[pulseaudio-discuss] [PATCH] alsa: get avail, delay, timestamps in a single kernel call
David Henningsson
david.henningsson at canonical.com
Tue Nov 6 23:40:28 PST 2012
On 11/05/2012 10:35 PM, Pierre-Louis Bossart wrote:
>
>> sorry for the long delay.
>
> no worries. looks like this mailing list went back 4 months in time with
> the switch from Daylight Savings time?
Naah, we more decided that there were a few things we wanted to review
for 3.0 and this was one of them :-)
>> This seems to make a lot of sense, but a question - does this impose a
>> requirement on a certain version of alsa-lib and/or the linux kernel?
>
> no, these changes are using standard calls.
> The changes I pushed at the kernel level since my presentation at the
> Plumbers conference will require a specific version of alsa-lib to read
> the audio timestamps, everything below should work right off the bat
> without dependencies.
Ok. I took your patch for a small test spin and it didn't seem to have
broken anything, so I pushed this patch now. Thanks for your contribution!
>
>> 2012-07-30 02:46, Pierre-Louis Bossart skrev:
>>> 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.
>>>
>>> Signed-off-by: Pierre-Louis Bossart
>>> <pierre-louis.bossart at linux.intel.com>
>>> ---
>>> src/modules/alsa/alsa-sink.c | 12 ++++--------
>>> src/modules/alsa/alsa-source.c | 12 ++++--------
>>> src/modules/alsa/alsa-util.c | 17 ++++++++++++-----
>>> src/modules/alsa/alsa-util.h | 2 +-
>>> 4 files changed, 21 insertions(+), 22 deletions(-)
>>>
>>> diff --git a/src/modules/alsa/alsa-sink.c b/src/modules/alsa/alsa-sink.c
>>> index 544399e..e4baa1f 100644
>>> --- a/src/modules/alsa/alsa-sink.c
>>> +++ b/src/modules/alsa/alsa-sink.c
>>> @@ -825,6 +825,7 @@ static void update_smoother(struct userdata *u) {
>>> int err;
>>> pa_usec_t now1 = 0, now2;
>>> snd_pcm_status_t *status;
>>> + snd_htimestamp_t htstamp = { 0, 0 };
>>> snd_pcm_status_alloca(&status);
>>> @@ -833,18 +834,13 @@ static void update_smoother(struct userdata *u) {
>>> /* Let's update the time smoother */
>>> - if (PA_UNLIKELY((err = pa_alsa_safe_delay(u->pcm_handle, &delay,
>>> u->hwbuf_size, &u->sink->sample_spec, FALSE)) < 0)) {
>>> + if (PA_UNLIKELY((err = pa_alsa_safe_delay(u->pcm_handle, status,
>>> &delay, u->hwbuf_size, &u->sink->sample_spec, FALSE)) < 0)) {
>>> pa_log_warn("Failed to query DSP status data: %s",
>>> pa_alsa_strerror(err));
>>> return;
>>> }
>>> - if (PA_UNLIKELY((err = snd_pcm_status(u->pcm_handle, status)) < 0))
>>> - pa_log_warn("Failed to get timestamp: %s",
>>> pa_alsa_strerror(err));
>>> - else {
>>> - snd_htimestamp_t htstamp = { 0, 0 };
>>> - snd_pcm_status_get_htstamp(status, &htstamp);
>>> - now1 = pa_timespec_load(&htstamp);
>>> - }
>>> + snd_pcm_status_get_htstamp(status, &htstamp);
>>> + now1 = pa_timespec_load(&htstamp);
>>> /* Hmm, if the timestamp is 0, then it wasn't set and we take
>>> the current time */
>>> if (now1 <= 0)
>>> diff --git a/src/modules/alsa/alsa-source.c
>>> b/src/modules/alsa/alsa-source.c
>>> index 94d4a09..17566e7 100644
>>> --- a/src/modules/alsa/alsa-source.c
>>> +++ b/src/modules/alsa/alsa-source.c
>>> @@ -767,6 +767,7 @@ static void update_smoother(struct userdata *u) {
>>> int err;
>>> pa_usec_t now1 = 0, now2;
>>> snd_pcm_status_t *status;
>>> + snd_htimestamp_t htstamp = { 0, 0 };
>>> snd_pcm_status_alloca(&status);
>>> @@ -775,18 +776,13 @@ static void update_smoother(struct userdata *u) {
>>> /* Let's update the time smoother */
>>> - if (PA_UNLIKELY((err = pa_alsa_safe_delay(u->pcm_handle, &delay,
>>> u->hwbuf_size, &u->source->sample_spec, TRUE)) < 0)) {
>>> + if (PA_UNLIKELY((err = pa_alsa_safe_delay(u->pcm_handle, status,
>>> &delay, u->hwbuf_size, &u->source->sample_spec, TRUE)) < 0)) {
>>> pa_log_warn("Failed to get delay: %s", pa_alsa_strerror(err));
>>> return;
>>> }
>>> - if (PA_UNLIKELY((err = snd_pcm_status(u->pcm_handle, status)) < 0))
>>> - pa_log_warn("Failed to get timestamp: %s",
>>> pa_alsa_strerror(err));
>>> - else {
>>> - snd_htimestamp_t htstamp = { 0, 0 };
>>> - snd_pcm_status_get_htstamp(status, &htstamp);
>>> - now1 = pa_timespec_load(&htstamp);
>>> - }
>>> + snd_pcm_status_get_htstamp(status, &htstamp);
>>> + now1 = pa_timespec_load(&htstamp);
>>> /* Hmm, if the timestamp is 0, then it wasn't set and we take
>>> the current time */
>>> if (now1 <= 0)
>>> diff --git a/src/modules/alsa/alsa-util.c b/src/modules/alsa/alsa-util.c
>>> index bb4e307..4a29a9a 100644
>>> --- a/src/modules/alsa/alsa-util.c
>>> +++ b/src/modules/alsa/alsa-util.c
>>> @@ -1141,10 +1141,11 @@ snd_pcm_sframes_t pa_alsa_safe_avail(snd_pcm_t
>>> *pcm, size_t hwbuf_size, const pa
>>> return n;
>>> }
>>> -int pa_alsa_safe_delay(snd_pcm_t *pcm, snd_pcm_sframes_t *delay,
>>> size_t hwbuf_size, const pa_sample_spec *ss, pa_bool_t capture) {
>>> +int pa_alsa_safe_delay(snd_pcm_t *pcm, snd_pcm_status_t *status,
>>> snd_pcm_sframes_t *delay, size_t hwbuf_size, const pa_sample_spec *ss,
>>> + pa_bool_t capture) {
>>> ssize_t k;
>>> size_t abs_k;
>>> - int r;
>>> + int err;
>>> snd_pcm_sframes_t avail = 0;
>>> pa_assert(pcm);
>>> @@ -1154,10 +1155,16 @@ int pa_alsa_safe_delay(snd_pcm_t *pcm,
>>> snd_pcm_sframes_t *delay, size_t hwbuf_si
>>> /* Some ALSA driver expose weird bugs, let's inform the user about
>>> * what is going on. We're going to get both the avail and delay
>>> values so
>>> - * that we can compare and check them for capture */
>>> + * that we can compare and check them for capture.
>>> + * This is done with snd_pcm_status() which provides
>>> + * avail, delay and timestamp values in a single kernel call to
>>> improve
>>> + * timer-based scheduling */
>>> - if ((r = snd_pcm_avail_delay(pcm, &avail, delay)) < 0)
>>> - return r;
>>> + if ((err = snd_pcm_status(pcm, status)) < 0)
>>> + return err;
>>> +
>>> + avail = snd_pcm_status_get_avail(status);
>>> + *delay = snd_pcm_status_get_delay(status);
>>> k = (ssize_t) *delay * (ssize_t) pa_frame_size(ss);
>>> diff --git a/src/modules/alsa/alsa-util.h b/src/modules/alsa/alsa-util.h
>>> index a4beed2..236a329 100644
>>> --- a/src/modules/alsa/alsa-util.h
>>> +++ b/src/modules/alsa/alsa-util.h
>>> @@ -125,7 +125,7 @@ int pa_alsa_recover_from_poll(snd_pcm_t *pcm, int
>>> revents);
>>> pa_rtpoll_item* pa_alsa_build_pollfd(snd_pcm_t *pcm, pa_rtpoll
>>> *rtpoll);
>>> snd_pcm_sframes_t pa_alsa_safe_avail(snd_pcm_t *pcm, size_t
>>> hwbuf_size, const pa_sample_spec *ss);
>>> -int pa_alsa_safe_delay(snd_pcm_t *pcm, snd_pcm_sframes_t *delay,
>>> size_t hwbuf_size, const pa_sample_spec *ss, pa_bool_t capture);
>>> +int pa_alsa_safe_delay(snd_pcm_t *pcm, snd_pcm_status_t *status,
>>> snd_pcm_sframes_t *delay, size_t hwbuf_size, const pa_sample_spec *ss,
>>> pa_bool_t capture);
>>> int pa_alsa_safe_mmap_begin(snd_pcm_t *pcm, const
>>> snd_pcm_channel_area_t **areas, snd_pcm_uframes_t *offset,
>>> snd_pcm_uframes_t *frames, size_t hwbuf_size, const pa_sample_spec *ss);
>>> char *pa_alsa_get_driver_name(int card);
>>
>>
>
--
David Henningsson, Canonical Ltd.
https://launchpad.net/~diwic
More information about the pulseaudio-discuss
mailing list