On 03/16/23 08:43, Greg KH wrote:
On Wed, Mar 15, 2023 at 12:53:04PM +0000, Qais Yousef wrote:
On 03/15/23 09:53, Greg KH wrote:
On Wed, Mar 15, 2023 at 09:03:45AM +0100, Greg KH wrote:
On Wed, Mar 08, 2023 at 04:22:00PM +0000, Qais Yousef wrote:
Portion of the fixes were ported in 5.15 but missed some.
This ports the remainder of the fixes.
Based on 5.15.98.
Build tested on x86 with and without uclamp config enabled.
Now queued up, thanks.
Wait, should we also queue up a2e90611b9f4 ("sched/fair: Remove capacity inversion detection") to 5.10 and 5.15 now as well?
It has a dependency on e5ed0550c04c ("sched/fair: unlink misfit task from cpu overutilized") which is nice to have but not strictly required. It improves the search for best CPU under adverse thermal pressure to try harder. And the new search effectively replaces the capacity inversion detection, so it is removed afterwards.
I'd like to have it but find_energy_efficient_cpu() looks a bit different for a straightforward backport and opted to avoid the risk.
Happy to reconsider though.
Ok, just wanting to verify as that commit looked like it was needed after these.
But, as I've just now dropped both 5.10.y and 5.15.y sets of patches, this is going to need to be redone anyway, so perhaps put that note in the 00/XX email when you resubmit so that I don't get confused again.
Will do. There were no failure for the 5.15 set though. The reports were against 5.10 only.
I retested 5.15 series against:
1. default ubuntu config which has uclamp + smp 2. default ubuntu config without uclamp + smp 3. default ubunto config without smp (which automatically disabled uclamp) 4. reported riscv-allnoconfig, mips-randconfig, x86_64-randocnfigs
And they all compiled without an error. Happy to retake them or prefer a resend anyway though nothing has changed?
Just finishing testing 5.10 and will send v2 with the updated cover-letter.
Thanks!
-- Qais Yousef