Pass a std::string to PathTraitsUTF8::Relative(), implicitly casting
it to std::string_view. This selects the right overload which returns
std::string_view instead of `const char *`; the latter could return
`nullptr` which would cause the implicit conversion of the return
value to std::string_view to crash.
Regression caused by commits ead208987d and a98d627c0b.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/995
This allows users to disable the "CUE files as directories" feature
without having to disable the CUE playlist plugin completely. This
feature has been annoying some users.
The bug https://bugzilla.samba.org/show_bug.cgi?id=11413 makes MPD
crash after at most a minute of using the plugin. Since this bug is
five years old already and it doesn't look like it will ever be fixed,
all libsmbclient code in MPD is scheduled for removal. For now, the
plugin is disabled by default so people are less likely to hit the
crash bug.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/991
This got lost in commit 5d597a3646 (v0.21.19), but it was never
noticed because the state_file_interval was way too short due to
commit 3413d1bf23, fixed recently by commit 27cc7b352d
Fixes regression by commit 23d5a2b862 -
that commit always pretended that any Opus file has both track and
album gain, and thus disabled the fallback to the other if one is not
set.
This patch changes the logic to only submit ReplayGain if at least one
value is set, and apply the offset only to that value. If none is
available, then the new check in HandleAudio() will submit only the
output gain.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/977
Oh no, 3413d1bf23 was broken! Instead of passing a number as
"seconds" to the duration constructor, it just abused the duration
constructor as cast operator, which caused custom state_file_interval
settings to be extremely short.
This allows canceling the blocking method LockWaitWriteAvailable(),
and thus allows breaking free of misbehaving ALSA drivers, avoiding a
MPD lockup.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/966
This fixes a spurious "single" mode bug which occurs when using "play"
or "seek" to start playback on the song that is currently paused: in
that case, the main thread never queues the next song, and at the end
of the song, the player thread exits Run(), stopping playback, and
after that, the main thread starts the next song without considering
"single" mode.
By calling OnPlayerSync(), we ensure that the main thread gets a
chance to queue the next song before the player thread exits the Run()
loop.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/850
The log levels have always been very confusing (and badly named), but
this was most confusing: if there's a log level called "default", why
is it not the default?
Closes https://github.com/MusicPlayerDaemon/MPD/issues/926
Oh the horror! This plugin cannot possibly ever have worked. It was
broken from the start, when it was added in commit 37796699cf nearly
twelve (!) years ago.
The plugin would always read at sector boundaries, so it could only
ever work at multiples of 2 kB.
While libsndfile doesn't like partial reads in the middle of a file
(see commit 95ac6071b9), it allows partial reads at the end of a file.
It doesn't pay attention to the file size when issuing a read.
Commit ecb67a1ed1 (MPD 0.18.12) was a regression: previously,
partial reads at the end of a file were possible, but switching to
decoder_read_full() made this an error condition. This way, a portion
at the end of each file was lost, leading to corruption with gapless
playback (https://github.com/MusicPlayerDaemon/MPD/issues/936).
This fix switches to the newly introduced function
decoder_read_much(), which does the same as the code before commit
ecb67a1ed1.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/936
Our AudioObjectGetPropertyDataT() wrapper throws exception on error,
and calling it from OSXOutput::Disable() can cause MPD crash due to
std::terminate().
Closes https://github.com/MusicPlayerDaemon/MPD/issues/932
When mounting something over a directory that is already a mount
point, CompositeStorage::Mount() silently overwrites the previously
mounted storage, disposing it. After that, SimpleDatabase::Mount()
will fail and handle_mount() will roll back the
CompositeStorage::Mount() command, effectively unmounting what was
there before (and also leaking memory).
Closes https://github.com/MusicPlayerDaemon/MPD/issues/918
Bug #915 is about an I/O exception thrown where none was allowed,
leading to crash via std::terminate(). However, instead of catching
and logging the error inside the decoder plugin, it should be able to
propagate the I/O error to the MPD core, so MPD can avoid trying other
decoder plugins.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/915
The formula in osx_output_score_sample_rate() to detect multiples of
the source sample rate was broken: when given a 44.1 kHz input file,
it preferred 16 kHz over 48 kHz, because its `frac_portion(16)=0.75`
is smaller than `frac_portion(48)=0.91`.
That formula, introduced by commit 40a1ebee29, looks completely
wrong. It doesn't do what the code comment pretends it does.
Instead of using that `frac_portion` to calculate a score, this patch
adds to the score only if `frac_portion` is nearly `0` or `1`. This
means that the factor is nearly integer.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/904
A bug report (https://github.com/MusicPlayerDaemon/MPD/issues/912)
suggests that on Linux, reading on `cifs` files may rarely return 0 (=
end of file) before the end of the file has really been reached. But
that's just a theory which I need to validate, so this runtime check
shall catch this condition before the assertion in
DecoderBridge::Read() crashes MPD. Let's see.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/912
An assertion failure in UpdateQueuedSong() could trigger because the
`prev` parameter is always `nullptr`, but `queued` may be set. And in
fact, calling UpdateQueuedSong() is only necessary when the queued
song was edited, to re-queue it with the new range.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/901
When the client wants to seek, but the decoder has already finished
decoding the current song, the player restarts the decoder with an
initial seek at the new position. When this initial seek fails, MPD
pretends nothing has happened and plays this song from the start.
With this new flag, a restarted decoder marks the initial seek as
"essential" and fails the decoder if that seek fails.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/895
The WildMidi project added the pkg-config file in version 0.3.3, but
unfortunately, Debian still doesn't ship it 4 years later:
https://bugs.debian.org/916631
However, for cross-compiling, the pkg-config file is very helpful.
This is the final piece of the series to establish io_uring support on
Linux.
MPD doesn't need io_uring for its efficient bulk I/O support, but to
allow file I/O to be cancelled. This is a big problem on CIFS/NFS
mounts where processes sleep uninterruptable if the file server
disappears, deadlocking MPD.
With io_uring, a flaky NFS connection allows MPD to continue to work
(even though there are still deadlocks inside MPD which need to be
addressed).
This plugin does not yet use cancellable `open()` using
`IORING_OP_OPENAT`. This will be implemented later.
Lots of other optimization opportunities for io_uring are still
missing as well - for example the database update could benefit a lot,
but unfortunately, io_uring doesn't have `readdir()` support just yet.
The command is used to configure the client's connection, and this
shouldn't require any permissions. The client should be able to do
that before sending a password.
This fixes a freeze bug in the NFS input/storage plugins: when libnfs
auto-reconnets after a failure, it installs the new socket on the same
file descriptor number. MPD's attempt to unregister the old socket by
calling SocketMonitor::Steal() from NfsConnection::ScheduleSocket()
fails because the new/old socket number is not registered in epoll, so
epoll_ctl() returns ENOENT. The problem is that it left
`scheduled_flags`, and so subsequent Schedule() calls will use
`EPOLL_CTL_MOD`, which will fail again and again. Instead, we need to
use `EPOLL_CTL_ADD` to register the new socket.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/806
Closes https://github.com/MusicPlayerDaemon/MPD/issues/756
Unescape the base path and the path coming from the server (href) to fix the
comparison when the server uses different escaped characters.
The outputted name need to be unescaped. Doing that before or after the
HrefToEscapedName() call should not change the current behavior.
If the file name is "Hello & bye", 3 CharacterData events will be sent with the
State::HREF state:
- "Hello%20"
- "&"
- "%20bye"
Reproduced with files hosted on an apache2 DAV server: 2.4.38-3+deb10u3.
Documentation says the limit is 5, but it was really 10 (at least
since 2004). But since MPD wants to promote using many small clients
idling around, and these clients consume only very few resources, it
seems reasonable to raise this limit's default value.
Aparently, libcdio sometimes returns empty filenames, causing MPD
crashes. This shouldn't really happen, and I consider this a libcdio
bug - but if it happens, people blame MPD, so let's add a check.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/776
Using libgme 0.6.2 on macOS, it appears that gme_info_t strings can be
empty, which creates weird track titles: (001/050)
This adds an additional check for an empty string.
"The issue is that ParseCommandArgSignedSongTime parses with
SongTime::FromS, not SignedSongTime::FromS, before casting back to a
SignedSongTime for the return. With x86 overflow rules this doesn't
matter, but on ARM the first cast turns negative values to zero."
Closes https://github.com/MusicPlayerDaemon/MPD/issues/757
This reverts commit c84bae739a. A
configuration option is not necessary, because the PcmConvert
constructor knows already whether integer or floating point is needed.
It appears that [[fallthrough]] is valid in C++ but not in C. And
in some Clang versions (e.g. Clang 11 on macOS), Clang is pedantic
about this and considers it an error to use [[fallthrough]] in a
.c file such as src/util/format.c.
This changes makes gcc_fallthrough a no-op under Clang in C files.
This prepares the migration away from strptime() for Windows
portability.
But the real reason I'm doing this is that strptime() on Apple is
buggy: strptime("14", "%H%M%S") (without separating colons) succeeds
even though only the hour has been parsed. This fixes recent Travis
failures in the ParseISO8601() unit test.
This can cause request completion in the I/O thread before this
constructor returns, leaving the object in an abstract state, causing
a crash due to pure virtual method call. We should not start the
request until this object is fully constructed.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/665
The ALSA "null" driver opens /dev/null and returns the file handle
from snd_pcm_poll_descriptors(), but /dev/null cannot be used with
epoll, the epoll_ctl() system call returns -EPERM. This means that
the ALSA output hangs, eventually freezing the whole MPD process.
This commit adds a workaround to the MultiSocketMonitor class which is
used by the ALSA output plugin.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/695
Most importantly, this commit translates ZZIP_ENOENT to
std::system_error(ENOENT) so IsFileNotFound() returns true and
find_stream_art() can suppress the log line.
Commit 13208bf5a7 added range support to
the `move` command, but applied the wrong offset to the `to` variable.
When the source range is before the current song, and the song thus
gets decremented by the range size, then the final destination offset
must also be decremented by the range size.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/663
On linux-rt, kernel IRQ threads are configured with priority=50, and
this change configures MPD somewhat below that priority, leaving some
room for other programs to be configured in between.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/643
Without requesting the property, "good" WebDAV servers would not send
it, and so MPD could never recognize a directory, failing the database
update.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/660
Apparently, libmpcdec sets gain/peak variables to zero if they are not
present. This clashes with our formula and results in bogus values
which cause noise during playback.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/640
This commit adds a PlaylistPlugin attribute "as_folder" which for now
is only enabled in the "CUE" playlist plugin (which handles separate
"*.cue" files). If a playlist with this flag set is being scanned
during database update, it will be parsed and its contents will be
added to the database. This allows clients to inspect them like
directories and its contents will be searchable.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/39
The new Response instance in the `catch` block didn't have the
`command` attribute set, so the error response didn't indicate which
command had failed, which however is required in the MPD protocol.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/628
SID files are generally collections of tunes, so a SID name field makes
sense as an MPD album. The SID tune information list (STIL) has name
and title fields for individual tunes, when such are known, but MPD is
currently not using the STIL.
High Voltage SID Collection (HVSC) metadata fields are encoded in
windows-1252, as described in DOCUMENTS/SID_file_format.txt:
https://www.hvsc.c64.org/download/C64Music/DOCUMENTS/SID_file_format.txt
If utf-8 transcoding fails, or the ICU library is unavailable, fall
back to plain ASCII and replace other characters with '?'.
Version 10.2+0.93+1 was released five years ago in 2014 and is the
first version to feature cdio_cddap_free_messages(). There is no way
to check the libcdio-paranoia version at compile time, so let's just
remove support for older versions instead of attempting to fix the
cdio_cddap_free_messages() check at build time.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/613
This reverts commit f7ed7446ae. It was
a bad idea, because MAD_F_MIN and MAD_F_MAX do not represent the
clamping limits, but the theoretical minimum and maximum values of the
mad_fixed_t data type.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/617
This is useful in multiple mpd instances scenario, or multiple pulse outputs defined on the same mpd instance.
It is actually a more flexible way to route flows than the "sink" parameter, letting the PulseAudio routing do its job, but with the ability to isolate routing for each output.
If not specified, the role remains like it was before this commit, ie "music"
The check IsSeekableCurrentSong() was added by commit
44b200240f in version 0.20.19, but it
caused a regression: by doing the branch only if the current song is
seekable, the player would restart the current song if it was not
seekable, and later the initial seek would fail; but we already know
it's not seekable, and so we should fail early.
Applying software volume to S16 samples means several bits of
precision are lost; at 25% volume, two bits are lost. Additionally,
dithering adds some noise.
The problem gets worse when you apply the software volume code twice:
for the software mixer volume, and again for the replay gain. This
loses some more precision and adds even more dithering noise, which
can become audible (see
https://github.com/MusicPlayerDaemon/MPD/issues/542).
By converting everything to 24 bit, we need to shift only two bits to
the right instead of ten, losing nearly no precision, and dithering is
not needed. Even if the output device is unable to play S24 directly,
we can convert back to S16 with only one stage of dithering.
Closes https://github.com/MusicPlayerDaemon/MPD/issues/542
MPD used to do that when this code lived in the player thread, but it
was removed by commit 98a7c62d7a4f716d90af6d78e18d1a3b10bc54b3; and
the replacement code in the ALSA output plugin didn't have it.
Without this timer, DispatchSockets() may disable the
MultiSocketMonitor and if Play() doesn't get called soon, it never
gets a chance to generate silence. However if Play() gets called,
generating silence isn't necessary anymore...
Resulting from this misdesign (added by commit ccafe3f3cf in 0.21.3),
the silence generator didn't work reliably.
Apparently, if snd_pcm_drain() returns EAGAIN, it does not actually
want to be called again; the next call will snd_pcm_drain() will also
return EAGAIN, forever, even though the PCM state has meanwhile
switched to SND_PCM_STATE_SETUP. This causes a busy loop; to fix
this, we should always check snd_pcm_state() to see if draining is
really required.
Stapper