Dan McCombs discovered that snd_pcm_close() invocations are not leading
to associated timers being closed, which results in successively more
timers being created but not freed.
Original patch from Daniel T Chen <crimsun@ubuntu.com>.
BugLink: https://bugs.launchpad.net/bugs/451893
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
Call the slave snd_pcm_info() as long as possible in the direct plugins
(i.e. when the PCM device could be opened with O_APPEND mode).
This allows dmix/dsnoop as a salve for PCM hook controls.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
As the result of snd_pcm_delay() is affected not only by hw_ptr
and appl_ptr, but also by 'runtime->delay' property,
either SNDRV_PCM_IOCTL_DELAY or SNDRV_PCM_IOCTL_STATUS ioctl
must be used to get the correct result.
Previously 'runtime->delay' was ignored in case 'hw->sync_ptr'
was used.
Signed-off-by: Kai Vehmanen <kvehmanen@eca.cx>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
In snd_pcm_hw_params_choose(), set the buffer size before the period
size and time as default. This will give more useful configuration for
most of apps, i.e. larger buffer size.
For apps that require the old behavior, now the function checks the
environment variable $LIBASOUND_COMPAT. If this variable is set to
non-empty, the hw_params is determined in the old way, first period
then buffer sizes.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
When snd_pcm_start() is called in the invalid state, it should return
-EBADFD. But ioplug plugin returns -EAGAIN. Let's fix it.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
In the normal mmap mode, the stream isn't started automatically even after
the data >= start_threshold has been written. However, in the
mmap-emulation mode, the stream is started because it uses
snd_pcm_write_areas() internally.
As a workaround for this inconsistency, start_threshold value is changed
dynamically in sw_parmams and mmap_commit callbacks in mmap-emul plugin.
Meanwhile, start_threshold for slave PCM is set to boundary so that only
this plugin (or the one over it) can control the start of the stream.
This will fix problems in some apps using pulse plugin in the mmap mode.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The contents of the snd_pcm_scope_ops structure are not going to be
changed, so we might as well declare is as constant. This change is
backwards compatible, and avoids warnings if some ops structure is
actually defined as const.
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
The shared object may be still needed depending on the implementation
of hook-installation functions. When any hooks are registered in the
installation function, the dlobj has to be kept opened until closing
the PCM instance.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Extend the PCM-rate plugin protocol to allow the host to query the
supported sample rates. The protocol version is bumped to 0x010002,
and the version number negotiaion is slightly changed.
Now the plugin is supposed to fill the version it supports in return.
The old versioned plugins are still supported, but they may spew
version-mismatch warning prints.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The documentation for the latency parameter of pcm_set_params() says
that the value 0 uses an optimal value. This is wrong, as there is no
special handling for 0, and the result will be a buffer of minimal size.
Therefore, remove that incorrect statement.
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Shut up misc compile warnings from gcc:
pcm_plug.c: In function ‘snd_pcm_plug_change_mmap’:
pcm_plug.c:608: warning: enumeration value ‘SND_PCM_ACCESS_MMAP_INTERLEAVED’ not handled in switch
pcm_plug.c:608: warning: enumeration value ‘SND_PCM_ACCESS_MMAP_NONINTERLEAVED’ not handled in switch
pcm_plug.c:608: warning: enumeration value ‘SND_PCM_ACCESS_MMAP_COMPLEX’ not handled in switch
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The revents parameter is ambiguously defined whether it's a pointer
to a single event or an arary.
This patch defines the behavior of revents more strictly (in the
function description): it's a pointer of a single event.
Also fixed snd_pcm_wait_nocheck() to follow that rule.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
* added support for including pcm stream params in the output filename
* added support for piping the stream to a shell command if the filename
string starts with a pipe char
Signed-off-by: Pavel Hofman <pavel.hofman@insite.cz>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
If the slave PCM supports only another type of interleaved format,
plug did convert it wrongly and resulted in an unused access type
error. For example, if a slave PCM supporst only RW_NONINTERLEAVE
access and you try to play an interleaved format file, it resulted
in an error.
This patch fixes the conversion rule.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The commit a13707da6b
pcm_softvol plugin: remove access type change for refine
breaks the softvol in the case of RW -> MMAP. The slave of softvol
must be an mmap although the previous fix forces RW access.
This patch reverts the commit, and the fixed access refine method
to hanle non-interleaved <-> interleaved changes.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This makes it possible to write them to .data.rel.ro or to .rodata if
there is no relocation involved (arrays of character arrays).
Signed-off-by: Diego E. 'Flameeyes' Pettenò <flameeyes@gmail.com>
Use "const char *const []" as type for string arrays, or convert to
"const char [][x]" when it makes sense.
Signed-off-by: Diego E. 'Flameeyes' Pettenò <flameeyes@gmail.com>
This ensures they are emitted in .data.rel.ro rather than .data.rel,
which should make a nice difference when using prelink.
Signed-off-by: Diego E. 'Flameeyes' Pettenò <flameeyes@gmail.com>
By doing this we move them from the .data section to .rodata setion,
or from .data.rel to .data.rel.ro.
The .rodata section is mapped directly from the on-disk file, which is
always a save, while .data.rel.ro is mapped directly when using
prelink, which is a save in a lot of cases.
Signed-off-by: Diego E. 'Flameeyes' Pettenò <flameeyes@gmail.com>
snd_pcm_meter_add_scope_conf() may cause a segfault when pcm_scope_type
isn't defined.
Initialize type_conf properly to avoid it.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The softvol plugin does in-place sample recalculation. The access type
cannot be changed like in linear plugins. Remove access type change
in refine functions.
This bug was detected with PulseAudio on Sony PS3 platform.
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
As proposed in http://mailman.alsa-project.org/pipermail/alsa-devel/2008-June/008558.html
the snd_pcm_avail() and snd_pcm_avail_delay() functions are now available
to get accurate stream position in a straight way. The snd_pcm_avail_delay()
function was added to ensure full sync between avail and delay values. It's
actually implemented using delay() + avail_update() calls but it might be
changed in future.
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
snd_pcm_delay() is for synchronization purposes, it returns the overall latency
of the stream, not just the latency induced by the hardware playback buffer.
The documentation is a bit misleading about this fact, and some people already
misunderstood it. So let's reword this to make the explanation clearer and
explicit.
This is a result of the discussions of the thread this mail belongs to:
http://mailman.alsa-project.org/pipermail/alsa-devel/2008-June/008456.html
From: Lennart Poettering <mznyfn@0pointer.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
The plug PCM copies the 'mode' field from the slave PCM. If blocking mode is
enabled for the plug PCM the mode is subsequently overwritten with the original
requested 'mode'. If non-blocking mode is requested this does not happen.
Because the hw PCM synthesizes the 'mode' from the actual file descriptor flags
no SND_PCM_NO_xxx will ever be set for it. This has the effect that the 'mode'
of the plug PCM will also not include those flags anymore -- unless they are
overwritten as mentioned above. This basically means SND_PCM_NO_xxx is ignored
for plug:hw:4711 style device strings opened in non-blocking mode.
You can easily test this with "aplay --channels 7 --disable-channels -f S16_LE
-r 44100 -D plug:hw:0" on a device that cannot do 7 channels. Normally this
call should fail, however if you add "-N" to the command line this call will
succeed.
This patch simply copies the SND_PCM_NO_xxx flags back into the 'mode' field in
case we don't overwrite it with the original anyway.
Probably closes bug 3571 for good.
From: Lennart Poettering <mznyfn@0pointer.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
Make snd_pcm_plugin_rewind()/_forward() actually return how much has
been rewound/forwarded instead of how much could have been
rewounded/forwarded. This makes the code actually do what the
documentation of snd_pcm_rewind() suggests.
Signed-off-by: Lennart Poettering <lennart@poettering.net>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This minor patch fixes the type of the sframes variable in
snd_pcm_plugin_forward(). With this fix we need to cast less and the
code is less confusing.
Signed-off-by: Lennart Poettering <lennart@poettering.net>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The clamping of the input parameter in snd_pcm_rewind()/_forward() is
in the wrong direction.
Signed-off-by: Lennart Poettering <lennart@poettering.net>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
When the resolution is set to 2, a boolean control is created as a
mute switch instead of a volume control.
Also, fixed the possible zero-division error.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The i386 and x86-64 dmix may cause segfaults when 3-bytes formats are used
due to btsl asm code, which may overcome the buffer end-boundary.
The patch changes btsl to btsw so that it doesn't happen.
ALSA bug#3341:
https://bugtrack.alsa-project.org/alsa-bug/view.php?id=3341
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The plug plugin has a long-standing problem that it can handle only
slaves that support mmap because of format/rate/access conversions
(these corresponding plugins work only with mmap).
This patch adds the support of automatic mmap->rw conversion via
mmap_emul plugin.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The check in snd_pcm_mmap_emul_refine() is bogus and buggy.
Since the changed access type is took back at snd_pcm_mmap_emul_refine
again, it has to check the availability of mmap at each time.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Two trivial compile warning fixes:
- Add a missing return to snd_pcm_hw_clear_timer_queue()
- params->info is no long but int
The second one might have hit already on 64bit machine, but alas,
no one didn't notice it.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
