On X1E80100 (Dell XPS 13 9345) this patch seems to cause poweroff to reboot instead. Reverting just the FW_START hunk avoids it, and so does keeping the GPU awake until shutdown, so I suspect the PDC being asleep at system off.
I've posted a fix, "drm/msm/adreno: Leave the GPU RSC awake for system poweroff": https://lore.kernel.org/all/[email protected]/ On this machine it resolves the reboot: poweroff and reboot work on AC and battery, and deep suspend still drops the GPU votes. There may be a better approach, e.g. waking only the RSC; happy to test. Thanks, Michael
