From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Yan, Zheng" Subject: [PATCH] namei: revalidate negative dentry before return -ENOENT Date: Fri, 9 Oct 2015 17:01:03 +0800 Message-ID: <1444381263-6505-1-git-send-email-zyan@redhat.com> Cc: linux-fsdevel@vger.kernel.org, "Yan, Zheng" To: viro@ZenIV.linux.org.uk Return-path: Received: from mx1.redhat.com ([209.132.183.28]:43525 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932632AbbJIJBZ (ORCPT ); Fri, 9 Oct 2015 05:01:25 -0400 Sender: linux-fsdevel-owner@vger.kernel.org List-ID: After commit 766c4cbf "namei: d_is_negative() should be checked before ->d_seq validation", lookup_fast() can return -ENOENT without revalidating the negative dentry. This breaks network filesystems that allow creating/deleting files from multiple hosts. Signed-off-by: Yan, Zheng --- fs/namei.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/fs/namei.c b/fs/namei.c index 726d211..5ffbf53 100644 --- a/fs/namei.c +++ b/fs/namei.c @@ -1558,8 +1558,6 @@ static int lookup_fast(struct nameidata *nd, negative = d_is_negative(dentry); if (read_seqcount_retry(&dentry->d_seq, seq)) return -ECHILD; - if (negative) - return -ENOENT; /* * This sequence count validates that the parent had no @@ -1580,6 +1578,10 @@ static int lookup_fast(struct nameidata *nd, goto unlazy; } } + + if (negative) + return -ENOENT; + path->mnt = mnt; path->dentry = dentry; if (likely(__follow_mount_rcu(nd, path, inode, seqp))) -- 1.9.3