APPLIED[K]: [PATCH 0/1][SRU][OEM-5.17/Unstable] Support AMD P-State cpufreq control mechanism
Andrea Righi
andrea.righi at canonical.com
Tue Jul 19 05:27:26 UTC 2022
On Fri, Apr 08, 2022 at 10:39:22AM +0800, You-Sheng Yang wrote:
> BugLink: https://bugs.launchpad.net/bugs/1956509
>
> [Impact]
>
> AMD P-State CPU performance scaling driver not yet supported on AMD Zen
> based CPU series.
>
> [Fix]
>
> Mainline commits starting at commit d341db8f48ea ("x86/cpufeatures: Add
> AMD Collaborative Processor Performance Control feature flag") from v5.17.
>
> Even this committed, the defconfig will build it as module, and that
> prevents AMD P-State driver from actually taking effect. Therefore an
> additional fix to the kernel configs is necessary.
>
> [Test Case]
>
> $ sudo cpupower frequency-info
> analyzing CPU 0:
> driver: amd-pstate
> CPUs which run at the same hardware frequency: 0
> CPUs which need to have their frequency coordinated by software: 0
> maximum transition latency: 131 us
> hardware limits: 400 MHz - 3.10 GHz
> available cpufreq governors: conservative ondemand userspace powersave
> performance schedutil
> current policy: frequency should be within 1.60 GHz and 3.10 GHz.
> The governor "ondemand" may decide which speed to use
> within this range.
> current CPU frequency: Unable to call hardware
> current CPU frequency: 1.59 GHz (asserted by call to kernel)
> boost state support:
> Supported: yes
> Active: yes
> Boost States: 0
> Total States: 3
> Pstate-P0: 3100MHz
> Pstate-P1: 1800MHz
> Pstate-P2: 1600MHz
>
> [Where problems could occur]
>
> The CPPC feature is implemented in two ways depending on underlying HW
> details for the APU/CPU:
> * Dedicated MSR
> * Shared memory
>
> The metrics used to measure power/performance efficiency don't show any
> improvement on the designs that use shared memory. They do show
> improvements in the MSR designs.
>
> So the code will only enable it by default on the MSR implementation.
>
> [Other Info]
>
> Althought the v5.17 kernel and newer should have already this committed,
> an additional fix to the kernel configs is necessary, and
> unstable/oem-5.17 are also nominated (in a separate thread for they don't
> need other parts but the config).
Applied to kinetic/linux, thanks!
-Andrea
More information about the kernel-team
mailing list