[PATCH 0/2] PM: sleep: Fix hibernation resume issues

[Abel Vesa]

Two things the PM core gets wrong on the hibernation paths, found while
getting the Qualcomm Glymur CRD to resume from hibernation.

The first one marks devices without PM callbacks as active when
restoring from hibernation, even though nothing resumes them, so the
next runtime suspend of such a device releases resources that are
already off. Seen with the CoreSight devices, whose AMBA bus clock gets
disabled twice.

The second one arms the wakeup interrupts for the freeze and quiesce
phases of a hibernation, which can only abort the operation, since the
system never goes to sleep there. The WWAN slot WAKE# of the Glymur CRD
fires as soon as the modem is powered down, and the hibernation never
gets to write the image.

Tested on the Qualcomm Glymur CRD, where hibernation and restore now
complete.

Prepared with help from an LLM-based assistant.

Signed-off-by: Abel Vesa <abel.vesa@xxxxxxxxxxxxxxxx>
---
Abel Vesa (2):
      PM: sleep: Do not mark devices without PM callbacks active on resume
      PM: sleep: Do not arm wakeup IRQs for hibernation freeze and quiesce

 drivers/base/power/main.c | 28 +++++++++++++++++++++++++---
 1 file changed, 25 insertions(+), 3 deletions(-)
---
base-commit: 6628e7d4c6b25c8d18f1345e3a20d21829aae7e4
change-id: 20261023-b4-pm-sleep-hibernation-fixes-c39e49aa6600

Best regards,
--  

Abel