ACK/Cmnt: [SRU][B/D/OEM-B/OEM-OSP1-B][PATCH 0/1] x86/timer: Skip PIT initialization on modern chipsets

[Stefan Bader]

On 07.11.19 09:05, You-Sheng Yang wrote:
> From: You-Sheng Yang <vicamo at gmail.com>
> 
> BugLink: https://bugs.launchpad.net/bugs/1851216
> 
> [Impact]
> Some recent platform seem to have an unstable HPET, which is used to
> calibrate TSC but turns out to mark an actual stable TSC clocksource as
> unstable, and then fallback to that unstable HPET as main clocksource. A
> previous fix was proposed in bug 1840239, which is then observed to hang
> platforms with modern chipsets including Skylake and ApolloLake at early
> boot stage when HPET is disabled.
> 
> [Fix]
> Commit c8c4076723da "x86/timer: Skip PIT initialization on modern chipsets"
> included in v5.3-rc1 to be backported.
> 
> [Test Case]
> Boot with kernel built with this change, and see if it will hang right
> after grub loaded initramfs image.
> 
> [Regression Risk]
> Low. This change detects the hardware that requires explicit PIT
> initialization skip and skip it. For those unaffected systems, they
> remain unaffected.
> 
> Thomas Gleixner (1):
>   x86/timer: Skip PIT initialization on modern chipsets
> 
>  arch/x86/include/asm/apic.h    |  2 ++
>  arch/x86/include/asm/time.h    |  1 +
>  arch/x86/kernel/apic/apic.c    | 27 +++++++++++++++++++++++++++
>  arch/x86/kernel/apic/io_apic.c |  4 ++++
>  arch/x86/kernel/i8253.c        | 25 ++++++++++++++++++++++++-
>  arch/x86/kernel/time.c         |  7 +++++--
>  6 files changed, 63 insertions(+), 3 deletions(-)
> 

You should have pointed out much more visibly that this is a critical fix to a
previously submitted SRU request:

[SRU] ... Disable unreliable HPET on CFL/ICL systems

Even more importantly as thi missed the deadline for the cycle. Not really keen
on adding even more delta but hoping this time at least regression testing was
done more carefully.

Acked-by: Stefan Bader <stefan.bader at canonical.com>