lkml.org 
[lkml]   [2010]   [Jul]   [18]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    Date
    From
    SubjectRe: [PATCH 5/6] kvm, x86: use ro page and don't copy shared page
    On 07/17/2010 07:31 AM, Gleb Natapov wrote:
    >>>
    >>> Currently pages allocated for guest memory are required to be RW, so after your series
    >>> behaviour will remain exactly the same as before.
    >>>
    >> Except KSM pages.
    >>
    >>
    > KSM page will be COWed by __get_user_pages_fast(addr, 1, 1, page) in
    > get_user_page_and_protection() just like it COWed now, no?
    >

    Well, we don't want to COW it on write faults.

    The optimal behaviour is:

    - write faults: COW and instantiate a writeable spte
    - read faults: instantiate a readable spte; if likely(page is
    writeable), make it a writeable spte; if likely(page is dirty) make it a
    dirty spte
    - speculative spte instantiations: if likely(page is present)
    instantiate a pte; if accessed, mark it accessed, if writeable, mark it
    writeable; if dirty, mark it dirty

    --
    error compiling committee.c: too many arguments to function



    \
     
     \ /
      Last update: 2010-07-18 17:17    [W:0.021 / U:32.004 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site