> The problem with the interrupts is not solved in rc9:
Yes, by the time we got this all resolved, it was already rc8 and I
though 2.6.31 was right around the corner ... the issue didn't seem
severe enough to stick in so late in the cycle, so I queued the patch
for .32 with a cc to stable to get in 2.6.31.1.
The workaround of limiting possible cpus to 16 should be OK for the time
between 2.6.31 and 2.6.31.1.
- R.