usbmon.rst 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375
  1. ======
  2. usbmon
  3. ======
  4. Introduction
  5. ============
  6. The name "usbmon" in lowercase refers to a facility in kernel which is
  7. used to collect traces of I/O on the USB bus. This function is analogous
  8. to a packet socket used by network monitoring tools such as tcpdump(1)
  9. or Ethereal. Similarly, it is expected that a tool such as usbdump or
  10. USBMon (with uppercase letters) is used to examine raw traces produced
  11. by usbmon.
  12. The usbmon reports requests made by peripheral-specific drivers to Host
  13. Controller Drivers (HCD). So, if HCD is buggy, the traces reported by
  14. usbmon may not correspond to bus transactions precisely. This is the same
  15. situation as with tcpdump.
  16. Two APIs are currently implemented: "text" and "binary". The binary API
  17. is available through a character device in /dev namespace and is an ABI.
  18. The text API is deprecated since 2.6.35, but available for convenience.
  19. How to use usbmon to collect raw text traces
  20. ============================================
  21. Unlike the packet socket, usbmon has an interface which provides traces
  22. in a text format. This is used for two purposes. First, it serves as a
  23. common trace exchange format for tools while more sophisticated formats
  24. are finalized. Second, humans can read it in case tools are not available.
  25. To collect a raw text trace, execute following steps.
  26. 1. Prepare
  27. ----------
  28. Mount debugfs (it has to be enabled in your kernel configuration), and
  29. load the usbmon module (if built as module). The second step is skipped
  30. if usbmon is built into the kernel::
  31. # mount -t debugfs none_debugs /sys/kernel/debug
  32. # modprobe usbmon
  33. #
  34. Verify that bus sockets are present:
  35. # ls /sys/kernel/debug/usb/usbmon
  36. 0s 0u 1s 1t 1u 2s 2t 2u 3s 3t 3u 4s 4t 4u
  37. #
  38. Now you can choose to either use the socket '0u' (to capture packets on all
  39. buses), and skip to step #3, or find the bus used by your device with step #2.
  40. This allows to filter away annoying devices that talk continuously.
  41. 2. Find which bus connects to the desired device
  42. ------------------------------------------------
  43. Run "cat /sys/kernel/debug/usb/devices", and find the T-line which corresponds
  44. to the device. Usually you do it by looking for the vendor string. If you have
  45. many similar devices, unplug one and compare the two
  46. /sys/kernel/debug/usb/devices outputs. The T-line will have a bus number.
  47. Example::
  48. T: Bus=03 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#= 2 Spd=12 MxCh= 0
  49. D: Ver= 1.10 Cls=00(>ifc ) Sub=00 Prot=00 MxPS= 8 #Cfgs= 1
  50. P: Vendor=0557 ProdID=2004 Rev= 1.00
  51. S: Manufacturer=ATEN
  52. S: Product=UC100KM V2.00
  53. "Bus=03" means it's bus 3. Alternatively, you can look at the output from
  54. "lsusb" and get the bus number from the appropriate line. Example:
  55. Bus 003 Device 002: ID 0557:2004 ATEN UC100KM V2.00
  56. 3. Start 'cat'
  57. --------------
  58. ::
  59. # cat /sys/kernel/debug/usb/usbmon/3u > /tmp/1.mon.out
  60. to listen on a single bus, otherwise, to listen on all buses, type::
  61. # cat /sys/kernel/debug/usb/usbmon/0u > /tmp/1.mon.out
  62. This process will read until it is killed. Naturally, the output can be
  63. redirected to a desirable location. This is preferred, because it is going
  64. to be quite long.
  65. 4. Perform the desired operation on the USB bus
  66. -----------------------------------------------
  67. This is where you do something that creates the traffic: plug in a flash key,
  68. copy files, control a webcam, etc.
  69. 5. Kill cat
  70. -----------
  71. Usually it's done with a keyboard interrupt (Control-C).
  72. At this point the output file (/tmp/1.mon.out in this example) can be saved,
  73. sent by e-mail, or inspected with a text editor. In the last case make sure
  74. that the file size is not excessive for your favourite editor.
  75. Raw text data format
  76. ====================
  77. Two formats are supported currently: the original, or '1t' format, and
  78. the '1u' format. The '1t' format is deprecated in kernel 2.6.21. The '1u'
  79. format adds a few fields, such as ISO frame descriptors, interval, etc.
  80. It produces slightly longer lines, but otherwise is a perfect superset
  81. of '1t' format.
  82. If it is desired to recognize one from the other in a program, look at the
  83. "address" word (see below), where '1u' format adds a bus number. If 2 colons
  84. are present, it's the '1t' format, otherwise '1u'.
  85. Any text format data consists of a stream of events, such as URB submission,
  86. URB callback, submission error. Every event is a text line, which consists
  87. of whitespace separated words. The number or position of words may depend
  88. on the event type, but there is a set of words, common for all types.
  89. Here is the list of words, from left to right:
  90. - URB Tag. This is used to identify URBs, and is normally an in-kernel address
  91. of the URB structure in hexadecimal, but can be a sequence number or any
  92. other unique string, within reason.
  93. - Timestamp in microseconds, a decimal number. The timestamp's resolution
  94. depends on available clock, and so it can be much worse than a microsecond
  95. (if the implementation uses jiffies, for example).
  96. - Event Type. This type refers to the format of the event, not URB type.
  97. Available types are: S - submission, C - callback, E - submission error.
  98. - "Address" word (formerly a "pipe"). It consists of four fields, separated by
  99. colons: URB type and direction, Bus number, Device address, Endpoint number.
  100. Type and direction are encoded with two bytes in the following manner:
  101. == == =============================
  102. Ci Co Control input and output
  103. Zi Zo Isochronous input and output
  104. Ii Io Interrupt input and output
  105. Bi Bo Bulk input and output
  106. == == =============================
  107. Bus number, Device address, and Endpoint are decimal numbers, but they may
  108. have leading zeros, for the sake of human readers.
  109. - URB Status word. This is either a letter, or several numbers separated
  110. by colons: URB status, interval, start frame, and error count. Unlike the
  111. "address" word, all fields save the status are optional. Interval is printed
  112. only for interrupt and isochronous URBs. Start frame is printed only for
  113. isochronous URBs. Error count is printed only for isochronous callback
  114. events.
  115. The status field is a decimal number, sometimes negative, which represents
  116. a "status" field of the URB. This field makes no sense for submissions, but
  117. is present anyway to help scripts with parsing. When an error occurs, the
  118. field contains the error code.
  119. In case of a submission of a Control packet, this field contains a Setup Tag
  120. instead of an group of numbers. It is easy to tell whether the Setup Tag is
  121. present because it is never a number. Thus if scripts find a set of numbers
  122. in this word, they proceed to read Data Length (except for isochronous URBs).
  123. If they find something else, like a letter, they read the setup packet before
  124. reading the Data Length or isochronous descriptors.
  125. - Setup packet, if present, consists of 5 words: one of each for bmRequestType,
  126. bRequest, wValue, wIndex, wLength, as specified by the USB Specification 2.0.
  127. These words are safe to decode if Setup Tag was 's'. Otherwise, the setup
  128. packet was present, but not captured, and the fields contain filler.
  129. - Number of isochronous frame descriptors and descriptors themselves.
  130. If an Isochronous transfer event has a set of descriptors, a total number
  131. of them in an URB is printed first, then a word per descriptor, up to a
  132. total of 5. The word consists of 3 colon-separated decimal numbers for
  133. status, offset, and length respectively. For submissions, initial length
  134. is reported. For callbacks, actual length is reported.
  135. - Data Length. For submissions, this is the requested length. For callbacks,
  136. this is the actual length.
  137. - Data tag. The usbmon may not always capture data, even if length is nonzero.
  138. The data words are present only if this tag is '='.
  139. - Data words follow, in big endian hexadecimal format. Notice that they are
  140. not machine words, but really just a byte stream split into words to make
  141. it easier to read. Thus, the last word may contain from one to four bytes.
  142. The length of collected data is limited and can be less than the data length
  143. reported in the Data Length word. In the case of an Isochronous input (Zi)
  144. completion where the received data is sparse in the buffer, the length of
  145. the collected data can be greater than the Data Length value (because Data
  146. Length counts only the bytes that were received whereas the Data words
  147. contain the entire transfer buffer).
  148. Examples:
  149. An input control transfer to get a port status::
  150. d5ea89a0 3575914555 S Ci:1:001:0 s a3 00 0000 0003 0004 4 <
  151. d5ea89a0 3575914560 C Ci:1:001:0 0 4 = 01050000
  152. An output bulk transfer to send a SCSI command 0x28 (READ_10) in a 31-byte
  153. Bulk wrapper to a storage device at address 5::
  154. dd65f0e8 4128379752 S Bo:1:005:2 -115 31 = 55534243 ad000000 00800000 80010a28 20000000 20000040 00000000 000000
  155. dd65f0e8 4128379808 C Bo:1:005:2 0 31 >
  156. Raw binary format and API
  157. =========================
  158. The overall architecture of the API is about the same as the one above,
  159. only the events are delivered in binary format. Each event is sent in
  160. the following structure (its name is made up, so that we can refer to it)::
  161. struct usbmon_packet {
  162. u64 id; /* 0: URB ID - from submission to callback */
  163. unsigned char type; /* 8: Same as text; extensible. */
  164. unsigned char xfer_type; /* ISO (0), Intr, Control, Bulk (3) */
  165. unsigned char epnum; /* Endpoint number and transfer direction */
  166. unsigned char devnum; /* Device address */
  167. u16 busnum; /* 12: Bus number */
  168. char flag_setup; /* 14: Same as text */
  169. char flag_data; /* 15: Same as text; Binary zero is OK. */
  170. s64 ts_sec; /* 16: gettimeofday */
  171. s32 ts_usec; /* 24: gettimeofday */
  172. int status; /* 28: */
  173. unsigned int length; /* 32: Length of data (submitted or actual) */
  174. unsigned int len_cap; /* 36: Delivered length */
  175. union { /* 40: */
  176. unsigned char setup[SETUP_LEN]; /* Only for Control S-type */
  177. struct iso_rec { /* Only for ISO */
  178. int error_count;
  179. int numdesc;
  180. } iso;
  181. } s;
  182. int interval; /* 48: Only for Interrupt and ISO */
  183. int start_frame; /* 52: For ISO */
  184. unsigned int xfer_flags; /* 56: copy of URB's transfer_flags */
  185. unsigned int ndesc; /* 60: Actual number of ISO descriptors */
  186. }; /* 64 total length */
  187. These events can be received from a character device by reading with read(2),
  188. with an ioctl(2), or by accessing the buffer with mmap. However, read(2)
  189. only returns first 48 bytes for compatibility reasons.
  190. The character device is usually called /dev/usbmonN, where N is the USB bus
  191. number. Number zero (/dev/usbmon0) is special and means "all buses".
  192. Note that specific naming policy is set by your Linux distribution.
  193. If you create /dev/usbmon0 by hand, make sure that it is owned by root
  194. and has mode 0600. Otherwise, unprivileged users will be able to snoop
  195. keyboard traffic.
  196. The following ioctl calls are available, with MON_IOC_MAGIC 0x92:
  197. MON_IOCQ_URB_LEN, defined as _IO(MON_IOC_MAGIC, 1)
  198. This call returns the length of data in the next event. Note that majority of
  199. events contain no data, so if this call returns zero, it does not mean that
  200. no events are available.
  201. MON_IOCG_STATS, defined as _IOR(MON_IOC_MAGIC, 3, struct mon_bin_stats)
  202. The argument is a pointer to the following structure::
  203. struct mon_bin_stats {
  204. u32 queued;
  205. u32 dropped;
  206. };
  207. The member "queued" refers to the number of events currently queued in the
  208. buffer (and not to the number of events processed since the last reset).
  209. The member "dropped" is the number of events lost since the last call
  210. to MON_IOCG_STATS.
  211. MON_IOCT_RING_SIZE, defined as _IO(MON_IOC_MAGIC, 4)
  212. This call sets the buffer size. The argument is the size in bytes.
  213. The size may be rounded down to the next chunk (or page). If the requested
  214. size is out of [unspecified] bounds for this kernel, the call fails with
  215. -EINVAL.
  216. MON_IOCQ_RING_SIZE, defined as _IO(MON_IOC_MAGIC, 5)
  217. This call returns the current size of the buffer in bytes.
  218. MON_IOCX_GET, defined as _IOW(MON_IOC_MAGIC, 6, struct mon_get_arg)
  219. MON_IOCX_GETX, defined as _IOW(MON_IOC_MAGIC, 10, struct mon_get_arg)
  220. These calls wait for events to arrive if none were in the kernel buffer,
  221. then return the first event. The argument is a pointer to the following
  222. structure::
  223. struct mon_get_arg {
  224. struct usbmon_packet *hdr;
  225. void *data;
  226. size_t alloc; /* Length of data (can be zero) */
  227. };
  228. Before the call, hdr, data, and alloc should be filled. Upon return, the area
  229. pointed by hdr contains the next event structure, and the data buffer contains
  230. the data, if any. The event is removed from the kernel buffer.
  231. The MON_IOCX_GET copies 48 bytes to hdr area, MON_IOCX_GETX copies 64 bytes.
  232. MON_IOCX_MFETCH, defined as _IOWR(MON_IOC_MAGIC, 7, struct mon_mfetch_arg)
  233. This ioctl is primarily used when the application accesses the buffer
  234. with mmap(2). Its argument is a pointer to the following structure::
  235. struct mon_mfetch_arg {
  236. uint32_t *offvec; /* Vector of events fetched */
  237. uint32_t nfetch; /* Number of events to fetch (out: fetched) */
  238. uint32_t nflush; /* Number of events to flush */
  239. };
  240. The ioctl operates in 3 stages.
  241. First, it removes and discards up to nflush events from the kernel buffer.
  242. The actual number of events discarded is returned in nflush.
  243. Second, it waits for an event to be present in the buffer, unless the pseudo-
  244. device is open with O_NONBLOCK.
  245. Third, it extracts up to nfetch offsets into the mmap buffer, and stores
  246. them into the offvec. The actual number of event offsets is stored into
  247. the nfetch.
  248. MON_IOCH_MFLUSH, defined as _IO(MON_IOC_MAGIC, 8)
  249. This call removes a number of events from the kernel buffer. Its argument
  250. is the number of events to remove. If the buffer contains fewer events
  251. than requested, all events present are removed, and no error is reported.
  252. This works when no events are available too.
  253. FIONBIO
  254. The ioctl FIONBIO may be implemented in the future, if there's a need.
  255. In addition to ioctl(2) and read(2), the special file of binary API can
  256. be polled with select(2) and poll(2). But lseek(2) does not work.
  257. * Memory-mapped access of the kernel buffer for the binary API
  258. The basic idea is simple:
  259. To prepare, map the buffer by getting the current size, then using mmap(2).
  260. Then, execute a loop similar to the one written in pseudo-code below::
  261. struct mon_mfetch_arg fetch;
  262. struct usbmon_packet *hdr;
  263. int nflush = 0;
  264. for (;;) {
  265. fetch.offvec = vec; // Has N 32-bit words
  266. fetch.nfetch = N; // Or less than N
  267. fetch.nflush = nflush;
  268. ioctl(fd, MON_IOCX_MFETCH, &fetch); // Process errors, too
  269. nflush = fetch.nfetch; // This many packets to flush when done
  270. for (i = 0; i < nflush; i++) {
  271. hdr = (struct ubsmon_packet *) &mmap_area[vec[i]];
  272. if (hdr->type == '@') // Filler packet
  273. continue;
  274. caddr_t data = &mmap_area[vec[i]] + 64;
  275. process_packet(hdr, data);
  276. }
  277. }
  278. Thus, the main idea is to execute only one ioctl per N events.
  279. Although the buffer is circular, the returned headers and data do not cross
  280. the end of the buffer, so the above pseudo-code does not need any gathering.