Date: Sun, 28 May 1995 02:57:39 -0400
From: dj (DJ Delorie)
To: eliz AT is DOT elta DOT co DOT il
Cc: junaid AT dino DOT eng DOT monash DOT edu DOT AU, djgpp AT sun DOT soe DOT clarkson DOT edu
Subject: Re: Compiling gdb, dpmi DS:VRAM hack.


> > implemented it.  It's easier to just not call malloc() while you're
> > updating conventional memory.
> 
> Does this mean that applications shouldn't call malloc() while accessing
> conventional memory?  I thought that the segment value stays constant, it's
> only the base address/limit which change, so a ``well-behaved'' application
> shouldn't bother.  If I'm wrong, this should be documented.

If you are updating conventional memory using the selector provided
(_go32_info_block.selector_for_linear_memory), you're all set, because
its base does NOT change.

If you are using the big-limit-hack to access conventional memory with
your own %ds, and malloc() changes the base of your %ds, you lose
unless you recalculate the offset to conventional memory.