memory leak in xskq_create

memory leak in xskq_create

[syzbot]

Hello,

syzbot found the following issue on:

HEAD commit:    a68a0262 mm/madvise: remove racy mm ownership check
git tree:       upstream
console output: https://syzkaller.appspot.com/x/log.txt?x=165b9413500000
kernel config:  https://syzkaller.appspot.com/x/.config?x=4305fa9ea70c7a9f
dashboard link: https://syzkaller.appspot.com/bug?extid=cfa88ddd0655afa88763
compiler:       gcc (GCC) 10.1.0-syz 20200507
syz repro:      https://syzkaller.appspot.com/x/repro.syz?x=1180a237500000
C reproducer:   https://syzkaller.appspot.com/x/repro.c?x=114067cf500000

IMPORTANT: if you fix the issue, please add the following tag to the commit:
Reported-by: syzbot+cfa88ddd0655afa88763@syzkaller.appspotmail.com

Debian GNU/Linux 9 syzkaller ttyS0
Warning: Permanently added '10.128.0.50' (ECDSA) to the list of known hosts.
executing program
executing program
BUG: memory leak
unreferenced object 0xffff88810f897940 (size 64):
  comm "syz-executor991", pid 8502, jiffies 4294942194 (age 14.080s)
  hex dump (first 32 bytes):
    7f 00 00 00 80 00 00 00 00 00 00 00 00 00 00 00  ................
    00 a0 37 0c 81 88 ff ff 00 00 00 00 00 00 00 00  ..7.............
  backtrace:
    [<00000000639d0dd1>] xskq_create+0x23/0xd0 include/linux/slab.h:552
    [<00000000b680b035>] xsk_init_queue net/xdp/xsk.c:508 [inline]
    [<00000000b680b035>] xsk_setsockopt+0x1c4/0x590 net/xdp/xsk.c:875
    [<000000002b302260>] __sys_setsockopt+0x1b0/0x360 net/socket.c:2132
    [<00000000ae03723e>] __do_sys_setsockopt net/socket.c:2143 [inline]
    [<00000000ae03723e>] __se_sys_setsockopt net/socket.c:2140 [inline]
    [<00000000ae03723e>] __x64_sys_setsockopt+0x22/0x30 net/socket.c:2140
    [<0000000005c2b4a0>] do_syscall_64+0x2d/0x70 arch/x86/entry/common.c:46
    [<0000000003db140f>] entry_SYSCALL_64_after_hwframe+0x44/0xa9

BUG: memory leak
unreferenced object 0xffff88810f8979c0 (size 64):
  comm "syz-executor991", pid 8503, jiffies 4294942194 (age 14.080s)
  hex dump (first 32 bytes):
    ff 03 00 00 00 04 00 00 00 00 00 00 00 00 00 00  ................
    00 00 13 12 81 88 ff ff 00 00 00 00 00 00 00 00  ................
  backtrace:
    [<00000000639d0dd1>] xskq_create+0x23/0xd0 include/linux/slab.h:552
    [<00000000b680b035>] xsk_init_queue net/xdp/xsk.c:508 [inline]
    [<00000000b680b035>] xsk_setsockopt+0x1c4/0x590 net/xdp/xsk.c:875
    [<000000002b302260>] __sys_setsockopt+0x1b0/0x360 net/socket.c:2132
    [<00000000ae03723e>] __do_sys_setsockopt net/socket.c:2143 [inline]
    [<00000000ae03723e>] __se_sys_setsockopt net/socket.c:2140 [inline]
    [<00000000ae03723e>] __x64_sys_setsockopt+0x22/0x30 net/socket.c:2140
    [<0000000005c2b4a0>] do_syscall_64+0x2d/0x70 arch/x86/entry/common.c:46
    [<0000000003db140f>] entry_SYSCALL_64_after_hwframe+0x44/0xa9



---
This report is generated by a bot. It may contain errors.
See https://goo.gl/tpsmEJ for more information about syzbot.
syzbot engineers can be reached at syzkaller@googlegroups.com.

syzbot will keep track of this issue. See:
https://goo.gl/tpsmEJ#status for how to communicate with syzbot.
syzbot can test patches for this issue, for details see:
https://goo.gl/tpsmEJ#testing-patches

Re: memory leak in xskq_create

[Peilin Ye]

Hi all,

On Sun, Dec 13, 2020 at 06:53:10AM -0800, syzbot wrote:
> BUG: memory leak
> unreferenced object 0xffff88810f897940 (size 64):
>   comm "syz-executor991", pid 8502, jiffies 4294942194 (age 14.080s)
>   hex dump (first 32 bytes):
>     7f 00 00 00 80 00 00 00 00 00 00 00 00 00 00 00  ................
>     00 a0 37 0c 81 88 ff ff 00 00 00 00 00 00 00 00  ..7.............
>   backtrace:
>     [<00000000639d0dd1>] xskq_create+0x23/0xd0 include/linux/slab.h:552
>     [<00000000b680b035>] xsk_init_queue net/xdp/xsk.c:508 [inline]
>     [<00000000b680b035>] xsk_setsockopt+0x1c4/0x590 net/xdp/xsk.c:875
>     [<000000002b302260>] __sys_setsockopt+0x1b0/0x360 net/socket.c:2132
>     [<00000000ae03723e>] __do_sys_setsockopt net/socket.c:2143 [inline]
>     [<00000000ae03723e>] __se_sys_setsockopt net/socket.c:2140 [inline]
>     [<00000000ae03723e>] __x64_sys_setsockopt+0x22/0x30 net/socket.c:2140
>     [<0000000005c2b4a0>] do_syscall_64+0x2d/0x70 arch/x86/entry/common.c:46
>     [<0000000003db140f>] entry_SYSCALL_64_after_hwframe+0x44/0xa9

I have tested the following diff locally against syzbot's reproducer,
and sent a patch to it [1] for testing.  I will send a real patch here
tomorrow if syzbot is happy about it.  Please see explanation below.

--- a/net/xdp/xsk_buff_pool.c
+++ b/net/xdp/xsk_buff_pool.c
@@ -37,6 +37,9 @@ void xp_destroy(struct xsk_buff_pool *pool)
        if (!pool)
                return;

+       xskq_destroy(pool->fq);
+       xskq_destroy(pool->cq);
+
        kvfree(pool->heads);
        kvfree(pool);
 }
@@ -234,16 +237,6 @@ static void xp_release_deferred(struct work_struct *work)
        xp_clear_dev(pool);
        rtnl_unlock();

-       if (pool->fq) {
-               xskq_destroy(pool->fq);
-               pool->fq = NULL;
-       }
-
-       if (pool->cq) {
-               xskq_destroy(pool->cq);
-               pool->cq = NULL;
-       }
-
        xdp_put_umem(pool->umem, false);
        xp_destroy(pool);
 }

When xsk_bind() allocates `xs->pool` but something else goes wrong:

		xs->pool = xp_create_and_assign_umem(xs, xs->umem);
		[...]
		if (err) {
			xp_destroy(xs->pool);

xp_destroy() doesn't free `pool->{f,c}q`, causing a memory leak.  Move
`xskq_destroy(pool->{f,c}q)` from xp_release_deferred() to xp_destroy().

Also, since xskq_destroy() already does null check, I think it's
unnecessary to do it again here?

Thanks,
Peilin Ye

[1] https://syzkaller.appspot.com/bug?id=fea808dfe3c6dfdd6ba9778becbffe0b14e91294

Re: memory leak in xskq_create

[Björn Töpel]

On 2020-12-16 19:11, Peilin Ye wrote:
> Hi all,
> 
> On Sun, Dec 13, 2020 at 06:53:10AM -0800, syzbot wrote:
>> BUG: memory leak
>> unreferenced object 0xffff88810f897940 (size 64):
>>    comm "syz-executor991", pid 8502, jiffies 4294942194 (age 14.080s)
>>    hex dump (first 32 bytes):
>>      7f 00 00 00 80 00 00 00 00 00 00 00 00 00 00 00  ................
>>      00 a0 37 0c 81 88 ff ff 00 00 00 00 00 00 00 00  ..7.............
>>    backtrace:
>>      [<00000000639d0dd1>] xskq_create+0x23/0xd0 include/linux/slab.h:552
>>      [<00000000b680b035>] xsk_init_queue net/xdp/xsk.c:508 [inline]
>>      [<00000000b680b035>] xsk_setsockopt+0x1c4/0x590 net/xdp/xsk.c:875
>>      [<000000002b302260>] __sys_setsockopt+0x1b0/0x360 net/socket.c:2132
>>      [<00000000ae03723e>] __do_sys_setsockopt net/socket.c:2143 [inline]
>>      [<00000000ae03723e>] __se_sys_setsockopt net/socket.c:2140 [inline]
>>      [<00000000ae03723e>] __x64_sys_setsockopt+0x22/0x30 net/socket.c:2140
>>      [<0000000005c2b4a0>] do_syscall_64+0x2d/0x70 arch/x86/entry/common.c:46
>>      [<0000000003db140f>] entry_SYSCALL_64_after_hwframe+0x44/0xa9
> 
> I have tested the following diff locally against syzbot's reproducer,
> and sent a patch to it [1] for testing.  I will send a real patch here
> tomorrow if syzbot is happy about it.  Please see explanation below.
>

Hi Peilin Ye!

Thanks for taking a look! Magnus has already addressed this problem in
another patch [1].


Cheers,
Björn



[1] 
https://lore.kernel.org/bpf/20201214085127.3960-1-magnus.karlsson@gmail.com/

Re: memory leak in xskq_create

[Peilin Ye]

Hi Björn,

On Thu, Dec 17, 2020 at 08:12:26AM +0100, Björn Töpel wrote:
> On 2020-12-16 19:11, Peilin Ye wrote:
> > I have tested the following diff locally against syzbot's reproducer,
> > and sent a patch to it [1] for testing.  I will send a real patch here
> > tomorrow if syzbot is happy about it.  Please see explanation below.
> 
> Hi Peilin Ye!
> 
> Thanks for taking a look! Magnus has already addressed this problem in
> another patch [1].

No problem!  Oops, I searched for "cfa88ddd0655afa88763" on LKML and
didn't notice Magnus's patch.  I should have searched netdev too.

Thanks,
Peilin Ye