Date: Thu, 13 Jun 2024 10:38:47 +0300
From: Amir Goldstein <amir73il@...il.com>
To: NeilBrown <neilb@...e.de>
Cc: Alexander Viro <viro@...iv.linux.org.uk>, Christian Brauner <brauner@...nel.org>, Jan Kara <jack@...e.cz>, 
	James Clark <james.clark@....com>, ltp@...ts.linux.it, linux-nfs@...r.kernel.org, 
	LKML <linux-kernel@...r.kernel.org>, linux-fsdevel@...r.kernel.org, 
	Miklos Szeredi <miklos@...redi.hu>
Subject: Re: [PATCH v2] VFS: generate FS_CREATE before FS_OPEN when
 ->atomic_open used.

On Wed, Jun 12, 2024 at 2:47 PM NeilBrown <neilb@...e.de> wrote:
>
> On Wed, 12 Jun 2024, Amir Goldstein wrote:
> > On Wed, Jun 12, 2024 at 10:10 AM NeilBrown <neilb@...e.de> wrote:
> > >
> > >
> > > When a file is opened and created with open(..., O_CREAT) we get
> > > both the CREATE and OPEN fsnotify events and would expect them in that
> > > order.   For most filesystems we get them in that order because
> > > open_last_lookups() calls fsnofify_create() and then do_open() (from
> > > path_openat()) calls vfs_open()->do_dentry_open() which calls
> > > fsnotify_open().
> > >
> > > However when ->atomic_open is used, the
> > >    do_dentry_open() -> fsnotify_open()
> > > call happens from finish_open() which is called from the ->atomic_open
> > > handler in lookup_open() which is called *before* open_last_lookups()
> > > calls fsnotify_create.  So we get the "open" notification before
> > > "create" - which is backwards.  ltp testcase inotify02 tests this and
> > > reports the inconsistency.
> > >
> > > This patch lifts the fsnotify_open() call out of do_dentry_open() and
> > > places it higher up the call stack.  There are three callers of
> > > do_dentry_open().
> > >
> > > For vfs_open() and kernel_file_open() the fsnotify_open() is placed
> > > directly in that caller so there should be no behavioural change.
> > >
> > > For finish_open() there are two cases:
> > >  - finish_open is used in ->atomic_open handlers.  For these we add a
> > >    call to fsnotify_open() at the top of do_open() if FMODE_OPENED is
> > >    set - which means do_dentry_open() has been called.
> > >  - finish_open is used in ->tmpfile() handlers.  For these a similar
> > >    call to fsnotify_open() is added to vfs_tmpfile()
> >
> > Any handlers other than ovl_tmpfile()?
>
> Local filesystems tend to call finish_open_simple() which is a trivial
> wrapper around finish_open().
> Every .tmpfile handler calls either finish_open() or finish_open_simple().
>
> > This one is a very recent and pretty special case.
> > Did open(O_TMPFILE) used to emit an OPEN event before that change?
>
> I believe so, yes.
>

Right. Thanks for clarifying.

> Thanks,
> NeilBrown
>
> >
> > >
> > > With this patch NFSv3 is restored to its previous behaviour (before
> > > ->atomic_open support was added) of generating CREATE notifications
> > > before OPEN, and NFSv4 now has that same correct ordering that is has
> > > not had before.  I haven't tested other filesystems.
> > >
> > > Fixes: 7c6c5249f061 ("NFS: add atomic_open for NFSv3 to handle O_TRUNC correctly.")
> > > Reported-by: James Clark <james.clark@....com>
> > > Closes: https://lore.kernel.org/all/01c3bf2e-eb1f-4b7f-a54f-d2a05dd3d8c8@arm.com
> > > Signed-off-by: NeilBrown <neilb@...e.de>
> > > ---
> > >  fs/namei.c |  5 +++++
> > >  fs/open.c  | 19 ++++++++++++-------
> > >  2 files changed, 17 insertions(+), 7 deletions(-)
> > >
> > > diff --git a/fs/namei.c b/fs/namei.c
> > > index 37fb0a8aa09a..057afacc4b60 100644
> > > --- a/fs/namei.c
> > > +++ b/fs/namei.c
> > > @@ -3612,6 +3612,9 @@ static int do_open(struct nameidata *nd,
> > >         int acc_mode;
> > >         int error;
> > >
> > > +       if (file->f_mode & FMODE_OPENED)
> > > +               fsnotify_open(file);
> > > +
> > >         if (!(file->f_mode & (FMODE_OPENED | FMODE_CREATED))) {
> > >                 error = complete_walk(nd);
> > >                 if (error)
> > > @@ -3700,6 +3703,8 @@ int vfs_tmpfile(struct mnt_idmap *idmap,
> > >         mode = vfs_prepare_mode(idmap, dir, mode, mode, mode);
> > >         error = dir->i_op->tmpfile(idmap, dir, file, mode);
> > >         dput(child);
> > > +       if (file->f_mode & FMODE_OPENED)
> > > +               fsnotify_open(file);
> > >         if (error)
> > >                 return error;
> > >         /* Don't check for other permissions, the inode was just created */
> > > diff --git a/fs/open.c b/fs/open.c
> > > index 89cafb572061..970f299c0e77 100644
> > > --- a/fs/open.c
> > > +++ b/fs/open.c
> > > @@ -1004,11 +1004,6 @@ static int do_dentry_open(struct file *f,
> > >                 }
> > >         }
> > >
> > > -       /*
> > > -        * Once we return a file with FMODE_OPENED, __fput() will call
> > > -        * fsnotify_close(), so we need fsnotify_open() here for symmetry.
> > > -        */
> > > -       fsnotify_open(f);
> > >         return 0;
> > >
> > >  cleanup_all:
> > > @@ -1085,8 +1080,17 @@ EXPORT_SYMBOL(file_path);
> > >   */
> > >  int vfs_open(const struct path *path, struct file *file)
> > >  {
> > > +       int ret;
> > > +
> > >         file->f_path = *path;
> > > -       return do_dentry_open(file, NULL);
> > > +       ret = do_dentry_open(file, NULL);
> > > +       if (!ret)
> > > +               /*
> > > +                * Once we return a file with FMODE_OPENED, __fput() will call
> > > +                * fsnotify_close(), so we need fsnotify_open() here for symmetry.
> > > +                */
> > > +               fsnotify_open(file);
> >
> > I agree that this change preserves the logic, but (my own) comment
> > above is inconsistent with the case of:
> >
> >         if ((f->f_flags & O_DIRECT) && !(f->f_mode & FMODE_CAN_ODIRECT))
> >                 return -EINVAL;
> >
> > Which does set FMODE_OPENED, but does not emit an OPEN event.
>
> If I understand correctly, that case doesn't emit an OPEN event before
> my patch, but will result in a CLOSE event.
> After my patch ... I think it still doesn't emit OPEN.
>
> I wonder if, instead of adding the the fsnotify_open() in do_open(), we
> should put it in the\
>         if (file->f_mode & (FMODE_OPENED | FMODE_CREATED)) {
> case of open_last_lookups().
>

We cannot do that.
See the reasoning for 7b8c9d7bb457 ("fsnotify: move fsnotify_open() hook into
do_dentry_open()") - we need the events for other callers of vfs_open(),
like overlayfs and nfsd.

> Or maybe it really doesn't hurt to have a CLOSE event without and OPEN.
> OPEN without CLOSE would be problematic, but the other way around
> shouldn't matter....  It feels untidy, but maybe we don't care.
>

We have had unmatched CLOSE events for a very long time before
7b8c9d7bb457 ("fsnotify: move fsnotify_open() hook into do_dentry_open()")
and I do not know of any complaints.

When I made this change, its purpose was not to match all OPEN/CLOSE
but to add missing OPEN events. However, I did try to avoid unmatched
CLOSE at least for the common cases.

Thanks,
Amir.

Powered by blists - more mailing lists