Messages in this thread | | | Date | Fri, 10 Jun 2011 12:30:56 +0900 | From | KOSAKI Motohiro <> | Subject | Re: Change in functionality of futex() system call. |
| |
(2011/06/10 2:58), Peter Zijlstra wrote: > On Thu, 2011-06-09 at 14:05 +0200, Peter Zijlstra wrote: >> On Thu, 2011-06-09 at 20:37 +0900, KOSAKI Motohiro wrote: >>> 1) check the pte of the target address is RW attribute. >>> 2-1) if yes, it has no COW chance. we can safely use gup result. >>> 2-2) if no, we have to fallback slow vma walk. maybe, it's okey. >>> both RO mappings and COW are rare situation. >> >> Not so, clean file pages are RO, even if the vma is writable. We use the >> fault to track dirty state with. > > Why can't something like > http://marc.info/?l=linux-kernel&m=130737669810421 work once you restore > the rw argument?
Yeah, I agree this is best way. Private RO mappings is not sane and no real world users.
| |