list if the user specified a number; otherwise, if we couldn't get the
list, just use what the user specified, so they can specify interfaces
in order to test them to see why they're not showing up in the list.
svn path=/trunk/; revision=46112
because of an incorrect exit status but because of the error message not
being what we expected.
Include the name of the program being tested in the capture
filter/interface options tests.
svn path=/trunk/; revision=46110
anyway, so that, for example, if an interface doesn't show up in the
list for some reason, the user can try specifying it explicitly for
testing purposes.
svn path=/trunk/; revision=46109
should be used (on success, have it return 0). Exit with that exit
status; if the problem is that we couldn't get the interface list or if
there are no interfaces in that list, return 2, as that's not a
command-line syntax error.
svn path=/trunk/; revision=46108
that if dumpcap isn't working right, we don't try TShark tests, as
TShark tests that involve capturing probably won't work.
svn path=/trunk/; revision=46104
consumed by that program, and that program only ends up reporting one of
those errors. Therefore, only log one of them; we're seeing errors in
the buildbot, and we want the more detailed error, so don't log a
message after GetInterfaceFriendlyNameFromDeviceGuid() returns an error,
just rely on the error logged by
GetInterfaceFriendlyNameFromDeviceGuid().
Ultimately, we don't want to print anything to the standard output or
error at all, as that stuff ends up being consumed by the program in
question; we should either ignore the error (if it's an "expected"
error) or report it through the API so the the caller can, in turn,
report it appropriately.
svn path=/trunk/; revision=46099
Enhancement to support MPLS-TP FM and LI payloads as per RFC 6427 and RFC 6435
Fixed some errors found by checkapi, changed filter names to use
proto abbr.
svn path=/trunk/; revision=46084
Friendly Names for interfaces on Windows
Notes on the changes the patch covers:
* if_info_t struct: addition of friendly_name
* Dumpcap Interface list format changes:
+ Win32: "dumpcap -D" shows friendly_name in place of descript if known
+ All: machine interface "dumpcap -D -Z none" includes friendly_name in the
list in addition to the existing parameters
* interface_options struct: addition of console_display_name
+ When an interface name is displayed in a console, it will typically be the
console_display_name (instead of name).
+ console_display_name is used as the basis of the autogenerated temp
filenames
+ console_display_name is typically set to the friendly_name if known,
otherwise it is set to the interface name
* Enhancements to capture_opts_add_iface_opt() (the function which process -i
options).
+ Can now specify the interface using its name and friendly_name
+ Interface name matching is case insenstive
+ Name matching first attempts exact matching, then falls back to prefix
matching
(e.g. dumpcap -i local)
+ Validates interface names, instead of blindly sending them off to
winpcap/libpcap
+ Interface specification by number is still supported.
* capture_opts_trim_iface() has been refactored:
+ Instead of repeating a decent chunk of the cost in
capture_opts_add_iface_opt(), it calls capture_opts_trim_iface() to specify the
interface.
* introduction of capture_win_ifnames.[ch] (windows only code)
+ Implements static function GetInterfaceFriendlyNameFromDeviceGuid() - a
windows version independant function to convert an interface guid into its
friendly name. Uses published api functions on windows vista and higher, but
falls back to unpublished API functions on older windows releases.
+ void get_windows_interface_friendlyname(/* IN */ char
*interface_devicename, /* OUT */char **interface_friendlyname); - extracts the
GUID from the interface_devicename, then uses
GetInterfaceFriendlyNameFromDeviceGuid() to do the resolution
* Auto temp filename generation:
+ Now uses wireshark_pcapng_* or wireshark_pcap_* depending on file format
+ Basis temp filename format on console_display_name
+ Win32: if console_display_name is a windows interface guid, extracts
numbers from GUID here (instead of in interface option processing)
GUI CHANGES:
* Dialog that displays when you click the "Manage Interfaces" button (within
Capture Options dialog) has been renamed from "Add new interfaces" to
"Interface Management"
* ui/gtk/capture_dlg.c: new_interfaces_w variable renamed to
interface_management_w
* Win32: Local Interfaces tab on Interface Management dialog, shows includes
friendly name as far left column
* Interface Management dialog defaults to larger size on win32 - so it fits
without resizing local interfaces tab
* Interface Management dialog now saves preferences when you click the apply
button (local hidden interfaces was not persisting across restarts)
* Tweaks: "Interface Details" dialog (Interface list->Capture Interfaces ->
Details):
+ "Friendly Name" renamed to "NDIS Friendly Name"
+ Added "OS Friendly Name" to the top of the list
* Win32: The "Capture Interfaces" dialog now shows the friendly name instead of
device guid
* Welcome screen:
+ The height of the interface list scrollbox dynamically adjusts & updates to
the number visible interfaces.
Up to 10 interfaces can be listed without a scroll bar, the minimum height
is for 2 interfaces.
+ Win32: now shows just the Friendly Name if known - in place of
"Interfacename_Guid:(Description)"
svn path=/trunk/; revision=46083
- In de_bssgp_pdu_in_error() use current_offset isof 0.
- in function de_bssgp_ran_inf_error_rim_cont. Error rim
container doesn't have field rim sequence number. This field exists in the rest
of containers but not in this one. So I have removed this field.
In this function we have field pdu in error and this field is mandatory, not
optional
https://bugs.wireshark.org/bugzilla/show_bug.cgi?id=8000
svn path=/trunk/; revision=46081
- Restore zooming out with middle-button + shift
- Lose unused graph->next and register_tap_listener_tcp_graph()
- Fix up some whitespace
svn path=/trunk/; revision=46072
We hand the user data in a CR or CC packet to the subdissectors,
but don't tell the subdissectors that - do we need to?
We don't hand the data in an ED packet to the subdissectors -
should we, and do we need to tell them that it's an ED packet?
svn path=/trunk/; revision=46067
epan/exceptions.h; it crashes).
Try BER first, including both the test dissection and the check of the
results. If that fails due to an exception being thrown, or due to the
results not indicating that it's a BER-encoded T.125 packet, try PER, so
that if the BER dissection doesn't work we don't just give up.
Get rid of some _U_s attached to parameters that are, in fact, used.
svn path=/trunk/; revision=46066
ReportedBoundsError exceptions when attempting to dissect the first bit
of the packet and, if either is thrown, assume the packet isn't a packet
for what we were trying to dissect, rather than just completely failing.
Return TRUE if the heuristic T.124 dissector recognizes the packet.
Get rid of some _U_s attached to parameters that are, in fact, used.
svn path=/trunk/; revision=46064
Also simplify a hash function - haven't measured collisions, but before
change sometimes wrong results would get attached. Don't think I fixed
it, and don't understand how equality function could have matched even
if hash function were colliding...
svn path=/trunk/; revision=46063
the source and destination address lengths and data, and adding them in,
rather than by copying those fields to a fixed-size static buffer - that
gets rid of a global variable (global variables considered harmful), and
also means that we don't try to copy 24 bytes of length+address from
packets that don't *have* 24 bytes of length+address (which caused
exceptions to be thrown on some OSI captures I have).
Construct some LI #defines out of other LI #defines, to make it a bit
clearer why they have the values they do.
Support the "additional information related to the clearing of the
connection" variable part parameter of the COTP DR packet (which just
means giving it a name, as its contents are user-defined - some HP-UX
OSI stack appears to just stick in a string saying that it's said
stack).
Make the code that decodes the variable part of a DR packet look like
the code that decodes the variable part of most other packets.
For COTP CR packets, determine the class up front by checking whether
the length is > 2. (At some point we might want to associate a class
indication with the COTP connection, if we see the connection setup,
and, if we have that indication, use it in preference to the
heuristics.)
Make the code to handle various length indicator values in the ATN case
more like the code in the non-ATN case.
Dissect the variable part of COTP ER packets.
Fix tpyos (TDPU->TPDU, tdpu->tpdu) and typpoes (accross->across).
Clean up white space.
svn path=/trunk/; revision=46060
Guy Harris