[PATCH 1/2] ASoC: max98088: fix dai1/2_hw_params access
From: Marco Felsch m.felsch@pengutronix.de
According the reference manuals [1], [2] updating register 0x14/1c should only be done after #shdn bit is set to 0.
[1] https://datasheets.maximintegrated.com/en/ds/MAX98089.pdf [2] https://datasheets.maximintegrated.com/en/ds/MAX98088.pdf
Signed-off-by: Marco Felsch m.felsch@pengutronix.de --- sound/soc/codecs/max98088.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/sound/soc/codecs/max98088.c b/sound/soc/codecs/max98088.c index 405ec16be2b6..7f108e147355 100644 --- a/sound/soc/codecs/max98088.c +++ b/sound/soc/codecs/max98088.c @@ -977,6 +977,8 @@ static int max98088_dai1_hw_params(struct snd_pcm_substream *substream,
rate = params_rate(params);
+ snd_soc_component_update_bits(component, M98088_REG_51_PWR_SYS, M98088_SHDNRUN, 0); + switch (params_width(params)) { case 16: snd_soc_component_update_bits(component, M98088_REG_14_DAI1_FORMAT, @@ -990,8 +992,6 @@ static int max98088_dai1_hw_params(struct snd_pcm_substream *substream, return -EINVAL; }
- snd_soc_component_update_bits(component, M98088_REG_51_PWR_SYS, M98088_SHDNRUN, 0); - if (rate_value(rate, ®val)) return -EINVAL;
@@ -1047,6 +1047,8 @@ static int max98088_dai2_hw_params(struct snd_pcm_substream *substream,
rate = params_rate(params);
+ snd_soc_component_update_bits(component, M98088_REG_51_PWR_SYS, M98088_SHDNRUN, 0); + switch (params_width(params)) { case 16: snd_soc_component_update_bits(component, M98088_REG_1C_DAI2_FORMAT, @@ -1060,8 +1062,6 @@ static int max98088_dai2_hw_params(struct snd_pcm_substream *substream, return -EINVAL; }
- snd_soc_component_update_bits(component, M98088_REG_51_PWR_SYS, M98088_SHDNRUN, 0); - if (rate_value(rate, ®val)) return -EINVAL;
From: Marco Felsch m.felsch@pengutronix.de
According the datasheets [1], [2] the initial value of register 0x2f/0x31 (dai1/dai2) is 0x00 which means that dai is unmuted. So upon the first playback request the register is not touched since it is cached by regmap. But the device output keeps silent. After ending the playback session the mute() callback updates the register. Now the 2nd playback request updates the register again (-> unmute the device) and now we can really hear the output signal.
I've checked the register initial value which is '0x00' so the driver is correct. Accroding the above inspections it seems that the hardware does not update the register correctly on power up because the output is muted.
To fix that we need to explicit set the mute state. Now the first playback request gets played correctly.
[1] https://datasheets.maximintegrated.com/en/ds/MAX98089.pdf [2] https://datasheets.maximintegrated.com/en/ds/MAX98088.pdf
Signed-off-by: Marco Felsch m.felsch@pengutronix.de --- sound/soc/codecs/max98088.c | 5 +++++ 1 file changed, 5 insertions(+)
diff --git a/sound/soc/codecs/max98088.c b/sound/soc/codecs/max98088.c index 7f108e147355..c00d7726ac04 100644 --- a/sound/soc/codecs/max98088.c +++ b/sound/soc/codecs/max98088.c @@ -1710,6 +1710,11 @@ static int max98088_probe(struct snd_soc_component *component) snd_soc_component_write(component, M98088_REG_1E_DAI2_IOCFG, M98088_S2NORMAL|M98088_SDATA);
+ snd_soc_component_update_bits(component, M98088_REG_2F_LVL_DAI1_PLAY, + M98088_DAI_MUTE_MASK, M98088_DAI_MUTE); + snd_soc_component_update_bits(component, M98088_REG_31_LVL_DAI2_PLAY, + M98088_DAI_MUTE_MASK, M98088_DAI_MUTE); + max98088_handle_pdata(component);
err_access:
On Tue, Dec 13, 2022 at 10:53:28AM +0100, Rouven Czerwinski wrote:
To fix that we need to explicit set the mute state. Now the first playback request gets played correctly.
+++ b/sound/soc/codecs/max98088.c @@ -1710,6 +1710,11 @@ static int max98088_probe(struct snd_soc_component *component) snd_soc_component_write(component, M98088_REG_1E_DAI2_IOCFG, M98088_S2NORMAL|M98088_SDATA);
snd_soc_component_update_bits(component, M98088_REG_2F_LVL_DAI1_PLAY,
M98088_DAI_MUTE_MASK, M98088_DAI_MUTE);
snd_soc_component_update_bits(component, M98088_REG_31_LVL_DAI2_PLAY,
M98088_DAI_MUTE_MASK, M98088_DAI_MUTE);
Won't this be broken again after suspend? The device gets powered off over suspend, then when it powers on again with the output unmuted nothing will do another write since the register is already in the state in the cache.
Hi Mark,
On 22-12-13, Mark Brown wrote:
On Tue, Dec 13, 2022 at 10:53:28AM +0100, Rouven Czerwinski wrote:
To fix that we need to explicit set the mute state. Now the first playback request gets played correctly.
+++ b/sound/soc/codecs/max98088.c @@ -1710,6 +1710,11 @@ static int max98088_probe(struct snd_soc_component *component) snd_soc_component_write(component, M98088_REG_1E_DAI2_IOCFG, M98088_S2NORMAL|M98088_SDATA);
snd_soc_component_update_bits(component, M98088_REG_2F_LVL_DAI1_PLAY,
M98088_DAI_MUTE_MASK, M98088_DAI_MUTE);
snd_soc_component_update_bits(component, M98088_REG_31_LVL_DAI2_PLAY,
M98088_DAI_MUTE_MASK, M98088_DAI_MUTE);
Won't this be broken again after suspend? The device gets powered off over suspend, then when it powers on again with the output unmuted nothing will do another write since the register is already in the state in the cache.
I didn't found any suspend logic within the driver. Is this handled within the ASoC core?
Regards, Marco
On Thu, Dec 15, 2022 at 10:17:47AM +0100, Marco Felsch wrote:
On 22-12-13, Mark Brown wrote:
snd_soc_component_update_bits(component, M98088_REG_2F_LVL_DAI1_PLAY,
M98088_DAI_MUTE_MASK, M98088_DAI_MUTE);
snd_soc_component_update_bits(component, M98088_REG_31_LVL_DAI2_PLAY,
M98088_DAI_MUTE_MASK, M98088_DAI_MUTE);
Won't this be broken again after suspend? The device gets powered off over suspend, then when it powers on again with the output unmuted nothing will do another write since the register is already in the state in the cache.
I didn't found any suspend logic within the driver. Is this handled within the ASoC core?
Register save and restore for the device won't be.
On Tue, Dec 13, 2022 at 10:53:27AM +0100, Rouven Czerwinski wrote:
From: Marco Felsch m.felsch@pengutronix.de
According the reference manuals [1], [2] updating register 0x14/1c should only be done after #shdn bit is set to 0.
[1] https://datasheets.maximintegrated.com/en/ds/MAX98089.pdf [2] https://datasheets.maximintegrated.com/en/ds/MAX98088.pdf
Signed-off-by: Marco Felsch m.felsch@pengutronix.de
sound/soc/codecs/max98088.c | 8 ++++----
You've not provided a Signed-off-by for this so I can't do anything with it, please see Documentation/process/submitting-patches.rst for details on what this is and why it's important.
participants (3)
-
Marco Felsch
-
Mark Brown
-
Rouven Czerwinski