1* Introduction
   3The name "usbmon" in lowercase refers to a facility in kernel which is
   4used to collect traces of I/O on the USB bus. This function is analogous
   5to a packet socket used by network monitoring tools such as tcpdump(1)
   6or Ethereal. Similarly, it is expected that a tool such as usbdump or
   7USBMon (with uppercase letters) is used to examine raw traces produced
   8by usbmon.
  10The usbmon reports requests made by peripheral-specific drivers to Host
  11Controller Drivers (HCD). So, if HCD is buggy, the traces reported by
  12usbmon may not correspond to bus transactions precisely. This is the same
  13situation as with tcpdump.
  15* How to use usbmon to collect raw text traces
  17Unlike the packet socket, usbmon has an interface which provides traces
  18in a text format. This is used for two purposes. First, it serves as a
  19common trace exchange format for tools while more sophisticated formats
  20are finalized. Second, humans can read it in case tools are not available.
  22To collect a raw text trace, execute following steps.
  241. Prepare
  26Mount debugfs (it has to be enabled in your kernel configuration), and
  27load the usbmon module (if built as module). The second step is skipped
  28if usbmon is built into the kernel.
  30# mount -t debugfs none_debugs /sys/kernel/debug
  31# modprobe usbmon
  34Verify that bus sockets are present.
  36# ls /sys/kernel/debug/usbmon
  370s  0t  0u  1s  1t  1u  2s  2t  2u  3s  3t  3u  4s  4t  4u
  40Now you can choose to either use the sockets numbered '0' (to capture packets on
  41all buses), and skip to step #3, or find the bus used by your device with step #2.
  432. Find which bus connects to the desired device
  45Run "cat /proc/bus/usb/devices", and find the T-line which corresponds to
  46the device. Usually you do it by looking for the vendor string. If you have
  47many similar devices, unplug one and compare two /proc/bus/usb/devices outputs.
  48The T-line will have a bus number. Example:
  50T:  Bus=03 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#=  2 Spd=12  MxCh= 0
  51D:  Ver= 1.10 Cls=00(>ifc ) Sub=00 Prot=00 MxPS= 8 #Cfgs=  1
  52P:  Vendor=0557 ProdID=2004 Rev= 1.00
  53S:  Manufacturer=ATEN
  54S:  Product=UC100KM V2.00
  56Bus=03 means it's bus 3.
  583. Start 'cat'
  60# cat /sys/kernel/debug/usbmon/3u > /tmp/1.mon.out
  62to listen on a single bus, otherwise, to listen on all buses, type:
  64# cat /sys/kernel/debug/usbmon/0u > /tmp/1.mon.out
  66This process will be reading until killed. Naturally, the output can be
  67redirected to a desirable location. This is preferred, because it is going
  68to be quite long.
  704. Perform the desired operation on the USB bus
  72This is where you do something that creates the traffic: plug in a flash key,
  73copy files, control a webcam, etc.
  755. Kill cat
  77Usually it's done with a keyboard interrupt (Control-C).
  79At this point the output file (/tmp/1.mon.out in this example) can be saved,
  80sent by e-mail, or inspected with a text editor. In the last case make sure
  81that the file size is not excessive for your favourite editor.
  83* Raw text data format
  85Two formats are supported currently: the original, or '1t' format, and
  86the '1u' format. The '1t' format is deprecated in kernel 2.6.21. The '1u'
  87format adds a few fields, such as ISO frame descriptors, interval, etc.
  88It produces slightly longer lines, but otherwise is a perfect superset
  89of '1t' format.
  91If it is desired to recognize one from the other in a program, look at the
  92"address" word (see below), where '1u' format adds a bus number. If 2 colons
  93are present, it's the '1t' format, otherwise '1u'.
  95Any text format data consists of a stream of events, such as URB submission,
  96URB callback, submission error. Every event is a text line, which consists
  97of whitespace separated words. The number or position of words may depend
  98on the event type, but there is a set of words, common for all types.
 100Here is the list of words, from left to right:
 102- URB Tag. This is used to identify URBs is normally a kernel mode address
 103 of the URB structure in hexadecimal.
 105- Timestamp in microseconds, a decimal number. The timestamp's resolution
 106  depends on available clock, and so it can be much worse than a microsecond
 107  (if the implementation uses jiffies, for example).
 109- Event Type. This type refers to the format of the event, not URB type.
 110  Available types are: S - submission, C - callback, E - submission error.
 112- "Address" word (formerly a "pipe"). It consists of four fields, separated by
 113  colons: URB type and direction, Bus number, Device address, Endpoint number.
 114  Type and direction are encoded with two bytes in the following manner:
 115    Ci Co   Control input and output
 116    Zi Zo   Isochronous input and output
 117    Ii Io   Interrupt input and output
 118    Bi Bo   Bulk input and output
 119  Bus number, Device address, and Endpoint are decimal numbers, but they may
 120  have leading zeros, for the sake of human readers.
 122- URB Status word. This is either a letter, or several numbers separated
 123  by colons: URB status, interval, start frame, and error count. Unlike the
 124  "address" word, all fields save the status are optional. Interval is printed
 125  only for interrupt and isochronous URBs. Start frame is printed only for
 126  isochronous URBs. Error count is printed only for isochronous callback
 127  events.
 129  The status field is a decimal number, sometimes negative, which represents
 130  a "status" field of the URB. This field makes no sense for submissions, but
 131  is present anyway to help scripts with parsing. When an error occurs, the
 132  field contains the error code.
 134  In case of a submission of a Control packet, this field contains a Setup Tag
 135  instead of an group of numbers. It is easy to tell whether the Setup Tag is
 136  present because it is never a number. Thus if scripts find a set of numbers
 137  in this word, they proceed to read Data Length (except for isochronous URBs).
 138  If they find something else, like a letter, they read the setup packet before
 139  reading the Data Length or isochronous descriptors.
 141- Setup packet, if present, consists of 5 words: one of each for bmRequestType,
 142  bRequest, wValue, wIndex, wLength, as specified by the USB Specification 2.0.
 143  These words are safe to decode if Setup Tag was 's'. Otherwise, the setup
 144  packet was present, but not captured, and the fields contain filler.
 146- Number of isochronous frame descriptors and descriptors themselves.
 147  If an Isochronous transfer event has a set of descriptors, a total number
 148  of them in an URB is printed first, then a word per descriptor, up to a
 149  total of 5. The word consists of 3 colon-separated decimal numbers for
 150  status, offset, and length respectively. For submissions, initial length
 151  is reported. For callbacks, actual length is reported.
 153- Data Length. For submissions, this is the requested length. For callbacks,
 154  this is the actual length.
 156- Data tag. The usbmon may not always capture data, even if length is nonzero.
 157  The data words are present only if this tag is '='.
 159- Data words follow, in big endian hexadecimal format. Notice that they are
 160  not machine words, but really just a byte stream split into words to make
 161  it easier to read. Thus, the last word may contain from one to four bytes.
 162  The length of collected data is limited and can be less than the data length
 163  report in Data Length word.
 165Here is an example of code to read the data stream in a well known programming
 168class ParsedLine {
 169        int data_len;           /* Available length of data */
 170        byte data[];
 172        void parseData(StringTokenizer st) {
 173                int availwords = st.countTokens();
 174                data = new byte[availwords * 4];
 175                data_len = 0;
 176                while (st.hasMoreTokens()) {
 177                        String data_str = st.nextToken();
 178                        int len = data_str.length() / 2;
 179                        int i;
 180                        int b;  // byte is signed, apparently?! XXX
 181                        for (i = 0; i < len; i++) {
 182                                // data[data_len] = Byte.parseByte(
 183                                //     data_str.substring(i*2, i*2 + 2),
 184                                //     16);
 185                                b = Integer.parseInt(
 186                                     data_str.substring(i*2, i*2 + 2),
 187                                     16);
 188                                if (b >= 128)
 189                                        b *= -1;
 190                                data[data_len] = (byte) b;
 191                                data_len++;
 192                        }
 193                }
 194        }
 199An input control transfer to get a port status.
 201d5ea89a0 3575914555 S Ci:1:001:0 s a3 00 0000 0003 0004 4 <
 202d5ea89a0 3575914560 C Ci:1:001:0 0 4 = 01050000
 204An output bulk transfer to send a SCSI command 0x5E in a 31-byte Bulk wrapper
 205to a storage device at address 5:
 207dd65f0e8 4128379752 S Bo:1:005:2 -115 31 = 55534243 5e000000 00000000 00000600 00000000 00000000 00000000 000000
 208dd65f0e8 4128379808 C Bo:1:005:2 0 31 >
 210* Raw binary format and API
 212The overall architecture of the API is about the same as the one above,
 213only the events are delivered in binary format. Each event is sent in
 214the following structure (its name is made up, so that we can refer to it):
 216struct usbmon_packet {
 217        u64 id;                 /*  0: URB ID - from submission to callback */
 218        unsigned char type;     /*  8: Same as text; extensible. */
 219        unsigned char xfer_type; /*    ISO (0), Intr, Control, Bulk (3) */
 220        unsigned char epnum;    /*     Endpoint number and transfer direction */
 221        unsigned char devnum;   /*     Device address */
 222        u16 busnum;             /* 12: Bus number */
 223        char flag_setup;        /* 14: Same as text */
 224        char flag_data;         /* 15: Same as text; Binary zero is OK. */
 225        s64 ts_sec;             /* 16: gettimeofday */
 226        s32 ts_usec;            /* 24: gettimeofday */
 227        int status;             /* 28: */
 228        unsigned int length;    /* 32: Length of data (submitted or actual) */
 229        unsigned int len_cap;   /* 36: Delivered length */
 230        unsigned char setup[8]; /* 40: Only for Control 'S' */
 231};                              /* 48 bytes total */
 233These events can be received from a character device by reading with read(2),
 234with an ioctl(2), or by accessing the buffer with mmap.
 236The character device is usually called /dev/usbmonN, where N is the USB bus
 237number. Number zero (/dev/usbmon0) is special and means "all buses".
 238However, this feature is not implemented yet. Note that specific naming
 239policy is set by your Linux distribution.
 241If you create /dev/usbmon0 by hand, make sure that it is owned by root
 242and has mode 0600. Otherwise, unpriviledged users will be able to snoop
 243keyboard traffic.
 245The following ioctl calls are available, with MON_IOC_MAGIC 0x92:
 247 MON_IOCQ_URB_LEN, defined as _IO(MON_IOC_MAGIC, 1)
 249This call returns the length of data in the next event. Note that majority of
 250events contain no data, so if this call returns zero, it does not mean that
 251no events are available.
 253 MON_IOCG_STATS, defined as _IOR(MON_IOC_MAGIC, 3, struct mon_bin_stats)
 255The argument is a pointer to the following structure:
 257struct mon_bin_stats {
 258        u32 queued;
 259        u32 dropped;
 262The member "queued" refers to the number of events currently queued in the
 263buffer (and not to the number of events processed since the last reset).
 265The member "dropped" is the number of events lost since the last call
 268 MON_IOCT_RING_SIZE, defined as _IO(MON_IOC_MAGIC, 4)
 270This call sets the buffer size. The argument is the size in bytes.
 271The size may be rounded down to the next chunk (or page). If the requested
 272size is out of [unspecified] bounds for this kernel, the call fails with
 275 MON_IOCQ_RING_SIZE, defined as _IO(MON_IOC_MAGIC, 5)
 277This call returns the current size of the buffer in bytes.
 279 MON_IOCX_GET, defined as _IOW(MON_IOC_MAGIC, 6, struct mon_get_arg)
 281This call waits for events to arrive if none were in the kernel buffer,
 282then returns the first event. Its argument is a pointer to the following
 285struct mon_get_arg {
 286        struct usbmon_packet *hdr;
 287        void *data;
 288        size_t alloc;           /* Length of data (can be zero) */
 291Before the call, hdr, data, and alloc should be filled. Upon return, the area
 292pointed by hdr contains the next event structure, and the data buffer contains
 293the data, if any. The event is removed from the kernel buffer.
 295 MON_IOCX_MFETCH, defined as _IOWR(MON_IOC_MAGIC, 7, struct mon_mfetch_arg)
 297This ioctl is primarily used when the application accesses the buffer
 298with mmap(2). Its argument is a pointer to the following structure:
 300struct mon_mfetch_arg {
 301        uint32_t *offvec;       /* Vector of events fetched */
 302        uint32_t nfetch;        /* Number of events to fetch (out: fetched) */
 303        uint32_t nflush;        /* Number of events to flush */
 306The ioctl operates in 3 stages.
 308First, it removes and discards up to nflush events from the kernel buffer.
 309The actual number of events discarded is returned in nflush.
 311Second, it waits for an event to be present in the buffer, unless the pseudo-
 312device is open with O_NONBLOCK.
 314Third, it extracts up to nfetch offsets into the mmap buffer, and stores
 315them into the offvec. The actual number of event offsets is stored into
 316the nfetch.
 318 MON_IOCH_MFLUSH, defined as _IO(MON_IOC_MAGIC, 8)
 320This call removes a number of events from the kernel buffer. Its argument
 321is the number of events to remove. If the buffer contains fewer events
 322than requested, all events present are removed, and no error is reported.
 323This works when no events are available too.
 327The ioctl FIONBIO may be implemented in the future, if there's a need.
 329In addition to ioctl(2) and read(2), the special file of binary API can
 330be polled with select(2) and poll(2). But lseek(2) does not work.
 332* Memory-mapped access of the kernel buffer for the binary API
 334The basic idea is simple:
 336To prepare, map the buffer by getting the current size, then using mmap(2).
 337Then, execute a loop similar to the one written in pseudo-code below:
 339   struct mon_mfetch_arg fetch;
 340   struct usbmon_packet *hdr;
 341   int nflush = 0;
 342   for (;;) {
 343      fetch.offvec = vec; // Has N 32-bit words
 344      fetch.nfetch = N;   // Or less than N
 345      fetch.nflush = nflush;
 346      ioctl(fd, MON_IOCX_MFETCH, &fetch);   // Process errors, too
 347      nflush = fetch.nfetch;       // This many packets to flush when done
 348      for (i = 0; i < nflush; i++) {
 349         hdr = (struct ubsmon_packet *) &mmap_area[vec[i]];
 350         if (hdr->type == '@')     // Filler packet
 351            continue;
 352         caddr_t data = &mmap_area[vec[i]] + 64;
 353         process_packet(hdr, data);
 354      }
 355   }
 357Thus, the main idea is to execute only one ioctl per N events.
 359Although the buffer is circular, the returned headers and data do not cross
 360the end of the buffer, so the above pseudo-code does not need any gathering.
 361 kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.