On 3/30/22 20:36, syzbot wrote:
> Hello,
>
> syzbot found the following issue on:
>
> HEAD commit: 52d543b5497c Merge tag 'for-linus-5.17-1' of https://githu..
> git tree: upstream
> console output: https://syzkaller.appspot.com/x/log.txt?x=17b804fb700000
> kernel config: https://syzkaller.appspot.com/x/.config?x=7094767cefc58fb9
> dashboard link: https://syzkaller.appspot.com/bug?extid=99d6c66dbbc484f50e1c
> compiler: gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2
> syz repro: https://syzkaller.appspot.com/x/repro.syz?x=161c4739700000
> C reproducer: https://syzkaller.appspot.com/x/repro.c?x=16432d51700000
>
> IMPORTANT: if you fix the issue, please add the following tag to the commit:
> Reported-by: [email protected]
>
> em28xx 5-1:0.130: Config register raw data: 0xfffffffb
> em28xx 5-1:0.130: AC97 chip type couldn't be determined
> em28xx 5-1:0.130: No AC97 audio processor
> ==================================================================
> BUG: KASAN: use-after-free in __list_add_valid+0x93/0xa0 lib/list_debug.c:26
> Read of size 8 at addr ffff888027458250 by task kworker/1:1/40
>
Just want to warn anyone looking into this bug.
I came up with the fix, that passed syzbot testing and patch has been in
Linus' tree for couple of months: see commit 2c98b8a3458d ("media:
em28xx: add missing em28xx_close_extension").
After some time Maximilian sent a report about kernel hung bug caused by
my fix [1]. Just random hung caused by wrong reference counting
somewhere. No idea how to reproduce it locally or how to fix it.
I had to revert my fix, of course. That's why this bug appeared one more
time.
So, if you are going to send a fix, please, check that it does not have
same problem as mine.
Thanks
[1]
https://lore.kernel.org/all/[email protected]/
With regards,
Pavel Skripkin