On 08/09/2017 11:55 PM, John Stultz wrote:
On Wed, Aug 9, 2017 at 6:28 AM, Ulf Hansson <ulf.hansson@xxxxxxxxxx> wrote:The commit 8503ff166504 ("i2c: designware: Avoid unnecessary resuming during system suspend"), may suggest to the PM core to try out the so called direct_complete path for system sleep. In this path, the PM core treats a runtime suspended device as it's already in a proper low power state for system sleep, which makes it skip calling the system sleep callbacks for the device, except for the ->prepare() and the ->complete() callbacks. However, the PM core may unset the direct_complete flag for a parent device, in case its child device are being system suspended before. In this scenario, the PM core invokes the system sleep callbacks, no matter if the device is runtime suspended or not. Particularly in cases of an existing i2c slave device, the above path is triggered, which breaks the assumption that the i2c device is always runtime resumed whenever the dw_i2c_plat_suspend() is being called. More precisely, dw_i2c_plat_suspend() calls clk_core_disable() and clk_core_unprepare(), for an already disabled/unprepared clock, leading to a splat in the log about clocks calls being wrongly balanced and breaking system sleep. To still allow the direct_complete path in cases when it's possible, but also to keep the fix simple, let's runtime resume the i2c device in the ->suspend() callback, before continuing to put the device into low power state. Note, in cases when the i2c device is attached to the ACPI PM domain, this problem doesn't occur, because ACPI's ->suspend() callback, assigned to acpi_subsys_suspend(), already calls pm_runtime_resume() for the device. It should also be noted that this change does not fix commit 8503ff166504 ("i2c: designware: Avoid unnecessary resuming during system suspend"). Because for the non-ACPI case, the system sleep support was already broken prior that point. Cc: <stable@xxxxxxxxxxxxxxx> # v4.4+ Signed-off-by: Ulf Hansson <ulf.hansson@xxxxxxxxxx> --- I decided to post this as a separate change, instead of as earlier folding it in the series that extends the ACPI PM domain to cope with the runtime PM centric path for system sleep. This change applies on a fresh v4.4+. If someone wants it's applied for earlier version, please send a backport yourself. It's based on 4.13 rc4 and I assume it should go as a fix via the i2c tree. Changes in v2: - Updated changelog. - Runtime resume the device in ->suspend() instead of in ->prepare().This avoids the suspend/resume warning I've seen w/o the patch on HiKey. Tested-by: John Stultz <john.stultz@xxxxxxxxxx>
Cool, no issues from acpi_lpss.c nor mfd/intel-lpss.c platforms. Tested-by: Jarkko Nikula <jarkko.nikula@xxxxxxxxxxxxxxx> Acked-by: Jarkko Nikula <jarkko.nikula@xxxxxxxxxxxxxxx>
