- 01 Jun, 2018 1 commit
-
-
Bo Chen authored
When 'kzalloc()' fails in 'snd_hda_attach_pcm_stream()', a new pcm instance is created without setting its operators via 'snd_pcm_set_ops()'. Following operations on the new pcm instance can trigger kernel null pointer dereferences and cause kernel oops. This bug was found with my work on building a gray-box fault-injection tool for linux-kernel-module binaries. A kernel null pointer dereference was confirmed from line 'substream->ops->open()' in function 'snd_pcm_open_substream()' in file 'sound/core/pcm_native.c'. This patch fixes the bug by calling 'snd_device_free()' in the error handling path of 'kzalloc()', which removes the new pcm instance from the snd card before returns with an error code. Signed-off-by:
Bo Chen <chenbo@pdx.edu> Cc: <stable@vger.kernel.org> Signed-off-by:
Takashi Iwai <tiwai@suse.de>
-
- 31 May, 2018 2 commits
-
-
Yisheng Xie authored
match_string() returns the index of an array for a matching string, which can be used instead of open coded variant. Signed-off-by:
Yisheng Xie <xieyisheng1@huawei.com> Signed-off-by:
-
Dan Carpenter authored
We want the loop to exit when "to" is set to zero, but in the current code it's set to -1. Also I tweaked the indenting so it doesn't look like we're passing "--to" to xenbus_read_unsigned(). Fixes: cc3196ae ("ALSA: xen-front: Introduce Xen para-virtualized sound frontend driver") Signed-off-by:
Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by:
Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com> Signed-off-by:
-
- 30 May, 2018 2 commits
-
-
Dan Carpenter authored
kfree() doesn't accept error pointers so I've set "str" to NULL on these paths. Fixes: fd3b3604 ("ALSA: xen-front: Read sound driver configuration from Xen store") Signed-off-by:
-
Hui Wang authored
We have several Lenovo AIOs like M810z, M820z and M920z, they have the same design for mic-mute hotkey and led and they use the same codec with the same pin configuration, so use the pin conf table to apply fix to all of them. Fixes: 29693efc ("ALSA: hda - Fix micmute hotkey problem for a lenovo AIO machine") Cc: <stable@vger.kernel.org> Signed-off-by:
Hui Wang <hui.wang@canonical.com> Signed-off-by:
-
- 29 May, 2018 6 commits
-
-
Tom Briden authored
Added a new helper file for these fixups due to requiring a huge number of coefs being set to get the top speakers to work, as well as setting pin 0x17 for the top speakers and the correct input source of 0x17 for volume control [ Note: this is a revised work based on Tom's fixup patch with the replacement of the full COEF tables provided by Realtek. Also, the fixup function has a proper HDA_FIXUP_ACT_* handling now. The credit for the new COEF table goes to Kailang -- tiwai ] Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=189331 Cc: Kailang Yang <kailang@realtek.com> Signed-off-by:
Tom Briden <tom@decompile.me.uk> Tested-by:
-
Takashi Iwai authored
Just a code refactoring to use the common helper for the all three functions. Signed-off-by: -
Tom Briden authored
This patch adds the mute LED control for HP Spectre x360 Kabylake model. The mute LED is controlled via VREF bits on NID 0x1b, so we need a new fixup function. Note that this doesn't fix the other issues like the missing speaker output on the machine. They will be addressed by later patches. Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=189331Signed-off-by:
-
Takashi Iwai authored
Currently, USB-audio driver allocates the PCM buffer via vmalloc(), as this serves merely as an intermediate buffer that is copied to each URB transfer buffer. This works well in general on x86, but on some archs this may result in cache coherency issues when mmap is used. OTOH, it works also on such arch unless mmap is used. This patch is a step for mitigating the inconvenience; a new module option "use_vmalloc" is provided so that user can choose to allocate the DMA coherent buffer instead of the existing vmalloc buffer. The drawback is that it'd be the standard dma_alloc_coherent() calls and the system would require contiguous pages on non-x86 archs. Note that it's a global option and not dynamically switchable since the buffer is pre-allocated at the probe time. In theory, it's possible to be switchable, but it'd be trickier and racier. As default use_vmalloc option is set to true, so that the old behavior is kept. For allowing the coherent mmap on ARM or MIPS, pass use_vmalloc=0 option explicitly. Reported-and-tested-by:
Daniel Danzberger <daniel@dd-wrt.com> Signed-off-by:
-
Hans de Goede authored
Power-saving is causing a humming sound when active on the Intel NUC5i7RY, add it to the blacklist. BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=199607Signed-off-by:
Hans de Goede <hdegoede@redhat.com> Signed-off-by:
-
Arnd Bergmann authored
gcc-8 warns that pcm_instance->name is not necessarily terminated correctly if the input is more than 80 characters long or lacks a termination byte itself: In function 'strncpy', inlined from 'cfg_device' at sound/xen/xen_snd_front_cfg.c:399:3, inlined from 'xen_snd_front_cfg_card' at sound/xen/xen_snd_front_cfg.c:509:9: include/linux/string.h:254:9: error: '__builtin_strncpy' specified bound 80 equals destination size [-Werror=stringop-truncation] return __builtin_strncpy(p, q, size); Using strlcpy() instead of strncpy() makes this a bit safer. Fixes: fd3b3604 ("ALSA: xen-front: Read sound driver configuration from Xen store") Signed-off-by:Arnd Bergmann <arnd@arndb.de> Reviewed-by:
-
- 28 May, 2018 8 commits
-
-
Takashi Iwai authored
Simplify the device management by replacing the lowlevel device object allocation with the card->private_data. Nowadays there is almost no advantage by the lowlevel device, and with card->private_data, the code becomes cleaner. Signed-off-by: -
Takashi Iwai authored
Avoid if ((err = ...) style and expand to multiple lines instead. No change in the end result, but just the beautification. Signed-off-by: -
Takashi Iwai authored
... so that we can avoid the extra goto lines. Also beautify the code to follow the standard codex. No functional changes. Signed-off-by: -
Takashi Iwai authored
The stream direction in open and close callbacks can be retrieved from substream->direction, hence we don't have to stick with the unique PCM ops hard-coded for each direction. Rewrite the common open/close callback functions. Signed-off-by: -
Takashi Iwai authored
The PCM ops defined for playback and capture are identical. Just use the single one for both. Signed-off-by: -
Colin Ian King authored
The negative error return from the call to to_sndif_format is being assigned to an unsigned 8 bit integer and hence the check for a negative value is always going to be false. Fix this by using ret as the error return and hence the negative error can be detected and assign the u8 sndif_format to ret if there is no error. Detected by CoverityScan, CID#1469385 ("Unsigned compared against 0") Signed-off-by:Colin Ian King <colin.king@canonical.com> Reviewed-by:
Takashi Sakamoto <o-takashi@sakamoccchi.jp> Signed-off-by:
-
Colin Ian King authored
The error for a -ve value in ret is redundant as all previous assignments to ret have an associated -ve check and hence it is impossible for ret to be less that zero at the point of the check. Remove this redundant error check. Detected by CoveritScan, CID#1469407 ("Logically Dead code") Signed-off-by:Takashi Sakamoto <o-takashi@sakamocchi.jp> Signed-off-by:
-
Joe Perches authored
Convert the S_<FOO> symbolic permissions to their octal equivalents as using octal and not symbolic permissions is preferred by many as more readable. see: https://lkml.org/lkml/2016/8/2/1945 Done with automated conversion via: $ ./scripts/checkpatch.pl -f --types=SYMBOLIC_PERMS --fix-inplace <files...> Miscellanea: o Wrapped one multi-line call to a single line Signed-off-by:
Joe Perches <joe@perches.com> Acked-by:
Vinod Koul <vkoul@kernel.org> Signed-off-by:
-
- 27 May, 2018 2 commits
-
-
Takashi Sakamoto authored
At present, all of models produced by TC Electronic except for Konnekt Live are supported with hard-coded their stream formats. Studio Konnekt 48 is sore model to support dual streams for both directions. The second stream has no MIDI conformant data channel in its data block. But current implementation transfers the second stream with MIDI conformant data channel. This commit fixes this issue. Signed-off-by:
-
Takashi Sakamoto authored
TC Electronic Studio Konnekt 48 is an application of combination of WaveFront Dice II STD and TC Applied Technologies (TCAT) TCD2210 (Dice Mini). The latter is on a board with BNC and optical interfaces, thus used for signal processing for word clock, S/PDIF and ADAT. This model doesn't support TCAT extended application protocol. For such devices, ALSA dice driver needs to have hard-coded parameters for stream formats. This commit fixes stream format parameters for this model. Unfortunately, at sampling transmission frequencies over 48.0kHz, I confirmed that current ALSA dice driver doesn't drive the device appropriately to generate sounds (silence). I guess that this comes from timestamping quirk of Dice-based devices, which I reported. [alsa-devel] Dice packet sequence quirk and ALSA firewire stack in Linux 4.6 http://mailman.alsa-project.org/pipermail/alsa-devel/2016-May/107715.html $ cd linux-firewire-utils/src $ python2 crpp < /sys/bus/firewire/devices/fw1/config_rom ROM header and bus information block ----------------------------------------------------------------- 400 04044a26 bus_info_length 4, crc_length 4, crc 18982 404 31333934 bus_name "1394" 408 e0ff8112 irmc 1, cmc 1, isc 1, bmc 0, pmc 0, cyc_clk_acc 255, max_rec 8 (512), max_rom 1, gen 1, spd 2 (S400) 40c 00016604 company_id 000166 | 410 08a65810 device_id 0408a65810 | EUI-64 0001660408a65810 root directory ----------------------------------------------------------------- 414 00062ab9 directory_length 6, crc 10937 418 03000166 vendor 41c 8100000a --> descriptor leaf at 444 420 17000022 model 424 8100000f --> descriptor leaf at 460 428 0c0087c0 node capabilities per IEEE 1394 42c d1000001 --> unit directory at 430 unit directory at 430 ----------------------------------------------------------------- 430 0004d5c5 directory_length 4, crc 54725 434 12000166 specifier id 438 13000001 version 43c 17000022 model 440 8100000f --> descriptor leaf at 47c descriptor leaf at 444 ----------------------------------------------------------------- 444 0006c490 leaf_length 6, crc 50320 448 00000000 textual descriptor 44c 00000000 minimal ASCII 450 54432045 "TC E" 454 6c656374 "lect" 458 726f6e69 "roni" 45c 63000000 "c" descriptor leaf at 460 ----------------------------------------------------------------- 460 0006e08e leaf_length 6, crc 57486 464 00000000 textual descriptor 468 00000000 minimal ASCII 46c 53747564 "Stud" 470 696f4b6f "ioKo" 474 6e6e656b "nnek" 478 74343800 "t48" descriptor leaf at 47c ----------------------------------------------------------------- 47c 0006e08e leaf_length 6, crc 57486 480 00000000 textual descriptor 484 00000000 minimal ASCII 488 53747564 "Stud" 48c 696f4b6f "ioKo" 490 6e6e656b "nnek" 494 74343800 "t48" Signed-off-by:
-
- 26 May, 2018 1 commit
-
-
Colin Ian King authored
Trivial fix to spelling mistake in string Signed-off-by:
-
- 24 May, 2018 3 commits
-
-
Takashi Iwai authored
Drop pci_device() macro that just leads to chip->pci->dev, and pass it directly to request_firmware(). It was introduced for allowing the external alsa-driver kernel module builds. Since it was discontinued years ago, we should clean it up now. Signed-off-by: -
Takashi Iwai authored
Drop the superfluous #ifndef checks that had been put just for allowing building the alsa-driver kernel modules externally. Since the external build was discontinued years ago, let's clean up the old kludges. Signed-off-by: -
Takashi Iwai authored
Drop the superfluous #ifndef check in memalloc.h that had been put just for allowing building the alsa-driver kernel modules externally. Since the external build was discontinued years ago, let's clean up the old kludges. Signed-off-by:
-
- 23 May, 2018 4 commits
-
-
Hans de Goede authored
Power-saving is causing plops on audio start/stop on ASRock H81M-HDS machines, add these to the power_save blacklist. BugLink: https://bugzilla.redhat.com/show_bug.cgi?id=1525104Signed-off-by:
-
Hans de Goede authored
Power-saving is causing plops on audio start/stop on Gigabyte P55A-UD3 and Gigabyte Z87-D3HP machines, add these to the power_save blacklist. Note these 2 boards both use 1458:a002 as subsystem ids, so they share a single entry. BugLink: https://bugzilla.redhat.com/show_bug.cgi?id=1525104Signed-off-by:
-
Hans de Goede authored
Power-saving is causing a plop and silences the first 2 seconds (give or take) of audio, silencing notifications sounds on Medion / Clevo W35xSS_370SS laptops. Add the Clevo W35xSS_370SS to the power_save blacklist. BugLink: https://bugzilla.redhat.com/show_bug.cgi?id=1581607Signed-off-by:
-
Hans de Goede authored
Power-saving is causing a humming sound when active on the Intel NUC7i3BNB, add it to the blacklist. BugLink: https://bugzilla.redhat.com/show_bug.cgi?id=1520902Signed-off-by:
-
- 22 May, 2018 3 commits
-
-
Takashi Iwai authored
Add "hp-mic-fix" model string for Conexant codecs so that user can test the quirk without recompiling. Signed-off-by: -
Takashi Iwai authored
HP Z2 G4 requires the same workaround as other HP machines that have no mic-pin detection. Cc: <stable@vger.kernel.org> Signed-off-by: -
Yisheng Xie authored
match_string() returns the index of an array for a matching string, which can be used intead of open coded variant. Signed-off-by:
-
- 21 May, 2018 2 commits
-
-
Takashi Iwai authored
Merge Xen para-virtualized frontend driver from Oleksandr Andrushchenko. Signed-off-by: -
Takashi Sakamoto authored
TC Electronic Digital Konnekt x32 is an application of WaveFront DiceII STD and doesn't support TCAT extended application protocol. For such devices, ALSA dice driver needs to have hard-coded parameters for stream formats. This commit adds stream format parameters for this model. Unfortunately, at sampling transmission frequencies of 88.2/96.0kHz, I confirmed that current ALSA dice driver doesn't drive the device appropriately due to detecting packet discontinuities. $ journalctl kernel: snd_dice fw1.0: Detect discontinuity of CIP: 90 80 At the frequencies, the device transfers 16 data blocks per packet and 16 data channels per data block, as a result one packet includes 1032 bytes if it's not NODATA. However, as long as I checked, the device often postpone packet transmission and continue with truncated payload than metadata in isochronous packet header. Below is a sample of sequence I got. sec cycle bytes CIP1 CIP2 37 3314 1032 0x01100090 0x900449E2 37 3315 8 0x011000A0 0x9004FFFF 37 3316 1032 0x011000A0 0x900461E2 37 3317 1032 0x011000B0 0x900475E2 37 3318 1032 0x011000C0 0x900489E2 37 3319 8 0x011000D0 0x9004FFFF 37 3320 1032 0x011000D0 0x9004A1E2 37 3321 1032 0x011000E0 0x9004B5E2 37 3322 1032 0x011000F0 0x9004C9E2 37 3323 8 0x01100000 0x9004FFFF 37 3324 1032 0x01100000 0x9004E1E2 37 3325 1032 0x01100010 0x9004F5E2 37 3326 1032 0x01100020 0x900409E2 37 3327 8 0x01100030 0x9004FFFF 37 3328 1032 0x01100030 0x900421E2 37 3329 1032 0x01100040 0x900435E2 37 3330 (skip) 37 3331 (skip) 37 3332 (skip) 37 3333 (skip) 37 3334 (skip) 37 3335 (skip) 37 3336 (skip) 37 3337 (skip) 37 3338 (skip) 37 3339 (skip) 37 3340 (skip) 37 3341 (skip) 37 3342 (skip) 37 3343 (skip) 37 3344 (skip) 37 3345 (skip) 37 3346 (skip) 37 3347 (skip) 37 3348 (skip) 37 3349 (skip) 37 3350 (skip) 37 3351 (skip) 37 3352 (skip) 37 3353 (skip) 37 3354 (skip) 37 3355 (skip) 37 3356 (skip) 37 3357 (skip) 37 3358 (skip) 37 3359 (skip) 37 3360 (skip) 37 3361 (skip) 37 3362 (skip) 37 3363 (skip) 37 3364 (skip) 37 3365 (skip) 37 3366 (skip) 37 3367 1032 0x01100050 0x900461E1 37 3368 1032 0x01100060 0x900475E1 37 3369 1032 0x01100070 0x9004A1E1 37 3370 1032 0x01100080 0x9004A1E1 but content of payload is truncated. 37 3371 (skip) 37 3371 1032 0x01100080 0x9004B5E0 detect discontinuity 37 3372 1032 0x01100090 0x9004C9E0 37 3373 1032 0x011000A0 0x9004E1E0 37 3374 1032 0x011000B0 0x9004F5E0 37 3375 1032 0x011000C0 0x900409E0 37 3376 1032 0x011000D0 0x900421E0 37 3377 1032 0x011000E0 0x900435E0 37 3378 1032 0x011000F0 0x900449DF 37 3379 8 0x01100000 0x9004FFFF 37 3380 1032 0x01100000 0x900461DF 37 3381 1032 0x01100010 0x900475DF 37 3382 1032 0x01100020 0x900489DF 37 3383 8 0x01100030 0x9004FFFF 37 3384 1032 0x01100030 0x9004A1DF 37 3385 1032 0x01100040 0x9004B5DF 37 3386 1032 0x01100050 0x9004C9DF 37 3387 8 0x01100060 0x9004FFFF I cannot confirm this quirks with Windows driver. ALSA dice driver has a cause if assumed differences between these two drivers are ways of timestampling to RX packets from the drivers to the device. I've already reported timestamping quirk of Dice-based devices and this might bring this issue. [alsa-devel] Dice packet sequence quirk and ALSA firewire stack in Linux 4.6 http://mailman.alsa-project.org/pipermail/alsa-devel/2016-May/107715.html Well, nevertheless, I enable ALSA dice driver to work at the frequencies. This may brings inconvenience to users but I expect developers and users to fix it. $ cd linux-firewire-utils/src $ python2 crpp < /sys/bus/firewire/devices/fw1/config_rom ROM header and bus information block ----------------------------------------------------------------- 400 040423bb bus_info_length 4, crc_length 4, crc 9147 404 31333934 bus_name "1394" 408 e0ff8112 irmc 1, cmc 1, isc 1, bmc 0, pmc 0, cyc_clk_acc 255, max_rec 8 (512), max_rom 1, gen 1, spd 2 (S400) 40c 00016604 company_id 000166 | 410 0c232c28 device_id 040c232c28 | EUI-64 000166040c232c28 root directory ----------------------------------------------------------------- 414 0006b6cb directory_length 6, crc 46795 418 03000166 vendor 41c 8100000a --> descriptor leaf at 444 420 17000030 model 424 8100000f --> descriptor leaf at 460 428 0c0087c0 node capabilities per IEEE 1394 42c d1000001 --> unit directory at 430 unit directory at 430 ----------------------------------------------------------------- 430 000476c2 directory_length 4, crc 30402 434 12000166 specifier id 438 13000001 version 43c 17000030 model 440 81000010 --> descriptor leaf at 480 descriptor leaf at 444 ----------------------------------------------------------------- 444 0006c490 leaf_length 6, crc 50320 448 00000000 textual descriptor 44c 00000000 minimal ASCII 450 54432045 "TC E" 454 6c656374 "lect" 458 726f6e69 "roni" 45c 63000000 "c" descriptor leaf at 460 ----------------------------------------------------------------- 460 000772b4 leaf_length 7, crc 29364 464 00000000 textual descriptor 468 00000000 minimal ASCII 46c 44696769 "Digi" 470 74616c4b "talK" 474 6f6e6e65 "onne" 478 6b747833 "ktx3" 47c 32000000 "2" descriptor leaf at 480 ----------------------------------------------------------------- 480 000772b4 leaf_length 7, crc 29364 484 00000000 textual descriptor 488 00000000 minimal ASCII 48c 44696769 "Digi" 490 74616c4b "talK" 494 6f6e6e65 "onne" 498 6b747833 "ktx3" 49c 32000000 "2" Signed-off-by:
-
- 18 May, 2018 6 commits
-
-
Dan Carpenter authored
The "entry" pointer is always non-NULL so this test for out of bounds won't work. Fixes: f1f0f330 ("ALSA: dice: add parameters of stream formats for models produced by TC Electronic") Signed-off-by:
-
Takashi Iwai authored
Pull the fixes for possible races in the resolution callback. Signed-off-by: -
Melvin Vermeeren authored
--nextPart3916812.EicPReet6m Content-Transfer-Encoding: 7Bit Content-Type: text/plain; charset="us-ascii" Mytek manufactures some equipment with DICE-based firewire ports. These devices contain old versions of DICE firmware which lacks detailed stream format reporting for all sampling clock modes. Building upon the recent work by Takashi Sakamoto, hard-coded parameters are added for the Stereo 192 DSD-DAC. When the device vendor and model match the coded parameters are copied into the stream format cache. Signed-off-by:
Melvin Vermeeren <mail@mel.vin> Reviewed-by:
-
Takashi Iwai authored
There are still many places calling the timer's hw.c_resolution callback without lock, and this may lead to some races, as we faced in the commit a820ccbe ("ALSA: pcm: Fix UAF at PCM release via PCM timer access"). This patch changes snd_timer_resolution() to take the timer->lock for avoiding the races. A place calling this function already inside the lock (from the notifier) is replaced with the snd_timer_hw_resolution() accordingly, as well as wrapping with the lock around another place calling snd_timer_hw_resolution(), too. Reported-by:
Ben Hutchings <ben.hutchings@codethink.co.uk> Signed-off-by:
-
Takashi Iwai authored
Instead of open-coding for getting the timer resolution, use the standard snd_timer_resolution() helper. The original code falls back to the callback function when the resolution is zero, but it must be always so when the callback function is defined. So this should be no functional change. Signed-off-by: -
Takashi Iwai authored
There multiple open-codes to get the hardware timer resolution. Make a local helper function snd_timer_hw_resolution() and call it from all relevant places. There is no functional change by this, just a preliminary work for the following timer resolution hardening patch. Signed-off-by:
-
