From: Patrick McHardy <kaber@trash.net>
To: Eric Dumazet <dada1@cosmosbay.com>
Cc: mbizon@freebox.fr,
"Paul E. McKenney" <paulmck@linux.vnet.ibm.com>,
Joakim Tjernlund <Joakim.Tjernlund@transmode.se>,
avorontsov@ru.mvista.com, netdev@vger.kernel.org,
Netfilter Developers <netfilter-devel@vger.kernel.org>
Subject: Re: [PATCH] conntrack: use SLAB_DESTROY_BY_RCU for nf_conn structs
Date: Wed, 25 Mar 2009 21:10:19 +0100 [thread overview]
Message-ID: <49CA8FAB.8090903@trash.net> (raw)
In-Reply-To: <49CA8CD0.2010907@cosmosbay.com>
Eric Dumazet wrote:
> Patrick McHardy a écrit :
>> I think double entries are not a problem, as you say, there
>> are already other cases where this can happen. But I think we
>> should try our best that every entry present at the start and
>> still present at the end of a dump is also contained in the
>> dump, otherwise the guantees seem to weak to still be useful.
>> Your first proposal would do exactly that, right?
>
> If your concern is to not forget entries, and we are allowed to print some entries several times,
> then we can just check the final "nulls" value, and if we find a different value than expected for
> chain N, go back to begining of chain N.
>
> No need to check hash value (this could help not print several time same entry, we dont care that much)
>
> + while (is_a_nulls(head)) {
> + if (likely(get_nulls_value(head) == st->bucket)) {
> + if (++st->bucket >= nf_conntrack_htable_size)
> + return NULL;
> + }
Looks perfect :)
> [PATCH] conntrack: use SLAB_DESTROY_BY_RCU and get rid of call_rcu()
>
> Use "hlist_nulls" infrastructure we added in 2.6.29 for RCUification of UDP & TCP.
>
> This permits an easy conversion from call_rcu() based hash lists to a
> SLAB_DESTROY_BY_RCU one.
>
> Avoiding call_rcu() delay at nf_conn freeing time has numerous gains.
>
> First, it doesnt fill RCU queues (up to 10000 elements per cpu).
> This reduces OOM possibility, if queued elements are not taken into account
> This reduces latency problems when RCU queue size hits hilimit and triggers
> emergency mode.
>
> - It allows fast reuse of just freed elements, permitting better use of
> CPU cache.
>
> - We delete rcu_head from "struct nf_conn", shrinking size of this structure
> by 8 or 16 bytes.
>
> This patch only takes care of "struct nf_conn".
> call_rcu() is still used for less critical conntrack parts, that may
> be converted later if necessary.
Applied, thanks a lot.
next prev parent reply other threads:[~2009-03-25 20:10 UTC|newest]
Thread overview: 44+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-03-23 10:42 ucc_geth: nf_conntrack: table full, dropping packet Joakim Tjernlund
2009-03-23 12:15 ` Patrick McHardy
2009-03-23 12:25 ` Joakim Tjernlund
2009-03-23 12:29 ` Patrick McHardy
2009-03-23 12:59 ` Joakim Tjernlund
[not found] ` <OF387EC803.F810F72A-ONC1257582.00468C6E-C1257582.00475783@LocalDomain>
2009-03-23 13:09 ` Joakim Tjernlund
2009-03-23 17:42 ` Joakim Tjernlund
2009-03-23 17:49 ` Patrick McHardy
2009-03-24 8:22 ` Joakim Tjernlund
2009-03-24 9:12 ` Eric Dumazet
2009-03-24 10:55 ` Joakim Tjernlund
2009-03-24 12:07 ` [PATCH] conntrack: Reduce conntrack count in nf_conntrack_free() Eric Dumazet
2009-03-24 12:25 ` Eric Dumazet
2009-03-24 12:43 ` Patrick McHardy
2009-03-24 13:32 ` Eric Dumazet
2009-03-24 13:38 ` Patrick McHardy
2009-03-24 13:47 ` Eric Dumazet
[not found] ` <49C8F871.9070600@cosmosbay.com>
[not found] ` <49C8F8E0.9050502@trash.net>
2009-03-25 3:53 ` Eric Dumazet
2009-03-25 13:39 ` Patrick McHardy
2009-03-25 13:44 ` Eric Dumazet
2009-03-24 13:20 ` Joakim Tjernlund
2009-03-24 13:28 ` Patrick McHardy
2009-03-24 13:29 ` Eric Dumazet
2009-03-24 13:41 ` Joakim Tjernlund
2009-03-24 15:17 ` Maxime Bizon
2009-03-24 15:21 ` Patrick McHardy
2009-03-24 15:27 ` Eric Dumazet
2009-03-24 19:54 ` [PATCH] netfilter: Use hlist_add_head_rcu() in nf_conntrack_set_hashsize() Eric Dumazet
2009-03-25 16:26 ` Patrick McHardy
2009-03-25 17:53 ` [PATCH] conntrack: use SLAB_DESTROY_BY_RCU for nf_conn structs Eric Dumazet
2009-03-25 18:05 ` Patrick McHardy
2009-03-25 18:06 ` Patrick McHardy
2009-03-25 18:15 ` Eric Dumazet
2009-03-25 18:24 ` Patrick McHardy
2009-03-25 18:53 ` Eric Dumazet
2009-03-25 19:00 ` Patrick McHardy
2009-03-25 19:17 ` Eric Dumazet
2009-03-25 19:41 ` Patrick McHardy
2009-03-25 19:58 ` Eric Dumazet
2009-03-25 20:10 ` Patrick McHardy [this message]
2009-03-24 18:29 ` [PATCH] conntrack: Reduce conntrack count in nf_conntrack_free() Joakim Tjernlund
2009-03-23 17:49 ` ucc_geth: nf_conntrack: table full, dropping packet Eric Dumazet
2009-03-23 18:04 ` Joakim Tjernlund
2009-03-23 18:08 ` Eric Dumazet
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=49CA8FAB.8090903@trash.net \
--to=kaber@trash.net \
--cc=Joakim.Tjernlund@transmode.se \
--cc=avorontsov@ru.mvista.com \
--cc=dada1@cosmosbay.com \
--cc=mbizon@freebox.fr \
--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).