Arun Raghavan [Mon, 3 Dec 2012 05:57:27 +0000 (11:27 +0530)]
alsa: Try to support non-standard rates in alsa-sink/source
We inadvertantly stopped supporting non-standard rates when the
passthrough work was done. This makes sure that if no standard rates are
supported, we try to fallback to whatever ALSA gives us.
Mikel Astiz [Wed, 28 Nov 2012 18:20:58 +0000 (19:20 +0100)]
bluetooth: Fix unacquired transports during sink resume
The sink can be resumed while the source is still in PA_SOURCE_INIT.
This is the case if a module such as module-stream-restore routes the
audio to the sink during pa_sink_put(), leading to an inconsistent
state: the sink stays RUNNING but the transport is not actually
acquired.
Mikel Astiz [Mon, 3 Dec 2012 10:03:59 +0000 (11:03 +0100)]
bluetooth: Request headset audio during profile switch
When a headset is having a profile switch, we can either leave the
SCO state unmodified (as it was before this patch) or we can
alternatively request it (as older versions of PA).
This patch tries to avoid a potential regression in case a module
such as module-suspend-on-idle is not present, due to the provided
resume-on-running policy. Without this patch, and without such a policy,
the sink and sources would stay suspended until the user manually
performed another profile switch (i.e. hsp->off->hsp).
There are several other solutions currently being discussed as a longer
term solution, some of which require extendind the core. This patch is
therefore proposed as a short-term workaround to avoid the regression.
Mikel Astiz [Mon, 3 Dec 2012 10:03:58 +0000 (11:03 +0100)]
bluetooth: Do not setup stream before thread starts
bt_transport_acquire() might get called from the main thread, in case
the IO thread hasn't been started yet. In this case, we should not call
setup_stream() since this is going to be called in the beginning of
thread_func().
Mikel Astiz [Mon, 3 Dec 2012 10:03:57 +0000 (11:03 +0100)]
bluetooth: Trivially refactor to call setup_stream() directly
If the transport is already acquired and the stream needs to be started,
call setup_stream() directly instead of bt_transport_acquire(u, TRUE).
Both calls are identical in these conditions, with the exception of the
log trace which has now been moved to setup_stream().
Mikel Astiz [Mon, 26 Nov 2012 17:32:08 +0000 (18:32 +0100)]
bluetooth: Disable profile auto-switch policy for headsets
Given that headsets have just one single port exposing whether the
audio is streaming (playing) or not, it's not possible that
module-bluetooth-policy would distinguish A2DP/HSP cases, and thus
the automatic selection of the card profile is not deterministic.
For this reason, disable the policy entirely for headsets and focus
only on HFGW and A2DP source profiles.
Mikel Astiz [Mon, 26 Nov 2012 17:32:07 +0000 (18:32 +0100)]
bluetooth: Merge headset ports into one
Merge the former "hsp-output" and "a2dp-output" ports into one single
port, in order to fix the regression of having several independent
entries in the UI.
Niels Ole Salscheider [Sat, 24 Nov 2012 11:32:51 +0000 (12:32 +0100)]
virtual-surround: check if resampled memblock is not equal to input
Since commit e32a408b3cdd46857fdf12210c1bf5bdbf3a96f8, we silence the
input memblock in order to give the resampler enough input samples, if
necessary.
But if there is no need to resample the hrir, the resampled memblock is
actually the same as the input memblock. Thus, we have to make sure that
we do not silence it in this case.
Tanu Kaskinen [Thu, 22 Nov 2012 14:20:28 +0000 (15:20 +0100)]
bluetooth: Unload device module when no audio profiles connected
Without this patch, device modules will be left around after the
device has been disconnected and when they are reconnected, the
discovery module will load duplicate device module instances.
Mikel Astiz [Thu, 22 Nov 2012 14:20:27 +0000 (15:20 +0100)]
bluetooth: Run the discovery hook only when necessary
This is a minor optimization too, but the main benefit is that it's
makes the code easier to understand (I hope), since run_callback()
won't be called at times when it's not needed.
The new helper function makes it easier to check whether any audio
profiles are connected. That information is needed by the discovery
module for deciding whether a new device module should be loaded. The
device module should use this information too to unload itself at the
right time, but that's currently not implemented.
Mikel Astiz [Thu, 22 Nov 2012 14:20:25 +0000 (15:20 +0100)]
bluetooth: Rename former device_is_audio()
Use a more accurate name for the function since it doesn't just check
if it is an audio device (which can be detected quite early), but it
also checks if the most relevant properties (device info, etc.) have
been received.
Besides, add the const qualifier to the pointer since it's not going to
be modified.
Tanu Kaskinen [Thu, 22 Nov 2012 14:20:24 +0000 (15:20 +0100)]
bluetooth: Ignore Device.Connected
The Device.Connected was only used for tracking whether a device module
should be loaded, but that information is already included in the
individual profile state properties. The property can therefore be
completely ignored without any loss in functionality.
Arun Raghavan [Mon, 19 Nov 2012 07:22:28 +0000 (12:52 +0530)]
sink-input, source-output: Check rate update success for passthrough
This makes sure we don't try to plug in a passthrough stream if the
final sink/source sample spec doesn't match what we want. In the future,
we might want to change rate updates to try a full sample spec update
for passthrough streams.
Arun Raghavan [Sat, 17 Nov 2012 05:25:05 +0000 (10:55 +0530)]
build-sys: Drop ChangeLog generation
This drops ChangeLog generation from git. It does not make sense to
distribute 500 kB of ChangeLog given how easy it is these days to browse
history with git.
We might replace this with a release-annoucement-esque high-level
summary of changes.
Arun Raghavan [Fri, 16 Nov 2012 17:39:15 +0000 (23:09 +0530)]
modules: Micro-optimisation for rewind_requested paths
Since some devices can be chatty with regards to how often they return
from poll(), this adds a PA_UNLIKELY() to all the the rewind_requested
checks in our sink modules to make the general case (no rewind was
requested) the fast path.
Tanu Kaskinen [Thu, 30 Aug 2012 13:50:13 +0000 (16:50 +0300)]
sink: Process rewind requests also when suspended.
When a rewind is requested on a sink input, the request parameters are
stored in the pa_sink_input struct. The parameters are reset during
rewind processing, and if the sink decides to ignore the rewind
request due to being suspended, stale parameters are left in
pa_sink_input. It's particularly problematic if the rewrite_bytes
parameter is left at -1, because that will prevent all future rewind
processing on that sink input. So, in order to avoid stale parameters,
every rewind request needs to be processed, even if the sink is
suspended.
mainloop: Don't care about the mainloop state variable when waking up the mainloop.
If the mainloop is just about to enter polling, but m->state
is not POLLING yet when some other thread calls
pa_mainloop_wakeup(), the mainloop will not be woken up.
It's safe to write to the wakeup pipe at any time, so let's
just remove the check.
Tanu Kaskinen [Wed, 7 Nov 2012 14:52:37 +0000 (16:52 +0200)]
pulse: Fix hole handling in pa_stream_peek().
Previously, if there was a hole in a recording stream,
pa_stream_peek() would crash. Holes could be handled silently inside
pa_stream_peek() by generating silence (wouldn't work for compressed
streams, though) or by skipping any holes. However, I think it's
better to let the caller decide how the holes should be handled, so
in case of holes, pa_stream_peek() will return NULL data pointer and
the length of the hole in the nbytes argument.
This change is technically an interface break, because previously the
documentation didn't mention the possibility of holes that need
special handling. However, since holes caused crashing anyway in the
past, it's not a regression if applications keep misbehaving due to
not handing holes properly.
Some words about when holes can appear in recording streams: I think
it would be reasonable behavior if overruns due to the application
reading data too slowly would cause holes. Currently that's not the
case - overruns will just cause audio to be skipped. But the point is
that this might change some day. I'm not sure how holes can occur
with the current code, but as the linked bug shows, they can happen.
It's most likely due to recording from a monitor source where the
thing being monitored has holes in its playback stream.
alsa: get avail, delay, timestamps in a single kernel call
Refactor code to fetch avail, delay and timestamp values
in a single call to snd_pcm_status().
The information reported is exactly the same as before,
however it is extracted in a more atomic manner to
improve timer-based scheduling.
Rethinking this, it makes more sense to not add this to the check
framework. This is mostly useful for exposing ALSA driver issues, and
it's handy to be able to build this as a standalone executable.
Frédéric Dalleau [Wed, 22 Aug 2012 13:42:16 +0000 (15:42 +0200)]
resampler: Fix crash if 'auto' resampler chooses ffmpeg with variable rate
To reproduce, add resampler-method = ffmpeg in daemon.conf
then start PA, and load module-loopback
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0xb2f1db40 (LWP 23047)]
0x00000000 in ?? ()
(gdb) bt
0 0x00000000 in ?? ()
1 0xb7c463cb in pa_resampler_set_input_rate (r=0x80e9438, rate=44011) at pulsecore/resampler.c:365
2 0xb7c6321d in pa_sink_input_process_msg (o=0x80e87a0, code=3, userdata=0xabeb, offset=0, chunk=0x0)
at pulsecore/sink-input.c:1833
3 0xb7e9840b in sink_input_process_msg_cb (obj=0x80e87a0, code=3, data=0xabeb, offset=0, chunk=0x0)
at modules/module-loopback.c:538
4 0xb7c2709b in pa_asyncmsgq_dispatch (object=0x80e87a0, code=3, userdata=0xabeb, offset=0, memchunk=0xb2f1d17c)
at pulsecore/asyncmsgq.c:322
5 0xb7c4c6e3 in asyncmsgq_read_work (i=0x80dd580) at pulsecore/rtpoll.c:564
6 0xb7c4b34a in pa_rtpoll_run (p=0x80fb7e0, wait_op=true) at pulsecore/rtpoll.c:238
7 0xb7dd90af in thread_func (userdata=0x80afe88) at modules/alsa/alsa-sink.c:1785
8 0xb7bf3291 in internal_thread_func (userdata=0x8095d08) at pulsecore/thread-posix.c:83
9 0xb7ab9d4c in start_thread (arg=0xb2f1db40) at pthread_create.c:308
10 0xb79f3ace in clone () at ../sysdeps/unix/sysv/linux/i386/clone.S:130
Tanu Kaskinen [Sat, 6 Oct 2012 12:05:08 +0000 (15:05 +0300)]
build: Add .gitignore files to EXTRA_DIST.
As a packager, I want to be able to do "git format-patch
v3.0..origin/master" and I want the generated patches to apply cleanly
against the 3.0 tarball. The patches don't currently apply cleanly if
there are any changes to the .gitignore files, because the .gitignore
files are not shipped in the tarball. This patch fixes that problem.
core: adjust playing_for and underrun_for at rewind
A rewind may erase data that sink_input counted in playing_for or
underrun_for earlier. Add code adjusting those values after a rewind.
One visible symptom of this bug was problems recovering from an
underrun. When a client calls pa_stream_write() with a large block of
memory, the function can split that into smaller pieces before sending
it to the server. When receiving new data for a stream that had
silence queued due to underrun, the server would do a rewind to
replace the queued-but-not-played silence with the new data. Because
of the bug, this rewind itself would not change underrun_for. It's
possible for multiple rewinds to be done without filling the sink
buffer in between (which is what would eventually reset underrun_for).
In this case, the server rapidly processing the split packets would
rewind the stream for _each_ of them (as underrun_for would stay set),
erasing valid audio as a result.
Arun Raghavan [Tue, 30 Oct 2012 06:34:42 +0000 (12:04 +0530)]
svolume: Fix ARM alignment issues
As Peter Meerwald <p.meerwald@bct-electronic.com> discovered, our ARM
svolume code performance is quite terrible when the incoming samples are
not word-aligned. This can very easily be the case, since the
architecture only requires that the samples be 16-bit aligned, and we
might end up running the innermost loop after processing modulo-4
samples. The performance degradation was ~50x on a Cortex A9
(Pandaboard).
This reworks the svolume logic to first consume enough samples to make
sure the rest is word aligned, and reordering the processing to work
with 4 samples at a time first, and then finally deal with the
remainder.
With this, performance is comparable for arbitrary alignments (~3x
faster than the C code).
match: Use the SINK_INPUT_FIXATE hook instead of NEW.
The callback relies on the sample spec being finalized, which is not
true with the NEW hook.
In case you're wondering about the "hook EARLY - 1, to match before
stream-restore" comment that was not changed even though the code that
the comment concerned was changed: the comment was apparently written
at a time when module-stream-restore used the NEW hook too, and later
stream-restore has been changed to use the FIXATE hook. So, the
comment was wrong/nonsensical before this patch. Since these two
modules now use the same hook again, the comment makes sense again.
combine: Keep the timer active in the null mode only when running.
Previously thread_func() used PA_SINK_IS_OPENED() to check whether
some data should be rendered. process_render_null() used a different
check: it would return immediately if the sink was not in the RUNNING
state. This caused a busy loop when the sink was in the IDLE state,
because process_render_null() didn't update the timestamp, and
thread_func() still kept the timer active using the old timestamp.
pa_rtpoll_run() would return immediately because of the old timestamp.
This is fixed by using the same check in both thread_func() and
process_render_null(). Since the checks are the same, it's actually
redundant to have the check in process_render_null(), so it is now an
assertion.
Tanu Kaskinen [Wed, 22 Aug 2012 06:00:25 +0000 (09:00 +0300)]
virtual-surround-sink: Fix setting max_request and max_rewind.
The sink has different frame size than the sink input, so
the max_request and max_rewind values of the sink input need
to be converted when setting the sink max_request and
max_rewind values.
The conversion is already done correctly in
sink_input_update_max_request_cb() and
sink_input_update_max_rewind_cb().
Tanu Kaskinen [Tue, 14 Aug 2012 14:08:35 +0000 (17:08 +0300)]
loopback: Don't fix the source output format/rate/channels.
Once the sink input has been routed in pa_sink_input_new(),
the sample spec and channel map have already become fixed.
The sink input and source output must use the same stream
format, because the data is copied as-is.
Tanu Kaskinen [Tue, 14 Aug 2012 14:08:34 +0000 (17:08 +0300)]
loopback: Use the real sample spec once it's known.
When module-loopback is loaded without arguments, the ss and
map variables are initialized with dummy values. This caused
a problem, because also pa_memblockq_new() was called with
the dummy values, making it work incorrectly. The base was
set to 1 instead of the real frame size, which in turn
caused alignment related crashes.
resampler: Make sure that there are no overflows when multiplying potentially big numbers.
This fixes at least one crash that has been observed. The
multiplication in trivial_resample() overflowed when
resampling from 96 kHz to 48 kHz, causing an assertion
error:
Assertion 'o_index * fz < pa_memblock_get_length(output->memblock)' failed at pulsecore/resampler.c:1521, function trivial_resample(). Aborting.
Without the assertion, the memcpy() after the assertion
would have overwritten some random heap memory.
Thomas Martitz [Tue, 30 Oct 2012 10:35:26 +0000 (11:35 +0100)]
core-util: Don't error out on existing runtime directory.
When compiling without HAVE_SYMLINK the runtime dir is a real directory,
which is attempted to be created. In the case it already exists we shouldn't
error out. The HAVE_SYMLINK-enabled code already does this.
Thomas Martitz [Tue, 30 Oct 2012 10:35:24 +0000 (11:35 +0100)]
core: Proper poll() emulation to fix pacat and friends on Windows
Currently, Windows versions of pacat and friends fail because the current
poll emulation is not sufficient (it only works for socket fds).
Luckily Gnulib has a much better emulation that seems to work good enough.
The implementation has been largely copied (except a few bug fix
regarding timeout handling, to be pushed upstream) and works on pipes
and files as well. The copy has been obtained through their gnulib-tool utility,
which gives a LGPLv2.1+ licensed file.
This fixes the "Assertion (!e->dead) failed" error coming and lets pacat
and friends stream happily to/from a server (I didn't actually test parec).
Arun Raghavan [Fri, 26 Oct 2012 11:09:11 +0000 (16:39 +0530)]
tests: Allow off-by-one error in sconv test
With some optimised sconv implementations (read NEON), rounding
inaccuracy might lead to a difference of 1 with the reference
implementation. The inaccuracy is worth the performance gain.
Also increases floating-point accuracy while printing errors to make
errors easier to analyse.
Arun Raghavan [Fri, 26 Oct 2012 11:06:49 +0000 (16:36 +0530)]
sconv: Fix NEON sconv rounding code
Rounding with 0.5 causes us to always round up for any value of the form
x.5. IEEE754 specifies round-to-nearest-even as the behaviour in this
case. This might not always be possible with NEON code, but this change
gets us much closer to it.
Peter Meerwald [Tue, 23 Oct 2012 15:54:57 +0000 (17:54 +0200)]
core: Add ARM NEON optimized sample conversion code
final:
* includes some minor style fixes and build-time changes to allow
building a single binary for neon and non-neon systems
v4:
* fix for sample length < 4
v3:
* convert from intrinsics to inline assembly
v2:
* load and store data with vld1/vld1q and vst1/vst1q, resp., to work
around alignment issues of compiler-generated vldmia instruction
* remove redundant check for NEON flags
Arun Raghavan [Fri, 26 Oct 2012 03:11:08 +0000 (08:41 +0530)]
build-sys: Add volume code to libpulsecommon
Commit dd31d652a ("utils: Adding a function to get volume from string")
uses pa_sw_volume_from_dB(), which is part of libpulse, in libpulsecore.
This breaks as-needed builds. We fix this by also building the volume
code in libpulsecommon.
Mikel Astiz [Tue, 23 Oct 2012 14:41:59 +0000 (16:41 +0200)]
bluetooth: Add hook to announce late UUIDs
UUIDs might be announced at any time, so a hook is needed to notify any
interested module. In practice, the UUIDs are quite stable with the
exception of the pairing procedure, where the UUIDs are reported by
BlueZ as soon as they are discovered.
Arun Raghavan [Wed, 24 Oct 2012 07:18:02 +0000 (12:48 +0530)]
tests: Run sconv tests with multiple alignments
This allows us to test the sconv code with the incoming samples at
various byte alignments. The test is also now split into correctness and
performance checks.
David Henningsson [Tue, 16 Oct 2012 09:35:00 +0000 (11:35 +0200)]
alsa-mixer: Don't let "Mic Jack Mode" alone create a "Line In" path
The IDT/Sigmatel codec driver often creates a "Mic Jack Mode" for
every mic jack, so it can change functionality between Mic and Line In.
However, as the "Mic Jack" is the standard naming, our current solution
does not make the Line In port unavailable when nothing is plugged in.
This patch makes the "Line In" port not to be created just because there
is a "Mic Jack Mode" that could be set to "Line". This makes the behaviour
consistent with e g "Dock Mic Jack Mode", "Front Mic Jack Mode" etc, where
we don't create a "Dock Line" or "Dock Mic" port either.
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
Arun Raghavan [Tue, 23 Oct 2012 12:56:39 +0000 (18:26 +0530)]
tests: Reorganise cpu-test to reuse code
This factors out the basic measurement code for each test into a
separate block so that each test can be broken down into a basic
correctness test, and a performance comparison with minimum effort.
Thomas Martitz [Tue, 23 Oct 2012 10:10:45 +0000 (12:10 +0200)]
gccmacro: Disable printf-like format checking on mingw32 compilers.
Am 23.10.2012 08:25, schrieb Arun Raghavan:
> On Tue, 2012-08-21 at 13:32 +0200, Thomas Martitz wrote:
>> Am 21.08.2012 08:51, schrieb Rémi Denis-Courmont:
>>> Le mardi 21 août 2012 00:50:34 Thomas Martitz, vous avez écrit :
>>>> There are tons of warnings, most of them because the function is not
>>>> recognized as printf-like.
>>> Removing checks looks very fishy.
>>>
>>> To use C99 and/or GNU format specifiers on MingW, you need to use the
>>> gnuprintf attribute instead of printf. With printf, the format string is
>>> validated according to the antiquated MSVC rules.
>>>
>> Interesting, I didn't know about gnuprintf. FWIW, what are those
>> antiquated MSVC rules? I assumed the return value which isn't int for
>> some affected functions?
> Is this one going to be respun?
>
Yes, here you go.
>From c5f15eec69bf95c9a1261e0d82abbd039156e75e Mon Sep 17 00:00:00 2001
From: Thomas Martitz <kuge@rockbox.org>
Date: Wed, 8 Aug 2012 17:38:04 +0200
Subject: [PATCH 1/3] gccmacro: Work around warnings due to printf redirection
by libintl.
Libintl defines printf as libintl_printf, which breaks the format
attribue. Unfortunately the workaround around provided by libintl
is only enabled for cygwin, but not for mingw builds. Therefore
install the workaround manually.
Peter Meerwald [Mon, 22 Oct 2012 21:48:27 +0000 (23:48 +0200)]
echo-cancel: Fix false warning in webrtc AEC.
CXX libwebrtc_util_la-webrtc.lo
modules/echo-cancel/webrtc.cc: In function 'pa_bool_t pa_webrtc_ec_init(pa_core*, pa_echo_canceller*, pa_sample_spec*, pa_channel_map*, pa_sample_spec*, pa_channel_map*, uint32_t*, const char*)':
modules/echo-cancel/webrtc.cc:196:9: warning: 'rm' may be used uninitialized in this function [-Wuninitialized]
Signed-off-by: Peter Meerwald <p.meerwald@bct-electronic.com>