From: Jonathan Corbet <corbet@lwn.net>
To: Jarek Poplawski <jarkao2@gmail.com>
Cc: "Markus Trippelsdorf" <markus@trippelsdorf.de>,
"Ilpo Järvinen" <ilpo.jarvinen@helsinki.fi>,
Netdev <netdev@vger.kernel.org>,
LKML <linux-kernel@vger.kernel.org>
Subject: Re: WARNING: at net/ipv4/tcp_input.c:2927 tcp_ack+0xd55/0x1991()
Date: Mon, 30 Mar 2009 08:01:50 -0600 [thread overview]
Message-ID: <20090330080150.724378b0@bike.lwn.net> (raw)
In-Reply-To: <20090330085107.GA5822@ff.dom.local>
On Mon, 30 Mar 2009 08:51:07 +0000
Jarek Poplawski <jarkao2@gmail.com> wrote:
> Probably I miss something, but generally in a case like this "a_lock"
> doesn't have to be taken in IRQ mode to be dangerous. Eg. if one cpu
> is trying to take this lock after fasync_lock (with IRQs disabled),
> while another cpu is waiting for fasync_lock in IRQ, which preempted
> such "a_lock".
The possibility exists, I guess, yes.
> Could you give some details of this fix?
I just reverse the order of lock acquisition in fasync_helper(). Patch
is attached. I'll be sending up a pull request shortly.
jon
>From 4a6a4499693a419a20559c41e33a7bd70bf20a6f Mon Sep 17 00:00:00 2001
From: Jonathan Corbet <corbet@lwn.net>
Date: Fri, 27 Mar 2009 12:24:31 -0600
Subject: [PATCH] Fix a lockdep warning in fasync_helper()
Lockdep gripes if file->f_lock is taken in a no-IRQ situation, since that
is not always the case. We don't really want to disable IRQs for every
acquisition of f_lock; instead, just move it outside of fasync_lock.
Reported-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Reported-by: Larry Finger <Larry.Finger@lwfinger.net>
Reported-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
---
fs/fcntl.c | 10 +++++++---
include/linux/fs.h | 2 +-
2 files changed, 8 insertions(+), 4 deletions(-)
diff --git a/fs/fcntl.c b/fs/fcntl.c
index d865ca6..cc8e4de 100644
--- a/fs/fcntl.c
+++ b/fs/fcntl.c
@@ -531,6 +531,12 @@ int fasync_helper(int fd, struct file * filp, int on, struct fasync_struct **fap
if (!new)
return -ENOMEM;
}
+
+ /*
+ * We need to take f_lock first since it's not an IRQ-safe
+ * lock.
+ */
+ spin_lock(&filp->f_lock);
write_lock_irq(&fasync_lock);
for (fp = fapp; (fa = *fp) != NULL; fp = &fa->fa_next) {
if (fa->fa_file == filp) {
@@ -555,14 +561,12 @@ int fasync_helper(int fd, struct file * filp, int on, struct fasync_struct **fap
result = 1;
}
out:
- /* Fix up FASYNC bit while still holding fasync_lock */
- spin_lock(&filp->f_lock);
if (on)
filp->f_flags |= FASYNC;
else
filp->f_flags &= ~FASYNC;
- spin_unlock(&filp->f_lock);
write_unlock_irq(&fasync_lock);
+ spin_unlock(&filp->f_lock);
return result;
}
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 7428c6d..2f13c1d 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -848,7 +848,7 @@ struct file {
#define f_dentry f_path.dentry
#define f_vfsmnt f_path.mnt
const struct file_operations *f_op;
- spinlock_t f_lock; /* f_ep_links, f_flags */
+ spinlock_t f_lock; /* f_ep_links, f_flags, no IRQ */
atomic_long_t f_count;
unsigned int f_flags;
fmode_t f_mode;
--
1.6.2
next prev parent reply other threads:[~2009-03-30 14:02 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-03-27 21:12 Markus Trippelsdorf
2009-03-27 23:05 ` Ilpo Järvinen
2009-03-28 4:50 ` Markus Trippelsdorf
2009-03-28 8:29 ` Ilpo Järvinen
2009-03-28 9:55 ` Markus Trippelsdorf
2009-03-28 13:56 ` Jonathan Corbet
2009-03-30 8:51 ` Jarek Poplawski
2009-03-30 14:01 ` Jonathan Corbet [this message]
2009-03-30 16:40 ` Jarek Poplawski
2009-03-30 16:01 ` Ilpo Järvinen
2009-03-30 16:40 ` Markus Trippelsdorf
2009-03-30 18:52 ` Ilpo Järvinen
2009-03-31 7:10 ` Markus Trippelsdorf
2009-03-31 9:16 ` Ilpo Järvinen
2009-03-31 18:49 ` Markus Trippelsdorf
2009-03-31 19:12 ` Ilpo Järvinen
2009-04-01 7:00 ` Ilpo Järvinen
2009-04-01 11:09 ` Ilpo Järvinen
2009-04-01 11:35 ` Markus Trippelsdorf
2009-04-02 9:15 ` [PATCH 1/2] tcp: add helper for counter tweaking due mid-wq change Ilpo Järvinen
2009-04-02 9:18 ` [PATCH 2/2] tcp: miscounts due to tcp_fragment pcount reset Ilpo Järvinen
2009-04-02 23:32 ` David Miller
2009-04-02 23:32 ` [PATCH 1/2] tcp: add helper for counter tweaking due mid-wq change David Miller
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=20090330080150.724378b0@bike.lwn.net \
--to=corbet@lwn.net \
--cc=ilpo.jarvinen@helsinki.fi \
--cc=jarkao2@gmail.com \
--cc=linux-kernel@vger.kernel.org \
--cc=markus@trippelsdorf.de \
--cc=netdev@vger.kernel.org \
/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
all inboxes | Powered by JetHome®