Messages in this thread | | | From | Vitaly Mayatskikh <> | Subject | Re: [PATCH 1/2] Introduce copy_user_handle_tail routine | Date | Mon, 07 Jul 2008 14:09:10 +0200 |
| |
Linus Torvalds <torvalds@linux-foundation.org> writes:
> Now, the stuff that comes *before* that point is the "try to fix up one > byte at a time" thing, which I'd like to be simple and dumb. At least to > start with.
Just to be clear: do these patches are good enough now (to start with)? Or, may be, it needs to be further improved?
> Of course, I also suspect that *eventually* we might want to make it > smarter and more complex. For example, while performance isn't a primary > issue, we might want to eventually avoid having to do _two_ faults (once > in the fast unrolled or word-at-a-time loop, and once in the byte-for-byte > one), by limiting the byte-for-byte one to be within a page, but that > would be a "future enhancement" thing.
Btw, how much does it cost to CPU to do a fault? Can it be compared with average time of find_vma()?
-- wbr, Vitaly
| |