Re: [PATCH 10/16] drm/i915/dp_mst: Rework pipe joiner logic in compute_config

[Imre Deak <imre.deak@intel.com>]

On Wed, Jan 28, 2026 at 07:36:30PM +0530, Ankit Nautiyal wrote:
> Similar to the DP SST, refactor `mst_stream_compute_config()` to iterate
> over joiner candidates and select the minimal joiner configuration that
> satisfies the mode requirements. This prepares the logic for future changes
> that will consider DSC slice overhead.
> 
> Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
> ---
>  drivers/gpu/drm/i915/display/intel_dp_mst.c | 47 +++++++++++++++++----
>  1 file changed, 38 insertions(+), 9 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/display/intel_dp_mst.c b/drivers/gpu/drm/i915/display/intel_dp_mst.c
> index 29ac7b2e1e9c..7a83af89ef03 100644
> --- a/drivers/gpu/drm/i915/display/intel_dp_mst.c
> +++ b/drivers/gpu/drm/i915/display/intel_dp_mst.c
> @@ -687,6 +687,7 @@ static int mst_stream_compute_config(struct intel_encoder *encoder,
>  	const struct drm_display_mode *adjusted_mode =
>  		&pipe_config->hw.adjusted_mode;
>  	int num_joined_pipes;
> +	int num_pipes;
>  	int ret = 0;
>  
>  	if (pipe_config->fec_enable &&
> @@ -700,17 +701,45 @@ static int mst_stream_compute_config(struct intel_encoder *encoder,
>  	pipe_config->output_format = INTEL_OUTPUT_FORMAT_RGB;
>  	pipe_config->has_pch_encoder = false;
>  
> -	num_joined_pipes = intel_dp_num_joined_pipes(intel_dp, connector,
> -						     adjusted_mode->crtc_hdisplay,
> -						     adjusted_mode->crtc_clock);
> +	for (num_pipes = 0; num_pipes < I915_MAX_PIPES; num_pipes++) {
> +		int max_dotclk = display->cdclk.max_dotclk_freq;
>  
> -	if (num_joined_pipes > 1)
> -		pipe_config->joiner_pipes = GENMASK(crtc->pipe + num_joined_pipes - 1, crtc->pipe);
> +		ret = -EINVAL;

I suppose in case no valid configuration is found the error code
propagated from this function should be either the error code returned
by the last mst_stream_compute_link_for_joined_pipes() call (which
should also handle the adjusted_mode->clock <= max_dotclk check
internally) for the last valid joined-pipe it was called, or -EINVAL if
there wasn't any valid joined-pipe candidate (and so
mst_stream_compute_link_for_joined_pipes() was never called). The above
ret = -EINVAL could overwrite ret which was set by the last call of
mst_stream_compute_link_for_joined_pipes(). To avoid that I'd init ret =
-EINVAL before the loop.

> +
> +		if (num_pipes == 0) {
> +			if (!connector->force_joined_pipes)
> +				continue;
> +			num_joined_pipes = connector->force_joined_pipes;
> +		} else {
> +			num_joined_pipes = num_pipes;
> +		}

I'd simplify the above as mentioned in earlier patches.

> +
> +		if (!intel_dp_can_join(display, num_joined_pipes))
> +			continue;
> +
> +		if (adjusted_mode->hdisplay >
> +		    num_joined_pipes * intel_dp_max_hdisplay_per_pipe(display))
> +			continue;
> +
> +		if (num_joined_pipes > 1)
> +			pipe_config->joiner_pipes = GENMASK(crtc->pipe + num_joined_pipes - 1,
> +							    crtc->pipe);
> +
> +		ret = mst_stream_compute_link_for_joined_pipes(encoder,
> +							       pipe_config,
> +							       conn_state,
> +							       num_joined_pipes);
> +		if (ret)
> +			continue;
> +
> +		max_dotclk *= num_joined_pipes;
> +
> +		if (adjusted_mode->clock <= max_dotclk) {
> +			ret = 0;

ret stays 0 if adjusted_mode->clock > max_dotclk (and num_pipes ==
I915_MAX_PIPES - 1). Also, this max_dotclk check should be moved to
mst_stream_compute_link_for_joined_pipes() and checked for both non-dsc
and dsc (falling back to dsc if needed), similarly to the SST case.

> +			break;
> +		}
> +	}
>  
> -	ret = mst_stream_compute_link_for_joined_pipes(encoder,
> -						       pipe_config,
> -						       conn_state,
> -						       num_joined_pipes);
>  	if (ret)
>  		return ret;
>  
> -- 
> 2.45.2
>