linux-kernel - Re: [PATCH v2 07/35] sched: define *_tsk_need_resched

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: Thu, 30 May 2024 02:08:19 -0700
From: Ankur Arora <ankur.a.arora@...cle.com>
To: Peter Zijlstra <peterz@...radead.org>
Cc: Ankur Arora <ankur.a.arora@...cle.com>, linux-kernel@...r.kernel.org,
        tglx@...utronix.de, torvalds@...ux-foundation.org, paulmck@...nel.org,
        rostedt@...dmis.org, mark.rutland@....com, juri.lelli@...hat.com,
        joel@...lfernandes.org, raghavendra.kt@....com, sshegde@...ux.ibm.com,
        boris.ostrovsky@...cle.com, konrad.wilk@...cle.com,
        Ingo Molnar
 <mingo@...hat.com>,
        Vincent Guittot <vincent.guittot@...aro.org>
Subject: Re: [PATCH v2 07/35] sched: define *_tsk_need_resched_lazy() helpers


Peter Zijlstra <peterz@...radead.org> writes:

> On Mon, May 27, 2024 at 05:34:53PM -0700, Ankur Arora wrote:
>
>>  static inline void clear_tsk_need_resched(struct task_struct *tsk)
>>  {
>> -	clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
>> +	clear_tsk_thread_flag(tsk, tif_resched(RESCHED_NOW));
>> +
>> +	if (IS_ENABLED(CONFIG_PREEMPT_AUTO))
>> +		clear_tsk_thread_flag(tsk, tif_resched(RESCHED_LAZY));
>> +}
>
> (using tif_resched() here is really uncalled for)
>
> So this will generate rather sub-optimal code, namely 2 atomics that
> really should be one.
>
> Ideally we'd write this something like:
>
> 	unsigned long mask = _TIF_NEED_RESCHED;
> 	if (IS_ENABLED(CONFIG_PREEMPT_AUTO))
> 		mask |= _TIF_NEED_RESCHED_LAZY;
>
> 	atomic_long_andnot(mask, (atomic_long_t *)task_thread_info(tsk)->flags);
>
> Which will clear both bits with a single atomic.

Much better. Will fix.

--
ankur