[Pascal Quantin <pascal.quantin@xxxxxxxxx>]

2014-11-21 17:34 GMT+01:00 Semjon <semgo@xxxxxx>:

Am 21.11.2014 um 15:42 schrieb Bill Meier:
> On 11/21/2014 9:29 AM, Pascal Quantin wrote:
>>
>>
>> 2014-11-21 14:06 GMT+01:00 Semjon <semgo-S0/GAf8tV78@xxxxxxxxxxxxxxxx
>> <mailto:semgo-S0/GAf8tV78@xxxxxxxxxxxxxxxx>>:
>>
>>
>>
>>
>>     Am 21.11.2014 um 10:06 schrieb Guy Harris:
>>      >
>>      > On Nov 21, 2014, at 12:48 AM, Semjon
>>     <semgo-S0/GAf8tV78-XMD5yJDbdMReXY1tMh2IBg@xxxxxxxxxxxxxxxx

>>     <mailto:GAf8tV78-XMD5yJDbdMReXY1tMh2IBg@xxxxxxxxxxxxxxxx>> wrote:
>>      >
>>      >> One of my current problems is with
>>      >>
>>      >> tvb_get_faked_unicode(...)
>>      >>
>>      >> which isn't available anymore.
>>      >> In my Protocol I have some Ascii-encoded String but which comes
>>     as two
>>      >> bytes per character. Example:
>>      >> {0x0031, 0x0032, 0x0033, 0x0034, 0x0000} in tvb should display in
>>      >> GUI/Tree/PacketList as "1234"
>>      >
>>      > If that's truly ASCII-encoded, that would be a significant waste
>>     of bytes - you could just use one byte per character for ASCII; if
>>     the second byte is always zero, that byte serves no useful purpose.
>>      >
>>      > So I'll assume it's a *superset* of ASCII, and that you mean
>>     either "UTF-16 encoded string" or "UCS-2 encoded string" rather than
>>     "ASCII-encoded string which comes as two bytes per character".
>>      >
>>      > So:
>>      >
>>      >> I used to call:
>>      >>
>>      >> tvb_get_faked_unicode(NULL,tvb, 20,
>>     ((tvb_length(tvb)-20)/2),ENC_BIG_ENDIAN)
>>      >>
>>      >> and display result as %s in col_append_fstr() or as FT_STRING in
>>      >> proto_tree_add_string().
>>      >>
>>      >> So could anyone give me a hint, is there a function still
>>     available for
>>      >> this type of encoding
>>      >
>>      >       tvb_get_string_enc(tvb, {offset}, {length of string},
>>     ENC_UTF_16|ENC_BIG_ENDIAN)
>>      >
>>      > or
>>      >
>>      >       tvb_get_string_enc(tvb, {offset}, {length of string},
>>     ENC_UCS_2|ENC_BIG_ENDIAN)
>>      >
>>      > depending on whether it's UTF-16 (with surrogate pairs to handle
>>     Unicode characters that don't fit in 16 bits) or UCS-2 (supporting
>>     only characters in the Unicode Basic Multilingual Plane, without
>>     surrogate pairs).
>>      >
>>      > Note that tvb_get_string_enc() returns a UTF-8-encoded string;
>>     octet sequences that can't be mapped to UTF-8 strings will be
>>     replaced by the Unicode "replacement character".
>>      >
>>      >> In general is there a fast/convenient way - other than manually
>>     looking
>>      >> through the sources after functions that might do what i want -
>>     to check
>>      >> if this function X is now replaced by function Y.
>>      >
>>      > No.  You could check doc/README.developer, etc. to see if
>>     anything is mentioned.
>>      >
>>      >> Other examples I need to replace are:
>>      >> abs_time_to_ep_str()
>>      >
>>      >       abs_time_to_str({wmem scope}, ...)
>>      >
>>      > The old "ephemeral" and "session" memory mechanisms are
>>     deprecated in favor of the new wmem mechanisms.  The scope that's
>>     equivalent to "ephemeral" scope is, I think, packet scope (right,
>>     Evan?), so you'd want
>>      >
>>      >       abs_time_to_str(wmem_packet_scope(), ...)
>>      >
>>      >> nstime_delta()
>>      >
>>      > Its replacement is called nstime_delta() and has the exact same
>>     arguments. :-)
>>      >
>>      > However, you need to include <wsutil/nstime.h> to get it declared.
>>      >
>>
>>     Well thanks a lot everybody for helping. I could resolve almost
>> all of
>>     my Problems with Your help. In fact the "ASCII encoded
>> 2-byte-string" is
>>     a Unicode String shame on me :-)
>>
>>     Unfortunately no luck with nstime_delta().
>>
>>     I already had included  <wsutil/nstime.h>
>>
>>     My call looks like this:
>>
>>              proto_item *it;
>>              nstime_t ns;
>>
>>              it=proto_tree_add_uint(xyz_tree, hf_xyz_response_to, tvb,
>> 0, 0,
>>     xyz_trans->req_frame);
>>              PROTO_ITEM_SET_GENERATED(it);
>>
>>              nstime_delta(&ns, &pinfo->fd->abs_ts, &xyz_trans->req_time);
>>              it=proto_tree_add_time(xyz_tree, hf_xyz_response_time,
>> tvb, 0,
>>     0, &ns);
>>              PROTO_ITEM_SET_GENERATED(it);
>>
>>     It always generates errors LNK2019/LNK1120 ... unresolved external
>>     symbol "__imp__nstime_delta" in function ...
>>
>>     Hope You have an idea here. I'm not really good in finding the
>> necessary
>>     functions/files to include in such a big project and my search on the
>>     www on this was not successful.
>>
>>
>> Hi,
>>
>> assuming that your proprietary dissector is a plugin, ensure that your
>> makefile indicates the path to libwsutil. I guess you are on Windows, so
>> your Makefile.nmake file should contain:
>>
>> !IFDEF ENABLE_LIBWIRESHARK
>> LINK_PLUGIN_WITH= ..\..\wsutil\libwsutil.lib
>> CFLAGS=$(CFLAGS)
>>
>>
>
>
> See plugins\ethercat for a dissector which uses nstime_delta()  [in
> packet-esl.c].
>
> Also: proto.h (#included by packet.h) #includes nstime.h so you need not
> explicitly include same.
>
>
>
> ___________________________________________________________________________
> Sent via:    Wireshark-dev mailing list

> <wireshark-dev-IZ8446WsY0/dtAWm4Da02A@xxxxxxxxxxxxxxxx>
> Archives:    http://www.wireshark.org/lists/wireshark-dev
> Unsubscribe: https://wireshark.org/mailman/options/wireshark-dev
>
> mailto:wireshark-dev-request-IZ8446WsY0/dtAWm4Da02A@xxxxxxxxxxxxxxxx?subject=unsubscribe
>
>

This is what I have in my Makefile.nmake, all automatic generated:

CFLAGS=/WX /DHAVE_CONFIG_H /I../.. $(GLIB_CFLAGS) \
        /I$(PCAP_DIR)\include -D_U_="" $(LOCAL_CFLAGS)

!IFDEF ENABLE_LIBWIRESHARK
LINK_PLUGIN_WITH=..\..\epan\libwireshark.lib
CFLAGS=/D_NEED_VAR_IMPORT_ $(CFLAGS)

if I change it to what You suggested Pascal it throws loads of the
LNK2010/LNK1120 errors so I'd rather stick with one of these errors and
change it back :-)

I'm running windows 7 pro 64 with MSVC2010EE and building the dissector
as a dll-plugin win32 variant which worked fine until some wireshark
source version.
Yesterday I started using Wireshark 1.99.1 which itself without the
plugin compiled without problems.
Unfortunately I can't remember when nstime_delta startet making problems
in my plugin as I didn't need it this bad so I just commented it out for
later.
I think it startet with the 1.11.x sources...

You're right Bill even packet-tcp uses nstime_delta without including
nstime.h explicitly. I compared ethercat to my dissector. Only
difference I see is that Ethercat includes glib.h in packet-esl.c where
nstime_delta is called. I don't include glib.h but it doesn't make a
difference if I do I tried it.

If there's some additional Info I could provide just tell me as I have
no clue where to look at the moment :-(

Edit Makefile.nmake and replace:
LINK_PLUGIN_WITH=..\..\epan\libwireshark.lib
by
LINK_PLUGIN_WITH=..\..\epan\libwireshark.lib ..\..\wsutil\libwsutil.lib

Pascal.