Message-ID: <MAPI.Id.0016.00333138303633303030303930303039@MAPI.to.RFC822>
Read-Receipt-To: "Erik Berglund" <erik2 DOT berglund AT telia DOT com>
X-MSMail-Priority: Normal
X-Priority: 3
To: djgpp-workers AT delorie DOT com
MIME-Version: 1.0
From: "Erik Berglund" <erik2 DOT berglund AT telia DOT com>
Subject: Re: Re: gcc-crash - and a possible solution
Date: Wed, 16 Jun 99 18:25:44 +0100 (DJG)
Content-Type: text/plain; charset="ISO-8859-1"; X-MAPIextension=".TXT"
Content-Transfer-Encoding: 7bit
Reply-To: djgpp-workers AT delorie DOT com


Eli Zaretskii wrote:

> I think we need to investigate further, at least to know where does it
> crash, exactly.

I agree on that. Would it be possible for you to rebuild GCC 2.7.2.1 with 
DJGPP 2.01 on your site and use my register dump (posted on 14 Jun 1999)
to tell where the code fails?

Otherwise, I can install for instance GCC 2.8.1 with DJGPP 2.02 and try to 
reproduce the error.

> Do you, by any chance, have Win32s installed on your system?  Also, do
> you have 32-bit file access enabled?  If the answer to both these
> questions is YES, please try disabling one of these features and see
> if that makes the crashes go away.

1) Yes, I have Win32s installed.
2) No, 32-bit file access is not enabled.
I disabled Win32s for a while, but the problem was still there.

I'm not sure, but maybe the error has something to do with win3.11 DPMI
sometimes failing to serve a malloc() or sbrk() call?? If there are too many
requests for small chunks of RAM, some DPMI servers might fail,
according to FAQ section 6.6.

In my own djgpp-programs, I use to call malloc() only once, in the
beginning,
and they use to work well, also in DOS box, after being debugged :).

As a temporary solution, setting _CRT0_FLAGS_UNIX_SBRK
in GCC still works for me, since 2 Jun 1999.

Thanks for any help,
Erik