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, 28 insertions, 6 deletions

diff --git a/fs/9p/vfs_inode_dotl.c b/fs/9p/vfs_inode_dotl.c
index 08f2e089fb0e..823c2eb5f1bf 100644
--- a/fs/9p/vfs_inode_dotl.c
+++ b/fs/9p/vfs_inode_dotl.c
@@ -296,6 +296,7 @@ v9fs_vfs_atomic_open_dotl(struct inode *dir, struct dentry *dentry,
 
 	/* instantiate inode and assign the unopened fid to the dentry */
 	fid = p9_client_walk(dfid, 1, &name, 1);
+	p9_client_clunk(dfid);
 	if (IS_ERR(fid)) {
 		err = PTR_ERR(fid);
 		p9_debug(P9_DEBUG_VFS, "p9_client_walk failed %d\n", err);
@@ -408,7 +409,6 @@ static int v9fs_vfs_mkdir_dotl(struct inode *dir,
 	err = p9_client_mkdir_dotl(dfid, name, mode, gid, &qid);
 	if (err < 0)
 		goto error;
-
 	fid = p9_client_walk(dfid, 1, &name, 1);
 	if (IS_ERR(fid)) {
 		err = PTR_ERR(fid);
@@ -452,6 +452,7 @@ error:
 	if (fid)
 		p9_client_clunk(fid);
 	v9fs_put_acl(dacl, pacl);
+	p9_client_clunk(dfid);
 	return err;
 }
 
@@ -479,6 +480,7 @@ v9fs_vfs_getattr_dotl(const struct path *path, struct kstat *stat,
 	 */
 
 	st = p9_client_getattr_dotl(fid, P9_STATS_ALL);
+	p9_client_clunk(fid);
 	if (IS_ERR(st))
 		return PTR_ERR(st);
 
@@ -540,7 +542,7 @@ static int v9fs_mapped_iattr_valid(int iattr_valid)
 
 int v9fs_vfs_setattr_dotl(struct dentry *dentry, struct iattr *iattr)
 {
-	int retval;
+	int retval, use_dentry = 0;
 	struct p9_fid *fid = NULL;
 	struct p9_iattr_dotl p9attr;
 	struct inode *inode = d_inode(dentry);
@@ -565,8 +567,10 @@ int v9fs_vfs_setattr_dotl(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);
 
@@ -575,8 +579,11 @@ int v9fs_vfs_setattr_dotl(struct dentry *dentry, struct iattr *iattr)
 		filemap_write_and_wait(inode->i_mapping);
 
 	retval = p9_client_setattr(fid, &p9attr);
-	if (retval < 0)
+	if (retval < 0) {
+		if (use_dentry)
+			p9_client_clunk(fid);
 		return retval;
+	}
 
 	if ((iattr->ia_valid & ATTR_SIZE) &&
 	    iattr->ia_size != i_size_read(inode))
@@ -588,9 +595,15 @@ int v9fs_vfs_setattr_dotl(struct dentry *dentry, struct iattr *iattr)
 	if (iattr->ia_valid & ATTR_MODE) {
 		/* We also want to update ACL when we update mode bits */
 		retval = v9fs_acl_chmod(inode, fid);
-		if (retval < 0)
+		if (retval < 0) {
+			if (use_dentry)
+				p9_client_clunk(fid);
 			return retval;
+		}
 	}
+	if (use_dentry)
+		p9_client_clunk(fid);
+
 	return 0;
 }
 
@@ -742,6 +755,7 @@ error:
 	if (fid)
 		p9_client_clunk(fid);
 
+	p9_client_clunk(dfid);
 	return err;
 }
 
@@ -770,11 +784,15 @@ v9fs_vfs_link_dotl(struct dentry *old_dentry, struct inode *dir,
 		return PTR_ERR(dfid);
 
 	oldfid = v9fs_fid_lookup(old_dentry);
-	if (IS_ERR(oldfid))
+	if (IS_ERR(oldfid)) {
+		p9_client_clunk(dfid);
 		return PTR_ERR(oldfid);
+	}
 
 	err = p9_client_link(dfid, oldfid, dentry->d_name.name);
 
+	p9_client_clunk(dfid);
+	p9_client_clunk(oldfid);
 	if (err < 0) {
 		p9_debug(P9_DEBUG_VFS, "p9_client_link failed %d\n", err);
 		return err;
@@ -789,6 +807,7 @@ v9fs_vfs_link_dotl(struct dentry *old_dentry, struct inode *dir,
 			return PTR_ERR(fid);
 
 		v9fs_refresh_inode_dotl(fid, d_inode(old_dentry));
+		p9_client_clunk(fid);
 	}
 	ihold(d_inode(old_dentry));
 	d_instantiate(dentry, d_inode(old_dentry));
@@ -887,6 +906,8 @@ error:
 	if (fid)
 		p9_client_clunk(fid);
 	v9fs_put_acl(dacl, pacl);
+	p9_client_clunk(dfid);
+
 	return err;
 }
 
@@ -915,6 +936,7 @@ v9fs_vfs_get_link_dotl(struct dentry *dentry,
 	if (IS_ERR(fid))
 		return ERR_CAST(fid);
 	retval = p9_client_readlink(fid, &target);
+	p9_client_clunk(fid);
 	if (retval)
 		return ERR_PTR(retval);
 	set_delayed_call(done, kfree_link, target);