From: Herbert Xu <herbert@gondor.apana.org.au>
To: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>,
akpm@linux-foundation.org,
Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
linuxppc-dev@ozlabs.org, Ingo Molnar <mingo@elte.hu>,
Thomas Gleixner <tglx@linutronix.de>
Subject: [IRQ]: Fix synchronize_irq races with IRQ handler
Date: Tue, 23 Oct 2007 11:26:25 +0800 [thread overview]
Message-ID: <20071023032625.GA26559@gondor.apana.org.au> (raw)
In-Reply-To: <1193001005.6745.34.camel@pasglop>
On Mon, Oct 22, 2007 at 07:10:05AM +1000, Benjamin Herrenschmidt wrote:
>
> Hrm... not on yet. Herbert, care to resend, looks like it fell down the
> wrong hole in Linus mailbox :-)
Thanks for the reminder Ben. Here it is again:
[IRQ]: Fix synchronize_irq races with IRQ handler
As it is some callers of synchronize_irq rely on memory barriers
to provide synchronisation against the IRQ handlers. For example,
the tg3 driver does
tp->irq_sync = 1;
smp_mb();
synchronize_irq();
and then in the IRQ handler:
if (!tp->irq_sync)
netif_rx_schedule(dev, &tp->napi);
Unfortunately memory barriers only work well when they come in
pairs. Because we don't actually have memory barriers on the
IRQ path, the memory barrier before the synchronize_irq() doesn't
actually protect us.
In particular, synchronize_irq() may return followed by the
result of netif_rx_schedule being made visible.
This patch (mostly written by Linus) fixes this by using spin
locks instead of memory barries on the synchronize_irq() path.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cheers,
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
--
diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c
index 80eab7a..1f31422 100644
--- a/kernel/irq/manage.c
+++ b/kernel/irq/manage.c
@@ -29,12 +29,28 @@
void synchronize_irq(unsigned int irq)
{
struct irq_desc *desc = irq_desc + irq;
+ unsigned int status;
if (irq >= NR_IRQS)
return;
- while (desc->status & IRQ_INPROGRESS)
- cpu_relax();
+ do {
+ unsigned long flags;
+
+ /*
+ * Wait until we're out of the critical section. This might
+ * give the wrong answer due to the lack of memory barriers.
+ */
+ while (desc->status & IRQ_INPROGRESS)
+ cpu_relax();
+
+ /* Ok, that indicated we're done: double-check carefully. */
+ spin_lock_irqsave(&desc->lock, flags);
+ status = desc->status;
+ spin_unlock_irqrestore(&desc->lock, flags);
+
+ /* Oops, that failed? */
+ } while (status & IRQ_INPROGRESS);
}
EXPORT_SYMBOL(synchronize_irq);
next prev parent reply other threads:[~2007-10-23 3:27 UTC|newest]
Thread overview: 48+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-10-18 1:25 [PATCH] synchronize_irq needs a barrier Benjamin Herrenschmidt
2007-10-18 1:45 ` Andrew Morton
2007-10-18 1:55 ` Benjamin Herrenschmidt
2007-10-18 2:12 ` Linus Torvalds
2007-10-18 2:40 ` Benjamin Herrenschmidt
2007-10-18 2:57 ` Benjamin Herrenschmidt
2007-10-18 14:56 ` Herbert Xu
2007-10-18 22:05 ` Benjamin Herrenschmidt
2007-10-18 22:52 ` Linus Torvalds
2007-10-18 23:17 ` Benjamin Herrenschmidt
2007-10-18 23:39 ` Linus Torvalds
2007-10-18 23:52 ` Benjamin Herrenschmidt
2007-10-19 2:32 ` Herbert Xu
2007-10-19 2:52 ` Nick Piggin
2007-10-19 3:28 ` Herbert Xu
2007-10-19 4:49 ` Nick Piggin
2007-10-19 2:55 ` Linus Torvalds
2007-10-19 3:26 ` Linus Torvalds
2007-10-19 4:11 ` Benjamin Herrenschmidt
2007-10-19 4:26 ` Benjamin Herrenschmidt
2007-10-19 5:53 ` Herbert Xu
2007-10-19 4:20 ` Herbert Xu
2007-10-19 4:29 ` Benjamin Herrenschmidt
2007-10-19 4:35 ` Benjamin Herrenschmidt
2007-10-19 4:48 ` Herbert Xu
2007-10-19 4:58 ` Benjamin Herrenschmidt
2007-10-21 21:10 ` Benjamin Herrenschmidt
2007-10-23 3:26 ` Herbert Xu [this message]
2007-10-19 5:36 ` [NET]: Fix possible dev_deactivate race condition Herbert Xu
2007-10-19 5:38 ` David Miller
2007-10-19 7:35 ` Peter Zijlstra
2007-10-19 9:29 ` Herbert Xu
2007-10-18 14:35 ` [PATCH] synchronize_irq needs a barrier Herbert Xu
2007-10-18 21:35 ` Benjamin Herrenschmidt
2007-10-20 2:02 ` Maxim Levitsky
2007-10-20 2:25 ` Linus Torvalds
2007-10-20 3:10 ` Maxim Levitsky
2007-10-20 4:06 ` Benjamin Herrenschmidt
2007-10-20 4:04 ` Benjamin Herrenschmidt
2007-10-20 4:09 ` Benjamin Herrenschmidt
2007-10-20 3:37 ` Herbert Xu
2007-10-20 3:56 ` Benjamin Herrenschmidt
2007-10-20 4:24 ` Maxim Levitsky
2007-10-20 5:04 ` Benjamin Herrenschmidt
2007-10-20 5:36 ` Maxim Levitsky
2007-10-20 5:46 ` Benjamin Herrenschmidt
2007-10-20 6:06 ` Maxim Levitsky
2007-10-20 6:13 ` Benjamin Herrenschmidt
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=20071023032625.GA26559@gondor.apana.org.au \
--to=herbert@gondor.apana.org.au \
--cc=akpm@linux-foundation.org \
--cc=benh@kernel.crashing.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linuxppc-dev@ozlabs.org \
--cc=mingo@elte.hu \
--cc=tglx@linutronix.de \
--cc=torvalds@linux-foundation.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;
as well as URLs for NNTP newsgroup(s).