Closed edwakuwaku closed 1 month ago
Hi @ewoodev @kishore-sn @ritesh55555 , let's discuss about the PM + SMP sleep/wakeup operation in this PR.
Hello @kishore-sn @ewoodev @ritesh55555 , I have pushed the latest implementation. Instead of pause/resume, I changed to use gating as I think this method is not affecting the task/scheduler block. When I tested with pause/resume, there is some deadlock happening after secondary core is paused. Thus, I think there are still some misunderstanding from our side regarding the usage of cpu pause/resume. I tested this changes with both cores in idle thread, it works fine as previously. Please help to test this changes on the condition of "cpu0: idle thread, cpu1: not idle thread". The sleep should be aborted under this condition.
@ritesh55555 @kishore-sn , Could you please test with this PR on internal branch and leave the test results here? Thanks.
@edwakuwaku Please check merge conflicts
@ritesh55555 @kishore-sn , Could you please test with this PR on internal branch and leave the test results here? Thanks.
hi @edwakuwaku , Can you please rebase the PR, when i merge and test i am getting stuck just after booting several times. May be if you merge yourself and update, it would be better to test?
I have rebased to latest master.
This PR is working fine with product code. There is no more hang issue during sleep and wakeup. It is working fine
Can we keep SMP Hotplug
logic outside pm_idle
by creating new APIs ?
Can we keep
SMP Hotplug
logic outsidepm_idle
by creating new APIs ?
I have talked with @ritesh55555 about this, after this PR is merged, cleanup can be done from Samsung team on PM + SMP which you think it is necessary.