Re: [PATCH] OMAPDSS: TFP410: use gpio_set_value_cansleep

[Tomi Valkeinen <tomi.valkeinen@ti.com>]

[-- Attachment #1: Type: text/plain, Size: 2013 bytes --]

On Wed, 2012-05-09 at 16:37 -0700, Tony Lindgren wrote:
> * Russ Dill <Russ.Dill@ti.com> [120509 15:53]:
> > On Wed, May 9, 2012 at 3:14 PM, Tony Lindgren <tony@atomide.com> wrote:
> > > * Russ Dill <Russ.Dill@ti.com> [120509 15:12]:
> > >> The Beagleboard xM gpio used for TFP410 powerdown is connected through
> > >> an I2C attached chip which means setting the GPIO can sleep. Code that
> > >> calls tfp410_power_on/off holds a mutex, so sleeping should be fine.
> > >
> > > What's the error without this patch? Or just no display?
> > >
> > > Just wondering if it's safe to merge Tomi's clean up series to
> > > arm-soc tree..
> > 
> > The only platform that has a problem is Beagleboard xM, and that is
> > only after 'ARM: OMAP: Cleanup Beagleboard DVI reset gpio' is applied.
> > Since the context actually can sleep, the only consequence is a
> > WARN_ON statement.
> > 
> > So yes, it should be safe.
> 
> Well since I have not actually merged it with other branches yet, I'll wait
> for Tomi to apply that and repull his for-l-o-3.5 branch.

You can pull for-l-o-3.5 as it is now, there's no need to change it.
This _cansleep change is a separate dss specific change.

So to summarize:

Currently the powerdown GPIO for tfp410 is handled in the board files
(and called reset gpio), with gpio_set_value(). My cleanup series moves
this to the tfp410 driver.

BB-xM needs to use gpio_set_value_cansleep() in tfp410 to function
properly, so the tfp410 driver needs to be changed, as this patch does.
I will take this patch to the dss tree.

So it's safe to pull my cleanup series, but if I understood correctly,
applying Russ's "[PATCH v4] ARM: OMAP: Cleanup Beagleboard DVI reset
gpio" will cause WARN_ONs on BB-xM until this patch to tfp410 is also
applied.

But it doesn't sound too serious, so I think it's safe to apply the
"cleanup beagleboard dvi" patch also. The warning will go away when both
l-o and dss trees are pulled in the merge window.

 Tomi


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]