Date: Tue, 9 Jun 1998 19:09:49 +0300 (IDT)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: HANRIGOU <HANRIGOU AT cgste DOT mq>
cc: spector AT EnchantedLearning DOT com, djgpp AT delorie DOT com
Subject: Re: Bash hangs -- to be continued...
In-Reply-To: <357D521A.DF04200B@CGSTE.MQ>
Message-ID: <Pine.SUN.3.91.980609190534.16324V-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Precedence: bulk


On Tue, 9 Jun 1998, HANRIGOU wrote:

> Does it really give CPU control to other programs?

My testing indicates that it does.  If you have any means to show CPU 
usage, please look at it and tell the results with the new variant of 
__dpmi_yield.

> How does __dpmi_int works? What happens after running a child 
> process (as opposed to previous __dpmi_yield definition)?

As far as I could see, everything works with the new version.  The old
version issued the INT 2Fh instruction directly, whereas the new one does
this via the DPMI server.  It seems like some versions of Windows have
subtle bugs in servicing the first call, but not the second, when nested
DPMI clients are involved.  (If you know a bit about the via delorosa
which each Int instruction goes through the various so-called ``thunking''
levels before it winds up in the module which actually services it, you
won't be surprised.)