Mail Archives: djgpp/1996/07/12/09:45:15
Charles Sandmann <sandmann AT clio DOT rice DOT edu> wrote:
>Certainly locking everything (or just all code) is easier and safer than
>doing it a routine at a time. But this will cause bad paging performance
>on systems with insufficient memory (or an outright failure with a
>pagefault if there isn't enough memory to lock). If you are willing to
>put a bigger minimum ram requirement to run your program, that's fine,
>but locking only the routines needed would allow the code to potentially
>run on a 1Mb box.
Hi. I'm fairly new to DPMI and this thread interests me. I have been
studying the ALLEGRO library (a fantastic resource) and I notice that
"critical" code and data are locked via function calls. When you
refer to paging, I assume you mean the potential of memory to be
swapped out to disk, creating a larger amount of virtual memory
than is actually available, yes? If this is the case, then I imagine
that any data (or code) associated with an ISR would cause
a system crash if the interrupt was generated after the page had
been swaped out (and was hence unavailable). Is this right?
Or am I totally off-track.
Regards, Bruce.
- Raw text -