Skip to contentSkip to conversation

nfsd fixes for 2.6.29

PATCH New J. Bruce Fields 1 participant 4 messages Jan 27, 2009

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

The following changes are available in the git repository at:

git://linux-nfs.org/~bfields/linux.git for-2.6.29

Mainly three regression fixes, but I also snuck in two minor documentation updates. (I can stop that if it's annoying.)

--b.

J. Bruce Fields (3):
nfs: note that CONFIG_SUNRPC_XPRT_RDMA turns on server side support too
nfsd: fix null dereference on error path
nfsd: fix cred leak on every rpc

James Lentini (1):
update port number in NFS/RDMA documentation

Jeff Layton (1):
nfsd: only set file_lock.fl_lmops in nfsd4_lockt if a stateowner is found

 Documentation/filesystems/nfs-rdma.txt |    4 ++--
 fs/nfsd/auth.c                         |    3 +++
 fs/nfsd/nfs4state.c                    |    1 -
 net/sunrpc/Kconfig                     |    5 ++---
 4 files changed, 7 insertions(+), 6 deletions(-)
#1 lore raw
Quoted 14 lines · bfields

The following changes are available in the git repository at:

git://linux-nfs.org/~bfields/linux.git for-2.6.29

Mainly three regression fixes, but I also snuck in two minor documentation updates. (I can stop that if it's annoying.)

--b.

J. Bruce Fields (3):
nfs: note that CONFIG_SUNRPC_XPRT_RDMA turns on server side support too
nfsd: fix null dereference on error path
nfsd: fix cred leak on every rpc

Quoted 5 lines

James Lentini (1):
update port number in NFS/RDMA documentation

Jeff Layton (1):
nfsd: only set file_lock.fl_lmops in nfsd4_lockt if a stateowner is found

Jeff's patch (together with the previous patch it fixes) are also appropriate for stable; I'll follow up with the patches in question.

--b.

#2 lore raw

PATCH 1/1 nfsd: Ensure nfsv4 calls the underlying filesystem on LOCKT

From: J. Bruce Fields <[email protected]>

Since nfsv4 allows LOCKT without an open, but the ->lock() method is a file method, we fake up a struct file in the nfsv4 code with just the fields we need initialized. But we forgot to initialize the file operations, with the result that LOCKT never results in a call to the filesystem's ->lock() method (if it exists).

We could just add that one more initialization. But this hack of faking up a struct file with only some fields initialized seems the kind of thing that might cause more problems in the future. We should either do an open and get a real struct file, or make lock-testing an inode (not a file) method.

This patch does the former.

---
 fs/nfsd/nfs4state.c |   30 ++++++++++++++++++++----------
 1 files changed, 20 insertions(+), 10 deletions(-)
fs/nfsd/nfs4state.c +20 -10
index 06b89df..e62d0e3 100644
@@ -2782,6 +2782,25 @@ out:
 }
 
 /*
+ * The NFSv4 spec allows a client to do a LOCKT without holding an OPEN,
+ * so we do a temporary open here just to get an open file to pass to
+ * vfs_test_lock.  (Arguably perhaps test_lock should be done with an
+ * inode operation.)
+ */
+static int nfsd_test_lock(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file_lock *lock)
+{
+	struct file *file;
+	int err;
+
+	err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, &file);
+	if (err)
+		return err;
+	err = vfs_test_lock(file, lock);
+	nfsd_close(file);
+	return err;
+}
+
+/*
  * LOCKT operation
  */
 __be32
@@ -2789,7 +2808,6 @@ nfsd4_lockt(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	    struct nfsd4_lockt *lockt)
 {
 	struct inode *inode;
-	struct file file;
 	struct file_lock file_lock;
 	int error;
 	__be32 status;
@@ -2847,16 +2865,8 @@ nfsd4_lockt(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 
 	nfs4_transform_lock_offset(&file_lock);
 
-	/* vfs_test_lock uses the struct file _only_ to resolve the inode.
-	 * since LOCKT doesn't require an OPEN, and therefore a struct
-	 * file may not exist, pass vfs_test_lock a struct file with
-	 * only the dentry:inode set.
-	 */
-	memset(&file, 0, sizeof (struct file));
-	file.f_path.dentry = cstate->current_fh.fh_dentry;
-
 	status = nfs_ok;
-	error = vfs_test_lock(&file, &file_lock);
+	error = nfsd_test_lock(rqstp, &cstate->current_fh, &file_lock);
 	if (error) {
 		status = nfserrno(error);
 		goto out;
Signature · 1 line
1.5.6.3
#3 lore raw

PATCH 1/1 nfsd: only set file_lock.fl_lmops in nfsd4_lockt if a stateowner is found

From: Jeff Layton <[email protected]>

nfsd4_lockt does a search for a lockstateowner when building the lock struct to test. If one is found, it'll set fl_owner to it. Regardless of whether that happens, it'll also set fl_lmops. Given that this lock is basically a "lightweight" lock that's just used for checking conflicts, setting fl_lmops is probably not appropriate for it.

This behavior exposed a bug in DLM's GETLK implementation where it wasn't clearing out the fields in the file_lock before filling in conflicting lock info. While we were able to fix this in DLM, it still seems pointless and dangerous to set the fl_lmops this way when we may have a NULL lockstateowner.

---
 fs/nfsd/nfs4state.c |    1 -
 1 files changed, 0 insertions(+), 1 deletions(-)
fs/nfsd/nfs4state.c +0 -1
index 88db7d3..b6f60f4 100644
@@ -2871,7 +2871,6 @@ nfsd4_lockt(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		file_lock.fl_owner = (fl_owner_t)lockt->lt_stateowner;
 	file_lock.fl_pid = current->tgid;
 	file_lock.fl_flags = FL_POSIX;
-	file_lock.fl_lmops = &nfsd_posix_mng_ops;
 
 	file_lock.fl_start = lockt->lt_offset;
 	file_lock.fl_end = last_byte_offset(lockt->lt_offset, lockt->lt_length);
Signature · 1 line
1.5.6.3