- 17 Aug, 2023 9 commits
-
-
Takashi Iwai authored
Align with the other components, and use snd_device_alloc() for the new sound device for sequencer, too. No functional changes. Reviewed-by:
Jaroslav Kysela <perex@perex.cz> Signed-off-by:
Curtis Malainey <cujomalainey@chromium.org> Tested-by:
Takashi Iwai <tiwai@suse.de>
-
Takashi Iwai authored
Align with the other components, and use snd_device_alloc() for the new sound device for timer, too. No functional changes. Reviewed-by:
-
Takashi Iwai authored
Embedding the struct device to snd_compr object may result in UAF when the delayed kobj release is used. Like other devices, let's detach the struct device from the snd_compr by allocating dynamically via snd_device_alloc(). Reviewed-by:
-
Takashi Iwai authored
This patch detaches the struct device from the snd_rawmidi object by allocating via snd_device_alloc(), just like done for other devices. Reviewed-by:
-
Takashi Iwai authored
Like control and PCM devices, it's better to avoid the embedded struct device for hwdep (although it's more or less well working), too. Change it to allocate via snd_device_alloc(), and free the memory at the common snd_hwdep_free(). Reviewed-by:
-
Takashi Iwai authored
So far we use the embedded struct device for each PCM substreams in struct snd_pcm. This may result in UAF when the delayed kobj release is used; each corresponding struct device is still accessed at the (delayed) device release, while the snd_pcm object may be already gone. As a workaround, detach the struct device from the snd_pcm object by allocating via the new snd_device_alloc() helper. A caveat is that we store the PCM substream pointer to drvdata since the device resume and others require the access to it. This patch is based on the fix Curtis posted initially. In this patch, the changes are split and use the new helper function instead. Link: https://lore.kernel.org/r/20230801171928.1460120-1-cujomalainey@chromium.orgReviewed-by:
-
Takashi Iwai authored
Embedding the ctl_dev in the snd_card object may result in UAF when the delayed kobj release is used; at the delayed kobj release, it still accesses the struct device itself while the card memory (that embeds the struct device) may be already gone. As a workaround, detach the struct device from the card object by allocating via the new snd_device_alloc() helper. The rest are just replacing ctl_dev access to the pointer. This is based on the fix Curtis posted initially. In this patch, the changes are split and use the new helper function instead. Link: https://lore.kernel.org/r/20230801171928.1460120-1-cujomalainey@chromium.orgReviewed-by:
-
Takashi Iwai authored
Introduce a new helper, snd_device_alloc(), for allocating a struct device that is bound with the sound class. It's a replacement of snd_device_initialize(). Reviewed-by:
-
Yang Yingliang authored
cs35l41_prop_model is only used in cs35l41_hda_property.c now, change it to static. Fixes: ef4ba63f ("ALSA: hda: cs35l41: Support systems with missing _DSD properties") Signed-off-by:
Yang Yingliang <yangyingliang@huawei.com> Link: https://lore.kernel.org/r/20230817014252.1511232-1-yangyingliang@huawei.comSigned-off-by:
-
- 16 Aug, 2023 2 commits
-
-
Brady Norander authored
AlderLake and RaptorLake Chromebooks currently use the HDA driver by default. Add a quirk to use the SOF driver on these platforms, which is needed for functional internal audio. Signed-off-by:
Brady Norander <bradynorander@gmail.com> Acked-by:
Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Acked-by:
-
Takashi Iwai authored
model->hid is a pointer, and should be rather NULL-checked in the loop of cs35l41_prop_model_table. Fixes: ef4ba63f ("ALSA: hda: cs35l41: Support systems with missing _DSD properties") Reported-by:
kernel test robot <lkp@intel.com> Closes: https://lore.kernel.org/oe-kbuild-all/202308160506.8lCEeFDG-lkp@intel.com/ Link: https://lore.kernel.org/r/20230816063525.23009-1-tiwai@suse.deSigned-off-by:
-
- 15 Aug, 2023 1 commit
-
-
Stefan Binding authored
Some systems using CS35L41 with HDA were released without some required _DSD properties in ACPI. To support these special cases, add an api to configure the correct properties for systems with this issue. This initial commit moves the no _DSD support for Lenovo Legion Laptops (CLSA0100, CLSA0101) into a new framework which can be extended to support additional laptops in the future. Signed-off-by:
Stefan Binding <sbinding@opensource.cirrus.com> Link: https://lore.kernel.org/r/20230815161033.3519-1-sbinding@opensource.cirrus.comSigned-off-by:
-
- 10 Aug, 2023 1 commit
-
-
Takashi Sakamoto authored
Hard-coded stream format parameters are added for Weiss Engineering FireWire devices. When the device vendor and model match, the parameters are copied into the stream format cache. This allows for setting all supported sampling rates up to 192kHz, and consequently adjusting the number of available I/O channels. Signed-off-by:
Rolf Anderegg <rolf.anderegg@weiss.ch> Signed-off-by:
Michele Perrone <michele.perrone@weiss.ch> Signed-off-by:
Takashi Sakamoto <o-takashi@sakamocchi.jp> Link: https://lore.kernel.org/r/20230809002631.750120-1-o-takashi@sakamocchi.jpSigned-off-by:
-
- 08 Aug, 2023 4 commits
-
-
Justin Stitt authored
`strncpy` is deprecated for use on NUL-terminated destination strings [1]. A suitable replacement is `strscpy` [2] due to the fact that it guarantees NUL-termination on its destination buffer argument which is _not_ the case for `strncpy`! It should be noted that the current implementation is unlikely to have a bug because `drv_name` is a string literal with a size of 9 while `card->driver` has a size of 16. However, it is probably worthwhile to switch to a more robust and less ambiguous interface. [1]: www.kernel.org/doc/html/latest/process/deprecated.html#strncpy-on-nul-terminated-strings [2]: https://manpages.debian.org/testing/linux-manual-4.8/strscpy.9.en.html Link: https://github.com/KSPP/linux/issues/90 Cc: linux-hardening@vger.kernel.org Signed-off-by:
Justin Stitt <justinstitt@google.com> Link: https://lore.kernel.org/r/20230807-sound-pci-hda-v1-1-6d9cdcd085ca@google.comSigned-off-by:
-
Yue Haibing authored
These declarations is never used since beginning of git history. Signed-off-by:
Yue Haibing <yuehaibing@huawei.com> Link: https://lore.kernel.org/r/20230807141513.31440-1-yuehaibing@huawei.comSigned-off-by:
-
Ivan Orlov authored
Remove redundant definitions of DEVNAME and CARD_NAME, as they're not useful. The former is not used anywhere, and the latter is used only in module parameters descriptions. Signed-off-by:
Ivan Orlov <ivan.orlov0322@gmail.com> Link: https://lore.kernel.org/r/20230804110740.9867-2-ivan.orlov0322@gmail.comSigned-off-by:
-
Ivan Orlov authored
Trigger callback is not the best place for buffer iterator initialization, so move it out to the prepare callback, where it have to be. Minor enhancement: remove blank line. Signed-off-by:
-
- 04 Aug, 2023 6 commits
-
-
Takashi Iwai authored
Pull ArrowLake-S PCI ID addition and config for MTL/LNL Link: https://lore.kernel.org/r/20230802150105.24604-1-pierre-louis.bossart@linux.intel.comSigned-off-by:
-
Kai Vehmanen authored
Expand the HDA/I915 connectivity check to correctly handle the PCI topology used in some Intel Arrow Lake products. Reviewed-by:
Bard Liao <yung-chuan.liao@linux.intel.com> Tested-by:
"T, Arun" <arun.t@intel.com> Signed-off-by:
Kai Vehmanen <kai.vehmanen@linux.intel.com> Signed-off-by:
-
Pierre-Louis Bossart authored
One more PCI ID for the road. Signed-off-by:
Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Reviewed-by:
-
Pierre-Louis Bossart authored
This was not updated in Commit 0cd0a7c2 ("ALSA: intel-dsp-config: Convert to PCI device IDs defines") Signed-off-by:
-
Kai Vehmanen authored
Add HD Audio PCI ID for Intel Arrow Lake-S platform. Reviewed-by:
-
Pierre-Louis Bossart authored
Add part ID to common include file Signed-off-by:
-
- 02 Aug, 2023 2 commits
-
-
Stefan Binding authored
Print amp configuration information to be able to confirm ACPI _DSD information (and other useful info) for each amp on each system using CS35L41, without having to get the acpidump. Signed-off-by:
-
Yu Liao authored
sparse reports sound/ac97/bus.c:465:1: sparse: sparse: symbol 'dev_attr_vendor_id' was not declared. Should it be static? This variable is only used in its defining file, so it should be static. Reported-by:
Yu Liao <liaoyu15@huawei.com> Link: https://lore.kernel.org/r/20230802022649.2514787-1-liaoyu15@huawei.comSigned-off-by:
-
- 01 Aug, 2023 10 commits
-
-
Yue Haibing authored
Commit 68e67f40 ("ALSA: snd-usb: move calls to usb_set_interface") leave this unused declaration. Signed-off-by:
-
Richard Fitzgerald authored
The ACPI nodes for CS35L56 can contain an extra I2CSerialBusV2 that is not a real device, it is an alias address. This alias address will not be in the cirrus,dev-index array, so reject any instantions with a device address not found in the array. Signed-off-by:
Richard Fitzgerald <rf@opensource.cirrus.com> Link: https://lore.kernel.org/r/20230731165726.7940-10-rf@opensource.cirrus.comSigned-off-by:
-
Richard Fitzgerald authored
A tuning patch is always needed to enable the ASP audio port. If the BIOS did not patch the firmware, then it is mandatory to have a .bin file. Signed-off-by:
-
Richard Fitzgerald authored
A RAM firmware can only be downloaded if the CS35L56 is currently running from ROM firmware. The driver must not try to overwrite the RAM if the CS35L56 is already running from that RAM. Firmware can be downloaded in these two cases: - The BIOS has already patched the firmware (secured mode). In this case the firmware files will only contain tunings that are safe to overwrite. - The CS35L56 is running the built-in ROM firmware. After a RAM firmware has been downloaded it can only be cleared by hard resetting CS35L56. Some systems only hard-reset during power-on and do not give the driver control of hard reset. Signed-off-by:
-
Richard Fitzgerald authored
If cs35l56_hda_fw_load() successfully called cs_dsp_power_up() the error path must balance that with a call to cs_dsp_power_down(). Signed-off-by:
-
Richard Fitzgerald authored
In cs35l56_hda_unbind() cs_dsp_power_down() must be called to cleanup before calling cs_dsp_remove cs_dsp_remove(). Signed-off-by:
-
Richard Fitzgerald authored
Always call cs_dsp_power_up() and cs_dsp_run() in cs35l56_hda_fw_load() even if there aren't any firmware files to download. Also, if there aren't any firmware files to download there is no need to do cs35l56_firmware_shutdown() and cs35l56_system_reset(). If there aren't any firmware files there's no need to write anything to the CS35L56 registers to make it work - it will already be running the ROM firmware. So it's not strictly necessary to start cs_dsp. But it's perfectly ok to call cs_dsp_power_up() and cs_dsp_run() without downloading any firmware. This avoids having to support a state where audio is playing but cs_dsp is not running. Signed-off-by:
-
Richard Fitzgerald authored
When firmware is reloaded after a system resume cs_dsp_power_down() should be called before calling cs_dsp_power_up(). The fw_patched flag should also be cleared and only set again if the firmware download succeeded. Signed-off-by:
-
Richard Fitzgerald authored
Only call regcache_mark_dirty() in cs35l56_hda_fw_load() if the CS35L56 was SYSTEM_RESET. recache_mark_dirty() changes the behaviour of regcache_sync() to write out cache values that are not the default value, and skip cache values that are the default. AUDIO_REINIT does not reset the registers. regcache_mark_dirty() after AUDIO_REINIT could cause the regcache_sync() to sync registers incorrectly because it will assume that all registers have reset to default. Signed-off-by:
-
Richard Fitzgerald authored
Move the call to cs_dsp_run() in cs35l56_hda_fw_load() so that it is after the CS35L56 has been reset/reinit'd and the regmap cache has been synced. cs_dsp_run() syncs up ALSA control cache values with the DSP memory so this must not be done until the firmware has reinitialized. Signed-off-by:
-
- 29 Jul, 2023 2 commits
-
-
Justin Stitt authored
`strncpy` is deprecated for use on NUL-terminated destination strings [1]. A suitable replacement is `strscpy` [2] due to the fact that it guarantees NUL-termination on its destination buffer argument which is _not_ always the case for `strncpy`! It should be noted that, in this case, the destination buffer has a length strictly greater than the source string. Moreover, the source string is NUL-terminated (and so is the destination) which means there was no real bug happening here. Nonetheless, this patch would get us one step closer to eliminating the `strncpy` API in the kernel, as its use is too ambiguous. We need to favor less ambiguous replacements such as: strscpy, strscpy_pad, strtomem and strtomem_pad (amongst others). Technically, my patch yields subtly different behavior. The original implementation with `strncpy` would fill the entire destination buffer with null bytes [3] while `strscpy` will leave the junk, uninitialized bytes trailing after the _mandatory_ NUL-termination. So, if somehow `card->driver` or `card->shortname` require this NUL-padding behavior then `strscpy_pad` should be used. My interpretation, though, is that the aforementioned fields are just fine as NUL-terminated strings. Please correct my assumptions if needed and I'll send in a v2. [1]: www.kernel.org/doc/html/latest/process/deprecated.html#strncpy-on-nul-terminated-strings [2]: manpages.debian.org/testing/linux-manual-4.8/strscpy.9.en.html [3]: https://linux.die.net/man/3/strncpy Link: https://github.com/KSPP/linux/issues/90 Link: https://lore.kernel.org/r/20230727-sound-xen-v1-1-89dd161351f1@google.com (related ALSA patch) Signed-off-by:
-
Justin Stitt authored
`strncpy` is deprecated for use on NUL-terminated destination strings [1]. A suitable replacement is `strscpy` [2] due to the fact that it guarantees NUL-termination on its destination buffer argument which is _not_ always the case for `strncpy`! It should be noted that, in this case, the destination buffer has a length strictly greater than the source string. Moreover, the source string is NUL-terminated (and so is the destination) which means there was no real bug happening here. Nonetheless, this patch would get us one step closer to eliminating the `strncpy` API in the kernel, as its use is too ambiguous. We need to favor less ambiguous replacements such as: strscpy, strscpy_pad, strtomem and strtomem_pad (amongst others). Technically, my patch yields subtly different behavior. The original implementation with `strncpy` would fill the entire destination buffer with null bytes [3] while `strscpy` will leave the junk, uninitialized bytes trailing after the _mandatory_ NUL-termination. So, if somehow `pcm->name` or `card->driver/shortname/longname` require this NUL-padding behavior then `strscpy_pad` should be used. My interpretation, though, is that the aforementioned fields are just fine as NUL-terminated strings. Please correct my assumptions if needed and I'll send in a v2. [1]: www.kernel.org/doc/html/latest/process/deprecated.html#strncpy-on-nul-terminated-strings [2]: manpages.debian.org/testing/linux-manual-4.8/strscpy.9.en.html [3]: https://linux.die.net/man/3/strncpy Link: https://github.com/KSPP/linux/issues/90Signed-off-by:
Kees Cook <keescook@chromium.org> Link: https://lore.kernel.org/r/20230727-sound-xen-v1-1-89dd161351f1@google.comSigned-off-by:
-
- 27 Jul, 2023 1 commit
-
-
Dan Carpenter authored
Smatch complains that this return should be a goto: sound/pci/hda/cs35l56_hda.c:910 cs35l56_hda_common_probe() warn: missing unwind goto? The goto error disables cansleep so that seems reasonable. Fixes: 73cfbfa9 ("ALSA: hda/cs35l56: Add driver for Cirrus Logic CS35L56 amplifier") Signed-off-by:Dan Carpenter <dan.carpenter@linaro.org> Reviewed-by:
-
- 25 Jul, 2023 1 commit
-
-
Takashi Iwai authored
Merge tag 'asoc-v6.6-early' of https://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound into for-next ASoC: Updates for v6.6 Here's an initial batch of updates for ASoC for this release cycle. We've got a bunch of new drivers in here, a bit of core work from Morimoto-san and quite a lot of janitorial work. There's several updates that pull in changes from other subsystems in order to build on them: - An adaptor to allow use of IIO DACs and ADCs in ASoC which pulls in some IIO changes. - Create a library function for intlog10() and use it in the NAU8825 driver. - Include the ASoC tests, including the topology tests, in the default KUnit full test coverage. This also involves enabling UML builds of ALSA since that's the default KUnit test environment which pulls in the addition of some stubs to the driver. - More factoring out from Morimoto-san. - Convert a lot of drivers to use the more modern maple tree register cache. - Support for AMD machines with MAX98388 and NAU8821, Cirrus Logic CS35L36, Intel AVS machines with ES8336 and RT5663 and NXP i.MX93.
-
- 24 Jul, 2023 1 commit
-
-
Mark Brown authored
Merge series from Richard Fitzgerald <rf@opensource.cirrus.com>: This set of patches adds support for using the CS35L56 boosted smart amplifier on HDA systems. In these systems the CS35L56 audio is routed through a HDA-to-I2S bridge codec. This doesn't include the changes to the Realtek driver to actually hook up the CS35L56 driver, because we don't yet have the QUIRK IDs to associate it with. But we want to publish the driver now so that it is available for bringing up hardware with the CS35L56. The first 9 patches are moving code out of the ASoC driver and into the shared library so that it can be shared with the HDA driver. Patch #10 fixes missing #includes in the HDA headers so that the CS35L56 driver doesn't have to #include headers that it doesn't use.
-
