netdev - Re: [syzbot] [wireless?] WARNING in ieee80211_link_release

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]

Message-ID: <cf6ab60c42705625a7b1d71e53db0edb78fe6e2d.camel@sipsolutions.net>
Date: Mon, 04 Sep 2023 08:57:11 +0200
From: Johannes Berg <johannes@...solutions.net>
To: syzbot <syzbot+9817a610349542589c42@...kaller.appspotmail.com>, 
	davem@...emloft.net, edumazet@...gle.com, kuba@...nel.org, 
	linux-kernel@...r.kernel.org, linux-wireless@...r.kernel.org, 
	netdev@...r.kernel.org, pabeni@...hat.com, syzkaller-bugs@...glegroups.com
Subject: Re: [syzbot] [wireless?] WARNING in ieee80211_link_release_channel

On Sat, 2023-09-02 at 18:48 -0700, syzbot wrote:
> 
> WARNING: CPU: 0 PID: 7597 at net/mac80211/chan.c:2021 ieee80211_link_release_channel+0x19f/0x200 net/mac80211/chan.c:2021
> 

That's a lockdep warning, so it's kind of a known issue. I don't know if
I've yet encountered this specific way to get the locking checks
screaming, but generally we knew for a while that the locking was a bit
messy and possibly to some extent broken (*).

Anyway, point is - this is basically the reason I did this series:
https://lore.kernel.org/linux-wireless/20230828115927.116700-41-johannes@sipsolutions.net/

so this should go away - though sadly only in -next. Per the footnote, I
believe that the issue in the current kernel is mostly not an issue, and
it should only happen with multi-link in the first place. There may be
an issue in what syzbot found, processing a multi-link response or
something though.

johannes

(*) in many cases it's not _actually_ broken because we hold some common
outer lock anyway, so no data races are possible