Upon trying it again, I'm actually seeing the load drop down in the "steady gpe17 after boot" case. I just didn't wait long enough for the boot load spike to get out of the average. Please disregard that second half of my previous comment.
Here are some more experiments: With processor.max_cstate=0 in boot args and without changing battery/external power state, I'm seeing a steady gpe17 interrupt count, and the 1-minute load after being booted for 5 minutes at 0.01. Upon disconnecting external power, I do not see the load jump and gpe17 remains steady. With idle=halt in boot args and without changing battery/external power state, I'm seeing a steady gpe17 interrupt count, and the 1-minute load after being booted for 5 minutes at 0.02. Upon disconnecting external power, I do see gpe17 spin and the 1-minute load jump to about 1.50 after two minutes and perf top showing: Samples: 21K of event 'cycles', Event count (approx.): 2163359670 Overhead Shared Object Symbol 45.50% [kernel] [k] acpi_os_read_port 5.73% [kernel] [k] acpi_ec_read_status.isra.4 With idle=mwait in boot args and without changing battery/external power state, I'm seeing a steady gpe17 interrupt count, and the 1-minute load after being booted for 5 minutes at 0.02. Upon disconnecting external power, I do see gpe17 spin and the 1-minute load jump to about 1.45 after two minutes and perf top showing: Samples: 100K of event 'cycles', Event count (approx.): 7259748699 Overhead Shared Object Symbol 34.33% [kernel] [k] acpi_os_read_port 20.85% [kernel] [k] read_hpet 6.56% [kernel] [k] acpi_processor_ffh_cstate_enter 4.28% [kernel] [k] acpi_ec_read_status.isra.4 In the two cases where gpe17 is spinning, disabling it drops the load back down to about 1.0 with perf top then showing: Samples: 127K of event 'cycles', Event count (approx.): 1193854375 Overhead Shared Object Symbol 40.05% [kernel] [k] acpi_processor_ffh_cstate_enter 8.25% [kernel] [k] read_hpet 5.06% [kernel] [k] _raw_spin_lock 2.46% [kernel] [k] acpi_os_read_port 0.99% [kernel] [k] hpet_legacy_next_event -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/1489622 Title: 15.10 20150826 snapshot -- MacBookPro3,1 has high load (and temperature) despite low CPU and low IO To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1489622/+subscriptions -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
