ACK/Cmnt: [SRU][PULL][Jammy] Add support for intel EMR cpu

Cory Todd cory.todd at canonical.com
Tue Apr 11 15:57:06 UTC 2023


On Tue, Apr 11, 2023 at 05:08:57PM +0200, Roxana Nicolescu wrote:
> BugLink: https://bugs.launchpad.net/bugs/2015372
> 
> SRU Justification
> 
> [Impact]
> 
> Intel has introduced support for their new Emerald Rapids CPU.
> This support is a nice feature to have in Jammy because it is an LTS
> release and enabling extra hardware adds to its usability.
> 
> Usually backporting new features to existing kernels it's not a
> common practice. But these commits were easy to backport. Most of
> them were clean cherry picks. Some of them were already picked up
> from upstream stable releases, and some required manual intervention
> because Jammy did not have some commits applied but the actual
> changes were very minimal.
> 
> More information about each commit is addressed inline.
> 
> [Commits]
> 
> 1. 9c252ecf3036:
> "platform/x86: intel-uncore-freq: add Emerald Rapids support"
> Adjusted the path to uncore-frequency.c
> In ce2645c458b5c83b0872ea9e39d2c3293445353a commit, this was moved to
> uncore-frequency dir
> 
> 2. 5a8a05f165fb18d37526062419774d9088c2a9b9
> "perf/x86/intel/cstate: Add Emerald Rapids"
> Small context adjustment in intel_cstates_match array
> because of missing SPR, RPL and MDL cpu models introduced in
> 528c9f1daf20d
> 
> 3. 57512b57dcfaf63c52d8ad2fb35321328cde31b0
> "perf/x86/rapl: Add support for Intel Emerald Rapids"
This commit is missing a bug link.

6fcc137ed272 ("perf/x86/rapl: Add support for Intel Emerald Rapids")

> 
> Context adjustment due to missing
> - 80275ca9e525c "perf/x86/rapl: Use standard Energy Unit for SPR Dram RAPL domain"
> - 1ab28f17eeeec "perf/x86/rapl: Add support for Intel AlderLake-N"
> - eff98a7421b3e "perf/x86/rapl: Add support for Intel Raptor Lake"
> - f52853a668bfe "perf/x86/rapl: Add support for Intel Meteor Lake"
> 
> 4. 74528edfbc664f9d2c927c4e5a44f1285598ed0f
> clean cherry-pick
> 
> 5. 61f9fdcdcd01f9a996b6db4e7092fcdfe8414ad5
> clean cherry-pick
> 
> 6. 93cac4150727dae0ee89f501dd75413b88eedec0
> clean cherry-pick
> 
> 7. 7adc6885259edd4ef5c9a7a62fd4270cf38fdbfb
> clean cherry-pick
> 
> 8. e4b2bc6616e21
> clean cherry-pick
> 
> [Testing]
> Kernel was built on cbd and boot tested on a VM.
> 
> [Regression potential]
> Very low, these are minimal new features, which actually reuse a lot
> of the existing cpu structures/functions.
> 
> ------------------------------------------------------------------------
> 
> The following changes since commit 8feeaa9039a290b978e7855ab82955cd9348fc13:
> 
> 	"platform/x86: intel-uncore-freq: add Emerald Rapids support"
> 
> are available in the Git repository at:
> 
> 	git://git.launchpad.net/~roxanan/ubuntu/+source/linux/+git/jammy lp2015372
> 
> for you to fetch changes up to 7d88565e6777082265b373733fd93eb3e0914e12:
> 
> 	"EDAC/i10nm: Add Intel Emerald Rapids server support"
> 
> ------------------------------------------------------------------------
> 
> Artem Bityutskiy (2):
>   platform/x86: intel-uncore-freq: add Emerald Rapids support
>   intel_idle: add Emerald Rapids Xeon support
> 
> Kan Liang (1):
>   perf/x86/intel/cstate: Add Emerald Rapids
> 
> Qiuxu Zhuo (1):
>   EDAC/i10nm: Add Intel Emerald Rapids server support
> 
> Zhang Rui (4):
>   perf/x86/rapl: Add support for Intel Emerald Rapids
>   tools/power/x86/intel-speed-select: Add Emerald Rapid quirk
>   tools/power turbostat: Introduce support for EMR
>   powercap: intel_rapl: add support for Emerald Rapids
> 
>  arch/x86/events/intel/cstate.c                   | 1 +
>  arch/x86/events/rapl.c                           | 1 +
>  drivers/edac/i10nm_base.c                        | 1 +
>  drivers/idle/intel_idle.c                        | 2 ++
>  drivers/platform/x86/intel/uncore-frequency.c    | 1 +
>  drivers/powercap/intel_rapl_common.c             | 1 +
>  tools/power/x86/intel-speed-select/isst-config.c | 2 +-
>  tools/power/x86/turbostat/turbostat.c            | 3 +++
>  8 files changed, 11 insertions(+), 1 deletion(-)
> 
All three backports make sense to me and the remaining cherry picks look
clean.

Acked-by: Cory Todd <cory.todd at canonical.com>




More information about the kernel-team mailing list