cvs commit: src/sys/i386/i386 pmap.c

[Andrew Gallatin]

Mike Silbersack writes:
 > 
 > On Thu, 7 Aug 2003, Andrew Gallatin wrote:
 > 
 > > It changed the failure somewhat.  Now make segfaults, and syntax
 > > errors which may not actually exist are found when building libstdc++.
 > >
 > > I'm going to back off to a known good kernel and make sure my srcs are
 > > not hosed.
 > >
 > > Drew
 > 
 > Are you sure this final bug isn't gcc's fault?  One of my machines has had
 > occasionaly buildworld failures in libstdc++, _but nowhere else_.  I was
 > even able to repeat the failure if I did make buildworld -DNOCLEAN.
 > 
 > (That doesn't cover make segfaulting, of course.)

I'm pretty sure.   The known good kernel (july 4th) just built the
world successfully with -j16.

I think I'm going to MFI386 Alan's pmap fix to pmap_insert_entry(),
that at least seems to help things out quite a bit.

I'm building now with an INVARIANTS/WITNESS kernel to see if its
perhaps related to  the memory-modified-after-free bug that somebody
(you?)  reported.

Drew