mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Paul Bolle <pebolle@tiscali.nl>
To: Jens Axboe <jaxboe@fusionio.com>
Cc: "paulmck@linux.vnet.ibm.com" <paulmck@linux.vnet.ibm.com>,
	Vivek Goyal <vgoyal@redhat.com>,
	linux kernel mailing list <linux-kernel@vger.kernel.org>
Subject: Re: Mysterious CFQ crash and RCU
Date: Sun, 05 Jun 2011 10:39:55 +0200	[thread overview]
Message-ID: <1307263197.28359.42.camel@t41.thuisdomein> (raw)
In-Reply-To: <4DEB28A1.5090109@fusionio.com>

On Sun, 2011-06-05 at 08:56 +0200, Jens Axboe wrote:
> Does this fix it? It will introduce a hierarchy that is queue -> ioc
> lock, but as far as I can remember (and tell from a quick look), we
> don't have any dependencies on that order of locking at this moment. So
> should be OK.
> 
> diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
> index 3c7b537..fa7ef54 100644
> --- a/block/cfq-iosched.c
> +++ b/block/cfq-iosched.c
> @@ -2772,8 +2772,11 @@ static void __cfq_exit_single_io_context(struct cfq_data *cfqd,
>  	smp_wmb();
>  	cic->key = cfqd_dead_key(cfqd);
>  
> -	if (ioc->ioc_data == cic)
> +	if (ioc->ioc_data == cic) {
> +		spin_lock(&ioc->lock);
>  		rcu_assign_pointer(ioc->ioc_data, NULL);
> +		spin_unlock(&ioc->lock);
> +	}
>  
>  	if (cic->cfqq[BLK_RW_ASYNC]) {
>  		cfq_exit_cfqq(cfqd, cic->cfqq[BLK_RW_ASYNC]);
> 

0) I'd guess not, as the last thing I tried before simply ripping
io_context.ioc_data out, was:

        spin_lock_irqsave(&ioc->lock, flags);
        rcu_read_lock();
        ioc_data = rcu_dereference(ioc->ioc_data);
        rcu_read_unlock();
        if (ioc_data == cic)
                rcu_assign_pointer(ioc->ioc_data, NULL);
        spin_unlock_irqrestore(&ioc->lock, flags);

(By this time I had already wrapped all access to io_context.ioc_data in
rcu_read_lock(), rcu_dereference(), and rcu_read_unlock() voodoo. I also
wrapped all access of io_context members - other than refcount and
nr_tasks - in a spin_lock_irqsave()/spin_unlock_irqrestore() on
io_context.lock. This gave no warnings, nor lockups, but the code just
kept crashing in the exact same location it always did!)

1) Of course, by now I have forgotten what I had in mind when I stopped
working on this last night. My first bet currently is to change the core
of cic_free_func() into something like:

        spin_lock_irqsave(&ioc->lock, flags);
        radix_tree_delete(&ioc->radix_root, dead_key >>
CIC_DEAD_INDEX_SHIFT);

        rcu_read_lock();
        ioc_data = rcu_dereference(ioc->ioc_data);
        rcu_read_unlock();
        if (ioc_data == cic)
		rcu_assign_pointer(ioc->ioc_data, NULL);

        hlist_del_rcu(&cic->cic_list);
        spin_unlock_irqrestore(&ioc->lock, flags);

2) But, I must admit I'm not yet at full speed today.


Paul Bolle


  reply	other threads:[~2011-06-05  8:39 UTC|newest]

Thread overview: 47+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-05-19 22:24 Vivek Goyal
2011-05-21 21:00 ` Paul E. McKenney
2011-05-21 22:23   ` Paul Bolle
2011-05-21 23:54     ` Paul E. McKenney
2011-05-22 19:30       ` Paul Bolle
2011-05-22 20:13         ` Paul E. McKenney
2011-05-23 15:21   ` Vivek Goyal
2011-05-23 15:38     ` Paul E. McKenney
2011-05-23 22:20       ` Paul Bolle
2011-05-24  4:14         ` Paul E. McKenney
2011-05-24  9:41         ` Jens Axboe
2011-05-24 14:35           ` Paul E. McKenney
2011-05-24 14:51             ` Jens Axboe
2011-05-24 15:42               ` Paul E. McKenney
2011-05-24 15:51                 ` Paul E. McKenney
2011-05-25  8:28           ` Paul Bolle
2011-05-25  8:46             ` Jens Axboe
2011-05-25  9:13               ` Paul Bolle
2011-05-25  9:30                 ` Jens Axboe
2011-05-25  9:40                   ` Paul Bolle
2011-05-25 12:48               ` Paul Bolle
2011-05-25 12:51                 ` Jens Axboe
2011-05-25 17:28               ` Paul Bolle
2011-05-25 18:59                 ` Jens Axboe
2011-05-25 10:17       ` Paul Bolle
2011-05-25 15:33         ` Paul E. McKenney
2011-05-25 17:44           ` Paul Bolle
2011-05-25 20:40             ` Paul E. McKenney
2011-05-26  9:15       ` Paul Bolle
2011-06-03  5:07         ` Paul E. McKenney
2011-06-03 13:45           ` Vivek Goyal
2011-06-03 15:33             ` Paul E. McKenney
2011-06-03 16:54               ` Paul E. McKenney
2011-06-04 12:22             ` Paul Bolle
2011-06-04 12:50           ` Paul Bolle
2011-06-04 16:03             ` Paul E. McKenney
2011-06-04 22:48               ` Paul Bolle
2011-06-04 23:06                 ` Paul E. McKenney
2011-08-04 15:05                   ` Vivek Goyal
2011-08-04 19:43                     ` Jens Axboe
2011-08-04 19:51                       ` Vivek Goyal
2011-06-05  6:56                 ` Jens Axboe
2011-06-05  8:39                   ` Paul Bolle [this message]
2011-06-05 10:38                   ` Paul Bolle
2011-06-05 22:51                     ` Jens Axboe
2011-06-06 14:28                   ` Vivek Goyal
2011-05-23 15:36   ` Vivek Goyal

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=1307263197.28359.42.camel@t41.thuisdomein \
    --to=pebolle@tiscali.nl \
    --cc=jaxboe@fusionio.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=paulmck@linux.vnet.ibm.com \
    --cc=vgoyal@redhat.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 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®