A latency offset variable was added to the port struct and a function to
set the latency offset.
The latency offset does nothing for now, but it will be later added to
the sink/source latency.
realpath() on OS X behaves GNUish and accepts NULL for resolved_name
only on 10.6 and higher. Older versions will crash, if resolved_name is
NULL.
All versions define PATH_MAX, though. Better play it safe and use the
generic PATH_MAX version of pa_realpath on Mac OS X systems.
Signed-off-by: Mihai Moldovan <ionic@ionic.de>
This makes pa_make_secure_dir() create any missing parent directories in
the given path as well. This is useful, for example, on a pristine
system with a clean $HOME that needs ~/.config/pulse/ to be created when
~/.config does not exist.
Specifying the volume when creating a new stream is not an
equivalent act as setting the volume with a volume control
application. When creating a new stream, stream-restore
shouldn't save the volume, but when changing the volume,
then saving it is ok. For example, when I say
"paplay --volume=10000 somefile.wav", I mean that I want the
new stream to have volume 10000. I don't mean that also
future paplay invocations (without the --volume option)
should have that same volume.
This patch effectively reverts
546bcf3f2f.
I doesn't make sense to require all callers of
pa_device_port_new() to create the hashmap themselves. There
are and there will be no cases where a port without any
profiles would be desired.
pacmd should allow unloading modules by name.
The command_unload() function was expanded to handle names while
unloading modules.
If there are multiple modules with the same name all
of them will be unloaded.
BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=48289
If ~/.pulse/ already exists we stick to it in order not to lose
configuration and so that pulse configuration may still be shared across
the network with different PA versions.
XDG_SESSION_COOKIE was supposed to be used for access control to the
session and should not be exposed to other users.
Replace usage of XDG_SESSION_COOKIE by XDG_SESSION_ID which is the right
choice here since it is actually a proper session ID that may be
published.
libsamplerate_resample() assumed that src_process() would
always consume the whole input buffer. That was an invalid
assumption leading to crashes.
This patch adds a leftover memchunk for storing any
non-consumed input. When pa_resampler_run() is called next
time, the leftover is prepended to the new input.
Changes in v3:
- Make the calculations in pa_resampler_result() and
pa_resampler_max_block_size() more readable and more
correct.
- Rework the leftover storing: instead of using a dedicated
buffer for it, store it in the beginning of remap_buf.
This can avoid some memory copying. (The idea was
suggested by Wang Xingchao.)
- Use a generic save_leftover() function instead of doing
the leftover copying in the resampler implementation.
- Use the leftover logic also with the speex and ffmpeg
resamplers.
[ed: dropped the speex bit since the API guarantees that
it will consume everything -- Arun]
Changes in v2:
- If add_leftover() is called with zero-length input while
the leftover length is non-zero, we don't try to acquire
the input memblock.
- Instead of taking a reference to the original input in
libsamplerate_resample(), we copy the leftover data to a
new memblock. This is done, because otherwise, if the
input is one of the internal buffers, the data can get
overwritten before reading it in add_leftover().
- Store add_leftover_buf size in bytes instead of samples
(more convenient, but less consistent with other code).
BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=47156
State which thread calls set_port in which scenario.
Reported-by: Jyri Sasha <oku@iki.fi>
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
Log in as user A, fast user switch to user B, let user B change
port, volume or mute status, then switch back to user A.
At this point we must make sure that the ALSA and PA volumes are
synchronised by writing to the ALSA mixer when the ALSA device
becomes available.
BugLink: https://bugs.launchpad.net/bugs/915035
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
Coverity thinks that expected_method_sig can be NULL when
it's dereferenced by pa_streq(). Adding assertions doesn't
hurt here (in my opinion), and that should get rid of the
warnings.
Add also an assertion for the sample spec validity. The
existing code already does crash in case of an invalid
sample spec, but the error would not be as obvious: the
crash would happen due to a divide-by-zero operation in
pa_frame_aligned().
Misbehaving clients can try to set a filter sink to output to
itself, leading to crashes later on. This patch protects us from that.
Thanks to Roman Beslik for testing and finding an error in the first
version of this patch.
Tested-by: Roman Beslik <rabeslik@gmail.com>
BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=44397
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
This causes problems with 24kHz audio (results in echoing)
when upscaling to 44.1kHz or 48kHz.
It can be reapplied when the optimisation works for all cases.
This reverts commit 8539fe9765.
When the runtime path gets long (which can happen on some NFS
mounts where $HOME is not just /home/$USER), it can grow
longer the 108 char limit imposed by sockaddr_un.sun_path.
This just calls realpath which should ultimately point into
/tmp in most cases and result in a much smaller path.
Only do this when we are adding on a name component to the
runtime path so creating the actual symlink will still get
the original, long name, but this shouldn't be a problem
as it never goes into the sockaddr_un.sun_path.
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=44680
As David points out, the previous commit made a couple of asserts
redundant (the XOR covers all cases that were previous tested for).
Remove these redundant commits now.
Commit 54cddc6ddf removed an assert that
looked redundant but was not. This commit reinstates it in a slightly
modified form. It is not stated as (a ^ b) instead of (!a || !b) in
order to make the condition more obvious.
For some reason, a badly behaving client was trying to set a NULL
port, which caused PulseAudio to crash. Add safeguards on two levels
just to be protected. (Also remove a redundant check.)
BugLink: https://bugs.launchpad.net/bugs/951273
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
While developing the new UI we had to ask ourselves the question of whether
"speakers" should be considered available when headphones are plugged in.
In most cases, they are not available and therefore we should list them
as such.
OTOH, we don't want unplugging the headphones to be considered an act of
wanting to use the speakers (the user might prefer HDMI), and there might
be line-outs that keeps the speakers from unmuting anyway. So, at this point,
I think the most reasonable would be to make the speakers have
PA_PORT_AVAILABLE_NO when headphones are plugged in and
PA_PORT_AVAILABLE_UNKNOWN when they are not. But we might want to revisit
this decision once we have the priority lists up and running.
The same reasoning applies for "Internal Mic", which should become unavailable
when any other mic is plugged in.
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
For volume control UIs to be able to show ports in inactive profiles,
expose all ports together with the card info. This includes updating
the protocol and the client API to show the connection between ports
and for which profiles the ports are relevant.
Update protocol to 26.
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
Our flist implementation suffers from the ABA problem
(see http://en.wikipedia.org/wiki/ABA_problem), causing PulseAudio
to crash very rarely, usually inside memblock operations.
By turning stored pointers into stored table indices, we have some
extra bits that we can use to store tag bits, which is a known
workaround for the ABA problem.
Buglink: https://bugs.launchpad.net/bugs/924416
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
This allows clients to get a "fake" sample space for compressed formats
that we can support. This should make size/time conversion for things
like calculating buffer attributes simpler.
These utility functions could be handy to clients.
pa_format_info_to_sample_spec_fake() isn't made public, but the return
value is changed to keep in sync with pa_format_info_to_sample_spec().
assuming RAND_MAX is around 1<<31, rand() >> 1 generates large numbers as
random volume data; these likely causes saturated sample values after
applying the volume function -- not a good test
