linux-kernel - Re: [PATCH 0/2] arm64: Allow the rescheduling IPI to bypass irq

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

Date:   Tue, 10 Nov 2020 15:48:58 +0000
From:   Valentin Schneider <valentin.schneider@....com>
To:     Peter Zijlstra <peterz@...radead.org>
Cc:     Marc Zyngier <maz@...nel.org>,
        LAK <linux-arm-kernel@...ts.infradead.org>,
        linux-kernel <linux-kernel@...r.kernel.org>,
        Will Deacon <will@...nel.org>,
        Catalin Marinas <catalin.marinas@....com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Android Kernel Team <kernel-team@...roid.com>
Subject: Re: [PATCH 0/2] arm64: Allow the rescheduling IPI to bypass irq_enter/exit


On 01/01/70 01:00, Valentin Schneider wrote:
> On 10/11/20 13:03, Peter Zijlstra wrote:
>> On Mon, Nov 02, 2020 at 10:30:50AM +0000, Valentin Schneider wrote:
>>
>>> Now, I'd like to pen exactly why we think it's okay to forgo irq_{enter,
>>> exit}() for that one IRQ and not any other.
>>
>> Thomas already said a few words on this, but basically scheduler_ipi()
>> is a NOP (*almost*), the IPI has no body. All it does is tickle the
>> return-from-interrupt path. So any setup and tear-down done for the
>> non-existing body is a waste of time.

Gotcha.

The pedant in me thinks this makes it more of a handler property than an
IRQ one, but I don't see a nice way to e.g. have this as a flag passed to
__request_percpu_irq() and not have it usable by random modules.