Date: Tue, 11 Mar 1997 23:45:33 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: Jih-Shin Ho <u7711501 AT bicmos DOT ee DOT nctu DOT edu DOT tw>
cc: djgpp AT delorie DOT com
Subject: Re: handle kbd & mouse without locking memory
In-Reply-To: <5g34v4$egl@ccnews.nctu.edu.tw>
Message-ID: <Pine.SUN.3.91.970311234506.2704S-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII


On 11 Mar 1997, Jih-Shin Ho wrote:

>   Do you know how DJGPP handles CTRL-C signal ? If you don't, please read
>   DJGPP faq. I use the same way to handle keyboard and mouse, and it works
>   fine in plain DOS. With this method, I don't need to lock any memory, but
>   I need to hack the library source.

Hack how?  Did you change the keyboard interrupt handler on crt0 to
catch more than just Ctrl-C?  Or did you do something else?

>   The selector for stack is equal to data selector whose size is changed to
>   4k only after kbd, mouse interrupt occurs (this is how DJGPP handles CTRL-C
>   signal).

Correct.

>   If I change SS to ___djgpp_ds_alias which is always valid, I get NO MORE
>   GPF. I don't know if this is bug in win95 DPMI server.

Is it a problem to set SS to `__djgpp_ds_alias'?  AFAIK, this is how
you should indeed work around such problems.