[lkml]   [2002]   [Sep]   [7]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
    SubjectRe: invalidate_inode_pages in 2.5.32/3
    Daniel Phillips wrote:
    > On Friday 06 September 2002 00:19, Andrew Morton wrote:
    > > I'm not sure what semantics we really want for this. If we were to
    > > "invalidate" a mapped page then it would become anonymous, which
    > > makes some sense.
    > There's no need to leave the page mapped, you can easily walk the rmap list
    > and remove the references.

    Yes, unmapping and forcing a subsequent major fault would make
    sense. It would require additional locking in the nopage pth
    to make this 100% accurate. I doubt if it's worth doing that,
    so the unmap-and-refault-for-invalidate feature would probably
    be best-effort. But more accurate than what we have now.

    > > If the VM wants to reclaim a page, and it has PG_private set then
    > > the vm will run mapping->releasepage() against the page. The mapping's
    > > releasepage must try to clear away whatever is held at ->private. If
    > > that was successful then releasepage() must clear PG_private, decrement
    > > page->count and return non-zero. If the info at ->private is not
    > > freeable, releasepage returns zero. ->releasepage() may not sleep in
    > > 2.5.
    > >
    > > So. NFS can put anything it likes at page->private. If you're not
    > > doing that then you don't need a releasepage. If you are doing that
    > > then you must have a releasepage().
    > Right now, there are no filesystems actually doing anything filesystem
    > specific here, are there? I really wonder if making this field, formerly
    > known as buffers, opaque to the vfs is the right idea.

    That's right - it is only used for buffers at present. I was using
    page->private in the delayed-allocate code for directly holding the
    disk mapping information. There was some talk of using it for <mumble>
    in XFS. Also it may be used in the NFS server for storing credential
    information. Also it could be used for MAP_SHARED pages for credential
    information - to fix the problem wherein kswapd (ie: root) is the
    one who instantiates the page's blocks, thus allowing non-root programs
    to consume the root-only reserved ext2/ext3 blocks.
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to
    More majordomo info at
    Please read the FAQ at

     \ /
      Last update: 2005-03-22 13:28    [W:0.025 / U:2.648 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site