[PATCHv3 0/6] ARM: OMAP3PLUS: IO daisychain support fixes

Tero Kristo [off-list ref] writes:

Hi,

The previous version of this set was sent by Vishwanath BS, but as he
is busy with other stuff, I am taking over.

Some basic info about this set based on comments received on set v2.
IO chain is now enabled always, and a re-trigger for the chain is
done during hwmod enable / disable. This means that IO chain is always
armed according to the latest pad specific WAKE_ENABLE setting, and
if it is modified, the IO chain needs to be re-armed (thus the hooks
within omap_hwmod code.) Having IO chain armed all the time allows us
to receive IO event interrupts even during non-sleep states if the
corresponding hardware module is disabled, and is unable to detect any
IO activity itself.

Version 3 contains the following changes:

- global: renamed *_trigger_wuclk_ctrl funcs as *_trigger_io_chain
- patch 3: added trigger WUCLKIN disable at the end of omap4_trigger_io_chain
  (fix pulled from the omap4 core retention set)
- patch 5: added function pointer for using machine specific io_chain trigger
  function, initialized during boot
- patch 7 dropped as the support capability for omap3 is now handled as
  part of patch 5 init

Tested on omap3 beagle + omap4 blaze boards. Suspend / resume works, and
hwmod_io interrupts are generated. Omap4 was tested with the core retention
support set.

Reviewed-by: Kevin Hilman <redacted>

I also tested this on 3430/n900 and verified IO wakeups are working as
expected.

Tested-by: Kevin Hilman <redacted>

Paul, as discussed on IRC, when you're OK with it, this can go through
your queue since it's mostly PRM core code.

Kevin