linux-kernel - Re: [RESEND PATCH v6] blk-throtl: Introduce sync and async queues for blk-throtl

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

Message-ID: <20230306170634.os744umla72yoryw@blackpad>
Date:   Mon, 6 Mar 2023 18:06:34 +0100
From:   Michal Koutný <mkoutny@...e.com>
To:     Jinke Han <hanjinke.666@...edance.com>
Cc:     tj@...nel.org, josef@...icpanda.com, axboe@...nel.dk,
        cgroups@...r.kernel.org, linux-block@...r.kernel.org,
        linux-kernel@...r.kernel.org, yinxin.x@...edance.com
Subject: Re: [RESEND PATCH v6] blk-throtl: Introduce sync and async queues
 for blk-throtl

Hello Jinke.

On Tue, Feb 28, 2023 at 04:59:35PM +0800, Jinke Han <hanjinke.666@...edance.com> wrote:
> and gives a huge priority to sync write ios. Sync queue only make sense
> for write ios as we treat all read io as sync io.
[...]
> +/* For write ios, dispatch 4 sync ios and 1 normal io in one loop */
> +#define THROTL_SYNC_FACTOR 4

This looks like an unexplained number wired in.
E.g. I'd expect that 1st iteration would have this factor==1, i.e. RR
between sync and async queue.

What is the reason for this value? E.g. I assume it'd depend on number
of synchronous writers and flusher threads and their activity, i.e.
could you describe why this value caters the best for wide range of
various sync and async situations? Or add a comment what is the
particular value based on (so that it can be reassessed should those
dependencies change in the future).

Thanks,
Michal

Download attachment "signature.asc" of type "application/pgp-signature" (229 bytes)