Re: insight/164: insight crashes on binaries from gpc (Gnu Pascal)2.1

[Keith Seitz <keiths at redhat dot com>]

The following reply was made to PR insight/164; it has been noted by GNATS.

From: Keith Seitz <keiths@redhat.com>
To: <r.scher@varilog.com>
Cc: <insight-gnats@sources.redhat.com>
Subject: Re: insight/164: insight crashes on binaries from gpc (Gnu Pascal)
 2.1
Date: Tue, 11 Jun 2002 10:45:44 -0700 (PDT)

 On 6 Jun 2002 r.scher@varilog.com wrote:
 
 > I built & installed per instructions in
 > p/INSTALL under the top-level source directory.
 >
 > I also built insight-5.2 from sources at your site.
 > The insight that I built works fine on C programs.
 >
 > Programs large & small that I compile and link using gpc 2.1 have some
 > slightly unusual stabs, perhaps.  In any case, when I try to use insight
 > to open such a file, insight crashes.  This only happens in graphical
 > mode, and I have tracked it down and can offer a prospective patch.
 > insight works more or less normally on a gpc2.1 generated binary, except
 > that all variables print as "void".  This is a separate problem and
 > looks to me like a gpc 2.1 defect.  But this crash situation, I'm not
 > sure who's not following the stabs spec., if there is such a thing...
 
 There is stabs docs, and they belong to the gdb group.
 
 I've never used Pascal with Insight. I didn't even know that gdb supported
 it! In any case, your going to run into a lot of problems with variable
 display, since that uses varobj in gdb, which only works with c, c++, and
 java. In order for it to work with Pascal, there might be some work to do.
 
 > After getting your initial mail, I'll e-mail you three files:  my patch
 > (dbxread.c.diff) and the pascal source and gpc2.1-produced executable of
 > "smalltest", a trivial program that will illustrate the crash.
 
 Insight crashes from something in dbxread.c and gdb does not? That doesn't
 sound very likely (unless it is crashing in varobj, which gdb proper
 doesn't use). Can you send me a backtrace of the crash?
 
 In any case, I guess the testcase is needed. If you have a patch for this,
 it'll probably have to go to gdb-patches@sources.redhat.com, unless it is
 a patch to a file in gdb/gdbtk.
 
 Keith