linux-kernel - Re: [RFC PATCH 1/2] bdev: add support for CPU latency PM QoS tuning

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]

Message-ID: <e38630e41353d083f7c0f4d726218aa5f3b36827.camel@linux.intel.com>
Date: Fri, 30 Aug 2024 14:55:56 +0300
From: Tero Kristo <tero.kristo@...ux.intel.com>
To: Jens Axboe <axboe@...nel.dk>
Cc: linux-block@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [RFC PATCH 1/2] bdev: add support for CPU latency PM QoS tuning

On Thu, 2024-08-29 at 05:37 -0600, Jens Axboe wrote:
> On 8/29/24 1:18 AM, Tero Kristo wrote:
> > diff --git a/block/bio.c b/block/bio.c
> > index e9e809a63c59..6c46d75345d7 100644
> > --- a/block/bio.c
> > +++ b/block/bio.c
> > @@ -282,6 +282,8 @@ void bio_init(struct bio *bio, struct
> > block_device *bdev, struct bio_vec *table,
> >  	bio->bi_max_vecs = max_vecs;
> >  	bio->bi_io_vec = table;
> >  	bio->bi_pool = NULL;
> > +
> > +	bdev_update_cpu_latency_pm_qos(bio->bi_bdev);
> >  }
> >  EXPORT_SYMBOL(bio_init);
> 
> This is entirely the wrong place to do this, presumably it should be
> done at IO dispatch time, not when something initializes a bio.
> 
> And also feels like entirely the wrong way to go about this, adding
> overhead to potentially each IO dispatch, of which there can be
> millions
> per second.

Any thoughts where it could/should be added?

I moved the bdev_* callback from bio_init to the below location and it
seems to work also:

diff --git a/block/blk-mq.c b/block/blk-mq.c
index 3b4df8e5ac9e..d97a3a4252de 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -2706,6 +2706,7 @@ static void __blk_mq_flush_plug_list(struct
request_queue *q,
 {
        if (blk_queue_quiesced(q))
                return;
+       bdev_update_cpu_latency_pm_qos(q->disk->part0);
        q->mq_ops->queue_rqs(&plug->mq_list);
 }