X-Authentication-Warning:	delorie.com: mail set sender to djgpp-bounces using -f
Subject:	Re: Random compile errors under WIndows (32 bit Windows Vista and
	Windows 10)
Newsgroups:	comp.os.msdos.djgpp
References:	<5611566C DOT 4020507 AT iki DOT fi> <561A4BCE DOT 7050606 AT iki DOT fi>
	<56213FE2 DOT 7090409 AT iki DOT fi> <562168C6 DOT 5050001 AT gmx DOT de> <5621D433 DOT 2020004 AT iki DOT fi>
From:	"Frank H. Sapone (fhsapone AT windstream DOT net) [via djgpp AT delorie DOT com]" <djgpp AT delorie DOT com>
Date:	Sat, 17 Oct 2015 01:39:31 -0400
User-Agent:	Mozilla/5.0 (Windows NT 5.1; rv:38.0) Gecko/20100101
	Thunderbird/38.3.0
MIME-Version:	1.0
In-Reply-To:	<5621D433.2020004@iki.fi>
Message-ID:	<55c9$5621df1b$d07bb077$21199@ALLTEL.NET>
X-Complaints-To:	abuse AT usenetserver DOT com
Organization:	UseNetServer.com
Lines:	35
X-Trace:	055c95621df1b75de15d221199
X-Received-Bytes:	2419
X-Received-Body-CRC:	2951412335
Bytes:	2667
To:	djgpp AT delorie DOT com
DJ-Gateway:	from newsgroup comp.os.msdos.djgpp
Reply-To:	djgpp AT delorie DOT com
Errors-To:	nobody AT delorie DOT com
X-Mailing-List:	djgpp AT delorie DOT com
X-Unsubscribes-To:	listserv AT delorie DOT com

On 10/17/2015 12:53 AM, Andris Pavenis (andris DOT pavenis AT iki DOT fi) [via 
djgpp AT delorie DOT com] wrote:
>
> After all that I'm more inclined to think that the problem lies on our
> side. There could be slight
> differences between different DPMI providers that could trigger
> problems. I have some more
> ideas to try but I would not like to discuss them before trying.

There was something sezero and I noticed that may possibly be related to 
alignment issues with SSE and DXEs.  But, we have not yet 100% confirmed.

Very briefly, we got the source code to Sage, a minigl like driver that 
also had a DJGPP codepath (written by Daniel Borca who was a DJGPP 
contributor in the past).  I noticed that static linked builds to Q2DOS 
worked fine with Pentium 3 processors, but builds that used Sage as a 
DXE would usually start with a black screen (but actually dropped back 
to DOS so it could be blindly attempted to try and restart).

I somehow found out by mistake adding some extra parameters at startup 
would allow it to start every time, but the string length was dependent 
within a certain window and differed from machine.

Talking to the author of the code he suggested changing movaps to movups 
and see if that resolved the issue and it did.

He suggested that DXE may be affecting alignment, but we haven't done 
much more testing yet.

Not a complete show stopper since we were able to work around it with no 
performance degradation, but thought it may be worth mentioning for 2.06 
and before I forget about it.

Frank

- Raw text -