9p: add refcount to p9_fid struct

Fix race issue in fid contention.

Eric's and Greg's patch offer a mechanism to fix open-unlink-f*syscall
bug in 9p. But there is race issue in fid parallel accesses.
As Greg's patch stores all of fids from opened files into according inode,
so all the lookup fid ops can retrieve fid from inode preferentially. But
there is no mechanism to handle the fid contention issue. For example,
there are two threads get the same fid in the same time and one of them
clunk the fid before the other thread ready to discard the fid. In this
scenario, it will lead to some fatal problems, even kernel core dump.

I introduce a mechanism to fix this race issue. A counter field introduced
into p9_fid struct to store the reference counter to the fid. When a fid
is allocated from the inode or dentry, the counter will increase, and
will decrease at the end of its occupation. It is guaranteed that the
fid won't be clunked before the reference counter go down to 0, then
we can avoid the clunked fid to be used.

tests:
race issue test from the old test case:
for file in {01..50}; do touch f.${file}; done
seq 1 1000 | xargs -n 1 -P 50 -I{} cat f.* > /dev/null

open-unlink-f*syscall test:
I have tested for f*syscall include: ftruncate fstat fchown fchmod faccessat.

Link: http://lkml.kernel.org/r/20200923141146.90046-5-jianyong.wu@arm.com
Fixes: 478ba09edc1f ("fs/9p: search open fids first")
Signed-off-by: Jianyong Wu <jianyong.wu@arm.com>
Signed-off-by: Dominique Martinet <asmadeus@codewreck.org>

1 files changed, 29 insertions, 8 deletions

diff --git a/fs/9p/vfs_inode.c b/fs/9p/vfs_inode.c
index 6b243ffcbcf0..4a937fac1acb 100644
--- a/fs/9p/vfs_inode.c
+++ b/fs/9p/vfs_inode.c
@@ -551,6 +551,7 @@ static int v9fs_remove(struct inode *dir, struct dentry *dentry, int flags)
 	if (v9fs_proto_dotl(v9ses))
 		retval = p9_client_unlinkat(dfid, dentry->d_name.name,
 					    v9fs_at_to_dotl_flags(flags));
+	p9_client_clunk(dfid);
 	if (retval == -EOPNOTSUPP) {
 		/* Try the one based on path */
 		v9fid = v9fs_fid_clone(dentry);
@@ -595,14 +596,12 @@ v9fs_create(struct v9fs_session_info *v9ses, struct inode *dir,
 {
 	int err;
 	const unsigned char *name;
-	struct p9_fid *dfid, *ofid, *fid;
+	struct p9_fid *dfid, *ofid = NULL, *fid = NULL;
 	struct inode *inode;
 
 	p9_debug(P9_DEBUG_VFS, "name %pd\n", dentry);
 
 	err = 0;
-	ofid = NULL;
-	fid = NULL;
 	name = dentry->d_name.name;
 	dfid = v9fs_parent_fid(dentry);
 	if (IS_ERR(dfid)) {
@@ -616,12 +615,14 @@ v9fs_create(struct v9fs_session_info *v9ses, struct inode *dir,
 	if (IS_ERR(ofid)) {
 		err = PTR_ERR(ofid);
 		p9_debug(P9_DEBUG_VFS, "p9_client_walk failed %d\n", err);
+		p9_client_clunk(dfid);
 		return ERR_PTR(err);
 	}
 
 	err = p9_client_fcreate(ofid, name, perm, mode, extension);
 	if (err < 0) {
 		p9_debug(P9_DEBUG_VFS, "p9_client_fcreate failed %d\n", err);
+		p9_client_clunk(dfid);
 		goto error;
 	}
 
@@ -633,6 +634,7 @@ v9fs_create(struct v9fs_session_info *v9ses, struct inode *dir,
 			p9_debug(P9_DEBUG_VFS,
 				   "p9_client_walk failed %d\n", err);
 			fid = NULL;
+			p9_client_clunk(dfid);
 			goto error;
 		}
 		/*
@@ -643,11 +645,13 @@ v9fs_create(struct v9fs_session_info *v9ses, struct inode *dir,
 			err = PTR_ERR(inode);
 			p9_debug(P9_DEBUG_VFS,
 				   "inode creation failed %d\n", err);
+			p9_client_clunk(dfid);
 			goto error;
 		}
 		v9fs_fid_add(dentry, fid);
 		d_instantiate(dentry, inode);
 	}
+	p9_client_clunk(dfid);
 	return ofid;
 error:
 	if (ofid)
@@ -760,6 +764,7 @@ struct dentry *v9fs_vfs_lookup(struct inode *dir, struct dentry *dentry,
 	 */
 	name = dentry->d_name.name;
 	fid = p9_client_walk(dfid, 1, &name, 1);
+	p9_client_clunk(dfid);
 	if (fid == ERR_PTR(-ENOENT))
 		inode = NULL;
 	else if (IS_ERR(fid))
@@ -910,7 +915,7 @@ v9fs_vfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 	struct inode *old_inode;
 	struct inode *new_inode;
 	struct v9fs_session_info *v9ses;
-	struct p9_fid *oldfid;
+	struct p9_fid *oldfid, *dfid;
 	struct p9_fid *olddirfid;
 	struct p9_fid *newdirfid;
 	struct p9_wstat wstat;
@@ -927,13 +932,20 @@ v9fs_vfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 	if (IS_ERR(oldfid))
 		return PTR_ERR(oldfid);
 
-	olddirfid = clone_fid(v9fs_parent_fid(old_dentry));
+	dfid = v9fs_parent_fid(old_dentry);
+	olddirfid = clone_fid(dfid);
+	if (dfid && !IS_ERR(dfid))
+		p9_client_clunk(dfid);
+
 	if (IS_ERR(olddirfid)) {
 		retval = PTR_ERR(olddirfid);
 		goto done;
 	}
 
-	newdirfid = clone_fid(v9fs_parent_fid(new_dentry));
+	dfid = v9fs_parent_fid(new_dentry);
+	newdirfid = clone_fid(dfid);
+	p9_client_clunk(dfid);
+
 	if (IS_ERR(newdirfid)) {
 		retval = PTR_ERR(newdirfid);
 		goto clunk_olddir;
@@ -990,6 +1002,7 @@ clunk_olddir:
 	p9_client_clunk(olddirfid);
 
 done:
+	p9_client_clunk(oldfid);
 	return retval;
 }
 
@@ -1022,6 +1035,7 @@ v9fs_vfs_getattr(const struct path *path, struct kstat *stat,
 		return PTR_ERR(fid);
 
 	st = p9_client_stat(fid);
+	p9_client_clunk(fid);
 	if (IS_ERR(st))
 		return PTR_ERR(st);
 
@@ -1042,7 +1056,7 @@ v9fs_vfs_getattr(const struct path *path, struct kstat *stat,
 
 static int v9fs_vfs_setattr(struct dentry *dentry, struct iattr *iattr)
 {
-	int retval;
+	int retval, use_dentry = 0;
 	struct v9fs_session_info *v9ses;
 	struct p9_fid *fid = NULL;
 	struct p9_wstat wstat;
@@ -1058,8 +1072,10 @@ static int v9fs_vfs_setattr(struct dentry *dentry, struct iattr *iattr)
 		fid = iattr->ia_file->private_data;
 		WARN_ON(!fid);
 	}
-	if (!fid)
+	if (!fid) {
 		fid = v9fs_fid_lookup(dentry);
+		use_dentry = 1;
+	}
 	if(IS_ERR(fid))
 		return PTR_ERR(fid);
 
@@ -1089,6 +1105,10 @@ static int v9fs_vfs_setattr(struct dentry *dentry, struct iattr *iattr)
 		filemap_write_and_wait(d_inode(dentry)->i_mapping);
 
 	retval = p9_client_wstat(fid, &wstat);
+
+	if (use_dentry)
+		p9_client_clunk(fid);
+
 	if (retval < 0)
 		return retval;
 
@@ -1213,6 +1233,7 @@ static const char *v9fs_vfs_get_link(struct dentry *dentry,
 		return ERR_PTR(-EBADF);
 
 	st = p9_client_stat(fid);
+	p9_client_clunk(fid);
 	if (IS_ERR(st))
 		return ERR_CAST(st);