Skip to content

Conversation

@tmleman
Copy link
Contributor

@tmleman tmleman commented Mar 8, 2023

This patch sets lowest clock as default for Meteorlake platform. In current version FW can do unnecessary clocks switching operations. FW should start from low value and scale-up if necessary.

This patch sets lowest clock as default for Meteorlake platform. In
current version FW can do unnecessary clocks switching operations. FW
should start from low value and scale-up if necessary.

Signed-off-by: Tomasz Leman <tomasz.m.leman@intel.com>
@kv2019i
Copy link
Collaborator

kv2019i commented Mar 8, 2023

We can't merge this until the KCPS_REGISTER stuff is merged to kernel and topologies and we have sane values. @jsarha @ujfalusi are we ready?

E.g. thesofproject/linux#4213 and friends.

@kv2019i kv2019i requested review from jsarha and ujfalusi March 8, 2023 13:41
@jsarha
Copy link
Contributor

jsarha commented Mar 9, 2023

We can't merge this until the KCPS_REGISTER stuff is merged to kernel and topologies and we have sane values. @jsarha @ujfalusi are we ready?

E.g. thesofproject/linux#4213 and friends.

#7160 is merged, but it currently only has the 0 kcps defined as the default value for all pipelines.

@abonislawski
Copy link
Member

This PR does not need any KCPS stuff. Our default clock is the lowest one, thats why we have to force HP clock in init (in current code).

#7160 is merged, but it currently only has the 0 kcps defined as the default value for all pipelines.

And we should keep it at 0. This is calculated by FW based on manifest data (currently only on MTL-004 branch).
Driver should use kcps_register to adjust this if necessary.

Copy link
Collaborator

@kv2019i kv2019i left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@abonislawski wrote:

And we should keep it at 0. This is calculated by FW based on manifest data (currently only on MTL-004 branch).

Right, so it's 0 in tplg and fw doesn't calculate so this patch will break MTL in main as DSP clock will be kept at lowest level? -1 until this is clarified. Shouldn't the calculation basd on FW data (or higher default in tplg) be merged first, before we do this PR.

Driver should use kcps_register to adjust this if necessary.

Hmm, how will topology authors know when it is necessary? This seems very fuzzy to me and we need to have this documented somewhere else then review comments..

Copy link
Member

@abonislawski abonislawski left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

yeap, actually we cant merge it now because temporary HP switch in init will not update clk info so its better to keep it as is for now

@lgirdwood lgirdwood added this to the v2.6 milestone Mar 17, 2023
@lgirdwood
Copy link
Member

Assigning milestone so this can be tracked and merged once dependencies are all in place.

@tmleman
Copy link
Contributor Author

tmleman commented Mar 29, 2023

Since it will not gonna be merge in the near future I'm closing this PR.

@tmleman tmleman closed this Mar 29, 2023
@tmleman tmleman deleted the topic/upstream/platform/mtl/clock branch May 23, 2024 11:18
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

7 participants