From: John McCutchan <ttb@tentacle.dhs.org>
To: viro@parcelfarce.linux.theplanet.co.uk
Cc: nautilus-list@gnome.org, linux-kernel@vger.kernel.org
Subject: Re: [RFC/PATCH] inotify -- a dnotify replacement
Date: Tue, 11 May 2004 15:02:31 -0400 [thread overview]
Message-ID: <20040511190228.GA12609@tentacle.dhs.org> (raw)
In-Reply-To: <20040511124647.GE17014@parcelfarce.linux.theplanet.co.uk>
On Tue, May 11, 2004 at 01:46:47PM +0100, viro@parcelfarce.linux.theplanet.co.uk wrote:
> On Tue, May 11, 2004 at 08:20:01AM -0400, John McCutchan wrote:
>
> > Inotify will support watching a hierarchy. The reason it was not
> > implemented yet is because the one app that I really care about is
> > nautilus and the maintainer of it says he doesn't care.
>
> How are you going to implement that?
>From a quick glance at someone elses implementation of it, I plan on
walking up the dentries and checking at each level if a watcher on that
level is interested in events from subdirectories. Is this good practice in
the kernel?
> > The big feature that inotify is trying to provide is not having to keep
> > a file open (So that unmounting is not affected). I asked for some
> > guidance from people more familiar with the kernel so that I can
> > implement this feature, it requires changes made to the inode cache, and
> > how unmounting is done.
>
> Bzzert. First of all, on quite a few filesystems inumbers are stable
> only when object is pinned down. What's more, if it's not pinned down
> you've got nothing even remotely resembling a reliable way to tell if
> two events had happened to the same object - inumbers can be reused.
The inode will be pinned down, I haven't implemented this yet but I am
going to change the inode cache (is this the right place? )
so that if inode->watcher_count > 0 the inode stays pinned. Then when
the filesystem is unmounted, we will kick off all the watchers on
each inode.
>
> Besides, your "doesn't pin down" is racy as hell - not to mention the
> way you manage the lists, pretty much every function is an exploitable
> hole. Hell, just take a look at your "find inode" stuff - you grab
> superblock, find an inode by inumber (great idea, that - especially
> since on a bunch of filesystems it will get you BUG() or equivalent)
> then drop refernce to superblock (at which point it can be destroyed by
> umount()) _and_ do iput() (which will do lovely, lovely things if that
> umount did happen). Moreover, you return a pointer to inode, even
> though there's nothing to hold it alive anymore. And dereference that
> pointer later on, not caring if it had been freed/reused/whatever.
Like I said above,as long as an inode has a watcher it will be pinned. As
for the races, I plan on implementing locking around all of the list operations.
Perhaps I wasn't very clear that this is very much a WIP and lots of work is
needed.
>
> Overall: hopeless crap. And that's a direct result of your main feature -
> it's really broken by design.
Having directory event notification without needing to keep a file open on the
device is not broken by design. It is the only reasonable solution to
a problem that needs fixing. You can't simply say that a file manager
needing to be notified when directories change is broken. How would
you solve this problem?
John
next prev parent reply other threads:[~2004-05-11 19:04 UTC|newest]
Thread overview: 29+ messages / expand[flat|nested] mbox.gz Atom feed top
2004-05-10 1:35 [RFC/PATCH] inotify -- a dnotify replacement John McCutchan
2004-05-10 2:11 ` Chris Wedgwood
2004-05-10 22:17 ` John McCutchan
2004-05-10 22:31 ` Davide Libenzi
2004-05-10 22:41 ` John McCutchan
2004-05-10 22:52 ` Davide Libenzi
2004-05-10 23:10 ` Valdis.Kletnieks
2004-05-10 23:42 ` Davide Libenzi
2004-05-11 4:11 ` Chris Wedgwood
2004-05-11 2:00 ` Ian Kent
2004-05-11 2:47 ` Chris Wedgwood
2004-05-11 11:52 ` nf
2004-05-11 12:17 ` Stephen Rothwell
2004-05-11 12:24 ` John McCutchan
[not found] ` <1084278605.3839.47.camel@carados.180sw.com>
[not found] ` <1084885604.4062.48.camel@lilota.lamp.priv>
[not found] ` <1085127066.20393.440.camel@localhost.localdomain>
2004-05-21 12:04 ` nf
2004-05-11 12:20 ` John McCutchan
2004-05-11 12:46 ` viro
2004-05-11 19:02 ` John McCutchan [this message]
2004-05-11 20:28 ` carbonated beverage
2004-05-11 21:28 ` John McCutchan
2004-05-11 15:02 ` Alexander Larsson
2004-05-11 21:41 ` Chris Wedgwood
2004-05-12 12:38 ` Jörn Engel
2004-05-13 15:36 ` raven
2004-05-13 19:04 ` Chris Wedgwood
2004-05-14 7:04 ` Ian Kent
2004-05-13 21:24 ` John McCutchan
2004-05-14 2:02 ` Ian Kent
2004-05-15 4:52 ` raven
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20040511190228.GA12609@tentacle.dhs.org \
--to=ttb@tentacle.dhs.org \
--cc=linux-kernel@vger.kernel.org \
--cc=nautilus-list@gnome.org \
--cc=viro@parcelfarce.linux.theplanet.co.uk \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.