May 2024 - Alsa-devel - mailman.alsa-project.org

[PATCH 0/2] ASoC: Intel: avs: Switch to acpi-nhlt
by Cezary Rojewski 09 May '24

09 May '24

The change is based on rafael/acpi-nhlt [1] immutable branch which Rafael kindly prepared for me. Without the topmost changes to ACPI/NHLT, the patches present will fail to compile. Recent changes for the ACPI tree [2] refactored interfaces of the NHLT table. Currently we have two implementations - one found in acpi subsystem (unused) and one in sound/hda/. As NHLT is part of ACPI, idea is to make the former useful and then switch all users of existing sound/hda/intel-nhlt.c to this new interface over time and remove the duplicate afterward. Two patches present here migrate the avs-driver and the intel-dsp-config module. [1]: https://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git/?h=acpi… [2]: https://lore.kernel.org/linux-acpi/20240319083018.3159716-1-cezary.rojewski… Amadeusz Sławiński (2): ALSA: hda: intel-dsp-config: Switch to ACPI NHLT ASoC: Intel: avs: Switch to ACPI NHLT sound/hda/Kconfig | 1 + sound/hda/intel-dsp-config.c | 16 ++++++++------ sound/soc/intel/Kconfig | 1 + sound/soc/intel/avs/avs.h | 1 - sound/soc/intel/avs/board_selection.c | 7 +++--- sound/soc/intel/avs/core.c | 10 ++++----- sound/soc/intel/avs/path.c | 32 +++++++++++++-------------- 7 files changed, 34 insertions(+), 34 deletions(-) base-commit: a640acab545b21ed1f347376f34d34e461ea92ba -- 2.25.1

3 8

Re: [syzbot] [kernel?] INFO: rcu detected stall in sys_wait4 (4)
by syzbot 09 May '24

09 May '24

syzbot has bisected this issue to: commit 51ea51b18904cd1a0fb244ce41dfd903c2ada628 Author: Shuming Fan <shumingf(a)realtek.com> Date: Fri Dec 23 05:58:46 2022 +0000 ASoC: rt711-sdca: add jack detection mode for JD2 100K bisection log: https://syzkaller.appspot.com/x/bisect.txt?x=123e6f88980000 start commit: fe46a7dd189e Merge tag 'sound-6.9-rc1' of git://git.kernel.. git tree: upstream final oops: https://syzkaller.appspot.com/x/report.txt?x=113e6f88980000 console output: https://syzkaller.appspot.com/x/log.txt?x=163e6f88980000 kernel config: https://syzkaller.appspot.com/x/.config?x=fe78468a74fdc3b7 dashboard link: https://syzkaller.appspot.com/bug?extid=6969434de600a6ba9f07 syz repro: https://syzkaller.appspot.com/x/repro.syz?x=1091a5f6180000 C reproducer: https://syzkaller.appspot.com/x/repro.c?x=17a22c13180000 Reported-by: syzbot+6969434de600a6ba9f07(a)syzkaller.appspotmail.com Fixes: 51ea51b18904 ("ASoC: rt711-sdca: add jack detection mode for JD2 100K") For information about bisection process see: https://goo.gl/tpsmEJ#bisection

1 0

[PATCH v2 0/7] ASoC: samsung: midas-audio: Add GPIO-based headset jack detection
by Artur Weber 09 May '24

09 May '24

Many of Samsung's Exynos 4 devices share the same midas-audio driver to handle the codec setup. While most of these devices, including the Midas itself, use the jack detection provided by the WM8994 driver, other devices such as the Samsung Galaxy Tab 3 8.0 (lt01) use two GPIOs and an ADC channel to determine jack insertion, the jack's type, and button presses (for headsets with volume up/down/play buttons). In the downstream kernel, this behavior is implemented in the sec-jack driver[1], and the per-device settings are configured in *-jack.c files in the mach folder (see e.g. the Tab 3's implementation[2]). This patchset implements this mechanism in the midas_wm1811.c driver, and adds new DTS options to allow for its configuration. It also enables jack detection for the Samsung Galaxy Tab 3 8.0. A very similar mechanism was already present in the aries_wm8994.c driver[3]; this implementation heavily borrows from it, though there are a few extra cleanups as well. Signed-off-by: Artur Weber <aweber.kernel(a)gmail.com> [1] https://github.com/gr8nole/android_kernel_samsung_smdk4x12/blob/lineage-14.… [2] https://github.com/gr8nole/android_kernel_samsung_smdk4x12/blob/lineage-14.… [3] https://github.com/torvalds/linux/blob/master/sound/soc/samsung/aries_wm899… --- Changes in v2: - Added vendor prefix to threshold properties - Added separate headset mic bias regulator - Changed some cases of dev_err + return with return dev_err_probe - Added an extra patch to replace some previous dev_err + return cases with dev_err_probe - Moved tab3 DTS wm1811 codec config changes to separate commit --- Artur Weber (7): ASoC: dt-bindings: samsung,midas-audio: Add headset mic bias supply ASoC: dt-bindings: samsung,midas-audio: Add GPIO-based headset jack detection ASoC: samsung: midas_wm1811: Add headset mic bias supply support ASoC: samsung: midas_wm1811: Add GPIO-based headset jack detection ASoC: samsung: midas_wm1811: Use dev_err_probe where appropriate ARM: dts: samsung: exynos4212-tab3: Fix headset mic, add jack detection ARM: dts: samsung: exynos4212-tab3: Fix up wm1811 codec config .../bindings/sound/samsung,midas-audio.yaml | 33 ++ arch/arm/boot/dts/samsung/exynos4212-tab3.dtsi | 31 +- sound/soc/samsung/Kconfig | 2 +- sound/soc/samsung/midas_wm1811.c | 343 +++++++++++++++++++-- 4 files changed, 379 insertions(+), 30 deletions(-) --- base-commit: e67572cd2204894179d89bd7b984072f19313b03 change-id: 20240502-midas-wm1811-gpio-jack-b10226b17ecc Best regards, -- Artur Weber <aweber.kernel(a)gmail.com>

3 9

[PATCH] ASoC: audio-graph-card2: remove unneeded of_node_get()
by Kuninori Morimoto 09 May '24

09 May '24

"lnk" is used as "port0", and it will be used to get "port1" by of_get_next_child(ports, lnk). It will call of_node_put() inside. This function is calling of_node_get() to make up for it, but it doesn't call paired of_node_put(port0) when it quit function. This of_node_get() itself is not needed, let's remove it. Reported-by: Dan Carpenter <dan.carpenter(a)linaro.org> Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx(a)renesas.com> --- sound/soc/generic/audio-graph-card2.c | 2 -- 1 file changed, 2 deletions(-) diff --git a/sound/soc/generic/audio-graph-card2.c b/sound/soc/generic/audio-graph-card2.c index b1c675c6b6db..825125d8985b 100644 --- a/sound/soc/generic/audio-graph-card2.c +++ b/sound/soc/generic/audio-graph-card2.c @@ -1021,8 +1021,6 @@ static int graph_count_c2c(struct asoc_simple_priv *priv, struct device_node *codec0 = of_graph_get_remote_port(ep0); struct device_node *codec1 = of_graph_get_remote_port(ep1); - of_node_get(lnk); - /* * codec2codec { * ports { -- 2.25.1

2 1

Motu 828 (28 x 32) support
by GitHub issues - opened 08 May '24

08 May '24

alsa-project/alsa-ucm-conf issue #415 was opened from beatboxa: The newest [Motu 828 (28x32)](https://motu.com/en-us/products/828/) does not appear to have a UCM yet. It is a class compliant USB device; so as a result, it generally "works" at a very basic level; but default channel mappings and profiles do not. The system appears to default to only 2 channels in and 2 channels out. So, for example, surround playback seems to mix into either the left or right channel. In qpwgraph, it shows 30 inputs (I believe this includes an additional 2-channel loopback at the end) and 32 outputs, all marked as "AUX#," where # starts with 0. I was able to identify which AUX# corresponds to which input or output on the device, through the use of qpwgraph and the device's built-in digital VU meter. I have tried to create a UCM or even an asoundrc; but I have no idea what I'm doing. The documentation and examples are not intuitive for me. Could I collaborate with someone in order to get this audio interface supported? I'd be happy to run any commands and tests you advise on. Issue URL : https://github.com/alsa-project/alsa-ucm-conf/issues/415 Repository URL: https://github.com/alsa-project/alsa-ucm-conf

1 0

ucm2: Qualcomm: x1e80100: add recording via DMIC01
by GitHub pull_request - opened 08 May '24

08 May '24

alsa-project/alsa-ucm-conf pull request #414 was opened from krzk: Add support for Digital Microphones DMIC01 on Qualcomm X1E80100 CRD board. Request URL : https://github.com/alsa-project/alsa-ucm-conf/pull/414 Patch URL : https://github.com/alsa-project/alsa-ucm-conf/pull/414.patch Repository URL: https://github.com/alsa-project/alsa-ucm-conf

1 0

[PATCH] ALSA: hda: cs35l56: Fix lifetime of cs_dsp instance
by Richard Fitzgerald 08 May '24

08 May '24

The cs_dsp instance is initialized in the driver probe() so it should be freed in the driver remove(). Also fix a missing call to cs_dsp_remove() in the error path of cs35l56_hda_common_probe(). The call to cs_dsp_remove() was being done in the component unbind callback cs35l56_hda_unbind(). This meant that if the driver was unbound and then re-bound it would be using an uninitialized cs_dsp instance. It is best to initialize the cs_dsp instance in probe() so that it can return an error if it fails. The component binding API doesn't have any error handling so there's no way to handle a failure if cs_dsp was initialized in the bind. Signed-off-by: Richard Fitzgerald <rf(a)opensource.cirrus.com> Fixes: 73cfbfa9caea ("ALSA: hda/cs35l56: Add driver for Cirrus Logic CS35L56 amplifier") --- sound/pci/hda/cs35l56_hda.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/sound/pci/hda/cs35l56_hda.c b/sound/pci/hda/cs35l56_hda.c index 558c1f38fe97..11b0570ff56d 100644 --- a/sound/pci/hda/cs35l56_hda.c +++ b/sound/pci/hda/cs35l56_hda.c @@ -732,8 +732,6 @@ static void cs35l56_hda_unbind(struct device *dev, struct device *master, void * if (cs35l56->base.fw_patched) cs_dsp_power_down(&cs35l56->cs_dsp); - cs_dsp_remove(&cs35l56->cs_dsp); - if (comps[cs35l56->index].dev == dev) memset(&comps[cs35l56->index], 0, sizeof(*comps)); @@ -1035,7 +1033,7 @@ int cs35l56_hda_common_probe(struct cs35l56_hda *cs35l56, int hid, int id) ARRAY_SIZE(cs35l56_hda_dai_config)); ret = cs35l56_force_sync_asp1_registers_from_cache(&cs35l56->base); if (ret) - goto err; + goto dsp_err; /* * By default only enable one ASP1TXn, where n=amplifier index, @@ -1061,6 +1059,8 @@ int cs35l56_hda_common_probe(struct cs35l56_hda *cs35l56, int hid, int id) pm_err: pm_runtime_disable(cs35l56->base.dev); +dsp_err: + cs_dsp_remove(&cs35l56->cs_dsp); err: gpiod_set_value_cansleep(cs35l56->base.reset_gpio, 0); @@ -1078,6 +1078,8 @@ void cs35l56_hda_remove(struct device *dev) component_del(cs35l56->base.dev, &cs35l56_hda_comp_ops); + cs_dsp_remove(&cs35l56->cs_dsp); + kfree(cs35l56->system_name); pm_runtime_put_noidle(cs35l56->base.dev); -- 2.39.2

2 1

[PATCH] ALSA: hda: hda_component: Initialize shared data during bind callback
by Richard Fitzgerald 08 May '24

08 May '24

Move the initialization of the shared struct hda_component array into hda_component_manager_bind(). The purpose of the manager bind() callback is to allow it to perform initialization before binding in the component drivers. This is the correct place to initialize the shared data. The original implementation initialized the shared data in hda_component_manager_init(). This is only done once during probe() of the manager driver. So if the component binding was unbound and then rebound, the shared data would not be re-initialized. Signed-off-by: Richard Fitzgerald <rf(a)opensource.cirrus.com> Fixes: fd895a74dc1d ("ALSA: hda: realtek: Move hda_component implementation to module") --- sound/pci/hda/hda_component.c | 16 +++++++++++++++- sound/pci/hda/hda_component.h | 7 ++----- sound/pci/hda/patch_realtek.c | 2 +- 3 files changed, 18 insertions(+), 7 deletions(-) diff --git a/sound/pci/hda/hda_component.c b/sound/pci/hda/hda_component.c index cd299d7d84ba..d02589014a3f 100644 --- a/sound/pci/hda/hda_component.c +++ b/sound/pci/hda/hda_component.c @@ -123,6 +123,21 @@ static int hda_comp_match_dev_name(struct device *dev, void *data) return !strcmp(d + n, tmp); } +int hda_component_manager_bind(struct hda_codec *cdc, + struct hda_component *comps, int count) +{ + int i; + + /* Init shared data */ + for (i = 0; i < count; ++i) { + memset(&comps[i], 0, sizeof(comps[i])); + comps[i].codec = cdc; + } + + return component_bind_all(hda_codec_dev(cdc), comps); +} +EXPORT_SYMBOL_NS_GPL(hda_component_manager_bind, SND_HDA_SCODEC_COMPONENT); + int hda_component_manager_init(struct hda_codec *cdc, struct hda_component *comps, int count, const char *bus, const char *hid, @@ -143,7 +158,6 @@ int hda_component_manager_init(struct hda_codec *cdc, sm->hid = hid; sm->match_str = match_str; sm->index = i; - comps[i].codec = cdc; component_match_add(dev, &match, hda_comp_match_dev_name, sm); } diff --git a/sound/pci/hda/hda_component.h b/sound/pci/hda/hda_component.h index c80a66691b5d..c70b3de68ab2 100644 --- a/sound/pci/hda/hda_component.h +++ b/sound/pci/hda/hda_component.h @@ -75,11 +75,8 @@ int hda_component_manager_init(struct hda_codec *cdc, void hda_component_manager_free(struct hda_codec *cdc, const struct component_master_ops *ops); -static inline int hda_component_manager_bind(struct hda_codec *cdc, - struct hda_component *comps) -{ - return component_bind_all(hda_codec_dev(cdc), comps); -} +int hda_component_manager_bind(struct hda_codec *cdc, + struct hda_component *comps, int count); static inline void hda_component_manager_unbind(struct hda_codec *cdc, struct hda_component *comps) diff --git a/sound/pci/hda/patch_realtek.c b/sound/pci/hda/patch_realtek.c index 8c157158f1cb..8c39bbdc6c5e 100644 --- a/sound/pci/hda/patch_realtek.c +++ b/sound/pci/hda/patch_realtek.c @@ -6785,7 +6785,7 @@ static int comp_bind(struct device *dev) struct alc_spec *spec = cdc->spec; int ret; - ret = hda_component_manager_bind(cdc, spec->comps); + ret = hda_component_manager_bind(cdc, spec->comps, ARRAY_SIZE(spec->comps)); if (ret) return ret; -- 2.39.2

2 1

[PATCH] ALSA: hda/cs_dsp_ctl: Use private_free for control cleanup
by Richard Fitzgerald 08 May '24

08 May '24

Use the control private_free callback to free the associated data block. This ensures that the memory won't leak, whatever way the control gets destroyed. The original implementation didn't actually remove the ALSA controls in hda_cs_dsp_control_remove(). It only freed the internal tracking structure. This meant it was possible to remove/unload the amp driver while leaving its ALSA controls still present in the soundcard. Obviously attempting to access them could cause segfaults or at least dereferencing stale pointers. Signed-off-by: Richard Fitzgerald <rf(a)opensource.cirrus.com> Fixes: 3233b978af23 ("ALSA: hda: hda_cs_dsp_ctl: Add Library to support CS_DSP ALSA controls") --- This replaces my earlier "keep it simple" attempt at cleanup. I decided that implementing private_free is the only sensible way to fix the cleanup. --- sound/pci/hda/hda_cs_dsp_ctl.c | 47 ++++++++++++++++++++++------------ 1 file changed, 31 insertions(+), 16 deletions(-) diff --git a/sound/pci/hda/hda_cs_dsp_ctl.c b/sound/pci/hda/hda_cs_dsp_ctl.c index 463ca06036bf..9db45d7c17e5 100644 --- a/sound/pci/hda/hda_cs_dsp_ctl.c +++ b/sound/pci/hda/hda_cs_dsp_ctl.c @@ -8,6 +8,7 @@ #include <linux/module.h> #include <sound/soc.h> +#include <linux/cleanup.h> #include <linux/firmware/cirrus/cs_dsp.h> #include <linux/firmware/cirrus/wmfw.h> #include "hda_cs_dsp_ctl.h" @@ -97,11 +98,23 @@ static unsigned int wmfw_convert_flags(unsigned int in) return out; } -static void hda_cs_dsp_add_kcontrol(struct hda_cs_dsp_coeff_ctl *ctl, const char *name) +static void hda_cs_dsp_free_kcontrol(struct snd_kcontrol *kctl) { + struct hda_cs_dsp_coeff_ctl *ctl = (struct hda_cs_dsp_coeff_ctl *)snd_kcontrol_chip(kctl); struct cs_dsp_coeff_ctl *cs_ctl = ctl->cs_ctl; + + /* NULL priv to prevent a double-free in hda_cs_dsp_control_remove() */ + cs_ctl->priv = NULL; + kfree(ctl); +} + +static void hda_cs_dsp_add_kcontrol(struct cs_dsp_coeff_ctl *cs_ctl, + const struct hda_cs_dsp_ctl_info *info, + const char *name) +{ struct snd_kcontrol_new kcontrol = {0}; struct snd_kcontrol *kctl; + struct hda_cs_dsp_coeff_ctl *ctl __free(kfree) = NULL; int ret = 0; if (cs_ctl->len > ADSP_MAX_STD_CTRL_SIZE) { @@ -110,6 +123,13 @@ static void hda_cs_dsp_add_kcontrol(struct hda_cs_dsp_coeff_ctl *ctl, const char return; } + ctl = kzalloc(sizeof(*ctl), GFP_KERNEL); + if (!ctl) + return; + + ctl->cs_ctl = cs_ctl; + ctl->card = info->card; + kcontrol.name = name; kcontrol.info = hda_cs_dsp_coeff_info; kcontrol.iface = SNDRV_CTL_ELEM_IFACE_MIXER; @@ -117,20 +137,22 @@ static void hda_cs_dsp_add_kcontrol(struct hda_cs_dsp_coeff_ctl *ctl, const char kcontrol.get = hda_cs_dsp_coeff_get; kcontrol.put = hda_cs_dsp_coeff_put; - /* Save ctl inside private_data, ctl is owned by cs_dsp, - * and will be freed when cs_dsp removes the control */ kctl = snd_ctl_new1(&kcontrol, (void *)ctl); if (!kctl) return; - ret = snd_ctl_add(ctl->card, kctl); + kctl->private_free = hda_cs_dsp_free_kcontrol; + ctl->kctl = kctl; + + /* snd_ctl_add() calls our private_free on error, which will kfree(ctl) */ + cs_ctl->priv = no_free_ptr(ctl); + ret = snd_ctl_add(info->card, kctl); if (ret) { dev_err(cs_ctl->dsp->dev, "Failed to add KControl %s = %d\n", kcontrol.name, ret); return; } dev_dbg(cs_ctl->dsp->dev, "Added KControl: %s\n", kcontrol.name); - ctl->kctl = kctl; } static void hda_cs_dsp_control_add(struct cs_dsp_coeff_ctl *cs_ctl, @@ -138,7 +160,6 @@ static void hda_cs_dsp_control_add(struct cs_dsp_coeff_ctl *cs_ctl, { struct cs_dsp *cs_dsp = cs_ctl->dsp; char name[SNDRV_CTL_ELEM_ID_NAME_MAXLEN]; - struct hda_cs_dsp_coeff_ctl *ctl; const char *region_name; int ret; @@ -163,15 +184,7 @@ static void hda_cs_dsp_control_add(struct cs_dsp_coeff_ctl *cs_ctl, " %.*s", cs_ctl->subname_len - skip, cs_ctl->subname + skip); } - ctl = kzalloc(sizeof(*ctl), GFP_KERNEL); - if (!ctl) - return; - - ctl->cs_ctl = cs_ctl; - ctl->card = info->card; - cs_ctl->priv = ctl; - - hda_cs_dsp_add_kcontrol(ctl, name); + hda_cs_dsp_add_kcontrol(cs_ctl, info, name); } void hda_cs_dsp_add_controls(struct cs_dsp *dsp, const struct hda_cs_dsp_ctl_info *info) @@ -203,7 +216,9 @@ void hda_cs_dsp_control_remove(struct cs_dsp_coeff_ctl *cs_ctl) { struct hda_cs_dsp_coeff_ctl *ctl = cs_ctl->priv; - kfree(ctl); + /* ctl and kctl may already have been removed by ALSA private_free */ + if (ctl && ctl->kctl) + snd_ctl_remove(ctl->card, ctl->kctl); } EXPORT_SYMBOL_NS_GPL(hda_cs_dsp_control_remove, SND_HDA_CS_DSP_CONTROLS); -- 2.39.2

2 1

[PATCH v1 1/1] ASoC: atmel: tse850-pcm5142: Remove unused of_gpio.h
by Andy Shevchenko 08 May '24

08 May '24

of_gpio.h is deprecated and subject to remove. The driver doesn't use it, simply remove the unused header. Signed-off-by: Andy Shevchenko <andriy.shevchenko(a)linux.intel.com> --- sound/soc/atmel/tse850-pcm5142.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/sound/soc/atmel/tse850-pcm5142.c b/sound/soc/atmel/tse850-pcm5142.c index 611da23325d3..5d208e0b4b90 100644 --- a/sound/soc/atmel/tse850-pcm5142.c +++ b/sound/soc/atmel/tse850-pcm5142.c @@ -35,10 +35,9 @@ // of the (filtered) output from the PCM5142 codec. #include <linux/clk.h> -#include <linux/gpio.h> +#include <linux/gpio/consumer.h> #include <linux/module.h> #include <linux/of.h> -#include <linux/of_gpio.h> #include <linux/regulator/consumer.h> #include <sound/soc.h> -- 2.43.0.rc1.1336.g36b5255a03ac

2 1