Merge tag 'nfs-for-4.8-4' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

Pull NFS client bugfixes from Trond Myklebust:
 "Highlights include:

  Stable patches:
   - We must serialise LAYOUTGET and LAYOUTRETURN to ensure correct
     state accounting
   - Fix the CREATE_SESSION slot number

  Bugfixes:
   - sunrpc: fix a UDP memory accounting regression
   - NFS: Fix an error reporting regression in nfs_file_write()
   - pNFS: Fix further layout stateid issues
   - RPC/rdma: Revert 3d4cf35bd4fa ("xprtrdma: Reply buffer
     exhaustion...")
   - RPC/rdma: Fix receive buffer accounting"

* tag 'nfs-for-4.8-4' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
  NFSv4.1: Fix the CREATE_SESSION slot number accounting
  xprtrdma: Fix receive buffer accounting
  xprtrdma: Revert 3d4cf35bd4fa ("xprtrdma: Reply buffer exhaustion...")
  pNFS: Don't forget the layout stateid if there are outstanding LAYOUTGETs
  pNFS: Clear out all layout segments if the server unsets lrp->res.lrs_present
  pNFS: Fix pnfs_set_layout_stateid() to clear NFS_LAYOUT_INVALID_STID
  pNFS: Ensure LAYOUTGET and LAYOUTRETURN are properly serialised
  NFS: Fix error reporting in nfs_file_write()
  sunrpc: fix UDP memory accounting

3 files changed, 44 insertions, 24 deletions

diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index 7d620970f2e1..ca699ddc11c1 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -657,7 +657,10 @@ ssize_t nfs_file_write(struct kiocb *iocb, struct iov_iter *from)
         if (result <= 0)
                 goto out;
 
-        written = generic_write_sync(iocb, result);
+        result = generic_write_sync(iocb, result);
+        if (result < 0)
+                goto out;
+        written = result;
         iocb->ki_pos += written;
 
         /* Return error values */
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index f5aecaabcb7c..a9dec32ba9ba 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -7570,12 +7570,20 @@ static int _nfs4_proc_create_session(struct nfs_client *clp,
         status = rpc_call_sync(session->clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
         trace_nfs4_create_session(clp, status);
 
+        switch (status) {
+        case -NFS4ERR_STALE_CLIENTID:
+        case -NFS4ERR_DELAY:
+        case -ETIMEDOUT:
+        case -EACCES:
+        case -EAGAIN:
+                goto out;
+        };
+
+        clp->cl_seqid++;
         if (!status) {
                 /* Verify the session's negotiated channel_attrs values */
                 status = nfs4_verify_channel_attrs(&args, &res);
                 /* Increment the clientid slot sequence id */
-                if (clp->cl_seqid == res.seqid)
-                        clp->cl_seqid++;
                 if (status)
                         goto out;
                 nfs4_update_session(session, &res);
@@ -8190,10 +8198,13 @@ static void nfs4_layoutreturn_release(void *calldata)
 
         dprintk("--> %s\n", __func__);
         spin_lock(&lo->plh_inode->i_lock);
-        pnfs_mark_matching_lsegs_invalid(lo, &freeme, &lrp->args.range,
-                        be32_to_cpu(lrp->args.stateid.seqid));
-        if (lrp->res.lrs_present && pnfs_layout_is_valid(lo))
+        if (lrp->res.lrs_present) {
+                pnfs_mark_matching_lsegs_invalid(lo, &freeme,
+                                &lrp->args.range,
+                                be32_to_cpu(lrp->args.stateid.seqid));
                 pnfs_set_layout_stateid(lo, &lrp->res.stateid, true);
+        } else
+                pnfs_mark_layout_stateid_invalid(lo, &freeme);
         pnfs_clear_layoutreturn_waitbit(lo);
         spin_unlock(&lo->plh_inode->i_lock);
         nfs4_sequence_free_slot(&lrp->res.seq_res);
diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index 6daf034645c8..2c93a85eda51 100644
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -365,7 +365,8 @@ pnfs_layout_remove_lseg(struct pnfs_layout_hdr *lo,
         /* Matched by pnfs_get_layout_hdr in pnfs_layout_insert_lseg */
         atomic_dec(&lo->plh_refcount);
         if (list_empty(&lo->plh_segs)) {
-                set_bit(NFS_LAYOUT_INVALID_STID, &lo->plh_flags);
+                if (atomic_read(&lo->plh_outstanding) == 0)
+                        set_bit(NFS_LAYOUT_INVALID_STID, &lo->plh_flags);
                 clear_bit(NFS_LAYOUT_BULK_RECALL, &lo->plh_flags);
         }
         rpc_wake_up(&NFS_SERVER(inode)->roc_rpcwaitq);
@@ -768,17 +769,32 @@ pnfs_destroy_all_layouts(struct nfs_client *clp)
         pnfs_destroy_layouts_byclid(clp, false);
 }
 
+static void
+pnfs_clear_layoutreturn_info(struct pnfs_layout_hdr *lo)
+{
+        lo->plh_return_iomode = 0;
+        lo->plh_return_seq = 0;
+        clear_bit(NFS_LAYOUT_RETURN_REQUESTED, &lo->plh_flags);
+}
+
 /* update lo->plh_stateid with new if is more recent */
 void
 pnfs_set_layout_stateid(struct pnfs_layout_hdr *lo, const nfs4_stateid *new,
                         bool update_barrier)
 {
         u32 oldseq, newseq, new_barrier = 0;
-        bool invalid = !pnfs_layout_is_valid(lo);
 
         oldseq = be32_to_cpu(lo->plh_stateid.seqid);
         newseq = be32_to_cpu(new->seqid);
-        if (invalid || pnfs_seqid_is_newer(newseq, oldseq)) {
+
+        if (!pnfs_layout_is_valid(lo)) {
+                nfs4_stateid_copy(&lo->plh_stateid, new);
+                lo->plh_barrier = newseq;
+                pnfs_clear_layoutreturn_info(lo);
+                clear_bit(NFS_LAYOUT_INVALID_STID, &lo->plh_flags);
+                return;
+        }
+        if (pnfs_seqid_is_newer(newseq, oldseq)) {
                 nfs4_stateid_copy(&lo->plh_stateid, new);
                 /*
                  * Because of wraparound, we want to keep the barrier
@@ -790,7 +806,7 @@ pnfs_set_layout_stateid(struct pnfs_layout_hdr *lo, const nfs4_stateid *new,
                 new_barrier = be32_to_cpu(new->seqid);
         else if (new_barrier == 0)
                 return;
-        if (invalid || pnfs_seqid_is_newer(new_barrier, lo->plh_barrier))
+        if (pnfs_seqid_is_newer(new_barrier, lo->plh_barrier))
                 lo->plh_barrier = new_barrier;
 }
 
@@ -886,19 +902,14 @@ void pnfs_clear_layoutreturn_waitbit(struct pnfs_layout_hdr *lo)
         rpc_wake_up(&NFS_SERVER(lo->plh_inode)->roc_rpcwaitq);
 }
 
-static void
-pnfs_clear_layoutreturn_info(struct pnfs_layout_hdr *lo)
-{
-        lo->plh_return_iomode = 0;
-        lo->plh_return_seq = 0;
-        clear_bit(NFS_LAYOUT_RETURN_REQUESTED, &lo->plh_flags);
-}
-
 static bool
 pnfs_prepare_layoutreturn(struct pnfs_layout_hdr *lo,
                 nfs4_stateid *stateid,
                 enum pnfs_iomode *iomode)
 {
+        /* Serialise LAYOUTGET/LAYOUTRETURN */
+        if (atomic_read(&lo->plh_outstanding) != 0)
+                return false;
         if (test_and_set_bit(NFS_LAYOUT_RETURN, &lo->plh_flags))
                 return false;
         pnfs_get_layout_hdr(lo);
@@ -1798,16 +1809,11 @@ pnfs_layout_process(struct nfs4_layoutget *lgp)
                  */
                 pnfs_mark_layout_stateid_invalid(lo, &free_me);
 
-                nfs4_stateid_copy(&lo->plh_stateid, &res->stateid);
-                lo->plh_barrier = be32_to_cpu(res->stateid.seqid);
+                pnfs_set_layout_stateid(lo, &res->stateid, true);
         }
 
         pnfs_get_lseg(lseg);
         pnfs_layout_insert_lseg(lo, lseg, &free_me);
-        if (!pnfs_layout_is_valid(lo)) {
-                pnfs_clear_layoutreturn_info(lo);
-                clear_bit(NFS_LAYOUT_INVALID_STID, &lo->plh_flags);
-        }
 
 
         if (res->return_on_close)