2 * I/O functions for libusb
3 * Copyright (C) 2007-2009 Daniel Drake <dsd@gentoo.org>
4 * Copyright (c) 2001 Johannes Erdfelt <johannes@erdfelt.com>
6 * This library is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * This library is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with this library; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
33 #ifdef USBI_TIMERFD_AVAILABLE
34 #include <sys/timerfd.h>
40 * \page io Synchronous and asynchronous device I/O
42 * \section intro Introduction
44 * If you're using libusb in your application, you're probably wanting to
45 * perform I/O with devices - you want to perform USB data transfers.
47 * libusb offers two separate interfaces for device I/O. This page aims to
48 * introduce the two in order to help you decide which one is more suitable
49 * for your application. You can also choose to use both interfaces in your
50 * application by considering each transfer on a case-by-case basis.
52 * Once you have read through the following discussion, you should consult the
53 * detailed API documentation pages for the details:
57 * \section theory Transfers at a logical level
59 * At a logical level, USB transfers typically happen in two parts. For
60 * example, when reading data from a endpoint:
61 * -# A request for data is sent to the device
62 * -# Some time later, the incoming data is received by the host
64 * or when writing data to an endpoint:
66 * -# The data is sent to the device
67 * -# Some time later, the host receives acknowledgement from the device that
68 * the data has been transferred.
70 * There may be an indefinite delay between the two steps. Consider a
71 * fictional USB input device with a button that the user can press. In order
72 * to determine when the button is pressed, you would likely submit a request
73 * to read data on a bulk or interrupt endpoint and wait for data to arrive.
74 * Data will arrive when the button is pressed by the user, which is
75 * potentially hours later.
77 * libusb offers both a synchronous and an asynchronous interface to performing
78 * USB transfers. The main difference is that the synchronous interface
79 * combines both steps indicated above into a single function call, whereas
80 * the asynchronous interface separates them.
82 * \section sync The synchronous interface
84 * The synchronous I/O interface allows you to perform a USB transfer with
85 * a single function call. When the function call returns, the transfer has
86 * completed and you can parse the results.
88 * If you have used the libusb-0.1 before, this I/O style will seem familar to
89 * you. libusb-0.1 only offered a synchronous interface.
91 * In our input device example, to read button presses you might write code
92 * in the following style:
94 unsigned char data[4];
96 int r = libusb_bulk_transfer(handle, EP_IN, data, sizeof(data), &actual_length, 0);
97 if (r == 0 && actual_length == sizeof(data)) {
98 // results of the transaction can now be found in the data buffer
99 // parse them here and report button press
105 * The main advantage of this model is simplicity: you did everything with
106 * a single simple function call.
108 * However, this interface has its limitations. Your application will sleep
109 * inside libusb_bulk_transfer() until the transaction has completed. If it
110 * takes the user 3 hours to press the button, your application will be
111 * sleeping for that long. Execution will be tied up inside the library -
112 * the entire thread will be useless for that duration.
114 * Another issue is that by tieing up the thread with that single transaction
115 * there is no possibility of performing I/O with multiple endpoints and/or
116 * multiple devices simultaneously, unless you resort to creating one thread
119 * Additionally, there is no opportunity to cancel the transfer after the
120 * request has been submitted.
122 * For details on how to use the synchronous API, see the
123 * \ref syncio "synchronous I/O API documentation" pages.
125 * \section async The asynchronous interface
127 * Asynchronous I/O is the most significant new feature in libusb-1.0.
128 * Although it is a more complex interface, it solves all the issues detailed
131 * Instead of providing which functions that block until the I/O has complete,
132 * libusb's asynchronous interface presents non-blocking functions which
133 * begin a transfer and then return immediately. Your application passes a
134 * callback function pointer to this non-blocking function, which libusb will
135 * call with the results of the transaction when it has completed.
137 * Transfers which have been submitted through the non-blocking functions
138 * can be cancelled with a separate function call.
140 * The non-blocking nature of this interface allows you to be simultaneously
141 * performing I/O to multiple endpoints on multiple devices, without having
144 * This added flexibility does come with some complications though:
145 * - In the interest of being a lightweight library, libusb does not create
146 * threads and can only operate when your application is calling into it. Your
147 * application must call into libusb from it's main loop when events are ready
148 * to be handled, or you must use some other scheme to allow libusb to
149 * undertake whatever work needs to be done.
150 * - libusb also needs to be called into at certain fixed points in time in
151 * order to accurately handle transfer timeouts.
152 * - Memory handling becomes more complex. You cannot use stack memory unless
153 * the function with that stack is guaranteed not to return until the transfer
154 * callback has finished executing.
155 * - You generally lose some linearity from your code flow because submitting
156 * the transfer request is done in a separate function from where the transfer
157 * results are handled. This becomes particularly obvious when you want to
158 * submit a second transfer based on the results of an earlier transfer.
160 * Internally, libusb's synchronous interface is expressed in terms of function
161 * calls to the asynchronous interface.
163 * For details on how to use the asynchronous API, see the
164 * \ref asyncio "asynchronous I/O API" documentation pages.
169 * \page packetoverflow Packets and overflows
171 * \section packets Packet abstraction
173 * The USB specifications describe how data is transmitted in packets, with
174 * constraints on packet size defined by endpoint descriptors. The host must
175 * not send data payloads larger than the endpoint's maximum packet size.
177 * libusb and the underlying OS abstract out the packet concept, allowing you
178 * to request transfers of any size. Internally, the request will be divided
179 * up into correctly-sized packets. You do not have to be concerned with
180 * packet sizes, but there is one exception when considering overflows.
182 * \section overflow Bulk/interrupt transfer overflows
184 * When requesting data on a bulk endpoint, libusb requires you to supply a
185 * buffer and the maximum number of bytes of data that libusb can put in that
186 * buffer. However, the size of the buffer is not communicated to the device -
187 * the device is just asked to send any amount of data.
189 * There is no problem if the device sends an amount of data that is less than
190 * or equal to the buffer size. libusb reports this condition to you through
191 * the \ref libusb_transfer::actual_length "libusb_transfer.actual_length"
194 * Problems may occur if the device attempts to send more data than can fit in
195 * the buffer. libusb reports LIBUSB_TRANSFER_OVERFLOW for this condition but
196 * other behaviour is largely undefined: actual_length may or may not be
197 * accurate, the chunk of data that can fit in the buffer (before overflow)
198 * may or may not have been transferred.
200 * Overflows are nasty, but can be avoided. Even though you were told to
201 * ignore packets above, think about the lower level details: each transfer is
202 * split into packets (typically small, with a maximum size of 512 bytes).
203 * Overflows can only happen if the final packet in an incoming data transfer
204 * is smaller than the actual packet that the device wants to transfer.
205 * Therefore, you will never see an overflow if your transfer buffer size is a
206 * multiple of the endpoint's packet size: the final packet will either
207 * fill up completely or will be only partially filled.
211 * @defgroup asyncio Asynchronous device I/O
213 * This page details libusb's asynchronous (non-blocking) API for USB device
214 * I/O. This interface is very powerful but is also quite complex - you will
215 * need to read this page carefully to understand the necessary considerations
216 * and issues surrounding use of this interface. Simplistic applications
217 * may wish to consider the \ref syncio "synchronous I/O API" instead.
219 * The asynchronous interface is built around the idea of separating transfer
220 * submission and handling of transfer completion (the synchronous model
221 * combines both of these into one). There may be a long delay between
222 * submission and completion, however the asynchronous submission function
223 * is non-blocking so will return control to your application during that
224 * potentially long delay.
226 * \section asyncabstraction Transfer abstraction
228 * For the asynchronous I/O, libusb implements the concept of a generic
229 * transfer entity for all types of I/O (control, bulk, interrupt,
230 * isochronous). The generic transfer object must be treated slightly
231 * differently depending on which type of I/O you are performing with it.
233 * This is represented by the public libusb_transfer structure type.
235 * \section asynctrf Asynchronous transfers
237 * We can view asynchronous I/O as a 5 step process:
238 * -# <b>Allocation</b>: allocate a libusb_transfer
239 * -# <b>Filling</b>: populate the libusb_transfer instance with information
240 * about the transfer you wish to perform
241 * -# <b>Submission</b>: ask libusb to submit the transfer
242 * -# <b>Completion handling</b>: examine transfer results in the
243 * libusb_transfer structure
244 * -# <b>Deallocation</b>: clean up resources
247 * \subsection asyncalloc Allocation
249 * This step involves allocating memory for a USB transfer. This is the
250 * generic transfer object mentioned above. At this stage, the transfer
251 * is "blank" with no details about what type of I/O it will be used for.
253 * Allocation is done with the libusb_alloc_transfer() function. You must use
254 * this function rather than allocating your own transfers.
256 * \subsection asyncfill Filling
258 * This step is where you take a previously allocated transfer and fill it
259 * with information to determine the message type and direction, data buffer,
260 * callback function, etc.
262 * You can either fill the required fields yourself or you can use the
263 * helper functions: libusb_fill_control_transfer(), libusb_fill_bulk_transfer()
264 * and libusb_fill_interrupt_transfer().
266 * \subsection asyncsubmit Submission
268 * When you have allocated a transfer and filled it, you can submit it using
269 * libusb_submit_transfer(). This function returns immediately but can be
270 * regarded as firing off the I/O request in the background.
272 * \subsection asynccomplete Completion handling
274 * After a transfer has been submitted, one of four things can happen to it:
276 * - The transfer completes (i.e. some data was transferred)
277 * - The transfer has a timeout and the timeout expires before all data is
279 * - The transfer fails due to an error
280 * - The transfer is cancelled
282 * Each of these will cause the user-specified transfer callback function to
283 * be invoked. It is up to the callback function to determine which of the
284 * above actually happened and to act accordingly.
286 * The user-specified callback is passed a pointer to the libusb_transfer
287 * structure which was used to setup and submit the transfer. At completion
288 * time, libusb has populated this structure with results of the transfer:
289 * success or failure reason, number of bytes of data transferred, etc. See
290 * the libusb_transfer structure documentation for more information.
292 * \subsection Deallocation
294 * When a transfer has completed (i.e. the callback function has been invoked),
295 * you are advised to free the transfer (unless you wish to resubmit it, see
296 * below). Transfers are deallocated with libusb_free_transfer().
298 * It is undefined behaviour to free a transfer which has not completed.
300 * \section asyncresubmit Resubmission
302 * You may be wondering why allocation, filling, and submission are all
303 * separated above where they could reasonably be combined into a single
306 * The reason for separation is to allow you to resubmit transfers without
307 * having to allocate new ones every time. This is especially useful for
308 * common situations dealing with interrupt endpoints - you allocate one
309 * transfer, fill and submit it, and when it returns with results you just
310 * resubmit it for the next interrupt.
312 * \section asynccancel Cancellation
314 * Another advantage of using the asynchronous interface is that you have
315 * the ability to cancel transfers which have not yet completed. This is
316 * done by calling the libusb_cancel_transfer() function.
318 * libusb_cancel_transfer() is asynchronous/non-blocking in itself. When the
319 * cancellation actually completes, the transfer's callback function will
320 * be invoked, and the callback function should check the transfer status to
321 * determine that it was cancelled.
323 * Freeing the transfer after it has been cancelled but before cancellation
324 * has completed will result in undefined behaviour.
326 * \section bulk_overflows Overflows on device-to-host bulk/interrupt endpoints
328 * If your device does not have predictable transfer sizes (or it misbehaves),
329 * your application may submit a request for data on an IN endpoint which is
330 * smaller than the data that the device wishes to send. In some circumstances
331 * this will cause an overflow, which is a nasty condition to deal with. See
332 * the \ref packetoverflow page for discussion.
334 * \section asyncctrl Considerations for control transfers
336 * The <tt>libusb_transfer</tt> structure is generic and hence does not
337 * include specific fields for the control-specific setup packet structure.
339 * In order to perform a control transfer, you must place the 8-byte setup
340 * packet at the start of the data buffer. To simplify this, you could
341 * cast the buffer pointer to type struct libusb_control_setup, or you can
342 * use the helper function libusb_fill_control_setup().
344 * The wLength field placed in the setup packet must be the length you would
345 * expect to be sent in the setup packet: the length of the payload that
346 * follows (or the expected maximum number of bytes to receive). However,
347 * the length field of the libusb_transfer object must be the length of
348 * the data buffer - i.e. it should be wLength <em>plus</em> the size of
349 * the setup packet (LIBUSB_CONTROL_SETUP_SIZE).
351 * If you use the helper functions, this is simplified for you:
352 * -# Allocate a buffer of size LIBUSB_CONTROL_SETUP_SIZE plus the size of the
353 * data you are sending/requesting.
354 * -# Call libusb_fill_control_setup() on the data buffer, using the transfer
355 * request size as the wLength value (i.e. do not include the extra space you
356 * allocated for the control setup).
357 * -# If this is a host-to-device transfer, place the data to be transferred
358 * in the data buffer, starting at offset LIBUSB_CONTROL_SETUP_SIZE.
359 * -# Call libusb_fill_control_transfer() to associate the data buffer with
360 * the transfer (and to set the remaining details such as callback and timeout).
361 * - Note that there is no parameter to set the length field of the transfer.
362 * The length is automatically inferred from the wLength field of the setup
364 * -# Submit the transfer.
366 * The multi-byte control setup fields (wValue, wIndex and wLength) must
367 * be given in little-endian byte order (the endianness of the USB bus).
368 * Endianness conversion is transparently handled by
369 * libusb_fill_control_setup() which is documented to accept host-endian
372 * Further considerations are needed when handling transfer completion in
373 * your callback function:
374 * - As you might expect, the setup packet will still be sitting at the start
375 * of the data buffer.
376 * - If this was a device-to-host transfer, the received data will be sitting
377 * at offset LIBUSB_CONTROL_SETUP_SIZE into the buffer.
378 * - The actual_length field of the transfer structure is relative to the
379 * wLength of the setup packet, rather than the size of the data buffer. So,
380 * if your wLength was 4, your transfer's <tt>length</tt> was 12, then you
381 * should expect an <tt>actual_length</tt> of 4 to indicate that the data was
382 * transferred in entirity.
384 * To simplify parsing of setup packets and obtaining the data from the
385 * correct offset, you may wish to use the libusb_control_transfer_get_data()
386 * and libusb_control_transfer_get_setup() functions within your transfer
389 * Even though control endpoints do not halt, a completed control transfer
390 * may have a LIBUSB_TRANSFER_STALL status code. This indicates the control
391 * request was not supported.
393 * \section asyncintr Considerations for interrupt transfers
395 * All interrupt transfers are performed using the polling interval presented
396 * by the bInterval value of the endpoint descriptor.
398 * \section asynciso Considerations for isochronous transfers
400 * Isochronous transfers are more complicated than transfers to
401 * non-isochronous endpoints.
403 * To perform I/O to an isochronous endpoint, allocate the transfer by calling
404 * libusb_alloc_transfer() with an appropriate number of isochronous packets.
406 * During filling, set \ref libusb_transfer::type "type" to
407 * \ref libusb_transfer_type::LIBUSB_TRANSFER_TYPE_ISOCHRONOUS
408 * "LIBUSB_TRANSFER_TYPE_ISOCHRONOUS", and set
409 * \ref libusb_transfer::num_iso_packets "num_iso_packets" to a value less than
410 * or equal to the number of packets you requested during allocation.
411 * libusb_alloc_transfer() does not set either of these fields for you, given
412 * that you might not even use the transfer on an isochronous endpoint.
414 * Next, populate the length field for the first num_iso_packets entries in
415 * the \ref libusb_transfer::iso_packet_desc "iso_packet_desc" array. Section
416 * 5.6.3 of the USB2 specifications describe how the maximum isochronous
417 * packet length is determined by the wMaxPacketSize field in the endpoint
419 * Two functions can help you here:
421 * - libusb_get_max_iso_packet_size() is an easy way to determine the max
422 * packet size for an isochronous endpoint. Note that the maximum packet
423 * size is actually the maximum number of bytes that can be transmitted in
424 * a single microframe, therefore this function multiplies the maximum number
425 * of bytes per transaction by the number of transaction opportunities per
427 * - libusb_set_iso_packet_lengths() assigns the same length to all packets
428 * within a transfer, which is usually what you want.
430 * For outgoing transfers, you'll obviously fill the buffer and populate the
431 * packet descriptors in hope that all the data gets transferred. For incoming
432 * transfers, you must ensure the buffer has sufficient capacity for
433 * the situation where all packets transfer the full amount of requested data.
435 * Completion handling requires some extra consideration. The
436 * \ref libusb_transfer::actual_length "actual_length" field of the transfer
437 * is meaningless and should not be examined; instead you must refer to the
438 * \ref libusb_iso_packet_descriptor::actual_length "actual_length" field of
439 * each individual packet.
441 * The \ref libusb_transfer::status "status" field of the transfer is also a
443 * - If the packets were submitted and the isochronous data microframes
444 * completed normally, status will have value
445 * \ref libusb_transfer_status::LIBUSB_TRANSFER_COMPLETED
446 * "LIBUSB_TRANSFER_COMPLETED". Note that bus errors and software-incurred
447 * delays are not counted as transfer errors; the transfer.status field may
448 * indicate COMPLETED even if some or all of the packets failed. Refer to
449 * the \ref libusb_iso_packet_descriptor::status "status" field of each
450 * individual packet to determine packet failures.
451 * - The status field will have value
452 * \ref libusb_transfer_status::LIBUSB_TRANSFER_ERROR
453 * "LIBUSB_TRANSFER_ERROR" only when serious errors were encountered.
454 * - Other transfer status codes occur with normal behaviour.
456 * The data for each packet will be found at an offset into the buffer that
457 * can be calculated as if each prior packet completed in full. The
458 * libusb_get_iso_packet_buffer() and libusb_get_iso_packet_buffer_simple()
459 * functions may help you here.
461 * \section asyncmem Memory caveats
463 * In most circumstances, it is not safe to use stack memory for transfer
464 * buffers. This is because the function that fired off the asynchronous
465 * transfer may return before libusb has finished using the buffer, and when
466 * the function returns it's stack gets destroyed. This is true for both
467 * host-to-device and device-to-host transfers.
469 * The only case in which it is safe to use stack memory is where you can
470 * guarantee that the function owning the stack space for the buffer does not
471 * return until after the transfer's callback function has completed. In every
472 * other case, you need to use heap memory instead.
474 * \section asyncflags Fine control
476 * Through using this asynchronous interface, you may find yourself repeating
477 * a few simple operations many times. You can apply a bitwise OR of certain
478 * flags to a transfer to simplify certain things:
479 * - \ref libusb_transfer_flags::LIBUSB_TRANSFER_SHORT_NOT_OK
480 * "LIBUSB_TRANSFER_SHORT_NOT_OK" results in transfers which transferred
481 * less than the requested amount of data being marked with status
482 * \ref libusb_transfer_status::LIBUSB_TRANSFER_ERROR "LIBUSB_TRANSFER_ERROR"
483 * (they would normally be regarded as COMPLETED)
484 * - \ref libusb_transfer_flags::LIBUSB_TRANSFER_FREE_BUFFER
485 * "LIBUSB_TRANSFER_FREE_BUFFER" allows you to ask libusb to free the transfer
486 * buffer when freeing the transfer.
487 * - \ref libusb_transfer_flags::LIBUSB_TRANSFER_FREE_TRANSFER
488 * "LIBUSB_TRANSFER_FREE_TRANSFER" causes libusb to automatically free the
489 * transfer after the transfer callback returns.
491 * \section asyncevent Event handling
493 * In accordance of the aim of being a lightweight library, libusb does not
494 * create threads internally. This means that libusb code does not execute
495 * at any time other than when your application is calling a libusb function.
496 * However, an asynchronous model requires that libusb perform work at various
497 * points in time - namely processing the results of previously-submitted
498 * transfers and invoking the user-supplied callback function.
500 * This gives rise to the libusb_handle_events() function which your
501 * application must call into when libusb has work do to. This gives libusb
502 * the opportunity to reap pending transfers, invoke callbacks, etc.
504 * The first issue to discuss here is how your application can figure out
505 * when libusb has work to do. In fact, there are two naive options which
506 * do not actually require your application to know this:
507 * -# Periodically call libusb_handle_events() in non-blocking mode at fixed
508 * short intervals from your main loop
509 * -# Repeatedly call libusb_handle_events() in blocking mode from a dedicated
512 * The first option is plainly not very nice, and will cause unnecessary
513 * CPU wakeups leading to increased power usage and decreased battery life.
514 * The second option is not very nice either, but may be the nicest option
515 * available to you if the "proper" approach can not be applied to your
516 * application (read on...).
518 * The recommended option is to integrate libusb with your application main
519 * event loop. libusb exposes a set of file descriptors which allow you to do
520 * this. Your main loop is probably already calling poll() or select() or a
521 * variant on a set of file descriptors for other event sources (e.g. keyboard
522 * button presses, mouse movements, network sockets, etc). You then add
523 * libusb's file descriptors to your poll()/select() calls, and when activity
524 * is detected on such descriptors you know it is time to call
525 * libusb_handle_events().
527 * There is one final event handling complication. libusb supports
528 * asynchronous transfers which time out after a specified time period, and
529 * this requires that libusb is called into at or after the timeout so that
530 * the timeout can be handled. So, in addition to considering libusb's file
531 * descriptors in your main event loop, you must also consider that libusb
532 * sometimes needs to be called into at fixed points in time even when there
533 * is no file descriptor activity.
535 * For the details on retrieving the set of file descriptors and determining
536 * the next timeout, see the \ref poll "polling and timing" API documentation.
540 * @defgroup poll Polling and timing
542 * This page documents libusb's functions for polling events and timing.
543 * These functions are only necessary for users of the
544 * \ref asyncio "asynchronous API". If you are only using the simpler
545 * \ref syncio "synchronous API" then you do not need to ever call these
548 * The justification for the functionality described here has already been
549 * discussed in the \ref asyncevent "event handling" section of the
550 * asynchronous API documentation. In summary, libusb does not create internal
551 * threads for event processing and hence relies on your application calling
552 * into libusb at certain points in time so that pending events can be handled.
553 * In order to know precisely when libusb needs to be called into, libusb
554 * offers you a set of pollable file descriptors and information about when
555 * the next timeout expires.
557 * If you are using the asynchronous I/O API, you must take one of the two
558 * following options, otherwise your I/O will not complete.
560 * \section pollsimple The simple option
562 * If your application revolves solely around libusb and does not need to
563 * handle other event sources, you can have a program structure as follows:
566 // find and open device
567 // maybe fire off some initial async I/O
569 while (user_has_not_requested_exit)
570 libusb_handle_events(ctx);
575 * With such a simple main loop, you do not have to worry about managing
576 * sets of file descriptors or handling timeouts. libusb_handle_events() will
577 * handle those details internally.
579 * \section pollmain The more advanced option
581 * In more advanced applications, you will already have a main loop which
582 * is monitoring other event sources: network sockets, X11 events, mouse
583 * movements, etc. Through exposing a set of file descriptors, libusb is
584 * designed to cleanly integrate into such main loops.
586 * In addition to polling file descriptors for the other event sources, you
587 * take a set of file descriptors from libusb and monitor those too. When you
588 * detect activity on libusb's file descriptors, you call
589 * libusb_handle_events_timeout() in non-blocking mode.
591 * What's more, libusb may also need to handle events at specific moments in
592 * time. No file descriptor activity is generated at these times, so your
593 * own application needs to be continually aware of when the next one of these
594 * moments occurs (through calling libusb_get_next_timeout()), and then it
595 * needs to call libusb_handle_events_timeout() in non-blocking mode when
596 * these moments occur. This means that you need to adjust your
597 * poll()/select() timeout accordingly.
599 * In pseudo-code, you want something that looks like:
603 libusb_get_pollfds(ctx)
604 while (user has not requested application exit) {
605 libusb_get_next_timeout(ctx);
606 poll(on libusb file descriptors plus any other event sources of interest,
607 using a timeout no larger than the value libusb just suggested)
608 if (poll() indicated activity on libusb file descriptors)
609 libusb_handle_events_timeout(ctx, 0);
610 if (time has elapsed to or beyond the libusb timeout)
611 libusb_handle_events_timeout(ctx, 0);
612 // handle events from other sources here
618 * \subsection polltime Notes on time-based events
620 * The above complication with having to track time and call into libusb at
621 * specific moments is a bit of a headache. For maximum compatibility, you do
622 * need to write your main loop as above, but you may decide that you can
623 * restrict the supported platforms of your application and get away with
624 * a more simplistic scheme.
626 * These time-based event complications are \b not required on the following
629 * - Linux, provided that the following version requirements are satisfied:
630 * - Linux v2.6.27 or newer, compiled with timerfd support
631 * - glibc v2.8 or newer
632 * - libusb v1.0.5 or newer
634 * Under these configurations, libusb_get_next_timeout() will \em always return
635 * 0, so your main loop can be simplified to:
639 libusb_get_pollfds(ctx)
640 while (user has not requested application exit) {
641 poll(on libusb file descriptors plus any other event sources of interest,
642 using any timeout that you like)
643 if (poll() indicated activity on libusb file descriptors)
644 libusb_handle_events_timeout(ctx, 0);
645 // handle events from other sources here
651 * Do remember that if you simplify your main loop to the above, you will
652 * lose compatibility with some platforms (including legacy Linux platforms,
653 * and <em>any future platforms supported by libusb which may have time-based
654 * event requirements</em>). The resultant problems will likely appear as
655 * strange bugs in your application.
657 * You can use the libusb_pollfds_handle_timeouts() function to do a runtime
658 * check to see if it is safe to ignore the time-based event complications.
659 * If your application has taken the shortcut of ignoring libusb's next timeout
660 * in your main loop, then you are advised to check the return value of
661 * libusb_pollfds_handle_timeouts() during application startup, and to abort
662 * if the platform does suffer from these timing complications.
664 * \subsection fdsetchange Changes in the file descriptor set
666 * The set of file descriptors that libusb uses as event sources may change
667 * during the life of your application. Rather than having to repeatedly
668 * call libusb_get_pollfds(), you can set up notification functions for when
669 * the file descriptor set changes using libusb_set_pollfd_notifiers().
671 * \subsection mtissues Multi-threaded considerations
673 * Unfortunately, the situation is complicated further when multiple threads
674 * come into play. If two threads are monitoring the same file descriptors,
675 * the fact that only one thread will be woken up when an event occurs causes
678 * The events lock, event waiters lock, and libusb_handle_events_locked()
679 * entities are added to solve these problems. You do not need to be concerned
680 * with these entities otherwise.
682 * See the extra documentation: \ref mtasync
685 /** \page mtasync Multi-threaded applications and asynchronous I/O
687 * libusb is a thread-safe library, but extra considerations must be applied
688 * to applications which interact with libusb from multiple threads.
690 * The underlying issue that must be addressed is that all libusb I/O
691 * revolves around monitoring file descriptors through the poll()/select()
692 * system calls. This is directly exposed at the
693 * \ref asyncio "asynchronous interface" but it is important to note that the
694 * \ref syncio "synchronous interface" is implemented on top of the
695 * asynchonrous interface, therefore the same considerations apply.
697 * The issue is that if two or more threads are concurrently calling poll()
698 * or select() on libusb's file descriptors then only one of those threads
699 * will be woken up when an event arrives. The others will be completely
700 * oblivious that anything has happened.
702 * Consider the following pseudo-code, which submits an asynchronous transfer
703 * then waits for its completion. This style is one way you could implement a
704 * synchronous interface on top of the asynchronous interface (and libusb
705 * does something similar, albeit more advanced due to the complications
706 * explained on this page).
709 void cb(struct libusb_transfer *transfer)
711 int *completed = transfer->user_data;
716 struct libusb_transfer *transfer;
717 unsigned char buffer[LIBUSB_CONTROL_SETUP_SIZE];
720 transfer = libusb_alloc_transfer(0);
721 libusb_fill_control_setup(buffer,
722 LIBUSB_REQUEST_TYPE_VENDOR | LIBUSB_ENDPOINT_OUT, 0x04, 0x01, 0, 0);
723 libusb_fill_control_transfer(transfer, dev, buffer, cb, &completed, 1000);
724 libusb_submit_transfer(transfer);
727 poll(libusb file descriptors, 120*1000);
728 if (poll indicates activity)
729 libusb_handle_events_timeout(ctx, 0);
731 printf("completed!");
736 * Here we are <em>serializing</em> completion of an asynchronous event
737 * against a condition - the condition being completion of a specific transfer.
738 * The poll() loop has a long timeout to minimize CPU usage during situations
739 * when nothing is happening (it could reasonably be unlimited).
741 * If this is the only thread that is polling libusb's file descriptors, there
742 * is no problem: there is no danger that another thread will swallow up the
743 * event that we are interested in. On the other hand, if there is another
744 * thread polling the same descriptors, there is a chance that it will receive
745 * the event that we were interested in. In this situation, <tt>myfunc()</tt>
746 * will only realise that the transfer has completed on the next iteration of
747 * the loop, <em>up to 120 seconds later.</em> Clearly a two-minute delay is
748 * undesirable, and don't even think about using short timeouts to circumvent
751 * The solution here is to ensure that no two threads are ever polling the
752 * file descriptors at the same time. A naive implementation of this would
753 * impact the capabilities of the library, so libusb offers the scheme
754 * documented below to ensure no loss of functionality.
756 * Before we go any further, it is worth mentioning that all libusb-wrapped
757 * event handling procedures fully adhere to the scheme documented below.
758 * This includes libusb_handle_events() and all the synchronous I/O functions -
759 * libusb hides this headache from you. You do not need to worry about any
760 * of these issues if you stick to that level.
762 * The problem is when we consider the fact that libusb exposes file
763 * descriptors to allow for you to integrate asynchronous USB I/O into
764 * existing main loops, effectively allowing you to do some work behind
765 * libusb's back. If you do take libusb's file descriptors and pass them to
766 * poll()/select() yourself, you need to be aware of the associated issues.
768 * \section eventlock The events lock
770 * The first concept to be introduced is the events lock. The events lock
771 * is used to serialize threads that want to handle events, such that only
772 * one thread is handling events at any one time.
774 * You must take the events lock before polling libusb file descriptors,
775 * using libusb_lock_events(). You must release the lock as soon as you have
776 * aborted your poll()/select() loop, using libusb_unlock_events().
778 * \section threadwait Letting other threads do the work for you
780 * Although the events lock is a critical part of the solution, it is not
781 * enough on it's own. You might wonder if the following is sufficient...
783 libusb_lock_events(ctx);
785 poll(libusb file descriptors, 120*1000);
786 if (poll indicates activity)
787 libusb_handle_events_timeout(ctx, 0);
789 libusb_unlock_events(ctx);
791 * ...and the answer is that it is not. This is because the transfer in the
792 * code shown above may take a long time (say 30 seconds) to complete, and
793 * the lock is not released until the transfer is completed.
795 * Another thread with similar code that wants to do event handling may be
796 * working with a transfer that completes after a few milliseconds. Despite
797 * having such a quick completion time, the other thread cannot check that
798 * status of its transfer until the code above has finished (30 seconds later)
799 * due to contention on the lock.
801 * To solve this, libusb offers you a mechanism to determine when another
802 * thread is handling events. It also offers a mechanism to block your thread
803 * until the event handling thread has completed an event (and this mechanism
804 * does not involve polling of file descriptors).
806 * After determining that another thread is currently handling events, you
807 * obtain the <em>event waiters</em> lock using libusb_lock_event_waiters().
808 * You then re-check that some other thread is still handling events, and if
809 * so, you call libusb_wait_for_event().
811 * libusb_wait_for_event() puts your application to sleep until an event
812 * occurs, or until a thread releases the events lock. When either of these
813 * things happen, your thread is woken up, and should re-check the condition
814 * it was waiting on. It should also re-check that another thread is handling
815 * events, and if not, it should start handling events itself.
817 * This looks like the following, as pseudo-code:
820 if (libusb_try_lock_events(ctx) == 0) {
821 // we obtained the event lock: do our own event handling
823 if (!libusb_event_handling_ok(ctx)) {
824 libusb_unlock_events(ctx);
827 poll(libusb file descriptors, 120*1000);
828 if (poll indicates activity)
829 libusb_handle_events_locked(ctx, 0);
831 libusb_unlock_events(ctx);
833 // another thread is doing event handling. wait for it to signal us that
834 // an event has completed
835 libusb_lock_event_waiters(ctx);
838 // now that we have the event waiters lock, double check that another
839 // thread is still handling events for us. (it may have ceased handling
840 // events in the time it took us to reach this point)
841 if (!libusb_event_handler_active(ctx)) {
842 // whoever was handling events is no longer doing so, try again
843 libusb_unlock_event_waiters(ctx);
847 libusb_wait_for_event(ctx);
849 libusb_unlock_event_waiters(ctx);
851 printf("completed!\n");
854 * A naive look at the above code may suggest that this can only support
855 * one event waiter (hence a total of 2 competing threads, the other doing
856 * event handling), because the event waiter seems to have taken the event
857 * waiters lock while waiting for an event. However, the system does support
858 * multiple event waiters, because libusb_wait_for_event() actually drops
859 * the lock while waiting, and reaquires it before continuing.
861 * We have now implemented code which can dynamically handle situations where
862 * nobody is handling events (so we should do it ourselves), and it can also
863 * handle situations where another thread is doing event handling (so we can
864 * piggyback onto them). It is also equipped to handle a combination of
865 * the two, for example, another thread is doing event handling, but for
866 * whatever reason it stops doing so before our condition is met, so we take
867 * over the event handling.
869 * Four functions were introduced in the above pseudo-code. Their importance
870 * should be apparent from the code shown above.
871 * -# libusb_try_lock_events() is a non-blocking function which attempts
872 * to acquire the events lock but returns a failure code if it is contended.
873 * -# libusb_event_handling_ok() checks that libusb is still happy for your
874 * thread to be performing event handling. Sometimes, libusb needs to
875 * interrupt the event handler, and this is how you can check if you have
876 * been interrupted. If this function returns 0, the correct behaviour is
877 * for you to give up the event handling lock, and then to repeat the cycle.
878 * The following libusb_try_lock_events() will fail, so you will become an
879 * events waiter. For more information on this, read \ref fullstory below.
880 * -# libusb_handle_events_locked() is a variant of
881 * libusb_handle_events_timeout() that you can call while holding the
882 * events lock. libusb_handle_events_timeout() itself implements similar
883 * logic to the above, so be sure not to call it when you are
884 * "working behind libusb's back", as is the case here.
885 * -# libusb_event_handler_active() determines if someone is currently
886 * holding the events lock
888 * You might be wondering why there is no function to wake up all threads
889 * blocked on libusb_wait_for_event(). This is because libusb can do this
890 * internally: it will wake up all such threads when someone calls
891 * libusb_unlock_events() or when a transfer completes (at the point after its
892 * callback has returned).
894 * \subsection fullstory The full story
896 * The above explanation should be enough to get you going, but if you're
897 * really thinking through the issues then you may be left with some more
898 * questions regarding libusb's internals. If you're curious, read on, and if
899 * not, skip to the next section to avoid confusing yourself!
901 * The immediate question that may spring to mind is: what if one thread
902 * modifies the set of file descriptors that need to be polled while another
903 * thread is doing event handling?
905 * There are 2 situations in which this may happen.
906 * -# libusb_open() will add another file descriptor to the poll set,
907 * therefore it is desirable to interrupt the event handler so that it
908 * restarts, picking up the new descriptor.
909 * -# libusb_close() will remove a file descriptor from the poll set. There
910 * are all kinds of race conditions that could arise here, so it is
911 * important that nobody is doing event handling at this time.
913 * libusb handles these issues internally, so application developers do not
914 * have to stop their event handlers while opening/closing devices. Here's how
915 * it works, focusing on the libusb_close() situation first:
917 * -# During initialization, libusb opens an internal pipe, and it adds the read
918 * end of this pipe to the set of file descriptors to be polled.
919 * -# During libusb_close(), libusb writes some dummy data on this control pipe.
920 * This immediately interrupts the event handler. libusb also records
921 * internally that it is trying to interrupt event handlers for this
922 * high-priority event.
923 * -# At this point, some of the functions described above start behaving
925 * - libusb_event_handling_ok() starts returning 1, indicating that it is NOT
926 * OK for event handling to continue.
927 * - libusb_try_lock_events() starts returning 1, indicating that another
928 * thread holds the event handling lock, even if the lock is uncontended.
929 * - libusb_event_handler_active() starts returning 1, indicating that
930 * another thread is doing event handling, even if that is not true.
931 * -# The above changes in behaviour result in the event handler stopping and
932 * giving up the events lock very quickly, giving the high-priority
933 * libusb_close() operation a "free ride" to acquire the events lock. All
934 * threads that are competing to do event handling become event waiters.
935 * -# With the events lock held inside libusb_close(), libusb can safely remove
936 * a file descriptor from the poll set, in the safety of knowledge that
937 * nobody is polling those descriptors or trying to access the poll set.
938 * -# After obtaining the events lock, the close operation completes very
939 * quickly (usually a matter of milliseconds) and then immediately releases
941 * -# At the same time, the behaviour of libusb_event_handling_ok() and friends
942 * reverts to the original, documented behaviour.
943 * -# The release of the events lock causes the threads that are waiting for
944 * events to be woken up and to start competing to become event handlers
945 * again. One of them will succeed; it will then re-obtain the list of poll
946 * descriptors, and USB I/O will then continue as normal.
948 * libusb_open() is similar, and is actually a more simplistic case. Upon a
949 * call to libusb_open():
951 * -# The device is opened and a file descriptor is added to the poll set.
952 * -# libusb sends some dummy data on the control pipe, and records that it
953 * is trying to modify the poll descriptor set.
954 * -# The event handler is interrupted, and the same behaviour change as for
955 * libusb_close() takes effect, causing all event handling threads to become
957 * -# The libusb_open() implementation takes its free ride to the events lock.
958 * -# Happy that it has successfully paused the events handler, libusb_open()
959 * releases the events lock.
960 * -# The event waiter threads are all woken up and compete to become event
961 * handlers again. The one that succeeds will obtain the list of poll
962 * descriptors again, which will include the addition of the new device.
964 * \subsection concl Closing remarks
966 * The above may seem a little complicated, but hopefully I have made it clear
967 * why such complications are necessary. Also, do not forget that this only
968 * applies to applications that take libusb's file descriptors and integrate
969 * them into their own polling loops.
971 * You may decide that it is OK for your multi-threaded application to ignore
972 * some of the rules and locks detailed above, because you don't think that
973 * two threads can ever be polling the descriptors at the same time. If that
974 * is the case, then that's good news for you because you don't have to worry.
975 * But be careful here; remember that the synchronous I/O functions do event
976 * handling internally. If you have one thread doing event handling in a loop
977 * (without implementing the rules and locking semantics documented above)
978 * and another trying to send a synchronous USB transfer, you will end up with
979 * two threads monitoring the same descriptors, and the above-described
980 * undesirable behaviour occuring. The solution is for your polling thread to
981 * play by the rules; the synchronous I/O functions do so, and this will result
982 * in them getting along in perfect harmony.
984 * If you do have a dedicated thread doing event handling, it is perfectly
985 * legal for it to take the event handling lock for long periods of time. Any
986 * synchronous I/O functions you call from other threads will transparently
987 * fall back to the "event waiters" mechanism detailed above. The only
988 * consideration that your event handling thread must apply is the one related
989 * to libusb_event_handling_ok(): you must call this before every poll(), and
990 * give up the events lock if instructed.
993 int usbi_io_init(struct libusb_context *ctx)
997 pthread_mutex_init(&ctx->flying_transfers_lock, NULL);
998 pthread_mutex_init(&ctx->pollfds_lock, NULL);
999 pthread_mutex_init(&ctx->pollfd_modify_lock, NULL);
1000 pthread_mutex_init(&ctx->events_lock, NULL);
1001 pthread_mutex_init(&ctx->event_waiters_lock, NULL);
1002 pthread_cond_init(&ctx->event_waiters_cond, NULL);
1003 list_init(&ctx->flying_transfers);
1004 list_init(&ctx->pollfds);
1006 /* FIXME should use an eventfd on kernels that support it */
1007 r = pipe(ctx->ctrl_pipe);
1009 return LIBUSB_ERROR_OTHER;
1011 r = usbi_add_pollfd(ctx, ctx->ctrl_pipe[0], POLLIN);
1015 #ifdef USBI_TIMERFD_AVAILABLE
1016 ctx->timerfd = timerfd_create(usbi_backend->get_timerfd_clockid(),
1018 if (ctx->timerfd >= 0) {
1019 usbi_dbg("using timerfd for timeouts");
1020 r = usbi_add_pollfd(ctx, ctx->timerfd, POLLIN);
1022 close(ctx->timerfd);
1026 usbi_dbg("timerfd not available (code %d error %d)", ctx->timerfd, errno);
1034 void usbi_io_exit(struct libusb_context *ctx)
1036 usbi_remove_pollfd(ctx, ctx->ctrl_pipe[0]);
1037 close(ctx->ctrl_pipe[0]);
1038 close(ctx->ctrl_pipe[1]);
1039 #ifdef USBI_TIMERFD_AVAILABLE
1040 if (usbi_using_timerfd(ctx)) {
1041 usbi_remove_pollfd(ctx, ctx->timerfd);
1042 close(ctx->timerfd);
1047 static int calculate_timeout(struct usbi_transfer *transfer)
1050 struct timespec current_time;
1051 unsigned int timeout =
1052 __USBI_TRANSFER_TO_LIBUSB_TRANSFER(transfer)->timeout;
1057 r = usbi_backend->clock_gettime(USBI_CLOCK_MONOTONIC, ¤t_time);
1059 usbi_err(ITRANSFER_CTX(transfer),
1060 "failed to read monotonic clock, errno=%d", errno);
1064 current_time.tv_sec += timeout / 1000;
1065 current_time.tv_nsec += (timeout % 1000) * 1000000;
1067 if (current_time.tv_nsec > 1000000000) {
1068 current_time.tv_nsec -= 1000000000;
1069 current_time.tv_sec++;
1072 TIMESPEC_TO_TIMEVAL(&transfer->timeout, ¤t_time);
1076 /* add a transfer to the (timeout-sorted) active transfers list.
1077 * returns 1 if the transfer has a timeout and it is the timeout next to
1079 static int add_to_flying_list(struct usbi_transfer *transfer)
1081 struct usbi_transfer *cur;
1082 struct timeval *timeout = &transfer->timeout;
1083 struct libusb_context *ctx = ITRANSFER_CTX(transfer);
1087 pthread_mutex_lock(&ctx->flying_transfers_lock);
1089 /* if we have no other flying transfers, start the list with this one */
1090 if (list_empty(&ctx->flying_transfers)) {
1091 list_add(&transfer->list, &ctx->flying_transfers);
1092 if (timerisset(timeout))
1097 /* if we have infinite timeout, append to end of list */
1098 if (!timerisset(timeout)) {
1099 list_add_tail(&transfer->list, &ctx->flying_transfers);
1103 /* otherwise, find appropriate place in list */
1104 list_for_each_entry(cur, &ctx->flying_transfers, list) {
1105 /* find first timeout that occurs after the transfer in question */
1106 struct timeval *cur_tv = &cur->timeout;
1108 if (!timerisset(cur_tv) || (cur_tv->tv_sec > timeout->tv_sec) ||
1109 (cur_tv->tv_sec == timeout->tv_sec &&
1110 cur_tv->tv_usec > timeout->tv_usec)) {
1111 list_add_tail(&transfer->list, &cur->list);
1118 /* otherwise we need to be inserted at the end */
1119 list_add_tail(&transfer->list, &ctx->flying_transfers);
1121 pthread_mutex_unlock(&ctx->flying_transfers_lock);
1125 /** \ingroup asyncio
1126 * Allocate a libusb transfer with a specified number of isochronous packet
1127 * descriptors. The returned transfer is pre-initialized for you. When the new
1128 * transfer is no longer needed, it should be freed with
1129 * libusb_free_transfer().
1131 * Transfers intended for non-isochronous endpoints (e.g. control, bulk,
1132 * interrupt) should specify an iso_packets count of zero.
1134 * For transfers intended for isochronous endpoints, specify an appropriate
1135 * number of packet descriptors to be allocated as part of the transfer.
1136 * The returned transfer is not specially initialized for isochronous I/O;
1137 * you are still required to set the
1138 * \ref libusb_transfer::num_iso_packets "num_iso_packets" and
1139 * \ref libusb_transfer::type "type" fields accordingly.
1141 * It is safe to allocate a transfer with some isochronous packets and then
1142 * use it on a non-isochronous endpoint. If you do this, ensure that at time
1143 * of submission, num_iso_packets is 0 and that type is set appropriately.
1145 * \param iso_packets number of isochronous packet descriptors to allocate
1146 * \returns a newly allocated transfer, or NULL on error
1148 API_EXPORTED struct libusb_transfer *libusb_alloc_transfer(int iso_packets)
1150 size_t os_alloc_size = usbi_backend->transfer_priv_size
1151 + (usbi_backend->add_iso_packet_size * iso_packets);
1152 int alloc_size = sizeof(struct usbi_transfer)
1153 + sizeof(struct libusb_transfer)
1154 + (sizeof(struct libusb_iso_packet_descriptor) * iso_packets)
1156 struct usbi_transfer *itransfer = malloc(alloc_size);
1160 memset(itransfer, 0, alloc_size);
1161 itransfer->num_iso_packets = iso_packets;
1162 pthread_mutex_init(&itransfer->lock, NULL);
1163 return __USBI_TRANSFER_TO_LIBUSB_TRANSFER(itransfer);
1166 /** \ingroup asyncio
1167 * Free a transfer structure. This should be called for all transfers
1168 * allocated with libusb_alloc_transfer().
1170 * If the \ref libusb_transfer_flags::LIBUSB_TRANSFER_FREE_BUFFER
1171 * "LIBUSB_TRANSFER_FREE_BUFFER" flag is set and the transfer buffer is
1172 * non-NULL, this function will also free the transfer buffer using the
1173 * standard system memory allocator (e.g. free()).
1175 * It is legal to call this function with a NULL transfer. In this case,
1176 * the function will simply return safely.
1178 * It is not legal to free an active transfer (one which has been submitted
1179 * and has not yet completed).
1181 * \param transfer the transfer to free
1183 API_EXPORTED void libusb_free_transfer(struct libusb_transfer *transfer)
1185 struct usbi_transfer *itransfer;
1189 if (transfer->flags & LIBUSB_TRANSFER_FREE_BUFFER && transfer->buffer)
1190 free(transfer->buffer);
1192 itransfer = __LIBUSB_TRANSFER_TO_USBI_TRANSFER(transfer);
1193 pthread_mutex_destroy(&itransfer->lock);
1197 /** \ingroup asyncio
1198 * Submit a transfer. This function will fire off the USB transfer and then
1199 * return immediately.
1201 * \param transfer the transfer to submit
1202 * \returns 0 on success
1203 * \returns LIBUSB_ERROR_NO_DEVICE if the device has been disconnected
1204 * \returns LIBUSB_ERROR_BUSY if the transfer has already been submitted.
1205 * \returns another LIBUSB_ERROR code on other failure
1207 API_EXPORTED int libusb_submit_transfer(struct libusb_transfer *transfer)
1209 struct libusb_context *ctx = TRANSFER_CTX(transfer);
1210 struct usbi_transfer *itransfer =
1211 __LIBUSB_TRANSFER_TO_USBI_TRANSFER(transfer);
1215 pthread_mutex_lock(&itransfer->lock);
1216 itransfer->transferred = 0;
1217 itransfer->flags = 0;
1218 r = calculate_timeout(itransfer);
1220 r = LIBUSB_ERROR_OTHER;
1224 first = add_to_flying_list(itransfer);
1225 r = usbi_backend->submit_transfer(itransfer);
1227 pthread_mutex_lock(&ctx->flying_transfers_lock);
1228 list_del(&itransfer->list);
1229 pthread_mutex_unlock(&ctx->flying_transfers_lock);
1231 #ifdef USBI_TIMERFD_AVAILABLE
1232 else if (first && usbi_using_timerfd(ctx)) {
1233 /* if this transfer has the lowest timeout of all active transfers,
1234 * rearm the timerfd with this transfer's timeout */
1235 const struct itimerspec it = { {0, 0},
1236 { itransfer->timeout.tv_sec, itransfer->timeout.tv_usec * 1000 } };
1237 usbi_dbg("arm timerfd for timeout in %dms (first in line)", transfer->timeout);
1238 r = timerfd_settime(ctx->timerfd, TFD_TIMER_ABSTIME, &it, NULL);
1240 r = LIBUSB_ERROR_OTHER;
1245 pthread_mutex_unlock(&itransfer->lock);
1249 /** \ingroup asyncio
1250 * Asynchronously cancel a previously submitted transfer.
1251 * This function returns immediately, but this does not indicate cancellation
1252 * is complete. Your callback function will be invoked at some later time
1253 * with a transfer status of
1254 * \ref libusb_transfer_status::LIBUSB_TRANSFER_CANCELLED
1255 * "LIBUSB_TRANSFER_CANCELLED."
1257 * \param transfer the transfer to cancel
1258 * \returns 0 on success
1259 * \returns LIBUSB_ERROR_NOT_FOUND if the transfer is already complete or
1261 * \returns a LIBUSB_ERROR code on failure
1263 API_EXPORTED int libusb_cancel_transfer(struct libusb_transfer *transfer)
1265 struct usbi_transfer *itransfer =
1266 __LIBUSB_TRANSFER_TO_USBI_TRANSFER(transfer);
1270 pthread_mutex_lock(&itransfer->lock);
1271 r = usbi_backend->cancel_transfer(itransfer);
1273 usbi_err(TRANSFER_CTX(transfer),
1274 "cancel transfer failed error %d", r);
1275 pthread_mutex_unlock(&itransfer->lock);
1279 #ifdef USBI_TIMERFD_AVAILABLE
1280 static int disarm_timerfd(struct libusb_context *ctx)
1282 const struct itimerspec disarm_timer = { { 0, 0 }, { 0, 0 } };
1286 r = timerfd_settime(ctx->timerfd, 0, &disarm_timer, NULL);
1288 return LIBUSB_ERROR_OTHER;
1293 /* iterates through the flying transfers, and rearms the timerfd based on the
1294 * next upcoming timeout.
1295 * must be called with flying_list locked.
1296 * returns 0 if there was no timeout to arm, 1 if the next timeout was armed,
1297 * or a LIBUSB_ERROR code on failure.
1299 static int arm_timerfd_for_next_timeout(struct libusb_context *ctx)
1301 struct usbi_transfer *transfer;
1303 list_for_each_entry(transfer, &ctx->flying_transfers, list) {
1304 struct timeval *cur_tv = &transfer->timeout;
1306 /* if we've reached transfers of infinite timeout, then we have no
1308 if (!timerisset(cur_tv))
1311 /* act on first transfer that is not already cancelled */
1312 if (!(transfer->flags & USBI_TRANSFER_TIMED_OUT)) {
1314 const struct itimerspec it = { {0, 0},
1315 { cur_tv->tv_sec, cur_tv->tv_usec * 1000 } };
1316 usbi_dbg("next timeout originally %dms", __USBI_TRANSFER_TO_LIBUSB_TRANSFER(transfer)->timeout);
1317 r = timerfd_settime(ctx->timerfd, TFD_TIMER_ABSTIME, &it, NULL);
1319 return LIBUSB_ERROR_OTHER;
1327 static int disarm_timerfd(struct libusb_context *ctx)
1331 static int arm_timerfd_for_next_timeout(struct libusb_context *ctx)
1337 /* Handle completion of a transfer (completion might be an error condition).
1338 * This will invoke the user-supplied callback function, which may end up
1339 * freeing the transfer. Therefore you cannot use the transfer structure
1340 * after calling this function, and you should free all backend-specific
1341 * data before calling it.
1342 * Do not call this function with the usbi_transfer lock held. User-specified
1343 * callback functions may attempt to directly resubmit the transfer, which
1344 * will attempt to take the lock. */
1345 int usbi_handle_transfer_completion(struct usbi_transfer *itransfer,
1346 enum libusb_transfer_status status)
1348 struct libusb_transfer *transfer =
1349 __USBI_TRANSFER_TO_LIBUSB_TRANSFER(itransfer);
1350 struct libusb_context *ctx = TRANSFER_CTX(transfer);
1354 /* FIXME: could be more intelligent with the timerfd here. we don't need
1355 * to disarm the timerfd if there was no timer running, and we only need
1356 * to rearm the timerfd if the transfer that expired was the one with
1357 * the shortest timeout. */
1359 pthread_mutex_lock(&ctx->flying_transfers_lock);
1360 list_del(&itransfer->list);
1361 r = arm_timerfd_for_next_timeout(ctx);
1362 pthread_mutex_unlock(&ctx->flying_transfers_lock);
1366 } else if (r == 0) {
1367 r = disarm_timerfd(ctx);
1372 if (status == LIBUSB_TRANSFER_COMPLETED
1373 && transfer->flags & LIBUSB_TRANSFER_SHORT_NOT_OK) {
1374 int rqlen = transfer->length;
1375 if (transfer->type == LIBUSB_TRANSFER_TYPE_CONTROL)
1376 rqlen -= LIBUSB_CONTROL_SETUP_SIZE;
1377 if (rqlen != itransfer->transferred) {
1378 usbi_dbg("interpreting short transfer as error");
1379 status = LIBUSB_TRANSFER_ERROR;
1383 flags = transfer->flags;
1384 transfer->status = status;
1385 transfer->actual_length = itransfer->transferred;
1386 if (transfer->callback)
1387 transfer->callback(transfer);
1388 /* transfer might have been freed by the above call, do not use from
1390 if (flags & LIBUSB_TRANSFER_FREE_TRANSFER)
1391 libusb_free_transfer(transfer);
1392 pthread_mutex_lock(&ctx->event_waiters_lock);
1393 pthread_cond_broadcast(&ctx->event_waiters_cond);
1394 pthread_mutex_unlock(&ctx->event_waiters_lock);
1398 /* Similar to usbi_handle_transfer_completion() but exclusively for transfers
1399 * that were asynchronously cancelled. The same concerns w.r.t. freeing of
1400 * transfers exist here.
1401 * Do not call this function with the usbi_transfer lock held. User-specified
1402 * callback functions may attempt to directly resubmit the transfer, which
1403 * will attempt to take the lock. */
1404 int usbi_handle_transfer_cancellation(struct usbi_transfer *transfer)
1406 /* if the URB was cancelled due to timeout, report timeout to the user */
1407 if (transfer->flags & USBI_TRANSFER_TIMED_OUT) {
1408 usbi_dbg("detected timeout cancellation");
1409 return usbi_handle_transfer_completion(transfer, LIBUSB_TRANSFER_TIMED_OUT);
1412 /* otherwise its a normal async cancel */
1413 return usbi_handle_transfer_completion(transfer, LIBUSB_TRANSFER_CANCELLED);
1417 * Attempt to acquire the event handling lock. This lock is used to ensure that
1418 * only one thread is monitoring libusb event sources at any one time.
1420 * You only need to use this lock if you are developing an application
1421 * which calls poll() or select() on libusb's file descriptors directly.
1422 * If you stick to libusb's event handling loop functions (e.g.
1423 * libusb_handle_events()) then you do not need to be concerned with this
1426 * While holding this lock, you are trusted to actually be handling events.
1427 * If you are no longer handling events, you must call libusb_unlock_events()
1428 * as soon as possible.
1430 * \param ctx the context to operate on, or NULL for the default context
1431 * \returns 0 if the lock was obtained successfully
1432 * \returns 1 if the lock was not obtained (i.e. another thread holds the lock)
1435 API_EXPORTED int libusb_try_lock_events(libusb_context *ctx)
1438 USBI_GET_CONTEXT(ctx);
1440 /* is someone else waiting to modify poll fds? if so, don't let this thread
1441 * start event handling */
1442 pthread_mutex_lock(&ctx->pollfd_modify_lock);
1443 r = ctx->pollfd_modify;
1444 pthread_mutex_unlock(&ctx->pollfd_modify_lock);
1446 usbi_dbg("someone else is modifying poll fds");
1450 r = pthread_mutex_trylock(&ctx->events_lock);
1454 ctx->event_handler_active = 1;
1459 * Acquire the event handling lock, blocking until successful acquisition if
1460 * it is contended. This lock is used to ensure that only one thread is
1461 * monitoring libusb event sources at any one time.
1463 * You only need to use this lock if you are developing an application
1464 * which calls poll() or select() on libusb's file descriptors directly.
1465 * If you stick to libusb's event handling loop functions (e.g.
1466 * libusb_handle_events()) then you do not need to be concerned with this
1469 * While holding this lock, you are trusted to actually be handling events.
1470 * If you are no longer handling events, you must call libusb_unlock_events()
1471 * as soon as possible.
1473 * \param ctx the context to operate on, or NULL for the default context
1476 API_EXPORTED void libusb_lock_events(libusb_context *ctx)
1478 USBI_GET_CONTEXT(ctx);
1479 pthread_mutex_lock(&ctx->events_lock);
1480 ctx->event_handler_active = 1;
1484 * Release the lock previously acquired with libusb_try_lock_events() or
1485 * libusb_lock_events(). Releasing this lock will wake up any threads blocked
1486 * on libusb_wait_for_event().
1488 * \param ctx the context to operate on, or NULL for the default context
1491 API_EXPORTED void libusb_unlock_events(libusb_context *ctx)
1493 USBI_GET_CONTEXT(ctx);
1494 ctx->event_handler_active = 0;
1495 pthread_mutex_unlock(&ctx->events_lock);
1497 /* FIXME: perhaps we should be a bit more efficient by not broadcasting
1498 * the availability of the events lock when we are modifying pollfds
1499 * (check ctx->pollfd_modify)? */
1500 pthread_mutex_lock(&ctx->event_waiters_lock);
1501 pthread_cond_broadcast(&ctx->event_waiters_cond);
1502 pthread_mutex_unlock(&ctx->event_waiters_lock);
1506 * Determine if it is still OK for this thread to be doing event handling.
1508 * Sometimes, libusb needs to temporarily pause all event handlers, and this
1509 * is the function you should use before polling file descriptors to see if
1512 * If this function instructs your thread to give up the events lock, you
1513 * should just continue the usual logic that is documented in \ref mtasync.
1514 * On the next iteration, your thread will fail to obtain the events lock,
1515 * and will hence become an event waiter.
1517 * This function should be called while the events lock is held: you don't
1518 * need to worry about the results of this function if your thread is not
1519 * the current event handler.
1521 * \param ctx the context to operate on, or NULL for the default context
1522 * \returns 1 if event handling can start or continue
1523 * \returns 0 if this thread must give up the events lock
1524 * \see \ref fullstory "Multi-threaded I/O: the full story"
1526 API_EXPORTED int libusb_event_handling_ok(libusb_context *ctx)
1529 USBI_GET_CONTEXT(ctx);
1531 /* is someone else waiting to modify poll fds? if so, don't let this thread
1532 * continue event handling */
1533 pthread_mutex_lock(&ctx->pollfd_modify_lock);
1534 r = ctx->pollfd_modify;
1535 pthread_mutex_unlock(&ctx->pollfd_modify_lock);
1537 usbi_dbg("someone else is modifying poll fds");
1546 * Determine if an active thread is handling events (i.e. if anyone is holding
1547 * the event handling lock).
1549 * \param ctx the context to operate on, or NULL for the default context
1550 * \returns 1 if a thread is handling events
1551 * \returns 0 if there are no threads currently handling events
1554 API_EXPORTED int libusb_event_handler_active(libusb_context *ctx)
1557 USBI_GET_CONTEXT(ctx);
1559 /* is someone else waiting to modify poll fds? if so, don't let this thread
1560 * start event handling -- indicate that event handling is happening */
1561 pthread_mutex_lock(&ctx->pollfd_modify_lock);
1562 r = ctx->pollfd_modify;
1563 pthread_mutex_unlock(&ctx->pollfd_modify_lock);
1565 usbi_dbg("someone else is modifying poll fds");
1569 return ctx->event_handler_active;
1573 * Acquire the event waiters lock. This lock is designed to be obtained under
1574 * the situation where you want to be aware when events are completed, but
1575 * some other thread is event handling so calling libusb_handle_events() is not
1578 * You then obtain this lock, re-check that another thread is still handling
1579 * events, then call libusb_wait_for_event().
1581 * You only need to use this lock if you are developing an application
1582 * which calls poll() or select() on libusb's file descriptors directly,
1583 * <b>and</b> may potentially be handling events from 2 threads simultaenously.
1584 * If you stick to libusb's event handling loop functions (e.g.
1585 * libusb_handle_events()) then you do not need to be concerned with this
1588 * \param ctx the context to operate on, or NULL for the default context
1591 API_EXPORTED void libusb_lock_event_waiters(libusb_context *ctx)
1593 USBI_GET_CONTEXT(ctx);
1594 pthread_mutex_lock(&ctx->event_waiters_lock);
1598 * Release the event waiters lock.
1599 * \param ctx the context to operate on, or NULL for the default context
1602 API_EXPORTED void libusb_unlock_event_waiters(libusb_context *ctx)
1604 USBI_GET_CONTEXT(ctx);
1605 pthread_mutex_unlock(&ctx->event_waiters_lock);
1609 * Wait for another thread to signal completion of an event. Must be called
1610 * with the event waiters lock held, see libusb_lock_event_waiters().
1612 * This function will block until any of the following conditions are met:
1613 * -# The timeout expires
1614 * -# A transfer completes
1615 * -# A thread releases the event handling lock through libusb_unlock_events()
1617 * Condition 1 is obvious. Condition 2 unblocks your thread <em>after</em>
1618 * the callback for the transfer has completed. Condition 3 is important
1619 * because it means that the thread that was previously handling events is no
1620 * longer doing so, so if any events are to complete, another thread needs to
1621 * step up and start event handling.
1623 * This function releases the event waiters lock before putting your thread
1624 * to sleep, and reacquires the lock as it is being woken up.
1626 * \param ctx the context to operate on, or NULL for the default context
1627 * \param tv maximum timeout for this blocking function. A NULL value
1628 * indicates unlimited timeout.
1629 * \returns 0 after a transfer completes or another thread stops event handling
1630 * \returns 1 if the timeout expired
1633 API_EXPORTED int libusb_wait_for_event(libusb_context *ctx, struct timeval *tv)
1635 struct timespec timeout;
1638 USBI_GET_CONTEXT(ctx);
1640 pthread_cond_wait(&ctx->event_waiters_cond, &ctx->event_waiters_lock);
1644 r = usbi_backend->clock_gettime(USBI_CLOCK_REALTIME, &timeout);
1646 usbi_err(ctx, "failed to read realtime clock, error %d", errno);
1647 return LIBUSB_ERROR_OTHER;
1650 timeout.tv_sec += tv->tv_sec;
1651 timeout.tv_nsec += tv->tv_usec * 1000;
1652 if (timeout.tv_nsec > 1000000000) {
1653 timeout.tv_nsec -= 1000000000;
1657 r = pthread_cond_timedwait(&ctx->event_waiters_cond,
1658 &ctx->event_waiters_lock, &timeout);
1659 return (r == ETIMEDOUT);
1662 static void handle_timeout(struct usbi_transfer *itransfer)
1664 struct libusb_transfer *transfer =
1665 __USBI_TRANSFER_TO_LIBUSB_TRANSFER(itransfer);
1668 itransfer->flags |= USBI_TRANSFER_TIMED_OUT;
1669 r = libusb_cancel_transfer(transfer);
1671 usbi_warn(TRANSFER_CTX(transfer),
1672 "async cancel failed %d errno=%d", r, errno);
1675 #ifdef USBI_OS_HANDLES_TIMEOUT
1676 static int handle_timeouts_locked(struct libusb_context *ctx)
1680 static int handle_timeouts(struct libusb_context *ctx)
1685 static int handle_timeouts_locked(struct libusb_context *ctx)
1688 struct timespec systime_ts;
1689 struct timeval systime;
1690 struct usbi_transfer *transfer;
1692 if (list_empty(&ctx->flying_transfers))
1695 /* get current time */
1696 r = usbi_backend->clock_gettime(USBI_CLOCK_MONOTONIC, &systime_ts);
1700 TIMESPEC_TO_TIMEVAL(&systime, &systime_ts);
1702 /* iterate through flying transfers list, finding all transfers that
1703 * have expired timeouts */
1704 list_for_each_entry(transfer, &ctx->flying_transfers, list) {
1705 struct timeval *cur_tv = &transfer->timeout;
1707 /* if we've reached transfers of infinite timeout, we're all done */
1708 if (!timerisset(cur_tv))
1711 /* ignore timeouts we've already handled */
1712 if (transfer->flags & USBI_TRANSFER_TIMED_OUT)
1715 /* if transfer has non-expired timeout, nothing more to do */
1716 if ((cur_tv->tv_sec > systime.tv_sec) ||
1717 (cur_tv->tv_sec == systime.tv_sec &&
1718 cur_tv->tv_usec > systime.tv_usec))
1721 /* otherwise, we've got an expired timeout to handle */
1722 handle_timeout(transfer);
1727 static int handle_timeouts(struct libusb_context *ctx)
1730 USBI_GET_CONTEXT(ctx);
1731 pthread_mutex_lock(&ctx->flying_transfers_lock);
1732 r = handle_timeouts_locked(ctx);
1733 pthread_mutex_unlock(&ctx->flying_transfers_lock);
1738 #ifdef USBI_TIMERFD_AVAILABLE
1739 static int handle_timerfd_trigger(struct libusb_context *ctx)
1743 r = disarm_timerfd(ctx);
1747 pthread_mutex_lock(&ctx->flying_transfers_lock);
1749 /* process the timeout that just happened */
1750 r = handle_timeouts_locked(ctx);
1754 /* arm for next timeout*/
1755 r = arm_timerfd_for_next_timeout(ctx);
1758 pthread_mutex_unlock(&ctx->flying_transfers_lock);
1763 /* do the actual event handling. assumes that no other thread is concurrently
1764 * doing the same thing. */
1765 static int handle_events(struct libusb_context *ctx, struct timeval *tv)
1768 struct usbi_pollfd *ipollfd;
1774 pthread_mutex_lock(&ctx->pollfds_lock);
1775 list_for_each_entry(ipollfd, &ctx->pollfds, list)
1778 /* TODO: malloc when number of fd's changes, not on every poll */
1779 fds = malloc(sizeof(*fds) * nfds);
1781 return LIBUSB_ERROR_NO_MEM;
1783 list_for_each_entry(ipollfd, &ctx->pollfds, list) {
1784 struct libusb_pollfd *pollfd = &ipollfd->pollfd;
1785 int fd = pollfd->fd;
1788 fds[i].events = pollfd->events;
1791 pthread_mutex_unlock(&ctx->pollfds_lock);
1793 timeout_ms = (tv->tv_sec * 1000) + (tv->tv_usec / 1000);
1795 /* round up to next millisecond */
1796 if (tv->tv_usec % 1000)
1799 usbi_dbg("poll() %d fds with timeout in %dms", nfds, timeout_ms);
1800 r = poll(fds, nfds, timeout_ms);
1801 usbi_dbg("poll() returned %d", r);
1804 return handle_timeouts(ctx);
1805 } else if (r == -1 && errno == EINTR) {
1807 return LIBUSB_ERROR_INTERRUPTED;
1810 usbi_err(ctx, "poll failed %d err=%d\n", r, errno);
1811 return LIBUSB_ERROR_IO;
1814 /* fd[0] is always the ctrl pipe */
1815 if (fds[0].revents) {
1816 /* another thread wanted to interrupt event handling, and it succeeded!
1817 * handle any other events that cropped up at the same time, and
1819 usbi_dbg("caught a fish on the control pipe");
1825 /* prevent OS backend from trying to handle events on ctrl pipe */
1831 #ifdef USBI_TIMERFD_AVAILABLE
1832 /* on timerfd configurations, fds[1] is the timerfd */
1833 if (usbi_using_timerfd(ctx) && fds[1].revents) {
1834 /* timerfd indicates that a timeout has expired */
1836 usbi_dbg("timerfd triggered");
1838 ret = handle_timerfd_trigger(ctx);
1840 /* return error code */
1843 } else if (r == 1) {
1844 /* no more active file descriptors, nothing more to do */
1848 /* more events pending...
1849 * prevent OS backend from trying to handle events on timerfd */
1856 r = usbi_backend->handle_events(ctx, fds, nfds, r);
1858 usbi_err(ctx, "backend handle_events failed with error %d", r);
1865 /* returns the smallest of:
1866 * 1. timeout of next URB
1867 * 2. user-supplied timeout
1868 * returns 1 if there is an already-expired timeout, otherwise returns 0
1871 static int get_next_timeout(libusb_context *ctx, struct timeval *tv,
1872 struct timeval *out)
1874 struct timeval timeout;
1875 int r = libusb_get_next_timeout(ctx, &timeout);
1877 /* timeout already expired? */
1878 if (!timerisset(&timeout))
1881 /* choose the smallest of next URB timeout or user specified timeout */
1882 if (timercmp(&timeout, tv, <))
1893 * Handle any pending events.
1895 * libusb determines "pending events" by checking if any timeouts have expired
1896 * and by checking the set of file descriptors for activity.
1898 * If a zero timeval is passed, this function will handle any already-pending
1899 * events and then immediately return in non-blocking style.
1901 * If a non-zero timeval is passed and no events are currently pending, this
1902 * function will block waiting for events to handle up until the specified
1903 * timeout. If an event arrives or a signal is raised, this function will
1906 * \param ctx the context to operate on, or NULL for the default context
1907 * \param tv the maximum time to block waiting for events, or zero for
1909 * \returns 0 on success, or a LIBUSB_ERROR code on failure
1911 API_EXPORTED int libusb_handle_events_timeout(libusb_context *ctx,
1915 struct timeval poll_timeout;
1917 USBI_GET_CONTEXT(ctx);
1918 r = get_next_timeout(ctx, tv, &poll_timeout);
1920 /* timeout already expired */
1921 return handle_timeouts(ctx);
1925 if (libusb_try_lock_events(ctx) == 0) {
1926 /* we obtained the event lock: do our own event handling */
1927 r = handle_events(ctx, &poll_timeout);
1928 libusb_unlock_events(ctx);
1932 /* another thread is doing event handling. wait for pthread events that
1933 * notify event completion. */
1934 libusb_lock_event_waiters(ctx);
1936 if (!libusb_event_handler_active(ctx)) {
1937 /* we hit a race: whoever was event handling earlier finished in the
1938 * time it took us to reach this point. try the cycle again. */
1939 libusb_unlock_event_waiters(ctx);
1940 usbi_dbg("event handler was active but went away, retrying");
1944 usbi_dbg("another thread is doing event handling");
1945 r = libusb_wait_for_event(ctx, &poll_timeout);
1946 libusb_unlock_event_waiters(ctx);
1951 return handle_timeouts(ctx);
1957 * Handle any pending events in blocking mode with a sensible timeout. This
1958 * timeout is currently hardcoded at 2 seconds but we may change this if we
1959 * decide other values are more sensible. For finer control over whether this
1960 * function is blocking or non-blocking, or the maximum timeout, use
1961 * libusb_handle_events_timeout() instead.
1963 * \param ctx the context to operate on, or NULL for the default context
1964 * \returns 0 on success, or a LIBUSB_ERROR code on failure
1966 API_EXPORTED int libusb_handle_events(libusb_context *ctx)
1971 return libusb_handle_events_timeout(ctx, &tv);
1975 * Handle any pending events by polling file descriptors, without checking if
1976 * any other threads are already doing so. Must be called with the event lock
1977 * held, see libusb_lock_events().
1979 * This function is designed to be called under the situation where you have
1980 * taken the event lock and are calling poll()/select() directly on libusb's
1981 * file descriptors (as opposed to using libusb_handle_events() or similar).
1982 * You detect events on libusb's descriptors, so you then call this function
1983 * with a zero timeout value (while still holding the event lock).
1985 * \param ctx the context to operate on, or NULL for the default context
1986 * \param tv the maximum time to block waiting for events, or zero for
1988 * \returns 0 on success, or a LIBUSB_ERROR code on failure
1991 API_EXPORTED int libusb_handle_events_locked(libusb_context *ctx,
1995 struct timeval poll_timeout;
1997 USBI_GET_CONTEXT(ctx);
1998 r = get_next_timeout(ctx, tv, &poll_timeout);
2000 /* timeout already expired */
2001 return handle_timeouts(ctx);
2004 return handle_events(ctx, &poll_timeout);
2008 * Determines whether your application must apply special timing considerations
2009 * when monitoring libusb's file descriptors.
2011 * This function is only useful for applications which retrieve and poll
2012 * libusb's file descriptors in their own main loop (\ref pollmain).
2014 * Ordinarily, libusb's event handler needs to be called into at specific
2015 * moments in time (in addition to times when there is activity on the file
2016 * descriptor set). The usual approach is to use libusb_get_next_timeout()
2017 * to learn about when the next timeout occurs, and to adjust your
2018 * poll()/select() timeout accordingly so that you can make a call into the
2019 * library at that time.
2021 * Some platforms supported by libusb do not come with this baggage - any
2022 * events relevant to timing will be represented by activity on the file
2023 * descriptor set, and libusb_get_next_timeout() will always return 0.
2024 * This function allows you to detect whether you are running on such a
2029 * \param ctx the context to operate on, or NULL for the default context
2030 * \returns 0 if you must call into libusb at times determined by
2031 * libusb_get_next_timeout(), or 1 if all timeout events are handled internally
2032 * or through regular activity on the file descriptors.
2033 * \see \ref pollmain "Polling libusb file descriptors for event handling"
2035 API_EXPORTED int libusb_pollfds_handle_timeouts(libusb_context *ctx)
2037 #if defined(USBI_OS_HANDLES_TIMEOUT)
2039 #elif defined(USBI_TIMERFD_AVAILABLE)
2040 USBI_GET_CONTEXT(ctx);
2041 return usbi_using_timerfd(ctx);
2048 * Determine the next internal timeout that libusb needs to handle. You only
2049 * need to use this function if you are calling poll() or select() or similar
2050 * on libusb's file descriptors yourself - you do not need to use it if you
2051 * are calling libusb_handle_events() or a variant directly.
2053 * You should call this function in your main loop in order to determine how
2054 * long to wait for select() or poll() to return results. libusb needs to be
2055 * called into at this timeout, so you should use it as an upper bound on
2056 * your select() or poll() call.
2058 * When the timeout has expired, call into libusb_handle_events_timeout()
2059 * (perhaps in non-blocking mode) so that libusb can handle the timeout.
2061 * This function may return 1 (success) and an all-zero timeval. If this is
2062 * the case, it indicates that libusb has a timeout that has already expired
2063 * so you should call libusb_handle_events_timeout() or similar immediately.
2064 * A return code of 0 indicates that there are no pending timeouts.
2066 * On some platforms, this function will always returns 0 (no pending
2067 * timeouts). See \ref polltime.
2069 * \param ctx the context to operate on, or NULL for the default context
2070 * \param tv output location for a relative time against the current
2071 * clock in which libusb must be called into in order to process timeout events
2072 * \returns 0 if there are no pending timeouts, 1 if a timeout was returned,
2073 * or LIBUSB_ERROR_OTHER on failure
2075 API_EXPORTED int libusb_get_next_timeout(libusb_context *ctx,
2078 #ifndef USBI_OS_HANDLES_TIMEOUT
2079 struct usbi_transfer *transfer;
2080 struct timespec cur_ts;
2081 struct timeval cur_tv;
2082 struct timeval *next_timeout;
2086 USBI_GET_CONTEXT(ctx);
2087 if (usbi_using_timerfd(ctx))
2090 pthread_mutex_lock(&ctx->flying_transfers_lock);
2091 if (list_empty(&ctx->flying_transfers)) {
2092 pthread_mutex_unlock(&ctx->flying_transfers_lock);
2093 usbi_dbg("no URBs, no timeout!");
2097 /* find next transfer which hasn't already been processed as timed out */
2098 list_for_each_entry(transfer, &ctx->flying_transfers, list) {
2099 if (!(transfer->flags & USBI_TRANSFER_TIMED_OUT)) {
2104 pthread_mutex_unlock(&ctx->flying_transfers_lock);
2107 usbi_dbg("all URBs have already been processed for timeouts");
2111 next_timeout = &transfer->timeout;
2113 /* no timeout for next transfer */
2114 if (!timerisset(next_timeout)) {
2115 usbi_dbg("no URBs with timeouts, no timeout!");
2119 r = usbi_backend->clock_gettime(USBI_CLOCK_MONOTONIC, &cur_ts);
2121 usbi_err(ctx, "failed to read monotonic clock, errno=%d", errno);
2122 return LIBUSB_ERROR_OTHER;
2124 TIMESPEC_TO_TIMEVAL(&cur_tv, &cur_ts);
2126 if (timercmp(&cur_tv, next_timeout, >=)) {
2127 usbi_dbg("first timeout already expired");
2130 timersub(next_timeout, &cur_tv, tv);
2131 usbi_dbg("next timeout in %d.%06ds", tv->tv_sec, tv->tv_usec);
2141 * Register notification functions for file descriptor additions/removals.
2142 * These functions will be invoked for every new or removed file descriptor
2143 * that libusb uses as an event source.
2145 * To remove notifiers, pass NULL values for the function pointers.
2147 * Note that file descriptors may have been added even before you register
2148 * these notifiers (e.g. at libusb_init() time).
2150 * Additionally, note that the removal notifier may be called during
2151 * libusb_exit() (e.g. when it is closing file descriptors that were opened
2152 * and added to the poll set at libusb_init() time). If you don't want this,
2153 * remove the notifiers immediately before calling libusb_exit().
2155 * \param ctx the context to operate on, or NULL for the default context
2156 * \param added_cb pointer to function for addition notifications
2157 * \param removed_cb pointer to function for removal notifications
2158 * \param user_data User data to be passed back to callbacks (useful for
2159 * passing context information)
2161 API_EXPORTED void libusb_set_pollfd_notifiers(libusb_context *ctx,
2162 libusb_pollfd_added_cb added_cb, libusb_pollfd_removed_cb removed_cb,
2165 USBI_GET_CONTEXT(ctx);
2166 ctx->fd_added_cb = added_cb;
2167 ctx->fd_removed_cb = removed_cb;
2168 ctx->fd_cb_user_data = user_data;
2171 /* Add a file descriptor to the list of file descriptors to be monitored.
2172 * events should be specified as a bitmask of events passed to poll(), e.g.
2173 * POLLIN and/or POLLOUT. */
2174 int usbi_add_pollfd(struct libusb_context *ctx, int fd, short events)
2176 struct usbi_pollfd *ipollfd = malloc(sizeof(*ipollfd));
2178 return LIBUSB_ERROR_NO_MEM;
2180 usbi_dbg("add fd %d events %d", fd, events);
2181 ipollfd->pollfd.fd = fd;
2182 ipollfd->pollfd.events = events;
2183 pthread_mutex_lock(&ctx->pollfds_lock);
2184 list_add_tail(&ipollfd->list, &ctx->pollfds);
2185 pthread_mutex_unlock(&ctx->pollfds_lock);
2187 if (ctx->fd_added_cb)
2188 ctx->fd_added_cb(fd, events, ctx->fd_cb_user_data);
2192 /* Remove a file descriptor from the list of file descriptors to be polled. */
2193 void usbi_remove_pollfd(struct libusb_context *ctx, int fd)
2195 struct usbi_pollfd *ipollfd;
2198 usbi_dbg("remove fd %d", fd);
2199 pthread_mutex_lock(&ctx->pollfds_lock);
2200 list_for_each_entry(ipollfd, &ctx->pollfds, list)
2201 if (ipollfd->pollfd.fd == fd) {
2207 usbi_dbg("couldn't find fd %d to remove", fd);
2208 pthread_mutex_unlock(&ctx->pollfds_lock);
2212 list_del(&ipollfd->list);
2213 pthread_mutex_unlock(&ctx->pollfds_lock);
2215 if (ctx->fd_removed_cb)
2216 ctx->fd_removed_cb(fd, ctx->fd_cb_user_data);
2220 * Retrieve a list of file descriptors that should be polled by your main loop
2221 * as libusb event sources.
2223 * The returned list is NULL-terminated and should be freed with free() when
2224 * done. The actual list contents must not be touched.
2226 * \param ctx the context to operate on, or NULL for the default context
2227 * \returns a NULL-terminated list of libusb_pollfd structures, or NULL on
2230 API_EXPORTED const struct libusb_pollfd **libusb_get_pollfds(
2231 libusb_context *ctx)
2233 struct libusb_pollfd **ret = NULL;
2234 struct usbi_pollfd *ipollfd;
2237 USBI_GET_CONTEXT(ctx);
2239 pthread_mutex_lock(&ctx->pollfds_lock);
2240 list_for_each_entry(ipollfd, &ctx->pollfds, list)
2243 ret = calloc(cnt + 1, sizeof(struct libusb_pollfd *));
2247 list_for_each_entry(ipollfd, &ctx->pollfds, list)
2248 ret[i++] = (struct libusb_pollfd *) ipollfd;
2252 pthread_mutex_unlock(&ctx->pollfds_lock);
2253 return (const struct libusb_pollfd **) ret;
2256 /* Backends call this from handle_events to report disconnection of a device.
2257 * The transfers get cancelled appropriately.
2259 void usbi_handle_disconnect(struct libusb_device_handle *handle)
2261 struct usbi_transfer *cur;
2262 struct usbi_transfer *to_cancel;
2264 usbi_dbg("device %d.%d",
2265 handle->dev->bus_number, handle->dev->device_address);
2267 /* terminate all pending transfers with the LIBUSB_TRANSFER_NO_DEVICE
2270 * this is a bit tricky because:
2271 * 1. we can't do transfer completion while holding flying_transfers_lock
2272 * 2. the transfers list can change underneath us - if we were to build a
2273 * list of transfers to complete (while holding look), the situation
2274 * might be different by the time we come to free them
2276 * so we resort to a loop-based approach as below
2277 * FIXME: is this still potentially racy?
2281 pthread_mutex_lock(&HANDLE_CTX(handle)->flying_transfers_lock);
2283 list_for_each_entry(cur, &HANDLE_CTX(handle)->flying_transfers, list)
2284 if (__USBI_TRANSFER_TO_LIBUSB_TRANSFER(cur)->dev_handle == handle) {
2288 pthread_mutex_unlock(&HANDLE_CTX(handle)->flying_transfers_lock);
2293 usbi_backend->clear_transfer_priv(to_cancel);
2294 usbi_handle_transfer_completion(to_cancel, LIBUSB_TRANSFER_NO_DEVICE);