linux-kernel - Re: [PATCH 8/9] irqentry: define irqentry_exit_allow

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:   Tue, 04 Apr 2023 22:29:15 -0700
From:   Ankur Arora <ankur.a.arora@...cle.com>
To:     Thomas Gleixner <tglx@...utronix.de>
Cc:     linux-kernel@...r.kernel.org, linux-mm@...ck.org, x86@...nel.org,
        torvalds@...ux-foundation.org, akpm@...ux-foundation.org,
        luto@...nel.org, bp@...en8.de, dave.hansen@...ux.intel.com,
        hpa@...or.com, mingo@...hat.com, juri.lelli@...hat.com,
        willy@...radead.org, mgorman@...e.de, peterz@...radead.org,
        rostedt@...dmis.org, vincent.guittot@...aro.org, jon.grimm@....com,
        bharata@....com, boris.ostrovsky@...cle.com,
        konrad.wilk@...cle.com, ankur.a.arora@...cle.com
Subject: Re: [PATCH 8/9] irqentry: define irqentry_exit_allow_resched()

Thomas Gleixner <tglx@...utronix.de> writes:

> On Sun, Apr 02 2023 at 22:22, Ankur Arora wrote:
>> Allow threads marked TIF_ALLOW_RESCHED to be rescheduled in irqexit.
>>
>> This is only necessary under !preempt_model_preemptible() for which
>> we reuse the same logic as irqentry_exit_code_resched().
>
> This tells what this patch is doing but completely fails to explain why
> this is necessary and useful.

Thanks. Yeah, it does seem to miss that completely.

Needs some massaging, but does something like this clarify it's purpose?

On kernels with PREEMPTION_NONE/_VOLUNTARY, rescheduling of kernel
tasks happens when they allow it -- for instance by synchronously
calling cond_resched() in a long running task.

There are cases where it is not convenient to periodically call
cond_resched() -- for instance when executing a potentially
long running instruction (such as REP STOSB on x86).

To handle kernel code sections which can be safely preempted, but
cannot explicitly call cond_resched(), allow them to mark themselves
TIF_ALLOW_RESCHED.

Contexts marked such (via allow_resched()) can be rescheduled in the
irqexit path. This is, of course only needed with
!preempt_model_preemptible() and the rescheduling logic is functionally
same as irqentry_exit_code_resched().

--
ankur