linux-kernel - Re: [syzbot] possible deadlock in smc_switch_to

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:   Sat, 20 Nov 2021 17:50:00 +0300
From:   Pavel Skripkin <paskripkin@...il.com>
To:     syzbot <syzbot+e979d3597f48262cb4ee@...kaller.appspotmail.com>,
        davem@...emloft.net, kgraul@...ux.ibm.com, kuba@...nel.org,
        linux-kernel@...r.kernel.org, linux-s390@...r.kernel.org,
        netdev@...r.kernel.org, syzkaller-bugs@...glegroups.com
Subject: Re: [syzbot] possible deadlock in smc_switch_to_fallback

On 11/20/21 05:47, syzbot wrote:
> Hello,
> 
> syzbot found the following issue on:
> 
> HEAD commit:    9539ba4308ad Merge tag 'riscv-for-linus-5.16-rc2' of git:/..
> git tree:       upstream
> console output: https://syzkaller.appspot.com/x/log.txt?x=17f79d01b00000
> kernel config:  https://syzkaller.appspot.com/x/.config?x=6d3b8fd1977c1e73
> dashboard link: https://syzkaller.appspot.com/bug?extid=e979d3597f48262cb4ee
> compiler:       gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2
> 
> Unfortunately, I don't have any reproducer for this issue yet.
> 
> IMPORTANT: if you fix the issue, please add the following tag to the commit:
> Reported-by: syzbot+e979d3597f48262cb4ee@...kaller.appspotmail.com
> 
> ============================================
> WARNING: possible recursive locking detected
> 5.16.0-rc1-syzkaller #0 Not tainted
> --------------------------------------------
> syz-executor.3/1337 is trying to acquire lock:
> ffff88809466ce58 (&ei->socket.wq.wait){..-.}-{2:2}, at: spin_lock include/linux/spinlock.h:349 [inline]
> ffff88809466ce58 (&ei->socket.wq.wait){..-.}-{2:2}, at: smc_switch_to_fallback+0x3d5/0x8c0 net/smc/af_smc.c:588
>  > but task is already holding lock:
> ffff88809466c258 (&ei->socket.wq.wait){..-.}-{2:2}, at: smc_switch_to_fallback+0x3ca/0x8c0 net/smc/af_smc.c:587
>  > other info that might help us debug this:
>   Possible unsafe locking scenario:
> 
>         CPU0
>         ----
>    lock(&ei->socket.wq.wait);
>    lock(&ei->socket.wq.wait);
> 

There is simple code block in net/smc/af_smc.c:

		spin_lock_irqsave(&smc_wait->lock, flags);
		spin_lock(&clc_wait->lock);
		list_splice_init(&smc_wait->head, &clc_wait->head);
		spin_unlock(&clc_wait->lock);
		spin_unlock_irqrestore(&smc_wait->lock, flags);

smc_wait and clc_wait are too different pointers (based on report), but 
these 2 different wait_queue locks registered to lockdep map via 
sock_alloc_inode(), where init_waitqueue_head(&ei->socket.wq.wait); is 
called. So any nested wait_queue_head_t locking will cause lockdep warning.

Have no idea how to handle it, just my thoughts about root case :)





With regards,
Pavel Skripkin