Linux 6.16 Lands Proper Power Management Fix For Code That Caused Power Regression
Linux 6.15 mistakenly shipped with a nasty power regression for some systems, such as those relying on the "nosmt" option to disable Simultaneous Multi-Threading / Hyper Threading. That idle power regression was fixed for Linux 6.15.2 and Linux 6.16 Git by reverting the troubled patch that introduced the regression. Now merged ahead of Linux 6.16-rc2 is a proper fix for that problematic patch so it could be re-merged without the power fallout.
Intel engineer and Linux power management subsystem maintainer Rafael Wysocki fixed up the code and sent it out as part of this week's Linux 6.16-rc2 power management fixes:
The reverted patch was re-applied while now adjusting the Intel Idle driver as well as the ACPI processor code to rescan "dead" SMT siblings during initialization. With this the power regression should be resolved as the proper solution beyond the original revert.
See this pull for all of the merged power management changes ahead of tomorrow's Linux 6.16-rc2 release.
Intel engineer and Linux power management subsystem maintainer Rafael Wysocki fixed up the code and sent it out as part of this week's Linux 6.16-rc2 power management fixes:
"Eliminate mwait_play_dead_cpuid_hint() again after reverting its elimination during the 6.16 merge window due to a problem with handling "dead" SMT siblings, but this time prevent leaving them in C1 after initialization by taking them online and back offline when a proper cpuidle driver for the platform has been registered (Rafael Wysocki)."
The reverted patch was re-applied while now adjusting the Intel Idle driver as well as the ACPI processor code to rescan "dead" SMT siblings during initialization. With this the power regression should be resolved as the proper solution beyond the original revert.
See this pull for all of the merged power management changes ahead of tomorrow's Linux 6.16-rc2 release.