aboutsummaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* directory: children leave parents before being free()edEric Wong2008-10-121-1/+4
| | | | | This prevents double-free()s when doing teardown of a directory.
* directory: always maintain sorted properties vectorsEric Wong2008-10-127-41/+12
| | | | | | | | | This allows clients to see sorted results while we're updating the DB and removes the need for us to have to sort manually. We'll have to write separate routines for managing stored playlists with songvecs eventually; but that's for another day.
* update: simplify the serialized_delete usage a bitEric Wong2008-10-121-20/+17
| | | | | Pass a callback and argument to it instead of requiring explicit type.
* update: remove delete_each_song and clear_directoryEric Wong2008-10-121-21/+3
| | | | | Our beefier directory_free takes care of it now in the main task.
* directory: directory_free kills all that it containsEric Wong2008-10-121-0/+14
| | | | | Only call this in the main thread; it is NOT thread safe and not designed to be.
* update: serialize directory deletionsEric Wong2008-10-121-11/+28
| | | | | We only delete directory object in the main thread to prevent access to individual elements (mainly path).
* update: serialize song_free in main threadEric Wong2008-10-121-3/+1
| | | | | | It's actually possible to have a traverser accessing a song while we're freeing it in the main thread, as the songvec iterators don't lock the individual elements.
* dirvec: introduce locking for all iteratorsEric Wong2008-10-121-7/+29
| | | | | | | Like the songvec nr_lock, only one lock is used for all traversals since they're rarely changed. This only projects traversals, but not the individual structures themselves.
* dirvec: use dirvec_for_each where it makes senseEric Wong2008-10-123-52/+65
| | | | | This way we can introduce locking to allow safe traversals from the main thread while we're updating.
* dirvec: add dirvec_for_each iteratorEric Wong2008-10-122-0/+19
| | | | This will make it easier to introduce locking
* songvec: avoid holding nr_lock during free(3)Eric Wong2008-10-121-3/+5
| | | | | | | We only need to lock sv->nr changes to prevent traversals ( why it's called "nr_lock"). free(3) is a "slow" function on my system; so we can avoid unnecessarily holding a lock long for longer than needed.
* update: allow music_root updates to be queuedEric Wong2008-10-121-3/+2
| | | | | | Previously only updates with subdirectories being specified could be queued. No harm in queueing full updates.
* update: validate in command.c and fix small memory leakEric Wong2008-10-122-6/+17
| | | | | | | | If update_task was called with "" as its path argument, that string would never get freed because it false positived. Instead, never pass "" to update_task and trip an assertion if we do.
* directory: rename isRootDirectory => path_is_music_rootEric Wong2008-10-123-6/+6
| | | | down with camelCase!
* Avoid calling isRootDirectory when we have a directory objectEric Wong2008-10-124-6/+6
| | | | | There is only one music_root and we can just compare addresses.
* directory: make music_root global and avoid runtime initializationEric Wong2008-10-124-34/+21
| | | | | | | | | | mpd can't function without music_root; so don't bother allocating it on the heap nor checking to see if it's initialized. Don't allow directory_new() to create a directory w/o a parent or with an empty path, either: root is root and there can be only one</highlander>.
* directory: use mpd_sizeof_str_flex_array for path, tooEric Wong2008-10-122-8/+9
| | | | This way we avoid unnecessary heap allocations.
* tag_item: avoid wasting space when struct is unpackableEric Wong2008-10-122-2/+4
| | | | | Not all compilers support struct packing, and those that don't shouldn't be punished for it.
* song: use mpd_sizeof_str_flex_array for song.urlEric Wong2008-10-121-2/+2
| | | | | | This can potentially save a few bytes on 64-bit architectures and makes it more obvious what we're doing.
* gcc: define mpd_sizeof_str_flex_arrayEric Wong2008-10-121-0/+5
| | | | | | This can save a few bytes here on newer gcc and there and will hopefully make it more obvious what we're doing with that last struct element
* Merge branch 'mk/directory'Eric Wong2008-10-1131-1298/+1471
|\ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * mk/directory: (59 commits) directory: don't use identical struct and variable names update: replaced update_return with global "modified" flag update: make the variable "progress" static update: don't print debug message when song was not modified update: fix memory leak in directory_update_init() update: make the job id unsigned update: job ID must be positive update: check progress!=IDLE in reap_update_task() update: fixed stack corruption due to pthread_join() call updated: always call removeDeletedFromDirectory() update: eliminated addSubDirectoryToDirectory() update: make the "song" variable more local update: do the recursive directory check only once update: copy stat to new directory update: avoid duplicate stat() calls update: rewrote updatePath() using updateInDirectory() update: don't export updateDirectory() update: pass const pointer to addSubDirectoryToDirectory() update: never pass root path to updatePath() update: merged addDirectoryPathToDB() into addParentPathToDB() ... Conflicts: src/song.c
| * directory: don't use identical struct and variable namesEric Wong2008-10-1110-130/+128
| | | | | | | | | | | | Duplicated tokens in close proximity takes too long for my head to parse; and "dir" is an easy and common abbreviation for "directory".
| * update: replaced update_return with global "modified" flagMax Kellermann2008-10-111-68/+39
| | | | | | | | | | | | | | There is only once update thread at a time. Make the "modified" flag global and remove the return values of most functions. Propagating an error is only useful for updateDirectory(), since updateInDirectory() will delete failed subdirectories.
| * update: make the variable "progress" staticMax Kellermann2008-10-111-1/+1
| |
| * update: don't print debug message when song was not modifiedMax Kellermann2008-10-111-4/+5
| | | | | | | | | | When a song file was not modified, MPD printed the debug message "not a directory or music", because the first "if" branch did not return.
| * update: fix memory leak in directory_update_init()Max Kellermann2008-10-111-1/+4
| | | | | | | | | | When the update queue is full, directory_update_init() did not free the path argument.
| * update: make the job id unsignedMax Kellermann2008-10-112-7/+7
| | | | | | | | Since the return value cannot be -1 anymore, we can make it unsigned.
| * update: job ID must be positiveMax Kellermann2008-10-114-10/+8
| | | | | | | | | | | | | | The documentation for directory_update_init() was incorrect: a job ID must be positive, not non-negative. If the update queue is full and no job was created, it makes more sense to return 0 instead of -1, because it is more consistent with the return value of isUpdatingDB().
| * update: check progress!=IDLE in reap_update_task()Max Kellermann2008-10-111-0/+3
| | | | | | | | | | When the update task is idle, there is no need to check for deleted songs. Return early from reap_update_task().
| * update: fixed stack corruption due to pthread_join() callMax Kellermann2008-10-111-1/+3
| | | | | | | | | | | | pthread_join() expects a "pointer to a pointer" parameter, but it got a "pointer to an enum". On AMD64, an enum is smaller than a pointer, leading to a buffer overflow.
| * updated: always call removeDeletedFromDirectory()Max Kellermann2008-10-111-3/+1
| | | | | | | | | | Removed the local variable "was_empty": don't remember if the directory is new. Always call removeDeletedFromDirectory().
| * update: eliminated addSubDirectoryToDirectory()Max Kellermann2008-10-111-27/+8
| | | | | | | | | | | | In updateInDirectory(), add new directories immediately and delete them when they turn out to be empty. This simplifies the code and allows us to eliminate addSubDirectoryToDirectory().
| * update: make the "song" variable more localMax Kellermann2008-10-111-2/+1
| |
| * update: do the recursive directory check only onceMax Kellermann2008-10-111-9/+6
| | | | | | | | | | The recursive checks were performed in several functions, and sometimes a directory was checked twice.
| * update: copy stat to new directoryMax Kellermann2008-10-111-1/+3
| | | | | | | | | | | | When reading a new directory, copy the stat data (which we have anyway) to the directory struct. This may save a stat() in the future.
| * update: avoid duplicate stat() callsMax Kellermann2008-10-111-28/+38
| | | | | | | | Pass a pointer to the stat struct to more functions.
| * update: rewrote updatePath() using updateInDirectory()Max Kellermann2008-10-111-70/+19
| | | | | | | | | | | | updatePath() duplicated a lot of code from the more generic updateInDirectory(). Eliminate most of updatePath() and call updateInDirectory().
| * update: don't export updateDirectory()Max Kellermann2008-10-114-13/+20
| | | | | | | | | | | | | | | | If the user requests database update during startup, call directory_update_init(). This should be changed to fully asynchronous update later. For this to work, main_notify has to be initialized before db_init().
| * update: pass const pointer to addSubDirectoryToDirectory()Max Kellermann2008-10-111-1/+1
| | | | | | | | The stat struct isn't going to be modified, make it const.
| * update: never pass root path to updatePath()Max Kellermann2008-10-111-6/+1
| | | | | | | | | | update_task() already checks if it has got a root path. Extend this check and in turn remove a check in the inner function updatePath().
| * update: merged addDirectoryPathToDB() into addParentPathToDB()Max Kellermann2008-10-111-23/+3
| | | | | | | | | | | | | | The algorithm in addDirectoryPathToDB() can be simplified further if it is combined with the function addParentPathToDB(). Since there is no other caller of addDirectoryPathToDB(), we can do that. This saves another large stack buffer.
| * update: make addDirectoryPathToDB() non-recursiveMax Kellermann2008-10-111-11/+13
| | | | | | | | | | | | This recursive function is very dangerous because it allocates a large buffer on the stack in every iteration. That may be misused to generate a stack overflow.
| * update: delete directory after failed updateMax Kellermann2008-10-111-1/+8
| | | | | | | | | | When a directory cannot be updated, there must be something wrong with it, and the database contains stale data. Remove it.
| * update: moved code to directory_make_child_checked()Max Kellermann2008-10-111-22/+25
| | | | | | | | | | | | The branching looks a bit complicated in addDirectoryPathToDB() - improve its readability by moving code to a simplified separate function.
| * update: clear root after errorMax Kellermann2008-10-111-0/+1
| | | | | | | | | | When the root directory fails to update, its contents are invalid. Clear it then.
| * update: locked delete after update errorMax Kellermann2008-10-111-1/+37
| | | | | | | | | | | | | | When a directory failed to update, it was removed from the database, without freeing all children and songs (memory leak), and without locking (race condition). Introduce the functions clear_directory() and delete_directory(), which do both.
| * dirvec: added dirvec_clear()Max Kellermann2008-10-111-0/+5
| |
| * update: removed addToDirectory()Max Kellermann2008-10-111-38/+10
| | | | | | | | | | | | Use updateInDirectory() instead of addToDirectory(). Eliminate a duplicate stat() in updateInDirectory() by calling song_file_update() directly.
| * directory: don't query database during loadMax Kellermann2008-10-111-2/+1
| | | | | | | | | | | | Don't use db_get_directory() and traverse the full path with every directory being loaded. Just see if the current parent contains the entry. Everything else would be invalid anyway..
| * directory: check the absolute path of a subdirectory while loadingMax Kellermann2008-10-111-0/+4
| | | | | | | | | | | | | | A manipulated database could trigger an assertion failure, because the parent didn't match. Do a proper check if the new directory is within the parent's. This uses FATAL() to bail out, so MPD still dies, but it doesn't crash.