The analog-output path should be suppressed when there are more
specific paths available. Currently that usually doesn't happen. The
suppression can be done with the path subset detection, and this patch
makes that work (another approach would be to mark the elements as
required-absent, like analog-input does, but I like the subset
suppression more, because it requires less stuff in the configuration
files). The problem with listing the now-removed elements in
analog-output.conf was that if the sound card had e.g. a Speaker
element, then the switch behaviour was different between analog-output
and analog-output-speakers, so analog-output was not considered a
subset of analog-output-speakers.
BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=74609
We weren't writing out one character from the "OK" response, and the
"AT" part of the "+VGS" and "+VGM" commands was missing. Also, the spec
says that the command is terminated by only a CR and not an LF (probably
doesn't hurt, but let's adhere to the spec for now).
Parse the gain changed AT commands from the headset and fire 2 new
hooks as a result. The device will connect to those hooks and change the
source/sink volumes.
When the source/sink volume changes, set the gain on the microphone or
speaker respectively. Make sure we do nothing if the transport can not
handle the gain changes.
Add a simple native headset backend that implements support for the
blutooth HSP profile.
This allows pulseaudio to output audio to a Headset using the HSP profile.
Make the native backend the default.
Bash-completion 1.90 introduced support for on-demand loading
of completions. Install the completion file as 'pulseaudio' to match
the main command, and create symlinks as aliases for other supported
commands in order to support the new system.
There is no use in trying to load data in legacy format, if we
already know that there is no data at all.
Also clarify in the debug message whether there is invalid data
or no data at all.
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
So far the command name has been figured out by looking one or two
items back in the $words array, but I needed a way to figure out the
command given an arbitrary number of parameters. I was implementing
a command for removing devices from the device-manager database, and
the command would take a list of devices. Since the number of devices
that need to be completed can be arbitrarily large, the previous "look
one or two words back" approach didn't work.
This new approach is more verbose, but I think it's also easier to
follow. There's some duplication that would be easy to avoid by
merging some of the commands, but I decided to not do that, to make
it more obvious what the code does.
Apparently "WIBBLE" is just a test, and maybe the test was "How
long does it take until somebody notices a strange row in configure.ac
and tries to remove it", if so, the test result is "a little over
three years". :-)
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
We will just ignore the memblock if this happens. We already have
a check for this in the client library, so this one is just for
security reasons.
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
If somebody tries to push a non-frame-aligned memblock onto the
memblockq, then we should fail the write. Otherwise the daemon will
crash, see https://bugs.freedesktop.org/show_bug.cgi?id=77595
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
Since we don't allow lengths that are not frame aligned,
it does not make sense to allow indices that are not frame aligned
either.
Also, allowing such a thing to be added causes the daemon to crash
later instead (see https://bugs.freedesktop.org/show_bug.cgi?id=77595 ).
Also drop _se from assert (there is no side effect).
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
This fixes a bug in latency configuration. The wrong type in the cast
caused UINT64_MAX being not treated as special, so the configured
latency was set to UINT64_MAX usecs, which of course is absurdly huge
latency.
This should not have any effect on behaviour. The goal is to align
with the pattern that I think we should follow:
Object initialization:
- put() is the place to create references from other objects to the
newly created object. In this case, adding the transport to
discovery->transports was moved from new() to put, and adding the
transport to device->transports was moved from set_state() to
put().
Object destruction:
- unlink() undoes put() and removes all references from other objects
to the object being unlinked. In this case setting the
device->transports pointer to NULL was moved from set_state() to
unlink(), and setting the discovery->transports pointer to NULL was
moved from free() to unlink().
- free() undoes new(), but also calls unlink() so that object owners
don't need to remember to call unlink() before free().
This fixes getting the binary name in the Hurd, or any other port using
the GNU C library, but only in the case where the library is directly
linked to. Opening with dlopen will not work.
Change in v3: reorder header includes and definitions
Change in v2: use a weak reference to main, so that we
don't crash when main cannot be found.
This test broke when PA_MAX_INPUTS_PER_SINK was increased from 32 to 256.
Because we currently don't have time to figure out why, let's just set
NSTREAMS to 20 in the meantime.
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
The default maximum latency is 10 seconds, which is not good,
especially since the tunnel sink doesn't support rewinding. Due to the
lack of rewinding, e.g. volume changes take a long time with large
latencies.
It seems at some point the code migrated to use the entry_write calls,
but fill_db is still using the old syntax, causing the entry to be
invalid.
The crash happens when clean_up_db gets called, which then calls
entry_read, causing the crash.
Signed-off-by: Ricardo Salveti de Araujo <ricardo.salveti@canonical.com>
If the keyboard is unplugged, it looks like the kernel is reporting
back -ENODEV when trying to close the fd. This is probably a kernel
error, but still, it's better to complain than to crash.
Buglink: https://bugs.freedesktop.org/show_bug.cgi?id=80867
Reported-by: Stelios Bounanos
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
It is possible that the chosen active_port doesn't equal
new_data->active_port, using p->name is more accurate.
Please refer to sink_new_hook_callback()
Signed-off-by: Hui Wang <hui.wang@canonical.com>
Now a2dp and hsp sinks and sources will have different names which means that
applications and other modules can use sink/source to distinguish selected
profile.
Module module-device-restore uses sink/source name and port name as identifier,
so if different profiles have different names module-device-restore can store
volume settings for each profile.
So with this patch it is possible to configure different volume settings for
a2dp and hsp profiles.
This patch does not change port names so gnome applications will be happy.
Note that similar patch is needed also for bluez5, but I'm not using bluez5
so I cannot write or test it.
Signed-off-by: Pali Rohár <pali.rohar@gmail.com>
On a machine without fixed connecting audio devices like internal
microphone or internal speaker, and when there is no external audio
devices plugging in, the default source/sink is alsa_input/alsa_output
and there is no input devices/output devices listed in the gnome
sound-setting.
Under this situation, if we connect a bluetooth headset, the gnome
sound-setting will list bluez input/output devices, but they are not
active devices by default. This looks very weird that sound-setting
lists only one input device and one output device, but they are not
active. To change this situation, we add an argument, the policy is
if a new source/sink is connected and current default source/sink's
active_port is AVAILABLE_NO, we let the new added one switch to
default one.
BugLink: http://bugs.launchpad.net/bugs/1369476
Signed-off-by: Hui Wang <hui.wang@canonical.com>
Calling the callback while setting it up can make things
complicated for clients, as the callback can do arbitrarily
things.
In this case, a protocol error caused the srbchannel to be
owned by both the pstream and the native connection.
Now the read callback is deferred, making sure the callback
is called from a cleaner context where errors are handled
appropriately.
Reported-by: Tanu Kaskinen <tanu.kaskinen@linux.intel.com>
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
As a third parameter, add the number of samples to read/write in
every iteration. This will help slow CPUs.
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
This allow 'off' profile to be choosen when no other profile is available
which is considered better since it requires less resources than other
profiles.
Now that we have switched to using the mixer handle only,
there is no use for sending hctl handles around.
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
