Re: [PATCH] drm: mali-dp: Set encoder possible_clones

["Ville Syrjälä" <ville.syrjala@linux.intel.com>]

On Thu, Jul 12, 2018 at 01:54:07PM +0100, Alexandru-Cosmin Gheorghe wrote:
> On Thu, Jul 12, 2018 at 03:37:36PM +0300, Ville Syrjälä wrote:
> > On Thu, Jul 12, 2018 at 09:48:56AM +0100, Alexandru Gheorghe wrote:
> > > Set possible_clones field to report that the writeback connector and
> > > the one driving the display could be enabled at the same time.
> > > 
> > > Signed-off-by: Alexandru Gheorghe <alexandru-cosmin.gheorghe@arm.com>
> > > ---
> > >  drivers/gpu/drm/arm/malidp_drv.c | 10 ++++++++++
> > >  1 file changed, 10 insertions(+)
> > > 
> > > diff --git a/drivers/gpu/drm/arm/malidp_drv.c b/drivers/gpu/drm/arm/malidp_drv.c
> > > index 5b72605..3664da9 100644
> > > --- a/drivers/gpu/drm/arm/malidp_drv.c
> > > +++ b/drivers/gpu/drm/arm/malidp_drv.c
> > > @@ -616,6 +616,8 @@ static int malidp_bind(struct device *dev)
> > >  	struct malidp_hw_device *hwdev;
> > >  	struct platform_device *pdev = to_platform_device(dev);
> > >  	struct of_device_id const *dev_id;
> > > +	struct drm_encoder *encoder;
> > > +	int num_encoders = 0;
> > >  	/* number of lines for the R, G and B output */
> > >  	u8 output_width[MAX_OUTPUT_CHANNELS];
> > >  	int ret = 0, i;
> > > @@ -737,6 +739,14 @@ static int malidp_bind(struct device *dev)
> > >  		goto bind_fail;
> > >  	}
> > >  
> > > +	list_for_each_entry(encoder, &drm->mode_config.encoder_list, head) {
> > > +		num_encoders++;
> > > +	}
> > 
> > dev->mode_config.num_encoder ?
> 
> Thanks for that, I will use it.
> 
> > 
> > > +
> > > +	list_for_each_entry(encoder, &drm->mode_config.encoder_list, head) {
> > > +		encoder->possible_clones = (1 << num_encoders) -  1;
> > > +	}
> > 
> > Your commit message speaks only about two specific encoders but here
> > you're claiming any encoder can be cloned with any other. I can't
> > actually figure out how many encoders there can be because the encoder
> > registration seems to happen by magic means. Some explanation would
> > be nice.
> > 
> > I'm interested because I'd like to clean up the possible_clones stuff
> > to actually reflect reality so that userspace could actually use that
> > information for something. See
> > https://patchwork.freedesktop.org/series/44849/
> > 
> 
> There are just two encoders, encoder->possible_clones = 0x3 would've
> done the job, do you think I should use it to better convey intent?

Are there any direct pointers to those encoders? If there are I
might recommend a more explicit
 a.possible_clones = drm_encoder_mask(a) | drm_encoder_mask(b);
 b.possible_clones = drm_encodrr_mask(a) | drm_encoder_mask(b);
or something like that.

But if there aren't then I think the num_encoder thing should work
fine. But I would suggest adding a comment to explain the situation,
and maybe even a WARN_ON(num_encoder != 2) so that people will
realize that this code needs to be updated if more encoders show up
in the future.

-- 
Ville Syrjälä
Intel
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel