Sometimes PORT_EXIT messages are lost when a process is exiting.
This happens if you subscribe to the announce port with client A,
then subscribe to the announce port with client B, then kill client A.
Client B will not see the PORT_EXIT message because client A's port is
closing and is earlier in the announce port subscription list. The
for each loop will try to send the announcement to client A and fail,
then will stop trying to broadcast to other ports. Killing B works fine
since the announcement will already have gone to A. The CLIENT_EXIT
message does not get lost.
How to reproduce problem:
*** termA
$ aseqdump -p 0:1
0:1 Port subscribed 0:1 -> 128:0
*** termB
$ aseqdump -p 0:1
*** termA
0:1 Client start client 129
0:1 Port start 129:0
0:1 Port subscribed 0:1 -> 129:0
*** termB
0:1 Port subscribed 0:1 -> 129:0
*** termA
^C
*** termB
0:1 Client exit client 128
<--- expected Port exit as well (before client exit)
Signed-off-by: Adam Goode <agoode@google.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Currently mutex_destroy() is called in module's cleanup function. But after
cleaned up, this mutex is automatically released. So this function call
is meaningless.
[fixed a typo in changelog by tiwai]
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The constants of enum snd_efw_grp_type is for struct snd_efw_phys_grp.type.
But this member is 1 byte. Although the value is between 0x00-0xff, a constant
has 0x10000. This constant is meaningless.
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
To reverse a pointer for the ring buffer, subtraction by buffer
size is better than assignment to the beginning of the buffer.
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
All assignment for local variables in these functions are not related to
critical section.
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
snd_seq_event_dup returns -ENOMEM in some buffer-full conditions,
but usually returns -EAGAIN. Make -EAGAIN trigger the overflow
condition in snd_seq_fifo_event_in so that the fifo is cleared
and -ENOSPC is returned to userspace as stated in the alsa-lib docs.
Signed-off-by: Adam Goode <agoode@google.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The commit [e1d4d3c8: ASoC: free jack GPIOs before the sound card is
freed] introduced snd_soc_card remove callbacks to a few drivers, but
they are implemented with a wrong argument type. The callback should
receive snd_soc_card pointer instead of snd_soc_pcm_runtime.
Fixes: e1d4d3c854 ('ASoC: free jack GPIOs before the sound card is freed')
Acked-by: Mark Brown <broonie@linaro.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
ASoC: Final updates for v3.16
A few more updates from the last week of development, nothing too
exciting. Highlights include:
- GPIO descriptor support for jacks
- More updates and fixes to the Freescale SSI, Intel and rsnd drivers.
- New drivers for Analog Devices ADAU1361, ADAU1381, ADAU1761 and
ADAU1781, and Realtek RT5677.
This is the same change as commit fb6b8e7144 "ASoC: tegra: free jack
GPIOs before the sound card is freed", but applied to all other ASoC
machine drivers where code inspection indicates the same problem exists.
That commit's description is:
==========
snd_soc_jack_add_gpios() schedules a work queue item to poll the GPIO to
generate an initial jack status report. If sound card initialization
fails, that work item needs to be cancelled, so it doesn't run after the
card has been freed. Specifically, freeing the card calls
snd_jack_dev_free() which calls snd_jack_dev_disconnect() which sets
jack->input_dev = NULL, and input_dev is used by snd_jack_report(), which
is called from the work queue item.
snd_soc_jack_free_gpios() cancels the work item. The Tegra ASoC machine
drivers do call this function in the platform driver remove() callback.
However, this happens after the sound card is freed, at least when the
card is freed due to errors late during snd_soc_instantiate_card(). This
leaves a window where the work item can execute after the card is freed.
In next-20140522, sound card initialization does fail for unrelated
reasons, and hits the problem described above.
To solve this, fix the Tegra ASoC machine drivers to clean up the Jack
GPIOs during the snd_soc_card's .remove() callback, which is executed
before the overall card object is freed. also, guard the cleanup call
based on whether we actually setup up the GPIOs in the first place.
Ideally, we'd do the cleanup in a struct snd_soc_dai_link .fini/remove
function to match where the GPIOs get set up. However, there is no such
callback.
==========
Note that I have not even compile-tested this in most cases, since most
of the drivers rely on specific mach-* support I don't have enabled, and
don't support COMPILE_TEST. Testing by the relevant board maintainers
would be useful.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Mark Brown <broonie@linaro.org>
Kailang Yang
Adam Goode
Takashi Sakamoto
Takashi Iwai
Stephen Warren
Mark Brown