NFSv4.1/pnfs: Fix a close/delegreturn hang when return-on-close is set

The helper pnfs_roc() has already verified that we have no delegations,
and no further open files, hence no outstanding I/O and it has marked
all the return-on-close lsegs as being invalid.
Furthermore, it sets the NFS_LAYOUT_RETURN bit, thus serialising the
close/delegreturn with all future layoutget calls on this inode.

The checks in pnfs_roc_drain() for valid layout segments are therefore
redundant: those cannot exist until another layoutget completes.
The other check for whether or not NFS_LAYOUT_RETURN is set, actually
causes a hang, since we already know that we hold that flag.

To fix, we therefore strip out all the functionality in pnfs_roc_drain()
except the retrieval of the barrier state, and then rename the function
accordingly.

Reported-by: Christoph Hellwig <hch@infradead.org>
Fixes: 5c4a79fb2b1c ("Don't prevent layoutgets when doing return-on-close")
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>

diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index f4e5816a77b070f337d9b653660aaeeba61ddc27..bda7837dfe6b98720c6b299fd77ede10202aae5b 100644 (file)
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -2737,11 +2737,8 @@ static void nfs4_close_prepare(struct rpc_task *task, void *data)
 
        if (calldata->arg.fmode == 0) {
                task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CLOSE];
-               if (calldata->roc &&
-                   pnfs_roc_drain(inode, &calldata->roc_barrier, task)) {
-                       nfs_release_seqid(calldata->arg.seqid);
-                       goto out_wait;
-                   }
+               if (calldata->roc)
+                       pnfs_roc_get_barrier(inode, &calldata->roc_barrier);
        }
        calldata->arg.share_access =
                nfs4_map_atomic_open_share(NFS_SERVER(inode),
@@ -5289,9 +5286,8 @@ static void nfs4_delegreturn_prepare(struct rpc_task *task, void *data)
 
        d_data = (struct nfs4_delegreturndata *)data;
 
-       if (d_data->roc &&
-           pnfs_roc_drain(d_data->inode, &d_data->roc_barrier, task))
-               return;
+       if (d_data->roc)
+               pnfs_roc_get_barrier(d_data->inode, &d_data->roc_barrier);
 
        nfs4_setup_sequence(d_data->res.server,
                        &d_data->args.seq_args,

diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index 6151f39c82916551dfbdeae8a2604ad596ce58ac..6aabbb6540216f9d23ff54cd79fcd402b2dd6e7a 100644 (file)
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -1141,25 +1141,14 @@ void pnfs_roc_set_barrier(struct inode *ino, u32 barrier)
        spin_unlock(&ino->i_lock);
 }
 
-bool pnfs_roc_drain(struct inode *ino, u32 *barrier, struct rpc_task *task)
+void pnfs_roc_get_barrier(struct inode *ino, u32 *barrier)
 {
        struct nfs_inode *nfsi = NFS_I(ino);
        struct pnfs_layout_hdr *lo;
-       struct pnfs_layout_segment *lseg;
        nfs4_stateid stateid;
        u32 current_seqid;
-       bool layoutreturn = false;
 
        spin_lock(&ino->i_lock);
-       list_for_each_entry(lseg, &nfsi->layout->plh_segs, pls_list) {
-               if (!test_bit(NFS_LSEG_ROC, &lseg->pls_flags))
-                       continue;
-               if (test_bit(NFS_LSEG_VALID, &lseg->pls_flags))
-                       continue;
-               rpc_sleep_on(&NFS_SERVER(ino)->roc_rpcwaitq, task, NULL);
-               spin_unlock(&ino->i_lock);
-               return true;
-       }
        lo = nfsi->layout;
        current_seqid = be32_to_cpu(lo->plh_stateid.seqid);
 
@@ -1168,18 +1157,7 @@ bool pnfs_roc_drain(struct inode *ino, u32 *barrier, struct rpc_task *task)
         */
        *barrier = current_seqid + atomic_read(&lo->plh_outstanding);
        stateid = lo->plh_stateid;
-       if (test_and_clear_bit(NFS_LAYOUT_RETURN_BEFORE_CLOSE,
-                                          &lo->plh_flags))
-               layoutreturn = pnfs_prepare_layoutreturn(lo);
-       if (test_bit(NFS_LAYOUT_RETURN, &lo->plh_flags))
-               rpc_sleep_on(&NFS_SERVER(ino)->roc_rpcwaitq, task, NULL);
-
        spin_unlock(&ino->i_lock);
-       if (layoutreturn) {
-               pnfs_send_layoutreturn(lo, stateid, IOMODE_ANY, false);
-               return true;
-       }
-       return false;
 }
 
 /*

diff --git a/fs/nfs/pnfs.h b/fs/nfs/pnfs.h
index 738672a0f8da975503a7318512397d35e213fb3c..a3d57a8fac762781a0c1bb50eb0c5e9da822ec37 100644 (file)
--- a/fs/nfs/pnfs.h
+++ b/fs/nfs/pnfs.h
@@ -267,7 +267,7 @@ int pnfs_mark_matching_lsegs_invalid(struct pnfs_layout_hdr *lo,
 bool pnfs_roc(struct inode *ino);
 void pnfs_roc_release(struct inode *ino);
 void pnfs_roc_set_barrier(struct inode *ino, u32 barrier);
-bool pnfs_roc_drain(struct inode *ino, u32 *barrier, struct rpc_task *task);
+void pnfs_roc_get_barrier(struct inode *ino, u32 *barrier);
 void pnfs_set_layoutcommit(struct inode *, struct pnfs_layout_segment *, loff_t);
 void pnfs_cleanup_layoutcommit(struct nfs4_layoutcommit_data *data);
 int pnfs_layoutcommit_inode(struct inode *inode, bool sync);
@@ -605,10 +605,9 @@ pnfs_roc_set_barrier(struct inode *ino, u32 barrier)
 {
 }
 
-static inline bool
-pnfs_roc_drain(struct inode *ino, u32 *barrier, struct rpc_task *task)
+static inline void
+pnfs_roc_get_barrier(struct inode *ino, u32 *barrier)
 {
-       return false;
 }
 
 static inline void set_pnfs_layoutdriver(struct nfs_server *s,