[PATCH v2 4/4] drm/msm/dp: Add support for LTTPR handling

[Abel Vesa]

On 24-12-11 15:56:53, Johan Hovold wrote:
> On Wed, Dec 11, 2024 at 03:04:15PM +0200, Abel Vesa wrote:
>  
> > +static void msm_dp_display_lttpr_init(struct msm_dp_display_private *dp)
> > +{
> > +	int lttpr_count;
> > +
> > +	if (drm_dp_read_lttpr_common_caps(dp->aux, dp->panel->dpcd,
> > +					  dp->lttpr_caps))
> > +		return;
> > +
> > +	lttpr_count = drm_dp_lttpr_count(dp->lttpr_caps);
> 
> I was gonna say shouldn't you handle errors here, but that explains the
> non-negative check I commented on the first patch in the series.

So lttpr_count is a bit weird. It's either between 0 and 8, or -ERANGE
if more than 8 LTTPRs are found, or -EINVAL if for some reason the
DP_PHY_REPEATER_CNT register contains an invalid value.
(see drm_dp_lttpr_count())

Now, I think I should just drop the lttr_count local variable here entirely.

> 
> This looks error prone, but I think you should at least update the
> kernel doc comment to drm_dp_lttpr_init() in the first patch so that
> it's clear that you pass in the number of LTTPRs *or* an errno.

Yes, I'll do that. Will mention all possible values and what they mean.
And will probably point to the drm_dp_lttpr_count() as well, just to be
safe.

> 
> > +
> > +	drm_dp_lttpr_init(dp->aux, lttpr_count);
> > +}
> > +
> >  static int msm_dp_display_process_hpd_high(struct msm_dp_display_private *dp)
> >  {
> >  	struct drm_connector *connector = dp->msm_dp_display.connector;
> >  	const struct drm_display_info *info = &connector->display_info;
> >  	int rc = 0;
> >  
> > +	msm_dp_display_lttpr_init(dp);
> 
> It looks like you ignore errors on purpose so I guess that's fine.

Maybe I should at least throw an error, just like the i915 does.
Will do that.

> 
> > +
> >  	rc = msm_dp_panel_read_sink_caps(dp->panel, connector);
> >  	if (rc)
> >  		goto end;
> 
> Either way, this is needed for external display on my x1e80100 machines,
> while not breaking the X13s:
> 
> Tested-by: Johan Hovold <johan+linaro at kernel.org>
> 
> Johan

Thanks for reviewing and testing,
Abel