From: Andris <pavenis AT lanet DOT lv>
Organization: Pavenis
To: Andrew Cagney <ac131313 AT ges DOT redhat DOT com>
Subject: Re: Internal GDB errors with current GDB snapshots and -gdwarf-23
Date: Sat, 28 Sep 2002 16:40:59 +0300
User-Agent: KMail/1.4.7
Cc: djgpp-workers AT delorie DOT com, gdb AT sources DOT redhat DOT com
References: <3D6E31F2 DOT 25704 DOT 9E195D AT localhost> <3D89F4C2 DOT 90809 AT ges DOT redhat DOT com>
In-Reply-To: <3D89F4C2.90809@ges.redhat.com>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Disposition: inline
Message-Id: <200209281640.59537.pavenis@lanet.lv>
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by delorie.com id g8SGbvS23062
Reply-To: djgpp-workers AT delorie DOT com

On Thursday 19 September 2002 19:01, Andrew Cagney wrote:
> Has anyone had a chance to look at this?
>
> Andrew

Tried today (with GDB-5.2.90_20020927) and reported results to bug report gdb/726. 
Currently I don't see other way for DJGPP as extremly ugly workaround (perhaps 
no need to get it in official GDB sources):

--- gdb+dejagnu-5.2.90_20020927/gdb/macrotab.c~1	2002-07-29 22:55:26.000000000 +0000
+++ gdb+dejagnu-5.2.90_20020927/gdb/macrotab.c	2002-09-28 16:24:58.000000000 +0000
@@ -272,7 +272,14 @@
     {
       /* They can't both be from #included files.  Then we shouldn't
          have walked up this far.  */
+      /* FIXME: Workaround for GCC bug: it misinterprets include from line 1
+         as from line2 when -imacros is used (in this case both lines
+         could be 2)   AP, 2002/09/28 */
+#if defined(__DJGPP__)
+      gdb_assert (! included1 || ! included2 || line1==2);
+#else
       gdb_assert (! included1 || ! included2);
+#endif
 
       /* Any #included position comes after a non-#included position
          with the same line number in the #including file.  */


Andris