On 2/25/20 5:38 PM, Jan Kara wrote:
Lockdep complains about a chain: sb_internal#2 --> &ei->xattr_sem#2 --> fs_reclaim and shrink_dentry_list -> ext2_evict_inode -> ext2_xattr_delete_inode -> down_write(ei->xattr_sem) creating a locking cycle in the reclaim path. This is however a false positive because when we are in ext2_evict_inode() we are the only holder of the inode reference and nobody else should touch xattr_sem of that inode. So we cannot ever block on acquiring the xattr_sem in the reclaim path. Silence the lockdep warning by using down_write_trylock() in ext2_xattr_delete_inode() to not create false locking dependency. Reported-by: "J. R. Okajima" <hooanon05g@xxxxxxxxx> Signed-off-by: Jan Kara <jack@xxxxxxx>
Agreed with evict() will only be called when it's the last reference going down and so we won't be blocked on xattr_sem.
Thanks for clearly explaining the problem in the cover letter. Reviewed-by: Ritesh Harjani <riteshh@xxxxxxxxxxxxx>
--- fs/ext2/xattr.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) Changes since v1: - changed WARN_ON to WARN_ON_ONCE diff --git a/fs/ext2/xattr.c b/fs/ext2/xattr.c index 0456bc990b5e..9ad07c7ef0b3 100644 --- a/fs/ext2/xattr.c +++ b/fs/ext2/xattr.c @@ -790,7 +790,15 @@ ext2_xattr_delete_inode(struct inode *inode) struct buffer_head *bh = NULL; struct ext2_sb_info *sbi = EXT2_SB(inode->i_sb); - down_write(&EXT2_I(inode)->xattr_sem); + /* + * We are the only ones holding inode reference. The xattr_sem should + * better be unlocked! We could as well just not acquire xattr_sem at + * all but this makes the code more futureproof. OTOH we need trylock + * here to avoid false-positive warning from lockdep about reclaim + * circular dependency. + */ + if (WARN_ON_ONCE(!down_write_trylock(&EXT2_I(inode)->xattr_sem))) + return; if (!EXT2_I(inode)->i_file_acl) goto cleanup;
