lkml.org 
[lkml]   [2014]   [Feb]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    Date
    SubjectRe: [PATCH v2] ceph: fix posix ACL hooks
    From
    On Mon, Feb 3, 2014 at 2:40 PM, Al Viro <viro@zeniv.linux.org.uk> wrote:
    >
    > Umm... Point, but that actually means that we get an extra pitfall for
    > filesystem writers here. foo_permission() passes dentry (now that it
    > has one) to foo_wank_a_lot(), with the latter using dentry->d_inode at
    > some point...

    I agree.

    The good news, though, is that in the RCU lookup case, we have that
    MAY_NOT_BLOCK thing, and most filesystems will have errored out for
    any complex operations.

    RCU lookup is special, and complicated, and sadly, the permissions
    checking is very much part of that. But for the really complex cases,
    at least we can punt.

    >> Look at gfs2_lookupi() in particular, and check how it is called.
    >
    > Yeowch... gfs2_ok_to_move() is particulary nasty... WTF do we need
    > it for and why is it not racy as hell?

    I don't know. And I suspect that for things like the journal index
    file lookup (which is actually worse - see gfs2_jindex_hold()) we
    don't really about the permissions, since this is just done at
    init_journal() time.

    So I think all of this is quite solvable for gfs2, it just wasn't the
    obvious kind of "we already have the dentry" case that every single
    other case I looked at was.

    Linus


    \
     
     \ /
      Last update: 2014-02-04 00:21    [W:3.270 / U:0.508 seconds]
    ©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site