1Filesystem 2========== 3 4Simple filesystem read/write is achieved using the ``uv_fs_*`` functions and the 5``uv_fs_t`` struct. 6 7.. note:: 8 9 The libuv filesystem operations are different from :doc:`socket operations 10 <networking>`. Socket operations use the non-blocking operations provided 11 by the operating system. Filesystem operations use blocking functions 12 internally, but invoke these functions in a `thread pool`_ and notify 13 watchers registered with the event loop when application interaction is 14 required. 15 16.. _thread pool: https://docs.libuv.org/en/v1.x/threadpool.html#thread-pool-work-scheduling 17 18All filesystem functions have two forms - *synchronous* and *asynchronous*. 19 20The *synchronous* forms automatically get called (and **block**) if the 21callback is null. The return value of functions is a :ref:`libuv error code 22<libuv-error-handling>`. This is usually only useful for synchronous calls. 23The *asynchronous* form is called when a callback is passed and the return 24value is 0. 25 26Reading/Writing files 27--------------------- 28 29A file descriptor is obtained using 30 31.. code-block:: c 32 33 int uv_fs_open(uv_loop_t* loop, uv_fs_t* req, const char* path, int flags, int mode, uv_fs_cb cb) 34 35``flags`` and ``mode`` are standard 36`Unix flags <https://man7.org/linux/man-pages/man2/open.2.html>`_. 37libuv takes care of converting to the appropriate Windows flags. 38 39File descriptors are closed using 40 41.. code-block:: c 42 43 int uv_fs_close(uv_loop_t* loop, uv_fs_t* req, uv_file file, uv_fs_cb cb) 44 45 46Filesystem operation callbacks have the signature: 47 48.. code-block:: c 49 50 void callback(uv_fs_t* req); 51 52Let's see a simple implementation of ``cat``. We start with registering 53a callback for when the file is opened: 54 55.. rubric:: uvcat/main.c - opening a file 56.. literalinclude:: ../../code/uvcat/main.c 57 :language: c 58 :linenos: 59 :lines: 41-53 60 :emphasize-lines: 4, 6-7 61 62The ``result`` field of a ``uv_fs_t`` is the file descriptor in case of the 63``uv_fs_open`` callback. If the file is successfully opened, we start reading it. 64 65.. rubric:: uvcat/main.c - read callback 66.. literalinclude:: ../../code/uvcat/main.c 67 :language: c 68 :linenos: 69 :lines: 26-39 70 :emphasize-lines: 2,8,12 71 72In the case of a read call, you should pass an *initialized* buffer which will 73be filled with data before the read callback is triggered. The ``uv_fs_*`` 74operations map almost directly to certain POSIX functions, so EOF is indicated 75in this case by ``result`` being 0. In the case of streams or pipes, the 76``UV_EOF`` constant would have been passed as a status instead. 77 78Here you see a common pattern when writing asynchronous programs. The 79``uv_fs_close()`` call is performed synchronously. *Usually tasks which are 80one-off, or are done as part of the startup or shutdown stage are performed 81synchronously, since we are interested in fast I/O when the program is going 82about its primary task and dealing with multiple I/O sources*. For solo tasks 83the performance difference usually is negligible and may lead to simpler code. 84 85Filesystem writing is similarly simple using ``uv_fs_write()``. *Your callback 86will be triggered after the write is complete*. In our case the callback 87simply drives the next read. Thus read and write proceed in lockstep via 88callbacks. 89 90.. rubric:: uvcat/main.c - write callback 91.. literalinclude:: ../../code/uvcat/main.c 92 :language: c 93 :linenos: 94 :lines: 17-24 95 :emphasize-lines: 6 96 97.. warning:: 98 99 Due to the way filesystems and disk drives are configured for performance, 100 a write that 'succeeds' may not be committed to disk yet. 101 102We set the dominos rolling in ``main()``: 103 104.. rubric:: uvcat/main.c 105.. literalinclude:: ../../code/uvcat/main.c 106 :language: c 107 :linenos: 108 :lines: 55- 109 :emphasize-lines: 2 110 111.. warning:: 112 113 The ``uv_fs_req_cleanup()`` function must always be called on filesystem 114 requests to free internal memory allocations in libuv. 115 116Filesystem operations 117--------------------- 118 119All the standard filesystem operations like ``unlink``, ``rmdir``, ``stat`` are 120supported asynchronously and have intuitive argument order. They follow the 121same patterns as the read/write/open calls, returning the result in the 122``uv_fs_t.result`` field. The full list: 123 124.. rubric:: Filesystem operations 125.. code-block:: c 126 127 int uv_fs_close(uv_loop_t* loop, uv_fs_t* req, uv_file file, uv_fs_cb cb); 128 int uv_fs_open(uv_loop_t* loop, uv_fs_t* req, const char* path, int flags, int mode, uv_fs_cb cb); 129 int uv_fs_read(uv_loop_t* loop, uv_fs_t* req, uv_file file, const uv_buf_t bufs[], unsigned int nbufs, int64_t offset, uv_fs_cb cb); 130 int uv_fs_unlink(uv_loop_t* loop, uv_fs_t* req, const char* path, uv_fs_cb cb); 131 int uv_fs_write(uv_loop_t* loop, uv_fs_t* req, uv_file file, const uv_buf_t bufs[], unsigned int nbufs, int64_t offset, uv_fs_cb cb); 132 int uv_fs_copyfile(uv_loop_t* loop, uv_fs_t* req, const char* path, const char* new_path, int flags, uv_fs_cb cb); 133 int uv_fs_mkdir(uv_loop_t* loop, uv_fs_t* req, const char* path, int mode, uv_fs_cb cb); 134 int uv_fs_mkdtemp(uv_loop_t* loop, uv_fs_t* req, const char* tpl, uv_fs_cb cb); 135 int uv_fs_mkstemp(uv_loop_t* loop, uv_fs_t* req, const char* tpl, uv_fs_cb cb); 136 int uv_fs_rmdir(uv_loop_t* loop, uv_fs_t* req, const char* path, uv_fs_cb cb); 137 int uv_fs_scandir(uv_loop_t* loop, uv_fs_t* req, const char* path, int flags, uv_fs_cb cb); 138 int uv_fs_scandir_next(uv_fs_t* req, uv_dirent_t* ent); 139 int uv_fs_opendir(uv_loop_t* loop, uv_fs_t* req, const char* path, uv_fs_cb cb); 140 int uv_fs_readdir(uv_loop_t* loop, uv_fs_t* req, uv_dir_t* dir, uv_fs_cb cb); 141 int uv_fs_closedir(uv_loop_t* loop, uv_fs_t* req, uv_dir_t* dir, uv_fs_cb cb); 142 int uv_fs_stat(uv_loop_t* loop, uv_fs_t* req, const char* path, uv_fs_cb cb); 143 int uv_fs_fstat(uv_loop_t* loop, uv_fs_t* req, uv_file file, uv_fs_cb cb); 144 int uv_fs_rename(uv_loop_t* loop, uv_fs_t* req, const char* path, const char* new_path, uv_fs_cb cb); 145 int uv_fs_fsync(uv_loop_t* loop, uv_fs_t* req, uv_file file, uv_fs_cb cb); 146 int uv_fs_fdatasync(uv_loop_t* loop, uv_fs_t* req, uv_file file, uv_fs_cb cb); 147 int uv_fs_ftruncate(uv_loop_t* loop, uv_fs_t* req, uv_file file, int64_t offset, uv_fs_cb cb); 148 int uv_fs_sendfile(uv_loop_t* loop, uv_fs_t* req, uv_file out_fd, uv_file in_fd, int64_t in_offset, size_t length, uv_fs_cb cb); 149 int uv_fs_access(uv_loop_t* loop, uv_fs_t* req, const char* path, int mode, uv_fs_cb cb); 150 int uv_fs_chmod(uv_loop_t* loop, uv_fs_t* req, const char* path, int mode, uv_fs_cb cb); 151 int uv_fs_utime(uv_loop_t* loop, uv_fs_t* req, const char* path, double atime, double mtime, uv_fs_cb cb); 152 int uv_fs_futime(uv_loop_t* loop, uv_fs_t* req, uv_file file, double atime, double mtime, uv_fs_cb cb); 153 int uv_fs_lutime(uv_loop_t* loop, uv_fs_t* req, const char* path, double atime, double mtime, uv_fs_cb cb); 154 int uv_fs_lstat(uv_loop_t* loop, uv_fs_t* req, const char* path, uv_fs_cb cb); 155 int uv_fs_link(uv_loop_t* loop, uv_fs_t* req, const char* path, const char* new_path, uv_fs_cb cb); 156 int uv_fs_symlink(uv_loop_t* loop, uv_fs_t* req, const char* path, const char* new_path, int flags, uv_fs_cb cb); 157 int uv_fs_readlink(uv_loop_t* loop, uv_fs_t* req, const char* path, uv_fs_cb cb); 158 int uv_fs_realpath(uv_loop_t* loop, uv_fs_t* req, const char* path, uv_fs_cb cb); 159 int uv_fs_fchmod(uv_loop_t* loop, uv_fs_t* req, uv_file file, int mode, uv_fs_cb cb); 160 int uv_fs_chown(uv_loop_t* loop, uv_fs_t* req, const char* path, uv_uid_t uid, uv_gid_t gid, uv_fs_cb cb); 161 int uv_fs_fchown(uv_loop_t* loop, uv_fs_t* req, uv_file file, uv_uid_t uid, uv_gid_t gid, uv_fs_cb cb); 162 int uv_fs_lchown(uv_loop_t* loop, uv_fs_t* req, const char* path, uv_uid_t uid, uv_gid_t gid, uv_fs_cb cb); 163 int uv_fs_statfs(uv_loop_t* loop, uv_fs_t* req, const char* path, uv_fs_cb cb); 164 165 166.. _buffers-and-streams: 167 168Buffers and Streams 169------------------- 170 171The basic I/O handle in libuv is the stream (``uv_stream_t``). TCP sockets, UDP 172sockets, and pipes for file I/O and IPC are all treated as stream subclasses. 173 174Streams are initialized using custom functions for each subclass, then operated 175upon using 176 177.. code-block:: c 178 179 int uv_read_start(uv_stream_t*, uv_alloc_cb alloc_cb, uv_read_cb read_cb); 180 int uv_read_stop(uv_stream_t*); 181 int uv_write(uv_write_t* req, uv_stream_t* handle, 182 const uv_buf_t bufs[], unsigned int nbufs, uv_write_cb cb); 183 184The stream based functions are simpler to use than the filesystem ones and 185libuv will automatically keep reading from a stream when ``uv_read_start()`` is 186called once, until ``uv_read_stop()`` is called. 187 188The discrete unit of data is the buffer -- ``uv_buf_t``. This is simply 189a collection of a pointer to bytes (``uv_buf_t.base``) and the length 190(``uv_buf_t.len``). The ``uv_buf_t`` is lightweight and passed around by value. 191What does require management is the actual bytes, which have to be allocated 192and freed by the application. 193 194.. ERROR:: 195 196 **THIS PROGRAM DOES NOT ALWAYS WORK, NEED SOMETHING BETTER** 197 198To demonstrate streams we will need to use ``uv_pipe_t``. This allows streaming 199local files [#]_. Here is a simple tee utility using libuv. Doing all operations 200asynchronously shows the power of evented I/O. The two writes won't block each 201other, but we have to be careful to copy over the buffer data to ensure we don't 202free a buffer until it has been written. 203 204The program is to be executed as:: 205 206 ./uvtee <output_file> 207 208We start off opening pipes on the files we require. libuv pipes to a file are 209opened as bidirectional by default. 210 211.. rubric:: uvtee/main.c - read on pipes 212.. literalinclude:: ../../code/uvtee/main.c 213 :language: c 214 :linenos: 215 :lines: 62-80 216 :emphasize-lines: 4,5,15 217 218The third argument of ``uv_pipe_init()`` should be set to 1 for IPC using named 219pipes. This is covered in :doc:`processes`. The ``uv_pipe_open()`` call 220associates the pipe with the file descriptor, in this case ``0`` (standard 221input). 222 223We start monitoring ``stdin``. The ``alloc_buffer`` callback is invoked as new 224buffers are required to hold incoming data. ``read_stdin`` will be called with 225these buffers. 226 227.. rubric:: uvtee/main.c - reading buffers 228.. literalinclude:: ../../code/uvtee/main.c 229 :language: c 230 :linenos: 231 :lines: 19-22,44-60 232 233The standard ``malloc`` is sufficient here, but you can use any memory allocation 234scheme. For example, node.js uses its own slab allocator which associates 235buffers with V8 objects. 236 237The read callback ``nread`` parameter is less than 0 on any error. This error 238might be EOF, in which case we close all the streams, using the generic close 239function ``uv_close()`` which deals with the handle based on its internal type. 240Otherwise ``nread`` is a non-negative number and we can attempt to write that 241many bytes to the output streams. Finally remember that buffer allocation and 242deallocation is application responsibility, so we free the data. 243 244The allocation callback may return a buffer with length zero if it fails to 245allocate memory. In this case, the read callback is invoked with error 246UV_ENOBUFS. libuv will continue to attempt to read the stream though, so you 247must explicitly call ``uv_close()`` if you want to stop when allocation fails. 248 249The read callback may be called with ``nread = 0``, indicating that at this 250point there is nothing to be read. Most applications will just ignore this. 251 252.. rubric:: uvtee/main.c - Write to pipe 253.. literalinclude:: ../../code/uvtee/main.c 254 :language: c 255 :linenos: 256 :lines: 9-13,23-42 257 258``write_data()`` makes a copy of the buffer obtained from read. This buffer 259does not get passed through to the write callback trigged on write completion. To 260get around this we wrap a write request and a buffer in ``write_req_t`` and 261unwrap it in the callbacks. We make a copy so we can free the two buffers from 262the two calls to ``write_data`` independently of each other. While acceptable 263for a demo program like this, you'll probably want smarter memory management, 264like reference counted buffers or a pool of buffers in any major application. 265 266.. WARNING:: 267 268 If your program is meant to be used with other programs it may knowingly or 269 unknowingly be writing to a pipe. This makes it susceptible to `aborting on 270 receiving a SIGPIPE`_. It is a good idea to insert:: 271 272 signal(SIGPIPE, SIG_IGN) 273 274 in the initialization stages of your application. 275 276.. _aborting on receiving a SIGPIPE: http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod#The_special_problem_of_SIGPIPE 277 278File change events 279------------------ 280 281All modern operating systems provide APIs to put watches on individual files or 282directories and be informed when the files are modified. libuv wraps common 283file change notification libraries [#fsnotify]_. This is one of the more 284inconsistent parts of libuv. File change notification systems are themselves 285extremely varied across platforms so getting everything working everywhere is 286difficult. To demonstrate, I'm going to build a simple utility which runs 287a command whenever any of the watched files change:: 288 289 ./onchange <command> <file1> [file2] ... 290 291.. note:: 292 293 Currently this example only works on OSX and Windows. 294 Refer to the `notes of uv_fs_event_start`_ function. 295 296.. _notes of uv_fs_event_start: https://docs.libuv.org/en/v1.x/fs_event.html#c.uv_fs_event_start 297 298The file change notification is started using ``uv_fs_event_init()``: 299 300.. rubric:: onchange/main.c - The setup 301.. literalinclude:: ../../code/onchange/main.c 302 :language: c 303 :linenos: 304 :lines: 26- 305 :emphasize-lines: 15 306 307The third argument is the actual file or directory to monitor. The last 308argument, ``flags``, can be: 309 310.. code-block:: c 311 312 /* 313 * Flags to be passed to uv_fs_event_start(). 314 */ 315 enum uv_fs_event_flags { 316 UV_FS_EVENT_WATCH_ENTRY = 1, 317 UV_FS_EVENT_STAT = 2, 318 UV_FS_EVENT_RECURSIVE = 4 319 }; 320 321``UV_FS_EVENT_WATCH_ENTRY`` and ``UV_FS_EVENT_STAT`` don't do anything (yet). 322``UV_FS_EVENT_RECURSIVE`` will start watching subdirectories as well on 323supported platforms. 324 325The callback will receive the following arguments: 326 327 #. ``uv_fs_event_t *handle`` - The handle. The ``path`` field of the handle 328 is the file on which the watch was set. 329 #. ``const char *filename`` - If a directory is being monitored, this is the 330 file which was changed. Only non-``null`` on Linux and Windows. May be ``null`` 331 even on those platforms. 332 #. ``int events`` - one of ``UV_RENAME`` or ``UV_CHANGE``, or a bitwise OR of 333 both. 334 #. ``int status`` - If ``status < 0``, there is an :ref:`libuv error<libuv-error-handling>`. 335 336In our example we simply print the arguments and run the command using 337``system()``. 338 339.. rubric:: onchange/main.c - file change notification callback 340.. literalinclude:: ../../code/onchange/main.c 341 :language: c 342 :linenos: 343 :lines: 9-24 344 345---- 346 347.. [#fsnotify] inotify on Linux, FSEvents on Darwin, kqueue on BSDs, 348 ReadDirectoryChangesW on Windows, event ports on Solaris, unsupported on Cygwin 349.. [#] see :ref:`pipes` 350