linux-kernel - Re: [RT] 5.9-rt14 softirq_ctrl.lock vs listening

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] [day] [month] [year] [list]

Date:   Wed, 9 Dec 2020 12:47:37 +0100
From:   Sebastian Andrzej Siewior <bigeasy@...utronix.de>
To:     Peter Zijlstra <peterz@...radead.org>
Cc:     Mike Galbraith <efault@....de>, tglx@...utronix.de,
        linux-rt-users@...r.kernel.org,
        lkml <linux-kernel@...r.kernel.org>,
        Boqun Feng <boqun.feng@...il.com>,
        Ingo Molnar <mingo@...nel.org>, Will Deacon <will@...nel.org>
Subject: Re: [RT] 5.9-rt14  softirq_ctrl.lock vs listening_hash[i].lock
 lockdep splat

On 2020-12-09 11:05:45 [+0100], Peter Zijlstra wrote:
> In general we have the rule that as long as a lock is only ever used
> from task context (like the above ilb->lock, afaict) then it doesn't
> matter if you also take it with (soft)irqs disabled or not. But this
> softirq scheme breaks that. If you ever take a lock with BH disabled,
> you must now always take it with BH disabled, otherwise you risk
> deadlocks against the softirq_ctrl lock.
> 
> Or am I missing something obvious (again) ?

No. With this explanation it makes sense. Thank you.

Sebastian