Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 8 Sep 2021 12:10:56 -0600
From:      Warner Losh <imp@bsdimp.com>
To:        Colin Percival <cperciva@freebsd.org>
Cc:        John Baldwin <jhb@freebsd.org>, src-committers <src-committers@freebsd.org>, "<dev-commits-src-all@freebsd.org>" <dev-commits-src-all@freebsd.org>, dev-commits-src-main@freebsd.org
Subject:   Re: git: a8b89dff6ac0 - main - Disable acpi_timer_test by default
Message-ID:  <CANCZdfpMmShmTz51bG-fH-LnhnXnkQzcEOwHJCjWWwb6ZfNF7g@mail.gmail.com>
In-Reply-To: <4b7bf983-8333-f4bc-6ce8-a59ac2fc7380@freebsd.org>
References:  <202109080132.1881WXPv069848@gitrepo.freebsd.org> <f1030cac-a17b-727a-472e-d3d90933da2b@FreeBSD.org> <4b7bf983-8333-f4bc-6ce8-a59ac2fc7380@freebsd.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Sep 8, 2021, 11:47 AM Colin Percival <cperciva@freebsd.org> wrote:

> On 9/8/21 9:01 AM, John Baldwin wrote:
> > On 9/7/21 6:32 PM, Colin Percival wrote:
> >>      Disable acpi_timer_test by default
> >>           This disables testing the ACPI timer by default, forcing the
> use of
> >>      ACPI-fast rather than ACPI-safe.  The broken-ACPI-timers workaround
> >>      can be re-enabled by setting the hw.acpi.timer_test_enabled=1
> tunable.
> >>           This speeds up the FreeBSD boot process by 140 ms on an EC2
> c5.xlarge
> >>      instance.
> >>           This change will not be MFCed.
> >>           Assuming no problems are reported, acpi_timer_test, the
> associated
> >>      tunable, and the ACPI-safe timecounter should be removed in
> FreeBSD 15.
> >>           Relnotes:       The ACPI-safe timer is disabled in favour of
> >> ACPI-fast;
> >>                      if timekeeping issues are observed, please test
> with
> >>                      hw.acpi.timer_test_enabled=1 in loader.conf and
> report
> >>                      if that fixes the problem.
> >
> > Perhaps it should default to '1' for i386 and '0' otherwise?  The
> relevant
> > chipsets were 32-bit only, so this would be a simple way to skip the
> test for
> > modern hardware, and you could probably MFC that safely.
>
> That option was discussed, but I decided that it was probably safer to
> keep it enabled by default in 13 in case the test was detecting systems
> which are broken in other ways.
>
> Googling for "ACPI-safe" (which shows up if the test fails) finds forum
> discussions from the mid-2010s, but it's not clear whether that's due to
> very old hardware, new ACPI timer issues, or other timekeeping problems
> -- I figured it was best to play it safe for something which would be
> going into a stable branch though.
>

I think John's suggestion is a good balance. It will also tell us if there
is a need on amd64. Having it tagged as i386 also will highlight it better
should the time come to remove it later.

Warner


-- 
> Colin Percival
> Security Officer Emeritus, FreeBSD | The power to serve
> Founder, Tarsnap | www.tarsnap.com | Online backups for the truly paranoid
>



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CANCZdfpMmShmTz51bG-fH-LnhnXnkQzcEOwHJCjWWwb6ZfNF7g>