Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1 | Revised: 2000-Dec-05. |
| 2 | Again: 2002-Jul-06 |
Alan Stern | 0fc084e | 2005-09-22 00:49:51 -0700 | [diff] [blame] | 3 | Again: 2005-Sep-19 |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 4 | |
| 5 | NOTE: |
| 6 | |
| 7 | The USB subsystem now has a substantial section in "The Linux Kernel API" |
| 8 | guide (in Documentation/DocBook), generated from the current source |
| 9 | code. This particular documentation file isn't particularly current or |
| 10 | complete; don't rely on it except for a quick overview. |
| 11 | |
| 12 | |
| 13 | 1.1. Basic concept or 'What is an URB?' |
| 14 | |
| 15 | The basic idea of the new driver is message passing, the message itself is |
| 16 | called USB Request Block, or URB for short. |
| 17 | |
| 18 | - An URB consists of all relevant information to execute any USB transaction |
| 19 | and deliver the data and status back. |
| 20 | |
| 21 | - Execution of an URB is inherently an asynchronous operation, i.e. the |
Alan Stern | 0fc084e | 2005-09-22 00:49:51 -0700 | [diff] [blame] | 22 | usb_submit_urb(urb) call returns immediately after it has successfully |
| 23 | queued the requested action. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 24 | |
| 25 | - Transfers for one URB can be canceled with usb_unlink_urb(urb) at any time. |
| 26 | |
| 27 | - Each URB has a completion handler, which is called after the action |
| 28 | has been successfully completed or canceled. The URB also contains a |
| 29 | context-pointer for passing information to the completion handler. |
| 30 | |
| 31 | - Each endpoint for a device logically supports a queue of requests. |
| 32 | You can fill that queue, so that the USB hardware can still transfer |
| 33 | data to an endpoint while your driver handles completion of another. |
| 34 | This maximizes use of USB bandwidth, and supports seamless streaming |
| 35 | of data to (or from) devices when using periodic transfer modes. |
| 36 | |
| 37 | |
| 38 | 1.2. The URB structure |
| 39 | |
| 40 | Some of the fields in an URB are: |
| 41 | |
| 42 | struct urb |
| 43 | { |
| 44 | // (IN) device and pipe specify the endpoint queue |
| 45 | struct usb_device *dev; // pointer to associated USB device |
| 46 | unsigned int pipe; // endpoint information |
| 47 | |
| 48 | unsigned int transfer_flags; // ISO_ASAP, SHORT_NOT_OK, etc. |
| 49 | |
| 50 | // (IN) all urbs need completion routines |
| 51 | void *context; // context for completion routine |
| 52 | void (*complete)(struct urb *); // pointer to completion routine |
| 53 | |
| 54 | // (OUT) status after each completion |
| 55 | int status; // returned status |
| 56 | |
| 57 | // (IN) buffer used for data transfers |
| 58 | void *transfer_buffer; // associated data buffer |
| 59 | int transfer_buffer_length; // data buffer length |
| 60 | int number_of_packets; // size of iso_frame_desc |
| 61 | |
| 62 | // (OUT) sometimes only part of CTRL/BULK/INTR transfer_buffer is used |
| 63 | int actual_length; // actual data buffer length |
| 64 | |
| 65 | // (IN) setup stage for CTRL (pass a struct usb_ctrlrequest) |
| 66 | unsigned char* setup_packet; // setup packet (control only) |
| 67 | |
| 68 | // Only for PERIODIC transfers (ISO, INTERRUPT) |
| 69 | // (IN/OUT) start_frame is set unless ISO_ASAP isn't set |
| 70 | int start_frame; // start frame |
| 71 | int interval; // polling interval |
| 72 | |
| 73 | // ISO only: packets are only "best effort"; each can have errors |
| 74 | int error_count; // number of errors |
| 75 | struct usb_iso_packet_descriptor iso_frame_desc[0]; |
| 76 | }; |
| 77 | |
| 78 | Your driver must create the "pipe" value using values from the appropriate |
| 79 | endpoint descriptor in an interface that it's claimed. |
| 80 | |
| 81 | |
| 82 | 1.3. How to get an URB? |
| 83 | |
| 84 | URBs are allocated with the following call |
| 85 | |
| 86 | struct urb *usb_alloc_urb(int isoframes, int mem_flags) |
| 87 | |
| 88 | Return value is a pointer to the allocated URB, 0 if allocation failed. |
| 89 | The parameter isoframes specifies the number of isochronous transfer frames |
| 90 | you want to schedule. For CTRL/BULK/INT, use 0. The mem_flags parameter |
| 91 | holds standard memory allocation flags, letting you control (among other |
| 92 | things) whether the underlying code may block or not. |
| 93 | |
| 94 | To free an URB, use |
| 95 | |
| 96 | void usb_free_urb(struct urb *urb) |
| 97 | |
Alan Stern | 0fc084e | 2005-09-22 00:49:51 -0700 | [diff] [blame] | 98 | You may free an urb that you've submitted, but which hasn't yet been |
| 99 | returned to you in a completion callback. It will automatically be |
| 100 | deallocated when it is no longer in use. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 101 | |
| 102 | |
| 103 | 1.4. What has to be filled in? |
| 104 | |
| 105 | Depending on the type of transaction, there are some inline functions |
| 106 | defined in <linux/usb.h> to simplify the initialization, such as |
| 107 | fill_control_urb() and fill_bulk_urb(). In general, they need the usb |
| 108 | device pointer, the pipe (usual format from usb.h), the transfer buffer, |
| 109 | the desired transfer length, the completion handler, and its context. |
| 110 | Take a look at the some existing drivers to see how they're used. |
| 111 | |
| 112 | Flags: |
| 113 | For ISO there are two startup behaviors: Specified start_frame or ASAP. |
| 114 | For ASAP set URB_ISO_ASAP in transfer_flags. |
| 115 | |
| 116 | If short packets should NOT be tolerated, set URB_SHORT_NOT_OK in |
| 117 | transfer_flags. |
| 118 | |
| 119 | |
| 120 | 1.5. How to submit an URB? |
| 121 | |
| 122 | Just call |
| 123 | |
| 124 | int usb_submit_urb(struct urb *urb, int mem_flags) |
| 125 | |
| 126 | The mem_flags parameter, such as SLAB_ATOMIC, controls memory allocation, |
| 127 | such as whether the lower levels may block when memory is tight. |
| 128 | |
| 129 | It immediately returns, either with status 0 (request queued) or some |
| 130 | error code, usually caused by the following: |
| 131 | |
| 132 | - Out of memory (-ENOMEM) |
| 133 | - Unplugged device (-ENODEV) |
| 134 | - Stalled endpoint (-EPIPE) |
| 135 | - Too many queued ISO transfers (-EAGAIN) |
| 136 | - Too many requested ISO frames (-EFBIG) |
| 137 | - Invalid INT interval (-EINVAL) |
| 138 | - More than one packet for INT (-EINVAL) |
| 139 | |
| 140 | After submission, urb->status is -EINPROGRESS; however, you should never |
| 141 | look at that value except in your completion callback. |
| 142 | |
| 143 | For isochronous endpoints, your completion handlers should (re)submit |
| 144 | URBs to the same endpoint with the ISO_ASAP flag, using multi-buffering, |
| 145 | to get seamless ISO streaming. |
| 146 | |
| 147 | |
| 148 | 1.6. How to cancel an already running URB? |
| 149 | |
Alan Stern | 0fc084e | 2005-09-22 00:49:51 -0700 | [diff] [blame] | 150 | There are two ways to cancel an URB you've submitted but which hasn't |
| 151 | been returned to your driver yet. For an asynchronous cancel, call |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 152 | |
| 153 | int usb_unlink_urb(struct urb *urb) |
| 154 | |
| 155 | It removes the urb from the internal list and frees all allocated |
Alan Stern | 0fc084e | 2005-09-22 00:49:51 -0700 | [diff] [blame] | 156 | HW descriptors. The status is changed to reflect unlinking. Note |
| 157 | that the URB will not normally have finished when usb_unlink_urb() |
| 158 | returns; you must still wait for the completion handler to be called. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 159 | |
Alan Stern | 0fc084e | 2005-09-22 00:49:51 -0700 | [diff] [blame] | 160 | To cancel an URB synchronously, call |
| 161 | |
| 162 | void usb_kill_urb(struct urb *urb) |
| 163 | |
| 164 | It does everything usb_unlink_urb does, and in addition it waits |
| 165 | until after the URB has been returned and the completion handler |
| 166 | has finished. It also marks the URB as temporarily unusable, so |
| 167 | that if the completion handler or anyone else tries to resubmit it |
| 168 | they will get a -EPERM error. Thus you can be sure that when |
| 169 | usb_kill_urb() returns, the URB is totally idle. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 170 | |
Alan Stern | da8bfb0 | 2012-03-28 16:13:28 -0400 | [diff] [blame] | 171 | There is a lifetime issue to consider. An URB may complete at any |
| 172 | time, and the completion handler may free the URB. If this happens |
| 173 | while usb_unlink_urb or usb_kill_urb is running, it will cause a |
| 174 | memory-access violation. The driver is responsible for avoiding this, |
| 175 | which often means some sort of lock will be needed to prevent the URB |
| 176 | from being deallocated while it is still in use. |
| 177 | |
| 178 | On the other hand, since usb_unlink_urb may end up calling the |
| 179 | completion handler, the handler must not take any lock that is held |
| 180 | when usb_unlink_urb is invoked. The general solution to this problem |
| 181 | is to increment the URB's reference count while holding the lock, then |
| 182 | drop the lock and call usb_unlink_urb or usb_kill_urb, and then |
| 183 | decrement the URB's reference count. You increment the reference |
| 184 | count by calling |
| 185 | |
| 186 | struct urb *usb_get_urb(struct urb *urb) |
| 187 | |
| 188 | (ignore the return value; it is the same as the argument) and |
| 189 | decrement the reference count by calling usb_free_urb. Of course, |
| 190 | none of this is necessary if there's no danger of the URB being freed |
| 191 | by the completion handler. |
| 192 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 193 | |
| 194 | 1.7. What about the completion handler? |
| 195 | |
| 196 | The handler is of the following type: |
| 197 | |
Alan Stern | 0fc084e | 2005-09-22 00:49:51 -0700 | [diff] [blame] | 198 | typedef void (*usb_complete_t)(struct urb *, struct pt_regs *) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 199 | |
Alan Stern | 0fc084e | 2005-09-22 00:49:51 -0700 | [diff] [blame] | 200 | I.e., it gets the URB that caused the completion call, plus the |
| 201 | register values at the time of the corresponding interrupt (if any). |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 202 | In the completion handler, you should have a look at urb->status to |
| 203 | detect any USB errors. Since the context parameter is included in the URB, |
| 204 | you can pass information to the completion handler. |
| 205 | |
| 206 | Note that even when an error (or unlink) is reported, data may have been |
| 207 | transferred. That's because USB transfers are packetized; it might take |
| 208 | sixteen packets to transfer your 1KByte buffer, and ten of them might |
Matt LaPlante | 53cb472 | 2006-10-03 22:55:17 +0200 | [diff] [blame] | 209 | have transferred successfully before the completion was called. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 210 | |
| 211 | |
| 212 | NOTE: ***** WARNING ***** |
Alan Stern | 0fc084e | 2005-09-22 00:49:51 -0700 | [diff] [blame] | 213 | NEVER SLEEP IN A COMPLETION HANDLER. These are normally called |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 214 | during hardware interrupt processing. If you can, defer substantial |
| 215 | work to a tasklet (bottom half) to keep system latencies low. You'll |
| 216 | probably need to use spinlocks to protect data structures you manipulate |
| 217 | in completion handlers. |
| 218 | |
| 219 | |
| 220 | 1.8. How to do isochronous (ISO) transfers? |
| 221 | |
| 222 | For ISO transfers you have to fill a usb_iso_packet_descriptor structure, |
| 223 | allocated at the end of the URB by usb_alloc_urb(n,mem_flags), for each |
| 224 | packet you want to schedule. You also have to set urb->interval to say |
| 225 | how often to make transfers; it's often one per frame (which is once |
| 226 | every microframe for highspeed devices). The actual interval used will |
| 227 | be a power of two that's no bigger than what you specify. |
| 228 | |
| 229 | The usb_submit_urb() call modifies urb->interval to the implemented interval |
| 230 | value that is less than or equal to the requested interval value. If |
| 231 | ISO_ASAP scheduling is used, urb->start_frame is also updated. |
| 232 | |
| 233 | For each entry you have to specify the data offset for this frame (base is |
| 234 | transfer_buffer), and the length you want to write/expect to read. |
| 235 | After completion, actual_length contains the actual transferred length and |
| 236 | status contains the resulting status for the ISO transfer for this frame. |
| 237 | It is allowed to specify a varying length from frame to frame (e.g. for |
| 238 | audio synchronisation/adaptive transfer rates). You can also use the length |
| 239 | 0 to omit one or more frames (striping). |
| 240 | |
| 241 | For scheduling you can choose your own start frame or ISO_ASAP. As explained |
| 242 | earlier, if you always keep at least one URB queued and your completion |
| 243 | keeps (re)submitting a later URB, you'll get smooth ISO streaming (if usb |
| 244 | bandwidth utilization allows). |
| 245 | |
| 246 | If you specify your own start frame, make sure it's several frames in advance |
| 247 | of the current frame. You might want this model if you're synchronizing |
| 248 | ISO data with some other event stream. |
| 249 | |
| 250 | |
| 251 | 1.9. How to start interrupt (INT) transfers? |
| 252 | |
| 253 | Interrupt transfers, like isochronous transfers, are periodic, and happen |
| 254 | in intervals that are powers of two (1, 2, 4 etc) units. Units are frames |
| 255 | for full and low speed devices, and microframes for high speed ones. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 256 | The usb_submit_urb() call modifies urb->interval to the implemented interval |
| 257 | value that is less than or equal to the requested interval value. |
Alan Stern | 0fc084e | 2005-09-22 00:49:51 -0700 | [diff] [blame] | 258 | |
| 259 | In Linux 2.6, unlike earlier versions, interrupt URBs are not automagically |
| 260 | restarted when they complete. They end when the completion handler is |
| 261 | called, just like other URBs. If you want an interrupt URB to be restarted, |
| 262 | your completion handler must resubmit it. |