oysteikt/mpd
0
0
Fork 0
Code Issues Pull Requests Releases Activity
11,880 Commits 1 Branch 144 Tags 56 MiB
c629ac7168
Commit Graph

11880 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Max Kellermann
78ca5491e6 db/Interface: GetUpdateStamp() returns std::chrono::system_clock::time_point 2017-02-11 23:02:17 +01:00
Max Kellermann
4146475c73 util/ChronoUtil: new utility library for std::chrono 2017-02-11 22:23:33 +01:00
Max Kellermann
329c3ab21b fs/FileInfo: use std::chrono::system_clock 2017-02-10 23:48:21 +01:00
Max Kellermann
3b7f6641d2 TimePrint: std::chrono::system_clock support 2017-02-10 23:48:21 +01:00
Max Kellermann
781487c4dd thread/Thread: use BoundMethod 2017-02-10 22:46:09 +01:00
Max Kellermann
5ba5bc8ba1 thread/Thread: move code to Run() 2017-02-10 22:43:55 +01:00
Max Kellermann
82c66ce078 thread/Thread: Start() returns void 
Since we switched to C++ exceptions, there is no code path which
returns false.
 2017-02-10 22:41:29 +01:00
Max Kellermann
9324fbf921 event/Thread: remove the Mutex 
We don't need to access Thread::handle early inside the EventThread,
so we don't need this trick anymore.
 2017-02-10 22:29:31 +01:00
Max Kellermann
5e081de14a IOThread: move EventThread instance into struct Instance 
Eliminate global variables.
 2017-02-10 22:25:06 +01:00
Max Kellermann
d1456ae039 test/*: use class EventThread instead of ScopeIOThread 2017-02-10 22:25:01 +01:00
Max Kellermann
115af4f565 event/Loop: Break() is no-op if "quit" is already set 2017-02-10 22:24:48 +01:00
Max Kellermann
822724d1aa event/Thread: auto-stop in the destructor 2017-02-10 22:24:46 +01:00
Max Kellermann
87c9856b20 input/alsa: use the EventLoop& passed to init() instead of io_thread_get() 2017-02-10 22:24:44 +01:00
Max Kellermann
835136dcd3 output/Thread: add assertion on the ao_plugin_play() result 2017-02-10 22:24:36 +01:00
Max Kellermann
4a80e9cb25 output/alsa: copy the PcmExport result to the ring_buffer 
.. and not the input data.

Regression from commit b1c7649edb (integer underflow).

 https://bugs.musicpd.org/view.php?id=4639
 2017-02-10 22:23:00 +01:00
Max Kellermann
de80c270bd IOThread: move code to class EventThread 2017-02-10 21:40:39 +01:00
Max Kellermann
b92bff2658 IOThread: remove unused function io_thread_inside() 2017-02-10 21:40:30 +01:00
Max Kellermann
42f1e26540 IOThread: remove unused Cond variable 2017-02-10 21:29:54 +01:00
Max Kellermann
cfd056231b output/alsa: use the EventLoop& parameter instead of io_thread_get() 2017-02-10 21:21:15 +01:00
Max Kellermann
20ae84bff9 {input,mixer}/alsa: cancel the DeferredMonitor in the destructor 
Yet another potential crash bug fix.
 2017-02-10 15:05:49 +01:00
Max Kellermann
b1c7649edb output/alsa: non-blocking mode 
Use SND_PCM_NONBLOCK, and perform all snd_pcm_writei() calls in the
IOThread.  Use a lockless queue to copy data from the OutputThread to
the IOThread.

This rather major change aims to improve MPD's internal latency.  All
waits are now under MPD's control, instead of blocking inside
libasound2.

As a side effect, an output's filter is now decoupled from the actual
device I/O, which solves a major latency problem with the conversion
filter on slow CPUs and small period buffers.  See:

 https://bugs.musicpd.org/view.php?id=3900
 2017-02-09 21:36:18 +01:00
Max Kellermann
853740f1e2 Main: use the IOThread for outputs and mixers 
The main EventLoop can block for a long time while a client's command
runs, and is therefore inappropriate for internal engine I/O.  This
fixes a serious regression for at least the "httpd" output, which used
to be hard-coded for the IOThread, but now receives the main EventLoop
as an initialization parameter.

For the mixers, this doesn't make much of a difference.  They are not
latency critical.
 2017-02-09 21:33:49 +01:00
Max Kellermann
14986b153a event/Loop: use std::lock_guard 2017-02-09 21:26:55 +01:00
Max Kellermann
9e503b21c1 {input,mixer}/alsa: move code to lib/alsa/NonBlock.cxx 2017-02-09 21:24:24 +01:00
Max Kellermann
67a958a326 Merge branch 'v0.20.x' 2017-02-09 21:24:20 +01:00
Max Kellermann
7372c931b3 event/Loop: make IsInsideOrNull() available in the NDEBUG build 
Fixes build breakage by commit 4e5271fcdf7; and this method does make
sense in non-debug builds.
 2017-02-09 21:21:49 +01:00
Max Kellermann
29e1b6e465 mixer/alsa: reset the MultiSocketMonitor in the destructor 
Fixes potential crash bug.
 2017-02-09 21:13:19 +01:00
Max Kellermann
eda06993f8 event/MultiSocketMonitor: add method Reset() 2017-02-09 21:12:23 +01:00
Max Kellermann
4b30ef1cf2 event/MultiSocketMonitor: use C++11 initializer 2017-02-09 21:12:23 +01:00
Max Kellermann
e92e5e8eb8 event/MultiSocketMonitor: more API documentation 
Now ClearSocketList() may only be called from PrepareSockets().
Calling it before destroying the object doesn't work properly, because
it doesn't unregister the TimeoutMonitor and the IdleMonitor.  Some of
its callers need to be fixed.
 2017-02-09 21:12:23 +01:00
Max Kellermann
4e5271fcdf event/Call: allow usage during shutdown 
Change EventLoop::IsInside() call to EventLoop::IsInsideOrNull().
This means that BlockingCall() may be used during shutdown, after the
main EventLoop::Run() has finished.  This is important because mixers
are currently registered in the main EventLoop.
 2017-02-09 21:12:23 +01:00
Max Kellermann
3c55487a16 configure.ac: don't require libsidutils when building with libsidplayfp 
The libsidplayfp fork has merged libsidutils into the main library.
The libsidutils we used to link with was part of the original
libsidplay project.
 2017-02-09 13:09:03 +01:00
Max Kellermann
76a1cae5d8 {input,mixer}/alsa: fix off-by-one bug in count check 
Doesn't make a practical difference - but it's more correct this way.
 2017-02-09 12:46:49 +01:00
Max Kellermann
3850716522 command/Database: add "sort" parameter to "find" and "search" 
Implement the second part of https://bugs.musicpd.org/view.php?id=3990
 2017-02-08 11:22:08 +01:00
Max Kellermann
1e0a60e73d doc/protocol: remove UTF-8 explanation 
These days, this is common knowledge, and doesn't need a lengthy
explanation.
 2017-02-08 11:22:08 +01:00
Max Kellermann
e9c99e0518 DetachedSong: add LightSong cast operator 2017-02-08 10:38:25 +01:00
Max Kellermann
332baa4f67 DetachedSong: don't declare empty destructor 
An explicit destructor prevents usage of implicit move operators, even
if it's empty.  Therefore, declaring a defaulted destructor with GCC
attribute "noinline" does what we want without preventing those
implicit operators.
 2017-02-08 10:24:45 +01:00
Max Kellermann
08879d2a20 DetachedSong: add move operator 2017-02-08 10:24:40 +01:00
Max Kellermann
1292af4768 Revert "DetachedSong: remove explicitly-defaulted copy/move constructors" 
This reverts commit 67b7d46432.

Turns out I was wrong, and mentioning these does make a difference:
the implicit move constructor is not defined in the presence of a
user-declared destructor.
 2017-02-08 10:21:35 +01:00
Max Kellermann
4d88a099f9 Compiler.h: add gcc_noinline 2017-02-08 10:17:21 +01:00
Max Kellermann
777e15bd78 db/DatabaseSong: make the Storage optional 
Some database plugins don't use a Storage (e.g. UPnP), and with this
plugin, DatabaseDetachSong() can crash.
 2017-02-08 10:05:55 +01:00
Max Kellermann
f689e28958 SongLoader: return instance, not pointer 2017-02-08 10:02:08 +01:00
Max Kellermann
d184231169 db/DatabaseSong: DatabaseDetachSong(uri) returns instance, not pointer 2017-02-08 09:59:12 +01:00
Max Kellermann
7225e919fc db/DatabaseSong: use AtScopeExit() 2017-02-08 09:58:40 +01:00
Max Kellermann
7a185f1ead queue/PlaylistUpdate: assert Database::GetSong()!=nullptr 2017-02-08 09:53:14 +01:00
Max Kellermann
c1fa5279f4 db/Interface: clarify GetSong() error handling 2017-02-08 09:50:30 +01:00
Max Kellermann
4f0fe66f69 DetachedSong: make the LightSong constructor public 
Sometimes, it's useful to construct a DetachedSong from a LightSong
even without having real_uri initialized.
 2017-02-08 09:45:37 +01:00
Max Kellermann
76380b2b45 DetachedSong: pass std::string&& to uri initializer 
Eliminate one temporary allocation.
 2017-02-08 09:42:09 +01:00
Max Kellermann
67b7d46432 DetachedSong: remove explicitly-defaulted copy/move constructors 
Mentioning these doesn't make a difference.
 2017-02-08 09:40:29 +01:00
Max Kellermann
29453ba196 client: add tag_mask attribute 
The "tagtypes" command now has several sub commands which can be used
to edit that mask.
 2017-02-08 09:06:11 +01:00