Comment 9 for bug 1856387

Revision history for this message
Anthony Buckley (tony-buckley) wrote :

Hello,
I have completed the bisect as requested and identified the problem commit. The bisect message is as follows:-

git bisect good
f723dd269d0740e09af47bb5590ffc4f61766153 is the first bad commit
commit f723dd269d0740e09af47bb5590ffc4f61766153
Author: Thomas Gleixner <email address hidden>
Date: Thu Nov 7 09:05:00 2019 +0100

    x86/timer: Skip PIT initialization on modern chipsets

    BugLink: https://bugs.launchpad.net/bugs/1851216

    Recent Intel chipsets including Skylake and ApolloLake have a special
    ITSSPRC register which allows the 8254 PIT to be gated. When gated, the
    8254 registers can still be programmed as normal, but there are no IRQ0
    timer interrupts.

    Some products such as the Connex L1430 and exone go Rugged E11 use this
    register to ship with the PIT gated by default. This causes Linux to fail
    to boot:

      Kernel panic - not syncing: IO-APIC + timer doesn't work! Boot with
      apic=debug and send a report.

    The panic happens before the framebuffer is initialized, so to the user, it
    appears as an early boot hang on a black screen.

    Affected products typically have a BIOS option that can be used to enable
    the 8254 and make Linux work (Chipset -> South Cluster Configuration ->
    Miscellaneous Configuration -> 8254 Clock Gating), however it would be best
    to make Linux support the no-8254 case.

    Modern sytems allow to discover the TSC and local APIC timer frequencies,
    so the calibration against the PIT is not required. These systems have
    always running timers and the local APIC timer works also in deep power
    states.

    So the setup of the PIT including the IO-APIC timer interrupt delivery
    checks are a pointless exercise.

    Skip the PIT setup and the IO-APIC timer interrupt checks on these systems,
    which avoids the panic caused by non ticking PITs and also speeds up the
    boot process.

    Thanks to Daniel for providing the changelog, initial analysis of the
    problem and testing against a variety of machines.

    Reported-by: Daniel Drake <email address hidden>
    Signed-off-by: Thomas Gleixner <email address hidden>
    Tested-by: Daniel Drake <email address hidden>
    Cc: <email address hidden>
    Cc: <email address hidden>
    Cc: <email address hidden>
    Cc: <email address hidden>
    Cc: <email address hidden>
    Link: https://<email address hidden>

    (backported from commit c8c4076723daca08bf35ccd68f22ea1c6219e207)
    Signed-off-by: You-Sheng Yang <email address hidden>
    Acked-by: Stefan Bader <email address hidden>
    Acked-by: Connor Kuehl <email address hidden>
    Signed-off-by: Stefan Bader <email address hidden>

:040000 040000 9c51f067713006f928684555c3254e89bdc10361 ad4d7a34eed39a733c78e630f4d9125f67e001bb M arch

Regards