mirrors/pulseaudio
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/pulseaudio/pulseaudio.git synced 2025-11-03 09:01:50 -05:00
Code Issues Projects Releases Packages Wiki Activity Actions

* rename "latency correction" to "write index correction"

Browse source 
* add read index invalidation code
* rename "ipol_event" stuff to "auto_timing_update"
* remove buffer_usec field from pa_timing_info, since it can be easily calculated from write_index and read_index anyway
* add read_index_corrupt field to "pa_timing_info", similar to the already existing write_index_corrupt field
* restart automatic timing update event every time a query is issued, not just when the last event elapsed
* proper invalidation code for pa_stream_flush()
* do tarsnport/sink/source latency correction for playback time only when device is not corked


git-svn-id: file:///home/lennart/svn/public/pulseaudio/trunk@686 fefdeb5f-60dc-0310-8127-8f9354f1896f
This commit is contained in:
Lennart Poettering 2006-04-12 17:17:23 +00:00
parent 77c2a1f561
commit 49b3150434
3 changed files with 217 additions and 135 deletions
Download patch file Download diff file Expand all files Collapse all files

23
src/polyp/def.h
View file

@ -204,13 +204,14 @@ typedef enum pa_subscription_event_type {
* pa_stream_update_timing_info() and pa_stream_get_timing_info(). The
* total output latency a sample that is written with
* pa_stream_write() takes to be played may be estimated by
* sink_usec+buffer_usec+transport_usec. The output buffer to which
* buffer_usec relates may be manipulated freely (with
* sink_usec+buffer_usec+transport_usec. (where buffer_usec is defined
* as pa_bytes_to_usec(write_index-read_index)) The output buffer
* which buffer_usec relates to may be manipulated freely (with
* pa_stream_write()'s seek argument, pa_stream_flush() and friends),
* the buffers sink_usec and source_usec relate to are first-in
* first-out (FIFO) buffers which cannot be flushed or manipulated in any
* way. The total input latency a sample that is recorded takes to be
* delivered to the application is:
* first-out (FIFO) buffers which cannot be flushed or manipulated in
* any way. The total input latency a sample that is recorded takes to
* be delivered to the application is:
* source_usec+buffer_usec+transport_usec-sink_usec. (Take care of
* sign issues!) When connected to a monitor source sink_usec contains
* the latency of the owning sink. The two latency estimations
@ -226,7 +227,6 @@ typedef struct pa_timing_info {
* limited und unreliable itself. \since
* 0.5 */
pa_usec_t buffer_usec; /**< Time in usecs the current buffer takes to play. For both playback and record streams. */
pa_usec_t sink_usec; /**< Time in usecs a sample takes to be played on the sink. For playback streams and record streams connected to a monitor source. */
pa_usec_t source_usec; /**< Time in usecs a sample takes from being recorded to being delivered to the application. Only for record streams. \since 0.5*/
pa_usec_t transport_usec; /**< Estimated time in usecs a sample takes to be transferred to/from the daemon. For both playback and record streams. \since 0.5 */
@ -240,14 +240,21 @@ typedef struct pa_timing_info {
* info was current . Only write
* commands with SEEK_RELATIVE_ON_READ
* and SEEK_RELATIVE_END can corrupt
* write_index. */
* write_index. \since 0.8 */
int64_t write_index; /**< Current write index into the
* playback buffer in bytes. Think twice before
* using this for seeking purposes: it
* might be out of date a the time you
* want to use it. Consider using
* PA_SEEK_RELATIVE instead. \since
* 0.8 */
* 0.8 */
int read_index_corrupt; /**< Non-zero if read_index is not
* up-to-date because a local pause or
* flush request that corrupted it has
* been issued in the time since this
* latency info was current. \since 0.8 */
int64_t read_index; /**< Current read index into the
* playback buffer in bytes. Think twice before
* using this for seeking purposes: it