Skip to contentSkip to conversation

ALSA: yet a few more fixes for AI bug reports

PATCH Superseded Takashi Iwai 2 participants 14 messages Oct 8 – Oct 9

Subsystems (4)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

cover ALSA: yet a few more fixes for AI bug reports

Hi,

here are a few more fixes I've worked quickly for issues Sashiko and other AI reported.

Takashi

===

Takashi Iwai (11):
ALSA: hda: intel: Cancel delayed work at shutdown, too
ALSA: hda: Disable unsol event handling at error and shutdown paths
ALSA: hda: Add lock around codec->registered flag manipulations
ALSA: hda: Add NULL check for the driver pointer at unsol event work
ALSA: caiaq: Register card at the final step
ALSA: caiaq: Fix races at MIDI URB and trigger accesses
ALSA: usb: us16x08: Fix racy accesses of mixer elements
ASoC: fsl_asrc_m2m: Fix bogus compress task pointer assignments
ALSA: line6: Reject too small max packet sizes
ALSA: line6: Fix potential OOB write in line6_capture_copy()
ALSA: line6: Fix handling of zero-length capture packets

 sound/hda/common/bind.c       |  2 ++
 sound/hda/common/codec.c      |  5 +++++
 sound/hda/controllers/intel.c |  7 ++++++-
 sound/hda/core/bus.c          |  2 +-
 sound/soc/fsl/fsl_asrc_m2m.c  |  2 ++
 sound/usb/caiaq/device.c      | 13 +++++++------
 sound/usb/caiaq/device.h      |  3 ++-
 sound/usb/caiaq/midi.c        |  2 ++
 sound/usb/line6/capture.c     |  3 ++-
 sound/usb/line6/pcm.c         | 10 ++++++++--
 sound/usb/line6/playback.c    |  4 ++--
 sound/usb/mixer_us16x08.c     | 16 ++++++++++++++++
 12 files changed, 55 insertions(+), 14 deletions(-)
Signature · 1 line
2.55.0
#1 lore raw

PATCH 1/11 ALSA: hda: intel: Cancel delayed work at shutdown, too

The shutdown procedure needs to cancel the delayed probe work in case where the probe work stalls or is slow but still possible to reactivated.

Fixes c0f1886de7e1 ("ALSA: hda: intel: Allow repeatedly probing on codec configuration errors")
---
 sound/hda/controllers/intel.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
sound/hda/controllers/intel.c +6 -1
index 88b0207fde9c..2b5b5d555f2c 100644
@@ -2484,12 +2484,17 @@ static void azx_remove(struct pci_dev *pci)
 static void azx_shutdown(struct pci_dev *pci)
 {
 	struct snd_card *card = pci_get_drvdata(pci);
+	struct hda_intel *hda;
 	struct azx *chip;
 
 	if (!card)
 		return;
 	chip = card->private_data;
-	if (chip && chip->running)
+	if (!chip)
+		return;
+	hda = container_of(chip, struct hda_intel, chip);
+	cancel_delayed_work_sync(&hda->probe_work);
+	if (chip->running)
 		__azx_shutdown_chip(chip, true);
 }
 
Signature · 1 line
2.55.0
#2 lore raw

PATCH 2/11 ALSA: hda: Disable unsol event handling at error and shutdown paths

In reply to Takashi Iwai

When the HD-audio controller driver probe fails, it still leaves the unsolicited event handling and jackpoll work active, hence if they are pending, they might fire up later after the resource gets released, which may lead to a UAF. A similar problem may be seen at shutdown, too.

Add the recently added helper to disable unsol events and the cancel of jackpoll_work at the appropriate places.

Fixes c3ec8ac82105 ("ASoC: hdac_hda: fix memleak on module unload")
---
 sound/hda/common/bind.c  | 2 ++
 sound/hda/common/codec.c | 1 +
 2 files changed, 3 insertions(+)
sound/hda/common/bind.c +2 -0
index 4772ca154a29..f2a498c78891 100644
@@ -147,6 +147,8 @@ static int hda_codec_driver_probe(struct device *dev)
 	module_put(owner);
 
  error:
+	snd_hdac_device_disable_unsol(&codec->core);
+	cancel_delayed_work_sync(&codec->jackpoll_work);
 	snd_hda_codec_cleanup_for_unbind(codec);
 	codec->preset = NULL;
 	return err;
sound/hda/common/codec.c +1 -0
index c99fe61c29db..94da4f3a21ed 100644
@@ -3039,6 +3039,7 @@ void snd_hda_codec_shutdown(struct hda_codec *codec)
 
 	codec->jackpoll_interval = 0; /* don't poll any longer */
 	cancel_delayed_work_sync(&codec->jackpoll_work);
+	snd_hdac_device_disable_unsol(&codec->core);
 	list_for_each_entry(cpcm, &codec->pcm_list_head, list)
 		snd_pcm_suspend_all(cpcm->pcm);
 
Signature · 1 line
2.55.0
#3 lore raw

PATCH 3/11 ALSA: hda: Add lock around codec->registered flag manipulations

In reply to Takashi Iwai

Currently the places setting or clearing codec->registered flag have no locking, while its reference in the unsol handling work snd_hdac_bus_process_unsol_events() takes the bus->reg_lock lock. Let's add the same bus->reg_lock at the places where codec->registered is set and cleared for avoiding the races.

---
 sound/hda/common/codec.c | 4 ++++
 1 file changed, 4 insertions(+)
sound/hda/common/codec.c +4 -0
index 94da4f3a21ed..7fbced571282 100644
@@ -761,7 +761,9 @@ void snd_hda_codec_cleanup_for_unbind(struct hda_codec *codec)
 		/* pm_runtime_put() is called in snd_hdac_device_exit() */
 		pm_runtime_get_noresume(hda_codec_dev(codec));
 		pm_runtime_disable(hda_codec_dev(codec));
+		spin_lock_irq(&bus->reg_lock);
 		codec->core.registered = false;
+		spin_unlock_irq(&bus->reg_lock);
 	}
 
 	snd_hda_codec_disconnect_pcms(codec);
@@ -815,7 +817,9 @@ void snd_hda_codec_register(struct hda_codec *codec)
 		pm_runtime_enable(hda_codec_dev(codec));
 		/* it was powered up in snd_hda_codec_new(), now all done */
 		snd_hda_power_down(codec);
+		spin_lock_irq(&bus->reg_lock);
 		codec->core.registered = true;
+		spin_unlock_irq(&bus->reg_lock);
 	}
 }
 EXPORT_SYMBOL_GPL(snd_hda_codec_register);
Signature · 1 line
2.55.0
#4 lore raw

Re: [PATCH 03/11] ALSA: hda: Add lock around codec->registered flag manipulations

kernel test robot noticed the following build errors:

Hi Takashi,

kernel test robot noticed the following build errors:

[auto build test ERROR on tiwai-sound/for-next]
[also build test ERROR on next-20261008]
[cannot apply to tiwai-sound/for-linus broonie-sound/for-next linus/master v7.3-rc6] [If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Takashi-Iwai/ALSA-hda-intel-Cancel-delayed-work-at-shutdown-too/20261008-212541
base:   https://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound.git for-next
patch link:    https://lore.kernel.org/r/20261008192553.300025-4-tiwai%40suse.de
patch subject: [PATCH 03/11] ALSA: hda: Add lock around codec->registered flag manipulations
config: x86_64-defconfig (https://download.01.org/0day-ci/archive/20261009/[email protected]/config)
compiler: gcc-14 (Debian 14.2.0-19) 14.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20261009/[email protected]/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/

All errors (new ones prefixed by >>):

sound/hda/common/codec.c: In function 'snd_hda_codec_cleanup_for_unbind':

>> sound/hda/common/codec.c:764:32: error: 'bus' undeclared (first use in this function)
     764 |                 spin_lock_irq(&bus->reg_lock);
         |                                ^~~
   sound/hda/common/codec.c:764:32: note: each undeclared identifier is reported only once for each function it appears in
   sound/hda/common/codec.c: In function 'snd_hda_codec_register':
   sound/hda/common/codec.c:820:32: error: 'bus' undeclared (first use in this function)
     820 |                 spin_lock_irq(&bus->reg_lock);
         |                                ^~~

vim +/bus +764 sound/hda/common/codec.c

753
754 /**
755 * snd_hda_codec_cleanup_for_unbind - Prepare codec for removal
756 * @codec: codec device to cleanup
757 */
758 void snd_hda_codec_cleanup_for_unbind(struct hda_codec *codec)
759 {
760 if (codec->core.registered) {
761 /* pm_runtime_put() is called in snd_hdac_device_exit() */
762 pm_runtime_get_noresume(hda_codec_dev(codec));
763 pm_runtime_disable(hda_codec_dev(codec));

 > 764			spin_lock_irq(&bus->reg_lock);

765 codec->core.registered = false;
766 spin_unlock_irq(&bus->reg_lock);
767 }
768
769 snd_hda_codec_disconnect_pcms(codec);
770 cancel_delayed_work_sync(&codec->jackpoll_work);
771 if (!codec->in_freeing)
772 snd_hda_ctls_clear(codec);
773 codec_release_pcms(codec);
774 snd_hda_detach_beep_device(codec);
775 snd_hda_jack_tbl_clear(codec);
776 codec->proc_widget_hook = NULL;
777 codec->spec = NULL;
778
779 /* free only driver_pins so that init_pins + user_pins are restored */
780 snd_array_free(&codec->driver_pins);
781 snd_array_free(&codec->cvt_setups);
782 snd_array_free(&codec->spdif_out);
783 snd_array_free(&codec->verbs);
784 codec->follower_dig_outs = NULL;
785 codec->spdif_status_reset = 0;
786 snd_array_free(&codec->mixers);
787 snd_array_free(&codec->nids);
788 remove_conn_list(codec);
789 snd_hdac_regmap_exit(&codec->core);
790 codec->configured = 0;
791 snd_refcount_init(&codec->pcm_ref); /* reset refcount */
792 }
793 EXPORT_SYMBOL_GPL(snd_hda_codec_cleanup_for_unbind);
794

--
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki

#5 lore raw

Re: [PATCH 03/11] ALSA: hda: Add lock around codec->registered flag manipulations

In reply to Takashi Iwai
kernel test robot noticed the following build errors:

Hi Takashi,

kernel test robot noticed the following build errors:

[auto build test ERROR on tiwai-sound/for-next]
[also build test ERROR on next-20261008]
[cannot apply to tiwai-sound/for-linus broonie-sound/for-next linus/master v7.3-rc6] [If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Takashi-Iwai/ALSA-hda-intel-Cancel-delayed-work-at-shutdown-too/20261008-212541
base:   https://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound.git for-next
patch link:    https://lore.kernel.org/r/20261008192553.300025-4-tiwai%40suse.de
patch subject: [PATCH 03/11] ALSA: hda: Add lock around codec->registered flag manipulations
config: i386-defconfig (https://download.01.org/0day-ci/archive/20261009/[email protected]/config)
compiler: clang version 22.1.3 (https://github.com/llvm/llvm-project e9846648fd6183ee6d8cbdb4502213fcf902a211)
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20261009/[email protected]/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/

All errors (new ones prefixed by >>):

>> sound/hda/common/codec.c:764:18: error: use of undeclared identifier 'bus'
     764 |                 spin_lock_irq(&bus->reg_lock);
         |                                ^~~
   sound/hda/common/codec.c:766:20: error: use of undeclared identifier 'bus'
     766 |                 spin_unlock_irq(&bus->reg_lock);
         |                                  ^~~
   sound/hda/common/codec.c:820:18: error: use of undeclared identifier 'bus'
     820 |                 spin_lock_irq(&bus->reg_lock);
         |                                ^~~
   sound/hda/common/codec.c:822:20: error: use of undeclared identifier 'bus'
     822 |                 spin_unlock_irq(&bus->reg_lock);
         |                                  ^~~

4 errors generated.

vim +/bus +764 sound/hda/common/codec.c

753
754 /**
755 * snd_hda_codec_cleanup_for_unbind - Prepare codec for removal
756 * @codec: codec device to cleanup
757 */
758 void snd_hda_codec_cleanup_for_unbind(struct hda_codec *codec)
759 {
760 if (codec->core.registered) {
761 /* pm_runtime_put() is called in snd_hdac_device_exit() */
762 pm_runtime_get_noresume(hda_codec_dev(codec));
763 pm_runtime_disable(hda_codec_dev(codec));

 > 764			spin_lock_irq(&bus->reg_lock);

765 codec->core.registered = false;
766 spin_unlock_irq(&bus->reg_lock);
767 }
768
769 snd_hda_codec_disconnect_pcms(codec);
770 cancel_delayed_work_sync(&codec->jackpoll_work);
771 if (!codec->in_freeing)
772 snd_hda_ctls_clear(codec);
773 codec_release_pcms(codec);
774 snd_hda_detach_beep_device(codec);
775 snd_hda_jack_tbl_clear(codec);
776 codec->proc_widget_hook = NULL;
777 codec->spec = NULL;
778
779 /* free only driver_pins so that init_pins + user_pins are restored */
780 snd_array_free(&codec->driver_pins);
781 snd_array_free(&codec->cvt_setups);
782 snd_array_free(&codec->spdif_out);
783 snd_array_free(&codec->verbs);
784 codec->follower_dig_outs = NULL;
785 codec->spdif_status_reset = 0;
786 snd_array_free(&codec->mixers);
787 snd_array_free(&codec->nids);
788 remove_conn_list(codec);
789 snd_hdac_regmap_exit(&codec->core);
790 codec->configured = 0;
791 snd_refcount_init(&codec->pcm_ref); /* reset refcount */
792 }
793 EXPORT_SYMBOL_GPL(snd_hda_codec_cleanup_for_unbind);
794

--
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki

#6 lore raw

PATCH 4/11 ALSA: hda: Add NULL check for the driver pointer at unsol event work

In reply to Takashi Iwai

snd_hdac_bus_process_unsol_events() assumes that the codec driver is always set when it's accessible from the codec table. It's true in general, but in a sheer timing, it might have been already NULLified at the driver unbinding.

As a safety, let's add a NULL check before dereferencing and calling the driver's unsol_event callback.

Fixes c637fa151259 ("ALSA: hda: Fix potential race in unsol event handler")
---
 sound/hda/core/bus.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
sound/hda/core/bus.c +1 -1
index 8d4ed9834aaa..96fd9bd221c9 100644
@@ -184,7 +184,7 @@ static void snd_hdac_bus_process_unsol_events(struct work_struct *work)
 			continue;
 		spin_unlock_irq(&bus->reg_lock);
 		drv = drv_to_hdac_driver(codec->dev.driver);
-		if (drv->unsol_event)
+		if (drv && drv->unsol_event)
 			drv->unsol_event(codec, res);
 		spin_lock_irq(&bus->reg_lock);
 	}
Signature · 1 line
2.55.0
#7 lore raw

PATCH 5/11 ALSA: caiaq: Register card at the final step

In reply to Takashi Iwai

caiaq driver calls snd_card_register() before creating the mixer elements, which is rather confusing to user-space, as the devices have been published before adding the control elements.

Swap the call order to make the card registration as the final step (as written in the comment).

Fixes 8e3cd08ed8e5 ("[ALSA] caiaq - add control API and more input features")
---
 sound/usb/caiaq/device.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)
sound/usb/caiaq/device.c +6 -6
index a596f5f763b8..c68ed16d7ac4 100644
@@ -395,6 +395,12 @@ static int setup_card(struct snd_usb_caiaqdev *cdev)
 		return ret;
 	}
 
+	ret = snd_usb_caiaq_control_init(cdev);
+	if (ret < 0) {
+		dev_err(dev, "Unable to set up control system (ret=%d)\n", ret);
+		return ret;
+	}
+
 	/* finally, register the card and all its sub-instances */
 	ret = snd_card_register(cdev->chip.card);
 	if (ret < 0) {
@@ -402,12 +408,6 @@ static int setup_card(struct snd_usb_caiaqdev *cdev)
 		return ret;
 	}
 
-	ret = snd_usb_caiaq_control_init(cdev);
-	if (ret < 0) {
-		dev_err(dev, "Unable to set up control system (ret=%d)\n", ret);
-		return ret;
-	}
-
 	return 0;
 }
 
Signature · 1 line
2.55.0
#8 lore raw

PATCH 6/11 ALSA: caiaq: Fix races at MIDI URB and trigger accesses

In reply to Takashi Iwai

URB completion and rawmidi trigger callbacks can race with each other, which may put the state inconsistent, causing double submissions, etc.

Guard both with a new spinlock for avoiding the races.

Fixes f3f80a9205da ("ALSA: caiaq - Fix Oops with MIDI")
---
 sound/usb/caiaq/device.c | 1 +
 sound/usb/caiaq/device.h | 3 ++-
 sound/usb/caiaq/midi.c   | 2 ++
 3 files changed, 5 insertions(+), 1 deletion(-)
sound/usb/caiaq/device.c +1 -0
index c68ed16d7ac4..634dc36f2ead 100644
@@ -455,6 +455,7 @@ static int create_card(struct usb_device *usb_dev,
 	cdev->chip.usb_id = USB_ID(le16_to_cpu(usb_dev->descriptor.idVendor),
 				  le16_to_cpu(usb_dev->descriptor.idProduct));
 	spin_lock_init(&cdev->spinlock);
+	spin_lock_init(&cdev->midi_lock);
 	mutex_init(&cdev->ep1_out_mutex);
 
 	*cardp = card;
sound/usb/caiaq/device.h +2 -1
index dd726ca68e80..16f65e7867d1 100644
@@ -77,7 +77,8 @@ struct snd_usb_caiaqdev {
 	unsigned char midi_out_buf[EP1_BUFSIZE];
 
 	struct caiaq_device_spec spec;
-	spinlock_t spinlock;
+	spinlock_t spinlock;	/* for PCM audio */
+	spinlock_t midi_lock;	/* midi_x_stream, midi_out_active */
 	wait_queue_head_t ep1_wait_queue;
 	wait_queue_head_t prepare_wait_queue;
 	int spec_received, audio_parm_answer;
sound/usb/caiaq/midi.c +2 -0
index 18529484c8dc..b7d7d24937a7 100644
@@ -79,6 +79,7 @@ static void snd_usb_caiaq_midi_output_trigger(struct snd_rawmidi_substream *subs
 {
 	struct snd_usb_caiaqdev *cdev = substream->rmidi->private_data;
 
+	guard(spinlock_irqsave)(&cdev->midi_lock);
 	if (up) {
 		cdev->midi_out_substream = substream;
 		if (!cdev->midi_out_active)
@@ -151,6 +152,7 @@ void snd_usb_caiaq_midi_output_done(struct urb* urb)
 {
 	struct snd_usb_caiaqdev *cdev = urb->context;
 
+	guard(spinlock_irqsave)(&cdev->midi_lock);
 	cdev->midi_out_active = 0;
 	if (urb->status != 0)
 		return;
Signature · 1 line
2.55.0
#9 lore raw

PATCH 7/11 ALSA: usb: us16x08: Fix racy accesses of mixer elements

In reply to Takashi Iwai

All get and put callbacks for us16x08 mixer have no proper protection, hence the concurrent accesses to multiple elements may face data races, resulting in unexpected values.

Put the new mixer->lock mutex for protecting the concurrent accesses.

Fixes d2bb390a2081 ("ALSA: usb-audio: Tascam US-16x08 DSP mixer quirk")
---
 sound/usb/mixer_us16x08.c | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)
sound/usb/mixer_us16x08.c +16 -0
index 8e5ccd3282a7..b447c4f7a161 100644
@@ -189,6 +189,7 @@ static int snd_us16x08_route_get(struct snd_kcontrol *kcontrol,
 	struct usb_mixer_elem_info *elem = snd_kcontrol_chip(kcontrol);
 	int index = ucontrol->id.index;
 
+	guard(mutex)(&elem->head.mixer->lock);
 	/* route has no bias */
 	ucontrol->value.enumerated.item[0] = elem->cache_val[index];
 
@@ -204,6 +205,7 @@ static int snd_us16x08_route_put(struct snd_kcontrol *kcontrol,
 	char buf[sizeof(route_msg)];
 	int val, val_org, err;
 
+	guard(mutex)(&elem->head.mixer->lock);
 	/*  get the new value (no bias for routes) */
 	val = ucontrol->value.enumerated.item[0];
 
@@ -258,6 +260,7 @@ static int snd_us16x08_master_get(struct snd_kcontrol *kcontrol,
 	struct usb_mixer_elem_info *elem = snd_kcontrol_chip(kcontrol);
 	int index = ucontrol->id.index;
 
+	guard(mutex)(&elem->head.mixer->lock);
 	ucontrol->value.integer.value[0] = elem->cache_val[index];
 
 	return 0;
@@ -272,6 +275,7 @@ static int snd_us16x08_master_put(struct snd_kcontrol *kcontrol,
 	int val, err;
 	int index = ucontrol->id.index;
 
+	guard(mutex)(&elem->head.mixer->lock);
 	/* new control value incl. bias*/
 	val = ucontrol->value.integer.value[0];
 
@@ -310,6 +314,7 @@ static int snd_us16x08_bus_put(struct snd_kcontrol *kcontrol,
 
 	val = ucontrol->value.integer.value[0];
 
+	guard(mutex)(&elem->head.mixer->lock);
 	/* prepare the message buffer from template */
 	switch (elem->head.id) {
 	case SND_US16X08_ID_BYPASS:
@@ -346,6 +351,7 @@ static int snd_us16x08_bus_get(struct snd_kcontrol *kcontrol,
 {
 	struct usb_mixer_elem_info *elem = snd_kcontrol_chip(kcontrol);
 
+	guard(mutex)(&elem->head.mixer->lock);
 	switch (elem->head.id) {
 	case SND_US16X08_ID_BUSS_OUT:
 		ucontrol->value.integer.value[0] = elem->cache_val[0];
@@ -368,6 +374,7 @@ static int snd_us16x08_channel_get(struct snd_kcontrol *kcontrol,
 	struct usb_mixer_elem_info *elem = snd_kcontrol_chip(kcontrol);
 	int index = ucontrol->id.index;
 
+	guard(mutex)(&elem->head.mixer->lock);
 	ucontrol->value.integer.value[0] = elem->cache_val[index];
 
 	return 0;
@@ -382,6 +389,7 @@ static int snd_us16x08_channel_put(struct snd_kcontrol *kcontrol,
 	int val, err;
 	int index = ucontrol->id.index;
 
+	guard(mutex)(&elem->head.mixer->lock);
 	val = ucontrol->value.integer.value[0];
 
 	/* sanity check */
@@ -428,6 +436,7 @@ static int snd_us16x08_comp_get(struct snd_kcontrol *kcontrol,
 	int index = ucontrol->id.index;
 	int val_idx = COMP_STORE_IDX(elem->head.id);
 
+	guard(mutex)(&elem->head.mixer->lock);
 	ucontrol->value.integer.value[0] = store->val[val_idx][index];
 
 	return 0;
@@ -445,6 +454,7 @@ static int snd_us16x08_comp_put(struct snd_kcontrol *kcontrol,
 	int threshold, ratio, attack, release, gain, switch_on;
 	int err;
 
+	guard(mutex)(&elem->head.mixer->lock);
 	val = ucontrol->value.integer.value[0];
 
 	/* sanity check */
@@ -521,6 +531,7 @@ static int snd_us16x08_eqswitch_get(struct snd_kcontrol *kcontrol,
 	struct snd_us16x08_eq_store *store = elem->private_data;
 	int index = ucontrol->id.index;
 
+	guard(mutex)(&elem->head.mixer->lock);
 	/* get low switch from cache is enough, cause all bands are together */
 	val = store->val[EQ_STORE_BAND_IDX(elem->head.id)]
 		[EQ_STORE_PARAM_IDX(elem->head.id)][index];
@@ -540,6 +551,7 @@ static int snd_us16x08_eqswitch_put(struct snd_kcontrol *kcontrol,
 	int val, err = 0;
 	int b_idx;
 
+	guard(mutex)(&elem->head.mixer->lock);
 	/* new control value incl. bias*/
 	val = ucontrol->value.integer.value[0] + SND_US16X08_KCBIAS(kcontrol);
 
@@ -582,6 +594,7 @@ static int snd_us16x08_eq_get(struct snd_kcontrol *kcontrol,
 	int b_idx = EQ_STORE_BAND_IDX(elem->head.id) - 1;
 	int p_idx = EQ_STORE_PARAM_IDX(elem->head.id);
 
+	guard(mutex)(&elem->head.mixer->lock);
 	val = store->val[b_idx][p_idx][index];
 
 	ucontrol->value.integer.value[0] = val;
@@ -601,6 +614,7 @@ static int snd_us16x08_eq_put(struct snd_kcontrol *kcontrol,
 	int b_idx = EQ_STORE_BAND_IDX(elem->head.id) - 1;
 	int p_idx = EQ_STORE_PARAM_IDX(elem->head.id);
 
+	guard(mutex)(&elem->head.mixer->lock);
 	val = ucontrol->value.integer.value[0];
 
 	/* sanity check */
@@ -729,6 +743,7 @@ static int snd_us16x08_meter_get(struct snd_kcontrol *kcontrol,
 	struct snd_us16x08_meter_store *store = elem->private_data;
 	u8 meter_urb[64] = {0};
 
+	guard(mutex)(&elem->head.mixer->lock);
 	switch (kcontrol->private_value) {
 	case 0: {
 		char tmp[sizeof(mix_init_msg1)];
@@ -788,6 +803,7 @@ static int snd_us16x08_meter_put(struct snd_kcontrol *kcontrol,
 	struct snd_us16x08_meter_store *store = elem->private_data;
 	int val;
 
+	guard(mutex)(&elem->head.mixer->lock);
 	val = ucontrol->value.integer.value[0];
 
 	/* sanity check */
Signature · 1 line
2.55.0
#10 lore raw

PATCH 8/11 ASoC: fsl_asrc_m2m: Fix bogus compress task pointer assignments

In reply to Takashi Iwai

When a creation of input or output compress-offload task fails in fsl-asrc driver, the task->input or task->output pointer is left with ERR_PTR(), which is non-NULL. Then it's handled in the compress offload core and it tries to release via dma_buf_put(), resulting in an access to a wrong address.

Clear the bogus pointers properly before returning an error.

Fixes 24a01710f627 ("ASoC: fsl_asrc_m2m: Add memory to memory function")
---
 sound/soc/fsl/fsl_asrc_m2m.c | 2 ++
 1 file changed, 2 insertions(+)
sound/soc/fsl/fsl_asrc_m2m.c +2 -0
index 4bc40f328f58..42b0e21f281c 100644
@@ -475,6 +475,7 @@ static int fsl_asrc_m2m_comp_task_create(struct snd_compr_stream *stream,
 	task->input = dma_buf_export(&exp_info_in);
 	if (IS_ERR(task->input)) {
 		ret = PTR_ERR(task->input);
+		task->input = NULL;
 		return ret;
 	}
 
@@ -485,6 +486,7 @@ static int fsl_asrc_m2m_comp_task_create(struct snd_compr_stream *stream,
 	task->output = dma_buf_export(&exp_info_out);
 	if (IS_ERR(task->output)) {
 		ret = PTR_ERR(task->output);
+		task->output = NULL;
 		return ret;
 	}
 
Signature · 1 line
2.55.0
#11 lore raw

PATCH 9/11 ALSA: line6: Reject too small max packet sizes

In reply to Takashi Iwai

Although the LINE6 driver has a sanity check for the given max packet sizes, it still has an implicit requirement of the minimal size being bytes-per-frame; e.g. the impulse test signal assuming the fixed size, and when a too small size is specified by a malformed USB descriptor, this may lead to an OOB access.

Change the sanity check conditions to reject too small max packet sizes for avoiding the scenario above.

Fixes 3450121997ce ("ALSA: line6: Fix write on zero-sized buffer")
---
 sound/usb/line6/pcm.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
sound/usb/line6/pcm.c +5 -1
index 2932eaf157f4..998869dc4613 100644
@@ -554,7 +554,11 @@ int line6_init_pcm(struct usb_line6 *line6,
 	line6pcm->max_packet_size_out =
 		usb_maxpacket(line6->usbdev,
 			usb_sndisocpipe(line6->usbdev, ep_write));
-	if (!line6pcm->max_packet_size_in || !line6pcm->max_packet_size_out) {
+	/* reject max packet sizes smaller than bytes-per-frame;
+	 * (the magic number 6 is taken from the playback case)
+	 */
+	if (line6pcm->max_packet_size_in < 6 ||
+	    line6pcm->max_packet_size_out < 6) {
 		dev_err(line6pcm->line6->ifcdev,
 			"cannot get proper max packet size\n");
 		return -EINVAL;
Signature · 1 line
2.55.0
#12 lore raw

PATCH 10/11 ALSA: line6: Fix potential OOB write in line6_capture_copy()

In reply to Takashi Iwai

line6_capture_copy() copies the data for the original byte size, but the buffer overwrap is checked against the frame size. Because of it, when the data size isn't aligned in frames, the remaining bytes might be still copied above the buffer size.

For avoiding the potential OOB write, correct the copied data size in line6_capture_copy() to be aligned with frames.

Fixes 1027f476f507 ("staging: line6: sync with upstream")
---
 sound/usb/line6/capture.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
sound/usb/line6/capture.c +2 -1
index 6dbaf30232f9..3e1f40d6b481 100644
@@ -117,7 +117,8 @@ void line6_capture_copy(struct snd_line6_pcm *line6pcm, char *fbuf, int fsize)
 	} else {
 		/* copy single chunk */
 		memcpy(runtime->dma_area +
-		       line6pcm->in.pos_done * bytes_per_frame, fbuf, fsize);
+		       line6pcm->in.pos_done * bytes_per_frame, fbuf,
+		       frames * bytes_per_frame);
 	}
 
 	line6pcm->in.pos_done += frames;
Signature · 1 line
2.55.0
#13 lore raw

PATCH 11/11 ALSA: line6: Fix handling of zero-length capture packets

In reply to Takashi Iwai

The LINE6 playback engine assumes that line6pcm->prev_fsize=0 indicates that there was no input packet to be processed, and synthesizes the frames. But, when a capture stream receives a zero-length (or a very small size) packet, it sets 0 to prev_fsize, while playback engine still believes it's for synthesis, hence it tries to copy the larger data than the actual input data.

As prev_fsize=0 can't be a good indication for "no input", change the meaning slightly: now prev_fsize=-1 means no input, while prev_fsize=0 means it's a zero-length packet.

Fixes 1027f476f507 ("staging: line6: sync with upstream")
---
 sound/usb/line6/pcm.c      | 4 +++-
 sound/usb/line6/playback.c | 4 ++--
 2 files changed, 5 insertions(+), 3 deletions(-)
sound/usb/line6/pcm.c +3 -1
index 998869dc4613..e2b99e7a08e0 100644
@@ -217,7 +217,7 @@ static void line6_stream_stop(struct snd_line6_pcm *line6pcm, int direction,
 	if (direction == SNDRV_PCM_STREAM_CAPTURE) {
 		guard(spinlock_irqsave)(&pstr->lock);
 		line6pcm->prev_fbuf = NULL;
-		line6pcm->prev_fsize = 0;
+		line6pcm->prev_fsize = -1;
 	}
 }
 
@@ -538,6 +538,8 @@ int line6_init_pcm(struct usb_line6 *line6,
 	line6pcm->volume_playback[0] = line6pcm->volume_playback[1] = 255;
 	line6pcm->volume_monitor = 255;
 	line6pcm->line6 = line6;
+	line6pcm->prev_fbuf = NULL;
+	line6pcm->prev_fsize = -1;
 
 	spin_lock_init(&line6pcm->out.lock);
 	spin_lock_init(&line6pcm->in.lock);
sound/usb/line6/playback.c +2 -2
index aa6ddf8746a0..6682ebaac646 100644
@@ -171,7 +171,7 @@ static int submit_audio_out_urb(struct snd_line6_pcm *line6pcm)
 		    &urb_out->iso_frame_desc[i];
 
 		fsize = line6pcm->prev_fsize;
-		if (fsize == 0) {
+		if (fsize == -1) {
 			int n;
 
 			line6pcm->out.count += frame_increment;
@@ -271,7 +271,7 @@ static int submit_audio_out_urb(struct snd_line6_pcm *line6pcm)
 						   bytes_per_frame);
 		}
 		line6pcm->prev_fbuf = NULL;
-		line6pcm->prev_fsize = 0;
+		line6pcm->prev_fsize = -1;
 	}
 	spin_unlock(&line6pcm->in.lock);
 
Signature · 1 line
2.55.0