netfilter-devel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jon Masters <jonathan@jonmasters.org>
To: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Patrick McHardy <kaber@trash.net>,
	Eric Dumazet <eric.dumazet@gmail.com>,
	linux-kernel <linux-kernel@vger.kernel.org>,
	netdev <netdev@vger.kernel.org>,
	netfilter-devel <netfilter-devel@vger.kernel.org>,
	"Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Subject: Re: [PATCH] netfilter: per netns nf_conntrack_cachep
Date: Tue, 02 Feb 2010 13:34:36 -0500	[thread overview]
Message-ID: <1265135676.2861.196.camel@tonnant> (raw)
In-Reply-To: <1265134598.2861.191.camel@tonnant>

On Tue, 2010-02-02 at 13:16 -0500, Jon Masters wrote:
> On Tue, 2010-02-02 at 19:58 +0200, Alexey Dobriyan wrote:
> 
> > Yes, moving to init_net-only function is fine.
> 
> So moving the "setup up fake conntrack" bits to init_init_net from
> init_net still results in the panic, which means that the use count
> really is dropping to zero and we really are trying to free it when
> using multiple namespaces. Per ns is probably an easier way to go.
> 
> Just for kicks, I'll have it error out on attempting to free to see if I
> can get this box to stay up for a while.

Confirmed. It boots and the hashsize is not inadvertedly corrupted if I
do the following:

void nf_conntrack_destroy(struct nf_conntrack *nfct)
{
        void (*destroy)(struct nf_conntrack *);

        if ((struct nf_conn *)nfct == &nf_conntrack_untracked) {
                printk("JCM: nf_conntrack_destroy: trying to destroy
nf_conntrack_untracked! CONTINUING...\n");
                //panic("JCM: nf_conntrack_destroy: trying to destroy
nf_conntrack_untracked!\n");
                return; /* refuse to free nf_conntrack_untracked */
        }

        rcu_read_lock();
        destroy = rcu_dereference(nf_ct_destroy);
        BUG_ON(destroy == NULL);
        destroy(nfct);
        rcu_read_unlock();
}
EXPORT_SYMBOL(nf_conntrack_destroy);

Clearly that's just a hack (though catching the specific attempt to free
the untracked conntrack sounds like a very good idea in general). I will
leave this running for a while, but so far no problems:

[jcm@perihelion jcm_26]$ dmesg|grep JCM
[   29.952717] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   30.207091] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   30.403248] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   31.403319] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   32.977106] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   33.347100] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   33.966092] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   34.967111] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   35.404323] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   35.911430] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   35.912442] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   38.967061] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   39.403342] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   41.288392] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   42.966063] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   44.036451] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   95.621174] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   95.673061] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   95.741180] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   96.273429] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   96.741296] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   97.246108] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[   98.246486] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  100.747170] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  102.252067] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  104.747066] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  105.762433] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  106.252084] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  147.260998] JCM: nf_conntrack_destroy: trying to destroy
nf_conntrack_untracked! CONTINUING...
[  155.485160] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  156.403661] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  157.402928] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  161.402477] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  163.270622] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  163.277368] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  167.718677] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  176.901877] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  182.554024] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  182.616693] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  183.616932] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  187.617026] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  188.362002] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  191.617001] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  193.574899] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  197.455338] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  214.163491] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  214.311528] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  215.311577] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  219.317496] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  220.693493] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  223.317095] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  240.326198] JCM: nf_conntrack_destroy: trying to destroy
nf_conntrack_untracked! CONTINUING...
[  252.481116] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  252.881087] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  253.881251] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  257.922816] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  258.882989] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  261.413093] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  261.921943] JCM: icmpv6_error: attaching to nf_conntrack_untracked.
[  266.350983] JCM: icmpv6_error: attaching to nf_conntrack_untracked.

So, over to you :)

Jon.

  reply	other threads:[~2010-02-02 18:34 UTC|newest]

Thread overview: 63+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-01-30  1:10 debug: nt_conntrack and KVM crash Jon Masters
2010-01-30  1:57 ` Jon Masters
2010-01-30  1:59   ` Jon Masters
2010-01-30  6:58     ` Eric Dumazet
2010-01-30  7:36       ` Jon Masters
2010-01-30  7:40         ` Jon Masters
2010-01-30  8:33         ` Eric Dumazet
2010-01-30 10:03           ` Jon Masters
2010-02-01  9:32       ` Jon Masters
2010-02-01  9:36         ` Alexey Dobriyan
2010-02-01 10:12           ` Eric Dumazet
2010-02-01 10:25             ` Alexey Dobriyan
2010-02-01 10:38               ` Jon Masters
2010-02-01 11:23               ` Eric Dumazet
2010-02-01 14:48                 ` Alexey Dobriyan
2010-02-01 14:57                   ` Eric Dumazet
2010-02-01 14:52                 ` [PATCH] netfilter: per netns nf_conntrack_cachep Eric Dumazet
2010-02-01 14:58                   ` Alexey Dobriyan
2010-02-01 15:02                     ` Eric Dumazet
2010-02-02 11:04                       ` Jon Masters
2010-02-02 11:35                         ` Jon Masters
2010-02-02 16:46                           ` Jon Masters
2010-02-02 16:48                             ` Patrick McHardy
2010-02-02 17:07                               ` Jon Masters
2010-02-02 17:58                                 ` Alexey Dobriyan
2010-02-02 18:16                                   ` Jon Masters
2010-02-02 18:34                                     ` Jon Masters [this message]
2010-02-02 18:36                                     ` Patrick McHardy
2010-02-02 18:39                                       ` Jon Masters
2010-02-02 18:42                                         ` Jon Masters
2010-02-03 12:10                                       ` Patrick McHardy
2010-02-03 18:38                                         ` Jon Masters
2010-02-03 19:09                                           ` Alexey Dobriyan
2010-02-03 19:43                                             ` Jon Masters
2010-02-03 19:46                                               ` Jon Masters
2010-02-03 19:53                                                 ` Alexey Dobriyan
2010-02-03 20:04                                                   ` Jon Masters
2010-02-03 19:51                                               ` Alexey Dobriyan
2010-02-03 19:53                                                 ` Jon Masters
2010-02-03 20:01                                                   ` Alexey Dobriyan
2010-02-04 12:25                                               ` Patrick McHardy
2010-02-04 12:27                                                 ` Alexey Dobriyan
2010-02-04 12:30                                                   ` Patrick McHardy
2010-02-04 12:35                                                     ` Alexey Dobriyan
2010-02-04 13:04                                                       ` Patrick McHardy
2010-02-04 13:18                                                         ` Jon Masters
2010-02-04 13:37                                                           ` Patrick McHardy
2010-02-04 13:42                                                             ` Jon Masters
2010-02-03 20:21                                         ` Jon Masters
2010-02-04 12:24                                           ` Patrick McHardy
2010-02-02 16:58                             ` PROBLEM with summary: " Jon Masters
2010-02-02 17:04                               ` Patrick McHardy
2010-02-02 17:16                                 ` Eric Dumazet
2010-02-02 17:23                                   ` Jon Masters
2010-02-02  4:36                   ` Jon Masters
2010-02-02  7:02                     ` Jon Masters
2010-02-02 10:47                   ` Jon Masters
2010-02-04 14:00                   ` Patrick McHardy
2010-02-01 10:35           ` debug: nt_conntrack and KVM crash Jon Masters
2010-02-01 10:44             ` Alexey Dobriyan
2010-02-01 10:47               ` Alexey Dobriyan
2010-02-01 10:49                 ` Alexey Dobriyan
2010-02-01 10:53                   ` Jon Masters

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=1265135676.2861.196.camel@tonnant \
    --to=jonathan@jonmasters.org \
    --cc=adobriyan@gmail.com \
    --cc=eric.dumazet@gmail.com \
    --cc=kaber@trash.net \
    --cc=linux-kernel@vger.kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=netfilter-devel@vger.kernel.org \
    --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 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).