Re: Regression in next with ext4 oops

[Al Viro <viro@ZenIV.linux.org.uk>]

On Tue, Oct 04, 2016 at 11:00:41AM +0200, Jan Kara wrote:
> Hi!
> 
> On Mon 03-10-16 16:30:55, Tony Lindgren wrote:
> > I'm seeing a repeatable oops with Linux next while running
> > update-initramfs, see below. I tried reverting commit 59aa5a3aeead
> > ("fscrypto: make filename crypto functions return 0 on success")
> > as that's the only commit changing ext4_htree_store_dirent, but
> > that did not help.
> > 
> > Anybody else seeing something like this?
> 
> Never seen this but I suspect it is a fallout from Al's directory locking
> changes. In particular ext4_htree_fill_tree() builds rb-tree of found
> directory entries in file->private_data (and generally modifies the
> structure stored there) but after Al's changes we don't have exclusive
> access to struct file if I'm right so if two processes end up calling
> getdents() for the same 'struct file' we are doomed.

RTFS.  We sure as hell *do* have exclusive access to struct file.  See
        /* POSIX.1-2008/SUSv4 Section XSI 2.9.7 */
        if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode))
                f->f_mode |= FMODE_ATOMIC_POS;
in do_dentry_open() as well as
        if (file && (file->f_mode & FMODE_ATOMIC_POS)) {
                if (file_count(file) > 1) {
                        v |= FDPUT_POS_UNLOCK;
                        mutex_lock(&file->f_pos_lock);
                }
        }
in __fdget_pos() and
        f = fdget_pos(fd);
        if (!f.file)
                return -EBADF;
in getdents(2).