[alsa-devel] [PATCH - JACK IO plug v2 0/3] snd_pcm_drain support
Now in a clean manner as composed patches.
pcm: ioplug: Provide avail helper function --------- No changes
jack: Avoid call to snd_pcm_avail_update() from JACK thread --------- v2 uses snd_pcm_uframes_t as type for avail and does not check for avail < 0 becasue snd_pcm_ioplug_avail() will never retrun a value < 0
jack: Update poll_fd also in draining state --------- This patch is depending on [1]. Therefore it was rebased on top of [1]. Additionally the commit message was extended.
v1 of this patch can be found in [2].
[1] http://mailman.alsa-project.org/pipermail/alsa-devel/2018-July/137775.html [2] http://mailman.alsa-project.org/pipermail/alsa-devel/2018-July/137769.html
From: Timo Wischer twischer@de.adit-jv.com
This function can be called without calling snd_pcm_avail_update().
The call to snd_pcm_avail_update() can take some time. Therefore some developers would not like to call it from a real-time context (e.g. from JACK client context).
Signed-off-by: Timo Wischer twischer@de.adit-jv.com
diff --git a/include/pcm_ioplug.h b/include/pcm_ioplug.h index c1310e3..b16fc8b 100644 --- a/include/pcm_ioplug.h +++ b/include/pcm_ioplug.h @@ -235,6 +235,9 @@ int snd_pcm_ioplug_set_param_list(snd_pcm_ioplug_t *io, int type, unsigned int n int snd_pcm_ioplug_set_state(snd_pcm_ioplug_t *ioplug, snd_pcm_state_t state);
/* calucalte the available frames */ +snd_pcm_uframes_t snd_pcm_ioplug_avail(const snd_pcm_ioplug_t * const ioplug, + const snd_pcm_uframes_t hw_ptr, + const snd_pcm_uframes_t appl_ptr); snd_pcm_uframes_t snd_pcm_ioplug_hw_avail(const snd_pcm_ioplug_t * const ioplug, const snd_pcm_uframes_t hw_ptr, const snd_pcm_uframes_t appl_ptr); diff --git a/src/pcm/pcm_ioplug.c b/src/pcm/pcm_ioplug.c index 4d44ae2..6d52c27 100644 --- a/src/pcm/pcm_ioplug.c +++ b/src/pcm/pcm_ioplug.c @@ -1221,6 +1221,21 @@ int snd_pcm_ioplug_set_state(snd_pcm_ioplug_t *ioplug, snd_pcm_state_t state) * \param ioplug the ioplug handle * \param hw_ptr hardware pointer in frames * \param appl_ptr application pointer in frames + * \return available frames for the application + */ +snd_pcm_uframes_t snd_pcm_ioplug_avail(const snd_pcm_ioplug_t * const ioplug, + const snd_pcm_uframes_t hw_ptr, + const snd_pcm_uframes_t appl_ptr) +{ + return __snd_pcm_avail(ioplug->pcm, hw_ptr, appl_ptr); +} + +/** + * \brief Get the available frames. This function can be used to calculate the + * the available frames before calling #snd_pcm_avail_update() + * \param ioplug the ioplug handle + * \param hw_ptr hardware pointer in frames + * \param appl_ptr application pointer in frames * \return available frames for the hardware */ snd_pcm_uframes_t snd_pcm_ioplug_hw_avail(const snd_pcm_ioplug_t * const ioplug, @@ -1230,8 +1245,9 @@ snd_pcm_uframes_t snd_pcm_ioplug_hw_avail(const snd_pcm_ioplug_t * const ioplug, /* available data/space which can be transferred by the user * application */ - const snd_pcm_uframes_t user_avail = __snd_pcm_avail(ioplug->pcm, - hw_ptr, appl_ptr); + const snd_pcm_uframes_t user_avail = snd_pcm_ioplug_avail(ioplug, + hw_ptr, + appl_ptr);
if (user_avail > ioplug->pcm->buffer_size) { /* there was an Xrun */
Iwai-san,
On Jul 3 2018 22:59, twischer@de.adit-jv.com wrote:
From: Timo Wischer twischer@de.adit-jv.com
This function can be called without calling snd_pcm_avail_update().
The call to snd_pcm_avail_update() can take some time. Therefore some developers would not like to call it from a real-time context (e.g. from JACK client context).
Signed-off-by: Timo Wischer twischer@de.adit-jv.com
diff --git a/include/pcm_ioplug.h b/include/pcm_ioplug.h index c1310e3..b16fc8b 100644 --- a/include/pcm_ioplug.h +++ b/include/pcm_ioplug.h @@ -235,6 +235,9 @@ int snd_pcm_ioplug_set_param_list(snd_pcm_ioplug_t *io, int type, unsigned int n int snd_pcm_ioplug_set_state(snd_pcm_ioplug_t *ioplug, snd_pcm_state_t state);
/* calucalte the available frames */ +snd_pcm_uframes_t snd_pcm_ioplug_avail(const snd_pcm_ioplug_t * const ioplug,
const snd_pcm_uframes_t hw_ptr,const snd_pcm_uframes_t appl_ptr);diff --git a/src/pcm/pcm_ioplug.c b/src/pcm/pcm_ioplug.c index 4d44ae2..6d52c27 100644 --- a/src/pcm/pcm_ioplug.c +++ b/src/pcm/pcm_ioplug.c @@ -1221,6 +1221,21 @@ int snd_pcm_ioplug_set_state(snd_pcm_ioplug_t *ioplug, snd_pcm_state_t state)
- \param ioplug the ioplug handle
- \param hw_ptr hardware pointer in frames
- \param appl_ptr application pointer in frames
- \return available frames for the application
- */
+snd_pcm_uframes_t snd_pcm_ioplug_avail(const snd_pcm_ioplug_t * const ioplug,
const snd_pcm_uframes_t hw_ptr,const snd_pcm_uframes_t appl_ptr)+{
- return __snd_pcm_avail(ioplug->pcm, hw_ptr, appl_ptr);
+}
+/**
- \brief Get the available frames. This function can be used to calculate the
- the available frames before calling #snd_pcm_avail_update()
- \param ioplug the ioplug handle
- \param hw_ptr hardware pointer in frames
*/ snd_pcm_uframes_t snd_pcm_ioplug_hw_avail(const snd_pcm_ioplug_t * const ioplug,
- \param appl_ptr application pointer in frames
- \return available frames for the hardware
@@ -1230,8 +1245,9 @@ snd_pcm_uframes_t snd_pcm_ioplug_hw_avail(const snd_pcm_ioplug_t * const ioplug, /* available data/space which can be transferred by the user * application */
- const snd_pcm_uframes_t user_avail = __snd_pcm_avail(ioplug->pcm,
hw_ptr, appl_ptr);
const snd_pcm_uframes_t user_avail = snd_pcm_ioplug_avail(ioplug,
hw_ptr,appl_ptr);if (user_avail > ioplug->pcm->buffer_size) { /* there was an Xrun */
I have a question about maintenance policy of this library to update version script for GNU Linker (ld) when introducing new symbols. The version script is not updated since node name as 'ALSA_0.9.7'. The newly intduced symbol, 'snd_pcm_ioplug_hw_avail' is in node name of 'ALSA_0.9'.
``` $ readelf -s src/.libs/libasound.so.2.0.0 | grep snd_pcm_ioplug_hw_avail 1254: 000000000009bfe0 40 FUNC GLOBAL DEFAULT 13 snd_pcm_ioplug_hw_avail@@ALSA_0.9 3840: 000000000009bfe0 40 FUNC GLOBAL DEFAULT 13 snd_pcm_ioplug_hw_avail ```
The last node name is 'ALSA_1.1.6':
``` $ tail -n 10 src/Versions.in global:
@SYMBOL_PREFIX@alsa_lisp_*; } ALSA_0.9.5;
ALSA_1.1.6 { global:
@SYMBOL_PREFIX@snd_dlopen; } ALSA_0.9.7; ```
Practically this brings no issue, but theoretically the newly introduced symbol should be in a new node name next to ALSA_1.1.6. I'm not so strict in this point, but it's better to decide maintenance policy (because I've added some APIs in recent years).
Regards
Takashi Sakamoto
On Wed, 04 Jul 2018 02:34:47 +0200, Takashi Sakamoto wrote:
Iwai-san,
On Jul 3 2018 22:59, twischer@de.adit-jv.com wrote:
From: Timo Wischer twischer@de.adit-jv.com
This function can be called without calling snd_pcm_avail_update().
The call to snd_pcm_avail_update() can take some time. Therefore some developers would not like to call it from a real-time context (e.g. from JACK client context).
Signed-off-by: Timo Wischer twischer@de.adit-jv.com
diff --git a/include/pcm_ioplug.h b/include/pcm_ioplug.h index c1310e3..b16fc8b 100644 --- a/include/pcm_ioplug.h +++ b/include/pcm_ioplug.h @@ -235,6 +235,9 @@ int snd_pcm_ioplug_set_param_list(snd_pcm_ioplug_t *io, int type, unsigned int n int snd_pcm_ioplug_set_state(snd_pcm_ioplug_t *ioplug, snd_pcm_state_t state); /* calucalte the available frames */ +snd_pcm_uframes_t snd_pcm_ioplug_avail(const snd_pcm_ioplug_t * const ioplug,
const snd_pcm_uframes_t hw_ptr,const snd_pcm_uframes_t appl_ptr);diff --git a/src/pcm/pcm_ioplug.c b/src/pcm/pcm_ioplug.c index 4d44ae2..6d52c27 100644 --- a/src/pcm/pcm_ioplug.c +++ b/src/pcm/pcm_ioplug.c @@ -1221,6 +1221,21 @@ int snd_pcm_ioplug_set_state(snd_pcm_ioplug_t *ioplug, snd_pcm_state_t state)
- \param ioplug the ioplug handle
- \param hw_ptr hardware pointer in frames
- \param appl_ptr application pointer in frames
- \return available frames for the application
- */
+snd_pcm_uframes_t snd_pcm_ioplug_avail(const snd_pcm_ioplug_t * const ioplug,
const snd_pcm_uframes_t hw_ptr,const snd_pcm_uframes_t appl_ptr)+{
- return __snd_pcm_avail(ioplug->pcm, hw_ptr, appl_ptr);
+}
+/**
- \brief Get the available frames. This function can be used to calculate the
- the available frames before calling #snd_pcm_avail_update()
- \param ioplug the ioplug handle
- \param hw_ptr hardware pointer in frames
*/ snd_pcm_uframes_t snd_pcm_ioplug_hw_avail(const snd_pcm_ioplug_t * const ioplug,
- \param appl_ptr application pointer in frames
- \return available frames for the hardware
@@ -1230,8 +1245,9 @@ snd_pcm_uframes_t snd_pcm_ioplug_hw_avail(const snd_pcm_ioplug_t * const ioplug, /* available data/space which can be transferred by the user * application */
- const snd_pcm_uframes_t user_avail = __snd_pcm_avail(ioplug->pcm,
hw_ptr, appl_ptr);
- const snd_pcm_uframes_t user_avail = snd_pcm_ioplug_avail(ioplug,
hw_ptr,appl_ptr); if (user_avail > ioplug->pcm->buffer_size) {I have a question about maintenance policy of this library to update version script for GNU Linker (ld) when introducing new symbols. The version script is not updated since node name as 'ALSA_0.9.7'. The newly intduced symbol, 'snd_pcm_ioplug_hw_avail' is in node name of 'ALSA_0.9'.
$ readelf -s src/.libs/libasound.so.2.0.0 | grep snd_pcm_ioplug_hw_avail 1254: 000000000009bfe0 40 FUNC GLOBAL DEFAULT 13 snd_pcm_ioplug_hw_avail@@ALSA_0.9 3840: 000000000009bfe0 40 FUNC GLOBAL DEFAULT 13 snd_pcm_ioplug_hw_availThe last node name is 'ALSA_1.1.6':
$ tail -n 10 src/Versions.in global: @SYMBOL_PREFIX@alsa_lisp_*; } ALSA_0.9.5; ALSA_1.1.6 { global: @SYMBOL_PREFIX@snd_dlopen; } ALSA_0.9.7;Practically this brings no issue, but theoretically the newly introduced symbol should be in a new node name next to ALSA_1.1.6. I'm not so strict in this point, but it's better to decide maintenance policy (because I've added some APIs in recent years).
Yes, that's an open question.
I myself am no big fan of the versioned symbols. This has been a PITA for many applications. The versioned smybols is useful only if the function signature may change. But what's the difference from renaming the function, then?
So we've stopped putting the new symbols into the new versioned section; the snd_dlopen was an exception because it really changed the signature.
thanks,
Takashi
Hi,
On Jul 4 2018 15:21, Takashi Iwai wrote:
On Wed, 04 Jul 2018 02:34:47 +0200, Takashi Sakamoto wrote:
Iwai-san,
On Jul 3 2018 22:59, twischer@de.adit-jv.com wrote:
From: Timo Wischer twischer@de.adit-jv.com
This function can be called without calling snd_pcm_avail_update().
The call to snd_pcm_avail_update() can take some time. Therefore some developers would not like to call it from a real-time context (e.g. from JACK client context).
Signed-off-by: Timo Wischer twischer@de.adit-jv.com
diff --git a/include/pcm_ioplug.h b/include/pcm_ioplug.h index c1310e3..b16fc8b 100644 --- a/include/pcm_ioplug.h +++ b/include/pcm_ioplug.h @@ -235,6 +235,9 @@ int snd_pcm_ioplug_set_param_list(snd_pcm_ioplug_t *io, int type, unsigned int n int snd_pcm_ioplug_set_state(snd_pcm_ioplug_t *ioplug, snd_pcm_state_t state); /* calucalte the available frames */ +snd_pcm_uframes_t snd_pcm_ioplug_avail(const snd_pcm_ioplug_t * const ioplug,
const snd_pcm_uframes_t hw_ptr,const snd_pcm_uframes_t appl_ptr);diff --git a/src/pcm/pcm_ioplug.c b/src/pcm/pcm_ioplug.c index 4d44ae2..6d52c27 100644 --- a/src/pcm/pcm_ioplug.c +++ b/src/pcm/pcm_ioplug.c @@ -1221,6 +1221,21 @@ int snd_pcm_ioplug_set_state(snd_pcm_ioplug_t *ioplug, snd_pcm_state_t state) * \param ioplug the ioplug handle * \param hw_ptr hardware pointer in frames * \param appl_ptr application pointer in frames
- \return available frames for the application
- */
+snd_pcm_uframes_t snd_pcm_ioplug_avail(const snd_pcm_ioplug_t * const ioplug,
const snd_pcm_uframes_t hw_ptr,const snd_pcm_uframes_t appl_ptr)+{
- return __snd_pcm_avail(ioplug->pcm, hw_ptr, appl_ptr);
+}
+/**
- \brief Get the available frames. This function can be used to calculate the
- the available frames before calling #snd_pcm_avail_update()
- \param ioplug the ioplug handle
- \param hw_ptr hardware pointer in frames
snd_pcm_uframes_t snd_pcm_ioplug_hw_avail(const snd_pcm_ioplug_t * const ioplug,
- \param appl_ptr application pointer in frames
*/
- \return available frames for the hardware
@@ -1230,8 +1245,9 @@ snd_pcm_uframes_t snd_pcm_ioplug_hw_avail(const snd_pcm_ioplug_t * const ioplug, /* available data/space which can be transferred by the user * application */
- const snd_pcm_uframes_t user_avail = __snd_pcm_avail(ioplug->pcm,
hw_ptr, appl_ptr);
- const snd_pcm_uframes_t user_avail = snd_pcm_ioplug_avail(ioplug,
hw_ptr,appl_ptr); if (user_avail > ioplug->pcm->buffer_size) { /* there was an Xrun */I have a question about maintenance policy of this library to update version script for GNU Linker (ld) when introducing new symbols. The version script is not updated since node name as 'ALSA_0.9.7'. The newly intduced symbol, 'snd_pcm_ioplug_hw_avail' is in node name of 'ALSA_0.9'.
$ readelf -s src/.libs/libasound.so.2.0.0 | grep snd_pcm_ioplug_hw_avail 1254: 000000000009bfe0 40 FUNC GLOBAL DEFAULT 13 snd_pcm_ioplug_hw_avail@@ALSA_0.9 3840: 000000000009bfe0 40 FUNC GLOBAL DEFAULT 13 snd_pcm_ioplug_hw_availThe last node name is 'ALSA_1.1.6':
$ tail -n 10 src/Versions.in global: @SYMBOL_PREFIX@alsa_lisp_*; } ALSA_0.9.5; ALSA_1.1.6 { global: @SYMBOL_PREFIX@snd_dlopen; } ALSA_0.9.7;Practically this brings no issue, but theoretically the newly introduced symbol should be in a new node name next to ALSA_1.1.6. I'm not so strict in this point, but it's better to decide maintenance policy (because I've added some APIs in recent years).
Yes, that's an open question.
I myself am no big fan of the versioned symbols. This has been a PITA for many applications. The versioned smybols is useful only if the function signature may change. But what's the difference from renaming the function, then?
So we've stopped putting the new symbols into the new versioned section; the snd_dlopen was an exception because it really changed the signature.
Nowadays I have no positive reason to use versioned symbols as you said. I'm OK and thanks for your explanation.
Takashi Sakamoto
From: Timo Wischer twischer@de.adit-jv.com
snd_pcm_avail_update() can call snd_pcm_jack_stop() but snd_pcm_jack_stop() should not be called by the JACK thread. It should only be called by the thread how has called snd_pcm_jack_start().
In addition the execution of snd_pcm_avail_update() can take a while. Therefore it should not be called by the JACK thread to not block this thread.
Signed-off-by: Timo Wischer twischer@de.adit-jv.com
diff --git a/jack/pcm_jack.c b/jack/pcm_jack.c index e3df4d2..5a57e13 100644 --- a/jack/pcm_jack.c +++ b/jack/pcm_jack.c @@ -62,13 +62,13 @@ static int snd_pcm_jack_stop(snd_pcm_ioplug_t *io); static int pcm_poll_block_check(snd_pcm_ioplug_t *io) { static char buf[32]; - snd_pcm_sframes_t avail; + snd_pcm_uframes_t avail; snd_pcm_jack_t *jack = io->private_data;
if (io->state == SND_PCM_STATE_RUNNING || (io->state == SND_PCM_STATE_PREPARED && io->stream == SND_PCM_STREAM_CAPTURE)) { - avail = snd_pcm_avail_update(io->pcm); - if (avail >= 0 && avail < jack->min_avail) { + avail = snd_pcm_ioplug_avail(io, jack->hw_ptr, io->appl_ptr); + if (avail < jack->min_avail) { while (read(io->poll_fd, &buf, sizeof(buf)) == sizeof(buf)) ; return 1; @@ -81,11 +81,11 @@ static int pcm_poll_block_check(snd_pcm_ioplug_t *io) static int pcm_poll_unblock_check(snd_pcm_ioplug_t *io) { static char buf[1]; - snd_pcm_sframes_t avail; + snd_pcm_uframes_t avail; snd_pcm_jack_t *jack = io->private_data;
- avail = snd_pcm_avail_update(io->pcm); - if (avail < 0 || avail >= jack->min_avail) { + avail = snd_pcm_ioplug_avail(io, jack->hw_ptr, io->appl_ptr); + if (avail >= jack->min_avail) { write(jack->fd, &buf, 1); return 1; }
From: Timo Wischer twischer@de.adit-jv.com
to support snd_pcm_drain for the JACK IO plugin. With this changes there will be an poll_fd event in DRAINING state even if the min_avail was not yet reached. Otherwise the application would never recognize that all samples were processed by JACK. In addition the JACK real-time thread is also processing when in DRAINING state and not only when in RUNNING or PREPARE state.
Signed-off-by: Timo Wischer twischer@de.adit-jv.com
diff --git a/jack/pcm_jack.c b/jack/pcm_jack.c index 5a57e13..98a6f7e 100644 --- a/jack/pcm_jack.c +++ b/jack/pcm_jack.c @@ -66,6 +66,7 @@ static int pcm_poll_block_check(snd_pcm_ioplug_t *io) snd_pcm_jack_t *jack = io->private_data;
if (io->state == SND_PCM_STATE_RUNNING || + io->state == SND_PCM_STATE_DRAINING || (io->state == SND_PCM_STATE_PREPARED && io->stream == SND_PCM_STREAM_CAPTURE)) { avail = snd_pcm_ioplug_avail(io, jack->hw_ptr, io->appl_ptr); if (avail < jack->min_avail) { @@ -85,7 +86,12 @@ static int pcm_poll_unblock_check(snd_pcm_ioplug_t *io) snd_pcm_jack_t *jack = io->private_data;
avail = snd_pcm_ioplug_avail(io, jack->hw_ptr, io->appl_ptr); - if (avail >= jack->min_avail) { + /* In draining state poll_fd is used to wait + * till all pending frames are played. + * Therefore it has to be guarantee that a poll event is also generated + * if the buffer contains less than min_avail frames + */ + if (avail >= jack->min_avail || io->state == SND_PCM_STATE_DRAINING) { write(jack->fd, &buf, 1); return 1; } @@ -161,7 +167,8 @@ snd_pcm_jack_process_cb(jack_nframes_t nframes, snd_pcm_ioplug_t *io) jack->areas[channel].step = jack->sample_bits; }
- if (io->state == SND_PCM_STATE_RUNNING) { + if (io->state == SND_PCM_STATE_RUNNING || + io->state == SND_PCM_STATE_DRAINING) { snd_pcm_uframes_t hw_ptr = jack->hw_ptr; const snd_pcm_uframes_t hw_avail = snd_pcm_ioplug_hw_avail(io, hw_ptr, io->appl_ptr);
On Tue, 03 Jul 2018 15:59:20 +0200, twischer@de.adit-jv.com wrote:
Now in a clean manner as composed patches.
Thanks, now applied all patches.
But just another reminder: don't forget to put "v2" to each patch subject line, too, not only in the cover letter.
Takashi
participants (3)
-
Takashi Iwai -
Takashi Sakamoto -
twischer＠de.adit-jv.com