arm/arm64: Unify PSCI client support

Quoted 39 lines · Will Deacon

Quoted 18 lines · Mark Rutland

This series unifies the 32-bit and 64-bit PSCI client code, moving the bulk of the FW invocation and probing out to a common location in drivers/firmware. The bulk of the PSCI 0.2 cleanups have hit mainline now, so this is just the unification portion.

This results in a reasonable saving in terms of lines of code, and will allow for PSCI 1.0 support to be unified form the beginning, avoiding further duplication.

Since v4 [1]:
* Apply Rob Herring's ack
* Rebase to v4.2-rc2 to handle a trivial conflict
* Reorder the series to keep arch/arm patches together Since v3 [2]:
* Drop the PSCI 0.2 patches as they're in mainline
* s/__pa/virt_to_idmap/ from Grygorii Strashko
* Use macros for Calxeda CPU_SUSPEND parameters

Quoted 6 lines

Russell, are you happy with the arch/arm patches? If so, are you happy for them to go via another tree, or would you prefer that I set up a stable branch for merging?

I was under the impression that you had already taken Grygorii's patch but I couldn't spot it in any branches. I can drop that if you already have it.

I've been looking at merging this, but it's a tad fiddly touching all of arm, arm64 and drivers. One way to do it would be:

(1) I create a branch containing patches 1,2 and 5 based on -rc2 and merge that into the arm64/for-next/core branch. There's a minor conflict, but it's easy to resolve.

(2) I create another branch, which is just the branch merged in (1) + patches 3 and 4 on top. I send a pull request for that to rmk.

(3) Torvalds will get the minor conflict resolved in (1) when he merges the arm and arm64 trees.

Quoted 4 lines

If we're not comfortable with (3), then the whole lot could go via arm-soc instead (including the conflict resolution).

Russell, Olof, any preferences?

Wouldn't it be easiest to just get acks on patch 3 and 4 and just merge it all through your tree?