linux-kernel - [PATCH AUTOSEL 5.16 19/42] cifs: unlock chan_lock before calling cifs_put_tcp

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Wed,  9 Feb 2022 13:32:51 -0500
From:   Sasha Levin <sashal@...nel.org>
To:     linux-kernel@...r.kernel.org, stable@...r.kernel.org
Cc:     Shyam Prasad N <sprasad@...rosoft.com>,
        Steve French <stfrench@...rosoft.com>,
        Sasha Levin <sashal@...nel.org>, sfrench@...ba.org,
        linux-cifs@...r.kernel.org, samba-technical@...ts.samba.org
Subject: [PATCH AUTOSEL 5.16 19/42] cifs: unlock chan_lock before calling cifs_put_tcp_session

From: Shyam Prasad N <sprasad@...rosoft.com>

[ Upstream commit 489f710a738e24d887823a010b8b206b4124e26f ]

While removing an smb session, we need to free up the
tcp session for each channel for that session. We were
doing this with chan_lock held. This results in a
cyclic dependency with cifs_tcp_ses_lock.

For now, unlock the chan_lock temporarily before calling
cifs_put_tcp_session. This should not cause any problem
for now, since we do not remove channels anywhere else.
And this code segment will not be called by two threads.

When we do implement the code for removing channels, we
will need to execute proper ref counting here.

Signed-off-by: Shyam Prasad N <sprasad@...rosoft.com>
Signed-off-by: Steve French <stfrench@...rosoft.com>
Signed-off-by: Sasha Levin <sashal@...nel.org>
---
 fs/cifs/connect.c | 8 ++------
 1 file changed, 2 insertions(+), 6 deletions(-)

diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c
index 1060164b984a7..4c8db1a91e940 100644
--- a/fs/cifs/connect.c
+++ b/fs/cifs/connect.c
@@ -1796,13 +1796,9 @@ void cifs_put_smb_ses(struct cifs_ses *ses)
 		int i;
 
 		for (i = 1; i < chan_count; i++) {
-			/*
-			 * note: for now, we're okay accessing ses->chans
-			 * without chan_lock. But when chans can go away, we'll
-			 * need to introduce ref counting to make sure that chan
-			 * is not freed from under us.
-			 */
+			spin_unlock(&ses->chan_lock);
 			cifs_put_tcp_session(ses->chans[i].server, 0);
+			spin_lock(&ses->chan_lock);
 			ses->chans[i].server = NULL;
 		}
 	}
-- 
2.34.1