On Wednesday, October 17, 2018 2:01:59 PM PDT Maxime Ripard wrote:
On Wed, Oct 17, 2018 at 08:36:26AM -0700, Vasily Khoruzhick wrote:
On Wednesday, October 17, 2018 8:25:23 AM PDT Maxime Ripard wrote:
On Wed, Oct 17, 2018 at 07:41:55AM -0700, Vasily Khoruzhick wrote:
On Wed, Oct 17, 2018 at 12:45 AM Maxime Ripard
maxime.ripard@bootlin.com wrote:
On Tue, Oct 16, 2018 at 08:08:41PM -0700, Vasily Khoruzhick wrote:
On Tue, Oct 16, 2018 at 12:16 AM Maxime Ripard
maxime.ripard@bootlin.com wrote: > On Mon, Oct 15, 2018 at 08:11:41PM -0700, Vasily Khoruzhick
wrote:
> > startup() and shutdown() hooks are called for both substreams, > > so stopping either substream when another is running breaks > > the > > latter. > > > > E.g. playback breaks if capture is stopped when playback is > > running. > > > > Move code from startup() and shutdown() to resume() and > > suspend() > > hooks respectively to fix this issue > > > > Signed-off-by: Vasily Khoruzhick anarsoul@gmail.com > > --- > > v2: - rename from "ASoC: sun4i-i2s: don't try to start up > > > > or shut down DAI if it's active" > > > > - move code from startup/shutdown hooks into pm_runtime > > > > sound/soc/sunxi/sun4i-i2s.c | 61 > > +++++++++++++++---------------------- > > 1 file changed, 25 insertions(+), 36 deletions(-) > > > > diff --git a/sound/soc/sunxi/sun4i-i2s.c > > b/sound/soc/sunxi/sun4i-i2s.c > > index a4aa931ebfae..daa6c08cffbc 100644 > > --- a/sound/soc/sunxi/sun4i-i2s.c > > +++ b/sound/soc/sunxi/sun4i-i2s.c > > @@ -644,40 +644,6 @@ static int sun4i_i2s_trigger(struct > > snd_pcm_substream *substream, int cmd,> > > > > > > > > return 0; > > > > } > > > > -static int sun4i_i2s_startup(struct snd_pcm_substream > > *substream, > > - struct snd_soc_dai *dai) > > -{ > > - struct sun4i_i2s *i2s = snd_soc_dai_get_drvdata(dai); > > - > > - /* Enable the whole hardware block */ > > - regmap_update_bits(i2s->regmap, SUN4I_I2S_CTRL_REG, > > - SUN4I_I2S_CTRL_GL_EN, > > SUN4I_I2S_CTRL_GL_EN); > > - > > - /* Enable the first output line */ > > - regmap_update_bits(i2s->regmap, SUN4I_I2S_CTRL_REG, > > - SUN4I_I2S_CTRL_SDO_EN_MASK, > > - SUN4I_I2S_CTRL_SDO_EN(0)); > > - > > - > > - return clk_prepare_enable(i2s->mod_clk); > > -} > > - > > -static void sun4i_i2s_shutdown(struct snd_pcm_substream > > *substream, > > - struct snd_soc_dai *dai) > > -{ > > - struct sun4i_i2s *i2s = snd_soc_dai_get_drvdata(dai); > > - > > - clk_disable_unprepare(i2s->mod_clk); > > - > > - /* Disable our output lines */ > > - regmap_update_bits(i2s->regmap, SUN4I_I2S_CTRL_REG, > > - SUN4I_I2S_CTRL_SDO_EN_MASK, 0); > > - > > - /* Disable the whole hardware block */ > > - regmap_update_bits(i2s->regmap, SUN4I_I2S_CTRL_REG, > > - SUN4I_I2S_CTRL_GL_EN, 0); > > -} > > - > > > > static int sun4i_i2s_set_sysclk(struct snd_soc_dai *dai, int > > clk_id, > > > > unsigned int freq, int dir) > > > > { > > > > @@ -695,8 +661,6 @@ static const struct snd_soc_dai_ops > > sun4i_i2s_dai_ops = {> > > > > > > > > .hw_params = sun4i_i2s_hw_params, > > .set_fmt = sun4i_i2s_set_fmt, > > .set_sysclk = sun4i_i2s_set_sysclk, > > > > - .shutdown = sun4i_i2s_shutdown, > > - .startup = sun4i_i2s_startup, > > > > .trigger = sun4i_i2s_trigger, > > > > }; > > > > @@ -869,6 +833,21 @@ static int > > sun4i_i2s_runtime_resume(struct > > device *dev)> > > > > > > > > goto err_disable_clk; > > > > } > > > > + /* Enable the whole hardware block */ > > + regmap_update_bits(i2s->regmap, SUN4I_I2S_CTRL_REG, > > + SUN4I_I2S_CTRL_GL_EN, > > SUN4I_I2S_CTRL_GL_EN); > > + > > So this is definitely needed > > > + /* Enable the first output line */ > > + regmap_update_bits(i2s->regmap, SUN4I_I2S_CTRL_REG, > > + SUN4I_I2S_CTRL_SDO_EN_MASK, > > + SUN4I_I2S_CTRL_SDO_EN(0)); > > + > > + ret = clk_prepare_enable(i2s->mod_clk); > > + if (ret) { > > + dev_err(dev, "Failed to enable module clock\n"); > > + goto err_disable_clk; > > + } > > But we really don't want to enable the module clock before the > playback / capture actually starts. And I'm guessing it would > make > more sense to keep the output enable configuration there as > well, > since it's likely that we will have support for it in the > future, > and > we'll need additional information from ASoC when we'll do.
Should we go with v1 version of my patch then? I see no reason to shuffle code around if we keep startup() and shutdown() hooks.
I'd go for something in between. runtime_pm already provides whether the device is active or not, so you can just move the SUN4I_I2S_CTRL_GL_EN write to the runtime_resume and runtime_suspend functions, and that should fix your issue.
I still need to guard enabling output line with "if (dai->active) return", so it doesn't fix the issue unfortunately.
Why? The configuration is static, so it probably doesn't change much?
startup() and shutdown() are called once for each substream, i.e. for capture and playback.
If we start playback and capture and the stop either we end up with calling startup() twice and then shutdown() once. Since it disables I2S output it breaks substream that is still working.
Ah, right, I forgot that case. In this case, yeah, your patch makes sense
Acked-by: Maxime Ripard maxime.ripard@bootlin.com
Mark, can you pick up this one please? Thanks!
Regards, Vasily
Thanks! Maxime