Merge tag 'nfs-for-5.2-1' of git://git.linux-nfs.org/projects/anna/linux-nfs

Pull NFS client updates from Anna Schumaker:
 "Highlights include:

  Stable bugfixes:
   - Fall back to MDS if no deviceid is found rather than aborting   # v4.11+
   - NFS4: Fix v4.0 client state corruption when mount

  Features:
   - Much improved handling of soft mounts with NFS v4.0:
       - Reduce risk of false positive timeouts
       - Faster failover of reads and writes after a timeout
       - Added a "softerr" mount option to return ETIMEDOUT instead of
         EIO to the application after a timeout
   - Increase number of xprtrdma backchannel requests
   - Add additional xprtrdma tracepoints
   - Improved send completion batching for xprtrdma

  Other bugfixes and cleanups:
   - Return -EINVAL when NFS v4.2 is passed an invalid dedup mode
   - Reduce usage of GFP_ATOMIC pages in SUNRPC
   - Various minor NFS over RDMA cleanups and bugfixes
   - Use the correct container namespace for upcalls
   - Don't share superblocks between user namespaces
   - Various other container fixes
   - Make nfs_match_client() killable to prevent soft lockups
   - Don't mark all open state for recovery when handling recallable
     state revoked flag"

* tag 'nfs-for-5.2-1' of git://git.linux-nfs.org/projects/anna/linux-nfs: (69 commits)
  SUNRPC: Rebalance a kref in auth_gss.c
  NFS: Fix a double unlock from nfs_match,get_client
  nfs: pass the correct prototype to read_cache_page
  NFSv4: don't mark all open state for recovery when handling recallable state revoked flag
  SUNRPC: Fix an error code in gss_alloc_msg()
  SUNRPC: task should be exit if encode return EKEYEXPIRED more times
  NFS4: Fix v4.0 client state corruption when mount
  PNFS fallback to MDS if no deviceid found
  NFS: make nfs_match_client killable
  lockd: Store the lockd client credential in struct nlm_host
  NFS: When mounting, don't share filesystems between different user namespaces
  NFS: Convert NFSv2 to use the container user namespace
  NFSv4: Convert the NFS client idmapper to use the container user namespace
  NFS: Convert NFSv3 to use the container user namespace
  SUNRPC: Use namespace of listening daemon in the client AUTH_GSS upcall
  SUNRPC: Use the client user namespace when encoding creds
  NFS: Store the credential of the mount process in the nfs_server
  SUNRPC: Cache cred of process creating the rpc_client
  xprtrdma: Remove stale comment
  xprtrdma: Update comments that reference ib_drain_qp
  ...

1 files changed, 7 insertions, 24 deletions

diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index 4899b85f9b3c..144e183250c3 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -147,7 +147,7 @@ nfs_file_flush(struct file *file, fl_owner_t id)
 		return 0;
 
 	/* Flush writes to the server and return any errors */
-	return vfs_fsync(file, 0);
+	return nfs_wb_all(inode);
 }
 
 ssize_t
@@ -199,13 +199,6 @@ EXPORT_SYMBOL_GPL(nfs_file_mmap);
  * Flush any dirty pages for this process, and check for write errors.
  * The return status from this call provides a reliable indication of
  * whether any write errors occurred for this process.
- *
- * Notice that it clears the NFS_CONTEXT_ERROR_WRITE before synching to
- * disk, but it retrieves and clears ctx->error after synching, despite
- * the two being set at the same time in nfs_context_set_write_error().
- * This is because the former is used to notify the _next_ call to
- * nfs_file_write() that a write error occurred, and hence cause it to
- * fall back to doing a synchronous write.
  */
 static int
 nfs_file_fsync_commit(struct file *file, int datasync)
@@ -220,11 +213,8 @@ nfs_file_fsync_commit(struct file *file, int datasync)
 	nfs_inc_stats(inode, NFSIOS_VFSFSYNC);
 	do_resend = test_and_clear_bit(NFS_CONTEXT_RESEND_WRITES, &ctx->flags);
 	status = nfs_commit_inode(inode, FLUSH_SYNC);
-	if (test_bit(NFS_CONTEXT_ERROR_WRITE, &ctx->flags)) {
-		ret = xchg(&ctx->error, 0);
-		if (ret)
-			goto out;
-	}
+	if (status == 0)
+		status = file_check_and_advance_wb_err(file);
 	if (status < 0) {
 		ret = status;
 		goto out;
@@ -245,13 +235,7 @@ nfs_file_fsync(struct file *file, loff_t start, loff_t end, int datasync)
 	trace_nfs_fsync_enter(inode);
 
 	do {
-		struct nfs_open_context *ctx = nfs_file_open_context(file);
-		ret = filemap_write_and_wait_range(inode->i_mapping, start, end);
-		if (test_and_clear_bit(NFS_CONTEXT_ERROR_WRITE, &ctx->flags)) {
-			int ret2 = xchg(&ctx->error, 0);
-			if (ret2)
-				ret = ret2;
-		}
+		ret = file_write_and_wait_range(file, start, end);
 		if (ret != 0)
 			break;
 		ret = nfs_file_fsync_commit(file, datasync);
@@ -600,8 +584,7 @@ static int nfs_need_check_write(struct file *filp, struct inode *inode)
 	struct nfs_open_context *ctx;
 
 	ctx = nfs_file_open_context(filp);
-	if (test_bit(NFS_CONTEXT_ERROR_WRITE, &ctx->flags) ||
-	    nfs_ctx_key_to_expire(ctx, inode))
+	if (nfs_ctx_key_to_expire(ctx, inode))
 		return 1;
 	return 0;
 }
@@ -655,7 +638,7 @@ ssize_t nfs_file_write(struct kiocb *iocb, struct iov_iter *from)
 
 	/* Return error values */
 	if (nfs_need_check_write(file, inode)) {
-		int err = vfs_fsync(file, 0);
+		int err = nfs_wb_all(inode);
 		if (err < 0)
 			result = err;
 	}
@@ -709,7 +692,7 @@ do_unlk(struct file *filp, int cmd, struct file_lock *fl, int is_local)
 	 * Flush all pending writes before doing anything
 	 * with locks..
 	 */
-	vfs_fsync(filp, 0);
+	nfs_wb_all(inode);
 
 	l_ctx = nfs_get_lock_context(nfs_file_open_context(filp));
 	if (!IS_ERR(l_ctx)) {