November 2022 - Alsa-devel - mailman.alsa-project.org

[PATCH v2 00/11] ASoC: dt-bindings: Rework Qualcomm APR/GPR Sound nodes for SM8450
by Krzysztof Kozlowski 28 Nov '22

28 Nov '22

Changes since v1 ================ 1. Patch 2: Keep compatibles in qcom,apr.yaml. 2. Patch 3: New patch. Description =========== Adding sound support for Qualcomm SM8450 SoC (and later for SC8280XP) brought some changes to APR/GPR services bindings. These bindings are part of qcom,apr.yaml: apr-or-gpr-device-node <- qcom,apr.yaml apr-gpr-service@[0-9] <- qcom,apr.yaml service-specific-components <- /schemas/sound/qcom,q6*.yaml The schema for services (apr-gpr-service@[0-9]) was not complete and is still quite not specific. It allows several incorrect combinations, like adding a clock-controller to a APM device. Restricting it would complicate the schema even more. Bringing new support for sound on Qualcomm SM8450 and SC8280XP SoC would grow it as well. Refactor the bindings before extending them for Qualcomm SM8450 SoC. DTS was sent here: https://lore.kernel.org/linux-arm-msm/20221115110053.110319-1-krzysztof.koz… Best regards, Krzysztof --- Cc: quic_srivasam(a)quicinc.com Cc: quic_plai(a)quicinc.com Krzysztof Kozlowski (11): ASoC: dt-bindings: qcom,apr: Add GLINK channel name for SM8450 ASoC: dt-bindings: qcom,apr: Split services to shared schema ASoC: dt-bindings: qcom,apr: Correct and extend example ASoC: dt-bindings: qcom,q6afe: Split to separate schema ASoC: dt-bindings: qcom,q6apm: Split to separate schema ASoC: dt-bindings: qcom,q6adm: Split to separate schema ASoC: dt-bindings: qcom,q6asm: Split to separate schema ASoC: dt-bindings: qcom,q6prm: Split to separate schema ASoC: dt-bindings: qcom,q6core: Split to separate schema ASoC: dt-bindings: qcom,q6apm-lpass-dais: Split to separate schema ASoC: dt-bindings: qcom,q6apm: Add SM8450 bedais node .../bindings/soc/qcom/qcom,apr-services.yaml | 54 ++++++ .../bindings/soc/qcom/qcom,apr.yaml | 170 ++++++------------ .../bindings/sound/qcom,q6adm-routing.yaml | 22 +-- .../devicetree/bindings/sound/qcom,q6adm.yaml | 51 ++++++ .../devicetree/bindings/sound/qcom,q6afe.yaml | 68 +++++++ .../bindings/sound/qcom,q6apm-dai.yaml | 19 +- .../bindings/sound/qcom,q6apm-lpass-dais.yaml | 32 ++++ .../devicetree/bindings/sound/qcom,q6apm.yaml | 67 +++++++ .../bindings/sound/qcom,q6asm-dais.yaml | 48 ++--- .../devicetree/bindings/sound/qcom,q6asm.yaml | 68 +++++++ .../bindings/sound/qcom,q6core.yaml | 39 ++++ .../sound/qcom,q6dsp-lpass-clocks.yaml | 40 +---- .../sound/qcom,q6dsp-lpass-ports.yaml | 57 +----- .../devicetree/bindings/sound/qcom,q6prm.yaml | 50 ++++++ MAINTAINERS | 2 +- 15 files changed, 515 insertions(+), 272 deletions(-) create mode 100644 Documentation/devicetree/bindings/soc/qcom/qcom,apr-services.yaml create mode 100644 Documentation/devicetree/bindings/sound/qcom,q6adm.yaml create mode 100644 Documentation/devicetree/bindings/sound/qcom,q6afe.yaml create mode 100644 Documentation/devicetree/bindings/sound/qcom,q6apm-lpass-dais.yaml create mode 100644 Documentation/devicetree/bindings/sound/qcom,q6apm.yaml create mode 100644 Documentation/devicetree/bindings/sound/qcom,q6asm.yaml create mode 100644 Documentation/devicetree/bindings/sound/qcom,q6core.yaml create mode 100644 Documentation/devicetree/bindings/sound/qcom,q6prm.yaml -- 2.34.1

3 23

[PATCH v1 0/4] ADD support for reading position updates from
by V sujith kumar Reddy 28 Nov '22

28 Nov '22

This series consists of 1.Reading postion update from stream box 2.selecting clock source as external ACLK clock 3.ADD virtual HS and SP DAI 4.Add AMD virtual DAI configuration support in sof core. V sujith kumar Reddy (4): ASoC: SOF: amd: Fix for reading position updates from stream box. ASoC: SOF: amd: Fix for selecting clock source as external clock. ASoC: SOF: amd: ADD HS and SP virtual DAI. ASoC: SOF: Add DAI configuration support for AMD platforms. include/sound/sof/dai-amd.h | 1 + include/sound/sof/dai.h | 2 ++ include/uapi/sound/sof/tokens.h | 5 +++++ sound/soc/sof/amd/acp-common.c | 1 + sound/soc/sof/amd/acp-ipc.c | 30 ++++++++++++++++++++++++++- sound/soc/sof/amd/acp.c | 4 +++- sound/soc/sof/amd/acp.h | 12 +++++++++++ sound/soc/sof/amd/rembrandt.c | 16 +++++++++++++++ sound/soc/sof/amd/renoir.c | 16 +++++++++++++++ sound/soc/sof/ipc3-pcm.c | 2 ++ sound/soc/sof/ipc3-topology.c | 36 +++++++++++++++++++++++++-------- sound/soc/sof/sof-audio.h | 1 + sound/soc/sof/topology.c | 10 +++++++++ 13 files changed, 126 insertions(+), 10 deletions(-) -- 2.25.1

2 6

[PATCH] ASoC: max9867: Implement exact integer mode
by Pavel Dobias 28 Nov '22

28 Nov '22

For 8kHz and 16kHz sample rates and certain PCLK values the codec can be programmed to operate in exact integer mode. If available, use it to achieve the exact sample rate. Signed-off-by: Pavel Dobias <dobias(a)2n.com> --- sound/soc/codecs/max9867.c | 33 ++++++++++++++++++++++++++++++--- 1 file changed, 30 insertions(+), 3 deletions(-) diff --git a/sound/soc/codecs/max9867.c b/sound/soc/codecs/max9867.c index 6d2941a9dbd6..e161ab037bf7 100644 --- a/sound/soc/codecs/max9867.c +++ b/sound/soc/codecs/max9867.c @@ -323,7 +323,7 @@ static int max9867_startup(struct snd_pcm_substream *substream, static int max9867_dai_hw_params(struct snd_pcm_substream *substream, struct snd_pcm_hw_params *params, struct snd_soc_dai *dai) { - int value; + int value, freq = 0; unsigned long int rate, ratio; struct snd_soc_component *component = dai->component; struct max9867_priv *max9867 = snd_soc_component_get_drvdata(component); @@ -373,6 +373,35 @@ static int max9867_dai_hw_params(struct snd_pcm_substream *substream, } regmap_update_bits(max9867->regmap, MAX9867_IFC1B, MAX9867_IFC1B_BCLK_MASK, value); + + /* Exact integer mode available for 8kHz and 16kHz sample rates + * and certain PCLK (prescaled MCLK) values. + */ + if (params_rate(params) == 8000 || + params_rate(params) == 16000) { + switch (max9867->pclk) { + case 12000000: + freq = 0x08; + break; + case 13000000: + freq = 0x0A; + break; + case 16000000: + freq = 0x0C; + break; + case 19200000: + freq = 0x0E; + break; + } + } + if (freq && params_rate(params) == 16000) + freq++; + + /* If exact integer mode not available, the freq value + * remains zero, i.e. normal mode is used. + */ + regmap_update_bits(max9867->regmap, MAX9867_SYSCLK, + MAX9867_FREQ_MASK, freq); } else { /* * digital pll locks on to any externally supplied LRCLK signal @@ -428,8 +457,6 @@ static int max9867_set_dai_sysclk(struct snd_soc_dai *codec_dai, freq); max9867->sysclk = freq; value = value << MAX9867_PSCLK_SHIFT; - /* exact integer mode is not supported */ - value &= ~MAX9867_FREQ_MASK; regmap_update_bits(max9867->regmap, MAX9867_SYSCLK, MAX9867_PSCLK_MASK, value); return 0; -- 2.20.1

2 1

[PATCH] ALSA: pcm: fix tracing reason in hw_ptr_error
by John Keeping 28 Nov '22

28 Nov '22

Strings need to be specially marked in trace events to ensure the content is captured, othewise the trace just shows the value of the pointer. Signed-off-by: John Keeping <john(a)metanate.com> --- sound/core/pcm_trace.h | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/sound/core/pcm_trace.h b/sound/core/pcm_trace.h index f18da2050772..350b40b906ca 100644 --- a/sound/core/pcm_trace.h +++ b/sound/core/pcm_trace.h @@ -88,19 +88,19 @@ TRACE_EVENT(hw_ptr_error, __field( unsigned int, device ) __field( unsigned int, number ) __field( unsigned int, stream ) - __field( const char *, reason ) + __string( reason, why ) ), TP_fast_assign( __entry->card = (substream)->pcm->card->number; __entry->device = (substream)->pcm->device; __entry->number = (substream)->number; __entry->stream = (substream)->stream; - __entry->reason = (why); + __assign_str(reason, why); ), TP_printk("pcmC%dD%d%s/sub%d: ERROR: %s", __entry->card, __entry->device, __entry->stream == SNDRV_PCM_STREAM_PLAYBACK ? "p" : "c", - __entry->number, __entry->reason) + __entry->number, __get_str(reason)) ); TRACE_EVENT(applptr, -- 2.38.1

3 3

[PATCH] ALSA: asihpi: fix missing pci_disable_device()
by Liu Shixin 28 Nov '22

28 Nov '22

pci_disable_device() need be called while module exiting, switch to use pcim_enable(), pci_disable_device() will be called in pcim_release(). Fixes: 3285ea10e9b0 ("ALSA: asihpi - Interrelated HPI tidy up.") Signed-off-by: Liu Shixin <liushixin2(a)huawei.com> --- sound/pci/asihpi/hpioctl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/sound/pci/asihpi/hpioctl.c b/sound/pci/asihpi/hpioctl.c index bb31b7fe867d..477a5b4b50bc 100644 --- a/sound/pci/asihpi/hpioctl.c +++ b/sound/pci/asihpi/hpioctl.c @@ -361,7 +361,7 @@ int asihpi_adapter_probe(struct pci_dev *pci_dev, pci_dev->device, pci_dev->subsystem_vendor, pci_dev->subsystem_device, pci_dev->devfn); - if (pci_enable_device(pci_dev) < 0) { + if (pcim_enable_device(pci_dev) < 0) { dev_err(&pci_dev->dev, "pci_enable_device failed, disabling device\n"); return -EIO; -- 2.25.1

2 1

[PATCH v3 0/2] ALSA: core: Fix deadlock when shutdown a frozen userspace
by Ricardo Ribalda 28 Nov '22

28 Nov '22

Since 83bfc7e793b5 ("ASoC: SOF: core: unregister clients and machine drivers in .shutdown") we wait for userspace to close its fds. But that will never occur with a frozen userspace (like during kexec()). Lets detect the frozen userpace and act accordingly. To: Jaroslav Kysela <perex(a)perex.cz> To: Takashi Iwai <tiwai(a)suse.com> To: "Rafael J. Wysocki" <rafael(a)kernel.org> To: Pavel Machek <pavel(a)ucw.cz> To: Len Brown <len.brown(a)intel.com> To: Kai Vehmanen <kai.vehmanen(a)linux.intel.com> To: Ranjani Sridharan <ranjani.sridharan(a)linux.intel.com> To: Pierre-Louis Bossart <pierre-louis.bossart(a)linux.intel.com> To: Mark Brown <broonie(a)kernel.org> Cc: alsa-devel(a)alsa-project.org Cc: linux-kernel(a)vger.kernel.org Cc: "Joel Fernandes (Google)" <joel(a)joelfernandes.org> Cc: linux-pm(a)vger.kernel.org Signed-off-by: Ricardo Ribalda <ribalda(a)chromium.org> --- Changes in v3: - Wrap pm_freezing in a function - Link to v2: https://lore.kernel.org/r/20221127-snd-freeze-v2-0-d8a425ea9663@chromium.org Changes in v2: - Only use pm_freezing if CONFIG_FREEZER - Link to v1: https://lore.kernel.org/r/20221127-snd-freeze-v1-0-57461a366ec2@chromium.org --- Ricardo Ribalda (2): freezer: Add processes_frozen() ALSA: core: Fix deadlock when shutdown a frozen userspace include/linux/freezer.h | 2 ++ kernel/freezer.c | 11 +++++++++++ sound/core/init.c | 13 +++++++++++++ 3 files changed, 26 insertions(+) --- base-commit: 4312098baf37ee17a8350725e6e0d0e8590252d4 change-id: 20221127-snd-freeze-1ee143228326 Best regards, -- Ricardo Ribalda <ribalda(a)chromium.org>

3 7

[PATCH v2] ALSA: core: Fix deadlock when shutdown a frozen userspace
by Ricardo Ribalda 28 Nov '22

28 Nov '22

If the user space is frozen, we cannot wait for it to complete. This fixes: [ 84.943749] Freezing user space processes ... (elapsed 0.111 seconds) done. [ 246.784446] INFO: task kexec-lite:5123 blocked for more than 122 seconds. [ 246.819035] Call Trace: [ 246.821782] <TASK> [ 246.824186] __schedule+0x5f9/0x1263 [ 246.828231] schedule+0x87/0xc5 [ 246.831779] snd_card_disconnect_sync+0xb5/0x127 ... [ 246.889249] snd_sof_device_shutdown+0xb4/0x150 [ 246.899317] pci_device_shutdown+0x37/0x61 [ 246.903990] device_shutdown+0x14c/0x1d6 [ 246.908391] kernel_kexec+0x45/0xb9 And [ 246.893222] INFO: task kexec-lite:4891 blocked for more than 122 seconds. [ 246.927709] Call Trace: [ 246.930461] <TASK> [ 246.932819] __schedule+0x5f9/0x1263 [ 246.936855] ? fsnotify_grab_connector+0x5c/0x70 [ 246.942045] schedule+0x87/0xc5 [ 246.945567] schedule_timeout+0x49/0xf3 [ 246.949877] wait_for_completion+0x86/0xe8 [ 246.954463] snd_card_free+0x68/0x89 ... [ 247.001080] platform_device_unregister+0x12/0x35 Signed-off-by: Ricardo Ribalda <ribalda(a)chromium.org> --- To: Jaroslav Kysela <perex(a)perex.cz> To: Takashi Iwai <tiwai(a)suse.com> Cc: alsa-devel(a)alsa-project.org Cc: linux-kernel(a)vger.kernel.org Cc: "Joel Fernandes (Google)" <joel(a)joelfernandes.org> --- Changes in v2: - Only use pm_freezing if CONFIG_FREEZER - Link to v1: https://lore.kernel.org/r/20221127-snd-freeze-v1-0-57461a366ec2@chromium.org --- sound/core/init.c | 17 +++++++++++++++++ 1 file changed, 17 insertions(+) diff --git a/sound/core/init.c b/sound/core/init.c index 5377f94eb211..35fb71177fea 100644 --- a/sound/core/init.c +++ b/sound/core/init.c @@ -9,6 +9,7 @@ #include <linux/module.h> #include <linux/device.h> #include <linux/file.h> +#include <linux/freezer.h> #include <linux/slab.h> #include <linux/time.h> #include <linux/ctype.h> @@ -573,6 +574,13 @@ void snd_card_disconnect_sync(struct snd_card *card) return; } +#ifdef CONFIG_FREEZER + if (pm_freezing) { + dev_err(card->dev, "Userspace is frozen, skipping sync\n"); + return; + } +#endif + spin_lock_irq(&card->files_lock); wait_event_lock_irq(card->remove_sleep, list_empty(&card->files_list), @@ -658,6 +666,15 @@ int snd_card_free(struct snd_card *card) ret = snd_card_free_when_closed(card); if (ret) return ret; + +#ifdef CONFIG_FREEZER + /* + * If userspace is frozen the wait from completion will never end. + */ + if (pm_freezing) + return 0; +#endif + /* wait, until all devices are ready for the free operation */ wait_for_completion(&released); --- base-commit: 4312098baf37ee17a8350725e6e0d0e8590252d4 change-id: 20221127-snd-freeze-1ee143228326 Best regards, -- Ricardo Ribalda <ribalda(a)chromium.org>

2 1

[PATCH] ALSA: core: Fix deadlock when shutdown a frozen userspace
by Ricardo Ribalda 28 Nov '22

28 Nov '22

If the user space is frozen, we cannot wait for it to complete. This fixes: [ 84.943749] Freezing user space processes ... (elapsed 0.111 seconds) done. [ 246.784446] INFO: task kexec-lite:5123 blocked for more than 122 seconds. [ 246.819035] Call Trace: [ 246.821782] <TASK> [ 246.824186] __schedule+0x5f9/0x1263 [ 246.828231] schedule+0x87/0xc5 [ 246.831779] snd_card_disconnect_sync+0xb5/0x127 ... [ 246.889249] snd_sof_device_shutdown+0xb4/0x150 [ 246.899317] pci_device_shutdown+0x37/0x61 [ 246.903990] device_shutdown+0x14c/0x1d6 [ 246.908391] kernel_kexec+0x45/0xb9 And [ 246.893222] INFO: task kexec-lite:4891 blocked for more than 122 seconds. [ 246.927709] Call Trace: [ 246.930461] <TASK> [ 246.932819] __schedule+0x5f9/0x1263 [ 246.936855] ? fsnotify_grab_connector+0x5c/0x70 [ 246.942045] schedule+0x87/0xc5 [ 246.945567] schedule_timeout+0x49/0xf3 [ 246.949877] wait_for_completion+0x86/0xe8 [ 246.954463] snd_card_free+0x68/0x89 ... [ 247.001080] platform_device_unregister+0x12/0x35 Signed-off-by: Ricardo Ribalda <ribalda(a)chromium.org> --- To: Jaroslav Kysela <perex(a)perex.cz> To: Takashi Iwai <tiwai(a)suse.com> Cc: alsa-devel(a)alsa-project.org Cc: linux-kernel(a)vger.kernel.org Cc: "Joel Fernandes (Google)" <joel(a)joelfernandes.org> --- sound/core/init.c | 13 +++++++++++++ 1 file changed, 13 insertions(+) diff --git a/sound/core/init.c b/sound/core/init.c index 5377f94eb211..f614a2cdea3d 100644 --- a/sound/core/init.c +++ b/sound/core/init.c @@ -9,6 +9,7 @@ #include <linux/module.h> #include <linux/device.h> #include <linux/file.h> +#include <linux/freezer.h> #include <linux/slab.h> #include <linux/time.h> #include <linux/ctype.h> @@ -573,6 +574,11 @@ void snd_card_disconnect_sync(struct snd_card *card) return; } + if (pm_freezing) { + dev_err(card->dev, "Userspace is frozen, skipping sync\n"); + return; + } + spin_lock_irq(&card->files_lock); wait_event_lock_irq(card->remove_sleep, list_empty(&card->files_list), @@ -658,6 +664,13 @@ int snd_card_free(struct snd_card *card) ret = snd_card_free_when_closed(card); if (ret) return ret; + + /* + * If userspace is frozen the wait from completion will never end. + */ + if (pm_freezing) + return 0; + /* wait, until all devices are ready for the free operation */ wait_for_completion(&released); --- base-commit: 4312098baf37ee17a8350725e6e0d0e8590252d4 change-id: 20221127-snd-freeze-1ee143228326 Best regards, -- Ricardo Ribalda <ribalda(a)chromium.org>

2 1

[PATCH 1/2] ASoC: tas5805m: rework to avoid scheduling while atomic.
by Daniel Beer 27 Nov '22

27 Nov '22

There's some setup we need to do in order to get the DSP initialized, and this can't be done until a bit-clock is ready. In an earlier version of this driver, this work was done in a DAPM callback. The DAPM callback doesn't guarantee that the bit-clock is running, so the work was moved instead to the trigger callback. Unfortunately this callback runs in atomic context, and the setup code needs to do I2C transactions. Here we use a work_struct to kick off the setup in a thread instead. Fixes: ec45268467f4 ("ASoC: add support for TAS5805M digital amplifier") Signed-off-by: Daniel Beer <daniel.beer(a)igorinstitute.com> --- sound/soc/codecs/tas5805m.c | 126 ++++++++++++++++++++++++------------ 1 file changed, 86 insertions(+), 40 deletions(-) diff --git a/sound/soc/codecs/tas5805m.c b/sound/soc/codecs/tas5805m.c index beb4ec629a03..d720c67b9675 100644 --- a/sound/soc/codecs/tas5805m.c +++ b/sound/soc/codecs/tas5805m.c @@ -154,6 +154,7 @@ static const uint32_t tas5805m_volume[] = { #define TAS5805M_VOLUME_MIN 0 struct tas5805m_priv { + struct i2c_client *i2c; struct regulator *pvdd; struct gpio_desc *gpio_pdn_n; @@ -165,6 +166,9 @@ struct tas5805m_priv { int vol[2]; bool is_powered; bool is_muted; + + struct work_struct work; + struct mutex lock; }; static void set_dsp_scale(struct regmap *rm, int offset, int vol) @@ -181,13 +185,11 @@ static void set_dsp_scale(struct regmap *rm, int offset, int vol) regmap_bulk_write(rm, offset, v, ARRAY_SIZE(v)); } -static void tas5805m_refresh(struct snd_soc_component *component) +static void tas5805m_refresh(struct tas5805m_priv *tas5805m) { - struct tas5805m_priv *tas5805m = - snd_soc_component_get_drvdata(component); struct regmap *rm = tas5805m->regmap; - dev_dbg(component->dev, "refresh: is_muted=%d, vol=%d/%d\n", + dev_dbg(&tas5805m->i2c->dev, "refresh: is_muted=%d, vol=%d/%d\n", tas5805m->is_muted, tas5805m->vol[0], tas5805m->vol[1]); regmap_write(rm, REG_PAGE, 0x00); @@ -226,8 +228,11 @@ static int tas5805m_vol_get(struct snd_kcontrol *kcontrol, struct tas5805m_priv *tas5805m = snd_soc_component_get_drvdata(component); + mutex_lock(&tas5805m->lock); ucontrol->value.integer.value[0] = tas5805m->vol[0]; ucontrol->value.integer.value[1] = tas5805m->vol[1]; + mutex_unlock(&tas5805m->lock); + return 0; } @@ -243,11 +248,13 @@ static int tas5805m_vol_put(struct snd_kcontrol *kcontrol, snd_soc_kcontrol_component(kcontrol); struct tas5805m_priv *tas5805m = snd_soc_component_get_drvdata(component); + int ret = 0; if (!(volume_is_valid(ucontrol->value.integer.value[0]) && volume_is_valid(ucontrol->value.integer.value[1]))) return -EINVAL; + mutex_lock(&tas5805m->lock); if (tas5805m->vol[0] != ucontrol->value.integer.value[0] || tas5805m->vol[1] != ucontrol->value.integer.value[1]) { tas5805m->vol[0] = ucontrol->value.integer.value[0]; @@ -256,9 +263,10 @@ static int tas5805m_vol_put(struct snd_kcontrol *kcontrol, tas5805m->vol[0], tas5805m->vol[1], tas5805m->is_powered); if (tas5805m->is_powered) - tas5805m_refresh(component); - return 1; + tas5805m_refresh(tas5805m); + ret = 1; } + mutex_unlock(&tas5805m->lock); return 0; } @@ -294,54 +302,83 @@ static int tas5805m_trigger(struct snd_pcm_substream *substream, int cmd, struct snd_soc_component *component = dai->component; struct tas5805m_priv *tas5805m = snd_soc_component_get_drvdata(component); - struct regmap *rm = tas5805m->regmap; - unsigned int chan, global1, global2; switch (cmd) { case SNDRV_PCM_TRIGGER_START: case SNDRV_PCM_TRIGGER_RESUME: case SNDRV_PCM_TRIGGER_PAUSE_RELEASE: - dev_dbg(component->dev, "DSP startup\n"); - - /* We mustn't issue any I2C transactions until the I2S - * clock is stable. Furthermore, we must allow a 5ms - * delay after the first set of register writes to - * allow the DSP to boot before configuring it. - */ - usleep_range(5000, 10000); - send_cfg(rm, dsp_cfg_preboot, - ARRAY_SIZE(dsp_cfg_preboot)); - usleep_range(5000, 15000); - send_cfg(rm, tas5805m->dsp_cfg_data, - tas5805m->dsp_cfg_len); - - tas5805m->is_powered = true; - tas5805m_refresh(component); + dev_dbg(component->dev, "clock start\n"); + schedule_work(&tas5805m->work); break; case SNDRV_PCM_TRIGGER_STOP: case SNDRV_PCM_TRIGGER_SUSPEND: case SNDRV_PCM_TRIGGER_PAUSE_PUSH: - dev_dbg(component->dev, "DSP shutdown\n"); + break; - tas5805m->is_powered = false; + default: + return -EINVAL; + } - regmap_write(rm, REG_PAGE, 0x00); - regmap_write(rm, REG_BOOK, 0x00); + return 0; +} - regmap_read(rm, REG_CHAN_FAULT, &chan); - regmap_read(rm, REG_GLOBAL_FAULT1, &global1); - regmap_read(rm, REG_GLOBAL_FAULT2, &global2); +static void do_work(struct work_struct *work) +{ + struct tas5805m_priv *tas5805m = + container_of(work, struct tas5805m_priv, work); + struct regmap *rm = tas5805m->regmap; - dev_dbg(component->dev, - "fault regs: CHAN=%02x, GLOBAL1=%02x, GLOBAL2=%02x\n", - chan, global1, global2); + dev_dbg(&tas5805m->i2c->dev, "DSP startup\n"); - regmap_write(rm, REG_DEVICE_CTRL_2, DCTRL2_MODE_HIZ); - break; + mutex_lock(&tas5805m->lock); + /* We mustn't issue any I2C transactions until the I2S + * clock is stable. Furthermore, we must allow a 5ms + * delay after the first set of register writes to + * allow the DSP to boot before configuring it. + */ + usleep_range(5000, 10000); + send_cfg(rm, dsp_cfg_preboot, ARRAY_SIZE(dsp_cfg_preboot)); + usleep_range(5000, 15000); + send_cfg(rm, tas5805m->dsp_cfg_data, tas5805m->dsp_cfg_len); + + tas5805m->is_powered = true; + tas5805m_refresh(tas5805m); + mutex_unlock(&tas5805m->lock); +} - default: - return -EINVAL; +static int tas5805m_dac_event(struct snd_soc_dapm_widget *w, + struct snd_kcontrol *kcontrol, int event) +{ + struct snd_soc_component *component = snd_soc_dapm_to_component(w->dapm); + struct tas5805m_priv *tas5805m = + snd_soc_component_get_drvdata(component); + struct regmap *rm = tas5805m->regmap; + + if (event & SND_SOC_DAPM_PRE_PMD) { + unsigned int chan, global1, global2; + + dev_dbg(component->dev, "DSP shutdown\n"); + cancel_work_sync(&tas5805m->work); + + mutex_lock(&tas5805m->lock); + if (tas5805m->is_powered) { + tas5805m->is_powered = false; + + regmap_write(rm, REG_PAGE, 0x00); + regmap_write(rm, REG_BOOK, 0x00); + + regmap_read(rm, REG_CHAN_FAULT, &chan); + regmap_read(rm, REG_GLOBAL_FAULT1, &global1); + regmap_read(rm, REG_GLOBAL_FAULT2, &global2); + + dev_dbg(component->dev, "fault regs: CHAN=%02x, " + "GLOBAL1=%02x, GLOBAL2=%02x\n", + chan, global1, global2); + + regmap_write(rm, REG_DEVICE_CTRL_2, DCTRL2_MODE_HIZ); + } + mutex_unlock(&tas5805m->lock); } return 0; @@ -354,7 +391,8 @@ static const struct snd_soc_dapm_route tas5805m_audio_map[] = { static const struct snd_soc_dapm_widget tas5805m_dapm_widgets[] = { SND_SOC_DAPM_AIF_IN("DAC IN", "Playback", 0, SND_SOC_NOPM, 0, 0), - SND_SOC_DAPM_DAC("DAC", NULL, SND_SOC_NOPM, 0, 0), + SND_SOC_DAPM_DAC_E("DAC", NULL, SND_SOC_NOPM, 0, 0, + tas5805m_dac_event, SND_SOC_DAPM_PRE_PMD), SND_SOC_DAPM_OUTPUT("OUT") }; @@ -375,11 +413,14 @@ static int tas5805m_mute(struct snd_soc_dai *dai, int mute, int direction) struct tas5805m_priv *tas5805m = snd_soc_component_get_drvdata(component); + mutex_lock(&tas5805m->lock); dev_dbg(component->dev, "set mute=%d (is_powered=%d)\n", mute, tas5805m->is_powered); + tas5805m->is_muted = mute; if (tas5805m->is_powered) - tas5805m_refresh(component); + tas5805m_refresh(tas5805m); + mutex_unlock(&tas5805m->lock); return 0; } @@ -434,6 +475,7 @@ static int tas5805m_i2c_probe(struct i2c_client *i2c) if (!tas5805m) return -ENOMEM; + tas5805m->i2c = i2c; tas5805m->pvdd = devm_regulator_get(dev, "pvdd"); if (IS_ERR(tas5805m->pvdd)) { dev_err(dev, "failed to get pvdd supply: %ld\n", @@ -507,6 +549,9 @@ static int tas5805m_i2c_probe(struct i2c_client *i2c) gpiod_set_value(tas5805m->gpio_pdn_n, 1); usleep_range(10000, 15000); + INIT_WORK(&tas5805m->work, do_work); + mutex_init(&tas5805m->lock); + /* Don't register through devm. We need to be able to unregister * the component prior to deasserting PDN# */ @@ -527,6 +572,7 @@ static void tas5805m_i2c_remove(struct i2c_client *i2c) struct device *dev = &i2c->dev; struct tas5805m_priv *tas5805m = dev_get_drvdata(dev); + cancel_work_sync(&tas5805m->work); snd_soc_unregister_component(dev); gpiod_set_value(tas5805m->gpio_pdn_n, 0); usleep_range(10000, 15000); -- 2.38.1

2 1

[PATCH 2/2] ASoC: tas5805m: add missing page switch.
by Daniel Beer 27 Nov '22

27 Nov '22

In tas5805m_refresh, we switch pages to update the DSP volume control, but we need to switch back to page 0 before trying to alter the soft-mute control. This latter page-switch was missing. Fixes: ec45268467f4 ("ASoC: add support for TAS5805M digital amplifier") Signed-off-by: Daniel Beer <daniel.beer(a)igorinstitute.com> --- sound/soc/codecs/tas5805m.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/sound/soc/codecs/tas5805m.c b/sound/soc/codecs/tas5805m.c index d720c67b9675..5cf7726904c2 100644 --- a/sound/soc/codecs/tas5805m.c +++ b/sound/soc/codecs/tas5805m.c @@ -203,6 +203,9 @@ static void tas5805m_refresh(struct tas5805m_priv *tas5805m) set_dsp_scale(rm, 0x24, tas5805m->vol[0]); set_dsp_scale(rm, 0x28, tas5805m->vol[1]); + regmap_write(rm, REG_PAGE, 0x00); + regmap_write(rm, REG_BOOK, 0x00); + /* Set/clear digital soft-mute */ regmap_write(rm, REG_DEVICE_CTRL_2, (tas5805m->is_muted ? DCTRL2_MUTE : 0) | -- 2.38.1

1 0