mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Andrea Mayer <andrea.mayer@uniroma2.it>
To: Yuya Kusakabe <yuya.kusakabe@gmail.com>
Cc: "David S. Miller" <davem@davemloft.net>,
	Eric Dumazet <edumazet@kernel.org>,
	Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>,
	Simon Horman <horms@kernel.org>,
	netdev@vger.kernel.org, linux-kernel@vger.kernel.org,
	Gabriel Goller <g.goller@proxmox.com>,
	Andrea Mayer <andrea.mayer@uniroma2.it>
Subject: Re: [PATCH net-next v3 1/2] seg6: skip the route refcount in seg6local IPv6 lookups
Date: Tue, 20 Oct 2026 01:04:29 +0200	[thread overview]
Message-ID: <20261020010429.eeffd6f140ec6157609c310a@uniroma2.it> (raw)
In-Reply-To: <20261015-seg6-lookup-noref-v3-1-6d51a60a67d3@gmail.com>

On Thu, 15 Oct 2026 23:39:27 +0900
Yuya Kusakabe <yuya.kusakabe@gmail.com> wrote:

Hi Yuya,

Thanks for v3 and considering my comments.

> [snip]
>  		if (!table)
>  			goto out;
>  
> +		if (noref)
> +			flags |= RT6_LOOKUP_F_DST_NOREF;

Even with the change below, the "out:" branch still takes a reference when
fib6_get_table() returns NULL, because RT6_LOOKUP_F_DST_NOREF is set only
after that call. Setting RT6_LOOKUP_F_DST_NOREF before fib6_get_table(),
inside the table branch, would avoid that reference too. The oif branch
needs the flag clear, because ip6_route_output() returns the dst with a
reference.

> [snip]
>  		rt = net->ipv6.ip6_blk_hole_entry;
>  		dst = &rt->dst;
>  		dst_hold(dst);
> +		flags &= ~RT6_LOOKUP_F_DST_NOREF;
>  	}

With this patch applied, the end of seg6_lookup_any_nexthop() reads:

out:
	if (!dst) {
		rt = net->ipv6.ip6_blk_hole_entry;
		dst = &rt->dst;
		dst_hold(dst);
		flags &= ~RT6_LOOKUP_F_DST_NOREF;
	}

	skb_dst_drop(skb);
	if ((flags & RT6_LOOKUP_F_DST_NOREF) && !dst->rt_uncached_list)
		skb_dst_set_noref(skb, dst);
	else
		skb_dst_set(skb, dst);
	return dst->error;

After a noref lookup, the "out:" branch still takes a reference on
ip6_blk_hole_entry and then clears RT6_LOOKUP_F_DST_NOREF.

This branch can be taken at steady state. The noref lookups, except the
ones in input_action_end_dt6(), run with local_delivery false. For example,
with End and a next segment that is a local address of the node, the lookup
returns a local route, seg6_lookup_any_nexthop() discards it, and every
packet takes and drops a reference on ip6_blk_hole_entry in the "out:"
branch. The commit message describes the same pattern for ip6_null_entry.

For a blackhole rule, __fib6_rule_action() picks ip6_blk_hole_entry and
takes no reference on it when the lookup has RT6_LOOKUP_F_DST_NOREF. The
"out:" branch could do the same, for example:

	if (!dst) {
		dst = &net->ipv6.ip6_blk_hole_entry->dst;
		if (!(flags & RT6_LOOKUP_F_DST_NOREF))
			dst_hold(dst);
	}

RT6_LOOKUP_F_DST_NOREF then stays set, and the skb_dst_set_noref() at the
end of the function attaches ip6_blk_hole_entry without a reference. The
line that clears RT6_LOOKUP_F_DST_NOREF is removed as well.

What do you think?

Nit: seg6_lookup_any_nexthop_noref() and seg6_lookup_nexthop_noref() can
attach the dst without a reference. They rely on being called under RCU. A
one line comment about this, like the one ip6_route_input() has, could help
the next reader.

Ciao,
Andrea

  reply	other threads:[~2026-10-19 23:05 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-10-15 14:39 [PATCH net-next v3 0/2] seg6: skip route refcounts in seg6local lookups Yuya Kusakabe
2026-10-15 14:39 ` [PATCH net-next v3 1/2] seg6: skip the route refcount in seg6local IPv6 lookups Yuya Kusakabe
2026-10-19 23:04   ` Andrea Mayer [this message]
2026-10-15 14:39 ` [PATCH net-next v3 2/2] seg6: skip the route refcount in seg6local IPv4 lookups Yuya Kusakabe

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=20261020010429.eeffd6f140ec6157609c310a@uniroma2.it \
    --to=andrea.mayer@uniroma2.it \
    --cc=davem@davemloft.net \
    --cc=edumazet@kernel.org \
    --cc=g.goller@proxmox.com \
    --cc=horms@kernel.org \
    --cc=kuba@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    --cc=yuya.kusakabe@gmail.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®