The idea with this change is threefold: 1. Shave some time off processing of the IPIs. 2. Reduce cacheline/bus traffic on systems with many CPUs. 3. Minimise time spent at IPL_VM. diff: http://www.netbsd.org/~ad/2019/tlb.diff libmicro comparison: http://www.netbsd.org/~ad/2019/tlb.html Comments? Thanks, Andrew