[PATCH 1/2] ALSA: hda/realtek: Accelerate the front MIC charging and discharging of the 897 codec

[PATCH 1/2] ALSA: hda/realtek: Accelerate the front MIC charging and discharging of the 897 codec

[wangdich9700]

From: wangdicheng <wangdicheng@kylinos.cn>

The front MIC capacitance of the Inspur board is too large, resulting in a longer charging and discharging time required,
which in turn prevents recording for 5 seconds before startup.
Therefore, software is used to increase the delay by 2 seconds to avoid this issue.

Signed-off-by: wangdicheng <wangdicheng@kylinos.cn>
---
 sound/pci/hda/patch_realtek.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/sound/pci/hda/patch_realtek.c b/sound/pci/hda/patch_realtek.c
index 877137cb09ac..88675eae9447 100644
--- a/sound/pci/hda/patch_realtek.c
+++ b/sound/pci/hda/patch_realtek.c
@@ -1001,6 +1001,14 @@ static int alc_resume(struct hda_codec *codec)
 	if (!spec->no_depop_delay)
 		msleep(150); /* to avoid pop noise */
 	codec->patch_ops.init(codec);
+
+	if (codec->core.subsystem_id == 0x10ec1304) {
+		snd_hda_codec_write(codec, 0x19, 0, AC_VERB_SET_PIN_WIDGET_CONTROL, 0x04);
+		msleep(50);
+		snd_hda_codec_write(codec, 0x19, 0, AC_VERB_SET_PIN_WIDGET_CONTROL, 0x44);
+		msleep(2000);
+		snd_hda_codec_write(codec, 0x19, 0, AC_VERB_SET_PIN_WIDGET_CONTROL, PIN_VREF80);
+	}
 	snd_hda_regmap_sync(codec);
 	hda_call_check_power_status(codec, 0x01);
 	return 0;
-- 
2.25.1

[PATCH 2/2] ALSA: hda/realtek: Fix ALC662 codec pop noise on recording

[wangdich9700]

From: wangdicheng <wangdicheng@kylinos.cn>

Pop noise originates from unstable mic_vref during power_save transitions.
Remediation: Disable VREF at resume, delay 1000ms before re-enabling to guarantee circuit stabilization pre-capture.

Signed-off-by: wangdicheng <wangdicheng@kylinos.cn>
---
 sound/pci/hda/patch_realtek.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/sound/pci/hda/patch_realtek.c b/sound/pci/hda/patch_realtek.c
index 88675eae9447..665e291dea75 100644
--- a/sound/pci/hda/patch_realtek.c
+++ b/sound/pci/hda/patch_realtek.c
@@ -997,16 +997,21 @@ static int alc_suspend(struct hda_codec *codec)
 static int alc_resume(struct hda_codec *codec)
 {
 	struct alc_spec *spec = codec->spec;
+	int vref_delay = 0;
 
 	if (!spec->no_depop_delay)
 		msleep(150); /* to avoid pop noise */
 	codec->patch_ops.init(codec);
 
-	if (codec->core.subsystem_id == 0x10ec1304) {
+	if (codec->core.subsystem_id == 0x10ec1304)
+		vref_delay = 2000;
+	else if (codec->core.subsystem_id == 0x1b505809)
+		vref_delay = 1000;
+	if (vref_delay > 0) {
 		snd_hda_codec_write(codec, 0x19, 0, AC_VERB_SET_PIN_WIDGET_CONTROL, 0x04);
 		msleep(50);
 		snd_hda_codec_write(codec, 0x19, 0, AC_VERB_SET_PIN_WIDGET_CONTROL, 0x44);
-		msleep(2000);
+		msleep(vref_delay);
 		snd_hda_codec_write(codec, 0x19, 0, AC_VERB_SET_PIN_WIDGET_CONTROL, PIN_VREF80);
 	}
 	snd_hda_regmap_sync(codec);
-- 
2.25.1

Re: [PATCH 1/2] ALSA: hda/realtek: Accelerate the front MIC charging and discharging of the 897 codec

[Takashi Iwai]

On Mon, 04 Aug 2025 05:05:17 +0200,
wangdich9700@163.com wrote:
> 
> From: wangdicheng <wangdicheng@kylinos.cn>
> 
> The front MIC capacitance of the Inspur board is too large, resulting in a longer charging and discharging time required,
> which in turn prevents recording for 5 seconds before startup.
> Therefore, software is used to increase the delay by 2 seconds to avoid this issue.

First off, the HD-audio code was replaced in the latest Linus tree to
sound/hda/* directories, so please adapt the patches accordingly.

About the code change:

- The 2.5 seconds wait is way too long and inappropriate to be put in
  the resume callback.  It has to be fixed differently.

- Does it apply really to all devices with 0x10ec1304?  It's Realtek's
  vendor id, so this can match to many devices.

- Last but not least, it's not nice to have a code that depends on the
  certain vendor id there.  Again, please implement in a different way
  so that it can be handled in more device-specific code rather than
  the common function.


thanks,

Takashi

> 
> Signed-off-by: wangdicheng <wangdicheng@kylinos.cn>
> ---
>  sound/pci/hda/patch_realtek.c | 8 ++++++++
>  1 file changed, 8 insertions(+)
> 
> diff --git a/sound/pci/hda/patch_realtek.c b/sound/pci/hda/patch_realtek.c
> index 877137cb09ac..88675eae9447 100644
> --- a/sound/pci/hda/patch_realtek.c
> +++ b/sound/pci/hda/patch_realtek.c
> @@ -1001,6 +1001,14 @@ static int alc_resume(struct hda_codec *codec)
>  	if (!spec->no_depop_delay)
>  		msleep(150); /* to avoid pop noise */
>  	codec->patch_ops.init(codec);
> +
> +	if (codec->core.subsystem_id == 0x10ec1304) {
> +		snd_hda_codec_write(codec, 0x19, 0, AC_VERB_SET_PIN_WIDGET_CONTROL, 0x04);
> +		msleep(50);
> +		snd_hda_codec_write(codec, 0x19, 0, AC_VERB_SET_PIN_WIDGET_CONTROL, 0x44);
> +		msleep(2000);
> +		snd_hda_codec_write(codec, 0x19, 0, AC_VERB_SET_PIN_WIDGET_CONTROL, PIN_VREF80);
> +	}
>  	snd_hda_regmap_sync(codec);
>  	hda_call_check_power_status(codec, 0x01);
>  	return 0;
> -- 
> 2.25.1
> 

Re: [PATCH 1/2] ALSA: hda/realtek: Accelerate the front MIC charging and discharging of the 897 codec

[Takashi Iwai]

On Mon, 04 Aug 2025 08:17:58 +0200,
Takashi Iwai wrote:
> 
> On Mon, 04 Aug 2025 05:05:17 +0200,
> wangdich9700@163.com wrote:
> > 
> > From: wangdicheng <wangdicheng@kylinos.cn>
> > 
> > The front MIC capacitance of the Inspur board is too large, resulting in a longer charging and discharging time required,
> > which in turn prevents recording for 5 seconds before startup.
> > Therefore, software is used to increase the delay by 2 seconds to avoid this issue.
> 
> First off, the HD-audio code was replaced in the latest Linus tree to
> sound/hda/* directories, so please adapt the patches accordingly.
> 
> About the code change:
> 
> - The 2.5 seconds wait is way too long and inappropriate to be put in
>   the resume callback.  It has to be fixed differently.
> 
> - Does it apply really to all devices with 0x10ec1304?  It's Realtek's
>   vendor id, so this can match to many devices.
> 
> - Last but not least, it's not nice to have a code that depends on the
>   certain vendor id there.  Again, please implement in a different way
>   so that it can be handled in more device-specific code rather than
>   the common function.

One more thing: the mailing list for Linux kernel sound stuff is
linux-sound@vger.kernel.org.  Please use this one instead of
alsa-devel ML.


thanks,

Takashi