This is a note to let you know that I've just added the patch titled
net: avoid double iput when sock_alloc_file fails
to the 6.2-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary
The filename of the patch is:
net-avoid-double-iput-when-sock_alloc_file-fails.patch
and it can be found in the queue-6.2 subdirectory.
If you, or anyone else, feels it should not be added to the stable tree,
please let <stable@xxxxxxxxxxxxxxx> know about it.
>From 649c15c7691e9b13cbe9bf6c65c365350e056067 Mon Sep 17 00:00:00 2001
From: Thadeu Lima de Souza Cascardo <cascardo@xxxxxxxxxxxxx>
Date: Tue, 7 Mar 2023 14:37:07 -0300
Subject: net: avoid double iput when sock_alloc_file fails
From: Thadeu Lima de Souza Cascardo <cascardo@xxxxxxxxxxxxx>
commit 649c15c7691e9b13cbe9bf6c65c365350e056067 upstream.
When sock_alloc_file fails to allocate a file, it will call sock_release.
__sys_socket_file should then not call sock_release again, otherwise there
will be a double free.
[ 89.319884] ------------[ cut here ]------------
[ 89.320286] kernel BUG at fs/inode.c:1764!
[ 89.320656] invalid opcode: 0000 [#1] PREEMPT SMP NOPTI
[ 89.321051] CPU: 7 PID: 125 Comm: iou-sqp-124 Not tainted 6.2.0+ #361
[ 89.321535] RIP: 0010:iput+0x1ff/0x240
[ 89.321808] Code: d1 83 e1 03 48 83 f9 02 75 09 48 81 fa 00 10 00 00 77 05 83 e2 01 75 1f 4c 89 ef e8 fb d2 ba 00 e9 80 fe ff ff c3 cc cc cc cc <0f> 0b 0f 0b e9 d0 fe ff ff 0f 0b eb 8d 49 8d b4 24 08 01 00 00 48
[ 89.322760] RSP: 0018:ffffbdd60068bd50 EFLAGS: 00010202
[ 89.323036] RAX: 0000000000000000 RBX: ffff9d7ad3cacac0 RCX: 0000000000001107
[ 89.323412] RDX: 000000000003af00 RSI: 0000000000000000 RDI: ffff9d7ad3cacb40
[ 89.323785] RBP: ffffbdd60068bd68 R08: ffffffffffffffff R09: ffffffffab606438
[ 89.324157] R10: ffffffffacb3dfa0 R11: 6465686361657256 R12: ffff9d7ad3cacb40
[ 89.324529] R13: 0000000080000001 R14: 0000000080000001 R15: 0000000000000002
[ 89.324904] FS: 00007f7b28516740(0000) GS:ffff9d7aeb1c0000(0000) knlGS:0000000000000000
[ 89.325328] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 89.325629] CR2: 00007f0af52e96c0 CR3: 0000000002a02006 CR4: 0000000000770ee0
[ 89.326004] PKRU: 55555554
[ 89.326161] Call Trace:
[ 89.326298] <TASK>
[ 89.326419] __sock_release+0xb5/0xc0
[ 89.326632] __sys_socket_file+0xb2/0xd0
[ 89.326844] io_socket+0x88/0x100
[ 89.327039] ? io_issue_sqe+0x6a/0x430
[ 89.327258] io_issue_sqe+0x67/0x430
[ 89.327450] io_submit_sqes+0x1fe/0x670
[ 89.327661] io_sq_thread+0x2e6/0x530
[ 89.327859] ? __pfx_autoremove_wake_function+0x10/0x10
[ 89.328145] ? __pfx_io_sq_thread+0x10/0x10
[ 89.328367] ret_from_fork+0x29/0x50
[ 89.328576] RIP: 0033:0x0
[ 89.328732] Code: Unable to access opcode bytes at 0xffffffffffffffd6.
[ 89.329073] RSP: 002b:0000000000000000 EFLAGS: 00000202 ORIG_RAX: 00000000000001a9
[ 89.329477] RAX: 0000000000000000 RBX: 0000000000000000 RCX: 00007f7b28637a3d
[ 89.329845] RDX: 00007fff4e4318a8 RSI: 00007fff4e4318b0 RDI: 0000000000000400
[ 89.330216] RBP: 00007fff4e431830 R08: 00007fff4e431711 R09: 00007fff4e4318b0
[ 89.330584] R10: 0000000000000000 R11: 0000000000000202 R12: 00007fff4e441b38
[ 89.330950] R13: 0000563835e3e725 R14: 0000563835e40d10 R15: 00007f7b28784040
[ 89.331318] </TASK>
[ 89.331441] Modules linked in:
[ 89.331617] ---[ end trace 0000000000000000 ]---
Fixes: da214a475f8b ("net: add __sys_socket_file()")
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@xxxxxxxxxxxxx>
Reviewed-by: Jens Axboe <axboe@xxxxxxxxx>
Reviewed-by: Eric Dumazet <edumazet@xxxxxxxxxx>
Reviewed-by: Kuniyuki Iwashima <kuniyu@xxxxxxxxxx>
Link: https://lore.kernel.org/r/20230307173707.468744-1-cascardo@xxxxxxxxxxxxx
Signed-off-by: Jakub Kicinski <kuba@xxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
---
net/socket.c | 11 ++++-------
1 file changed, 4 insertions(+), 7 deletions(-)
--- a/net/socket.c
+++ b/net/socket.c
@@ -449,7 +449,9 @@ static struct file_system_type sock_fs_t
*
* Returns the &file bound with @sock, implicitly storing it
* in sock->file. If dname is %NULL, sets to "".
- * On failure the return is a ERR pointer (see linux/err.h).
+ *
+ * On failure @sock is released, and an ERR pointer is returned.
+ *
* This function uses GFP_KERNEL internally.
*/
@@ -1613,7 +1615,6 @@ static struct socket *__sys_socket_creat
struct file *__sys_socket_file(int family, int type, int protocol)
{
struct socket *sock;
- struct file *file;
int flags;
sock = __sys_socket_create(family, type, protocol);
@@ -1624,11 +1625,7 @@ struct file *__sys_socket_file(int famil
if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK))
flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK;
- file = sock_alloc_file(sock, flags, NULL);
- if (IS_ERR(file))
- sock_release(sock);
-
- return file;
+ return sock_alloc_file(sock, flags, NULL);
}
int __sys_socket(int family, int type, int protocol)
Patches currently in stable-queue which might be from cascardo@xxxxxxxxxxxxx are
queue-6.2/net-avoid-double-iput-when-sock_alloc_file-fails.patch
