> > [there is another x86 change that went in as well recently with
> > pre-2.3.11, the 'lazy TLB flush feature' which was planned for quite some
> > time already - bug the gdb bug predates the lazy-TLB-flush changes. The
> > lazy-TLB feature lets us cut down on the number of TLB flushes rather
> > heavily, in a typical x86 Linux box about 80-90% of TLB flushes are now
> > gone.]
>
> The two bugs that seem to be caused by this (or at least, are introduced
> with 2.3.11-pre4) are the weird things in /proc/{2,3} and the 'bad page
> middle entry 23232323'-type messages.
i saw those too, hm. One thing to keep in mind is that 'lazy TLB state'
makes bugs live much longer than they live now, so lazy-TLB might just
have triggered an existing bug.
btw., in case you can reproduce the 'bad page middle entry ...' messages,
do they go away if you remove the lazy-TLB support for bdflush? (the
PF_LAZY_TLB changes in buffer.c)
-- mingo
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/