[Intel-gfx] [RFC 06/14] drm/i915: Disable vlank interrupt for disabling MIPI cmd mode

[Daniel Vetter]

On Fri, Jun 19, 2015 at 03:19:47AM +0530, Gaurav K Singh wrote:
> vblank interrupt should be disabled before starting the disable
> sequence for MIPI command mode. Otherwise when pipe is disabled
> TE interurpt will be still handled and one memory write command
> will be sent with pipe disabled. This makes the pipe hw to get
> stuck and it doesn't recover in the next enable sequence causing
> display blank out.
> 
> v2: Use drm_blank_off instead of platform specific disable vblank functions (Daniel)
> 
> Signed-off-by: Yogesh Mohan Marimuthu <yogesh.mohan.marimuthu at intel.com>
> Signed-off-by: Gaurav K Singh <gaurav.k.singh at intel.com>
> ---
>  drivers/gpu/drm/i915/intel_dsi.c |   14 ++++++++++++++
>  1 file changed, 14 insertions(+)
> 
> diff --git a/drivers/gpu/drm/i915/intel_dsi.c b/drivers/gpu/drm/i915/intel_dsi.c
> index d378246..7021591 100644
> --- a/drivers/gpu/drm/i915/intel_dsi.c
> +++ b/drivers/gpu/drm/i915/intel_dsi.c
> @@ -513,11 +513,25 @@ static void intel_dsi_enable_nop(struct intel_encoder *encoder)
>  
>  static void intel_dsi_pre_disable(struct intel_encoder *encoder)
>  {
> +	struct drm_device *dev = encoder->base.dev;
>  	struct intel_dsi *intel_dsi = enc_to_intel_dsi(&encoder->base);
> +	struct intel_crtc *intel_crtc = to_intel_crtc(encoder->base.crtc);
> +	int pipe = intel_crtc->pipe;
>  	enum port port;
>  
>  	DRM_DEBUG_KMS("\n");
>  
> +	if (is_cmd_mode(intel_dsi)) {
> +		drm_vblank_off(dev, pipe);

Now you'll have 2 drm_vblank_off in the crtc disable path, which should
result in some pretty loud warning. If it doesn't we need to fix up
drm_vblank_off to WARN, and you also need to move around the existing
drm_vblank_off in the vlv/chv code into encoder callbacks for _all_
encoders.

Hm, looking at drm_irq.c we indeed seem to be missing these crucial
WARN_ONs. I'll float a patch to address this asap.
-Daniel

> +
> +		/*
> +		 * Make sure that the last frame is sent otherwise pipe can get
> +		 * stuck. Currently providing delay time for ~2 vblanks
> +		 * assuming 60fps.
> +		 */
> +		mdelay(40);
> +	}
> +
>  	if (is_vid_mode(intel_dsi)) {
>  		/* Send Shutdown command to the panel in LP mode */
>  		for_each_dsi_port(port, intel_dsi->ports)
> -- 
> 1.7.9.5
> 
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx

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