I leave it to those more familiar with the protocol to explain this
better or in more detail and, if possible, to come up with a better way
of doing this.
Some classes, most notably Mass Storage, use STALL (instead of ZLP) to
prematurely end transfer. Finish reassembly when dissecting STALL. For
IN transfers ending reassembly is pretty straightforward and resembles
ZLP packet. For OUT transfers the reassembled URB is opposite direction
than the STALL handshake itself and last data packet may not be part of
reassembled URB. The last OUT data packet is part of reassembled URB
only if packet was acknowledged with NYET before endpoint was STALLed.
According to the Bluetooth Core Specification v5.3, Volume 2, Part C,
Section 5.1, the Name_Offset parameter of the LMP_NAME_RES PDU should be at
offset 2 in the payload and Name_Length should be at offset 3.
Store the field filter strings in a wmem_map pointing to the
field flags for each string. This allows specifying multiple
filter options (-j or -J) on the command line, including some
of both.
Fix#17470
The kernel unfortunately doesn't indicate which payload type
it is. In particular, it might be an Ethernet packet or an IP one,
depending on how the SKB has been generated.
We work around this issue by guessing if the packet contains the
EtherType at the right offset to be an Ethernet packet, and decode
accordingly the payload.
If MSP_FLAGS_REASSEMBLE_ENTIRE_SEGMENT is set (because the dissector
asked for one more segment) but the new segment didn't result in
a completed reassembly (because it overlapped and didn't add new
data), don't clear the REASSEMBLE_ENTIRE_SEGMENT flag.
Related to #18411
If we have payload on TCP with a single octet that is not printable ASCII,
just reject it rather than waiting for the next CRLF line end (and
marking it as Continuation Data then). It is more likely to be a TCP Keep
Alive at the beginning of a capture file or connection.
At best, this change means that a correct SIP request or response is
interpreted rather than marking as Continuation Data. At worst,
Continuation Data has one less non printable octet at the beginning.
Fix#18411.
When device responds with STALL, the host will clear the halt using
ClearFeature(ENDPOINT_HALT) request. The request always results in data
toggle being reinitialized to DATA0. Because USBLL dissector does not
track all control transfers, it is unaware of the expected data toggle
change and thus would treat next DATA0 packet as retransmission if the
last data packet before STALL was DATA0.
USB transfer never spans across STALL, i.e. data packet after STALL
cannot be retransmission nor continuation of any earlier transfer.
Avoid continuing reassembly after a STALL by clearing active transfer
information from endpoint info on every STALL handshake.
Move the top-level user-guide.adoc and developer-guide.adoc to their
respective source directores. This is in preparation for a future
toolchain revamp.
Move the wsug_graphics directory to wsug_src/images and wsdg_graphics
directory to wsdg_src/images. Copy common_graphics/* to the each images
directory and remove common_graphics. We only have five admonition
graphics; duplicating them lets us remove some build config overhead.
Rename wsluarm.adoc to wsdg_src/wsdg_lua_support.adoc.
Remove a dummy file.
Tested visually and by enabling `--failure-level=WARN`.
Add a drop-down combobox for UATs, including User DLTs, that
have a choice of dissectors. Make the combobox editable, which
will provide suggestions, and pass things through to the existing
UAT validation for dissectors. (It's a very long list, especially
with 1717 entries, including 530 just from various BT GATT UUIDs,
so being able to still type it in seems useful.)
Dissectors are not protocols. Rename the UAT field from PROTO to
DISSECTOR where used. Update the column names and long descriptions
to use dissector instead of protocol in dissectors that used this.
There may at some point be UATs that want protocols instead of
dissectors, but that's not what the current behavior does and
none of the current dissectors that use the existing types want.
Update the documentation to use "dissector" instead of "protocol."
Put the names of the actual current three Ethernet dissectors.
Clarify that the "ip" dissector actually tries IPv4 and IPv6,
instead of just IPv4.
UAT entries are backwards and forwards compatible with versions
without this change.
Fix#18836.
wscbor_skip_next_item should not return to the beginning offset
if there are errors, because this makes wscbor_skip_if_errors
do the opposite of what it claims. In the case where the errors
involve having far too many items in a list, this can cause
memory exhaustion or infinite loops.
Fix#18782. Fix#18840.
Mass Storage Bulk Only Transport prohibits ending data transfers with
zero length packet. This is generally not problematic when capturing at
OS URB level, but it does raise issues when capturing at USB Link Layer.
USBLL dissector has no idea where the transfer ends. It will concatenate
SCSI Data IN with CSW and SCSI Data OUT with next CBW whenever SCSI Data
length is multiple of bulk endpoint max packet size (virtually all Read
and Write commands because most common sector sizes are 512 and 4096).
CBW and CSW always end transfer reassembly because they must start at
packet boundary and their size is not equal to bulk max packet size.
Merging Data IN with CSW poses no problems at all. The only end user
visible difference is that Data IN and CSW appear in single packet (the
packet where reassembly ends).
Merging Data OUT with next CBW is ok for practical purposes, because
host periodically issues TEST UNIT READY (which does not have data
transfer and thus is not subject to the issue). While the CSW (and thus
SCSI status) will appear before Data OUT (and next CBW), the packets
will be correctly linked.
Workaround USBLL reassembly limitation by anticipating that SCSI Data
can be concatenated with Bulk Only Transport wrappers. Proper solution
would involve implementing a framework to allow USB class dissectors to
signal expected transfer length on Bulk IN or Bulk OUT endpoint whenever
CBW is encountered.
Guy Harris