[Intel-gfx] [PATCH v3.1 1/3] drm/i915: Fix broken mst get_hw_state.

[Daniel Vetter]

On Thu, Aug 06, 2015 at 03:37:08PM +0200, Maarten Lankhorst wrote:
> Op 06-08-15 om 14:59 schreef Daniel Vetter:
> > On Thu, Aug 06, 2015 at 01:47:35PM +0200, Maarten Lankhorst wrote:
> >> This function always returned false because intel_connector->encoder
> >> is always NULL. Instead use the attached encoder from atomic.
> > Note that you've broken this since you removed the updating of
> > intel_connector->encoder somewhere in the 4.3 atomic series. So this is a
> > regression fix. Can you please digg out the right commit?
> 
> Looks like 'Use full atomic modeset' broke it. intel_connector->encoder was updated in its set_config function..
> 
> Would below amend for this commit be good?
> diff --git a/drivers/gpu/drm/i915/intel_dp_mst.c b/drivers/gpu/drm/i915/intel_dp_mst.c
> index ee2a77144373..671715ba8538 100644
> --- a/drivers/gpu/drm/i915/intel_dp_mst.c
> +++ b/drivers/gpu/drm/i915/intel_dp_mst.c
> @@ -173,6 +173,11 @@ static void intel_mst_pre_enable_dp(struct intel_encoder *encoder)
>  		return;
>  	}
>  
> +	/* MST encoders are bound to a crtc, not to a connector,
> +	 * force the mapping here for get_hw_state.
> +	 */
> +	found->encoder = encoder;
> +
>  	DRM_DEBUG_KMS("%d\n", intel_dp->active_mst_links);
>  	intel_mst->port = found->port;
>  
> 
> And corresponding amendment to the convert connector to atomic..
> 
> diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
> index a6617283b1bd..90caa62cf2c3 100644
> --- a/drivers/gpu/drm/i915/intel_display.c
> +++ b/drivers/gpu/drm/i915/intel_display.c
> @@ -6313,7 +6313,7 @@ static void intel_connector_check_state(struct intel_connector *connector)
>  		I915_STATE_WARN(!crtc->state->active,
>  		      "connector is active, but attached crtc isn't\n");
>  
> -		if (!encoder)
> +		if (!encoder || encoder->type == INTEL_OUTPUT_DP_MST)
>  			return;

Nah, I don't want to keep using the legacy stuff, your patch is good imo.
What I wanted is sha1 of the patches which broke the old
intel_connector->encoder trick.

And of course we get to chase around the code for all the other users of
intel_connector->encoder.
-Daniel

>  
>  		I915_STATE_WARN(conn_state->best_encoder != encoder,
> 
> >> Signed-off-by: Maarten Lankhorst <maarten.lankhorst at linux.intel.com>
> >> ---
> >>  drivers/gpu/drm/i915/intel_dp_mst.c | 7 +++++--
> >>  1 file changed, 5 insertions(+), 2 deletions(-)
> >>
> >> diff --git a/drivers/gpu/drm/i915/intel_dp_mst.c b/drivers/gpu/drm/i915/intel_dp_mst.c
> >> index e3a5864160fa..ff01569158ea 100644
> >> --- a/drivers/gpu/drm/i915/intel_dp_mst.c
> >> +++ b/drivers/gpu/drm/i915/intel_dp_mst.c
> >> @@ -395,9 +395,12 @@ static const struct drm_encoder_funcs intel_dp_mst_enc_funcs = {
> >>  
> >>  static bool intel_dp_mst_get_hw_state(struct intel_connector *connector)
> >>  {
> >> -	if (connector->encoder) {
> >> +	struct intel_encoder *encoder;
> >> +
> >> +	encoder = to_intel_encoder(connector->base.state->best_encoder);
> > Strictly speaking this won't work for hw state readout since what we
> > really need to do is ask the mst connector which mst stream it accepts and
> > then compare that with all the mst encoders ...
> It's good enough to make sure the encoder callbacks did run.
> > But that's been broken since forever. Only side-effect here is that
> > fastboot won't work with mst and that's imo totally ok. Please also add
> > this to your commit message.
> HW readout runs before mst probing, so you wouldn't get the mst connectors anyway with fastboot. Not sure it needs a fixme. :-)

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch