i386 VMI fix for lazy MMU bug, take 3
cover i386 VMI fix for lazy MMU bug, take 3
In light of objections to the original fix I proposed for lazy MMU interactions between VMI and kmap_atomic in interrupt context, I have an updated patch series.
The first is for 2.6.21; it is an urgent bugfix, please apply. The second backs out the first; it is required to avoid conflicts
with later work which has been done on the paravirt-ops code. It should be added to subsystem queues which are being held
for 2.6.22.
The third illustrates the form the fix will likely take after the paravirt-ops
queue is applied, but is for demonstration purposes only.
Zach