Re: umtx/libthr SMP fixes.

Bryan Liesner wrote:
> On Thu, 5 Jun 2003, Terry Lambert wrote:
> > As I said: I still think there is a lost serialization here
> > that's at the root of the problem.  I can't really dedicate
> > the equipment I have here to reproducing the issue at this
> > time, or I'd track down the race I think may be happening.
> 
> The original panic in kern/52718 is no longer reproduceable for me at
> this time.  After Jeff's latest changes, the panic moved from boot time to
> panicking when I did an init 6.  I could _not_ reproduce the new panic
> if I built a kernel with DDB, but a DDBless kernel would panic every time
> after init 6. Needles to say, that makes things really tough to track
> down...
> 
> After shit-canning the acpi module, it doesn't panic at all.  I'm sure
> the race conditions are still lying in wait, but due to a lack of
> interest and the incredulous attitude of most on the mailing list, I'm
> going to forget about it for the time being.
> 
> ACPI has always worked fine on my system, and I'm probably masking
> problems by not loading the module.  Since I'm running a desktop
> system here, acpi doesn't really buy me anything, so apm is back in my
> kernel.

I hesistate to suggest this because everyone always gives me
crap about me not disclosing the bug, but unless you are ready
to grovel around in locore, and figure out what the root cause
is for the difference in behaviour, I'm going to say that
the most likely cause is that a DDB kernel uses more memory.

Given that, I'm going to suggest you try again without DDB,
but with:

	options DISABLE_PSE
	options DISABLE_PG_G

If it doesn't fix the problem, then you haven't really lost
anything but the time it takes to compile, reboot, rename,
and reboot again.

-- Terry