Date: Thu, 25 Jul 1996 17:00:05 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: Charles Sandmann <sandmann AT clio DOT rice DOT edu>
Cc: DJ Delorie <dj AT delorie DOT com>, Demmer AT LStM DOT Ruhr-Uni-Bochum DOT De,
        djgpp-workers AT delorie DOT com
Subject: Re: open() / _open() bug
In-Reply-To: <9607250457.AA11737@clio.rice.edu>
Message-Id: <Pine.SUN.3.91.960725165114.17239B-100000@is>
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII


On Wed, 24 Jul 1996, Charles Sandmann wrote:

> > > The higher bits of omode are masked out before a possibly existing 
> > 
> > > !     r.x.bx = oflag & 0xff; /* :TOM: new open() doesn't mask out upper bits,
> > 
> > Charles - what does Win95 expect in BH?  Can we just leave it
> > unmasked, or will the spare bits we use confuse it?
> 
> I can't find my LFN docs right now (sigh) but I do know that additional bits
> were defined.  Which ones will have to wait a bit til I can find the DOCS.

I dug out the docs that Charles once sent to me (btw, I think the
Interrupt List is quite up-to-date on this now), and I think there *is* a
conflict.  The following bits are defined in 716Ch for the BX register:

	0000h	readonly
	0001h	writeonly
	0002h	readwrite
	0003h	"reserved; do not use"
	0004h	nomodlastaccess (readonly, doesn't modify last access time)
	0010h	deny-readwrite
	0020h	deny-write
	0030h	deny-read
	0040h	deny-none
	0080h	noinherit (child doesn't inherit the handle)
	0100h	no-buffering (read and write go directly to the disk)
	0200h	no-compress (not compressed on a compressed volume)
	0400h	alias-hint (use DI as the numeric tail for short name)
	2000h	n-criterr (don't call Int 24h handler when crit. err.)
	4000h	commit (commit file after each write)