Wireshark · Ethereal-dev: Re: [ethereal-dev] ICQ Crash - vprintf()

[Guy Harris <guy@xxxxxxxxxx>]

> >From the backtrace made in a core you wont see any symbols resolving
> so i guess it triggers a bug in an external lib (gtk ?) 

Any crash as a result of asking Ethereal to do something will involve a
lot of GTK+ routines on the stack; however, this particular crash was
*probably* due to an incorrect call to "proto_tree_add_item_format()" in
the dissector for the CMD_STATUS_CHANGE command - it didn't produce a
crash on Solaris/SPARC, but the line for the body of the command didn't
have a title.

The crash probably did happen in an external library ("libc", in
"vfprintf()"), but the ultimate cause was probably a bug in Ethereal.

I've attached a patch; I've checked in this change.

Index: packet-icq.c
===================================================================
RCS file: /usr/local/cvsroot/ethereal/packet-icq.c,v
retrieving revision 1.7
retrieving revision 1.8
diff -c -r1.7 -r1.8
*** packet-icq.c	1999/11/18 21:04:53	1.7
--- packet-icq.c	1999/12/01 23:58:44	1.8
***************
*** 1031,1036 ****
--- 1031,1037 ----
  					hf_icq_cmd,
  					offset,
  					4,
+ 					CMD_STATUS_CHANGE,
  					"Body");
  	subtree = proto_item_add_subtree(ti, ett_icq_body);
  	if (status!=-1)