kernel/git/stable/linux.git/sound/oss, branch linux-3.14.y

sound/oss: fix deadlock in sequencer_ioctl(SNDCTL_SEQ_OUTOFBAND)

2015-05-17T16:53:51+00:00

commit bc26d4d06e337ade069f33d3f4377593b24e6e36 upstream. A deadlock can be initiated by userspace via ioctl(SNDCTL_SEQ_OUTOFBAND) on /dev/sequencer with TMR_ECHO midi event. In this case the control flow is: sound_ioctl() -> case SND_DEV_SEQ: case SND_DEV_SEQ2: sequencer_ioctl() -> case SNDCTL_SEQ_OUTOFBAND: spin_lock_irqsave(&lock,flags); play_event(); -> case EV_TIMING: seq_timing_event() -> case TMR_ECHO: seq_copy_to_input() -> spin_lock_irqsave(&lock,flags); It seems that spin_lock_irqsave() around play_event() is not necessary, because the only other call location in seq_startplay() makes the call without acquiring spinlock. So, the patch just removes spinlocks around play_event(). By the way, it removes unreachable code in seq_timing_event(), since (seq_mode == SEQ_2) case is handled in the beginning. Compile tested only. Found by Linux Driver Verification project (linuxtesting.org). Signed-off-by: Alexey Khoroshilov Signed-off-by: Takashi Iwai Cc: Willy Tarreau Signed-off-by: Greg Kroah-Hartman

sound: oss: remove last sleep_on users

2014-01-14T15:12:07+00:00

There are three files in oss for which I could not find an easy way to replace interruptible_sleep_on_timeout with a non-racy version. This patch instead just adds a private implementation of the function, now named oss_broken_sleep_on, and changes over the remaining users in sound/oss/ so we can remove the global interface. [fixed coding style warnings by tiwai] Signed-off-by: Arnd Bergmann Signed-off-by: Takashi Iwai

sound: oss: dmasound: kill SLEEP() macro to avoid race

2014-01-14T15:12:07+00:00

The use of interruptible_sleep_on_timeout in the dmasound driver is questionable and we want to kill off all sleep_on variants. This replaces the calls with wait_event_interruptible_timeout where possible, to wait for a particular event instead of blocking in a racy way. In the sq_write function, the easiest solution is an open-coded prepare_to_wait loop. Signed-off-by: Arnd Bergmann Signed-off-by: Takashi Iwai

sound: oss: midibuf: fix sleep_on races

2014-01-14T15:11:44+00:00

sleep_on is known to be racy and going away because of this. All instances of interruptible_sleep_on and interruptible_sleep_on_timeout in the midibuf driver can trivially be replaced with wait_event_interruptible and wait_event_interruptible_timeout. [fixed coding style warnings by tiwai] Signed-off-by: Arnd Bergmann Signed-off-by: Takashi Iwai

sound: oss: vwsnd: avoid interruptible_sleep_on

2014-01-14T15:01:27+00:00

Interruptible_sleep_on is racy and we want to remove it. This replaces the use in the vwsnd driver with an open-coded prepare_to_wait loop that fixes the race between concurrent open() and close() calls, and also drops the global mutex while waiting here, which restores the original behavior that was changed during the BKL removal. Signed-off-by: Arnd Bergmann Signed-off-by: Takashi Iwai

sound: oss: msnd_pinnacle: avoid interruptible_sleep_on_timeout

2014-01-14T15:01:20+00:00

We want to remove all sleep_on variants from the kernel because they are racy. In case of the pinnacle driver, we can replace interruptible_sleep_on_timeout with wait_event_interruptible_timeout by changing the meaning of a few flags used in the driver so they are cleared at wakeup time, which is a somewhat more appropriate way to do the same, although probably still racy. Signed-off-by: Arnd Bergmann Signed-off-by: Takashi Iwai

sound: Remove unnecessary semicolons

2013-10-14T09:00:05+00:00

These aren't necessary after switch and if blocks. Signed-off-by: Joe Perches Signed-off-by: Takashi Iwai

Merge tag 'asoc-v3.12' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound into for-next

2013-08-23T12:12:22+00:00

ASoC: Updates for v3.12 - DAPM is now mandatory for CODEC drivers in order to avoid the repeated regressions in the special cases for non-DAPM CODECs and make it easier to integrate with other components on boards. All existing drivers have had some level of DAPM support added. - A lot of cleanups in DAPM plus support for maintaining controls in a specific state while a DAPM widget all contributed by Lars-Peter Clausen. - Core helpers for bitbanged AC'97 reset from Markus Pargmann. - New drivers and support for Analog Devices ADAU1702 and ADAU1401(a), Asahi Kasei Microdevices AK4554, Atmel AT91ASM9x5 and WM8904 based machines, Freescale S/PDIF and SSI AC'97, Renesas R-Car SoCs, Samsung Exynos5420 SoCs, Texas Instruments PCM1681 and PCM1792A and Wolfson Microelectronics WM8997. - Support for building drivers that can support it cross-platform for compile test.

sound: oss/dmabuf: remove an unneeded temporary variable

2013-08-09T10:38:33+00:00

We don't actually use the "go" variable so it can be removed. Signed-off-by: Dan Carpenter Signed-off-by: Takashi Iwai

sound: oss/vwsnd: Always define vwsnd_mutex

2013-07-15T10:07:58+00:00

While the conversion of BKL to mutex in commit 645ef9ef, the mutex definition was put in a wrong place inside #ifdef WSND_DEBUG, which leads to the build error. Just move it outside the ifdef. Reported-by: Fengguang Wu Signed-off-by: Takashi Iwai