[alsa-devel] [PATCH] ASoC: set idle_bias_off=1 for all platform DAPM contexts
From: Stephen Warren swarren@nvidia.com
The ASoC core currently defaults to using STANDBY rather than OFF for idle ASoC platform devices, which causes a permanent pm_runtime_get() on them. This keeps the device active unnecessarily. This can be especially problematic when the ASoC platform device and DAI device are the same device.
The distinction between OFF and STANDBY is likely not relevant for ASoC platform drivers, since they aren't analog devices. So, solve this issue by hard-coding idle_bias_off = 1 for all ASoC platform devices. If this turns out to be a problem, this value could be sourced from the snd_soc_platform_driver, similarly to soc_probe_codec().
Note: Prior to this change, this caused a large (10) runtime_active count for the Tegra I2S controller even when not in use, and a leak in that value as streams were started and stopped. This change probably hides a bug.
Signed-off-by: Stephen Warren swarren@nvidia.com --- sound/soc/soc-core.c | 2 ++ 1 files changed, 2 insertions(+), 0 deletions(-)
diff --git a/sound/soc/soc-core.c b/sound/soc/soc-core.c index 8a17048..63e6114 100644 --- a/sound/soc/soc-core.c +++ b/sound/soc/soc-core.c @@ -1072,6 +1072,8 @@ static int soc_probe_platform(struct snd_soc_card *card, snd_soc_dapm_new_dai_widgets(&platform->dapm, dai); }
+ platform->dapm.idle_bias_off = 1; + if (driver->probe) { ret = driver->probe(platform); if (ret < 0) {
On 04/05/2012 12:28 PM, Stephen Warren wrote:
From: Stephen Warren swarren@nvidia.com
The ASoC core currently defaults to using STANDBY rather than OFF for idle ASoC platform devices, which causes a permanent pm_runtime_get() on them. This keeps the device active unnecessarily. This can be especially problematic when the ASoC platform device and DAI device are the same device.
The distinction between OFF and STANDBY is likely not relevant for ASoC platform drivers, since they aren't analog devices. So, solve this issue by hard-coding idle_bias_off = 1 for all ASoC platform devices. If this turns out to be a problem, this value could be sourced from the snd_soc_platform_driver, similarly to soc_probe_codec().
Sorry, I forgot to mention that future Tegra ASoC changes will depend on this change at runtime for correct operation.
On Thu, Apr 05, 2012 at 12:28:01PM -0600, Stephen Warren wrote:
From: Stephen Warren swarren@nvidia.com
The ASoC core currently defaults to using STANDBY rather than OFF for idle ASoC platform devices, which causes a permanent pm_runtime_get() on them. This keeps the device active unnecessarily. This can be especially problematic when the ASoC platform device and DAI device are the same device.
Applied, thanks (for 3.4 as it looks like a bug fix to me).
participants (2)
-
Mark Brown
-
Stephen Warren