From: Jamie Lokier <jamie@shareable.org>
To: Linus Torvalds <torvalds@osdl.org>
Cc: Ingo Molnar <mingo@elte.hu>, Tridge <tridge@samba.org>,
Al Viro <viro@parcelfarce.linux.theplanet.co.uk>,
"H. Peter Anvin" <hpa@zytor.com>,
Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: Re: explicit dcache <-> user-space cache coherency, sys_mark_dir_clean(), O_CLEAN
Date: Fri, 20 Feb 2004 17:33:07 +0000 [thread overview]
Message-ID: <20040220173307.GF8994@mail.shareable.org> (raw)
In-Reply-To: <Pine.LNX.4.58.0402200733350.1107@ppc970.osdl.org>
Linus Torvalds wrote:
> > One 'user-space cache is valid/clean' bit should be enough - where all
> > non-Samba accesses clear the 'valid bit', and Samba sets the bit
> > manually.
>
> Yes, that, together with O_CLEAN would work.
>
> The problem is that you'd still need other system calls: it's not like
> open(O_CREAT) is the only way to create a file. So you'd have to add
> versions of "link()" etc, which means that O_CLEAN is really pretty
> pointless, and you might as well just do it in a new system call.
>
> Your version is also not multi-threaded: you can never allow more than one
> thread doing the "sys_mark_dir_clean()". That was the reason for having
> two bits: so that anybody can do a lookup in parallell, and only the
> "filldir" part needs to be serialized.
How about this: we clean up dnotify, so it can be used for
user<->kernel dcache coherency, efficiently, and implement O_CLEAN in
a different way, which works with multiple threads, without extra
system calls for rename/link etc. and where the scope of "this
process/thread/whatever doesn't make a directory unclean" is flexible
enough for Samba, multi-threaded file viewers, maildir mail trackers
and so on.
Ok, marketing aside:
1. open() a directory
2. fcntl() for dnotify on the directory, as with the current interface,
but adding a flag call DN_POLL.
Normally dnotify sends a queued signal with each event. It can listen
in one-short or multiple event modes. I'm surprised a signal was ever
used, because we have a perfectly good file descriptor to read from.
So DN_POLL means "register this dnotify but don't send a signal".
Instead, you'll call fcntl() again to read the dnotify status bits.
For Samba, the dnotify is equivalent to sys_mark_dir_clean().
Samba's dcache works like this, following Ingo's logic:
repeat:
if (fcntl(dirfd, F_NOTIFY, DN_CREATE | DN_RENAME | DN_POLL) == 0) {
... pure user-space fast path, use Samba dcache ...
return;
}
... fill Samba dcache ...
readdir() loop
goto repeat;
(Note that DN_DELETE isn't needed: the negative userspace dcache
doesn't care about deletions).
See, that is obviously equivalent and uses an obvious (and tiny)
improvement to the existing dnotify feature.
The argument that O_CLEAN _requires_ sys_mark_dir_clean() is obviously
bogus: if O_CLEAN will abort when _this_ process/thread/whatever has
the directory marked clean, than it can just as easily abort when
this process/thread/whatever has a dnotify listening.
We might, however, like to add a flag DN_CLEAN so that O_CLEAN only
aborts for dnotifies with that flag set. Just to stay friendly with
libraries.
-- Jamie
next prev parent reply other threads:[~2004-02-20 17:33 UTC|newest]
Thread overview: 123+ messages / expand[flat|nested] mbox.gz Atom feed top
2004-02-17 4:12 UTF-8 and case-insensitivity tridge
2004-02-17 5:11 ` Linus Torvalds
2004-02-17 6:54 ` tridge
2004-02-17 8:33 ` Neil Brown
2004-02-17 22:48 ` tridge
2004-02-18 0:06 ` Neil Brown
2004-02-18 9:47 ` Helge Hafting
2004-02-17 15:13 ` Linus Torvalds
2004-02-17 16:57 ` Linus Torvalds
2004-02-17 19:44 ` viro
2004-02-17 20:10 ` Linus Torvalds
2004-02-17 20:17 ` viro
2004-02-17 20:23 ` Linus Torvalds
2004-02-17 21:08 ` Robin Rosenberg
2004-02-17 21:17 ` Linus Torvalds
2004-02-17 22:27 ` Robin Rosenberg
2004-02-18 3:02 ` tridge
2004-02-17 23:57 ` tridge
2004-02-17 23:20 ` tridge
2004-02-17 23:43 ` Linus Torvalds
2004-02-18 3:26 ` tridge
2004-02-18 5:33 ` H. Peter Anvin
2004-02-18 7:54 ` Marc Lehmann
2004-02-18 2:37 ` H. Peter Anvin
2004-02-18 3:03 ` Linus Torvalds
2004-02-18 3:14 ` H. Peter Anvin
2004-02-18 3:27 ` Linus Torvalds
2004-02-18 21:31 ` tridge
2004-02-18 22:23 ` Linus Torvalds
2004-02-18 22:28 ` Linus Torvalds
2004-02-18 22:50 ` tridge
2004-02-18 22:59 ` Linus Torvalds
2004-02-18 23:09 ` tridge
2004-02-18 23:16 ` Linus Torvalds
2004-02-19 8:10 ` Jamie Lokier
2004-02-19 16:09 ` Linus Torvalds
2004-02-19 16:38 ` Jamie Lokier
2004-02-19 16:54 ` Linus Torvalds
2004-02-19 18:29 ` Jamie Lokier
2004-02-19 19:48 ` Eureka! (was Re: UTF-8 and case-insensitivity) Linus Torvalds
2004-02-19 19:51 ` Linus Torvalds
2004-02-19 19:48 ` H. Peter Anvin
2004-02-19 20:04 ` Linus Torvalds
2004-02-19 20:05 ` viro
2004-02-19 20:23 ` Linus Torvalds
2004-02-19 20:32 ` Linus Torvalds
2004-02-19 20:45 ` viro
2004-02-19 21:26 ` Linus Torvalds
2004-02-19 21:38 ` Linus Torvalds
2004-02-19 21:45 ` Linus Torvalds
2004-02-19 21:43 ` viro
2004-02-19 21:53 ` Linus Torvalds
2004-02-19 22:21 ` David Lang
2004-02-19 20:48 ` Jamie Lokier
2004-02-19 21:30 ` Linus Torvalds
2004-02-20 0:00 ` Jamie Lokier
2004-02-20 0:17 ` Linus Torvalds
2004-02-20 0:24 ` Linus Torvalds
2004-02-20 0:30 ` Trond Myklebust
2004-02-20 0:54 ` Jamie Lokier
2004-02-20 0:57 ` tridge
2004-02-20 1:07 ` Paul Wagland
2004-02-20 13:31 ` Chris Wedgwood
2004-02-20 0:46 ` Jamie Lokier
2004-02-23 10:13 ` Tim Connors
2004-02-20 1:39 ` Junio C Hamano
2004-02-20 12:54 ` Jamie Lokier
2004-02-19 23:37 ` tridge
2004-02-20 0:02 ` Linus Torvalds
2004-02-20 0:16 ` tridge
2004-02-20 0:37 ` Linus Torvalds
2004-02-20 1:26 ` tridge
2004-02-20 1:07 ` H. Peter Anvin
2004-02-20 2:30 ` Theodore Ts'o
2004-02-20 12:04 ` explicit dcache <-> user-space cache coherency, sys_mark_dir_clean(), O_CLEAN Ingo Molnar
2004-02-20 13:19 ` Jamie Lokier
2004-02-20 13:37 ` Ingo Molnar
2004-02-20 14:00 ` Ingo Molnar
2004-02-20 16:31 ` Jamie Lokier
2004-02-20 13:23 ` [patch] " Ingo Molnar
2004-02-20 18:00 ` viro
2004-02-20 15:41 ` Linus Torvalds
2004-02-20 17:04 ` Ingo Molnar
2004-02-20 17:19 ` Linus Torvalds
2004-02-20 18:48 ` Ingo Molnar
2004-02-21 1:44 ` Jamie Lokier
2004-02-21 7:58 ` Ingo Molnar
2004-02-21 8:04 ` viro
2004-02-21 17:46 ` Ingo Molnar
2004-02-21 18:15 ` Linus Torvalds
2004-02-21 8:26 ` Keith Owens
2004-02-23 10:59 ` Pavel Machek
2004-02-23 13:55 ` Jamie Lokier
2004-02-23 16:45 ` Ingo Molnar
2004-02-23 17:32 ` Jamie Lokier
2004-02-20 23:00 ` tridge
2004-02-20 17:33 ` Jamie Lokier [this message]
2004-02-20 18:22 ` Linus Torvalds
2004-02-21 0:38 ` Jamie Lokier
2004-02-21 1:10 ` Linus Torvalds
2004-02-21 3:01 ` Jamie Lokier
2004-02-20 17:47 ` Jamie Lokier
2004-02-20 20:38 ` Christer Weinigel
2004-02-22 15:07 ` Jamie Lokier
2004-02-22 16:55 ` Miquel van Smoorenburg
2004-02-19 19:08 ` UTF-8 and case-insensitivity Helge Hafting
2004-02-18 4:08 ` tridge
2004-02-18 10:05 ` Robin Rosenberg
2004-02-18 11:43 ` tridge
2004-02-18 12:31 ` Robin Rosenberg
2004-02-18 16:48 ` H. Peter Anvin
2004-02-18 20:00 ` H. Peter Anvin
2004-02-19 2:53 ` Daniel Newby
2004-02-17 5:25 ` Tim Connors
2004-02-17 7:43 ` H. Peter Anvin
2004-02-17 8:05 ` H. Peter Anvin
2004-02-17 14:25 ` Dave Kleikamp
2004-02-18 0:16 ` Robert White
2004-02-18 0:20 ` Linus Torvalds
2004-02-18 1:03 ` Robert White
2004-02-18 21:48 ` Ville Herva
2004-02-18 2:48 ` tridge
2004-02-18 20:56 ` Robert White
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=20040220173307.GF8994@mail.shareable.org \
--to=jamie@shareable.org \
--cc=hpa@zytor.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@elte.hu \
--cc=torvalds@osdl.org \
--cc=tridge@samba.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox