Browse Source

Fix xrun_recovery in ALSA clients.

According to the ALSA documentation, snd_pcm_readi() returns -ESTRPIPE
when a suspend event occured. But xrun_recovery() checks for -EAGAIN
instead.

If the ALSA client is connected to the ALSA loopback device, it leads to
an infinite loop in the client when the computer is resumed from sleep.
The client takes up to 99% of the CPU resources.

The problem is fixed by a proper error checking of snd_pcm_readi().

A word of caution however: on kernels prior to 3.8, it will trigger a
bug in the snd-aloop driver, and may lead to a kernel oops.
The kernel fix can be found there:
http://www.spinics.net/lists/stable-commits/msg23379.html

Signed-off-by: elboulangero <elboulangero@gmail.com>
tags/0.124.0
elboulangero Adrian Knoth 12 years ago
parent
commit
2a400f4214
2 changed files with 2 additions and 2 deletions
  1. +1
    -1
      alsa_in.c
  2. +1
    -1
      alsa_out.c

+ 1
- 1
alsa_in.c View File

@@ -112,7 +112,7 @@ static int xrun_recovery(snd_pcm_t *handle, int err) {
if (err < 0)
printf("Can't recover from underrun, prepare failed: %s\n", snd_strerror(err));
return 0;
} else if (err == -EAGAIN) {
} else if (err == -ESTRPIPE) {
while ((err = snd_pcm_resume(handle)) == -EAGAIN)
usleep(100); /* wait until the suspend flag is released */
if (err < 0) {


+ 1
- 1
alsa_out.c View File

@@ -111,7 +111,7 @@ static int xrun_recovery(snd_pcm_t *handle, int err) {
if (err < 0)
printf("Can't recovery from underrun, prepare failed: %s\n", snd_strerror(err));
return 0;
} else if (err == -EAGAIN) {
} else if (err == -ESTRPIPE) {
while ((err = snd_pcm_resume(handle)) == -EAGAIN)
usleep(100); /* wait until the suspend flag is released */
if (err < 0) {


Loading…
Cancel
Save