Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20230113210703.62107-1-nhuck@google.com> <Y8HI+42TxxlJxT6D@slm.duckdns.org>
 <CAJkfWY7vmvrU8sW3OWpSa9zygY=6e8BTTkktPe-VScdOcLL-sw@mail.gmail.com>
 <Y8gxu7l8BUWewuMg@slm.duckdns.org> <CAB=BE-Q9jtJnqPwGzSTQ6-soZ9STvqAebeONy=Eyo08H+eg-rQ@mail.gmail.com>
In-Reply-To: <CAB=BE-Q9jtJnqPwGzSTQ6-soZ9STvqAebeONy=Eyo08H+eg-rQ@mail.gmail.com>
From:   Nathan Huckleberry <nhuck@google.com>
Date:   Wed, 18 Jan 2023 14:04:34 -0800
Message-ID: <CAJkfWY5c=kT1CpQ+R34MpX6_Dyq_32zgkYDgrWU7Kuoz7d9TYg@mail.gmail.com>
Subject: Re: [PATCH] workqueue: Add WQ_SCHED_FIFO
To:     Sandeep Dhavale <dhavale@google.com>
Cc:     Tejun Heo <tj@kernel.org>, Daeho Jeong <daehojeong@google.com>,
        Eric Biggers <ebiggers@kernel.org>,
        Sami Tolvanen <samitolvanen@google.com>,
        Lai Jiangshan <jiangshanlai@gmail.com>,
        Jonathan Corbet <corbet@lwn.net>, linux-doc@vger.kernel.org,
        linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Wed, Jan 18, 2023 at 10:22 AM Sandeep Dhavale <dhavale@google.com> wrote:
>
> On Wed, Jan 18, 2023 at 9:52 AM Tejun Heo <tj@kernel.org> wrote:
> >
> > On Sat, Jan 14, 2023 at 01:00:00PM -0800, Nathan Huckleberry wrote:
> > > This sounds fine to me. How do you feel about a config flag to change
> > > the default WQ_HIGHPRI scheduler policy and a sysfs node to update the
> > > policy per workqueue?
> >
> > Yeah, sounds fine to me.
> >
> > Thanks.
> >
> Hi Tejun,
> If with the kernel config option, every WQ_HIGHPRI is elevated to
> sched_fifo_low, wouldn't that be kind of defeating the purpose? Having
> another class for even more urgent work is better in my opinion.

I agree, however most of the users of WQ_HIGHPRI that are relevant to
Android would probably have SCHED_FIFO enabled anyway.

I can write the patches such that WQ_HIGHPRI selects one of two
internal WQ flags. If using SCHED_FIFO for all workqueues is
problematic, Android can consider using the internal WQ flags directly
and carry those one-line patches in the Android tree.

Thanks,
Huck

>
> Thanks,
> Sandeep.
> > --
> > tejun