linux-kernel - Re: [PATCH] block/blk-mq: Don't complete locally if capacities are different

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: Wed, 24 Jan 2024 00:24:36 +0000
From: Qais Yousef <qyousef@...alina.io>
To: Jens Axboe <axboe@...nel.dk>
Cc: Christoph Hellwig <hch@...radead.org>, Ingo Molnar <mingo@...nel.org>,
	Peter Zijlstra <peterz@...radead.org>,
	Vincent Guittot <vincent.guittot@...aro.org>,
	Dietmar Eggemann <dietmar.eggemann@....com>,
	linux-kernel@...r.kernel.org, linux-block@...r.kernel.org,
	Sudeep Holla <sudeep.holla@....com>, Wei Wang <wvw@...gle.com>,
	Jaegeuk Kim <jaegeuk@...nel.org>,
	Bart Van Assche <bvanassche@....org>
Subject: Re: [PATCH] block/blk-mq: Don't complete locally if capacities are
 different

On 01/23/24 08:58, Jens Axboe wrote:
> On 1/23/24 1:47 AM, Christoph Hellwig wrote:
> > On Mon, Jan 22, 2024 at 10:42:20PM +0000, Qais Yousef wrote:
> >> The logic in blk_mq_complete_need_ipi() assumes SMP systems where all
> >> CPUs have equal capacities
> > 
> > What is a capacity here?
> 
> It seems to be the chosen word to describe the performance potential of
> the core in question, we use it elsewhere in the kernel. But yes, could
> do with a bit more of an explanation.

Is referring to it as compute capacity makes it clearer? Sorry I thought that's
a common term.

> 
> >> +	return arch_scale_cpu_capacity(this_cpu) >= arch_scale_cpu_capacity(that_cpu);
> > 
> > oerly long line here.

This is consistent with similar long lines in the same file and it's more
readable as one line. checkpatch doesn't complain about this being long;
I think they look for 100 or 120 now. This is 86.

> > 
> > Also pleas split patches for different subsystems.
> 
> Yes please, the sched/topology thing should be a separate prep patch.

Okay. I thought the norm to keep such small patches self contained as adding
dead code followed by a one liner isn't always seen as better. But a split it
is :)

I'll give the sched/arm folks time to have a look before I post a new version.


Thanks!

--
Qais Yousef