Path: utzoo!utgpu!watmath!uunet!tut.cis.ohio-state.edu!M.CS.UIUC.EDU!grunwald
From: grunwald@M.CS.UIUC.EDU (Dirk Grunwald)
Newsgroups: gnu.gdb.bug
Subject: glitch in gdb-3.0/encore
Message-ID: <8901101650.AA13668@m.cs.uiuc.edu>
Date: 10 Jan 89 16:50:51 GMT
Sender: daemon@tut.cis.ohio-state.edu
Reply-To: grunwald@m.cs.uiuc.edu
Distribution: gnu
Organization: GNUs Not Usenet
Lines: 21


There appears to be a bug in GDB-3.0, that I've been unable to narrow
down, although I would beleive it's with the COFF code.

I have a set of subroutines (the ``zopen'' routines mentioned on the net
that read/write compress-format files) that use ``setjmp'' and ``longjmp''.

There's an error in the code that is raised after a longjmp occurs.

On the Encore/UMAX, GDB-3.0 reports the error as occuring in a routine
called `sv', which is related to the setjmp/longjmp code. It gives no line
numbers in the stack trace.

SDB on the other hand, shows the error as occuring at a specific line in
the routine, with a backtrace.

I can send the code to anyone wishing to tackle this.

Dirk Grunwald
Univ. of Illinois
grunwald@m.cs.uiuc.edu