Date: Sat, 26 May 2001 10:01:24 +0300
From: "Eli Zaretskii" <eliz AT is DOT elta DOT co DOT il>
Sender: halo1 AT zahav DOT net DOT il
To: "Alex Oleynikov" <alex AT compuweigh DOT com>
Message-Id: <2593-Sat26May2001100124+0300-eliz@is.elta.co.il>
X-Mailer: Emacs 20.6 (via feedmail 8.3.emacs20_6 I) and Blat ver 1.8.9
CC: djgpp AT delorie DOT com, sandmann AT clio DOT rice DOT edu
In-reply-to: <003701c0e555$31fcb600$1400a8c0@alex2000>
Subject: Re: SRAM chip access problem
References:  <003701c0e555$31fcb600$1400a8c0 AT alex2000>
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
Precedence: bulk

> From: "Alex Oleynikov" <alex AT compuweigh DOT com>
> Date: Fri, 25 May 2001 15:59:23 -0400
> 
> >You don't check the return values of these three function calls.
> 
> Actually, I do, but for the simplicity of the code snippet I just stripped
> all extra stuff.

It is usually best not to remove parts of your code when posting it.

> >something _is_ wrong with the selector you got from
> >__dpmi_allocate_ldt_descriptors: it seems to be a selector for code,
> >not for data (bit 3 of the selector is 0 instead of 1).
> 
> I am a little bit confused here. DJGPP help says, that bit #3 of a
> descriptor has to be set for code segments and cleared for data segments.

You are right.

> (BTW, how do you know whether it is code or data segment just by looking at
> the selector's value? Isn't this information available only from the
> descriptor to which this selector points in the LDT?).

The selector itself (c7) looked to me like a code selector.  Charles,
did I miss something?

> What _is_ amazing, when I upgraded to CWSDPMI r5 after your suggestion and
> used "near" pointer access to the SRAM memory:
> 
> if( __djgpp_nearptr_enable() )
> 
> Window_Base=(char *)(mapping.address + __djgpp_conventional_base);
> 
> everything began to work fine! Before the upgrade this code was also
> breaking down with SIGSEGV exception.

As Charles explained, this particular example hits the bug in r4 which
was corrected in r5.  So upgrading is the right solution.