On Wed, Sep 06, 2017 at 11:55:01AM +0200, Rafael J. Wysocki wrote:

On Wednesday, September 6, 2017 11:16:46 AM CEST Johannes Stezenbach wrote:

quoted

In case it matters, what I tested was v4.13 + sound/topic/dollar-cove-ti-4.13-v5
(PMIC etc) + sound/topic/soc-cx2072x-4.13 (sound codec driver)
+ "S0ix blocker debug patch v3" from https://bugzilla.kernel.org/show_bug.cgi?id=193891
+ your 3 patches.

I suspend + resume a few times, no problems.  (Of course S0ix
still doesn't work according to /sys/kernel/debug/pmc_atom/sleep_state.)

Well, we'll need to address the S0ix thing, but that will require some
investigation.

Can you please open a bug entry for that at bugzilla.kernel.org
(against Power Management -> Hibernation/Suspend) and CC it to me
so that it doesn't fall of the radar?

I'll do that tomorrow, TIA for your interest in the issue.

Today I spent several hours to bisect the S0ix "regression"
on Asus E200HA, I put regression in quotes because S0ix
only ever worked with gross hacks like poking registers
using busybox devmem.  Anyway, the gross hack worked
on v4.12 and stopped working in v4.13-rc and the
culprit is d31fd43c0f9a4 "clk: x86: Do not gate clocks
enabled by the firmware" (I'll report this properly in a seperate mail).

But I guess that's not the part you're interested in,
the bug entry should cover interaction of dw i2c, acpi pm,
acpi-lpss and ACPI OpRegion, right?


Thanks,
Johannes