfsd4: extend state lock over seqid replay logic" I modified the exit logic of all the seqid-based procedures except nfsd4_locku(). Fix the oversight. The result of the bug was a double-unlock while handling the LOCKU procedure, and a warning like: [ 142.150014] WARNING: at kernel/mutex-debug.c:78 debug_mutex_unlock+0xda/0xe0() ... [ 142.152927] Pid: 742, comm: nfsd Not tainted 3.1.0-rc1-SLIM+ #9 [ 142.152927] Call Trace: [ 142.152927] [] warn_slowpath_common+0x7f/0xc0 [ 142.152927] [] warn_slowpath_null+0x1a/0x20 [ 142.152927] [] debug_mutex_unlock+0xda/0xe0 [ 142.152927] [] __mutex_unlock_slowpath+0x80/0x140 [ 142.152927] [] mutex_unlock+0xe/0x10 [ 142.152927] [] nfs4_lock_state+0x35/0x40 [nfsd] [ 142.152927] [] nfsd4_proc_compound+0x2a1/0x690 [nfsd] [ 142.152927] [] nfsd_dispatch+0xeb/0x230 [nfsd] [ 142.152927] [] svc_process_common+0x345/0x690 [sunrpc] [ 142.152927] [] ? try_to_wake_up+0x280/0x280 [ 142.152927] [] svc_process+0x102/0x150 [sunrpc] [ 142.152927] [] nfsd+0xbd/0x160 [nfsd] [ 142.152927] [] ? 0xffffffffa039efff [ 142.152927] [] kthread+0x8c/0xa0 [ 142.152927] [] kernel_thread_helper+0x4/0x10 [ 142.152927] [] ? kthread_worker_fn+0x190/0x190 [ 142.152927] [] ? gs_change+0x13/0x13 Reported-by: Bryan Schumaker Tested-by: Bryan Schumaker Signed-off-by: J. Bruce Fields --- diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index 931155f51ecc..8e253a347649 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -4185,7 +4185,8 @@ nfsd4_locku(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, memcpy(&locku->lu_stateid, &stp->st_stid.sc_stateid, sizeof(stateid_t)); out: - nfs4_unlock_state(); + if (!cstate->replay_owner) + nfs4_unlock_state(); return status; out_nfserr: