I can imagine that an architecture with TLBs will usually evictOne risk for i386 are the huge tlbs that AMD uses (512 entries?) - hot pages might stay in the TLB forever.
the entry from the TLB sooner or later and the accessed bit will end
up beeing set again. On PPC, that isn't the case, the entry can well
stay a loooong time in the hash and if not evicted, _PAGE_ACCESSED
will never be set again.
Or does it snoop accessesNo. AMD K8 cpu partially snoop PDE/PTE accesses and ignore tlb flush instructions if they are certain that the tlb is valid, but I'm not aware that anyone snoops the complete tlb cache.
to the PTE to "catch" somebody clearing the bits ?