4 Jan
2009
4 Jan
'09
10:51 p.m.
Hiho,
I am experiencing strange things with some code that uses the ALSA sequencer,
with the function
snd_seq_ev_schedule_real(evt, mQueue, 1, &time);
The timing of this is unreliable.
At times it goes as expected for a while, but then after some time large
delays (minutes, instead of the expected latency of 0.1 second) are
introduced.
Does anyone have any idea what the problem could be?
I have a code snippet below. In case anyone wants to try (the app is
SuperCollider, you need the latest svn).
I'm on a Debian kernel (2.6.24-1-amd64 #1 SMP Thu Mar 27 16:52:38 UTC 2008
x86_64 GNU/Linux), in case that is relevant.
sincerely,
Marije
int SC_AlsaMidiClient::sendEvent(int outputIndex, int uid, snd_seq_event_t*
evt, float late)
{
snd_seq_real_time time;
if ((outputIndex < 0) || (outputIndex >= mNumOutPorts)) return
errIndexOutOfRange;
snd_seq_ev_set_source(evt, mOutPorts[outputIndex]);
if (uid == 0) {
// send to all subscribed ports
snd_seq_ev_set_subs(evt);
} else {
// send to specific port
int cid, pid;
SC_AlsaParseUID(uid, cid, pid);
snd_seq_ev_set_dest(evt, cid, pid);
}
long latelong;
if (late > 0.f) {
latelong = (long) (late * 1000000000);
// new time calculation. The old one was not correct
time.tv_sec = (long)(latelong / 1000000000); // seconds
time.tv_nsec = (long)(latelong % 1000000000); // nanoseconds
} else {
time.tv_sec = time.tv_nsec = 0;
}
// evt->flags = evt->flags | SND_SEQ_TIME_STAMP_REAL;
post("MIDI (ALSA): sending event, time %i, %i, late %f, latelong %i\n",
time.tv_sec, time.tv_nsec, late, latelong);
snd_seq_ev_schedule_real(evt, mQueue, 1, &time);
snd_seq_event_output_direct(mHandle, evt);
// snd_seq_event_output(mHandle, evt);
return errNone;
}
10 Jan
10 Jan
4:59 p.m.
New subject: alsa sequencer latency ( snd_seq_ev_schedule_real ) - need help!
Hi,
since I get no response here...
which list you recommend I post this question to?
I also got no reaction on the alsa-dev list... alsa-user then?
sincerely,
Marije
On Sunday 04 January 2009 21:51:21 nescivi wrote:
Hiho,
I am experiencing strange things with some code that uses the ALSA
sequencer, with the function
snd_seq_ev_schedule_real(evt, mQueue, 1, &time);
The timing of this is unreliable.
At times it goes as expected for a while, but then after some time large
delays (minutes, instead of the expected latency of 0.1 second) are
introduced.
Does anyone have any idea what the problem could be?
I have a code snippet below. In case anyone wants to try (the app is
SuperCollider, you need the latest svn).
I'm on a Debian kernel (2.6.24-1-amd64 #1 SMP Thu Mar 27 16:52:38 UTC 2008
x86_64 GNU/Linux), in case that is relevant.
sincerely,
Marije
int SC_AlsaMidiClient::sendEvent(int outputIndex, int uid, snd_seq_event_t*
evt, float late)
{
snd_seq_real_time time;
if ((outputIndex < 0) || (outputIndex >= mNumOutPorts)) return
errIndexOutOfRange;
snd_seq_ev_set_source(evt, mOutPorts[outputIndex]);
if (uid == 0) {
// send to all subscribed ports
snd_seq_ev_set_subs(evt);
} else {
// send to specific port
int cid, pid;
SC_AlsaParseUID(uid, cid, pid);
snd_seq_ev_set_dest(evt, cid, pid);
}
long latelong;
if (late > 0.f) {
latelong = (long) (late * 1000000000);
// new time calculation. The old one was not correct
time.tv_sec = (long)(latelong / 1000000000); // seconds
time.tv_nsec = (long)(latelong % 1000000000); // nanoseconds
} else {
time.tv_sec = time.tv_nsec = 0;
}
// evt->flags = evt->flags | SND_SEQ_TIME_STAMP_REAL;
post("MIDI (ALSA): sending event, time %i, %i, late %f, latelong %i\n",
time.tv_sec, time.tv_nsec, late, latelong);
snd_seq_ev_schedule_real(evt, mQueue, 1, &time);
snd_seq_event_output_direct(mHandle, evt);
// snd_seq_event_output(mHandle, evt);
return errNone;
}
_______________________________________________
Linux-audio-dev mailing list
Linux-audio-dev(a)lists.linuxaudio.org
http://lists.linuxaudio.org/mailman/listinfo/linux-audio-dev
5:27 p.m.
New subject: alsa sequencer latency ( snd_seq_ev_schedule_real ) - need help!
On Sat, Jan 10, 2009 at 03:59:06PM -0500, nescivi wrote:
since I get no response here...
which list you recommend I post this question to?
I also got no reaction on the alsa-dev list... alsa-user then?
> int SC_AlsaMidiClient::sendEvent(int outputIndex,
int uid, snd_seq_event_t*
> evt, float late)
> {
> snd_seq_real_time time;
>
> if ((outputIndex < 0) || (outputIndex >= mNumOutPorts)) return
> errIndexOutOfRange;
>
> snd_seq_ev_set_source(evt, mOutPorts[outputIndex]);
> if (uid == 0) {
> // send to all subscribed ports
> snd_seq_ev_set_subs(evt);
> } else {
> // send to specific port
> int cid, pid;
> SC_AlsaParseUID(uid, cid, pid);
> snd_seq_ev_set_dest(evt, cid, pid);
> }
>
> long latelong;
> if (late > 0.f) {
> latelong = (long) (late * 1000000000);
> // new time calculation. The old one was not correct
> time.tv_sec = (long)(latelong / 1000000000); // seconds
> time.tv_nsec = (long)(latelong % 1000000000); // nanoseconds
> } else {
> time.tv_sec = time.tv_nsec = 0;
> }
>
> // evt->flags = evt->flags | SND_SEQ_TIME_STAMP_REAL;
>
> post("MIDI (ALSA): sending event, time %i, %i, late %f, latelong %i\n",
> time.tv_sec, time.tv_nsec, late, latelong);
>
> snd_seq_ev_schedule_real(evt, mQueue, 1, &time);
> snd_seq_event_output_direct(mHandle, evt);
> // snd_seq_event_output(mHandle, evt);
>
> return errNone;
> }
If 'late' > 2.147 then 'latelong' will go negative.
Could this happen when you schedule an event more than
2 seconds ahead of due time ?
Alternative:
time.tv_sec = (long)(floorf (late));
time.tv_nsec = (long)((late - time.tv_sec) * 1e9f);
Ciao,
--
FA
Laboratorio di Acustica ed Elettroacustica
Parma, Italia
O tu, che porte, correndo si ?
E guerra e morte !
11 Jan
11 Jan
1:38 a.m.
New subject: alsa sequencer latency ( snd_seq_ev_schedule_real ) - need help!
On Saturday 10 January 2009 16:13:44 Fons Adriaensen wrote:
On Sat, Jan 10, 2009 at 03:59:06PM -0500, nescivi
wrote:
no, it happens even if I set late to 0.1.
It is really strange. Sometimes it just works for a while, and then it just
takes minutes before any event comes out.
The values print out fine, just something weird seems to be happening in the
actual scheduling.
since I get no response here...
which list you recommend I post this question to?
I also got no reaction on the alsa-dev list... alsa-user then?
> int SC_AlsaMidiClient::sendEvent(int outputIndex, int uid,
> snd_seq_event_t* evt, float late)
> {
> snd_seq_real_time time;
>
> if ((outputIndex < 0) || (outputIndex >= mNumOutPorts)) return
> errIndexOutOfRange;
>
> snd_seq_ev_set_source(evt, mOutPorts[outputIndex]);
> if (uid == 0) {
> // send to all subscribed ports
> snd_seq_ev_set_subs(evt);
> } else {
> // send to specific port
> int cid, pid;
> SC_AlsaParseUID(uid, cid, pid);
> snd_seq_ev_set_dest(evt, cid, pid);
> }
>
> long latelong;
> if (late > 0.f) {
> latelong = (long) (late * 1000000000);
> // new time calculation. The old one was not correct
> time.tv_sec = (long)(latelong / 1000000000); // seconds
> time.tv_nsec = (long)(latelong % 1000000000); // nanoseconds
> } else {
> time.tv_sec = time.tv_nsec = 0;
> }
>
> // evt->flags = evt->flags | SND_SEQ_TIME_STAMP_REAL;
>
> post("MIDI (ALSA): sending event, time %i, %i, late %f, latelong
> %i\n", time.tv_sec, time.tv_nsec, late, latelong);
>
> snd_seq_ev_schedule_real(evt, mQueue, 1, &time);
> snd_seq_event_output_direct(mHandle, evt);
> // snd_seq_event_output(mHandle, evt);
>
> return errNone;
> }
If 'late' > 2.147 then 'latelong' will go negative.
Could this happen when you schedule an event more than
2 seconds ahead of due time ? Alternative:
time.tv_sec = (long)(floorf (late));
time.tv_nsec = (long)((late - time.tv_sec) * 1e9f);
thanks, I'll try that too.
sincerely,
Marije
4:59 a.m.
New subject: alsa sequencer latency ( snd_seq_ev_schedule_real ) -?need help!
On Sun, Jan 11, 2009 at 12:38:01AM -0500, nescivi wrote:
It is really strange. Sometimes it just works for a
while, and then it just
takes minutes before any event comes out.
The values print out fine, just something weird seems to be happening in the
actual scheduling.
Since you are using snd_seq_output_direct(), I *assume* you
don't need snd_seq_drain_output(). But have you ever tried
snd_seq_write();
snd_seq_drain_output();
instead of
snd_seq_write_direct(); ?
Ciao,
--
FA
Laboratorio di Acustica ed Elettroacustica
Parma, Italia
O tu, che porte, correndo si ?
E guerra e morte !
12:16 p.m.
New subject: alsa sequencer latency ( snd_seq_ev_schedule_real ) -?need help!
On Sunday 11 January 2009 03:45:34 Fons Adriaensen wrote:
On Sun, Jan 11, 2009 at 12:38:01AM -0500, nescivi
wrote:
yes, I did.
I tried it with:
snd_seq_event_output(mHandle, evt);
snd_seq_drain_output();
sincerely,
Marije
It is really strange. Sometimes it just works for
a while, and then it
just takes minutes before any event comes out.
The values print out fine, just something weird seems to be happening in
the actual scheduling.
Since you are using snd_seq_output_direct(), I *assume* you
don't need snd_seq_drain_output(). But have you ever tried
snd_seq_write();
snd_seq_drain_output();
instead of
snd_seq_write_direct(); ?
25 Jan
25 Jan
5:27 p.m.
New subject: alsa sequencer latency ( snd_seq_ev_schedule_real ) -?need help!
Hiho,
just a note.
It does seem to behave properly on a Debian kernel 2.6.26, whereas it failed
on 2.6.24
sincerely,
Marije
On Sunday 11 January 2009 11:16:54 nescivi wrote:
On Sunday 11 January 2009 03:45:34 Fons Adriaensen
wrote:
On Sun, Jan 11, 2009 at 12:38:01AM -0500, nescivi
wrote:
yes, I did.
I tried it with:
snd_seq_event_output(mHandle, evt);
snd_seq_drain_output();
sincerely,
Marije
_______________________________________________
Linux-audio-dev mailing list
Linux-audio-dev(a)lists.linuxaudio.org
http://lists.linuxaudio.org/mailman/listinfo/linux-audio-dev It is really strange. Sometimes it just works for
a while, and then it
just takes minutes before any event comes out.
The values print out fine, just something weird seems to be happening
in the actual scheduling.
Since you are using snd_seq_output_direct(), I *assume* you
don't need snd_seq_drain_output(). But have you ever tried
snd_seq_write();
snd_seq_drain_output();
instead of
snd_seq_write_direct(); ?
5580
days inactive
5601
days old
linux-audio-dev@lists.linuxaudio.org
6 comments
2 participants
participants (2)
-
Fons Adriaensen -
nescivi