Hi Guys;
I think I made a mistake in failing to recognize a fairly nasty (and
pretty obvious in hindsight) interrupt-related hazard in the interrupt
notification schema. In order to fix it, I'd like to muck about the
scheduler::notifyOfInterrupts and other related routines.
I notice that scheduler.cc contains some ifdefs related to NATIVE_CLOCK,
which looks like it blows off notifying Java threads if the interrupt is
number 8 (which is, BTW, unfortunately, a machine-dependent thing,
located in a machine-independent file). Doing this would preclude Java
drivers from waiting on the clock interrupt. In fact, I would like to
use the clock interrupt to help debug the problem. Unless I hear some
objections from you all out there in kernel-land, I'm going to elide
this logic prior to rewriting the notification handler(s).
Also, has anybody else seen similar problems? I know Todd must've been
bitten, because of the PARANOID printouts in the notifyOfInterrupts
method.
-jm
--
==== John Morrison ==== [EMAIL PROTECTED] == http://www.mak.com/welcome.html
==== MaK Technologies Inc., 185 Alewife Brook Pkwy, Cambridge, MA 02138
==== vox:617-876-8085 x115
==== fax:617-876-9208
_______________________________________________
Kernel maillist - [EMAIL PROTECTED]
http://jos.org/mailman/listinfo/kernel
