| Message ID | 20240429140531.210576-1-ulf.hansson@linaro.org (mailing list archive) |
|---|---|
| Headers | show |
| Series | pmdomain/cpuidle-psci: Support s2idle/s2ram on PREEMPT_RT | expand |
On 2024-04-29 16:05:25 [+0200], Ulf Hansson wrote: > The hierarchical PM domain topology and the corresponding domain-idle-states > are currently disabled on a PREEMPT_RT based configuration. The main reason is > because spinlocks are turned into sleepable locks on PREEMPT_RT, which means > genpd and runtime PM can't be use in the atomic idle-path when > selecting/entering an idle-state. > > For s2idle/s2ram this is an unnecessary limitation that this series intends to > address. Note that, the support for cpuhotplug is left to future improvements. > More information about this are available in the commit messages. > > I have tested this on a Dragonboard 410c. Have you tested this with PREEMPT_RT enabled and if so, which kernel? > Kind regards > Ulf Hansson Sebastian
On Tue, 30 Apr 2024 at 11:44, Sebastian Andrzej Siewior <bigeasy@linutronix.de> wrote: > > On 2024-04-29 16:05:25 [+0200], Ulf Hansson wrote: > > The hierarchical PM domain topology and the corresponding domain-idle-states > > are currently disabled on a PREEMPT_RT based configuration. The main reason is > > because spinlocks are turned into sleepable locks on PREEMPT_RT, which means > > genpd and runtime PM can't be use in the atomic idle-path when > > selecting/entering an idle-state. > > > > For s2idle/s2ram this is an unnecessary limitation that this series intends to > > address. Note that, the support for cpuhotplug is left to future improvements. > > More information about this are available in the commit messages. > > > > I have tested this on a Dragonboard 410c. > > Have you tested this with PREEMPT_RT enabled and if so, which kernel? Yes, of course. :-) I should have mentioned this in the cover-letter, sorry. I have used the linux-rt-devel.git, which had a branch based upon v6.8-rc7 a while ago, that I used when I did my tests. The series needed a small rebase on top of my linux-pm tree [1], before I could post it though. I also tested the rebased series, but then of course then not with PREEMPT_RT, but to make sure there are no regressions. Kind regards Uffe [1] https://git.kernel.org/pub/scm/linux/kernel/git/ulfh/linux-pm.git next
The hierarchical PM domain topology and the corresponding domain-idle-states are currently disabled on a PREEMPT_RT based configuration. The main reason is because spinlocks are turned into sleepable locks on PREEMPT_RT, which means genpd and runtime PM can't be use in the atomic idle-path when selecting/entering an idle-state. For s2idle/s2ram this is an unnecessary limitation that this series intends to address. Note that, the support for cpuhotplug is left to future improvements. More information about this are available in the commit messages. I have tested this on a Dragonboard 410c. Kind regards Ulf Hansson Ulf Hansson (6): pmdomain: core: Enable s2idle for CPU PM domains on PREEMPT_RT pmdomain: core: Don't hold the genpd-lock when calling dev_pm_domain_set() cpuidle: psci-domain: Enable system-wide suspend on PREEMPT_RT cpuidle: psci: Drop redundant assignment of CPUIDLE_FLAG_RCU_IDLE cpuidle: psci: Enable the hierarchical topology for s2ram on PREEMPT_RT cpuidle: psci: Enable the hierarchical topology for s2idle on PREEMPT_RT drivers/cpuidle/cpuidle-psci-domain.c | 10 ++++-- drivers/cpuidle/cpuidle-psci.c | 26 +++++++++----- drivers/pmdomain/core.c | 52 +++++++++++++++++++++++++-- include/linux/pm_domain.h | 5 ++- 4 files changed, 77 insertions(+), 16 deletions(-)