All of lore.kernel.org
 help / color / mirror / Atom feed
From: Jani Nikula <jani.nikula@linux.intel.com>
To: Mitul Golani <mitulkumar.ajitkumar.golani@intel.com>,
	intel-gfx@lists.freedesktop.org
Cc: jyri.sarha@linux.intel.com
Subject: Re: [Intel-gfx] [PATCH 3/3] drm/i915/display: Add wrapper to Compute SAD
Date: Fri, 18 Aug 2023 11:38:02 +0300	[thread overview]
Message-ID: <87cyzk9251.fsf@intel.com> (raw)
In-Reply-To: <20230817125007.2681331-4-mitulkumar.ajitkumar.golani@intel.com>

On Thu, 17 Aug 2023, Mitul Golani <mitulkumar.ajitkumar.golani@intel.com> wrote:
> Compute SADs that takes into account the supported rate
> and channel based on the capabilities of the audio source.
> This wrapper function should encapsulate the logic for
> determining the supported rate and channel and should
> return a set of SADs that are compatible with the source.
>
> --v1:
> - call intel_audio_compute_eld in this commit as it is
> defined here
>
> --v2:
> - Handle case when max frequency is less than 32k.
> - remove drm prefix.
> - name change for parse_sad to eld_to_sad.
>
> --v3:
> - Use signed int wherever required.
> - add debug trace when channel is limited.
>
> --v4:
> - remove inline from eld_to_sad.
> - declare index outside of for loop with int type.
> - Correct mask value calculation.
> - remove drm_err, instead just return if eld parsing failed.
> - remove unncessary typecast
> - reduce indentation while parsing sad
> - use intel_audio_compute_eld as static and call bandwidth
> calculation just before that.
>
> --v9:
> - Handling the case when, sink supported channel is less
> than max supported. In that case, rate needs to be calibrate
> in accordance with available bandwidth.
>
> Signed-off-by: Mitul Golani <mitulkumar.ajitkumar.golani@intel.com>
> ---
>  drivers/gpu/drm/i915/display/intel_audio.c | 107 ++++++++++++++++++++-
>  1 file changed, 106 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/i915/display/intel_audio.c b/drivers/gpu/drm/i915/display/intel_audio.c
> index 79377e33a59b..c90ac2608eef 100644
> --- a/drivers/gpu/drm/i915/display/intel_audio.c
> +++ b/drivers/gpu/drm/i915/display/intel_audio.c
> @@ -806,6 +806,111 @@ static void calc_audio_config_params(struct intel_crtc_state *pipe_config)
>  	pipe_config->audio.max_channel_count = 0;
>  }
>  

This was added in previous patch:

 +static void calc_audio_config_params(struct intel_crtc_state *pipe_config)
 +{
 +	struct drm_display_mode *adjusted_mode = &pipe_config->hw.adjusted_mode;
 +	int channel_count;
 +	int index, rate[] = { 192000, 176000, 96000, 88000, 48000, 44100, 32000 };
 +	int audio_req_bandwidth, available_blank_bandwidth, vblank, hblank;
 +
 +	hblank = adjusted_mode->htotal - adjusted_mode->hdisplay;
 +	vblank = adjusted_mode->vtotal - adjusted_mode->vdisplay;
 +	available_blank_bandwidth = hblank * vblank *
 +				    drm_mode_vrefresh(adjusted_mode) * pipe_config->pipe_bpp;
 +	for (channel_count = MAX_CHANNEL_COUNT; channel_count > 0; channel_count--) {
 +		for (index = 0; index < ARRAY_SIZE(rate); index++) {
 +			audio_req_bandwidth = calc_audio_bw(channel_count,
 +							    rate[index]);
 +			if (audio_req_bandwidth < available_blank_bandwidth) {
 +				pipe_config->audio.max_rate = rate[index];
 +				pipe_config->audio.max_channel_count = channel_count;
 +				return;
 +			}
 +		}
 +	}
 +
 +	pipe_config->audio.max_rate = 0;
 +	pipe_config->audio.max_channel_count = 0;
 +}

And this here:

> +static void calibrate_audio_config_params(struct intel_crtc_state *pipe_config, int channel)
> +{
> +	struct drm_display_mode *adjusted_mode = &pipe_config->hw.adjusted_mode;
> +	int channel_count;
> +	int index, rate[] = { 192000, 176000, 96000, 88000, 48000, 44100, 32000 };
> +	int audio_req_bandwidth, available_blank_bandwidth, vblank, hblank;
> +
> +	hblank = adjusted_mode->htotal - adjusted_mode->hdisplay;
> +	vblank = adjusted_mode->vtotal - adjusted_mode->vdisplay;
> +	available_blank_bandwidth = hblank * vblank *
> +				    drm_mode_vrefresh(adjusted_mode) * pipe_config->pipe_bpp;
> +
> +	for (index = 0; index < ARRAY_SIZE(rate); index++) {
> +		audio_req_bandwidth = calc_audio_bw(channel_count,
> +						    rate[index]);
> +		if (audio_req_bandwidth < available_blank_bandwidth) {
> +			pipe_config->audio.max_rate = rate[index];
> +			pipe_config->audio.max_channel_count = channel_count;
> +			return;
> +		}
> +	}
> +
> +	pipe_config->audio.max_rate = 0;
> +	pipe_config->audio.max_channel_count = 0;
> +}

This kind of stuff needs to be deduplicated.

> +
> +static int sad_to_channels(const u8 *sad)
> +{
> +	return 1 + (sad[0] & 0x7);
> +}
> +
> +static u8 *eld_to_sad(u8 *eld)
> +{
> +	int ver, mnl;
> +
> +	ver = (eld[DRM_ELD_VER] & DRM_ELD_VER_MASK) >> DRM_ELD_VER_SHIFT;
> +	if (ver != 2 && ver != 31)
> +		return NULL;
> +
> +	mnl = drm_eld_mnl(eld);
> +	if (mnl > 16)
> +		return NULL;
> +
> +	return eld + DRM_ELD_CEA_SAD(mnl, 0);
> +}

I'm still not happy with the copy-paste here. The parsing details should
be hidden in drm_edid.c and that should be the Single Point of Truth how
to deal with this.

> +
> +static int get_supported_freq_mask(struct intel_crtc_state *crtc_state)
> +{
> +	int rate[] = { 32000, 44100, 48000, 88000, 96000, 176000, 192000 };

This is like the third copy of the same array already... and the values
are also present in the existing dp/hdmi aud arrays in intel_audio.c.

And those have 88200 instead of 88000, so which one is correct?

> +	int mask = 0, index;
> +
> +	for (index = 0; index < ARRAY_SIZE(rate); index++) {
> +		if (rate[index] > crtc_state->audio.max_rate)
> +			break;
> +
> +		mask |= 1 << index;
> +
> +		if (crtc_state->audio.max_rate != rate[index])
> +			continue;
> +	}
> +
> +	return mask;
> +}
> +
> +static void intel_audio_compute_eld(struct intel_crtc_state *crtc_state)
> +{
> +	struct drm_i915_private *i915 = to_i915(crtc_state->uapi.crtc->dev);
> +	u8 *eld, *sad;
> +	int index, mask = 0;
> +
> +	eld = crtc_state->eld;
> +	if (!eld)
> +		return;
> +
> +	sad = eld_to_sad(eld);
> +	if (!sad)
> +		return;
> +
> +	calc_audio_config_params(crtc_state);
> +
> +	mask = get_supported_freq_mask(crtc_state);
> +	for (index = 0; index < drm_eld_sad_count(eld); index++, sad += 3) {
> +		/*
> +		 * Respect source restricitions. Limit capabilities to a subset that is
> +		 * supported both by the source and the sink.
> +		 */
> +		if (sad_to_channels(sad) >= crtc_state->audio.max_channel_count) {
> +			sad[0] &= ~0x7;

Also not happy with magic values here.

> +			sad[0] |= crtc_state->audio.max_channel_count - 1;
> +			drm_dbg_kms(&i915->drm, "Channel count is limited to %d\n",
> +				    crtc_state->audio.max_channel_count - 1);
> +		} else {
> +			/*
> +			 * calibrate rate when, sink supported channel
> +			 * count is slight less than max supported
> +			 * channel count.
> +			 */
> +			calibrate_audio_config_params(crtc_state, sad_to_channels(sad));
> +			mask = get_supported_freq_mask(crtc_state);
> +		}
> +
> +		sad[1] &= mask;
> +	}
> +}
> +
>  bool intel_audio_compute_config(struct intel_encoder *encoder,
>  				struct intel_crtc_state *crtc_state,
>  				struct drm_connector_state *conn_state)
> @@ -827,7 +932,7 @@ bool intel_audio_compute_config(struct intel_encoder *encoder,
>  
>  	crtc_state->eld[6] = drm_av_sync_delay(connector, adjusted_mode) / 2;
>  
> -	calc_audio_config_params(crtc_state);
> +	intel_audio_compute_eld(crtc_state);
>  
>  	return true;
>  }

-- 
Jani Nikula, Intel Open Source Graphics Center

  parent reply	other threads:[~2023-08-18  8:38 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-08-17 12:50 [Intel-gfx] [PATCH 0/3] Get optimal audio frequency and channels Mitul Golani
2023-08-17 12:50 ` [Intel-gfx] [PATCH 1/3] drm/i915: Add has_audio to separate audio parameter in crtc_state Mitul Golani
2023-08-17 12:50 ` [Intel-gfx] [PATCH 2/3] drm/i915/display: Configure and initialize HDMI audio capabilities Mitul Golani
2023-08-17 13:04   ` Ville Syrjälä
2023-08-17 13:50     ` Golani, Mitulkumar Ajitkumar
2023-08-17 12:50 ` [Intel-gfx] [PATCH 3/3] drm/i915/display: Add wrapper to Compute SAD Mitul Golani
2023-08-17 15:24   ` kernel test robot
2023-08-17 15:24     ` kernel test robot
2023-08-18  8:38   ` Jani Nikula [this message]
2023-08-21 16:10     ` Golani, Mitulkumar Ajitkumar
2023-08-17 14:46 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for Get optimal audio frequency and channels (rev9) Patchwork
2023-08-17 14:46 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork
2023-08-17 15:03 ` [Intel-gfx] ✗ Fi.CI.BAT: failure " Patchwork
2023-08-18 15:24 ` [Intel-gfx] [PATCH 0/3] Get optimal audio frequency and channels Kai Vehmanen
2023-08-22  6:38 ` [Intel-gfx] ✓ Fi.CI.BAT: success for Get optimal audio frequency and channels (rev9) Patchwork

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=87cyzk9251.fsf@intel.com \
    --to=jani.nikula@linux.intel.com \
    --cc=intel-gfx@lists.freedesktop.org \
    --cc=jyri.sarha@linux.intel.com \
    --cc=mitulkumar.ajitkumar.golani@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.