Mail Archives: djgpp/2000/09/27/07:52:45
Edmund Horner <ejrh AT paradise DOT net DOT nz> wrote:
> I understand that DPMI is not supposed to provide support for mmap(). Can
> we extend it? (or is that just _not_ done?)
We have absolutely no influence on DPMI, in general. It's a fixed
standard defined by Microsoft about a decade ago, and it's not going
to change. Ever.
> Windows: does DJGPP under Windows rely on the OS for virtual memory
> management?
Windows *is* DPMI (cum grano salis). So: yes, DJGPP relies on Windows
not only for virtual memory management, but for absolutely *every*
management, if running in a Windows DOS box. That's what DPMI is
about, in a nutshell.
> If so, then there's probably not much to be done about it,
> right?
Right.
> CSDPMI: can this server be extended to provide mmap() functionality?
It could, possibly. But what good would it really do us? The
fundamental idea behind mmap(), i.e. bypassing the file-related system
calls by directly talking to the page and HD caching mechanisms in the
OS kernel, still wouldn't be available (there *is* no such mechanism
in the DOS kernel, to begin with), so the benefits would be minimal,
if existant all.
In a nutshell, mmap() is slightly beyond that thin red line where
POSIX compatibility becomes impossible to achieve, with a system as
restricted as DOS as the basis of the effort.
--
Hans-Bernhard Broeker (broeker AT physik DOT rwth-aachen DOT de)
Even if all the snow were burnt, ashes would remain.
- Raw text -