linux-stable.git/include/sound, branch v5.16.3

ALSA: hda: Fix potential deadlock at codec unbinding

2022-01-27T11:02:02+00:00

[ Upstream commit 7206998f578d5553989bc01ea2e544b622e79539 ]

When a codec is unbound dynamically via sysfs while its stream is in
use, we may face a potential deadlock at the proc remove or a UAF.
This happens since the hda_pcm is managed by a linked list, as it
handles the hda_pcm object release via kref.

When a PCM is opened at the unbinding time, the release of hda_pcm
gets delayed and it ends up with the close of the PCM stream releasing
the associated hda_pcm object of its own.  The hda_pcm destructor
contains the PCM device release that includes the removal of procfs
entries.  And, this removal has the sync of the close of all in-use
files -- which would never finish because it's called from the PCM
file descriptor itself, i.e. it's trying to shoot its foot.

For addressing the deadlock above, this patch changes the way to
manage and release the hda_pcm object.  The kref of hda_pcm is
dropped, and instead a simple refcount is introduced in hda_codec for
keeping the track of the active PCM streams, and at each PCM open and
close, this refcount is adjusted accordingly.  At unbinding, the
driver calls snd_device_disconnect() for each PCM stream, then
synchronizes with the refcount finish, and finally releases the object
resources.

Fixes: bbbc7e8502c9 ("ALSA: hda - Allocate hda_pcm objects dynamically")
Link: https://lore.kernel.org/r/20211116072459.18930-1-tiwai@suse.de
Signed-off-by: Takashi Iwai 
Signed-off-by: Sasha Levin

Merge tag 'asoc-fix-v5.16-rc3' of https://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound into for-linus

2021-11-25T13:35:24+00:00

ASoC: Fixes for v5.16

There's a large but repetitive set of fixes here for issues with the
Tegra kcontrols not correctly reporting changes to userspace, a fix for
some issues with matching on older x86 platforms introduced during the
merge window together with a set of smaller fixes and one new system
quirk.

ASoC: soc-acpi: Set mach->id field on comp_ids matches

2021-11-22T15:40:01+00:00

Commit dac7cbd55dca ("ASoC: Intel: soc-acpi-byt: shrink tables using
compatible IDs") and commit 959ae8215a9e ("ASoC: Intel: soc-acpi-cht:
shrink tables using compatible IDs") simplified the match tables in
soc-acpi-intel-byt-match.c and soc-acpi-intel-cht-match.c by merging
identical entries using the new .comp_ids snd_soc_acpi_mach field to
point a single entry to multiple ACPI HIDs and clearing the previously
unique per entry .id field.

But various machine drivers from sound/soc/intel/boards rely on mach->id
in one or more ways, e.g. some drivers contain the following snippets:

	adev = acpi_dev_get_first_match_dev(mach->id, NULL, -1);

	pkg_found = snd_soc_acpi_find_package_from_hid(mach->id, ...

	if (!strncmp(snd_soc_cards[i].codec_id, mach->id, 8)) { ...

All of which are broken by the match table shrinking.

Make the snd_soc_acpi_mach.id field non const (the storage for the tables
already is non const) and on a comps_ids match copy the matching HID to
the id field to fix this.

Fixes: dac7cbd55dca ("ASoC: Intel: soc-acpi-byt: shrink tables using compatible IDs")
Fixes: 959ae8215a9e ("ASoC: Intel: soc-acpi-cht: shrink tables using compatible IDs")
Suggested-by: Pierre-Louis Bossart 
Cc: Pierre-Louis Bossart 
Cc: Brent Lu 
Signed-off-by: Hans de Goede 
Reviewed-by: Pierre-Louis Bossart 
Link: https://lore.kernel.org/r/20211118153014.349222-1-hdegoede@redhat.com
Signed-off-by: Mark Brown

Revert "ALSA: memalloc: Convert x86 SG-buffer handling with non-contiguous type"

2021-11-04T21:10:03+00:00

This reverts commit 2d9ea39917a4e4293bc2caea902c7059a330b611.

We've got a regression report showing that the audio got broken the
device over AMD IOMMU.  The conversion assumed the wrong pointer /
page mapping for the indirect mapping case, and we need to correct
this urgently, so let's revert it for now.

Fixes: 2d9ea39917a4 ("ALSA: memalloc: Convert x86 SG-buffer handling with non-contiguous type")
Reported-and-tested-by: Alex Xu (Hello71) 
Link: https://lore.kernel.org/r/20211104180846.16340-1-tiwai@suse.de
Signed-off-by: Takashi Iwai

Merge tag 'asoc-v5.16' of https://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound into for-linus

2021-11-01T15:58:27+00:00

ASoC: Updates for v5.16

This is an unusually large set of updates, mostly a large crop of
unusually big drivers coupled with extensive overhauls of existing code.
There's a SH change here for the DAI format terminology, the change is
straightforward and the SH maintainers don't seem very active.

 - A new version of the audio graph card which supports a wider range of
   systems.
 - Move of the Cirrus DSP framework into drivers/firmware to allow for
   future use by non-audio DSPs.
 - Several conversions to YAML DT bindings.
 - Continuing cleanups to the SOF and Intel code.
 - A very big overhaul of the cs42l42 driver, correcting many problems.
 - Support for AMD Vangogh and Yelow Cap, Cirrus CS35L41, Maxim
   MAX98520 and MAX98360A, Mediatek MT8195, Nuvoton NAU8821, nVidia
   Tegra210, NXP i.MX8ULP, Qualcomm AudioReach, Realtek ALC5682I-VS,
   RT5682S, and RT9120 and Rockchip RV1126 and RK3568

ASoC: soc-acpi: add comp_ids field for machine driver matching

2021-10-29T17:55:16+00:00

A machine driver needs to be enumerated by more than one ACPI HID if
it supports second headphone driver (i.e. rt5682 and rt5682s).
However, the id field in snd_soc_acpi_mach structure could contain
only one HID. By adding a 'comp_ids' field which can contain several
HIDs, we can enumerate a machine driver by multiple ACPI HIDs.

Signed-off-by: Brent Lu 
Link: https://lore.kernel.org/r/20211029171409.611600-2-brent.lu@intel.com
Signed-off-by: Mark Brown

ASoC: topology: Fix stub for snd_soc_tplg_component_remove()

2021-10-25T16:41:26+00:00

When removing the index argument from snd_soc_topology_component_remove()
commit a5b8f71c5477f (ASoC: topology: Remove multistep topology loading)
forgot to update the stub for !SND_SOC_TOPOLOGY use, causing build failures
for anything that tries to make use of it.

Fixes: a5b8f71c5477f (ASoC: topology: Remove multistep topology loading)
Signed-off-by: Mark Brown 
Link: https://lore.kernel.org/r/20211025154844.2342120-1-broonie@kernel.org
Signed-off-by: Mark Brown

ALSA: memalloc: Drop superfluous snd_dma_buffer_sync() declaration

2021-10-19T06:07:26+00:00

snd_dma_buffer_sync() is declared twice, and the one outside the ifdef
CONFIG_HAS_DMA could lead to a build error when CONFIG_HAS_DMA=n.
As it's an overlooked leftover after rebase, drop this line.

Fixes: a25684a95646 ("ALSA: memalloc: Support for non-contiguous page allocation")
Reported-by: Stephen Rothwell 
Link: https://lore.kernel.org/r/20211019165402.4fa82c38@canb.auug.org.au
Link: https://lore.kernel.org/r/20211019060536.26089-1-tiwai@suse.de
Signed-off-by: Takashi Iwai

ASoC: soc-component: add snd_soc_component_is_codec()

2021-10-18T12:56:37+00:00

Checking .non_legacy_dai_naming is not readable.
Let's add new snd_soc_component_is_codec().

Signed-off-by: Kuninori Morimoto 
Link: https://lore.kernel.org/r/87h7dft7dn.wl-kuninori.morimoto.gx@renesas.com
Signed-off-by: Mark Brown

Merge tag 'v5.15-rc6' into asoc-5.16

2021-10-18T12:50:42+00:00

Linux 5.15-rc6