tfing wrote:
> I have quite some files with an empty album tag as they do not come
> from a particular album.
>
> If I want to look for those files and browse them, this happens:
> :: nc localhost 6600
> OK MPD 0.12.0
> find album ""
> ACK [2@0] {find} too few arguments for "find"
>
> I'd like to be able to browse those files in a client like gmpc.
> So these 2 items would have to be developed:
> - list album should report that some files have an empty tag
> - it should be possible to search for an empty tag with the find command
Patch-by: Marc Pavot
ref: http://musicpd.org/mantis/view.php?id=464
This only breaks "update" under list command mode and
no other commands. This can be done more optimally
without the extra heap allocation via xstrdup(); but is
uncommon enough to not matter.
It was a huge confusing mess of parameter passing around
and around. Add a few extra assertions to ensure we're
handling parent/child relationships properly.
This is like basename(3) but with predictable semantics independent
of C library or build options used. This is also much more strict
and does not account for trailing slashes (mpd should never deal with
trailing slashes on internal functions).
If we updated the mpd metadata database; then there's a chance
some of those songs in the playlist will have updated metadata.
So be on the safe side and increment the playlist version number
if _any_ song changed (this is how all released versions of mpd
did it, too).
This bug was introduced recently when making "update" threaded.
Thanks to stonecrest for the bug report.
Make the code more readable by moving the range checks to pcm_range().
gcc does quite a good job at optimizing it: the resulting binary is
exactly the same, although it contains a parametrized shift instead of
hard-coded boundaries.
There was a known deadlocking bug in the notify library: when the
other thread set notify->pending after the according check in
notify_wait(), the latter thread was deadlocked. Resolve this by
synchronizing all accesses to notify->pending with the notify object's
mutex. Since notify_signal_sync() was never used, we can remove it.
As a consequence, we don't need notify_enter() and notify_leave()
anymore; eliminate them, too.
During debugging, I found a deadlock between flushAudioBuffer() and
the audio_output_task(): audio_output_task() didn't notice that there
is a command, and flushAudioBuffer() waited forever in notify_wait().
I am not sure yet what is the real cause; work around this for now by
waking up non-finished audio outputs in every iteration.
Due to a merge error, I broke the function handleUpdate(). It did not
do anything for the global update, and it did not send a proper
response to the client. This patch fixes both bugs.
To check whether a device is really on or off, we should rather check
audio_output.open, instead of managing another variable. Wrap
audio_output.open in the inline function audio_output_is_open() and
use it instead of DEVICE_ON and DEVICE_OFF.
Send an output buffer to all output plugins at the same time, instead
of waiting for each of them separately. Make several functions
non-blocking, and introduce the new function audio_output_wait_all()
to synchronize with all audio output threads.
We have eliminated direct accesses to the audio_output struct from
the all output plugins. Make it opaque for them, and move its real
declaration to output_internal.h, similar to decoder_internal.h.
Pass the opaque structure to plugin.init() only, which will return the
plugin's data pointer on success, and NULL on failure. This data
pointer will be passed to all other methods instead of the
audio_output struct.
The JACK output plugin needs to reset its "opened" flag when the JACK
server fails. To prevent it from accessing the audio_output struct
directly introduce the API function audio_output_closed().
Reduce direct accesses to the audio_output struct from the plugins:
this time, eliminate all accesses to audio_output.name. The name is
required by some plugins for log messages.
Pass the globally configured audio_format as a const pointer to
plugin.init(). plugin.open() gets a writable pointer which contains
the audio_format requested by the plugin. Its initial value is either
the configured audio_format or the input file's audio_format.
To keep I/O nastiness and latencies away from the core, move the audio
output code to a separate thread, one per output. The thread is
created on demand, and currently runs until mpd exits.
Since flacSendChunk() is a trivial function and is only used in one
location, move the code there. The advantage is that calling
decoder_data() directly returns the decoder_command value, so we can
eliminate one decoder_get_command() call.
Support for bit rates except 16 bits (and 8 bits on little endian) has
always been broken. Since we added optimized functions for 8, 16,
24/32 bits, we can remove the generic flac_convert() function.
Instead of removing it, convert it to a wrapper function for
flac_convert_*().
flac_convert_16() runs a lot faster than the generic (and quite buggy)
function flac_convert(). flac_convert_16() is only used for
non-stereo files, since there is already flac_convert_stereo16().
By mistake, I casted the sample value to uint16_t, which is wrong.
This patch simplifies the code by using a int16_t pointer instead of
casting to int16_t* every time.
There is still a lot of duplicated code in flac_plugin.c and
oggflac_plugin.c. Move code from flac_plugin.c to _flac_common.c, and
use the new function flac_common_write() also in oggflac_plugin.c,
porting lots of optimizations over to it.
The inline function audio_format_sample_size() calculates how many
bytes each sample consumes. This function already takes into account
that 24 bit samples are 4 bytes long, not 3.
Instead of letting ALSA block for us (and potentially allowing
something stupid on certain hardware or drivers), we do the
sleeping ourselves. We calculate the sleep to be a fraction of
period_time to avoid oversleeping (and thus audible skipping).
A lot of the preparation was needed (and done in previous
months) in making update thread-safe, but here it is.
This was the first thing I made work inside a thread when I
started mpd-uclinux many years ago, and also the last thing I've
done in mainline mpd to work inside a thread, go figure.
pthreads with our existing signal blocking/handling is broken,
for now just sleep a bit in the child to prevent the CHLD handler
from being called too early. Also, improve error reporting when
handling SIGCHLD by storing the status to be called in the main
task (which can be logged, since we can't do logging inside the
sig handler).
Our linked-list implementation is wasteful and the
SongList isn't modified enough to benefit from being a linked
list. So use a more compact array of song pointers which
saves ~200K on a library with ~9K songs (on x86-32).
It hasn't been used in many years
commit 3a89afdd80
Author: Warren Dukes <warren.dukes@gmail.com>
Date: Sat Nov 20 20:28:32 2004 +0000
remove --update-db option
(SVN r2719)
This allows us to avoid the nasty repetition in strncmp(foo,
bar, strlen(foo)). We'll miss out on the compiler optimizing
strlen() into sizeof() - 1 for string literals for this; but we
don't use this it for performance-critical functions anyways...
This should save a few thousand ops. Not worth it to malloc
for such a small (3-words on 32-bit ARM and x86) structures.
Signed-off-by: Eric Wong <normalperson@yhbt.net>
The function decodeFirstFrame() allocates memory based on data from
the mp3 header. This can make the buffer size allocation overflow, or
lead to a DoS attack with a very large buffer. Cap this buffer at 8
million frames, which should really be enough for reasonable files.
The assertion on "!client_is_expired(client)" was wrong, because
writing the command response may cause the client to become expired.
Replace that assertion with a check.
A crafted mp4 file could cause an integer overflow in mp4_decode
function in src/inputPlugins/mp4_plugin.c. mp4ff_num_samples()
function returns some tainted value. sizeof(float) * numSamples is an
integer overflow operation if numSamples is too huge, so xmalloc will
allocate a small memory region. I constructe a mp4 file, and use
faad2 to open the file. mp4ff_num_samples() returns -1. So I think mpd
bears from the same problem.
Since the buffer size is known at compile time, we can save an
indirection by declaring it as a char array instead of a pointer.
That saves an extra allocation, and we can calculate with the
compile-time constant sizeof(data) instead of the attribute "max_len".
Shout encoder plugins are known at compile time. There is no reason
to use a complex data structure as "List" to manage them at runtime -
just put the pointers into a static array.
[mk: moved this patch after "Refactor and cleanup of shout Ogg and MP3
audio outputs". The original commit message follows, although it is
outdated:]
Creation of shout_mp3 audio output plugin. Basically I just copied the
existing shout plugin and replaced ogg with lame. Uses lame for mp3
encoding. Next step is to pull common functionality out of each shout
plugin and share it between them.
Configuration options for "shout_mp3" are the same as for "shout".
I've perhaps gone a bit overboard, but here's the current rundown:
Both Ogg and MP3 use the "shout" audio output plugin. The shout audio
output plugin itself has two new plugins, one for the Ogg encoder,
and another for the MP3 (LAME) encoder.
Configuration for an Ogg stream doesn't change. For an MP3 stream,
configuration is the same as Ogg, with two exceptions. First, you must
specify the optional "encoding" parameter, which should be set to "mp3".
See mpd.conf(5) for more details. Second, the "quality" parameter is
reversed for LAME, such that 1 is high quality for LAME, whereas 10 is
high quality for Ogg.
I've decomposed the code so that all libshout related operations
are done in audioOutput_shout.c, all Ogg specific functions are in
audioOutput_shout_ogg.c, and of course then all LAME specific functions
are handled in audioOutput_shout_mp3.c.
To develop encoder plugins for the shout audio output plugin, I basically
just mimicked the plugin system used for audio outputs. This might be
overkill, but hopefully if anyone ever wants to support some other sort
of stream, like maybe AAC, FLAC, or WMA (hey it could happen), they will
hopefully be all set.
The Ogg encoder is slightly less optimal under this configuration.
It used to send shout data directly out of its ogg_page structures. Now,
in the interest of encapsulation, it copies the data from its ogg_page
structures into a buffer provided by the shout audio output plugin (see
audioOutput_shout_ogg.c, line 77.) I suspect the performance impact
is negligible.
As for metadata, I'm pretty sure they'll both work. I wrote up a test
scaffold that would create a fake tag, and tell the plugin to send it
out to the stream every few seconds. It seemed to work fine. Of course,
if something does break, I'll be glad to fix it.
Lastly, I've renamed lots of things into snake_case, in keeping with
normalperson's wishes in that regard.
[mk: moved the MP3 patch after this one. Splitted this patch into
several parts; the others were already applied before this one. Fixed
a bunch GCC warnings and wrong whitespace modifications. Made it
compile with mpd-mk by adapting to its prototypes]
Support sending metadata to a shout server using shout_metadata_new()
and shout_metadata_add(). The Ogg Vorbis encoder does not support
this currently.
[mk: this patch was separated from Eric's patch "Refactor and cleanup
of shout Ogg and MP3 audio outputs", I added a description]
Preparing the merge of Eric Wollesen's patch "Refactor and cleanup of
shout Ogg and MP3 audio outputs": we declare one of the struct types
here, to make the merge smoother.
The Ogg encoder is slightly less optimal under this configuration. It
used to send shout data directly out of its ogg_page structures. Now,
in the interest of encapsulation, it copies the data from its ogg_page
structures into a buffer provided by the shout audio output plugin
(see audioOutput_shout_ogg.c, line 77.) I suspect the performance
impact is negligible.
[mk: this patch and its description was separated from Eric's patch
"Refactor and cleanup of shout Ogg and MP3 audio outputs"]
Begin dividing audioOutput_shout.c: move everything OGG Vorbis related
to audioOutput_shout_ogg.c. The header audioOutput_shout.h has to
keep its dependency on vorbis/vorbisenc.h, because it needs the vorbis
encoder types.
For this patch, we have to export several internal functions with
generic names to the ABI; these will be removed later when the encoder
plugin patches are merged.
Remove unused code which is in comments. Remove that comment about
"stolen code", since the plugin has changed much, and it isn't obvious
which parts are derived.
If the output device is already open, it may have modified
outAudioFormat; in this case, outAudioFormat is still valid, and does
not need an overwrite.
As long as the device isn't open, both attributes are not used. Since
they will both be initialized in audio_output_open(), we do not need
the initialization in audio_output_init().
Storing pointers to immutable audio_format structs isn't worth it,
because the struct itself isn't much larger than the pointer. Since
the shout plugin requires the user to configure a fixed audio format,
we can simply copy it in myShout_initDriver().
Save one allocation, since the whole audio_format struct is nearly the
same size as the pointer to it. Check audio_format_defined(af)
instead of af!=NULL.
free(NULL) isn't explicitly forbidden, but isn't exactly good style.
Check the rare case that the audio buffer isn't initialized yet in
closeAudioDevice(). In this case, we also don't have to call
flushAudioBuffer().
To make openAudioDevice() smaller and more readable, move code to a
static function. Also don't use realloc(), since the old value of the
buffer isn't needed anymore, saving a memcpy().
There are too many static variables in audio.c - organize all
properties of the audio buffer in a struct. The current audio format
is also a property of the buffer, since it describes the buffer's
data format.
audio_format_clear() sets an audio_format struct to an cleared
(undefined) state, which is both faster and smaller than memset(0).
audio_format_defined() checks if the audio_format struct actually has
a defined value (i.e. non-zero). Both can be used to avoid pointers
to audio_format, replacing the "NULL" value with an "undefined"
audio_format.
Since the caller chain doesn't care about the return value (except for
COMMAND_RETURN_KILL, COMMAND_RETURN_CLOSE), just return 0 if there is
nothing special. This saves one local variable initialization, and
one access to it.
Also remove one unreachable "return 1" from client_read().
Don't close the client within client_process_line(), return
COMMAND_RETURN_CLOSE instead. This is the signal for the caller chain
to actually close it. This makes dealing with the client pointer a
lot safer, since the caller always knows whether it is still valid.
The "!src" check in copyAudioFormat() used to hide bugs - one should
never pass NULL to it. There is one caller which might pass NULL, add
a check in this caller.
Instead of doing mempcy(), we can simply assign the structures, which
looks more natural.
The way we used non-blocking mode was HORRIBLE.
It was non-blocking to ALSA, but we end up blocking in a busy
loop that does absolutely NOTHING but retry. We don't check
for playback cancellation (like we do in decoders) or anything.
This is seriously broken and I can imagine it affects people on
fast CPUs more because we do asynchronous output buffering and
our ALSA device will always have data ready.
This is safer than the patch in
http://www.musicpd.org/mantis/view.php?id=1542
with multiple audio outputs enabled.
Sadly, I only noticed that patch/problem when I googled for
"snd_config_update_free_global"
Apparently snd_pcm_hw_params_can_resume() can return false even
though my hardware does in fact support resuming. So stop
carrying that value in the canResume flag and just try to resume
when we're in the suspended state; falling back to
snd_pcm_prepare only if resuming fails. libao does something
similar on resume, too.
While we're at it, use the E() macro which will enable us to
have better error reporting.
[mk: remove the E() macro stuff]
With a large music database, the linear string collection in
tagTracker.c becomes very slow. We implemented that in a
quick'n'dirty fashion when we removed tree.c, and now we rewrite it
using the fast hashed string set.
"struct strset" is a hashed string set: you can add strings to this
library, and it stores them as a set of unique strings. You can get
the size of the set, and you can enumerate through all values.
This will be used to replace the linear tagTracker library.
Instead of having to register each output plugin, store them
statically in an array. This eliminates the need for the List library
here, and saves some small allocations during startup.
Due to clumsy layout, the audio_format struct took 12 bytes. Move the
"channels" to the end, so it can be merged into the same 32 bit slot
as "bits", which reduces the struct size to 8 bytes.
print_playlist_result() had an assert(0) at the end, in case there was
an invalid result value. With NDEBUG, this resulted in a function not
returning a value - add a dummy "return -1" at the end to keep gcc
quiet.
Since all callers of song_id_exists() will map it to a song position
after the check, introduce a new function called song_id_to_position()
which performs both the check and the map lookup, including nice
assertions.
volatile provides absolutely no guarantee thread-safety in SMP
environments. volatile was designed to access memory locations
in peripheral hardware directly; not for SMP. If volatile is
needed to work properly on SMP, then it is only hiding subtle
bugs.
volatile only prevents the /compiler/ from making optimizations
when accessing variables. CPUs do their own optimizations at
runtime so it cannot guarantee registers of CPUs are flushed
to memory cache-coherent access on different CPUs.
Furthermore, the thread-communication via condition variables
between threads sharing audio formats already results in memory
barriers.
The tag pool is a shared global resource that is infrequently
modified. However, it can occasionally be modified by several
threads, especially by the metadata_pipe for streaming metadata
(both reading/writing).
The bulk tag_item pool is NOT locked as currently only the
update thread uses it.
Trying to read or remember
"tag->numOfItems * sizeof(*tag->items)"
requires too much thinking and mental effort on my part.
Also, favor "sizeof(struct mpd_tag)" over "sizeof(*tag->items)"
because the former is easier to read and follow, even though
the latter is easier to modify if the items member changes
to a different type.
The previous patch enabled these warnings. In Eric's branch, they
were worked around with a generic deconst_ptr() function. There are
several places where we can add "const" to pointers, and in others,
libraries want non-const strings. In the latter, convert string
literals to "static char[]" variables - this takes the same space, and
seems safer than deconsting a string literal.
All callers of fdprintf() have been converted to client_printf() or
fprintf(); it is time to remove this clumsy hack now. We can also
remove client_print() which took a file descriptor as parameter.
Now that we have removed all invocations of client_get_fd(), we can
safely remove this transitional function. All access to the file
descriptor is now hidden behind the interface declared in client.h.
The shared code in showPlaylist() isn't worth it, because we aim to
remove fdprintf(). Duplicate this small function, and enable stdio
buffering for saved playlists.
The function loadPlaylist() wants to report incremental errors to the
client, for this reason we cannot remove its protocol dependency right
now. Instead, make it use the client struct instead of the raw file
descriptor.
Don't pass the raw file descriptor around. This migration patch is
rather large, because all of the sources have inter dependencies - we
have to change all of them at the same time.
Pass the client struct to CommandHandlerFunction and
CommandListHandlerFunction. Most commands cannot take real advantage
of that yet, since most of them still work with the raw file
descriptor.
These two functions take a client struct instead of the file
descriptor. We will now begin passing the client struct around
instead of a raw file descriptor (which needed a linear lookup in the
client list to be useful).
This patch continues the work of the previous patch: don't pass a file
descriptor at all to traverseAllIn(). Since this fd was only used to
report "directory not found" errors, we can easily move that check to
the caller. This is a great relief, since it removes the dependency
on a client connection from a lot of enumeration functions.
Database traversal should be generic, and not bound to a client
connection. This is the first step: no file descriptor for the
callback functions forEachSong() and forEachDir(). If a callback
needs the file descriptor, it has to be passed in the void*data
pointer somehow; some callbacks might need a new struct for passing
more than one parameter. This might look a bit cumbersome right now,
but our goal is to have a clean API.
Continuing the effort of removing protocol specific calls from the
core libraries: let the command.c code call commandError() based on
PlaylistInfo's return value.
The playlist library shouldn't talk to the client if possible.
Introduce the "enum playlist_result" type which the caller
(i.e. command.c) may use to generate an error message.
Client's input values should be validated by the command
implementation, and the core libraries shouldn't talk to the client
directly if possible. Thus, setPlaylistRepeatStatus() and
setPlaylistRandomStatus() don't get the file descriptor, and cannot
fail (return void).
The function valid_playlist_name() checks the name, but it insists on
reporting an eventual error to the client. The new function
is_valid_playlist_name() is more generic: it just returns a boolean,
and does not care what the caller will use it for. The old function
valid_playlist_name() will be removed later.
Currently, when the tag cache is being serialized to hard disk, the
stdio buffer is flushed before every song, because tag_print.c
performs unbuffered writes on the raw file descriptor. Unfortunately,
the fdprintf() API allows buffered I/O only for a client connection by
looking up the client pointer owning the file descriptor - for stdio,
this is not possible. To re-enable proper stdio buffering, we have to
duplicate the tag_print.c code without fprintf() instead of our custom
fdprintf() hack. Add this duplicated code to tag_save.c.
Move everything which dumps song information (via tag_print.c) to a
separate source file. song_print.c gets code which writes song data
to the client; song_save.c is responsible for serializing songs from
the tag cache.
Based on client_puts(), client_printf() is the successor of
fdprintf(). As soon as all fdprintf() callers have been rewritten to
use client_printf(), we can remove fdprintf().
client_write() writes a buffer to the client and buffers it if
required. client_puts() does the same for a C string. The next patch
will add more tools which will replace fdprintf() later.
clearMpdTag could be called on a tag that was still in a
tag_begin_add transaction before tag_end_add is called. This
was causing free() to attempt to operate on bulk.items; which is
un-free()-able. Now instead we unmark the bulk.busy to avoid
committing the tags to the heap only to be immediately freed.
Additionally, we need to remember to call tag_end_add() when
a song is updated before we NULL song->tag to avoid tripping
an assertion the next time tag_begin_add() is called.
Since client->fd==-1 has become our "expired" flag, it may already be
-1 when client_close() is called. Don't assert that it is still
non-negative, and call client_set_expired() instead.
During the tag library refactoring, the shout plugin was disabled, and
I forgot about adapting it to the new API. Apply the same fixes to
the oggflac decoder plugin.
While parsing the tag cache, don't allocate the directory name from
the heap, but copy it into a buffer on the stack. This reduces heap
fragmentation by 1%.
If many tag_items are added at once while the tag cache is being
loaded, manage these items in a static fixed list, instead of
reallocating the list with every newly created item. This reduces
heap fragmentation.
Massif results again:
mk before: total 12,837,632; useful 10,626,383; extra 2,211,249
mk now: total 12,736,720; useful 10,626,383; extra 2,110,337
The "useful" value is the same since this patch only changes the way
we allocate the same amount of memory, but heap fragmentation was
reduced by 5%.
Try to detect if the string needs Latin1-UTF8 conversion, or
whitespace cleanup. If not, we don't need to allocate temporary
memory, leading to decreased heap fragmentation.
At several places, we create temporary copies of non-null-terminated
strings, just to use them in functions like validUtf8String(). We can
save this temporary allocation and avoid heap fragmentation if we
add a length parameter instead of expecting a null-terminated string.
Since the inline function cannot modify its caller's variables (which
is a good thing for code readability), the new string pointer is the
return value. The resulting binary should be the same as with the
macro.
The new source tag_pool.c manages a pool of reference counted tag_item
objects. This is used to merge tag items of the same type and value,
saving lots of memory. Formerly, only the value itself was pooled,
wasting memory for all the pointers and tag_item structs.
The following results were measured with massif. Started MPD on
amd64, typed "mpc", no song being played. My music database contains
35k tagged songs. The results are what massif reports as "peak".
0.13.2: total 14,131,392; useful 11,408,972; extra 2,722,420
eric: total 18,370,696; useful 15,648,182; extra 2,722,514
mk f34f694: total 15,833,952; useful 13,111,470; extra 2,722,482
mk now: total 12,837,632; useful 10,626,383; extra 2,211,249
This patch set saves 20% memory, and does a good job in reducing heap
fragmentation.
The value is stored in the same memory allocation as the tag_item
struct; this saves memory because we do not store the value pointer
anymore. Also remove the getTagItemString()/removeTagItemString()
dummies.
This patch makes MPD consume much more memory because string pooling
is disabled, but it prepares the next bunch of patches. Replace the
code in tagTracker.c with naive algorithms without the tree code. For
now, this should do; later we should find better algorithms,
especially for getNumberOfTagItems(), which has become wasteful with
temporary memory.
Unfortunately, the C standard postulates that the argument to free()
must be non-const. This does not makes sense, and virtually prevents
every pointer which must be freed at some time to be non-const. Use
the deconst hack (sorry for that) to allow us to free constant
pointers.
Instead of passing the pointer to the "expired" flag to
processListOfCommands(), this function should use the client API to
check this flag. We can now remove the "global_expired" hack
introduced recently.
Start exporting the client struct as an opaque struct. For now, pass
it only to processCommand() and processListOfCommands(), and provide a
function to extract the socket handle. Later, we will propagate the
pointer to all command implementations, and of course to
client_print() etc.
The old code tried to write a response to the client, without even
checking if it was already closed. Now that we have added more
assertions, these may fail... perform the "expired" check earlier.
Patch bdeb8e14 ("client: moved "expired" accesses into inline
function") was created under the wrong assumption that
processListOfCommands() could modify the expired flag, which is not
the case. Although "expired" is a non-const pointer,
processListOfCommands() just reads it, using it as the break condition
in a "while" loop. I will address this issue with a better overall
solution, but for now provide a pointer to a global "expired" flag.
client_defer_output() was modified so that it can create the
deferred_send list. With this patch, the assertion on
"deferred_send!=NULL" has become invalid. Remove it.
Previously, when select() failed, we assumed that there was an invalid
file descriptor in one of the client structs. Thus we tried select()
one by one. This is bogus, because we should never have invalid file
descriptors. Remove it, and make select() errors fatal.
All of the client's resources are freed in client_close(). It is
enough to set the "expired" flag, no need to duplicate lots of
destruction code again and again.
Due to the large buffers in the client struct, the static client array
eats several megabytes of RAM with a maximum of only 10 clients. Stop
this waste and allocate each client struct from the heap.
Second patch: rename the internal struct name. We will eventually
export this type as an opaque forward-declared struct later, so we
can pass a struct pointer instead of a file descriptor, which would
save us an expensive linear lookup.
I don't believe "interface" is a good name for something like
"connection by a client to MPD", let's call it "client". This is the
first patch in the series which changes the name, beginning with the
file name.
linux/list.h is a nice doubly linked list library - it is lightweight
and powerful at the same time. It will be useful later, when we begin
to allocate client structures dynamically. Import it, and strip out
all the stuff which we are not going to use.
It should be obvious in which thread or context a function is being
executed at runtime. The code which was left in decode.c is for the
decoder thread itself; give the file a better name.
This releases several include file dependencies. As a side effect,
"CHUNK_SIZE" isn't defined by decoder_api.h anymore, so we have to
define it directly in the plugins which need it. It just isn't worth
it to add it to the decoder plugin API.
The decoder plugins need this type, so export it in the public API.
This allows is to remove "decode.h" from "decoder_api.h", uncluttering
the API namespace some more.
Unfortunately, we have to pass the DecoderControl pointer to these
inline functions, because the global variable "dc" may not be
available here. This will be fixed later.
The decoder thread is responsible for resetting dc->command after a
command was executed. As a consequence, we can assume that
dc->command is already NONE after decoder_stop().
There is no unlocked caller of clearPlayerQueue(), and the functions
lockPlaylistInteraction() and unlockPlaylistInteraction() are trivial
- merge them.
Since playerPlay() already calls playerStop(), we can remove its
invocation of playerStop() from playPlaylistOrderNumber().
We can also make playerStop a static function.
All (indirect) callers of queueSong() ensure that the queue state is
BLANK, so there is no need to check it in queueSong() again. As a
side effect, queueSong() cannot fail anymore, and can return void.
Also, playlist_queueError and all its error handling can go away.
playerKill() was marked as deprecated, but it seems like a good idea
to do proper cleanup in all threads (e.g. for usable valgrind
results). Introduce the command "EXIT" which makes the player thread
exit cleanly.
playerWait() stops the player thread (twice!) and closes the output
device. It should be well enough to just send CLOSE_AUDIO, without
STOP.
This requires a tiny change to the player thread code: make it break
when CLOSE_AUDIO is sent.
It was possible for the decoder thread to go into an endless loop
(flac and oggflac decoders): when a "STOP" command arrived, the Read()
callback would return 0, but the EOF() callback returned false. Fix:
when decoder_get_command()!=NONE, return EOF==true.
Storing local configuration in global (static) variables is obviously
a bad idea. Move all those variables into the JackData struct,
including the locks.
There is only one caller of freeJackData() left: jack_finishDriver().
This function is called by the mpd core, and is called exactly once
for every successful jack_initDriver(). We do not need to clear
audioOutput->data, since this variable is invalidated anyway.
Over the lifetime of the jack AudioOutput object, we want a single
valid JackData object, so we can persistently store data there
(configuration etc.). Allocate JackData in jack_initDriver(). After
that, we can safely remove all audioOutput->data==NULL checks (and
replace them with assertions).
No need to destroy the JackData object when an error occurs, since
jack_finishDriver() already frees it. Only deinitialize the jack
library, introduce freeJackClient() for that, and move code from
freeJackData().
Prepare the next patch: make the "!jd" check independent of the
jd->client initialization. This way we can change the "jd"
initialization semantics later.
connect_jack() invokes freeJackData() in every error handler, although
its caller also invokes this function after a failure. We can save a
lot of lines in connect_jack() by removing these redundant
freeJackData() invocations.
When we introduced decoder_read(), we added code which aborts the read
operation when a decoder command arrives. Several plugins however did
not expect that when they were converted to decoder_read(). Add
proper checks to the mp3 and flac decoder plugins.
The code said "decoder_command==STOP" because that was a conversion
from the old "dc->stop" test. As we can now check for all commands in
one test, we can simply rewrite that to decoder_command!=NONE.
This flag is used internally; it is set by decoder_seek_where(), and
indicates that the decoder plugin has begun the seek process. It is
used for the case that the decoder plugin has to read data during the
seek process. Before this patch, that was impossible, because
decoder_read() would refuse to read data unless dc->command is NONE.
This patch is kind of a dirty workaround, and needs to be redesigned
later.
The old code called can_seek() with the uninitialized pointer
"isp.is". Has this ever worked? Anyway, initialize "isp" first, then
call can_seek(&isp).
Move everything related to finding and initializing the WVC stream to
wavpack_open_wvc(). This greatly simplifies its error handling and
the function wavpack_streamdecode().
On our way to stabilize the decoder API, we will one day remove the
input stream functions. The most basic function, read() will be
provided by decoder_api.h with this patch. It already contains a loop
(still with manual polling), error/eof handling and decoder command
checks. This kind of code used to be duplicated in all decoder
plugins.
If the input stream is not seekable, the try_decode() function
consumes valuable data, which is not available to the decode()
function anymore. This means that the decode() function does not
parse the header correctly. Better skip the detection if we cannot
seek. Or implement better buffering, something like unread() or
buffered rewind().
The return value of audio_linear_dither() is always casted to
mpd_sint16. Returning long does not make sense, and consumed 8 bytes
on a 64 bit platform.
The output buffer is always flushed after being appended to, which
allows us to assume it is always empty. Always start writing at
outputBuffer, don't remember outputPtr.
Fill the whole output buffer at a time by using dither_buffer()'s
ability to decode blocks. Calculate how many samples fit into the
output buffer before each invocation.
Simplifying loops for performance: why check dropSamplesAtEnd in every
iteration, when we could modify the loop boundary? The (writable)
variable samplesLeft can be eliminated; add a write-once variable
pcm_length instead, which is used for the loop condition.
The variable samplesPerFrame is used only in one single closure. Make
it local to this closure. The compiler will probably convert it to a
register anyway.
Preparing for simplifying and thus speeding up the dithering code:
moved dithering to a separate function which contains a trivial loop.
With this patch, only one sample is dithered at a time, but the
following patches will allow us to dither a whole block at a time,
without complicated buffer length checks.
Copy some code from aac_decode() to aac_stream_decode() and apply
necessary changes to allow streaming audio data. Both functions might
be merged later.
initAacBuffer() should really only initialize the buffer; currently,
it also reads data from the input stream and parses the header. All
of the AAC buffer code should probably be moved to a separate library
anyway.
Shifting from the buffer queue is a common operation, and should be
provided as a separate function. Move code to aac_buffer_shift() and
add a bunch of assertions.
When checking for EOF, we should not check whether the read request
has been fully satisified. The InputStream API does not guarantee
that readFromInputStream() always fills the whole buffer, if EOF is
not reached. Since there is the function inputStreamAtEOF() dedicated
for this purpose, we should use it for EOF checking after
readFromInputStream()==0.
Fill the AacBuffer even when nothing has been consumed yet. The
function should not check for consumed data, but for free space at the
end of the buffer.
The flag "ready" indicates whether the input stream is ready and it
has parsed all meta data. Previously, it was impossible for
decodeStart() to see the content type of HTTP input streams, because
at that time, the HTTP response wasn't parsed yet.
With the functions decoder_plugin_register() and
decoder_plugin_unregister(), decoder plugins can register a
"secondary" plugin, like the flac input plugin does this for
"oggflac".
"decoder plugin" is a better name than "input plugin", since the
plugin does not actually do the input - InputStream does. Also don't
use typedef, so we can forward-declare it if required.
PlayerControl.command replaces the old attributes play, stop, pause,
closeAudio, lockQueue, unlockQueue, seek. The main thread waits for
each command synchronously, so there can only be one command enabled
at a time anyway.
The wavpack decoder plugin implements a hack, and it needs the song
URL for that. This API (and the hack) should be revised later, but
add that function for now.
Since we want to hide mpd internals from the decoder plugins, the
plugins should not check dc->state whether they have already called
decoder_initialized(). Use a local variable to track that.
Some decoder commands are implemented in the decoder plugins, thus
they need to have an API call to signal that their current command has
been finished. Let them use the new decoder_command_finished()
instead of the internal dc_command_finished().
Another big patch which hides internal mpd APIs from decoder plugins:
decoder plugins regularly poll dc->command; expose it with a
decoder_api.h function.
Since we moved all PCM conversions to decoder_data(), the attribute
convState isn't being used anymore by the OutputBuffer code. Move it
to struct decoder.
InputPlugin is the API which is implemented by a decoder plugin. This
belongs to the public API/ABI, so move it to decoder_api.h. It will
later be renamed to something like "decoder_plugin".
Since we have merged dc->stop, dc->seek into one variable, we don't
have to check both conditions at a time; we can replace "!stop &&
!seek" with "none".
dc->audioFormat is set once by the decoder plugins before invoking
decoder_initialized(); hide dc->audioFormat and let the decoder pass
an AudioFormat pointer to decoder_initialized().
We are now beginning to remove direct structure accesses from the
decoder plugins. decoder_clear() and decoder_flush() mask two very
common buffer functions.
Code simplification: since we are not using in-band signalling with
the chunk index anymore, we can just return a pointer to the tail
chunk instead of the index.
OutputBuffer should be a more generic low-level library, without
dependencies to the other headers. This patch adds the field
"notify", which is used to signal the player thread. It is passed in
the constructor, and removes the need to compile with the decode.h
header.
After the decoder has been initialized and the audio device has been
opened, don't sleep. The decoder plugin won't do anything special nor
will it care to wake us up for some reason.
decoder_initialized() sets the state to DECODE_STATE_DECODE and wakes
up the player thread. It is called by the decoder plugin after its
internal initialization is finished. More arguments will be added
later to prevent direct accesses to the DecoderControl struct.
The decoder struct should later be made opaque to the decoder plugin,
because maintaining a stable struct ABI is quite difficult. The ABI
should only consist of a small number of stable functions.
Don't use wrappers like player_wakeup_decoder_nb(). These have been
wrappers calling notify.c functions, for compatibility with the
existing code when we migrated to notify.c.
dc_command_finished() is invoked by the decoder thread when it has
finished a command (sent by the player thread). It resets dc.command
and wakes up the player thread. This combination was used at a lot of
places, and by introducing this function, the code will be more
readable.
Busy wait loops are a bad thing, especially when the response time can
be very long - busy waits eat a lot of CPU, and thus slow down the
other thread. Since the other thread will notify us when it's ready,
we can use notify_wait() instead.
Much of the existing code queries all three variables sequentially.
Since only one of them can be set at a time, this can be optimized and
unified by merging all of them into one enum variable. Later, the
"command" checks can be expressed in a "switch" statement.
Since pc->current_song denotes the song which the decoder should use
next, we should move it to DecoderControl. This removes one internal
PlayerControl struct access from the decoder code.
Also add pc.next_song, which is manipulated by the playlist code, and
gets copied to dc.next_song as soon as the decoder is started.
Also enable -Wunused-parameter - this forces us to add the gcc
"unused" attribute to a lot of parameters (mostly library callback
functions), but it's worth it during code refactorizations.
If nothing has been read from the input stream, we don't have to
rewind it.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7397 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The variable "to_read" is never modified except in the last iteration
of the while loop. This means the while condition will never become
false, as the body will break before that may be checked.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7396 09075e82-0dd4-0310-85a5-a0d7c8717e4f
This institutes the usage of a separate thread to buffer HTTP
input. It is basically practice code for using the ringbuffer
code which I plan on reusing for the OutputBuffer as well as
further input buffering for disk (networked filesystems over
WAN, laptops on battery, etc).
Each readFromInputStream() call on an HTTP stream can take
several seconds to complete, short reads are avoided.
A single-threaded solution for systems supporting large enough
SO_RCVBUF values should also be possible and will likely be done
in the future; but this lock-free(except when full/empty)
ringbuffer is cool :)
git-svn-id: https://svn.musicpd.org/mpd/trunk@7393 09075e82-0dd4-0310-85a5-a0d7c8717e4f
We'll be using pipes when waiting for I/O, and condition
variables at other times.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7391 09075e82-0dd4-0310-85a5-a0d7c8717e4f
This will allow both the reader and writer threads to
reset the ringbuffer in a thread-safe fashion.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7390 09075e82-0dd4-0310-85a5-a0d7c8717e4f
This will eliminate unnecessary calls to ringbuf_{read,write}_space
git-svn-id: https://svn.musicpd.org/mpd/trunk@7389 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The auth code also has some ugly usages of string generation
which I will eventually replace with something nicer...
git-svn-id: https://svn.musicpd.org/mpd/trunk@7387 09075e82-0dd4-0310-85a5-a0d7c8717e4f
This piece of code is from the JACK Audio Connection Kit
(trimmed down a bit for better readability).
The vector functions now reuse the common iovec struct used by
writev/readv instead of reinventing an identical but
differently-named struct.
From the comments:
> ISO/POSIX C version of Paul Davis's lock free ringbuffer C++ code.
> This is safe for the case of one read thread and one write thread.
License is LGPL 2.1 or later
git-svn-id: https://svn.musicpd.org/mpd/trunk@7386 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Initialize audioOutput->data with NULL in jack_initDriver().
Previously, this was never initialized, although the other functions
relied on it being NULL prior to jack_openDevice().
This patch addresses bug 0001641[1]. In contrast to the patch provided
by the bug reporter, it moves the initialization before the "!param"
check.
[1] - http://musicpd.org/mantis/view.php?id=1641
git-svn-id: https://svn.musicpd.org/mpd/trunk@7375 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Streaming was broken, beacuse the stream URL was never copied to
path_max_fs.
[ew: replaced strcpy with pathcpy_trunc for ease of auditing]
git-svn-id: https://svn.musicpd.org/mpd/trunk@7371 09075e82-0dd4-0310-85a5-a0d7c8717e4f
During the decoder thread main loop, dc.state must be
DECODE_STATE_STOP. Explicitly assigning it after the "dc.stop" check
is redundant.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7366 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The function wait_main_task() is racy: if the function
wakeup_via_cond() sees the mutex is locked just before
wait_main_task() executes pthread_cond_wait(), the main thread blocks
forever.
Work around this issue by adding a "pending" flag just like in my
notify.c code. A standards-compliant solution should be implemented
later.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7365 09075e82-0dd4-0310-85a5-a0d7c8717e4f
In lazy mode (previously the default), outputBuffer.c only wakes up
the player when it was previously empty. That caused a deadlock when
the player was waiting for buffered_before_play, since the decoder
wouldn't wake up the player when buffered_before_play was reached. In
non-lazy mode, always wake up the player when a new chunk was decoded.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7364 09075e82-0dd4-0310-85a5-a0d7c8717e4f
When we are in an input plugin, dc.current_song should already be
set. Use it instead of pc.current_song.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7363 09075e82-0dd4-0310-85a5-a0d7c8717e4f
We had functions names varied between
outputBufferFoo, fooOutputBuffer, and output_buffer_foo
That was too confusing for my little brain to handle.
And the global variable was somehow named 'cb' instead of
the more obvious 'ob'...
git-svn-id: https://svn.musicpd.org/mpd/trunk@7355 09075e82-0dd4-0310-85a5-a0d7c8717e4f
All of our main singleton data structures are implicitly shared,
so there's no reason to keep passing them around and around in
the stack and making our internal API harder to deal with.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7354 09075e82-0dd4-0310-85a5-a0d7c8717e4f
This at least makes the argument list to a lot of our plugin
functions shorter and removes a good amount of line nois^W^Wcode,
hopefully making things easier to read and follow.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7353 09075e82-0dd4-0310-85a5-a0d7c8717e4f
It actually increases our image size a small bit and may even
hurt performance a very small bit, but makes the code less
verbose and easier to manage.
I don't see a reason for mpd to ever support playing multiple
files at the same time (users can run multiple instances of mpd
if they really want to play Zaireeka, but that's such an edge
case it's not worth ever supporting in our code).
git-svn-id: https://svn.musicpd.org/mpd/trunk@7352 09075e82-0dd4-0310-85a5-a0d7c8717e4f
signal is all we need since we only have one waiter and
likely faster
git-svn-id: https://svn.musicpd.org/mpd/trunk@7349 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The select() in the main event loop blocks now (saving us many
unnecessary wakeups). This interacted badly with the threads
that were trying to wakeup the main task via
pthread_cond_signal() since the main task was not blocked
on a condition variable, but on select().
So now if we detect a need to wakeup the player, we write
to a pipe which select() is watching instead of blindly
calling pthread_cond_signal().
git-svn-id: https://svn.musicpd.org/mpd/trunk@7347 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Since tailChunk() automatically flushes full buffers, we do not have
to check this in every iteration of sendDataToOutputBuffer().
git-svn-id: https://svn.musicpd.org/mpd/trunk@7343 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Checking dc->stop in the while condition and again after the while
loop costs some CPU cycles we should save.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7340 09075e82-0dd4-0310-85a5-a0d7c8717e4f
OutputBuffer.currentChunk contains redundant data: it is either -1
when there is no chunk which is currently being written, or it equals
"end". If we always keep chunk[end] in a valid state, we can remove
OutputBuffer.currentChunk.
This patch may look a bit clumsy, especially flushOutputBuffer(), but
that will be fixed later with an major OutputBuffer API overhaul.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7339 09075e82-0dd4-0310-85a5-a0d7c8717e4f
output_buffer_expand() moves the cb->end to the new position (only its
current successor is allowed) and wakes up the player if is waiting
for the decoder. This simplifies flushOutputBuffer().
git-svn-id: https://svn.musicpd.org/mpd/trunk@7338 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The current OutputBuffer object is allocated statically, i.e. it is
zeroed. To be safe for other cases in the future, also initialize the
other elements.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7337 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The decoder should not wake up the player when it did not produce a
flushed chunk. Move the decoder_wakeup_player() call to
flushOutputBuffer() and invoke it only if the buffer was previously
empty.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7336 09075e82-0dd4-0310-85a5-a0d7c8717e4f
During my tests, it happened that data->position>newPosition. I have
not yet fully understood why this can happen; for now, replace this
with a run-time check.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7334 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The patch "convert blocks until the buffer is full" did not update
data->chunk_length correctly: it added the number of samples, not the
number of bytes. Multiply that with bytes_per_channel
git-svn-id: https://svn.musicpd.org/mpd/trunk@7332 09075e82-0dd4-0310-85a5-a0d7c8717e4f
In the patch "special optimized case for 16bit stereo", the check for
"num_channels==2" was missing.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7331 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Not having to loop for every sample byte (depending on a variable
unknown at compile time) saves a lot of CPU cycles. We could consider
reimplementing this function with liboil...
git-svn-id: https://svn.musicpd.org/mpd/trunk@7330 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Read frame->header.channels once, and pass only this integer to
flac_convert().
git-svn-id: https://svn.musicpd.org/mpd/trunk@7329 09075e82-0dd4-0310-85a5-a0d7c8717e4f
flacWrite() is the only function which sets data->chunk_length. If we
flush the buffer before we return, we can assume that it is always
empty upon entering flacWrite().
git-svn-id: https://svn.musicpd.org/mpd/trunk@7328 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Move the inner loop which converts samples to flac_convert(). There
it is isolated and easier to optimize. This function does not have to
worry about buffer boundaries; the caller (i.e. flacWrite())
calculates how much is left and is responsible for flushing. That
saves a lot of superfluous range checks within the loop.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7327 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Check for flushing the chunk buffer only once per sample, before
iterating over channels and bytes. This saves another 5% CPU cycles.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7326 09075e82-0dd4-0310-85a5-a0d7c8717e4f
AudioFormat.bits is volatile, and to read it, 3 pointers had to be
deferenced. Calculate this value once. This speeds up this function
by 5%.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7325 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Try to only include headers which are really needed. We should
particularly check all "headers including other headers". The
long-term goal is to have a manageable, small API for plugins
(decoders, output) without so many mpd internals cluttering the
namespace.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7319 09075e82-0dd4-0310-85a5-a0d7c8717e4f
There is no danger that gcc will optimize access to OutputBufferChunk
properties, since decoder and player work in different chunk objects.
It is safe to remove "volatile" here.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7318 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Less global variables: at any invocation of decoder_sleep(), we have a
reference to the DecoderControl anyway, so we should pass it. This
costs less than having to call getPlayerData() in every tiny
function. Maybe some day we will be able to have multiple decoders at
the same time...
git-svn-id: https://svn.musicpd.org/mpd/trunk@7316 09075e82-0dd4-0310-85a5-a0d7c8717e4f
To do proper cleanup before exiting, we have to provide a destructor
for OutputBuffer. One day, valgrind will not complain about memory
leaks!
git-svn-id: https://svn.musicpd.org/mpd/trunk@7315 09075e82-0dd4-0310-85a5-a0d7c8717e4f
"end" is not being used anymore, since we moved most OutputBuffer
struct accesses into methods.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7314 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Don't be mean with integer sizes. Although we will probably never
have more than 32k buffered chunks, we should use 32 bit integers for
addressing them. We do not save very much (some of the saved space is
eaten by alignment anyway), but we save at least one assembler
instruction for converting short to int.
This change requires some more explicit casts, because gcc was less
picky when comparing short with a full int.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7313 09075e82-0dd4-0310-85a5-a0d7c8717e4f
First patch without camelCase ;)
output_buffer_skip() lets us eliminate advanceOutputBufferTo(), and
removes yet another external OutputBuffer struct access.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7312 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Second patch to make OutputBuffer self-contained: since OutputBuffer
now knows its own size, we do not need the global variable
"buffered_chunks" anymore.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7311 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Try to make OutputBuffer self-contained, without depending on a global
variable.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7310 09075e82-0dd4-0310-85a5-a0d7c8717e4f
This is the first patch in a series which removes the shared memory,
and moves all the playerData objects into the normal libc heap.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7304 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Now that we do proper locking and signalling instead of continuous
polling for IPC, a deadlock was found: at the end of a song, the
player thread waits until the main thread sets pc->queueState from
PLAYER_QUEUE_DECODE to PLAYER_QUEUE_PLAY. He is never woken up, since
syncPlaylistWithQueue() does not activate the notification. I added
wakeup_player_nb() to setQueueState(), since the player must be
signalled whenever pc->queueState changes.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7303 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Buffer sizes should be size_t. This is safe here, at least not
unsafer than without the patch. I have no idea why audioBufferSize
and audioBufferPos were explicitly declared as signed integer.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7296 09075e82-0dd4-0310-85a5-a0d7c8717e4f
When growing the audioOutput->convBuffer, we can use free()+malloc()
instead of realloc(), which saves a memcpy().
git-svn-id: https://svn.musicpd.org/mpd/trunk@7295 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The audio output plugins should get a constant pointer, because they
must not modify the buffer. Since the size is a non-negative buffer
size in bytes, we should change its type to size_t.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7293 09075e82-0dd4-0310-85a5-a0d7c8717e4f
In my previous patch set, I forgot to change the
pcm_sizeOfConvBuffer() invocation in convertAudioFormat() to also use
size_t.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7292 09075e82-0dd4-0310-85a5-a0d7c8717e4f
calculateCrossFadeChunks() still returns int, although the caller uses
it as an unsigned value. Since the function body checks for negative
values, it is safe to cast to unsigned.
crossFade() takes signed parameters, although it callers pass unsigned
integers. Change declaration to unsigned.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7291 09075e82-0dd4-0310-85a5-a0d7c8717e4f
There were some const pointers missing in the previous const-cleanup
patch.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7290 09075e82-0dd4-0310-85a5-a0d7c8717e4f
libfaad wants uint32_t pointers. Passing a long pointer is bugged on
amd64.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7289 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The patch "Start using song pointers in core data structures" removed
dc->utf8url, and the adaption for wavpack_plugin.c was missing.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7288 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The new function successor() can be used to simplify a lot of code
lines and saves a lot of "i+>=buffered_chunks" checks.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7285 09075e82-0dd4-0310-85a5-a0d7c8717e4f
currentChunk is a global variable, which renders the whole output
buffer code non-reentrant. Although this is not a real problem since
there is only one global output buffer currently, we should move it to
the OutputBuffer struct.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7284 09075e82-0dd4-0310-85a5-a0d7c8717e4f
This patch rewrites notify.c to use the pthread API, namely
pthread_mutex and pthread_cond. This is a lot cheaper and easier than
the pipe() hack.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7280 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The code paths which return from the functions all have to call
quitDecode(). If we simply break instead of calling quitDecode()
explicitly, this function gets called in the last line of this
function anyway.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7278 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The multi-line expression which calculates sizeToTime is hard to read,
partly because "cb->audioFormat." is too long. Create a separate
inline function in audio.h for that.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7277 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Moved the decision whether to cross-fade the current song to
calculateCrossFadeChunks(). This simplifies the function
decoderParent() and eliminates one layer of indentation.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7276 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Similar to the crossFade() patch: pass chunk objects to playChunk(),
simplify decodeParent() by removing clutter.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7275 09075e82-0dd4-0310-85a5-a0d7c8717e4f
To unify the decoderParent() main loop some more, use it to wait for
the decoder to change the song. Only one single processDecodeInput()
caller left after this patch.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7273 09075e82-0dd4-0310-85a5-a0d7c8717e4f
When there are not enough decode cross-fade chunks in the buffer yet,
the current code does busy-wait, which will delay the decoder even
more. sleep instead, expecting the decoder to wake us up.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7272 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Calling crossFade() with the chunk objects is easier than unrolling
all the chunk properties manually, and decodeParent() can really use
more of these simplifications.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7271 09075e82-0dd4-0310-85a5-a0d7c8717e4f
sizeof() is the more "natural" or "direct" access to the buffer size,
instead of a macro happening to be used to the buffer declaration.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7270 09075e82-0dd4-0310-85a5-a0d7c8717e4f
To make access to OutputBuffer easier, move everything which belongs
to a chunk into its own structure, namely OutputBufferChunk.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7269 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The chunk size should be in outputBuffer.h since the output buffer
code is its primary user.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7268 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Hiding OutputBuffer internals, again. We get an extra assertion in
return.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7267 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The variable "quit" can be removed, since its only setter can use
"break" instead, just like the other code paths.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7266 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The variables "nextChunk" and "crossFadeChunks" are only used when
doCrossFade==1. This means that we do not have to reset these as long
as doCrossFade!=1.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7265 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The cross-fade check is still very complicated whenever it uses
OutputBuffer internals. Greatly simplify another check by introducing
outputBufferRelative().
git-svn-id: https://svn.musicpd.org/mpd/trunk@7264 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Another "don't use OutputBuffer internals" patch. This ignores the
copied "end" value, but I do not think that has ever been a real
issue.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7263 09075e82-0dd4-0310-85a5-a0d7c8717e4f
decoderParent() uses a lot of OutputBuffer internals to see whether
cross-fading should be started. Move these checks to outputBuffer.c,
which also simplifies decoderParent().
git-svn-id: https://svn.musicpd.org/mpd/trunk@7262 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Declare the variables "test" and "fadePosition" in the scope where
they are really used. This removes some of the clutter in the
function decodeParent().
git-svn-id: https://svn.musicpd.org/mpd/trunk@7261 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Eliminating some duplicated and. This also decreases the number of
lines calling processDecodeInput().
git-svn-id: https://svn.musicpd.org/mpd/trunk@7260 09075e82-0dd4-0310-85a5-a0d7c8717e4f
I have spent some time to understand decodeParent(), which does a lot
of obfuscated magic... I find it useful to help others to also
understand it, so I wrote a few comments.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7259 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Another patch indended to improve the CPP macro hell. This enlarges
the function decodeParent(), but it cannot be converted into a
standalone function easily, because it references so many local
variables.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7258 09075e82-0dd4-0310-85a5-a0d7c8717e4f
clearOutputBuffer() also resets currentChunk; this might resolve a
theoretical bug.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7257 09075e82-0dd4-0310-85a5-a0d7c8717e4f
dc->start cannot be true after the loop, because it was the loop
condition. dc->start could have been set by another thread between
the while loop and the if, but I suspect this is not the case the
author intended, so we just remove the dc->start check.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7256 09075e82-0dd4-0310-85a5-a0d7c8717e4f
In my opinion, the code becomes more readable when we explicitly check
"==NOERROR" instead of an implicit 0 check.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7255 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Macros are ugly, and multi-line macros are even more ugly. This patch
converts processDecodeInput() to a C function. The disadvantage may
be that the function does not have access to the caller's local
variables, which might be regarded as an advantage on the other hand.
For this reason, we have to pass variable references. This costs a
tiny bit of performance, but it's worth eliminating this monster
macro, and further patches will optimize this cost down.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7254 09075e82-0dd4-0310-85a5-a0d7c8717e4f
since clearPlayerQueue() is always called within
lockPlaylistInteraction() / unlockPlaylistInteraction(), it simplifies
the code to add another function which calls these three functions.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7253 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The block after "if" breaks out of the loop. To make the code a
little bit more readable, don't write the rest in an "else" block,
since this code path does not break.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7251 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The method availableOutputBuffer() calculates how many chunks are in
use. This simplifies code which needs this information, and it can
run without knowing OutputBuffer internals. The function knows how to
calculate this when begin>end; this might have been a bug in
decodeParent(), which does not.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7250 09075e82-0dd4-0310-85a5-a0d7c8717e4f
After the previous patch, it is clear that the loop in
advanceOutputBufferTo() can be replaced with a simple assignment.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7249 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The parameter "currentChunkSent" is not used and can be dropped.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7248 09075e82-0dd4-0310-85a5-a0d7c8717e4f
This patch removes some clutter from decodeParent() by moving some
code out.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7247 09075e82-0dd4-0310-85a5-a0d7c8717e4f
realloc() has to copy data to the new buffer. Since convBuffer
contains temporary data only, we can safely use free() plus a new
malloc(), which saves the mempy().
git-svn-id: https://svn.musicpd.org/mpd/trunk@7246 09075e82-0dd4-0310-85a5-a0d7c8717e4f
It is way more complicated than it should be; and
locking it for thread-safety is too difficult.
[merged r7183 from branches/ew]
git-svn-id: https://svn.musicpd.org/mpd/trunk@7241 09075e82-0dd4-0310-85a5-a0d7c8717e4f
I initially started to do a heavy rewrite that changed the way processes
communicated, but that was too much to do at once. So this change only
focuses on replacing the player and decode processes with threads and
using condition variables instead of polling in loops; so the changeset
itself is quiet small.
* The shared output buffer variables will still need locking
to guard against race conditions. So in this effect, we're probably
just as buggy as before. The reduced context-switching overhead of
using threads instead of processes may even make bugs show up more or
less often...
* Basic functionality appears to be working for playing local (and NFS)
audio, including:
play, pause, stop, seek, previous, next, and main playlist editing
* I haven't tested HTTP streams yet, they should work.
* I've only tested ALSA and Icecast. ALSA works fine, Icecast
metadata seems to get screwy at times and breaks song
advancement in the playlist at times.
* state file loading works, too (after some last-minute hacks with
non-blocking wakeup functions)
* The non-blocking (*_nb) variants of the task management functions are
probably overused. They're more lenient and easier to use because
much of our code is still based on our previous polling-based system.
* It currently segfaults on exit. I haven't paid much attention
to the exit/signal-handling routines other than ensuring it
compiles. At least the state file seems to work. We don't
do any cleanups of the threads on exit, yet.
* Update is still done in a child process and not in a thread.
To do this in a thread, we'll need to ensure it does proper
locking and communication with the main thread; but should
require less memory in the end because we'll be updating
the database "in-place" rather than updating a copy and
then bulk-loading when done.
* We're more sensitive to bugs in 3rd party libraries now.
My plan is to eventually use a master process which forks()
and restarts the child when it dies:
locking and communication with the main thread; but should
require less memory in the end because we'll be updating
the database "in-place" rather than updating a copy and
then bulk-loading when done.
* We're more sensitive to bugs in 3rd party libraries now.
My plan is to eventually use a master process which forks()
and restarts the child when it dies:
master - just does waitpid() + fork() in a loop
\- main thread
\- decoder thread
\- player thread
At the beginning of every song, the main thread will set
a dirty flag and update the state file. This way, if we
encounter a song that triggers a segfault killing the
main thread, the master will start the replacement main
on the next song.
* The main thread still wakes up every second on select()
to check for signals; which affects power management.
[merged r7138 from branches/ew]
git-svn-id: https://svn.musicpd.org/mpd/trunk@7240 09075e82-0dd4-0310-85a5-a0d7c8717e4f
autoconf flags for enabling and disabling TCP and unix domain socket
support. Embedded machines without a TCP stack may be better off
without TCP support.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7236 09075e82-0dd4-0310-85a5-a0d7c8717e4f
This trivial patch addresses bug 1639. When a bind_to_address
argument starts with a slash, assume that it is the address of a Unix
domain socket.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7235 09075e82-0dd4-0310-85a5-a0d7c8717e4f
It is a good practice to constify pointers when their dereferenced
data is not modified within the functions or its descendants.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7234 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The parameter "port" is not actually used by establishListen(), and
can be removed. This also allows establishListen() to be used for
socket addresses which have no port.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7233 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The number of buffered chunks can obviously not become negative. The
"buffered_before_play<0" therefore cannot be useful, so let's remove
it, too.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7232 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Do explicit casts before comparing signed with unsigned. The one in
log.c actually fixes another warning: in the expanded macro, there may
be a check "logLevel>=0", which is always true.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7230 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Unfortunately, the function iconv() wants a non-const input buffer.
In this context, we only have a const pointer, which emits a correct
gcc warning. Work around this ugliness with an union-deconst hack.
This is optimized away in the binary.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7229 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The counter variables c_samp and c_chan begin at zero and can never be
negative.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7228 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The local variable d_samp is initialized, but never actually used.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7227 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Use unsigned integers in decoderParent() for chunk numbers which
cannot be negative.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7226 09075e82-0dd4-0310-85a5-a0d7c8717e4f
We don't really care what that variable is, so it might as
well be uninitialized, but valgrind does...
git-svn-id: https://svn.musicpd.org/mpd/trunk@7220 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Don't bother initializing the junk buffer, we really don't care.
The array was also unnecessary and ugly.
Also, avoid returning the byte count we read/wrote since it
unnecessarily exposes internal details of the implementation to
the callers.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7219 09075e82-0dd4-0310-85a5-a0d7c8717e4f
* move set_nonblock{,ing}() into utils.c since we use it
elsewhere, too
* add proper error checking to set_nonblocking()
* use os_compat.h instead of individually #includ-ing system headers
git-svn-id: https://svn.musicpd.org/mpd/trunk@7217 09075e82-0dd4-0310-85a5-a0d7c8717e4f
When the decoder receives SIGCONT during waitNotify(), the kernel
restarts the read() system call. This lets the decoder process block
indefinitely, while the player process waits for it to react. This
should probably be solved with a proper signal handler which aborts
the read() system call, but for now, we just write to the pipe to make
it wake up.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7216 09075e82-0dd4-0310-85a5-a0d7c8717e4f
When the decoder process is faster than the player process, all
decodedd buffers are full at some point in time. The decoder has to
wait for buffers to become free (finished playing). It used to do
this by polling the buffer status 100 times a second.
This generates a lot of unnecessary CPU wakeups. This patch adds a
way for the player process to notify the decoder process that it may
continue its work.
We could use pthread_cond for that, unfortunately inter-process
mutexes/conds are not supported by some kernels (Linux), so we cannot
use this light-weight method until mpd moves to using threads instead
of processes. The other method would be semaphores, which
historically are global resources with a unique name; this historic
API is cumbersome, and I wanted to avoid it.
I came up with a quite naive solution for now: I create an anonymous
pipe with pipe(), and the decoder process reads on that pipe. Until
the player process sends data on it as a signal, the decoder process
blocks.
This can be optimized in a number of ways:
- if the decoder process is still working (instead of waiting for
buffers), we could save the write() system call, since there is
nobody waiting for the notification.
[ew: I tried this using a counter in shared memory, didn't help]
- the pipe buffer will be full at some point, when the decoder thread
is too slow. For this reason, the writer side of the pipe is
non-blocking, and mpd can ignore the resulting EWOULDBLOCK.
- since we have shared memory, we could check whether somebody is
actually waiting without a context switch, and we could just not
write the notification byte.
[ew: tried same method/result as first point above]
- if there is already a notification in the pipe, we could also not
write another one.
[ew: tried same method/result as first/third points above]
- the decoder will only consume 64 bytes at a time. If the pipe
buffer is full, this will result in a lot of read() invocations.
This does not hurt badly, but on a heavily loaded system, this might
add a little bit more load. The preceding optimizations however
are able eliminate the this.
- finally, we should use another method for inter process
notifications - maybe kill() or just make mpd use threads, finally.
In spite of all these possibilities to optimize this code further,
this pipe notification trick is faster than the 100 Hz poll. On my
machine, it reduced the number of wakeups to less than 30%.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7215 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Use unsigned variables for storing the count of items or for iteration
variables. Since there can never be a negative number of items, it
makes sense to use an unsigned data type here. This change is safe
because the unsigned values are only used for adddressing array items.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7214 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The interfaces main loop repeats the select() (non-blocking) after an
event was handled. I do not see any reason for that, since all events
should be handled after the first select(). This double select() does
nothing than consume more CPU cycles.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7213 09075e82-0dd4-0310-85a5-a0d7c8717e4f
mpd sets a 1s select() timeout for no reason. This makes mpd wake up
the CPU, consume some cycles just to see there is nothing to do. We
can save that by specifying NULL instead of a timeout.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7212 09075e82-0dd4-0310-85a5-a0d7c8717e4f
* malloc() => xmalloc() for error checking
* strncpy() replaced with memcpy(),
memcpy appears perfectly safe here and mpd
does not ever use strncpy() (see r4491)
git-svn-id: https://svn.musicpd.org/mpd/trunk@7211 09075e82-0dd4-0310-85a5-a0d7c8717e4f
This patch does the following:
-enables WVC support for streams as well,
-improves MPD inputStream <=> WavPack stream connector,
-fixes two compile warnings (which were caused by MPD API change).
Mantis #1660 <http://musicpd.org/mantis/view.php?id=1660>
git-svn-id: https://svn.musicpd.org/mpd/trunk@7210 09075e82-0dd4-0310-85a5-a0d7c8717e4f
Basically, I don't trust myself nor Max to not have bugs in our
code when switching over to unsigned types, so I've added more
assertions which will hopefully trip and force us to fix these
bugs before somebody can exploit them :)
Some cleanups for parameter parsing using strtol
and error reporting to the user. Also, fix some completely
garbled indentation in inputStream_http.c
git-svn-id: https://svn.musicpd.org/mpd/trunk@7209 09075e82-0dd4-0310-85a5-a0d7c8717e4f
This patch moves code which initializes the OutputBuffer struct to
outputBuffer.c. Although this is generally a good idea, it prepares
the following patch.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7206 09075e82-0dd4-0310-85a5-a0d7c8717e4f
When dealing with in-memory lengths, the standard type "size_t" should
be used. Missing one can be quite dangerous, because an attacker
could provoke an integer under-/overflow, which may provide an attack
vector.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7205 09075e82-0dd4-0310-85a5-a0d7c8717e4f
we do not save anything by limiting a variable to an unsigned char,
since the compiler aligns it at machine word size anyway. however by
using the full machine word, we save one instruction, and we remove
the useless artificial limitation to 255.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7203 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The local variable eof can actually be replaced with a simple "break".
With a negative ret, the value of chunkpos can be invalidated, I am
not sure if this might have been a bug.
[ew: no, a negative ret will correspond to ret == OV_HOLE and ret
will be reset to zero leaving chunkpos untouched (code cleaned up
to make this more obvious]
git-svn-id: https://svn.musicpd.org/mpd/trunk@7202 09075e82-0dd4-0310-85a5-a0d7c8717e4f
The database parser does not check whether the song object has been
initialized yet, which may lead to a NULL pointer dereference. Add
this check.
git-svn-id: https://svn.musicpd.org/mpd/trunk@7201 09075e82-0dd4-0310-85a5-a0d7c8717e4f
