Skip to content

L
linux
Switch branch/tag
  1. 08 Feb, 2021 13 commits
  3. 05 Feb, 2021 15 commits
  5. 04 Feb, 2021 8 commits
  7. 03 Feb, 2021 4 commits
    • Mark Brown's avatar
      Merge series "ASoC: soc-pcm: cleanup soc_new_pcm() and bugfix" from Kuninori... · 1c4273a5
      Mark Brown authored 
      Merge series "ASoC: soc-pcm: cleanup soc_new_pcm() and bugfix" from Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>:

Hi Mark

These are soc-pcm cleanup patchset.

	1) - 3) : cleanup soc_new_pcm() function
	4)      : cleanup dpcm_runtime_merge_xxx() function
	5)      : bugfix of snd_pcm_limit_hw_rates() order

Kuninori Morimoto (5):
  1) ASoC: soc-pcm: tidyup pcm setting
  2) ASoC: soc-pcm: add soc_get_playback_capture() and simplify soc_new_pcm()
  3) ASoC: soc-pcm: add soc_create_pcm() and simplify soc_new_pcm()
  4) ASoC: soc-pcm: use snd_pcm_hardware at dpcm_runtime_merge_xxx()
  5) ASoC: soc-pcm: fixup snd_pcm_limit_hw_rates() timing

 sound/soc/soc-pcm.c | 124 +++++++++++++++++++++++++++-----------------
 1 file changed, 75 insertions(+), 49 deletions(-)

--
2.25.1
      1c4273a5
    • Kuninori Morimoto's avatar
      ASoC: soc-pcm: fixup snd_pcm_limit_hw_rates() timing · dd5abc78
      Kuninori Morimoto authored 
      soc-pcm has snd_pcm_limit_hw_rates() which determine rate_min/rate_max.
It updates runtime->hw.rate_min/max (A) based on hw->rates (B).

	int snd_pcm_limit_hw_rates(...)
	{
		int i;
		for (...) {
(B)			if (runtime->hw.rates & (1 << i)) {
(A)				runtime->hw.rate_min = ...
				break;
			}
		}
		for (...) {
(B)			if (runtime->hw.rates & (1 << i)) {
(A)				runtime->hw.rate_max = ...
				break;
			}
		}
		return 0;
	}

This means, setup order is

	1) set hw->rates
	2) call snd_pcm_limit_hw_rates()
	3) update hw->rate_min/max

soc_pcm_init_runtime_hw() is calling it in good order

	static void soc_pcm_init_runtime_hw(xxx)
	{
		...
1)		hw->rates = snd_pcm_rate_mask_intersect(...);

2)		snd_pcm_limit_hw_rates(...);

3)		hw->rate_min = max(...);
		hw->rate_min = max(...);
		hw->rate_max = min_not_zero(...);
		hw->rate_max = min_not_zero(...);
	}

But, dpcm_fe_dai_startup() is not.

	static int dpcm_fe_dai_startup(xxx)
	{
		...
1) 3)		dpcm_set_fe_runtime(...);
2)		snd_pcm_limit_hw_rates(...);
		...
	}

More detail of dpcm_set_fe_runtime() is

	static void dpcm_set_fe_runtime()
	{
		...
		for_each_rtd_cpu_dais(rtd, i, cpu_dai) {
			...

3) 1)			dpcm_init_runtime_hw(...);
		}
		...
3) 1)		dpcm_runtime_merge_rate(...);
	}

This patch fixup these into

	static void dpcm_set_fe_runtime()
	{
		...
		for_each_rtd_cpu_dais(rtd, i, cpu_dai) {
			...

1) 2) 3)		dpcm_init_runtime_hw(...);
		}
		...
1) 2) 3)	dpcm_runtime_merge_rate(...);
	}

	static int dpcm_fe_dai_startup(xxx)
	{
		...
		dpcm_set_fe_runtime(...);
-		snd_pcm_limit_hw_rates(...);
		...
	}

Link: https://lore.kernel.org/r/87k15l7ewd.wl-kuninori.morimoto.gx@renesas.comSigned-off-by: default avatarKuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Link: https://lore.kernel.org/r/8735ytaig8.wl-kuninori.morimoto.gx@renesas.comSigned-off-by: default avatarMark Brown <broonie@kernel.org>
      dd5abc78
    • Kuninori Morimoto's avatar
      ASoC: soc-pcm: use snd_pcm_hardware at dpcm_runtime_merge_xxx() · 4b260f42
      Kuninori Morimoto authored 
      soc-pcm has dpcm_runtime_merge_xxx() functions,
but uses parameters are very verbose.

	dpcm_runtime_merge_format(...,	&runtime->hw.formats);
	dpcm_runtime_merge_chan(...,	&runtime->hw.channels_min,
					&runtime->hw.channels_max);
	dpcm_runtime_merge_rate(...,	&runtime->hw.rates,
					&runtime->hw.rate_min,
					&runtime->hw.rate_max);

We want to replace it into

	dpcm_runtime_merge_format(...,	runtime);
	dpcm_runtime_merge_chan(...,	runtime);
	dpcm_runtime_merge_rate(...,	runtime);

This patch do it.
Signed-off-by: default avatarKuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Link: https://lore.kernel.org/r/874kj9aigd.wl-kuninori.morimoto.gx@renesas.comSigned-off-by: default avatarMark Brown <broonie@kernel.org>
      4b260f42
    • Kuninori Morimoto's avatar
      ASoC: soc-pcm: add soc_create_pcm() and simplify soc_new_pcm() · 2b39123b
      Kuninori Morimoto authored 
      soc_new_pcm() implementation is very long / verbose / complex,
thus, it is very difficult to read.
If we read it carefully, we can notice that it is consisted by

	int soc_new_pcm(...)
	{
		(1) judging playback/caputre part
		(2) creating the PCM part
		(3) setup pcm/rtd part
	}

This patch adds new soc_create_pcm() for (2) part
and offload it from snd_pcm_new().
Signed-off-by: default avatarKuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Link: https://lore.kernel.org/r/875z3paigi.wl-kuninori.morimoto.gx@renesas.comSigned-off-by: default avatarMark Brown <broonie@kernel.org>
      2b39123b
GitLab Nexedi Edition | About GitLab | About Nexedi | 沪ICP备2021021310号-2 | 沪ICP备2021021310号-7