summaryrefslogtreecommitdiffstats
path: root/src/pulsecore
Commit message (Collapse)AuthorAgeFilesLines
* make sure we send a started messages when we are uncorkingLennart Poettering2008-10-062-0/+7
|
* disable valgrind macro usage for now since valgrind generates a lot of ↵Lennart Poettering2008-10-051-7/+11
| | | | spurious warnings as it seems
* do not cleanup staticly allocated memory unless we are in valgrind modeLennart Poettering2008-10-043-0/+9
|
* add new API function pa_in_valgrind() to check forLennart Poettering2008-10-042-0/+28
|
* allow - in sample namesLennart Poettering2008-10-032-12/+20
|
* make a few casts explicit to remove compiler warningsLennart Poettering2008-10-031-2/+2
|
* a bit of late pa_bool_t'izationLennart Poettering2008-10-032-2/+2
|
* user lrint() and friends in inner loops instead of normal C casts to speed ↵Lennart Poettering2008-10-034-12/+12
| | | | up a few things
* use PA_FLOAT32_SWAP where usefulLennart Poettering2008-10-031-8/+4
|
* optimize mixing routines a bit by pulling the multiplication with the global ↵Lennart Poettering2008-10-031-112/+85
| | | | volume out of the inner loop by applying it first to the per-stream volumes
* when we mix into a 16bit accumulator make sure we clamp before we scale with ↵Lennart Poettering2008-10-021-4/+4
| | | | a volume to avoid range faults when multiplying
* make simple protocol not crash when allocating a memory blockLennart Poettering2008-10-021-4/+8
|
* PA_WARN_REFERENCE works only for ELF targetsLennart Poettering2008-10-011-1/+1
|
* remove useless log message, re #367Lennart Poettering2008-10-011-1/+0
|
* a few FreeBSD fixes, from alexisLennart Poettering2008-10-013-2/+4
| | | | Signed-off-by: Lennart Poettering <lennart@poettering.net>
* Make the shared memory segment size configurableLennart Poettering2008-10-016-9/+42
| | | | | This is useful only on systems where memory overcommit is not available or disabled.
* have make_random_dir respect $TMPDIRStelian Ionescu2008-09-291-7/+19
| | | | | | | with the attached patch, make_random_dir first tries to use $TMPDIR, then $TMP finally defaulting to "/tmp" Signed-off-by: Lennart Poettering <lennart@poettering.net>
* fix compilation errors in priority queue codeLennart Poettering2008-09-272-3/+4
|
* add a generic priority queue implementationLennart Poettering2008-09-262-0/+319
|
* make pa_idxset_trivial_compare_func() do a full compare instea of just ↵Lennart Poettering2008-09-261-1/+1
| | | | equakity check
* make sure ~/.pulse exists before we create the runtime dir link beneath itLennart Poettering2008-09-121-3/+8
|
* make sure peaks resampler also works for very short input buffersLennart Poettering2008-09-091-11/+20
|
* unbreak pa_idxset_rrobinLennart Poettering2008-09-091-2/+1
|
* move autospawn lock to pulsecore/ since we don't need it in the client anymoreLennart Poettering2008-09-082-0/+362
|
* Add new option to disable remixing from/to LFE and set it to on by defaultLennart Poettering2008-09-056-7/+16
|
* include build and runtime host information in debug outputLennart Poettering2008-09-052-0/+9
|
* if we are exiting due to idleness, say soLennart Poettering2008-09-051-0/+1
|
* command_get_info() segv in some conditionsMarc-André Lureau2008-09-031-1/+1
| | | | Signed-off-by: Lennart Poettering <lennart@poettering.net>
* Implement "early requests" mode.Lennart Poettering2008-09-031-34/+138
| | | | | | | PA_STREAM_EARLY_REQUESTS is a new flag that will modify buffering metric selection behaviour a bit. This code is good for broken ALSA/OSS clients that ignore 'readability' on the fds in question and schedule audio via usleep() instead.
* Rework pa_machine_id() a bitLennart Poettering2008-09-031-10/+27
| | | | | Guarantee this function never fails, use POSIX gethostid as last resort. Add comments.
* avoid rounding errors on requested buffering metricsLennart Poettering2008-09-011-11/+18
|
* output relative timestamps in addition to absolute timestamps when loggingLennart Poettering2008-09-011-4/+14
|
* remove a few more gcc warningsLennart Poettering2008-08-306-10/+10
|
* fix a few compiler warnings on older gccLennart Poettering2008-08-295-6/+6
|
* reword amd64 messageLennart Poettering2008-08-291-1/+1
|
* change default log level for the library to PA_LOG_ERROR to avoid spamming ↵Lennart Poettering2008-08-291-1/+1
| | | | to stderr more often
* downgrade a few messagesLennart Poettering2008-08-291-4/+4
|
* the native atomic ops implementation for amd64 seems to work fineLennart Poettering2008-08-291-3/+3
|
* Change return value of cmpxchg atomic op to pa_bool_tLennart Poettering2008-08-291-14/+16
|
* optionally add timestampts to every line loggedLennart Poettering2008-08-292-9/+32
|
* fix up latency before calling into stream code, to make sure we don't ask ↵Lennart Poettering2008-08-294-13/+16
| | | | for too much data to early
* rework pa_ulog2 and base it on __builtin_clz if available, make ↵Lennart Poettering2008-08-291-15/+21
| | | | pa_make_power_of_two based on it
* add byte-to-usec and usec-to-byte converters that round up, not downLennart Poettering2008-08-282-0/+36
|
* call close() in a loop to catch EINTRLennart Poettering2008-08-261-1/+9
|
* add a few missing castsLennart Poettering2008-08-205-12/+12
|
* use final glibc eventfd() instead of our homegrown syscall invocationsLennart Poettering2008-08-191-37/+13
|
* rework cpu limit logic to use monotonic instead of wall clock timeLennart Poettering2008-08-191-0/+1
|
* simplify pa_start_child_for_read by using ↵Lennart Poettering2008-08-191-57/+3
| | | | pa_close_all()/pa_reset_sigs()/pa_unblock_sigs()
* add a few more gcc warning flags and fix quite a few problems found by doing soLennart Poettering2008-08-1937-326/+346
|
* Merge commit 'flameeyes/autoconf-2.62'Lennart Poettering2008-08-181-1/+1
|\
generated by cgit (git 2.34.1) at 2024-11-12 17:59:03 +0000