Increase amd-pmc timeout

[Mario Limonciello]

[Impact]

  amd-pmc times out communicating with the SMU over stress testing cycles like:
   > amd_pmc AMDI0005:00: SMU response timed out
   > amd_pmc AMDI0005:00: failed to talk to SMU
   > amd_pmc AMDI0005:00: suspend failed
   > PM: dpm_run_callback(): acpi_subsys_suspend_noirq+0x0/0x50 returns -110

[Fix]

  Increase the timeout to 2000ms that matches to GPU driver.

[Test]

  This is requested by AMD.

[Where problems could occur]

  Low risk. This only affects small number of AMD systems that would excess the original timeout with low fail rate 0.9%-1.7%. In worst case scenario it would only takes time up to 2 seconds.

==== original descriptions ==

On some customer designs there have been reports that amd-pmc times out communicating with the SMU over stress testing cycles.
When this happens the symptoms look something like this:

> amd_pmc AMDI0005:00: SMU response timed out
> amd_pmc AMDI0005:00: failed to talk to SMU
> amd_pmc AMDI0005:00: suspend failed
> PM: dpm_run_callback(): acpi_subsys_suspend_noirq+0x0/0x50 returns -110

Due to this behavior we have submitted a patch to extend the timeout which has shown to help the customer designs with problems.
https://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86.git/commit/?h=fixes&id=3c3c8e88c8712bfe06cd10d7ca77a94a33610cd6

This patch is destined for 5.15, but also CC'ed to stable. Can you please include in OEM 5.14 kernel?

[Mario Limonciello (superm1)]

FYI - this patch is at least "part" of the solution needed for https://bugzilla.kernel.org/show_bug.cgi?id=214091.

[Mario Limonciello (superm1)]

It's landed upstream as 3c3c8e88c8712bfe06cd10d7ca77a94a33610cd6

[Alex Hung (alexhung)]

https://lists.ubuntu.com/archives/kernel-team/2021-September/124287.html

[Timo Aaltonen (tjaalton)]

and applied via https://bugs.launchpad.net/ubuntu/focal/+source/linux-oem-5.14/+bug/1945656