Serious bug in interrupt masking fixed

[Frank Wille <frank%phoenix.owl.de@localhost>]

Hi!

Today macallan@ committed intr.c 1.23, which fixes an intmask calculation
bug. It is triggered by interrupt handlers at different IPLs, which share
the same hardware interrupt. Probably i8259 based systems had the highest
probability to be affected by this bug.

A well-known problem caused by it was the vr(4) LOCKDEBUG panic on the
Pegasos2. It is gone now.

Pullup to 6.0 will be requested.

-- 
Frank Wille