Re: [PATCH] drm/edid: Support type 7 timings

[Jani Nikula <jani.nikula@linux.intel.com>]

On Wed, 19 Jan 2022, Yaroslav Bolyukin <iam@lach.pw> wrote:
> Per VESA DisplayID Standard v2.0: Type VII Timing – Detailed Timing Data
>
> Definitions were already provided as type I, but not used

Thanks for the patch. Functionally I think it looks correct, and
something we'll want. I do have some nitpicks though, comments inline.

For the next version, please consider Cc'ing the intel-gfx list as well
to get our CI results on it too.

BR,
Jani.

>
> Signed-off-by: Yaroslav Bolyukin <iam@lach.pw>
> ---
>  drivers/gpu/drm/drm_edid.c  | 26 +++++++++++++++++---------
>  include/drm/drm_displayid.h |  6 +++---
>  2 files changed, 20 insertions(+), 12 deletions(-)
>
> diff --git a/drivers/gpu/drm/drm_edid.c b/drivers/gpu/drm/drm_edid.c
> index 12893e7be..5fcefd9b5 100644
> --- a/drivers/gpu/drm/drm_edid.c
> +++ b/drivers/gpu/drm/drm_edid.c
> @@ -5404,13 +5404,17 @@ u32 drm_add_display_info(struct drm_connector *connector, const struct edid *edi
>  	return quirks;
>  }
>  
> -static struct drm_display_mode *drm_mode_displayid_detailed(struct drm_device *dev,
> -							    struct displayid_detailed_timings_1 *timings)
> +static struct drm_display_mode *drm_mode_displayid_detailed_1_7(struct drm_device *dev,
> +								struct displayid_detailed_timings_1_7 *timings,
> +								bool type_7)

I think the function rename here is unnecessary.

>  {
>  	struct drm_display_mode *mode;
>  	unsigned pixel_clock = (timings->pixel_clock[0] |
>  				(timings->pixel_clock[1] << 8) |
>  				(timings->pixel_clock[2] << 16)) + 1;
> +	// type 7 allows higher precision pixel clock

Please don't use // style comments.

For the comment contents, I think you should just state the units for
each; 10 kHz for type I, kHz for type VII.

> +	if (!type_7)
> +		pixel_clock *= 10;

Please don't mix declarations and code.

>  	unsigned hactive = (timings->hactive[0] | timings->hactive[1] << 8) + 1;
>  	unsigned hblank = (timings->hblank[0] | timings->hblank[1] << 8) + 1;
>  	unsigned hsync = (timings->hsync[0] | (timings->hsync[1] & 0x7f) << 8) + 1;
> @@ -5426,7 +5430,7 @@ static struct drm_display_mode *drm_mode_displayid_detailed(struct drm_device *d
>  	if (!mode)
>  		return NULL;
>  
> -	mode->clock = pixel_clock * 10;
> +	mode->clock = pixel_clock;

Since we used to have the multiplication here (and we don't mix
declarations and code anyway) I'd keep it here.

Maybe:

	mode->clock = type_7 ? pixel_clock : pixel_clock * 10;

>  	mode->hdisplay = hactive;
>  	mode->hsync_start = mode->hdisplay + hsync;
>  	mode->hsync_end = mode->hsync_start + hsync_width;
> @@ -5449,10 +5453,12 @@ static struct drm_display_mode *drm_mode_displayid_detailed(struct drm_device *d
>  	return mode;
>  }
>  
> -static int add_displayid_detailed_1_modes(struct drm_connector *connector,
> -					  const struct displayid_block *block)
> +static int add_displayid_detailed_1_7_modes(struct drm_connector *connector,
> +					    const struct displayid_block *block,
> +					    bool type_7)
>  {
> -	struct displayid_detailed_timing_block *det = (struct displayid_detailed_timing_block *)block;
> +	struct displayid_detailed_timing_1_7_block *det =
> +		(struct displayid_detailed_timing_1_7_block *)block;

I think the displayid_detailed_timing_block ->
displayid_detailed_timing_1_7_block rename is unnecessary.

>  	int i;
>  	int num_timings;
>  	struct drm_display_mode *newmode;
> @@ -5463,9 +5469,9 @@ static int add_displayid_detailed_1_modes(struct drm_connector *connector,
>  
>  	num_timings = block->num_bytes / 20;
>  	for (i = 0; i < num_timings; i++) {
> -		struct displayid_detailed_timings_1 *timings = &det->timings[i];
> +		struct displayid_detailed_timings_1_7 *timings = &det->timings[i];
>  
> -		newmode = drm_mode_displayid_detailed(connector->dev, timings);
> +		newmode = drm_mode_displayid_detailed_1_7(connector->dev, timings, type_7);
>  		if (!newmode)
>  			continue;
>  
> @@ -5485,7 +5491,9 @@ static int add_displayid_detailed_modes(struct drm_connector *connector,
>  	displayid_iter_edid_begin(edid, &iter);
>  	displayid_iter_for_each(block, &iter) {
>  		if (block->tag == DATA_BLOCK_TYPE_1_DETAILED_TIMING)
> -			num_modes += add_displayid_detailed_1_modes(connector, block);
> +			num_modes += add_displayid_detailed_1_7_modes(connector, block, false);
> +		else if (block->tag == DATA_BLOCK_2_TYPE_7_DETAILED_TIMING)
> +			num_modes += add_displayid_detailed_1_7_modes(connector, block, true);

I'd probably not add a true/false parameter here, since we pass in block
anyway, and the function can have and initialize the bool local variable
internally based on block->tag.

>  	}
>  	displayid_iter_end(&iter);
>  
> diff --git a/include/drm/drm_displayid.h b/include/drm/drm_displayid.h
> index 7ffbd9f7b..268ff5e1f 100644
> --- a/include/drm/drm_displayid.h
> +++ b/include/drm/drm_displayid.h
> @@ -111,7 +111,7 @@ struct displayid_tiled_block {
>  	u8 topology_id[8];
>  } __packed;
>  
> -struct displayid_detailed_timings_1 {
> +struct displayid_detailed_timings_1_7 {
>  	u8 pixel_clock[3];
>  	u8 flags;
>  	u8 hactive[2];
> @@ -124,9 +124,9 @@ struct displayid_detailed_timings_1 {
>  	u8 vsw[2];
>  } __packed;
>  
> -struct displayid_detailed_timing_block {
> +struct displayid_detailed_timing_1_7_block {

Like I said, I wouldn't rename this.

>  	struct displayid_block base;
> -	struct displayid_detailed_timings_1 timings[];
> +	struct displayid_detailed_timings_1_7 timings[];
>  };
>  
>  #define DISPLAYID_VESA_MSO_OVERLAP	GENMASK(3, 0)
>
> base-commit: 99613159ad749543621da8238acf1a122880144e

-- 
Jani Nikula, Intel Open Source Graphics Center