Hello, Attached is the rest of the 64K page patch. It's been tested uniprocessor on Malta 20KC and 25KF. It also runs on a 6-way SMP functional simulator containing 5KF's. Ran tests with 16K and 64K page size. Question: Is there an SMP malta board? There are 2 areas which could use improvement: (1) Because 64K is larger than a 15 bit immediate operand, I could not get the asm-offsets mechanism to produce the correct constants. So I enlisted a fairly gruesome hack of using #define's for _PAGE_SIZE and _THREAD_SIZE, for that page size. Hopefully someone has a better idea. (2) In tlbex.c:build_adjust_context(), I suspect the change for shift = PAGE_SHIFT - 12 should be more generally true, rather than just for the CPU's mentioned in the case statement. I was conservative there because I'm not familiar with the CPU_VR41* machines. Hopefully someone more intimate with that code can comment. Signed-off-by: Peter Watkins