Re: [PATCH 1/2 v2] ceph: add ceph_lock_info support for file_lock
From: Xiubo Li
Date: Mon Nov 14 2022 - 08:01:31 EST
On 14/11/2022 19:24, Jeff Layton wrote:
On Mon, 2022-11-14 at 13:19 +0800, xiubli@xxxxxxxxxx wrote:
From: Xiubo Li <xiubli@xxxxxxxxxx>
When ceph releasing the file_lock it will try to get the inode pointer
from the fl->fl_file, which the memory could already be released by
another thread in filp_close(). Because in VFS layer the fl->fl_file
doesn't increase the file's reference counter.
Will switch to use ceph dedicate lock info to track the inode.
And in ceph_fl_release_lock() we should skip all the operations if
the fl->fl_u.ceph_fl.fl_inode is not set, which should come from
the request file_lock. And we will set fl->fl_u.ceph_fl.fl_inode when
inserting it to the inode lock list, which is when copying the lock.
Cc: stable@xxxxxxxxxxxxxxx
URL: https://tracker.ceph.com/issues/57986
Signed-off-by: Xiubo Li <xiubli@xxxxxxxxxx>
---
fs/ceph/locks.c | 18 +++++++++++++++---
include/linux/ceph/ceph_fs_fl.h | 26 ++++++++++++++++++++++++++
include/linux/fs.h | 2 ++
3 files changed, 43 insertions(+), 3 deletions(-)
create mode 100644 include/linux/ceph/ceph_fs_fl.h
diff --git a/fs/ceph/locks.c b/fs/ceph/locks.c
index 3e2843e86e27..d8385dd0076e 100644
--- a/fs/ceph/locks.c
+++ b/fs/ceph/locks.c
@@ -34,22 +34,34 @@ static void ceph_fl_copy_lock(struct file_lock *dst, struct file_lock *src)
{
struct ceph_file_info *fi = dst->fl_file->private_data;
struct inode *inode = file_inode(dst->fl_file);
+
atomic_inc(&ceph_inode(inode)->i_filelock_ref);
atomic_inc(&fi->num_locks);
+ dst->fl_u.ceph_fl.fl_inode = igrab(inode);
}
static void ceph_fl_release_lock(struct file_lock *fl)
{
struct ceph_file_info *fi = fl->fl_file->private_data;
- struct inode *inode = file_inode(fl->fl_file);
- struct ceph_inode_info *ci = ceph_inode(inode);
- atomic_dec(&fi->num_locks);
+ struct inode *inode = fl->fl_u.ceph_fl.fl_inode;
+ struct ceph_inode_info *ci;
+
+ /*
+ * If inode is NULL it should be a request file_lock,
+ * nothing we can do.
+ */
+ if (!inode)
+ return;
+
+ ci = ceph_inode(inode);
if (atomic_dec_and_test(&ci->i_filelock_ref)) {
/* clear error when all locks are released */
spin_lock(&ci->i_ceph_lock);
ci->i_ceph_flags &= ~CEPH_I_ERROR_FILELOCK;
spin_unlock(&ci->i_ceph_lock);
}
+ iput(inode);
+ atomic_dec(&fi->num_locks);
It doesn't look like this fixes the original issue. "fi" may be pointing
to freed memory at this point, right?
Yeah, I didn't fix this in the this patch. I fixed it in a dedicated 2/2
patch.
I think you may need to get rid of
the "num_locks" field in ceph_file_info, and do that in a different way?
This is a dedicated field for each 'file' struct. I have no idea how to
fix this in a different way yet.
Thanks!
- Xiubo
}
static const struct file_lock_operations ceph_fl_lock_ops = {
diff --git a/include/linux/ceph/ceph_fs_fl.h b/include/linux/ceph/ceph_fs_fl.h
new file mode 100644
index 000000000000..02a412b26095
--- /dev/null
+++ b/include/linux/ceph/ceph_fs_fl.h
@@ -0,0 +1,26 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * ceph_fs.h - Ceph constants and data types to share between kernel and
+ * user space.
+ *
+ * Most types in this file are defined as little-endian, and are
+ * primarily intended to describe data structures that pass over the
+ * wire or that are stored on disk.
+ *
+ * LGPL2
+ */
+
+#ifndef CEPH_FS_FL_H
+#define CEPH_FS_FL_H
+
+#include <linux/fs.h>
+
+/*
+ * Ceph lock info
+ */
+
+struct ceph_lock_info {
+ struct inode *fl_inode;
+};
+
+#endif
diff --git a/include/linux/fs.h b/include/linux/fs.h
index e654435f1651..db4810d19e26 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -1066,6 +1066,7 @@ bool opens_in_grace(struct net *);
/* that will die - we need it for nfs_lock_info */
#include <linux/nfs_fs_i.h>
+#include <linux/ceph/ceph_fs_fl.h>
/*
* struct file_lock represents a generic "file lock". It's used to represent
@@ -1119,6 +1120,7 @@ struct file_lock {
int state; /* state of grant or error if -ve */
unsigned int debug_id;
} afs;
+ struct ceph_lock_info ceph_fl;
} fl_u;
} __randomize_layout;