03.12.2019 19:27, Thierry Reding пишет:
From: Thierry Reding treding@nvidia.com
The Tegra DRM driver heavily relies on the implementations for runtime suspend/resume to be called at specific times. Unfortunately, there are some cases where that doesn't work. One example is if the user disables runtime PM for a given subdevice. Another example is that the PM core acquires a reference to runtime PM during system sleep, effectively preventing devices from going into low power modes. This is intentional to avoid nasty race conditions, but it also causes system sleep to not function properly on all Tegra systems.
Fix this by not implementing runtime PM at all. Instead, a minimal, reference-counted suspend/resume infrastructure is added to the host1x bus. This has the benefit that it can be used regardless of the system power state (or any transitions we might be in), or whether or not the user allows runtime PM.
Atomic modesetting guarantees that these functions will end up being called at the right point in time, so the pitfalls for the more generic runtime PM do not apply here.
Signed-off-by: Thierry Reding treding@nvidia.com
Couldn't we just use pm_runtime_force_suspend/resume whenever it is necessary to enforce the suspend/resume?
I briefly looked through the previous discussion and don't see why the forced suspend/resume isn't suitable. Please excuse me if I'm missing the point.
Why planes/outputs need to care about resuming DC controller at all? Doesn't DRM core take care of enabling DC for us by enabling CRTC before planes/outputs are enabled?