stable/9 mps(4) rev 254938 == BOOM!

[Hans Petter Selasky]

On 01/29/14 19:45, Garrett Wollman wrote:
> I've been trying to puzzle out why the USB stack hits a GPF during
> boot on 9-stable on one of my machines (which runs 9.2 just fine), and
> by trial and error I've found that it is actually caused by r254938,
> which is a mega-MFC of mps(4) driver changes, even though the fault
> always occurs at the same place in a harmless USB subroutine.  Details
> below; config available on request.
>
> -GAWollman

Hi,

To me this sounds like someone is writing outside their assigned area.

options 	DEBUG_REDZONE

Maybe some change that should have been MFC'ed did not get MFC'ed.

I think you can copy the mps driver from -current just keeping the 
following deltas intact:

-
-       if (curthread->td_no_sleeping != 0)
+
+       if(curthread->td_pflags & TDP_NOSLEEPING)
                 sleep_flags = NO_SLEEP;

--HPS