Skip to contentSkip to conversation

ASoC: sgtl5000: Convert mic bias to a supply widget

PATCH In review Mark Brown 4 participants 8 messages Mar 28, 2012 – Mar 29, 2012

Subsystems (3)
lore
Share & tools
lore

Mark Brown

author
#0 lore raw

PATCH 1/2 ASoC: sgtl5000: Convert mic bias to a supply widget

No current users and it's the last user of MICBIAS_E().

---
 sound/soc/codecs/sgtl5000.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)
sound/soc/codecs/sgtl5000.c +3 -3
index d192626..77beb6d 100644
@@ -197,9 +197,9 @@ static const struct snd_soc_dapm_widget sgtl5000_dapm_widgets[] = {
 	SND_SOC_DAPM_OUTPUT("HP_OUT"),
 	SND_SOC_DAPM_OUTPUT("LINE_OUT"),
 
-	SND_SOC_DAPM_MICBIAS_E("Mic Bias", SGTL5000_CHIP_MIC_CTRL, 8, 0,
-				mic_bias_event,
-				SND_SOC_DAPM_POST_PMU | SND_SOC_DAPM_PRE_PMD),
+	SND_SOC_DAPM_SUPPLY("Mic Bias", SGTL5000_CHIP_MIC_CTRL, 8, 0,
+			    mic_bias_event,
+			    SND_SOC_DAPM_POST_PMU | SND_SOC_DAPM_PRE_PMD),
 
 	SND_SOC_DAPM_PGA_E("HP", SGTL5000_CHIP_ANA_POWER, 4, 0, NULL, 0,
 			small_pop_event,
Signature · 1 line
1.7.9.1
author
#1 lore raw

PATCH 2/2 ASoC: dapm: Remove SND_SOC_DAPM_MICBIAS_E()

There are no users any more and new drivers should be using supply widgets which fully replace it anyway.

---
 include/sound/soc-dapm.h |    4 ----
 1 files changed, 0 insertions(+), 4 deletions(-)
include/sound/soc-dapm.h +0 -4
index f98dc7d..01e7ad1 100644
@@ -141,10 +141,6 @@ struct device;
 {       .id = snd_soc_dapm_mixer, .name = wname, .reg = wreg, .shift = wshift, \
 	.invert = winvert, .kcontrol_news = wcontrols, \
 	.num_kcontrols = wncontrols, .event = wevent, .event_flags = wflags}
-#define SND_SOC_DAPM_MICBIAS_E(wname, wreg, wshift, winvert, wevent, wflags) \
-{	.id = snd_soc_dapm_micbias, .name = wname, .reg = wreg, .shift = wshift, \
-	.invert = winvert, .kcontrol_news = NULL, .num_kcontrols = 0, \
-	.event = wevent, .event_flags = wflags}
 #define SND_SOC_DAPM_SWITCH_E(wname, wreg, wshift, winvert, wcontrols, \
 	wevent, wflags) \
 {	.id = snd_soc_dapm_switch, .name = wname, .reg = wreg, .shift = wshift, \
Signature · 1 line
1.7.9.1
#2 lore raw

WM8994 playback delay

Hi, Mark,

Currently, We are developing SMDK with Exynos5250 and WM8994 on 3.3-rc6. We try to enable CONFIG_PM_RUNTIME and on/off control of power domain. But It has playback delay in the first playback.(3~4sec and 6/10times)

In that time, Log maessage is
'wm8994-codec wm8994-codec: Timed out waiting for DC Servo 300' wait_for_dc_servo() function generate above message on sound/soc/codecs/wm_hubs.c

After removing wait_for_dc_servo for testing, playback delay is disappeared. So, Could you please explain the purporse of wait_for_dc_servo()? If wait_for_dc_servo() is removed, What kind of problem can be happened?

Thanks,
Sangbeom.

author
#3 lore raw

Re: WM8994 playback delay

Quoted 7 lines · Sangbeom Kim

Currently, We are developing SMDK with Exynos5250 and WM8994 on 3.3-rc6. We try to enable CONFIG_PM_RUNTIME and on/off control of power domain. But It has playback delay in the first playback.(3~4sec and 6/10times)

In that time, Log maessage is
'wm8994-codec wm8994-codec: Timed out waiting for DC Servo 300' wait_for_dc_servo() function generate above message on sound/soc/codecs/wm_hubs.c

This issue usually indicates that the clocking has an issue - are you following a similar pattern to that seen in the lowland machine driver in mainline? This is the best practice for WM8994 class parts, it should minimise power consumption when idle.

Quoted 3 lines

After removing wait_for_dc_servo for testing, playback delay is disappeared. So, Could you please explain the purporse of wait_for_dc_servo()? If wait_for_dc_servo() is removed, What kind of problem can be happened?

The DC servo is used to reduce pops on the headphone output. If the DC servo is not run the headphone outputs will produce audible artifacts when headphone is enabled which are generally considered unacceptable by users. There's a more detailed explanation of the feature in the WM8994 datasheet but that's the high level.

Acked-by on v1 1/2
#4 lore raw
In reply to Mark Brown
Quoted 32 lines · Mark Brown

No current users and it's the last user of MICBIAS_E().

---
 sound/soc/codecs/sgtl5000.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)
sound/soc/codecs/sgtl5000.c +3 -3
index d192626..77beb6d 100644
@@ -197,9 +197,9 @@ static const struct snd_soc_dapm_widget sgtl5000_dapm_widgets[] = {
 	SND_SOC_DAPM_OUTPUT("HP_OUT"),
 	SND_SOC_DAPM_OUTPUT("LINE_OUT"),
 
-	SND_SOC_DAPM_MICBIAS_E("Mic Bias", SGTL5000_CHIP_MIC_CTRL, 8, 0,
-				mic_bias_event,
-				SND_SOC_DAPM_POST_PMU | SND_SOC_DAPM_PRE_PMD),
+	SND_SOC_DAPM_SUPPLY("Mic Bias", SGTL5000_CHIP_MIC_CTRL, 8, 0,
+			    mic_bias_event,
+			    SND_SOC_DAPM_POST_PMU | SND_SOC_DAPM_PRE_PMD),
 
 	SND_SOC_DAPM_PGA_E("HP", SGTL5000_CHIP_ANA_POWER, 4, 0, NULL, 0,
 			small_pop_event,
Signature · 6 lines
1.7.9.1

_______________________________________________
Alsa-devel mailing list
[email protected]
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel

Regards
Dong Aisheng

#5 lore raw
In reply to Mark Brown
Quoted 12 lines · Mark Brown

No current users and it's the last user of MICBIAS_E().

---
 sound/soc/codecs/sgtl5000.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)
sound/soc/codecs/sgtl5000.c +0 -0
index d192626..77beb6d 100644

sound/soc/codecs/sgtl5000.c static const struct snd_soc_dapm_widget sgtl5000_dapm_widgets[] = { · line 203

Quoted 10 lines
sound/soc/codecs/sgtl5000.c +3 -3
@@ -197,9 +197,9 @@ static const struct snd_soc_dapm_widget sgtl5000_dapm_widgets[] = {
 	SND_SOC_DAPM_OUTPUT("HP_OUT"),
 	SND_SOC_DAPM_OUTPUT("LINE_OUT"),
 
-	SND_SOC_DAPM_MICBIAS_E("Mic Bias", SGTL5000_CHIP_MIC_CTRL, 8, 0,
-				mic_bias_event,
-				SND_SOC_DAPM_POST_PMU | SND_SOC_DAPM_PRE_PMD),
+	SND_SOC_DAPM_SUPPLY("Mic Bias", SGTL5000_CHIP_MIC_CTRL, 8, 0,
+			    mic_bias_event,
+			    SND_SOC_DAPM_POST_PMU | SND_SOC_DAPM_PRE_PMD),
 

It breaks my capture testing.

I'm not sure these two macros are equivalent, since they have different .id assignment. To quickly confirm the problem, I have the capture back to work with the following change.

#define SND_SOC_DAPM_SUPPLY(wname, wreg, wshift, winvert, wevent, wflags) \
-{ .id = snd_soc_dapm_supply, .name = wname, .reg = wreg, \

+{      .id = snd_soc_dapm_micbias, .name = wname, .reg = wreg, \
        .shift = wshift, .invert = winvert, .event = wevent, \
        .event_flags = wflags}
Signature · 2 lines
Regards,
Shawn
author
#6 lore raw
Quoted 10 lines · Shawn Guo

sound/soc/codecs/sgtl5000.c static const struct snd_soc_dapm_widget sgtl5000_dapm_widgets[] = { · line 202

Quoted 6 lines · Mark Brown
sound/soc/codecs/sgtl5000.c +3 -3
-	SND_SOC_DAPM_MICBIAS_E("Mic Bias", SGTL5000_CHIP_MIC_CTRL, 8, 0,
-				mic_bias_event,
-				SND_SOC_DAPM_POST_PMU | SND_SOC_DAPM_PRE_PMD),
+	SND_SOC_DAPM_SUPPLY("Mic Bias", SGTL5000_CHIP_MIC_CTRL, 8, 0,
+			    mic_bias_event,
+			    SND_SOC_DAPM_POST_PMU | SND_SOC_DAPM_PRE_PMD),

It breaks my capture testing.

Well, any capture testing that uses mic bias shouldn't work in mainline anyway - there's no hookup in the machine driver.

Quoted 3 lines

I'm not sure these two macros are equivalent, since they have different .id assignment. To quickly confirm the problem, I have the capture back to work with the following change.

You'll need to change the hookup in your machine to adapt to the change of widget type - see the previous conversions that have been done in git history for examples. No device tree bindings should be using old style mic bias hookups.

Tested-by
#7 lore raw
Quoted 15 lines · Mark Brown
Quoted 10 lines · Shawn Guo

sound/soc/codecs/sgtl5000.c static const struct snd_soc_dapm_widget sgtl5000_dapm_widgets[] = { · line 202

Quoted 6 lines · Mark Brown
sound/soc/codecs/sgtl5000.c +3 -3
-	SND_SOC_DAPM_MICBIAS_E("Mic Bias", SGTL5000_CHIP_MIC_CTRL, 8, 0,
-				mic_bias_event,
-				SND_SOC_DAPM_POST_PMU | SND_SOC_DAPM_PRE_PMD),
+	SND_SOC_DAPM_SUPPLY("Mic Bias", SGTL5000_CHIP_MIC_CTRL, 8, 0,
+			    mic_bias_event,
+			    SND_SOC_DAPM_POST_PMU | SND_SOC_DAPM_PRE_PMD),

It breaks my capture testing.

Well, any capture testing that uses mic bias shouldn't work in mainline anyway - there's no hookup in the machine driver.

Quoted 8 lines
Quoted 3 lines

I'm not sure these two macros are equivalent, since they have different .id assignment. To quickly confirm the problem, I have the capture back to work with the following change.

You'll need to change the hookup in your machine to adapt to the change of widget type - see the previous conversions that have been done in git history for examples. No device tree bindings should be using old style mic bias hookups.

Yes, with the proper machine driver adaption, it works now. Thanks, Mark.

Signature · 2 lines
Regards,
Shawn