|
From: Jeremy F. <je...@go...> - 2004-09-08 20:38:47
|
On Sat, 2004-09-04 at 00:08 +0100, Nicholas Nethercote wrote: > Originally we didn't use big-bang. Jeremy put it in with the FV memory > layout changes; AIUI he experimented with direct-offset shadow addressing > but the performance was no better, so the committed version didn't use it, > even though the code to support it was all present. It was part of my thinking about how to deal with a 64-bit address space. It would have been nice to get an unambiguous performance advantage for 32-bit processes, but it was about even, and the address space management problems are annoying. But the 2-level page table scheme doesn't scale well for 64-bit address spaces, and it duplicates something that the CPU will do for free anyway, so why not use the existing mechanism? Hence, direct mapping into a sparse(-ish) array. J |