Subject: Re: Are there any known problems with uvm or pmap in -current as of about Oct. 16?
To: Monroe Williams <monroe@criticalpath.com>
From: Chuck Silvers <chuq@chuq.com>
List: port-macppc
Date: 11/04/2001 23:55:49
hi,

matt and I spent a bunch of time working on the NEWPMAP problems today,
and we found and fixed several bugs.  our stress-test was "make -j 20" in
a kernel build dir.  when we started this morning, it took about 30 seconds
to trigger one of the assertions.  a few minutes ago a test run completed
without a problem after 60 minutes.

at this point the new pmap appears to be much more stable than the old one,
so I recommend that everyone turn on "options NEWPMAP" if you haven't
done so already.

-Chuck


On Tue, Oct 30, 2001 at 08:13:44PM -0800, Monroe Williams wrote:
> on 10/26/01 7:03 PM, Chuck Silvers at chuq@chuq.com wrote:
> 
> > the bug in the new pmap that people have complained about causes a panic
> > when some sanity-checking code notices that the internal pmap state
> > has become inconsistent.
> > 
> > if you haven't had any problems with the old pmap, don't switch to the new
> > one.
> > these pmap bugs are the sort that seem to only affect some people and not
> > others.  I've never had any problems with either pmap myself, but then
> > I don't do much with my powermac.
> 
> With a -current kernel and the old pmap, the machine I'm working with
> reliably panics in short order (less than 1/2 hour).  I never saw this panic
> in testing, but when I deployed the new machine it made for an exciting day
> until I reverted to the old server.
> 
> When you're talking about IS, exciting == bad.
> 
> This machine _does_ get used for a lot, and random panics just aren't
> acceptable.
> 
> Is someone actively working on fixing the bugs in the new pmap?  I really
> hate to hear that -current macppc kernels Just Aren't Reliable with either
> pmap, but that seems to be what the evidence suggests.
> 
> A 1.5.2 kernel just doesn't make good use of the 1.25G of RAM in this box,
> but that's what I'm planning to deploy this weekend, if I can figure out the
> proper settings for BUFPAGES & friends.  I may end up downgrading to 512M to
> get the kernel to boot...
> 
> -- monroe
> ------------------------------------------------------------------------
> Monroe Williams                                  monroe@criticalpath.com