linux-kernel - Re: [PATCH] firmware: arm_scmi: fix timeout value for send

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Thu, 11 Jun 2020 10:19:12 -0500
From:   Jassi Brar <jassisinghbrar@...il.com>
To:     Sudeep Holla <sudeep.holla@....com>
Cc:     linux-arm-kernel <linux-arm-kernel@...ts.infradead.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Viresh Kumar <viresh.kumar@...aro.org>,
        Rob Herring <robh@...nel.org>,
        Frank Rowand <frowand.list@...il.com>,
        Bjorn Andersson <bjorn.andersson@...aro.org>,
        Vincent Guittot <vincent.guittot@...aro.org>,
        "arnd@...db.de" <arnd@...db.de>,
        Jassi Brar <jaswinder.singh@...aro.org>
Subject: Re: [PATCH] firmware: arm_scmi: fix timeout value for send_message

On Thu, Jun 11, 2020 at 3:40 AM Sudeep Holla <sudeep.holla@....com> wrote:

> >
> > > > > >       if (xfer->hdr.poll_completion) {
> > > > > > -             ktime_t stop = ktime_add_ns(ktime_get(), SCMI_MAX_POLL_TO_NS);
> > > > > > +             ktime_t stop = ktime_add_ns(ktime_get(), 500 * 1000 * NSEC_PER_USEC);
> > > > > >
> > > > >
> > > > > This is unacceptable delay for schedutil fast_switch. So no for this one.
> > > > >
> > > > Increasing timeout does not increase latency.
> > >
> > > Agreed, but worst case you may be stuck here for 500ms which is not
> > > acceptable.
> > >
> > Not acceptable to who, you or the kernel? :)    Now that you said you
> > are fixing the scmi's fast_switch implementation.
> >
> Sorry, I meant to disable it for single channel implementation.
>
The single-channel platform may have only cpufreq as the user, or only
users that respond quickly ~3us. So maybe we leave the decision, to
enable fast_switch, totally to the platform. But of course, this
discussion is for another thread.


> > Even though I don't think 500ms would ruin our lives, but ok, I will
> > make it 30ms - same as you did in the 'else' block. And drop the other
> > change.
>
> I am fine if cpufreq maintainers allow that in the fast switch path that
> happens in the fast path.
>
Thanks, let me respin the patch and include some cpufreq folks.

Cheers!