Ai, With the introduction of cpuidle_device.power_specified in commit 71abbbf856a0e70ca478782505c800891260ba84 (cpuidle: extend cpuidle and menu governor to handle dynamic states) large positive numbers are now shown in sysfs rather than the power numbers specified by acpi_idle, intel_idle, and sh_idle. I expect the intent of the patch was actually to set the default to power_specified=1 for the drivers where really do specify power_usage rather than to override every driver, yes? But it has been on my clean-up list to delete power_usage from both intel_idle and acpi_idle anyway -- and I'd offer the same change to sh_idle, since it appears to be of no value there either. I planned to then to delete power_usage from cpuidle itself. But you have changed menu_select from assuming that states are in order of increasing power savings, to doing an exhaustive search of all states to find the one with the lowest power_usage that satisfies constraints. Does your platform set cpuidle_device.power_specified, or is it using the small negative integers for power_usage that you have bestowed upon the other architectures? Do you intend to modify power_usage at run-time, even when the new CPUIDLE_FLAG_IGNORE flag is available? thanks, Len Brown, Intel Open Source Technology Center _______________________________________________ linux-pm mailing list linux-pm@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linux-foundation.org/mailman/listinfo/linux-pm
