From: Davide Libenzi <davidel@xmailserver.org>
To: Gregory Haskins <ghaskins@novell.com>
Cc: mst@redhat.com, kvm@vger.kernel.org,
Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
avi@redhat.com, paulmck@linux.vnet.ibm.com,
Ingo Molnar <mingo@elte.hu>
Subject: Re: [PATCH 3/3] eventfd: add internal reference counting to fix notifier race conditions
Date: Fri, 19 Jun 2009 12:10:48 -0700 (PDT) [thread overview]
Message-ID: <alpine.DEB.1.10.0906191156580.14884@makko.or.mcafeemobile.com> (raw)
In-Reply-To: <20090619185138.31118.14916.stgit@dev.haskins.net>
On Fri, 19 Jun 2009, Gregory Haskins wrote:
> eventfd currently emits a POLLHUP wakeup on f_ops->release() to generate a
> notifier->release() callback. This lets notification clients know if
> the eventfd is about to go away and is very useful particularly for
> in-kernel clients. However, as it stands today it is not possible to
> use the notification API in a race-free way. This patch adds some
> additional logic to the notification subsystem to rectify this problem.
>
> Background:
> -----------------------
> Eventfd currently only has one reference count mechanism: fget/fput. This
> in of itself is normally fine. However, if a client expects to be
> notified if the eventfd is closed, it cannot hold a fget() reference
> itself or the underlying f_ops->release() callback will never be invoked
> by VFS. Therefore we have this somewhat unusual situation where we may
> hold a pointer to an eventfd object (by virtue of having a waiter registered
> in its wait-queue), but no reference. This makes it nearly impossible to
> design a mutual decoupling algorithm: you cannot unhook one side from the
> other (or vice versa) without racing.
And why is that?
struct xxx {
struct mutex mtx;
struct file *file;
...
};
struct file *xxx_get_file(struct xxx *x) {
struct file *file;
mutex_lock(&x->mtx);
file = x->file;
if (!file)
mutex_unlock(&x->mtx);
return file;
}
void xxx_release_file(struct xxx *x) {
mutex_unlock(&x->mtx);
}
void handle_POLLHUP(struct xxx *x) {
struct file *file;
file = xxx_get_file(x);
if (file) {
unhook_waitqueue(file, ...);
x->file = NULL;
xxx_release_file(x);
}
}
Every time you need to "use" file, you call xxx_get_file(), and if you get
NULL, it means it's gone and you handle it accordigly to your IRQ fd
policies. As soon as you done with the file, you call xxx_release_file().
Replace "mtx" with the lock that fits your needs.
- Davide
next prev parent reply other threads:[~2009-06-19 19:17 UTC|newest]
Thread overview: 104+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-06-16 2:29 [KVM-RFC PATCH 0/2] eventfd enhancements for irqfd/iosignalfd Gregory Haskins
2009-06-16 2:29 ` [KVM-RFC PATCH 1/2] eventfd: add an explicit srcu based notifier interface Gregory Haskins
2009-06-16 14:02 ` Michael S. Tsirkin
2009-06-16 14:11 ` Gregory Haskins
2009-06-16 14:38 ` Michael S. Tsirkin
2009-06-16 14:48 ` Gregory Haskins
2009-06-16 14:54 ` Gregory Haskins
2009-06-16 15:16 ` Michael S. Tsirkin
2009-06-16 14:55 ` Michael S. Tsirkin
2009-06-16 15:20 ` Gregory Haskins
2009-06-16 15:41 ` Michael S. Tsirkin
2009-06-16 16:17 ` Gregory Haskins
2009-06-16 16:19 ` Davide Libenzi
2009-06-16 17:01 ` Gregory Haskins
2009-06-17 16:38 ` Davide Libenzi
2009-06-17 17:28 ` Gregory Haskins
2009-06-17 17:44 ` Davide Libenzi
2009-06-17 19:17 ` Gregory Haskins
2009-06-17 19:50 ` Davide Libenzi
2009-06-17 21:48 ` Gregory Haskins
2009-06-17 23:21 ` Davide Libenzi
2009-06-18 6:23 ` Michael S. Tsirkin
2009-06-18 17:52 ` Davide Libenzi
2009-06-18 14:01 ` Gregory Haskins
2009-06-18 14:01 ` Gregory Haskins
2009-06-18 17:44 ` Davide Libenzi
2009-06-18 19:04 ` Gregory Haskins
2009-06-18 19:04 ` Gregory Haskins
2009-06-18 22:03 ` Davide Libenzi
2009-06-18 22:47 ` Gregory Haskins
2009-06-18 22:47 ` Gregory Haskins
2009-06-19 18:51 ` Gregory Haskins
2009-06-19 18:51 ` [PATCH 1/3] eventfd: Allow waiters to be notified about the eventfd file* going away Gregory Haskins
2009-06-19 18:51 ` [PATCH 2/3] eventfd: add generalized notifier interface Gregory Haskins
2009-06-19 18:51 ` [PATCH 3/3] eventfd: add internal reference counting to fix notifier race conditions Gregory Haskins
2009-06-19 19:10 ` Davide Libenzi [this message]
2009-06-19 21:16 ` Gregory Haskins
2009-06-19 21:26 ` Davide Libenzi
2009-06-19 21:49 ` Gregory Haskins
2009-06-19 21:54 ` Davide Libenzi
2009-06-19 22:47 ` Davide Libenzi
2009-06-20 2:09 ` Gregory Haskins
2009-06-20 21:17 ` Davide Libenzi
2009-06-20 22:11 ` Davide Libenzi
2009-06-20 23:48 ` Davide Libenzi
2009-06-21 1:14 ` Gregory Haskins
2009-06-21 16:51 ` Davide Libenzi
2009-06-21 18:39 ` Gregory Haskins
2009-06-21 23:54 ` Davide Libenzi
2009-06-22 16:05 ` Gregory Haskins
2009-06-22 16:05 ` Gregory Haskins
2009-06-22 17:01 ` Davide Libenzi
2009-06-22 17:43 ` Gregory Haskins
2009-06-22 17:43 ` Gregory Haskins
2009-06-22 18:03 ` Davide Libenzi
2009-06-22 18:31 ` Gregory Haskins
2009-06-22 18:31 ` Gregory Haskins
2009-06-22 18:40 ` Davide Libenzi
2009-06-22 18:41 ` Michael S. Tsirkin
2009-06-22 18:51 ` Davide Libenzi
2009-06-22 19:05 ` Michael S. Tsirkin
2009-06-22 19:26 ` Gregory Haskins
2009-06-22 19:29 ` Davide Libenzi
2009-06-22 20:06 ` Gregory Haskins
2009-06-22 22:53 ` Davide Libenzi
2009-06-23 1:03 ` Gregory Haskins
2009-06-23 1:17 ` Davide Libenzi
2009-06-23 1:26 ` Gregory Haskins
2009-06-23 1:26 ` Gregory Haskins
2009-06-23 14:29 ` Davide Libenzi
2009-06-23 14:37 ` Gregory Haskins
2009-06-23 14:37 ` Gregory Haskins
2009-06-23 14:35 ` Davide Libenzi
2009-06-23 14:42 ` Gregory Haskins
2009-06-23 14:42 ` Gregory Haskins
2009-06-23 15:04 ` Michael S. Tsirkin
2009-06-22 20:28 ` Michael S. Tsirkin
2009-06-22 19:16 ` Gregory Haskins
2009-06-22 19:54 ` Davide Libenzi
2009-06-24 3:25 ` Rusty Russell
2009-06-24 22:45 ` Davide Libenzi
2009-06-25 11:42 ` Rusty Russell
2009-06-25 16:34 ` Davide Libenzi
2009-06-25 17:32 ` Gregory Haskins
2009-06-25 18:26 ` Michael S. Tsirkin
2009-06-25 18:41 ` Gregory Haskins
2009-06-26 11:23 ` Michael S. Tsirkin
2009-06-23 3:25 ` Rusty Russell
2009-06-23 14:31 ` Davide Libenzi
2009-06-25 0:19 ` Davide Libenzi
2009-06-21 1:05 ` Gregory Haskins
2009-06-16 17:54 ` [KVM-RFC PATCH 1/2] eventfd: add an explicit srcu based notifier interface Michael S. Tsirkin
2009-06-16 18:09 ` Gregory Haskins
2009-06-17 14:45 ` Michael S. Tsirkin
2009-06-17 15:02 ` Gregory Haskins
2009-06-17 16:25 ` Michael S. Tsirkin
2009-06-17 16:41 ` Gregory Haskins
2009-06-16 14:17 ` Gregory Haskins
2009-06-16 14:22 ` Gregory Haskins
2009-06-16 14:40 ` Gregory Haskins
2009-06-16 14:46 ` Michael S. Tsirkin
2009-06-18 9:03 ` Avi Kivity
2009-06-18 11:43 ` Gregory Haskins
2009-06-16 2:30 ` [KVM-RFC PATCH 2/2] eventfd: add module reference counting support for registered notifiers Gregory Haskins
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=alpine.DEB.1.10.0906191156580.14884@makko.or.mcafeemobile.com \
--to=davidel@xmailserver.org \
--cc=avi@redhat.com \
--cc=ghaskins@novell.com \
--cc=kvm@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@elte.hu \
--cc=mst@redhat.com \
--cc=paulmck@linux.vnet.ibm.com \
/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.