Debian User wrote:
>
> I've improved the ASSERT macro in kernel/include/monitor.h to dump the
> expression that was asserted, as well as the file and line number of the
> failure of the assertion.
OK, incorporated, thanks.
> I've also fixed kernel/phymem-mon.c to allow physical memory access to
> inexistant physical memory pages, because in real life, such access *is*
> valid, and simply has undefined results. The panic that was previously there
> was commented out, and in the write function, "return;" was added. The panic
> did not allow my newly experiemental os to run since it tested the amount of
> physical memory by finding the first page that has undefined read/write
> results :)
Also incorporated. That might help EROS, which was sizing memory like this
too. For reads, I return 0xff in all bytes involved in the read which
were out of bounds. I think out of bounds memory access float high.
Have to check. Anyways, should be 0x00 or 0xff, rather than leave the
random value in the parameter pass on the stack.
Will upload to CVS soon...
-Kevin
--
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Kevin Lawton [EMAIL PROTECTED]
MandrakeSoft, Inc. Plex86 developer
http://www.linux-mandrake.com/ http://www.plex86.org/
