[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20191130020225.20239-3-yukuai3@huawei.com>
Date: Sat, 30 Nov 2019 10:02:24 +0800
From: yu kuai <yukuai3@...wei.com>
To: <gregkh@...uxfoundation.org>, <rafael@...nel.org>,
<viro@...iv.linux.org.uk>, <rostedt@...dmis.org>,
<oleg@...hat.com>, <mchehab+samsung@...nel.org>, <corbet@....net>,
<tytso@....edu>, <jmorris@...ei.org>
CC: <linux-kernel@...r.kernel.org>, <linux-fsdevel@...r.kernel.org>,
<yukuai3@...wei.com>, <zhengbin13@...wei.com>,
<yi.zhang@...wei.com>, <chenxiang66@...ilicon.com>,
<xiexiuqi@...wei.com>
Subject: [PATCH 2/3] fs/libfs.c: use 'spin_lock_nested' when taking 'd_lock' for dentry in simple_empty
simple_emtpty currently use 'spin_lock_nested' for 'child' to avoid
confusion for lockdep. However, it's not used for 'dentry'.
In that case, there will be a problem if the caller called 'simple_empty'
with a parent's 'd_lock' held:
spin_lock(&dentry->d_parent->d_lock)
call simple_empty(dentry)
spin_lock(&dentry->d_lock) --> lockdep will report this
spin_lock_nested(child->d_lock, spin_lock_nested)
spin_unlock(child_lock)
spin_unlock($dentry->d_lock)
return from simple_empty
spin_unlock(&dentry->d_patrent->d_lock)
So, use 'DENTRY_D_LOCK_NESTED' for 'dentry' and 'DENTRY_D_LOCK_NESTED_TWICE'
for child.
Signed-off-by: yu kuai <yukuai3@...wei.com>
---
fs/libfs.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/fs/libfs.c b/fs/libfs.c
index 1463b038ffc4..54a37444f7f9 100644
--- a/fs/libfs.c
+++ b/fs/libfs.c
@@ -336,9 +336,9 @@ int simple_empty(struct dentry *dentry)
struct dentry *child;
int ret = 0;
- spin_lock(&dentry->d_lock);
+ spin_lock_nested(&dentry->d_lock, DENTRY_D_LOCK_NESTED);
list_for_each_entry(child, &dentry->d_subdirs, d_child) {
- spin_lock_nested(&child->d_lock, DENTRY_D_LOCK_NESTED);
+ spin_lock_nested(&child->d_lock, DENTRY_D_LOCK_NESTED_TWICE);
if (simple_positive(child)) {
spin_unlock(&child->d_lock);
goto out;
--
2.17.2
Powered by blists - more mailing lists