[PATCH 0/9] Assorted patches for knfsd

[PATCH 0/9] Assorted patches for knfsd

[trondmy]

From: Trond Myklebust <trond.myklebust@hammerspace.com>

The following patchset is mainly for improving support for re-exporting
NFSv4 as NFSv3. However it also includes one generic bugfix for NFSv3 to
allow zero length writes. It also improves the writeback performance by
replacing the rwsem with a lock-free errseq_t-based method.

Jeff Layton (3):
  nfsd: Add errno mapping for EREMOTEIO
  nfsd: Retry once in nfsd_open on an -EOPENSTALE return
  nfsd: allow lockd to be forcibly disabled

Peng Tao (1):
  nfsd: map EBADF

Trond Myklebust (5):
  nfs: Add export support for weak cache consistency attributes
  nfsd: NFSv3 should allow zero length writes
  nfsd: Add a tracepoint for errors in nfsd4_clone_file_range()
  nfsd: Replace use of rwsem with errseq_t
  nfsd: Ignore rpcbind errors on nfsd startup

 fs/nfs/export.c                | 24 ++++++++++
 fs/nfsd/filecache.c            |  1 -
 fs/nfsd/filecache.h            |  1 -
 fs/nfsd/nfs3xdr.c              | 83 ++++++++++++++++++++++-----------
 fs/nfsd/nfs4proc.c             | 18 +++----
 fs/nfsd/nfs4xdr.c              |  6 +--
 fs/nfsd/nfsctl.c               |  7 ++-
 fs/nfsd/nfsd.h                 |  1 +
 fs/nfsd/nfsproc.c              |  3 ++
 fs/nfsd/nfssvc.c               | 29 +++++++++++-
 fs/nfsd/trace.h                | 50 ++++++++++++++++++++
 fs/nfsd/vfs.c                  | 85 +++++++++++++++++++++++-----------
 fs/nfsd/vfs.h                  |  8 ++--
 include/linux/exportfs.h       |  3 ++
 include/linux/sunrpc/svcsock.h |  5 +-
 net/sunrpc/svc.c               |  2 +-
 net/sunrpc/svcsock.c           | 14 +++---
 17 files changed, 257 insertions(+), 83 deletions(-)

-- 
2.33.1

[PATCH 1/9] nfsd: map EBADF

[trondmy]

From: Peng Tao <tao.peng@primarydata.com>

Now that we have open file cache, it is possible that another client
deletes the file and DP will not know about it. Then IO to MDS would
fail with BADSTATEID and knfsd would start state recovery, which
should fail as well and then nfs read/write will fail with EBADF.
And it triggers a WARN() in nfserrno().

-----------[ cut here ]------------
WARNING: CPU: 0 PID: 13529 at fs/nfsd/nfsproc.c:758 nfserrno+0x58/0x70 [nfsd]()
nfsd: non-standard errno: -9
modules linked in: nfsv3 nfs_layout_flexfiles rpcsec_gss_krb5 nfsv4 dns_resolver nfs fscache ip6t_rpfilter ip6t_REJECT nf_reject_ipv6 xt_connt
pata_acpi floppy
CPU: 0 PID: 13529 Comm: nfsd Tainted: G        W       4.1.5-00307-g6e6579b #7
Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 09/30/2014
 0000000000000000 00000000464e6c9c ffff88079085fba8 ffffffff81789936
 0000000000000000 ffff88079085fc00 ffff88079085fbe8 ffffffff810a08ea
 ffff88079085fbe8 ffff88080f45c900 ffff88080f627d50 ffff880790c46a48
 all Trace:
 [<ffffffff81789936>] dump_stack+0x45/0x57
 [<ffffffff810a08ea>] warn_slowpath_common+0x8a/0xc0
 [<ffffffff810a0975>] warn_slowpath_fmt+0x55/0x70
 [<ffffffff81252908>] ? splice_direct_to_actor+0x148/0x230
 [<ffffffffa02fb8c0>] ? fsid_source+0x60/0x60 [nfsd]
 [<ffffffffa02f9918>] nfserrno+0x58/0x70 [nfsd]
 [<ffffffffa02fba57>] nfsd_finish_read+0x97/0xb0 [nfsd]
 [<ffffffffa02fc7a6>] nfsd_splice_read+0x76/0xa0 [nfsd]
 [<ffffffffa02fcca1>] nfsd_read+0xc1/0xd0 [nfsd]
 [<ffffffffa0233af2>] ? svc_tcp_adjust_wspace+0x12/0x30 [sunrpc]
 [<ffffffffa03073da>] nfsd3_proc_read+0xba/0x150 [nfsd]
 [<ffffffffa02f7a03>] nfsd_dispatch+0xc3/0x210 [nfsd]
 [<ffffffffa0233af2>] ? svc_tcp_adjust_wspace+0x12/0x30 [sunrpc]
 [<ffffffffa0232913>] svc_process_common+0x453/0x6f0 [sunrpc]
 [<ffffffffa0232cc3>] svc_process+0x113/0x1b0 [sunrpc]
 [<ffffffffa02f740f>] nfsd+0xff/0x170 [nfsd]
 [<ffffffffa02f7310>] ? nfsd_destroy+0x80/0x80 [nfsd]
 [<ffffffff810bf3a8>] kthread+0xd8/0xf0
 [<ffffffff810bf2d0>] ? kthread_create_on_node+0x1b0/0x1b0
 [<ffffffff817912a2>] ret_from_fork+0x42/0x70
 [<ffffffff810bf2d0>] ? kthread_create_on_node+0x1b0/0x1b0

Signed-off-by: Peng Tao <tao.peng@primarydata.com>
Signed-off-by: Lance Shelton <lance.shelton@hammerspace.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
---
 fs/nfsd/nfsproc.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/fs/nfsd/nfsproc.c b/fs/nfsd/nfsproc.c
index 90fcd6178823..c2afe1afb6d7 100644
--- a/fs/nfsd/nfsproc.c
+++ b/fs/nfsd/nfsproc.c
@@ -851,6 +851,7 @@ nfserrno (int errno)
 		{ nfserr_io, -EIO },
 		{ nfserr_nxio, -ENXIO },
 		{ nfserr_fbig, -E2BIG },
+		{ nfserr_stale, -EBADF },
 		{ nfserr_acces, -EACCES },
 		{ nfserr_exist, -EEXIST },
 		{ nfserr_xdev, -EXDEV },
-- 
2.33.1

[PATCH 2/9] nfsd: Add errno mapping for EREMOTEIO

[trondmy]

From: Jeff Layton <jeff.layton@primarydata.com>

The NFS client can occasionally return EREMOTEIO when signalling issues
with the server.  ...map to NFSERR_IO.

Signed-off-by: Jeff Layton <jeff.layton@primarydata.com>
Signed-off-by: Lance Shelton <lance.shelton@hammerspace.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
---
 fs/nfsd/nfsproc.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/fs/nfsd/nfsproc.c b/fs/nfsd/nfsproc.c
index c2afe1afb6d7..1ebf02123368 100644
--- a/fs/nfsd/nfsproc.c
+++ b/fs/nfsd/nfsproc.c
@@ -880,6 +880,7 @@ nfserrno (int errno)
 		{ nfserr_toosmall, -ETOOSMALL },
 		{ nfserr_serverfault, -ESERVERFAULT },
 		{ nfserr_serverfault, -ENFILE },
+		{ nfserr_io, -EREMOTEIO },
 		{ nfserr_io, -EUCLEAN },
 		{ nfserr_perm, -ENOKEY },
 		{ nfserr_no_grace, -ENOGRACE},
-- 
2.33.1

[PATCH 3/9] nfsd: Retry once in nfsd_open on an -EOPENSTALE return

[trondmy]

From: Jeff Layton <jeff.layton@primarydata.com>

If we get back -EOPENSTALE from an NFSv4 open, then we either got some
unhandled error or the inode we got back was not the same as the one
associated with the dentry.

We really have no recourse in that situation other than to retry the
open, and if it fails to just return nfserr_stale back to the client.

Signed-off-by: Jeff Layton <jeff.layton@primarydata.com>
Signed-off-by: Lance Shelton <lance.shelton@hammerspace.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
---
 fs/nfsd/nfsproc.c |  1 +
 fs/nfsd/vfs.c     | 10 +++++++++-
 2 files changed, 10 insertions(+), 1 deletion(-)

diff --git a/fs/nfsd/nfsproc.c b/fs/nfsd/nfsproc.c
index 1ebf02123368..458307876ead 100644
--- a/fs/nfsd/nfsproc.c
+++ b/fs/nfsd/nfsproc.c
@@ -881,6 +881,7 @@ nfserrno (int errno)
 		{ nfserr_serverfault, -ESERVERFAULT },
 		{ nfserr_serverfault, -ENFILE },
 		{ nfserr_io, -EREMOTEIO },
+		{ nfserr_stale, -EOPENSTALE },
 		{ nfserr_io, -EUCLEAN },
 		{ nfserr_perm, -ENOKEY },
 		{ nfserr_no_grace, -ENOGRACE},
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 738d564ca4ce..a90cc08211a3 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -779,6 +779,7 @@ nfsd_open(struct svc_rqst *rqstp, struct svc_fh *fhp, umode_t type,
 		int may_flags, struct file **filp)
 {
 	__be32 err;
+	bool retried = false;
 
 	validate_process_creds();
 	/*
@@ -794,9 +795,16 @@ nfsd_open(struct svc_rqst *rqstp, struct svc_fh *fhp, umode_t type,
 	 */
 	if (type == S_IFREG)
 		may_flags |= NFSD_MAY_OWNER_OVERRIDE;
+retry:
 	err = fh_verify(rqstp, fhp, type, may_flags);
-	if (!err)
+	if (!err) {
 		err = __nfsd_open(rqstp, fhp, type, may_flags, filp);
+		if (err == nfserr_stale && !retried) {
+			retried = true;
+			fh_put(fhp);
+			goto retry;
+		}
+	}
 	validate_process_creds();
 	return err;
 }
-- 
2.33.1

[PATCH 4/9] nfs: Add export support for weak cache consistency attributes

[trondmy]

From: Trond Myklebust <trond.myklebust@primarydata.com>

Allow knfsd to request weak cache consistency attributes on files that
have delegations and/or have up to date attribute caches.

Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
Signed-off-by: Lance Shelton <lance.shelton@hammerspace.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
---
 fs/nfs/export.c          | 24 ++++++++++++
 fs/nfsd/nfs3xdr.c        | 83 +++++++++++++++++++++++++++-------------
 fs/nfsd/nfs4xdr.c        |  6 +--
 fs/nfsd/vfs.c            | 14 +++++++
 fs/nfsd/vfs.h            |  5 ++-
 include/linux/exportfs.h |  3 ++
 6 files changed, 103 insertions(+), 32 deletions(-)

diff --git a/fs/nfs/export.c b/fs/nfs/export.c
index 171c424cb6d5..967f8902c49b 100644
--- a/fs/nfs/export.c
+++ b/fs/nfs/export.c
@@ -151,10 +151,34 @@ static u64 nfs_fetch_iversion(struct inode *inode)
 	return inode_peek_iversion_raw(inode);
 }
 
+static int nfs_exp_getattr(struct path *path, struct kstat *stat, bool force)
+{
+	const unsigned long check_valid =
+		NFS_INO_INVALID_CHANGE | NFS_INO_INVALID_ATIME |
+		NFS_INO_INVALID_CTIME | NFS_INO_INVALID_MTIME |
+		NFS_INO_INVALID_SIZE | /* NFS_INO_INVALID_BLOCKS | */
+		NFS_INO_INVALID_OTHER | NFS_INO_INVALID_NLINK;
+	struct inode *inode = d_inode(path->dentry);
+	int flags = force ? AT_STATX_SYNC_AS_STAT : AT_STATX_DONT_SYNC;
+	int ret, ret2 = 0;
+
+	if (!force && nfs_check_cache_invalid(inode, check_valid))
+		ret2 = -EAGAIN;
+	ret = vfs_getattr(path, stat, STATX_BASIC_STATS & ~STATX_BLOCKS, flags);
+	if (ret < 0)
+		return ret;
+	stat->blocks = nfs_calc_block_size(stat->size);
+	if (S_ISDIR(inode->i_mode))
+		stat->blksize = NFS_SERVER(inode)->dtsize;
+	stat->result_mask |= STATX_BLOCKS;
+	return ret2;
+}
+
 const struct export_operations nfs_export_ops = {
 	.encode_fh = nfs_encode_fh,
 	.fh_to_dentry = nfs_fh_to_dentry,
 	.get_parent = nfs_get_parent,
+	.getattr = nfs_exp_getattr,
 	.fetch_iversion = nfs_fetch_iversion,
 	.flags = EXPORT_OP_NOWCC|EXPORT_OP_NOSUBTREECHK|
 		EXPORT_OP_CLOSE_BEFORE_UNLINK|EXPORT_OP_REMOTE_FS|
diff --git a/fs/nfsd/nfs3xdr.c b/fs/nfsd/nfs3xdr.c
index 0a5ebc52e6a9..81b6cf228517 100644
--- a/fs/nfsd/nfs3xdr.c
+++ b/fs/nfsd/nfs3xdr.c
@@ -415,21 +415,14 @@ svcxdr_encode_pre_op_attr(struct xdr_stream *xdr, const struct svc_fh *fhp)
 	return svcxdr_encode_wcc_attr(xdr, fhp);
 }
 
-/**
- * svcxdr_encode_post_op_attr - Encode NFSv3 post-op attributes
- * @rqstp: Context of a completed RPC transaction
- * @xdr: XDR stream
- * @fhp: File handle to encode
- *
- * Return values:
- *   %false: Send buffer space was exhausted
- *   %true: Success
- */
-bool
-svcxdr_encode_post_op_attr(struct svc_rqst *rqstp, struct xdr_stream *xdr,
-			   const struct svc_fh *fhp)
+static bool
+__svcxdr_encode_post_op_attr(struct svc_rqst *rqstp, struct xdr_stream *xdr,
+			     const struct svc_fh *fhp, bool force)
 {
 	struct dentry *dentry = fhp->fh_dentry;
+	struct path path = {
+		.dentry = dentry,
+	};
 	struct kstat stat;
 
 	/*
@@ -437,9 +430,10 @@ svcxdr_encode_post_op_attr(struct svc_rqst *rqstp, struct xdr_stream *xdr,
 	 * stale file handle. In this case, no attributes are
 	 * returned.
 	 */
-	if (fhp->fh_no_wcc || !dentry || !d_really_is_positive(dentry))
+	if (!dentry || !d_really_is_positive(dentry))
 		goto no_post_op_attrs;
-	if (fh_getattr(fhp, &stat) != nfs_ok)
+	path.mnt = fhp->fh_export->ex_path.mnt;
+	if (nfsd_getattr(&path, &stat, force) != nfs_ok)
 		goto no_post_op_attrs;
 
 	if (xdr_stream_encode_item_present(xdr) < 0)
@@ -454,6 +448,31 @@ svcxdr_encode_post_op_attr(struct svc_rqst *rqstp, struct xdr_stream *xdr,
 	return xdr_stream_encode_item_absent(xdr) > 0;
 }
 
+/**
+ * svcxdr_encode_post_op_attr - Encode NFSv3 post-op attributes
+ * @rqstp: Context of a completed RPC transaction
+ * @xdr: XDR stream
+ * @fhp: File handle to encode
+ *
+ * Return values:
+ *   %false: Send buffer space was exhausted
+ *   %true: Success
+ */
+bool
+svcxdr_encode_post_op_attr(struct svc_rqst *rqstp, struct xdr_stream *xdr,
+			   const struct svc_fh *fhp)
+{
+	return __svcxdr_encode_post_op_attr(rqstp, xdr, fhp, true);
+}
+
+static bool
+svcxdr_encode_post_op_attr_opportunistic(struct svc_rqst *rqstp,
+					 struct xdr_stream *xdr,
+					 const struct svc_fh *fhp)
+{
+	return __svcxdr_encode_post_op_attr(rqstp, xdr, fhp, !fhp->fh_no_wcc);
+}
+
 /*
  * Encode weak cache consistency data
  */
@@ -481,7 +500,7 @@ svcxdr_encode_wcc_data(struct svc_rqst *rqstp, struct xdr_stream *xdr,
 neither:
 	if (xdr_stream_encode_item_absent(xdr) < 0)
 		return false;
-	if (!svcxdr_encode_post_op_attr(rqstp, xdr, fhp))
+	if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr, fhp))
 		return false;
 
 	return true;
@@ -879,11 +898,13 @@ int nfs3svc_encode_lookupres(struct svc_rqst *rqstp, __be32 *p)
 			return 0;
 		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->fh))
 			return 0;
-		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->dirfh))
+		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
+							      &resp->dirfh))
 			return 0;
 		break;
 	default:
-		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->dirfh))
+		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
+							      &resp->dirfh))
 			return 0;
 	}
 
@@ -901,13 +922,15 @@ nfs3svc_encode_accessres(struct svc_rqst *rqstp, __be32 *p)
 		return 0;
 	switch (resp->status) {
 	case nfs_ok:
-		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->fh))
+		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
+							      &resp->fh))
 			return 0;
 		if (xdr_stream_encode_u32(xdr, resp->access) < 0)
 			return 0;
 		break;
 	default:
-		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->fh))
+		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
+							      &resp->fh))
 			return 0;
 	}
 
@@ -926,7 +949,8 @@ nfs3svc_encode_readlinkres(struct svc_rqst *rqstp, __be32 *p)
 		return 0;
 	switch (resp->status) {
 	case nfs_ok:
-		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->fh))
+		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
+							      &resp->fh))
 			return 0;
 		if (xdr_stream_encode_u32(xdr, resp->len) < 0)
 			return 0;
@@ -935,7 +959,8 @@ nfs3svc_encode_readlinkres(struct svc_rqst *rqstp, __be32 *p)
 			return 0;
 		break;
 	default:
-		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->fh))
+		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
+							      &resp->fh))
 			return 0;
 	}
 
@@ -954,7 +979,8 @@ nfs3svc_encode_readres(struct svc_rqst *rqstp, __be32 *p)
 		return 0;
 	switch (resp->status) {
 	case nfs_ok:
-		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->fh))
+		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
+							      &resp->fh))
 			return 0;
 		if (xdr_stream_encode_u32(xdr, resp->count) < 0)
 			return 0;
@@ -968,7 +994,8 @@ nfs3svc_encode_readres(struct svc_rqst *rqstp, __be32 *p)
 			return 0;
 		break;
 	default:
-		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->fh))
+		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
+							      &resp->fh))
 			return 0;
 	}
 
@@ -1065,7 +1092,8 @@ nfs3svc_encode_readdirres(struct svc_rqst *rqstp, __be32 *p)
 		return 0;
 	switch (resp->status) {
 	case nfs_ok:
-		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->fh))
+		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
+							      &resp->fh))
 			return 0;
 		if (!svcxdr_encode_cookieverf3(xdr, resp->verf))
 			return 0;
@@ -1077,7 +1105,8 @@ nfs3svc_encode_readdirres(struct svc_rqst *rqstp, __be32 *p)
 			return 0;
 		break;
 	default:
-		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->fh))
+		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
+							      &resp->fh))
 			return 0;
 	}
 
@@ -1221,7 +1250,7 @@ svcxdr_encode_entry3_plus(struct nfsd3_readdirres *resp, const char *name,
 	if (compose_entry_fh(resp, fhp, name, namlen, ino) != nfs_ok)
 		goto out_noattrs;
 
-	if (!svcxdr_encode_post_op_attr(resp->rqstp, xdr, fhp))
+	if (!svcxdr_encode_post_op_attr_opportunistic(resp->rqstp, xdr, fhp))
 		goto out;
 	if (!svcxdr_encode_post_op_fh3(xdr, fhp))
 		goto out;
diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c
index 7abeccb975b2..e14af0383b70 100644
--- a/fs/nfsd/nfs4xdr.c
+++ b/fs/nfsd/nfs4xdr.c
@@ -2865,9 +2865,9 @@ nfsd4_encode_fattr(struct xdr_stream *xdr, struct svc_fh *fhp,
 			goto out;
 	}
 
-	err = vfs_getattr(&path, &stat, STATX_BASIC_STATS, AT_STATX_SYNC_AS_STAT);
-	if (err)
-		goto out_nfserr;
+	status = nfsd_getattr(&path, &stat, true);
+	if (status)
+		goto out;
 	if ((bmval0 & (FATTR4_WORD0_FILES_AVAIL | FATTR4_WORD0_FILES_FREE |
 			FATTR4_WORD0_FILES_TOTAL | FATTR4_WORD0_MAXNAME)) ||
 	    (bmval1 & (FATTR4_WORD1_SPACE_AVAIL | FATTR4_WORD1_SPACE_FREE |
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index a90cc08211a3..4d57befdac6e 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -2412,3 +2412,17 @@ nfsd_permission(struct svc_rqst *rqstp, struct svc_export *exp,
 
 	return err? nfserrno(err) : 0;
 }
+
+
+__be32
+nfsd_getattr(struct path *p, struct kstat *stat, bool force)
+{
+	const struct export_operations *ops = p->dentry->d_sb->s_export_op;
+	int err;
+
+	if (ops->getattr)
+		err = ops->getattr(p, stat, force);
+	else
+		err = vfs_getattr(p, stat, STATX_BASIC_STATS, AT_STATX_SYNC_AS_STAT);
+	return err ? nfserrno(err) : 0;
+}
diff --git a/fs/nfsd/vfs.h b/fs/nfsd/vfs.h
index b21b76e6b9a8..6edae1b9a96e 100644
--- a/fs/nfsd/vfs.h
+++ b/fs/nfsd/vfs.h
@@ -132,6 +132,8 @@ __be32		nfsd_statfs(struct svc_rqst *, struct svc_fh *,
 __be32		nfsd_permission(struct svc_rqst *, struct svc_export *,
 				struct dentry *, int);
 
+__be32		nfsd_getattr(struct path *p, struct kstat *, bool);
+
 static inline int fh_want_write(struct svc_fh *fh)
 {
 	int ret;
@@ -156,8 +158,7 @@ static inline __be32 fh_getattr(const struct svc_fh *fh, struct kstat *stat)
 {
 	struct path p = {.mnt = fh->fh_export->ex_path.mnt,
 			 .dentry = fh->fh_dentry};
-	return nfserrno(vfs_getattr(&p, stat, STATX_BASIC_STATS,
-				    AT_STATX_SYNC_AS_STAT));
+	return nfsd_getattr(&p, stat, true);
 }
 
 static inline int nfsd_create_is_exclusive(int createmode)
diff --git a/include/linux/exportfs.h b/include/linux/exportfs.h
index 3260fe714846..58f36022787e 100644
--- a/include/linux/exportfs.h
+++ b/include/linux/exportfs.h
@@ -10,6 +10,8 @@ struct inode;
 struct iomap;
 struct super_block;
 struct vfsmount;
+struct path;
+struct kstat;
 
 /* limit the handle size to NFSv4 handle size now */
 #define MAX_HANDLE_SZ 128
@@ -224,6 +226,7 @@ struct export_operations {
 #define EXPORT_OP_SYNC_LOCKS		(0x20) /* Filesystem can't do
 						  asychronous blocking locks */
 	unsigned long	flags;
+	int (*getattr)(struct path *, struct kstat *, bool);
 };
 
 extern int exportfs_encode_inode_fh(struct inode *inode, struct fid *fid,
-- 
2.33.1

[PATCH 5/9] nfsd: NFSv3 should allow zero length writes

[trondmy]

From: Trond Myklebust <trond.myklebust@hammerspace.com>

According to RFC1813: "If count is 0, the WRITE will succeed
and return a count of 0, barring errors due to permissions checking."

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
---
 fs/nfsd/vfs.c    | 3 +++
 net/sunrpc/svc.c | 2 +-
 2 files changed, 4 insertions(+), 1 deletion(-)

diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 4d57befdac6e..38fdfcbb079e 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -969,6 +969,9 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct nfsd_file *nf,
 
 	trace_nfsd_write_opened(rqstp, fhp, offset, *cnt);
 
+	if (!*cnt)
+		return nfs_ok;
+
 	if (sb->s_export_op)
 		exp_op_flags = sb->s_export_op->flags;
 
diff --git a/net/sunrpc/svc.c b/net/sunrpc/svc.c
index a3bbe5ce4570..d1ccf37a4588 100644
--- a/net/sunrpc/svc.c
+++ b/net/sunrpc/svc.c
@@ -1692,7 +1692,7 @@ unsigned int svc_fill_write_vector(struct svc_rqst *rqstp, struct page **pages,
 	 * entirely in rq_arg.pages. In this case, @first is empty.
 	 */
 	i = 0;
-	if (first->iov_len) {
+	if (first->iov_len || !total) {
 		vec[i].iov_base = first->iov_base;
 		vec[i].iov_len = min_t(size_t, total, first->iov_len);
 		total -= vec[i].iov_len;
-- 
2.33.1

[PATCH 6/9] nfsd: Add a tracepoint for errors in nfsd4_clone_file_range()

[trondmy]

From: Trond Myklebust <trond.myklebust@hammerspace.com>

Since a clone error commit can cause the boot verifier to change,
we should trace those errors.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
---
 fs/nfsd/nfs4proc.c |  2 +-
 fs/nfsd/trace.h    | 50 ++++++++++++++++++++++++++++++++++++++++++++++
 fs/nfsd/vfs.c      | 18 +++++++++++++++--
 fs/nfsd/vfs.h      |  3 ++-
 4 files changed, 69 insertions(+), 4 deletions(-)

diff --git a/fs/nfsd/nfs4proc.c b/fs/nfsd/nfs4proc.c
index 486c5dba4b65..53c2a8f0d627 100644
--- a/fs/nfsd/nfs4proc.c
+++ b/fs/nfsd/nfs4proc.c
@@ -1102,7 +1102,7 @@ nfsd4_clone(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	if (status)
 		goto out;
 
-	status = nfsd4_clone_file_range(src, clone->cl_src_pos,
+	status = nfsd4_clone_file_range(rqstp, src, clone->cl_src_pos,
 			dst, clone->cl_dst_pos, clone->cl_count,
 			EX_ISSYNC(cstate->current_fh.fh_export));
 
diff --git a/fs/nfsd/trace.h b/fs/nfsd/trace.h
index f1e0d3c51bc2..001444d9829d 100644
--- a/fs/nfsd/trace.h
+++ b/fs/nfsd/trace.h
@@ -413,6 +413,56 @@ TRACE_EVENT(nfsd_dirent,
 	)
 )
 
+DECLARE_EVENT_CLASS(nfsd_copy_err_class,
+	TP_PROTO(struct svc_rqst *rqstp,
+		 struct svc_fh	*src_fhp,
+		 loff_t		src_offset,
+		 struct svc_fh	*dst_fhp,
+		 loff_t		dst_offset,
+		 u64		count,
+		 int		status),
+	TP_ARGS(rqstp, src_fhp, src_offset, dst_fhp, dst_offset, count, status),
+	TP_STRUCT__entry(
+		__field(u32, xid)
+		__field(u32, src_fh_hash)
+		__field(loff_t, src_offset)
+		__field(u32, dst_fh_hash)
+		__field(loff_t, dst_offset)
+		__field(u64, count)
+		__field(int, status)
+	),
+	TP_fast_assign(
+		__entry->xid = be32_to_cpu(rqstp->rq_xid);
+		__entry->src_fh_hash = knfsd_fh_hash(&src_fhp->fh_handle);
+		__entry->src_offset = src_offset;
+		__entry->dst_fh_hash = knfsd_fh_hash(&dst_fhp->fh_handle);
+		__entry->dst_offset = dst_offset;
+		__entry->count = count;
+		__entry->status = status;
+	),
+	TP_printk("xid=0x%08x src_fh_hash=0x%08x src_offset=%lld "
+			"dst_fh_hash=0x%08x dst_offset=%lld "
+			"count=%llu status=%d",
+		  __entry->xid, __entry->src_fh_hash, __entry->src_offset,
+		  __entry->dst_fh_hash, __entry->dst_offset,
+		  (unsigned long long)__entry->count,
+		  __entry->status)
+)
+
+#define DEFINE_NFSD_COPY_ERR_EVENT(name)		\
+DEFINE_EVENT(nfsd_copy_err_class, nfsd_##name,		\
+	TP_PROTO(struct svc_rqst	*rqstp,		\
+		 struct svc_fh		*src_fhp,	\
+		 loff_t			src_offset,	\
+		 struct svc_fh		*dst_fhp,	\
+		 loff_t			dst_offset,	\
+		 u64			count,		\
+		 int			status),	\
+	TP_ARGS(rqstp, src_fhp, src_offset, dst_fhp, dst_offset, \
+		count, status))
+
+DEFINE_NFSD_COPY_ERR_EVENT(clone_file_range_err);
+
 #include "state.h"
 #include "filecache.h"
 #include "vfs.h"
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 38fdfcbb079e..e761b2eff415 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -40,6 +40,7 @@
 #include "../internal.h"
 #include "acl.h"
 #include "idmap.h"
+#include "xdr4.h"
 #endif /* CONFIG_NFSD_V4 */
 
 #include "nfsd.h"
@@ -516,8 +517,15 @@ __be32 nfsd4_set_nfs4_label(struct svc_rqst *rqstp, struct svc_fh *fhp,
 }
 #endif
 
-__be32 nfsd4_clone_file_range(struct nfsd_file *nf_src, u64 src_pos,
-		struct nfsd_file *nf_dst, u64 dst_pos, u64 count, bool sync)
+static struct nfsd4_compound_state *nfsd4_get_cstate(struct svc_rqst *rqstp)
+{
+	return &((struct nfsd4_compoundres *)rqstp->rq_resp)->cstate;
+}
+
+__be32 nfsd4_clone_file_range(struct svc_rqst *rqstp,
+		struct nfsd_file *nf_src, u64 src_pos,
+		struct nfsd_file *nf_dst, u64 dst_pos,
+		u64 count, bool sync)
 {
 	struct file *src = nf_src->nf_file;
 	struct file *dst = nf_dst->nf_file;
@@ -541,6 +549,12 @@ __be32 nfsd4_clone_file_range(struct nfsd_file *nf_src, u64 src_pos,
 		if (!status)
 			status = commit_inode_metadata(file_inode(src));
 		if (status < 0) {
+			trace_nfsd_clone_file_range_err(rqstp,
+					&nfsd4_get_cstate(rqstp)->save_fh,
+					src_pos,
+					&nfsd4_get_cstate(rqstp)->current_fh,
+					dst_pos,
+					count, status);
 			nfsd_reset_boot_verifier(net_generic(nf_dst->nf_net,
 						 nfsd_net_id));
 			ret = nfserrno(status);
diff --git a/fs/nfsd/vfs.h b/fs/nfsd/vfs.h
index 6edae1b9a96e..3dba6397d452 100644
--- a/fs/nfsd/vfs.h
+++ b/fs/nfsd/vfs.h
@@ -57,7 +57,8 @@ __be32          nfsd4_set_nfs4_label(struct svc_rqst *, struct svc_fh *,
 		    struct xdr_netobj *);
 __be32		nfsd4_vfs_fallocate(struct svc_rqst *, struct svc_fh *,
 				    struct file *, loff_t, loff_t, int);
-__be32		nfsd4_clone_file_range(struct nfsd_file *nf_src, u64 src_pos,
+__be32		nfsd4_clone_file_range(struct svc_rqst *,
+				       struct nfsd_file *nf_src, u64 src_pos,
 				       struct nfsd_file *nf_dst, u64 dst_pos,
 				       u64 count, bool sync);
 #endif /* CONFIG_NFSD_V4 */
-- 
2.33.1

[PATCH 7/9] nfsd: Replace use of rwsem with errseq_t

[trondmy]

From: Trond Myklebust <trond.myklebust@hammerspace.com>

The nfsd_file nf_rwsem is currently being used to separate file write
and commit instances to ensure that we catch errors and apply them to
the correct write/commit.
We can improve scalability at the expense of a little accuracy (some
extra false positives) by replacing the nf_rwsem with more careful
use of the errseq_t mechanism to track errors across the different
operations.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
---
 fs/nfsd/filecache.c |  1 -
 fs/nfsd/filecache.h |  1 -
 fs/nfsd/nfs4proc.c  | 16 +++++++++-------
 fs/nfsd/vfs.c       | 40 +++++++++++++++-------------------------
 4 files changed, 24 insertions(+), 34 deletions(-)

diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c
index 7629248fdd53..1e894ddc7ac2 100644
--- a/fs/nfsd/filecache.c
+++ b/fs/nfsd/filecache.c
@@ -194,7 +194,6 @@ nfsd_file_alloc(struct inode *inode, unsigned int may, unsigned int hashval,
 				__set_bit(NFSD_FILE_BREAK_READ, &nf->nf_flags);
 		}
 		nf->nf_mark = NULL;
-		init_rwsem(&nf->nf_rwsem);
 		trace_nfsd_file_alloc(nf);
 	}
 	return nf;
diff --git a/fs/nfsd/filecache.h b/fs/nfsd/filecache.h
index 7872df5a0fe3..435ceab27897 100644
--- a/fs/nfsd/filecache.h
+++ b/fs/nfsd/filecache.h
@@ -46,7 +46,6 @@ struct nfsd_file {
 	refcount_t		nf_ref;
 	unsigned char		nf_may;
 	struct nfsd_file_mark	*nf_mark;
-	struct rw_semaphore	nf_rwsem;
 };
 
 int nfsd_file_cache_init(void);
diff --git a/fs/nfsd/nfs4proc.c b/fs/nfsd/nfs4proc.c
index 53c2a8f0d627..9ea2d2e554ce 100644
--- a/fs/nfsd/nfs4proc.c
+++ b/fs/nfsd/nfs4proc.c
@@ -1511,6 +1511,9 @@ static void nfsd4_init_copy_res(struct nfsd4_copy *copy, bool sync)
 
 static ssize_t _nfsd_copy_file_range(struct nfsd4_copy *copy)
 {
+	struct file *dst = copy->nf_dst->nf_file;
+	struct file *src = copy->nf_src->nf_file;
+	errseq_t since;
 	ssize_t bytes_copied = 0;
 	u64 bytes_total = copy->cp_count;
 	u64 src_pos = copy->cp_src_pos;
@@ -1523,9 +1526,8 @@ static ssize_t _nfsd_copy_file_range(struct nfsd4_copy *copy)
 	do {
 		if (kthread_should_stop())
 			break;
-		bytes_copied = nfsd_copy_file_range(copy->nf_src->nf_file,
-				src_pos, copy->nf_dst->nf_file, dst_pos,
-				bytes_total);
+		bytes_copied = nfsd_copy_file_range(src, src_pos, dst, dst_pos,
+						    bytes_total);
 		if (bytes_copied <= 0)
 			break;
 		bytes_total -= bytes_copied;
@@ -1535,11 +1537,11 @@ static ssize_t _nfsd_copy_file_range(struct nfsd4_copy *copy)
 	} while (bytes_total > 0 && !copy->cp_synchronous);
 	/* for a non-zero asynchronous copy do a commit of data */
 	if (!copy->cp_synchronous && copy->cp_res.wr_bytes_written > 0) {
-		down_write(&copy->nf_dst->nf_rwsem);
-		status = vfs_fsync_range(copy->nf_dst->nf_file,
-					 copy->cp_dst_pos,
+		since = READ_ONCE(dst->f_wb_err);
+		status = vfs_fsync_range(dst, copy->cp_dst_pos,
 					 copy->cp_res.wr_bytes_written, 0);
-		up_write(&copy->nf_dst->nf_rwsem);
+		if (!status)
+			status = filemap_check_wb_err(dst->f_mapping, since);
 		if (!status)
 			copy->committed = true;
 	}
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index e761b2eff415..19215b9f768e 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -529,10 +529,11 @@ __be32 nfsd4_clone_file_range(struct svc_rqst *rqstp,
 {
 	struct file *src = nf_src->nf_file;
 	struct file *dst = nf_dst->nf_file;
+	errseq_t since;
 	loff_t cloned;
 	__be32 ret = 0;
 
-	down_write(&nf_dst->nf_rwsem);
+	since = READ_ONCE(dst->f_wb_err);
 	cloned = vfs_clone_file_range(src, src_pos, dst, dst_pos, count, 0);
 	if (cloned < 0) {
 		ret = nfserrno(cloned);
@@ -546,6 +547,8 @@ __be32 nfsd4_clone_file_range(struct svc_rqst *rqstp,
 		loff_t dst_end = count ? dst_pos + count - 1 : LLONG_MAX;
 		int status = vfs_fsync_range(dst, dst_pos, dst_end, 0);
 
+		if (!status)
+			status = filemap_check_wb_err(dst->f_mapping, since);
 		if (!status)
 			status = commit_inode_metadata(file_inode(src));
 		if (status < 0) {
@@ -561,7 +564,6 @@ __be32 nfsd4_clone_file_range(struct svc_rqst *rqstp,
 		}
 	}
 out_err:
-	up_write(&nf_dst->nf_rwsem);
 	return ret;
 }
 
@@ -972,6 +974,7 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct nfsd_file *nf,
 	struct super_block	*sb = file_inode(file)->i_sb;
 	struct svc_export	*exp;
 	struct iov_iter		iter;
+	errseq_t		since;
 	__be32			nfserr;
 	int			host_err;
 	int			use_wgather;
@@ -1012,21 +1015,18 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct nfsd_file *nf,
 		flags |= RWF_SYNC;
 
 	iov_iter_kvec(&iter, WRITE, vec, vlen, *cnt);
+	since = READ_ONCE(file->f_wb_err);
 	if (flags & RWF_SYNC) {
-		down_write(&nf->nf_rwsem);
 		host_err = vfs_iter_write(file, &iter, &pos, flags);
 		if (host_err < 0)
 			nfsd_reset_boot_verifier(net_generic(SVC_NET(rqstp),
 						 nfsd_net_id));
-		up_write(&nf->nf_rwsem);
 	} else {
-		down_read(&nf->nf_rwsem);
 		if (verf)
 			nfsd_copy_boot_verifier(verf,
 					net_generic(SVC_NET(rqstp),
 					nfsd_net_id));
 		host_err = vfs_iter_write(file, &iter, &pos, flags);
-		up_read(&nf->nf_rwsem);
 	}
 	if (host_err < 0) {
 		nfsd_reset_boot_verifier(net_generic(SVC_NET(rqstp),
@@ -1036,6 +1036,9 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct nfsd_file *nf,
 	*cnt = host_err;
 	nfsd_stats_io_write_add(exp, *cnt);
 	fsnotify_modify(file);
+	host_err = filemap_check_wb_err(file->f_mapping, since);
+	if (host_err < 0)
+		goto out_nfserr;
 
 	if (stable && use_wgather) {
 		host_err = wait_for_concurrent_writes(file);
@@ -1116,19 +1119,6 @@ nfsd_write(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t offset,
 }
 
 #ifdef CONFIG_NFSD_V3
-static int
-nfsd_filemap_write_and_wait_range(struct nfsd_file *nf, loff_t offset,
-				  loff_t end)
-{
-	struct address_space *mapping = nf->nf_file->f_mapping;
-	int ret = filemap_fdatawrite_range(mapping, offset, end);
-
-	if (ret)
-		return ret;
-	filemap_fdatawait_range_keep_errors(mapping, offset, end);
-	return 0;
-}
-
 /*
  * Commit all pending writes to stable storage.
  *
@@ -1159,25 +1149,25 @@ nfsd_commit(struct svc_rqst *rqstp, struct svc_fh *fhp,
 	if (err)
 		goto out;
 	if (EX_ISSYNC(fhp->fh_export)) {
-		int err2 = nfsd_filemap_write_and_wait_range(nf, offset, end);
+		errseq_t since = READ_ONCE(nf->nf_file->f_wb_err);
+		int err2;
 
-		down_write(&nf->nf_rwsem);
-		if (!err2)
-			err2 = vfs_fsync_range(nf->nf_file, offset, end, 0);
+		err2 = vfs_fsync_range(nf->nf_file, offset, end, 0);
 		switch (err2) {
 		case 0:
 			nfsd_copy_boot_verifier(verf, net_generic(nf->nf_net,
 						nfsd_net_id));
+			err2 = filemap_check_wb_err(nf->nf_file->f_mapping,
+						    since);
 			break;
 		case -EINVAL:
 			err = nfserr_notsupp;
 			break;
 		default:
-			err = nfserrno(err2);
 			nfsd_reset_boot_verifier(net_generic(nf->nf_net,
 						 nfsd_net_id));
 		}
-		up_write(&nf->nf_rwsem);
+		err = nfserrno(err2);
 	} else
 		nfsd_copy_boot_verifier(verf, net_generic(nf->nf_net,
 					nfsd_net_id));
-- 
2.33.1

[PATCH 8/9] nfsd: allow lockd to be forcibly disabled

[trondmy]

From: Jeff Layton <jeff.layton@primarydata.com>

In some cases, we may want to use a userland NLM server which will
require that we turn off lockd.

Signed-off-by: Jeff Layton <jeff.layton@primarydata.com>
Signed-off-by: Lance Shelton <lance.shelton@hammerspace.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
---
 fs/nfsd/nfssvc.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/fs/nfsd/nfssvc.c b/fs/nfsd/nfssvc.c
index ccb59e91011b..7486a6f5fc21 100644
--- a/fs/nfsd/nfssvc.c
+++ b/fs/nfsd/nfssvc.c
@@ -340,8 +340,19 @@ static void nfsd_shutdown_generic(void)
 	nfsd_file_cache_shutdown();
 }
 
+/*
+ * Allow admin to disable lockd. This would typically be used to allow (e.g.)
+ * a userspace NLM server of some sort to be used.
+ */
+static bool nfsd_disable_lockd = false;
+module_param(nfsd_disable_lockd, bool, 0644);
+MODULE_PARM_DESC(nfsd_disable_lockd, "Allow lockd to be manually disabled.");
+
 static bool nfsd_needs_lockd(struct nfsd_net *nn)
 {
+	if (nfsd_disable_lockd)
+		return false;
+
 	return nfsd_vers(nn, 2, NFSD_TEST) || nfsd_vers(nn, 3, NFSD_TEST);
 }
 
-- 
2.33.1

[PATCH 9/9] nfsd: Ignore rpcbind errors on nfsd startup

[trondmy]

From: Trond Myklebust <trond.myklebust@hammerspace.com>

NFSv4 doesn't need rpcbind, so let's not refuse to start up just because
the rpcbind registration failed.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
---
 fs/nfsd/nfsctl.c               |  7 ++++++-
 fs/nfsd/nfsd.h                 |  1 +
 fs/nfsd/nfssvc.c               | 18 ++++++++++++++++--
 include/linux/sunrpc/svcsock.h |  5 +++--
 net/sunrpc/svcsock.c           | 14 ++++++++------
 5 files changed, 34 insertions(+), 11 deletions(-)

diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c
index c2c3d9077dc5..50f1db5ce59b 100644
--- a/fs/nfsd/nfsctl.c
+++ b/fs/nfsd/nfsctl.c
@@ -727,6 +727,7 @@ static ssize_t __write_ports_addfd(char *buf, struct net *net, const struct cred
 	char *mesg = buf;
 	int fd, err;
 	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
+	int flags = SVC_SOCK_DEFAULTS;
 
 	err = get_int(&mesg, &fd);
 	if (err != 0 || fd < 0)
@@ -741,7 +742,11 @@ static ssize_t __write_ports_addfd(char *buf, struct net *net, const struct cred
 	if (err != 0)
 		return err;
 
-	err = svc_addsock(nn->nfsd_serv, fd, buf, SIMPLE_TRANSACTION_LIMIT, cred);
+	if (!nfsd_rpcbind_error_is_fatal())
+		flags |= SVC_SOCK_RPCBIND_NOERR;
+
+	err = svc_addsock(nn->nfsd_serv, fd, buf, SIMPLE_TRANSACTION_LIMIT,
+			  flags, cred);
 	if (err < 0) {
 		nfsd_destroy(net);
 		return err;
diff --git a/fs/nfsd/nfsd.h b/fs/nfsd/nfsd.h
index 9664303afdaf..ebb87e43b965 100644
--- a/fs/nfsd/nfsd.h
+++ b/fs/nfsd/nfsd.h
@@ -132,6 +132,7 @@ int nfsd_vers(struct nfsd_net *nn, int vers, enum vers_op change);
 int nfsd_minorversion(struct nfsd_net *nn, u32 minorversion, enum vers_op change);
 void nfsd_reset_versions(struct nfsd_net *nn);
 int nfsd_create_serv(struct net *net);
+extern bool nfsd_rpcbind_error_is_fatal(void);
 
 extern int nfsd_max_blksize;
 
diff --git a/fs/nfsd/nfssvc.c b/fs/nfsd/nfssvc.c
index 7486a6f5fc21..9e27e3c842a6 100644
--- a/fs/nfsd/nfssvc.c
+++ b/fs/nfsd/nfssvc.c
@@ -289,17 +289,21 @@ static int nfsd_init_socks(struct net *net, const struct cred *cred)
 {
 	int error;
 	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
+	int flags = SVC_SOCK_DEFAULTS;
 
 	if (!list_empty(&nn->nfsd_serv->sv_permsocks))
 		return 0;
 
+	if (!nfsd_rpcbind_error_is_fatal())
+		flags |= SVC_SOCK_RPCBIND_NOERR;
+
 	error = svc_create_xprt(nn->nfsd_serv, "udp", net, PF_INET, NFS_PORT,
-					SVC_SOCK_DEFAULTS, cred);
+				flags, cred);
 	if (error < 0)
 		return error;
 
 	error = svc_create_xprt(nn->nfsd_serv, "tcp", net, PF_INET, NFS_PORT,
-					SVC_SOCK_DEFAULTS, cred);
+				flags, cred);
 	if (error < 0)
 		return error;
 
@@ -340,6 +344,16 @@ static void nfsd_shutdown_generic(void)
 	nfsd_file_cache_shutdown();
 }
 
+static bool nfsd_rpcbind_error_fatal = false;
+module_param(nfsd_rpcbind_error_fatal, bool, 0644);
+MODULE_PARM_DESC(nfsd_rpcbind_error_fatal,
+		 "rpcbind errors are fatal when starting nfsd.");
+
+bool nfsd_rpcbind_error_is_fatal(void)
+{
+	return nfsd_rpcbind_error_fatal;
+}
+
 /*
  * Allow admin to disable lockd. This would typically be used to allow (e.g.)
  * a userspace NLM server of some sort to be used.
diff --git a/include/linux/sunrpc/svcsock.h b/include/linux/sunrpc/svcsock.h
index bcc555c7ae9c..f34c222cee9d 100644
--- a/include/linux/sunrpc/svcsock.h
+++ b/include/linux/sunrpc/svcsock.h
@@ -61,8 +61,8 @@ void		svc_drop(struct svc_rqst *);
 void		svc_sock_update_bufs(struct svc_serv *serv);
 bool		svc_alien_sock(struct net *net, int fd);
 int		svc_addsock(struct svc_serv *serv, const int fd,
-					char *name_return, const size_t len,
-					const struct cred *cred);
+			    char *name_return, const size_t len, int flags,
+			    const struct cred *cred);
 void		svc_init_xprt_sock(void);
 void		svc_cleanup_xprt_sock(void);
 struct svc_xprt *svc_sock_create(struct svc_serv *serv, int prot);
@@ -74,5 +74,6 @@ void		svc_sock_destroy(struct svc_xprt *);
 #define SVC_SOCK_DEFAULTS	(0U)
 #define SVC_SOCK_ANONYMOUS	(1U << 0)	/* don't register with pmap */
 #define SVC_SOCK_TEMPORARY	(1U << 1)	/* flag socket as temporary */
+#define SVC_SOCK_RPCBIND_NOERR	(1U << 2)	/* Ignore pmap errors */
 
 #endif /* SUNRPC_SVCSOCK_H */
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 478f857cdaed..7f5b12a50bf9 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -1309,14 +1309,15 @@ static struct svc_sock *svc_setup_socket(struct svc_serv *serv,
 	inet = sock->sk;
 
 	/* Register socket with portmapper */
-	if (pmap_register)
+	if (pmap_register) {
 		err = svc_register(serv, sock_net(sock->sk), inet->sk_family,
 				     inet->sk_protocol,
 				     ntohs(inet_sk(inet)->inet_sport));
 
-	if (err < 0) {
-		kfree(svsk);
-		return ERR_PTR(err);
+		if (err < 0 && !(flags & SVC_SOCK_RPCBIND_NOERR)) {
+			kfree(svsk);
+			return ERR_PTR(err);
+		}
 	}
 
 	svsk->sk_sock = sock;
@@ -1364,6 +1365,7 @@ EXPORT_SYMBOL_GPL(svc_alien_sock);
  * @fd: file descriptor of the new listener
  * @name_return: pointer to buffer to fill in with name of listener
  * @len: size of the buffer
+ * @flags: flags argument for svc_setup_socket()
  * @cred: credential
  *
  * Fills in socket name and returns positive length of name if successful.
@@ -1371,7 +1373,7 @@ EXPORT_SYMBOL_GPL(svc_alien_sock);
  * value.
  */
 int svc_addsock(struct svc_serv *serv, const int fd, char *name_return,
-		const size_t len, const struct cred *cred)
+		const size_t len, int flags, const struct cred *cred)
 {
 	int err = 0;
 	struct socket *so = sockfd_lookup(fd, &err);
@@ -1395,7 +1397,7 @@ int svc_addsock(struct svc_serv *serv, const int fd, char *name_return,
 	err = -ENOENT;
 	if (!try_module_get(THIS_MODULE))
 		goto out;
-	svsk = svc_setup_socket(serv, so, SVC_SOCK_DEFAULTS);
+	svsk = svc_setup_socket(serv, so, flags);
 	if (IS_ERR(svsk)) {
 		module_put(THIS_MODULE);
 		err = PTR_ERR(svsk);
-- 
2.33.1

Re: [PATCH 8/9] nfsd: allow lockd to be forcibly disabled

[Chuck Lever III]

> On Dec 17, 2021, at 4:50 PM, trondmy@kernel.org wrote:
> 
> From: Jeff Layton <jeff.layton@primarydata.com>
> 
> In some cases, we may want to use a userland NLM server which will
> require that we turn off lockd.

Does this adminstrative interface need to be container-aware?


> Signed-off-by: Jeff Layton <jeff.layton@primarydata.com>
> Signed-off-by: Lance Shelton <lance.shelton@hammerspace.com>
> Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
> ---
> fs/nfsd/nfssvc.c | 11 +++++++++++
> 1 file changed, 11 insertions(+)
> 
> diff --git a/fs/nfsd/nfssvc.c b/fs/nfsd/nfssvc.c
> index ccb59e91011b..7486a6f5fc21 100644
> --- a/fs/nfsd/nfssvc.c
> +++ b/fs/nfsd/nfssvc.c
> @@ -340,8 +340,19 @@ static void nfsd_shutdown_generic(void)
> 	nfsd_file_cache_shutdown();
> }
> 
> +/*
> + * Allow admin to disable lockd. This would typically be used to allow (e.g.)
> + * a userspace NLM server of some sort to be used.
> + */
> +static bool nfsd_disable_lockd = false;
> +module_param(nfsd_disable_lockd, bool, 0644);
> +MODULE_PARM_DESC(nfsd_disable_lockd, "Allow lockd to be manually disabled.");
> +
> static bool nfsd_needs_lockd(struct nfsd_net *nn)
> {
> +	if (nfsd_disable_lockd)
> +		return false;
> +
> 	return nfsd_vers(nn, 2, NFSD_TEST) || nfsd_vers(nn, 3, NFSD_TEST);
> }
> 
> -- 
> 2.33.1
> 

--
Chuck Lever

Re: [PATCH 8/9] nfsd: allow lockd to be forcibly disabled

[Bruce Fields]

This seems like a really specialized requirement.  Is anyone outside
hammerspace doing this?

--b.

On Fri, Dec 17, 2021 at 5:02 PM <trondmy@kernel.org> wrote:
>
> From: Jeff Layton <jeff.layton@primarydata.com>
>
> In some cases, we may want to use a userland NLM server which will
> require that we turn off lockd.
>
> Signed-off-by: Jeff Layton <jeff.layton@primarydata.com>
> Signed-off-by: Lance Shelton <lance.shelton@hammerspace.com>
> Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
> ---
>  fs/nfsd/nfssvc.c | 11 +++++++++++
>  1 file changed, 11 insertions(+)
>
> diff --git a/fs/nfsd/nfssvc.c b/fs/nfsd/nfssvc.c
> index ccb59e91011b..7486a6f5fc21 100644
> --- a/fs/nfsd/nfssvc.c
> +++ b/fs/nfsd/nfssvc.c
> @@ -340,8 +340,19 @@ static void nfsd_shutdown_generic(void)
>         nfsd_file_cache_shutdown();
>  }
>
> +/*
> + * Allow admin to disable lockd. This would typically be used to allow (e.g.)
> + * a userspace NLM server of some sort to be used.
> + */
> +static bool nfsd_disable_lockd = false;
> +module_param(nfsd_disable_lockd, bool, 0644);
> +MODULE_PARM_DESC(nfsd_disable_lockd, "Allow lockd to be manually disabled.");
> +
>  static bool nfsd_needs_lockd(struct nfsd_net *nn)
>  {
> +       if (nfsd_disable_lockd)
> +               return false;
> +
>         return nfsd_vers(nn, 2, NFSD_TEST) || nfsd_vers(nn, 3, NFSD_TEST);
>  }
>
> --
> 2.33.1
>

Re: [PATCH 5/9] nfsd: NFSv3 should allow zero length writes

[Bruce Fields]

On Fri, Dec 17, 2021 at 5:07 PM <trondmy@kernel.org> wrote:
>
> From: Trond Myklebust <trond.myklebust@hammerspace.com>
>
> According to RFC1813: "If count is 0, the WRITE will succeed
> and return a count of 0, barring errors due to permissions checking."

Yes, I'm surprised we're not already doing this right.

I wonder how far back this bug goes.

The svc.c code is from 8154ef2776aa "NFSD: Clean up legacy NFS WRITE
argument XDR decoders", but the behavior might predate that code.  The
nfsd_vfs_write() logic, I'm not sure I understand.

We have a pynfs test for the v4 case (WRT4), but I guess we must have
nothing testing the v3 case.

--b.

>
> Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
> ---
>  fs/nfsd/vfs.c    | 3 +++
>  net/sunrpc/svc.c | 2 +-
>  2 files changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
> index 4d57befdac6e..38fdfcbb079e 100644
> --- a/fs/nfsd/vfs.c
> +++ b/fs/nfsd/vfs.c
> @@ -969,6 +969,9 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct nfsd_file *nf,
>
>         trace_nfsd_write_opened(rqstp, fhp, offset, *cnt);
>
> +       if (!*cnt)
> +               return nfs_ok;
> +
>         if (sb->s_export_op)
>                 exp_op_flags = sb->s_export_op->flags;
>
> diff --git a/net/sunrpc/svc.c b/net/sunrpc/svc.c
> index a3bbe5ce4570..d1ccf37a4588 100644
> --- a/net/sunrpc/svc.c
> +++ b/net/sunrpc/svc.c
> @@ -1692,7 +1692,7 @@ unsigned int svc_fill_write_vector(struct svc_rqst *rqstp, struct page **pages,
>          * entirely in rq_arg.pages. In this case, @first is empty.
>          */
>         i = 0;
> -       if (first->iov_len) {
> +       if (first->iov_len || !total) {
>                 vec[i].iov_base = first->iov_base;
>                 vec[i].iov_len = min_t(size_t, total, first->iov_len);
>                 total -= vec[i].iov_len;
> --
> 2.33.1
>

Re: [PATCH 5/9] nfsd: NFSv3 should allow zero length writes

[Chuck Lever III]

> On Dec 17, 2021, at 5:23 PM, Bruce Fields <bfields@redhat.com> wrote:
> 
> On Fri, Dec 17, 2021 at 5:07 PM <trondmy@kernel.org> wrote:
>> 
>> From: Trond Myklebust <trond.myklebust@hammerspace.com>
>> 
>> According to RFC1813: "If count is 0, the WRITE will succeed
>> and return a count of 0, barring errors due to permissions checking."
> 
> Yes, I'm surprised we're not already doing this right.
> 
> I wonder how far back this bug goes.
> 
> The svc.c code is from 8154ef2776aa "NFSD: Clean up legacy NFS WRITE
> argument XDR decoders", but the behavior might predate that code.  The
> nfsd_vfs_write() logic, I'm not sure I understand.

The new check in nfsd_vfs_write() might circumvent the VFS
layer's security checks, so I agree, that is a little suspect.


> We have a pynfs test for the v4 case (WRT4), but I guess we must have
> nothing testing the v3 case.

I guess cthon04 doesn't check this case.

But it seems to me WRT4 should already tickle any problems
with nfsd_vfs_write(), shouldn't it?


> --b.
> 
>> 
>> Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
>> ---
>> fs/nfsd/vfs.c    | 3 +++
>> net/sunrpc/svc.c | 2 +-
>> 2 files changed, 4 insertions(+), 1 deletion(-)
>> 
>> diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
>> index 4d57befdac6e..38fdfcbb079e 100644
>> --- a/fs/nfsd/vfs.c
>> +++ b/fs/nfsd/vfs.c
>> @@ -969,6 +969,9 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct nfsd_file *nf,
>> 
>>        trace_nfsd_write_opened(rqstp, fhp, offset, *cnt);
>> 
>> +       if (!*cnt)
>> +               return nfs_ok;
>> +
>>        if (sb->s_export_op)
>>                exp_op_flags = sb->s_export_op->flags;
>> 
>> diff --git a/net/sunrpc/svc.c b/net/sunrpc/svc.c
>> index a3bbe5ce4570..d1ccf37a4588 100644
>> --- a/net/sunrpc/svc.c
>> +++ b/net/sunrpc/svc.c
>> @@ -1692,7 +1692,7 @@ unsigned int svc_fill_write_vector(struct svc_rqst *rqstp, struct page **pages,
>>         * entirely in rq_arg.pages. In this case, @first is empty.
>>         */
>>        i = 0;
>> -       if (first->iov_len) {
>> +       if (first->iov_len || !total) {
>>                vec[i].iov_base = first->iov_base;
>>                vec[i].iov_len = min_t(size_t, total, first->iov_len);
>>                total -= vec[i].iov_len;
>> --
>> 2.33.1
>> 
> 

--
Chuck Lever

Re: [PATCH 5/9] nfsd: NFSv3 should allow zero length writes

[Bruce Fields]

On Sat, Dec 18, 2021 at 1:42 PM Chuck Lever III <chuck.lever@oracle.com> wrote:
> But it seems to me WRT4 should already tickle any problems
> with nfsd_vfs_write(), shouldn't it?

I was just taking Trond's word that this is NFSv3-specific, but it's
not clear to me why from the code, and I know that WRT4 is passing.
Something's weird.  I'm travelling or I'd test it.

--b.

Re: [PATCH 4/9] nfs: Add export support for weak cache consistency attributes

[Chuck Lever III]

> On Dec 17, 2021, at 4:50 PM, trondmy@kernel.org wrote:
> 
> From: Trond Myklebust <trond.myklebust@primarydata.com>
> 
> Allow knfsd to request weak cache consistency attributes on files that
> have delegations and/or have up to date attribute caches.
> 
> Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
> Signed-off-by: Lance Shelton <lance.shelton@hammerspace.com>
> Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

This one does not apply to v5.16-rc. Please rebase this series
and resend.

More below:


> ---
> fs/nfs/export.c          | 24 ++++++++++++
> fs/nfsd/nfs3xdr.c        | 83 +++++++++++++++++++++++++++-------------
> fs/nfsd/nfs4xdr.c        |  6 +--
> fs/nfsd/vfs.c            | 14 +++++++
> fs/nfsd/vfs.h            |  5 ++-
> include/linux/exportfs.h |  3 ++
> 6 files changed, 103 insertions(+), 32 deletions(-)
> 
> diff --git a/fs/nfs/export.c b/fs/nfs/export.c
> index 171c424cb6d5..967f8902c49b 100644
> --- a/fs/nfs/export.c
> +++ b/fs/nfs/export.c
> @@ -151,10 +151,34 @@ static u64 nfs_fetch_iversion(struct inode *inode)
> 	return inode_peek_iversion_raw(inode);
> }
> 
> +static int nfs_exp_getattr(struct path *path, struct kstat *stat, bool force)
> +{
> +	const unsigned long check_valid =
> +		NFS_INO_INVALID_CHANGE | NFS_INO_INVALID_ATIME |
> +		NFS_INO_INVALID_CTIME | NFS_INO_INVALID_MTIME |
> +		NFS_INO_INVALID_SIZE | /* NFS_INO_INVALID_BLOCKS | */
> +		NFS_INO_INVALID_OTHER | NFS_INO_INVALID_NLINK;
> +	struct inode *inode = d_inode(path->dentry);
> +	int flags = force ? AT_STATX_SYNC_AS_STAT : AT_STATX_DONT_SYNC;
> +	int ret, ret2 = 0;
> +
> +	if (!force && nfs_check_cache_invalid(inode, check_valid))
> +		ret2 = -EAGAIN;
> +	ret = vfs_getattr(path, stat, STATX_BASIC_STATS & ~STATX_BLOCKS, flags);
> +	if (ret < 0)
> +		return ret;
> +	stat->blocks = nfs_calc_block_size(stat->size);
> +	if (S_ISDIR(inode->i_mode))
> +		stat->blksize = NFS_SERVER(inode)->dtsize;
> +	stat->result_mask |= STATX_BLOCKS;
> +	return ret2;
> +}
> +
> const struct export_operations nfs_export_ops = {
> 	.encode_fh = nfs_encode_fh,
> 	.fh_to_dentry = nfs_fh_to_dentry,
> 	.get_parent = nfs_get_parent,
> +	.getattr = nfs_exp_getattr,
> 	.fetch_iversion = nfs_fetch_iversion,
> 	.flags = EXPORT_OP_NOWCC|EXPORT_OP_NOSUBTREECHK|
> 		EXPORT_OP_CLOSE_BEFORE_UNLINK|EXPORT_OP_REMOTE_FS|
> diff --git a/fs/nfsd/nfs3xdr.c b/fs/nfsd/nfs3xdr.c
> index 0a5ebc52e6a9..81b6cf228517 100644
> --- a/fs/nfsd/nfs3xdr.c
> +++ b/fs/nfsd/nfs3xdr.c
> @@ -415,21 +415,14 @@ svcxdr_encode_pre_op_attr(struct xdr_stream *xdr, const struct svc_fh *fhp)
> 	return svcxdr_encode_wcc_attr(xdr, fhp);
> }
> 
> -/**
> - * svcxdr_encode_post_op_attr - Encode NFSv3 post-op attributes
> - * @rqstp: Context of a completed RPC transaction
> - * @xdr: XDR stream
> - * @fhp: File handle to encode
> - *
> - * Return values:
> - *   %false: Send buffer space was exhausted
> - *   %true: Success
> - */
> -bool
> -svcxdr_encode_post_op_attr(struct svc_rqst *rqstp, struct xdr_stream *xdr,
> -			   const struct svc_fh *fhp)
> +static bool
> +__svcxdr_encode_post_op_attr(struct svc_rqst *rqstp, struct xdr_stream *xdr,
> +			     const struct svc_fh *fhp, bool force)
> {
> 	struct dentry *dentry = fhp->fh_dentry;
> +	struct path path = {
> +		.dentry = dentry,
> +	};
> 	struct kstat stat;
> 
> 	/*
> @@ -437,9 +430,10 @@ svcxdr_encode_post_op_attr(struct svc_rqst *rqstp, struct xdr_stream *xdr,
> 	 * stale file handle. In this case, no attributes are
> 	 * returned.
> 	 */
> -	if (fhp->fh_no_wcc || !dentry || !d_really_is_positive(dentry))
> +	if (!dentry || !d_really_is_positive(dentry))
> 		goto no_post_op_attrs;
> -	if (fh_getattr(fhp, &stat) != nfs_ok)
> +	path.mnt = fhp->fh_export->ex_path.mnt;
> +	if (nfsd_getattr(&path, &stat, force) != nfs_ok)
> 		goto no_post_op_attrs;
> 
> 	if (xdr_stream_encode_item_present(xdr) < 0)
> @@ -454,6 +448,31 @@ svcxdr_encode_post_op_attr(struct svc_rqst *rqstp, struct xdr_stream *xdr,
> 	return xdr_stream_encode_item_absent(xdr) > 0;
> }
> 
> +/**
> + * svcxdr_encode_post_op_attr - Encode NFSv3 post-op attributes
> + * @rqstp: Context of a completed RPC transaction
> + * @xdr: XDR stream
> + * @fhp: File handle to encode
> + *
> + * Return values:
> + *   %false: Send buffer space was exhausted
> + *   %true: Success
> + */
> +bool
> +svcxdr_encode_post_op_attr(struct svc_rqst *rqstp, struct xdr_stream *xdr,
> +			   const struct svc_fh *fhp)
> +{
> +	return __svcxdr_encode_post_op_attr(rqstp, xdr, fhp, true);
> +}
> +
> +static bool
> +svcxdr_encode_post_op_attr_opportunistic(struct svc_rqst *rqstp,
> +					 struct xdr_stream *xdr,
> +					 const struct svc_fh *fhp)
> +{
> +	return __svcxdr_encode_post_op_attr(rqstp, xdr, fhp, !fhp->fh_no_wcc);
> +}

I don't quite understand the need for splitting encode_post_op_attr.
At the least, more commentary in the patch description would help.
For now, I will study this and get back to you.


> +
> /*
>  * Encode weak cache consistency data
>  */
> @@ -481,7 +500,7 @@ svcxdr_encode_wcc_data(struct svc_rqst *rqstp, struct xdr_stream *xdr,
> neither:
> 	if (xdr_stream_encode_item_absent(xdr) < 0)
> 		return false;
> -	if (!svcxdr_encode_post_op_attr(rqstp, xdr, fhp))
> +	if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr, fhp))
> 		return false;
> 
> 	return true;
> @@ -879,11 +898,13 @@ int nfs3svc_encode_lookupres(struct svc_rqst *rqstp, __be32 *p)
> 			return 0;
> 		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->fh))
> 			return 0;
> -		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->dirfh))
> +		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
> +							      &resp->dirfh))
> 			return 0;
> 		break;
> 	default:
> -		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->dirfh))
> +		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
> +							      &resp->dirfh))
> 			return 0;
> 	}
> 
> @@ -901,13 +922,15 @@ nfs3svc_encode_accessres(struct svc_rqst *rqstp, __be32 *p)
> 		return 0;
> 	switch (resp->status) {
> 	case nfs_ok:
> -		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->fh))
> +		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
> +							      &resp->fh))
> 			return 0;
> 		if (xdr_stream_encode_u32(xdr, resp->access) < 0)
> 			return 0;
> 		break;
> 	default:
> -		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->fh))
> +		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
> +							      &resp->fh))
> 			return 0;
> 	}
> 
> @@ -926,7 +949,8 @@ nfs3svc_encode_readlinkres(struct svc_rqst *rqstp, __be32 *p)
> 		return 0;
> 	switch (resp->status) {
> 	case nfs_ok:
> -		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->fh))
> +		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
> +							      &resp->fh))
> 			return 0;
> 		if (xdr_stream_encode_u32(xdr, resp->len) < 0)
> 			return 0;
> @@ -935,7 +959,8 @@ nfs3svc_encode_readlinkres(struct svc_rqst *rqstp, __be32 *p)
> 			return 0;
> 		break;
> 	default:
> -		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->fh))
> +		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
> +							      &resp->fh))
> 			return 0;
> 	}
> 
> @@ -954,7 +979,8 @@ nfs3svc_encode_readres(struct svc_rqst *rqstp, __be32 *p)
> 		return 0;
> 	switch (resp->status) {
> 	case nfs_ok:
> -		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->fh))
> +		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
> +							      &resp->fh))
> 			return 0;
> 		if (xdr_stream_encode_u32(xdr, resp->count) < 0)
> 			return 0;
> @@ -968,7 +994,8 @@ nfs3svc_encode_readres(struct svc_rqst *rqstp, __be32 *p)
> 			return 0;
> 		break;
> 	default:
> -		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->fh))
> +		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
> +							      &resp->fh))
> 			return 0;
> 	}
> 
> @@ -1065,7 +1092,8 @@ nfs3svc_encode_readdirres(struct svc_rqst *rqstp, __be32 *p)
> 		return 0;
> 	switch (resp->status) {
> 	case nfs_ok:
> -		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->fh))
> +		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
> +							      &resp->fh))
> 			return 0;
> 		if (!svcxdr_encode_cookieverf3(xdr, resp->verf))
> 			return 0;
> @@ -1077,7 +1105,8 @@ nfs3svc_encode_readdirres(struct svc_rqst *rqstp, __be32 *p)
> 			return 0;
> 		break;
> 	default:
> -		if (!svcxdr_encode_post_op_attr(rqstp, xdr, &resp->fh))
> +		if (!svcxdr_encode_post_op_attr_opportunistic(rqstp, xdr,
> +							      &resp->fh))
> 			return 0;
> 	}
> 
> @@ -1221,7 +1250,7 @@ svcxdr_encode_entry3_plus(struct nfsd3_readdirres *resp, const char *name,
> 	if (compose_entry_fh(resp, fhp, name, namlen, ino) != nfs_ok)
> 		goto out_noattrs;
> 
> -	if (!svcxdr_encode_post_op_attr(resp->rqstp, xdr, fhp))
> +	if (!svcxdr_encode_post_op_attr_opportunistic(resp->rqstp, xdr, fhp))
> 		goto out;
> 	if (!svcxdr_encode_post_op_fh3(xdr, fhp))
> 		goto out;
> diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c
> index 7abeccb975b2..e14af0383b70 100644
> --- a/fs/nfsd/nfs4xdr.c
> +++ b/fs/nfsd/nfs4xdr.c
> @@ -2865,9 +2865,9 @@ nfsd4_encode_fattr(struct xdr_stream *xdr, struct svc_fh *fhp,
> 			goto out;
> 	}
> 
> -	err = vfs_getattr(&path, &stat, STATX_BASIC_STATS, AT_STATX_SYNC_AS_STAT);
> -	if (err)
> -		goto out_nfserr;
> +	status = nfsd_getattr(&path, &stat, true);
> +	if (status)
> +		goto out;
> 	if ((bmval0 & (FATTR4_WORD0_FILES_AVAIL | FATTR4_WORD0_FILES_FREE |
> 			FATTR4_WORD0_FILES_TOTAL | FATTR4_WORD0_MAXNAME)) ||
> 	    (bmval1 & (FATTR4_WORD1_SPACE_AVAIL | FATTR4_WORD1_SPACE_FREE |
> diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
> index a90cc08211a3..4d57befdac6e 100644
> --- a/fs/nfsd/vfs.c
> +++ b/fs/nfsd/vfs.c
> @@ -2412,3 +2412,17 @@ nfsd_permission(struct svc_rqst *rqstp, struct svc_export *exp,
> 
> 	return err? nfserrno(err) : 0;
> }
> +
> +
> +__be32
> +nfsd_getattr(struct path *p, struct kstat *stat, bool force)
> +{
> +	const struct export_operations *ops = p->dentry->d_sb->s_export_op;
> +	int err;
> +
> +	if (ops->getattr)
> +		err = ops->getattr(p, stat, force);
> +	else
> +		err = vfs_getattr(p, stat, STATX_BASIC_STATS, AT_STATX_SYNC_AS_STAT);
> +	return err ? nfserrno(err) : 0;
> +}

Please add nfsd_getattr() in a separate patch ("Refactor:").
And please include a kerneldoc comment for it.


> diff --git a/fs/nfsd/vfs.h b/fs/nfsd/vfs.h
> index b21b76e6b9a8..6edae1b9a96e 100644
> --- a/fs/nfsd/vfs.h
> +++ b/fs/nfsd/vfs.h
> @@ -132,6 +132,8 @@ __be32		nfsd_statfs(struct svc_rqst *, struct svc_fh *,
> __be32		nfsd_permission(struct svc_rqst *, struct svc_export *,
> 				struct dentry *, int);
> 
> +__be32		nfsd_getattr(struct path *p, struct kstat *, bool);
> +
> static inline int fh_want_write(struct svc_fh *fh)
> {
> 	int ret;
> @@ -156,8 +158,7 @@ static inline __be32 fh_getattr(const struct svc_fh *fh, struct kstat *stat)
> {
> 	struct path p = {.mnt = fh->fh_export->ex_path.mnt,
> 			 .dentry = fh->fh_dentry};
> -	return nfserrno(vfs_getattr(&p, stat, STATX_BASIC_STATS,
> -				    AT_STATX_SYNC_AS_STAT));
> +	return nfsd_getattr(&p, stat, true);
> }
> 
> static inline int nfsd_create_is_exclusive(int createmode)
> diff --git a/include/linux/exportfs.h b/include/linux/exportfs.h
> index 3260fe714846..58f36022787e 100644
> --- a/include/linux/exportfs.h
> +++ b/include/linux/exportfs.h
> @@ -10,6 +10,8 @@ struct inode;
> struct iomap;
> struct super_block;
> struct vfsmount;
> +struct path;
> +struct kstat;
> 
> /* limit the handle size to NFSv4 handle size now */
> #define MAX_HANDLE_SZ 128
> @@ -224,6 +226,7 @@ struct export_operations {
> #define EXPORT_OP_SYNC_LOCKS		(0x20) /* Filesystem can't do
> 						  asychronous blocking locks */
> 	unsigned long	flags;
> +	int (*getattr)(struct path *, struct kstat *, bool);
> };
> 
> extern int exportfs_encode_inode_fh(struct inode *inode, struct fid *fid,
> -- 
> 2.33.1
> 

--
Chuck Lever