kernel/linux.git/sound/usb/stream.c, branch v6.12.91

ALSA: usb-audio: Fix potential leak of pd at parsing UAC3 streams

2026-05-23T11:04:54+00:00

[ Upstream commit c39f0bc03f84ba64c9144c95714df1dc36150f6d ] At parsing UAC3 streams, we allocate a PD object at each time, and either assign or free it. But there is a case where the PD object may be leaked; namely, in __snd_usb_parse_audio_interface() loop, when an audioformat shares the same endpoint with others, it's put to a link and returns from snd_usb_add_audio_stream(), but the PD is forgotten afterwards. Overall, the treatment of PD object in the parser code is a bit flaky, and we should be more careful about the object ownership. This patch tries to fix the above case and improve the code a bit. The pd object is now managed with the auto-cleanup in the loop, and the ownership is updated when the pd object gets assigned to the stream, which guarantees the release of the leftover object. Fixes: 7edf3b5e6a45 ("ALSA: usb-audio: AudioStreaming Power Domain parsing") Link: https://patch.msgid.link/20260427151508.12544-1-tiwai@suse.de Signed-off-by: Takashi Iwai Signed-off-by: Sasha Levin

ALSA: usb-audio: Fix UAC3 cluster descriptor size check

2026-05-14T13:29:19+00:00

commit 26265dd69da32d88a88d21987853cec899d9e21f upstream. The UAC3 cluster descriptor length check in snd_usb_get_audioformat_uac3()was added to make sure that the buffer is large enough for a struct uac3_cluster_header_descriptor before the returned data is cast and used. However, the check uses sizeof(cluster), where cluster is a pointer, not the size of the descriptor header. This makes the validation depend on the architecture pointer size and does not match the intended object size. Check against sizeof(*cluster) instead. Fixes: fb4e2a6e8f28 ("ALSA: usb-audio: Fix out-of-bounds read in snd_usb_get_audioformat_uac3()") Cc: stable@vger.kernel.org Signed-off-by: Cássio Gabriel Link: https://patch.msgid.link/20260424-alsa-usb-uac3-cluster-size-v1-1-99a5808898a3@gmail.com Signed-off-by: Takashi Iwai Signed-off-by: Greg Kroah-Hartman

ALSA: usb-audio: Avoid potential endless loop in convert_chmap_v3()

2026-05-14T13:29:19+00:00

commit 6e7247d8f5fefeceb0bb9cc80a5388a636b219cd upstream. The convert_chmap_v3() has a loop with its increment size of cs_desc->wLength, but we forgot to validate cs_desc->wLength itself, which may lead to potential endless loop by a malformed descriptor. Add a proper size check to abort the loop for plugging the hole. Fixes: ecfd41166b72 ("ALSA: usb-audio: Validate UAC3 cluster segment descriptors") Cc: Link: https://patch.msgid.link/20260427152224.15276-1-tiwai@suse.de Signed-off-by: Takashi Iwai Signed-off-by: Greg Kroah-Hartman

ALSA: usb-audio: Fix size validation in convert_chmap_v3()

2025-08-28T14:31:13+00:00

[ Upstream commit 89f0addeee3cb2dc49837599330ed9c4612f05b0 ] The "p" pointer is void so sizeof(*p) is 1. The intent was to check sizeof(*cs_desc), which is 3, instead. Fixes: ecfd41166b72 ("ALSA: usb-audio: Validate UAC3 cluster segment descriptors") Signed-off-by: Dan Carpenter Link: https://patch.msgid.link/aKL5kftC1qGt6lpv@stanley.mountain Signed-off-by: Takashi Iwai Signed-off-by: Sasha Levin

ALSA: usb-audio: Validate UAC3 cluster segment descriptors

2025-08-20T16:30:12+00:00

commit ecfd41166b72b67d3bdeb88d224ff445f6163869 upstream. UAC3 class segment descriptors need to be verified whether their sizes match with the declared lengths and whether they fit with the allocated buffer sizes, too. Otherwise malicious firmware may lead to the unexpected OOB accesses. Fixes: 11785ef53228 ("ALSA: usb-audio: Initial Power Domain support") Reported-and-tested-by: Youngjun Lee Cc: Link: https://patch.msgid.link/20250814081245.8902-2-tiwai@suse.de Signed-off-by: Takashi Iwai Signed-off-by: Greg Kroah-Hartman

ALSA: usb-audio: Fix out-of-bounds read in snd_usb_get_audioformat_uac3()

2025-07-06T09:01:39+00:00

[ Upstream commit fb4e2a6e8f28a3c0ad382e363aeb9cd822007b8a ] In snd_usb_get_audioformat_uac3(), the length value returned from snd_usb_ctl_msg() is used directly for memory allocation without validation. This length is controlled by the USB device. The allocated buffer is cast to a uac3_cluster_header_descriptor and its fields are accessed without verifying that the buffer is large enough. If the device returns a smaller than expected length, this leads to an out-of-bounds read. Add a length check to ensure the buffer is large enough for uac3_cluster_header_descriptor. Signed-off-by: Youngjun Lee Fixes: 9a2fe9b801f5 ("ALSA: usb: initial USB Audio Device Class 3.0 support") Link: https://patch.msgid.link/20250623-uac3-oob-fix-v1-1-527303eaf40a@samsung.com Signed-off-by: Takashi Iwai Signed-off-by: Sasha Levin

ALSA: usb-audio: Fix NULL pointer deref in snd_usb_power_domain_set()

2024-10-10T11:58:44+00:00

Commit adding support for multiple control interfaces expanded struct snd_usb_power_domain with pointer to control interface for proper control message routing but missed one initialization point of this structure, which has left new field with NULL value. Standard mandates that each device has at least one control interface and code responsible for power domain does not check for NULL values when querying for control interface. This caused some USB devices to crash the kernel. Fixes: 6aa8700150f7 ("ALSA: usb-audio: Support multiple control interfaces") Signed-off-by: Karol Kosik Link: https://patch.msgid.link/AS8P190MB1285B563C6B5394DB274813FEC782@AS8P190MB1285.EURP190.PROD.OUTLOOK.COM Signed-off-by: Takashi Iwai

ALSA: usb-audio: Support multiple control interfaces

2024-08-12T14:17:46+00:00

Registering Numark Party Mix II fails with error 'bogus bTerminalLink 1'. The problem stems from the driver not being able to find input/output terminals required to configure audio streaming. The information about those terminals is stored in AudioControl Interface. Numark device contains 2 AudioControl Interfaces and the driver checks only one of them. According to the USB standard, a device can have multiple audio functions, each represented by Audio Interface Collection. Every audio function is considered to be closed box and will contain unique AudioControl Interface and zero or more AudioStreaming and MIDIStreaming Interfaces. The Numark device adheres to the standard and defines two audio functions: - MIDIStreaming function - AudioStreaming function It starts with MIDI function, followed by the audio function. The driver saves the first AudioControl Interface in `snd_usb_audio` structure associated with the entire device. It then attempts to use this interface to query for terminals and clocks. However, this fails because the correct information is stored in the second AudioControl Interface, defined in the second Audio Interface Collection. This patch introduces a structure holding association between each MIDI/Audio Interface and its corresponding AudioControl Interface, instead of relying on AudioControl Interface defined for the entire device. This structure is populated during usb probing phase and leveraged later when querying for terminals and when sending USB requests. Alternative solutions considered include: - defining a quirk for Numark where the order of interface is manually changed, or terminals are hardcoded in the driver. This solution would have fixed only this model, though it seems that device is USB compliant, and it also seems that other devices from this company may be affected. What's more, it looks like products from other manufacturers have similar problems, i.e. Rane One DJ console - keeping a list of all AudioControl Interfaces and querying all of them to find required information. That would have solved my problem and have low probability of breaking other devices, as we would always start with the same logic of querying first AudioControl Interface. This solution would not have followed the standard though. This patch preserves the `snd_usb_audio.ctrl_intf` variable, which holds the first AudioControl Interface, and uses it as a fallback when some interfaces are not parsed correctly and lack an associated AudioControl Interface, i.e., when configured via quirks. Link: https://bugzilla.kernel.org/show_bug.cgi?id=217865 Signed-off-by: Karol Kosik Link: https://patch.msgid.link/AS8P190MB1285893F4735C8B32AD3886BEC852@AS8P190MB1285.EURP190.PROD.OUTLOOK.COM Signed-off-by: Takashi Iwai

ALSA: usb-audio: Correct surround channels in UAC1 channel map

2024-07-31T14:22:51+00:00

USB-audio driver puts SNDRV_CHMAP_SL and _SR as left and right surround channels for UAC1 channel map, respectively. But they should have been SNDRV_CHMAP_RL and _RR; the current value *_SL and _SR are rather "side" channels, not "surround". I guess I took those mistakenly when I read the spec mentioning "surround left". This patch corrects those entries to be the right channels. Suggested-by: Sylvain BERTRAND Closes: https://lore.kernel.orgZ/qIyJD8lhd8hFhlC@freedom Fixes: 04324ccc75f9 ("ALSA: usb-audio: add channel map support") Cc: Link: https://patch.msgid.link/20240731142018.24750-1-tiwai@suse.de Signed-off-by: Takashi Iwai

ALSA: usb-audio: Stop parsing channels bits when all channels are found.

2024-03-13T13:29:22+00:00

If a usb audio device sets more bits than the amount of channels it could write outside of the map array. Signed-off-by: Johan Carlsson Fixes: 04324ccc75f9 ("ALSA: usb-audio: add channel map support") Message-ID: <20240313081509.9801-1-johan.carlsson@teenage.engineering> Signed-off-by: Takashi Iwai