Created attachment 136111 [details] acpidump.out.gz On an IVB core-i5 ultrabook... echo freeze > /sys/power/state works, but the system wakes up in under 10 seconds /sys/kernel/debug/wakeup_sources shows that BAT0 is ticking away... Looking at what is counting in /sys/firmware/acpi/interrupts... echo disable > /sys/firmware/acpi/interrups/gpe1B stops the spurious wakeups, and I'm able to stay frozen as long as desired, until button-press wakeup. Another workaround is to boot without the acpi battery driver.
Do you get spurious wakeups if you unload the xhci driver or build a kernel without it ?
It is the ACPI battery driver that receive an ACPI notification every 10 seconds, and it is registered a wakeup source by the power_supply class. I'm working on a fix for this issue.
rmmod xhci_hcd makes no difference rmmod battery makes the system work
Patches have been accepted by Rafael. https://patchwork.kernel.org/patch/4253111/ https://patchwork.kernel.org/patch/4253121/ https://patchwork.kernel.org/patch/4253131/ https://patchwork.kernel.org/patch/4253141/
in 3.16-rc1: commit 0c5ff0ef80c2561ef20721299ecfc39c5a42f694 Author: Zhang Rui <rui.zhang@intel.com> Date: Wed May 28 15:23:35 2014 +0800 PM / sleep: unregister wakeup source when disabling device wakeup presumably the rest of the series is too. closed.