[lkml]   [2008]   [Jan]   [31]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
    Patch in this message
    SubjectRe: mmu_notifier: close hole in fork
    On Thu, Jan 31, 2008 at 02:01:43PM -0800, Christoph Lameter wrote:
    > Talking to Robin and Jack we found taht we still have a hole during fork.
    > Fork may set a pte writeprotect. At that point the remote pte are
    > not marked readonly(!). Remote writes may occur to pages that are marked
    > readonly locally without this patch.

    Good catch! This was missing also in my #v5 (KVM doesn't need that
    because the only possible cows on sptes can be generated by ksm, but
    it would have been a problem for GRU). The more I think about it, the
    more I think GRU will run so nicely with follow_page taking zero
    additional locking and with the nicely scalable PT lock that doesn't
    require 1024 threads to trash on the same lock on 1024 different cpus
    if they access addresses that aren't in the same naturally aligned 2M
    chunk of virtual address space. I understood follow_page was
    performance critical for GRU, if yes, then I hope my _dual_ approach
    is by far the best for at least GRU (and KVM of course for the very
    same reason), and of course it'll fit XPMEM too the moment you add
    invalidate_range_start/end too.

    Signed-off-by: Andrea Arcangeli <>

    diff --git a/mm/memory.c b/mm/memory.c
    --- a/mm/memory.c
    +++ b/mm/memory.c
    @@ -494,6 +494,7 @@ static int copy_pte_range(struct mm_stru
    spinlock_t *src_ptl, *dst_ptl;
    int progress = 0;
    int rss[2];
    + unsigned long start;

    rss[1] = rss[0] = 0;
    @@ -505,6 +506,7 @@ again:
    spin_lock_nested(src_ptl, SINGLE_DEPTH_NESTING);

    + start = addr;
    do {
    * We are holding two locks at this point - either of them
    @@ -525,6 +527,8 @@ again:
    } while (dst_pte++, src_pte++, addr += PAGE_SIZE, addr != end);

    + if (is_cow_mapping(vma->vm_flags))
    + mmu_notifier(invalidate_pages, vma->vm_mm, start, addr);
    pte_unmap_nested(src_pte - 1);
    add_mm_rss(dst_mm, rss[0], rss[1]);

     \ /
      Last update: 2008-02-01 01:05    [W:0.020 / U:4.672 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site