From: sandmann AT clio DOT rice DOT edu (Charles Sandmann)
Message-Id: <10109201805.AA03298@clio.rice.edu>
Subject: Re: gcc-3.01 seems unstable
To: djgpp-workers AT delorie DOT com
Date: Thu, 20 Sep 2001 13:05:00 -0500 (CDT)
Cc: wojciech DOT galazka AT polkomtel DOT com DOT pl
In-Reply-To: <250B3114DA16D511B82C00E0094005F8023FC191@MSGWAW11> from "Wojciech Galazka" at Sep 20, 2001 04:39:50 PM
X-Mailer: ELM [version 2.5 PL2]
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Reply-To: djgpp-workers AT delorie DOT com
Errors-To: nobody AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk

> gcc -D_USE_LIBM_MATH_H -D_HAVE_STDC -fno-builtin -O2 -g -nostdinc
> -Id:/djgpp.2/tests/cygnus -Id:/djgpp.2/tests/cygnus/../../include  -c -o
> test.o test.c
> 
> Exiting due to signal SIGSEGV
> General Protection Fault at eip=00000214
> eax=ffc9de35 ebx=9fff6cc0 ecx=00000004 edx=ffecc77d esi=0024483a
> edi=0022e963
> ebp=ffc9de1a esp=0000075c program=D:\DJGPP.2\BIN/gcc.exe
> cs: sel=02b7  base=02100000  limit=002cffff
> ds: sel=02bf  base=02100000  limit=002cffff
> es: sel=02bf  base=02100000  limit=002cffff
> fs: sel=028f  base=00015fb0  limit=0000ffff
> gs: sel=0277  base=00000000  limit=0010ffff
> ss: sel=028f  base=00015fb0  limit=0000ffff
> App stack: [000ae5b0..0002e5b0]  Exceptn stack: [0002dff8..0002c0b8]

fs = ss and in DOS memory area.  The only time our SS is there is very
early in startup (stub and transition to 32-bit code), using unixy
sbrk (but it's cs is also low base unless during transition), and during
final exit (also cs is low except during transition).  Since EIP is
small, this is either a jump to bogus location or early in use of a
UPX image (are these UPXed?)

This may also indicate the exception handler is not set up yet in a
child and gcc caught the exception.  Looking at register contents 
we could probably find exactly where it's crashing.  (Our debuggers
suppress reading the first 4K page, which must be re-enabled to debug
what's happening with UPX images).

> Operating system MS Windows NT 4.0 Wks + SP 6 (both with LFN driver or
> without give the same result)

Great, if reproducible we can find and fix.

> The files in the \djgpp\tests\cygnus directory compile with gcc-2.953 with
> no errors

Does this error happen with gcc (w/o make) on the above file?

> How could I help track down the problem ?

This is good help, I can see if I can reproduce this evening.