On Mon, Nov 23, 2009 at 02:45:16PM +0900, Joonyoung Shim wrote:
If the widget was powered up when the device enters to the suspend, it remains on because the power of the widget doesn't change, but we need to power down the widget on suspend.
Hrm, yes - good spot, this got missed in the refactoring to make the bais maintinence less dependant on stream state. However...
power = w->power_check(w);
if (power)
if (power) { sys_power = 1;if (event == SND_SOC_DAPM_STREAM_SUSPEND)power = 0;}
We don't want to set sys_power here (since we want to bring the bias down to standby in preparatio for suspend) and we can skip the power check so something like the patch below ought to do the trick. Could you test and let me know, please?
That said, we probably want some more flexibility here for systems which want to suspend with bypass paths active. They'll want to pause streams being directly driven by the CPU but leave any other paths active. That would be a new feature, though.
diff --git a/sound/soc/soc-dapm.c b/sound/soc/soc-dapm.c index eaadb4b..0d294ef 100644 --- a/sound/soc/soc-dapm.c +++ b/sound/soc/soc-dapm.c @@ -977,9 +977,19 @@ static int dapm_power_widgets(struct snd_soc_codec *codec, int event) if (!w->power_check) continue;
- power = w->power_check(w); - if (power) - sys_power = 1; + /* If we're suspending then pull down all the + * power. */ + switch (event) { + case SND_SOC_DAPM_STREAM_SUSPEND: + power = 0; + break; + + default: + power = w->power_check(w); + if (power) + sys_power = 1; + break; + }
if (w->power == power) continue; @@ -1003,8 +1013,12 @@ static int dapm_power_widgets(struct snd_soc_codec *codec, int event) case SND_SOC_DAPM_STREAM_RESUME: sys_power = 1; break; + case SND_SOC_DAPM_STREAM_SUSPEND: + sys_power = 0; + break; case SND_SOC_DAPM_STREAM_NOP: sys_power = codec->bias_level != SND_SOC_BIAS_STANDBY; + break; default: break; }