2007/5/24, Joe Marcus Clarke <marcus_at_marcuscom.com>: > On Thu, 2007-05-24 at 18:30 +0200, Attilio Rao wrote: > > 2007/5/24, Joe Marcus Clarke <marcus_at_marcuscom.com>: > > > On Thu, 2007-05-24 at 20:13 +0200, Attilio Rao wrote: > > > > Jeff Roberson wrote: > > > > > Thanks, Attilio and I will look at this right away. > > > > > > > > > > Jeff > > > > > > > > > > On Wed, 23 May 2007, Joe Marcus Clarke wrote: > > > > > > > > > >> On a -CURRENT build today with Jeff's LDT locking changes, I get an > > > > >> instant, reproducible panic by doing: > > > > >> > > > > >> # cd /usr/src > > > > >> # make -DNO_CLEAN buildworld > > > > >> > > > > >> Backing out the LDT changes to /sys/i386/i386/*machdep* corrects the > > > > >> problem. Here is a dmesg and backtrace. In frame 6, it appears mdp is > > > > >> NULL (i.e. td->td_proc->p_md = 0x0). > > > > > > > > Did you enable WITNESS/INVARIANTS/INVARIANT_SUPPORT? > > > > > > No, this kernel does not have WITNESS or INVARIANTS. I can enable those > > > tomorrow if that would give you more useful information. > > > > It would be very appreciated, since I've been unable to reproduce your > > panic on a centrino dual core. > > Will do, but it may be that adding those options masks the bug. Have > you tried without them? Yes, I tried both. And please don't add WITNESS_SKIPSPIN since the dt_lock is a spinlock. Thanks a lot, Attilio -- Peace can only be achieved by understanding - A. EinsteinReceived on Thu May 24 2007 - 14:39:01 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:10 UTC