Skip to contentSkip to conversation

nfsd: fix handling of cached open files in nfsd4_open codepath

PATCH Superseded Jeff Layton 1 participant 1 message Jan 4, 2023

Subsystems (2)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Jeff Layton

author maintainer
#0 lore raw

PATCH 1/1 nfsd: fix handling of cached open files in nfsd4_open codepath

Commit fb70bf124b05 ("NFSD: Instantiate a struct file when creating a regular NFSv4 file") added the ability to cache an open fd over a compound. There are a couple of problems with the way this currently works:

It's racy, as a newly-created nfsd_file can end up with its PENDING bit cleared while the nf is hashed, and the nf_file pointer is still zeroed out. Other tasks can find it in this state and they expect to see a valid nf_file, and can oops if nf_file is NULL.

Also, there is no guarantee that we'll end up creating a new nfsd_file if one is already in the hash. If an extant entry is in the hash with a valid nf_file, nfs4_get_vfs_file will clobber its nf_file pointer with the value of op_file and the old nf_file will leak.

Fix both issues by changing nfsd_file_acquire to take an optional file pointer. If one is present when this is called, we'll take a new reference to it instead of trying to open the file. If the nfsd_file already has a valid nf_file, we'll just ignore the optional file and pass the nfsd_file back as-is.

Also rework the tracepoints a bit to allow for a cached open variant, and don't try to avoid counting acquisitions in the case where we already have a cached open file.

---
 fs/nfsd/filecache.c | 49 ++++++++++++++----------------------------
 fs/nfsd/filecache.h |  5 ++---
 fs/nfsd/nfs4proc.c  |  2 +-
 fs/nfsd/nfs4state.c | 20 ++++++-----------
 fs/nfsd/trace.h     | 52 ++++++++++++---------------------------------
 5 files changed, 38 insertions(+), 90 deletions(-)

Only lightly tested with pynfs so far, but this should fix the oops that was reported.

fs/nfsd/filecache.c +16 -33
index f55798cc0947..f0ca9501edb2 100644
@@ -1075,8 +1075,8 @@ nfsd_file_is_cached(struct inode *inode)
 
 static __be32
 nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp,
-		     unsigned int may_flags, struct nfsd_file **pnf,
-		     bool open, bool want_gc)
+		     unsigned int may_flags, struct file *file,
+		     struct nfsd_file **pnf, bool want_gc)
 {
 	struct nfsd_file_lookup_key key = {
 		.type	= NFSD_FILE_KEY_FULL,
@@ -1151,25 +1151,27 @@ nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp,
 
 out:
 	if (status == nfs_ok) {
-		if (open)
-			this_cpu_inc(nfsd_file_acquisitions);
+		this_cpu_inc(nfsd_file_acquisitions);
 		*pnf = nf;
 	}
 	put_cred(key.cred);
-	if (open)
-		trace_nfsd_file_acquire(rqstp, key.inode, may_flags, nf, status);
+	trace_nfsd_file_acquire(rqstp, key.inode, may_flags, nf, status);
 	return status;
 
 open_file:
 	trace_nfsd_file_alloc(nf);
 	nf->nf_mark = nfsd_file_mark_find_or_create(nf, key.inode);
 	if (nf->nf_mark) {
-		if (open) {
+		if (file) {
+			get_file(file);
+			nf->nf_file = file;
+			status = nfs_ok;
+			trace_nfsd_file_open_cached(nf, status);
+		} else {
 			status = nfsd_open_verified(rqstp, fhp, may_flags,
 						    &nf->nf_file);
 			trace_nfsd_file_open(nf, status);
-		} else
-			status = nfs_ok;
+		}
 	} else
 		status = nfserr_jukebox;
 	/*
@@ -1210,7 +1212,7 @@ __be32
 nfsd_file_acquire_gc(struct svc_rqst *rqstp, struct svc_fh *fhp,
 		     unsigned int may_flags, struct nfsd_file **pnf)
 {
-	return nfsd_file_do_acquire(rqstp, fhp, may_flags, pnf, true, true);
+	return nfsd_file_do_acquire(rqstp, fhp, may_flags, NULL, pnf, true);
 }
 
 /**
@@ -1218,6 +1220,7 @@ nfsd_file_acquire_gc(struct svc_rqst *rqstp, struct svc_fh *fhp,
  * @rqstp: the RPC transaction being executed
  * @fhp: the NFS filehandle of the file to be opened
  * @may_flags: NFSD_MAY_ settings for the file
+ * @file: cached, already-open file (may be NULL)
  * @pnf: OUT: new or found "struct nfsd_file" object
  *
  * The nfsd_file_object returned by this API is reference-counted
@@ -1229,30 +1232,10 @@ nfsd_file_acquire_gc(struct svc_rqst *rqstp, struct svc_fh *fhp,
  */
 __be32
 nfsd_file_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp,
-		  unsigned int may_flags, struct nfsd_file **pnf)
-{
-	return nfsd_file_do_acquire(rqstp, fhp, may_flags, pnf, true, false);
-}
-
-/**
- * nfsd_file_create - Get a struct nfsd_file, do not open
- * @rqstp: the RPC transaction being executed
- * @fhp: the NFS filehandle of the file just created
- * @may_flags: NFSD_MAY_ settings for the file
- * @pnf: OUT: new or found "struct nfsd_file" object
- *
- * The nfsd_file_object returned by this API is reference-counted
- * but not garbage-collected. The object is released immediately
- * one RCU grace period after the final nfsd_file_put().
- *
- * Returns nfs_ok and sets @pnf on success; otherwise an nfsstat in
- * network byte order is returned.
- */
-__be32
-nfsd_file_create(struct svc_rqst *rqstp, struct svc_fh *fhp,
-		 unsigned int may_flags, struct nfsd_file **pnf)
+		  unsigned int may_flags, struct file *file,
+		  struct nfsd_file **pnf)
 {
-	return nfsd_file_do_acquire(rqstp, fhp, may_flags, pnf, false, false);
+	return nfsd_file_do_acquire(rqstp, fhp, may_flags, file, pnf, false);
 }
 
 /*
fs/nfsd/filecache.h +2 -3
index b7efb2c3ddb1..ef0083cd4ea9 100644
@@ -59,8 +59,7 @@ bool nfsd_file_is_cached(struct inode *inode);
 __be32 nfsd_file_acquire_gc(struct svc_rqst *rqstp, struct svc_fh *fhp,
 		  unsigned int may_flags, struct nfsd_file **nfp);
 __be32 nfsd_file_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp,
-		  unsigned int may_flags, struct nfsd_file **nfp);
-__be32 nfsd_file_create(struct svc_rqst *rqstp, struct svc_fh *fhp,
-		  unsigned int may_flags, struct nfsd_file **nfp);
+		  unsigned int may_flags, struct file *file,
+		  struct nfsd_file **nfp);
 int nfsd_file_cache_stats_show(struct seq_file *m, void *v);
 #endif /* _FS_NFSD_FILECACHE_H */
fs/nfsd/nfs4proc.c +1 -1
index bd880d55f565..6b09cdd4b067 100644
@@ -735,7 +735,7 @@ nfsd4_commit(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	__be32 status;
 
 	status = nfsd_file_acquire(rqstp, &cstate->current_fh, NFSD_MAY_WRITE |
-				   NFSD_MAY_NOT_BREAK_LEASE, &nf);
+				   NFSD_MAY_NOT_BREAK_LEASE, NULL, &nf);
 	if (status != nfs_ok)
 		return status;
 
fs/nfsd/nfs4state.c +6 -14
index f7b9e12245dd..3df3ae84bd07 100644
@@ -5262,18 +5262,10 @@ static __be32 nfs4_get_vfs_file(struct svc_rqst *rqstp, struct nfs4_file *fp,
 	if (!fp->fi_fds[oflag]) {
 		spin_unlock(&fp->fi_lock);
 
-		if (!open->op_filp) {
-			status = nfsd_file_acquire(rqstp, cur_fh, access, &nf);
-			if (status != nfs_ok)
-				goto out_put_access;
-		} else {
-			status = nfsd_file_create(rqstp, cur_fh, access, &nf);
-			if (status != nfs_ok)
-				goto out_put_access;
-			nf->nf_file = open->op_filp;
-			open->op_filp = NULL;
-			trace_nfsd_file_create(rqstp, access, nf);
-		}
+		status = nfsd_file_acquire(rqstp, cur_fh, access,
+					   open->op_filp, &nf);
+		if (status != nfs_ok)
+			goto out_put_access;
 
 		spin_lock(&fp->fi_lock);
 		if (!fp->fi_fds[oflag]) {
@@ -6476,7 +6468,7 @@ nfs4_check_file(struct svc_rqst *rqstp, struct svc_fh *fhp, struct nfs4_stid *s,
 			goto out;
 		}
 	} else {
-		status = nfsd_file_acquire(rqstp, fhp, acc, &nf);
+		status = nfsd_file_acquire(rqstp, fhp, acc, NULL, &nf);
 		if (status)
 			return status;
 	}
@@ -7648,7 +7640,7 @@ static __be32 nfsd_test_lock(struct svc_rqst *rqstp, struct svc_fh *fhp, struct
 	struct inode *inode;
 	__be32 err;
 
-	err = nfsd_file_acquire(rqstp, fhp, NFSD_MAY_READ, &nf);
+	err = nfsd_file_acquire(rqstp, fhp, NFSD_MAY_READ, NULL, &nf);
 	if (err)
 		return err;
 	inode = fhp->fh_dentry->d_inode;
fs/nfsd/trace.h +13 -39

70 lines not on this page · raw message

Signature · 1 line
2.39.0