netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Vadim Fedorenko <vadim.fedorenko@linux.dev>
To: Ido Schimmel <idosch@nvidia.com>,
	Willem de Bruijn <willemdebruijn.kernel@gmail.com>
Cc: "David S. Miller" <davem@davemloft.net>,
	David Ahern <dsahern@kernel.org>,
	Eric Dumazet <edumazet@google.com>,
	Paolo Abeni <pabeni@redhat.com>, Simon Horman <horms@kernel.org>,
	Willem de Bruijn <willemb@google.com>,
	Jakub Kicinski <kuba@kernel.org>, Shuah Khan <shuah@kernel.org>,
	netdev@vger.kernel.org
Subject: Re: [PATCH net v2 1/2] net: fib: restore ECMP balance from loopback
Date: Sun, 21 Dec 2025 18:49:53 +0000	[thread overview]
Message-ID: <1c6ba073-79e5-461b-ae76-4ef22fe04632@linux.dev> (raw)
In-Reply-To: <aUgnGahB9uXbvrbh@shredder>

On 21/12/2025 16:58, Ido Schimmel wrote:
> On Sun, Dec 21, 2025 at 10:55:15AM -0500, Willem de Bruijn wrote:
>> Vadim Fedorenko wrote:
>>> Preference of nexthop with source address broke ECMP for packets with
>>> source addresses which are not in the broadcast domain, but rather added
>>> to loopback/dummy interfaces. Original behaviour was to balance over
>>> nexthops while now it uses the latest nexthop from the group.
>>>
>>> For the case with 198.51.100.1/32 assigned to dummy0 and routed using
>>> 192.0.2.0/24 and 203.0.113.0/24 networks:
>>>
>>> 2: dummy0: <BROADCAST,NOARP,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN group default qlen 1000
>>>      link/ether d6:54:8a:ff:78:f5 brd ff:ff:ff:ff:ff:ff
>>>      inet 198.51.100.1/32 scope global dummy0
>>>         valid_lft forever preferred_lft forever
>>> 7: veth1@if6: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
>>>      link/ether 06:ed:98:87:6d:8a brd ff:ff:ff:ff:ff:ff link-netnsid 0
>>>      inet 192.0.2.2/24 scope global veth1
>>>         valid_lft forever preferred_lft forever
>>>      inet6 fe80::4ed:98ff:fe87:6d8a/64 scope link proto kernel_ll
>>>         valid_lft forever preferred_lft forever
>>> 9: veth3@if8: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
>>>      link/ether ae:75:23:38:a0:d2 brd ff:ff:ff:ff:ff:ff link-netnsid 0
>>>      inet 203.0.113.2/24 scope global veth3
>>>         valid_lft forever preferred_lft forever
>>>      inet6 fe80::ac75:23ff:fe38:a0d2/64 scope link proto kernel_ll
>>>         valid_lft forever preferred_lft forever
>>>
>>> ~ ip ro list:
>>> default
>>> 	nexthop via 192.0.2.1 dev veth1 weight 1
>>> 	nexthop via 203.0.113.1 dev veth3 weight 1
>>> 192.0.2.0/24 dev veth1 proto kernel scope link src 192.0.2.2
>>> 203.0.113.0/24 dev veth3 proto kernel scope link src 203.0.113.2
>>>
>>> before:
>>>     for i in {1..255} ; do ip ro get 10.0.0.$i; done | grep veth | awk ' {print $(NF-2)}' | sort | uniq -c:
>>>      255 veth3
>>>
>>> after:
>>>     for i in {1..255} ; do ip ro get 10.0.0.$i; done | grep veth | awk ' {print $(NF-2)}' | sort | uniq -c:
>>>      122 veth1
>>>      133 veth3
> 
> The commit message only explains the problem, but not the solution...

Well, the solution is to try to restore original logic. But ok, I'll
explain it explicitly

> 
>>>
>>> Fixes: 32607a332cfe ("ipv4: prefer multipath nexthop that matches source address")
>>> Signed-off-by: Vadim Fedorenko <vadim.fedorenko@linux.dev>
>>> ---
>>> v1 -> v2:
>>>
>>> - add score calculation for nexthop to keep original logic
>>> - adjust commit message to explain the config
>>> - use dummy device instead of loopback
>>> ---
>>>
>>>   net/ipv4/fib_semantics.c | 24 ++++++++----------------
>>>   1 file changed, 8 insertions(+), 16 deletions(-)
>>>
>>> diff --git a/net/ipv4/fib_semantics.c b/net/ipv4/fib_semantics.c
>>> index a5f3c8459758..4d3650d20ff2 100644
>>> --- a/net/ipv4/fib_semantics.c
>>> +++ b/net/ipv4/fib_semantics.c
>>> @@ -2167,8 +2167,8 @@ void fib_select_multipath(struct fib_result *res, int hash,
>>>   {
>>>   	struct fib_info *fi = res->fi;
>>>   	struct net *net = fi->fib_net;
>>> -	bool found = false;
>>>   	bool use_neigh;
>>> +	int score = -1;
>>>   	__be32 saddr;
>>>   
>>>   	if (unlikely(res->fi->nh)) {
>>> @@ -2180,7 +2180,7 @@ void fib_select_multipath(struct fib_result *res, int hash,
>>>   	saddr = fl4 ? fl4->saddr : 0;
>>>   
>>>   	change_nexthops(fi) {
>>> -		int nh_upper_bound;
>>> +		int nh_upper_bound, nh_score = 0;
>>>   
>>>   		/* Nexthops without a carrier are assigned an upper bound of
>>>   		 * minus one when "ignore_routes_with_linkdown" is set.
>>> @@ -2190,24 +2190,16 @@ void fib_select_multipath(struct fib_result *res, int hash,
>>>   		    (use_neigh && !fib_good_nh(nexthop_nh)))
>>>   			continue;
>>>   
>>> -		if (!found) {
>>> +		if (saddr && nexthop_nh->nh_saddr == saddr)
>>> +			nh_score += 2;
>>> +		if (hash <= nh_upper_bound)
>>> +			nh_score++;
>>> +		if (score < nh_score) {
>>>   			res->nh_sel = nhsel;
>>>   			res->nhc = &nexthop_nh->nh_common;
>>> -			found = !saddr || nexthop_nh->nh_saddr == saddr;
>>
>> if score == 3 return immediately?
> 
> We can also return early in the input path (!saddr) when score is 1.
> This seems to work:
> 
> diff --git a/net/ipv4/fib_semantics.c b/net/ipv4/fib_semantics.c
> index 4d3650d20ff2..0caf38e44c73 100644
> --- a/net/ipv4/fib_semantics.c
> +++ b/net/ipv4/fib_semantics.c
> @@ -2197,6 +2197,8 @@ void fib_select_multipath(struct fib_result *res, int hash,
>   		if (score < nh_score) {
>   			res->nh_sel = nhsel;
>   			res->nhc = &nexthop_nh->nh_common;
> +			if (nh_score == 3 || (!saddr && nh_score == 1))
> +				return;
>   			score = nh_score;
>   		}
> 

It makes sense to amortize the loop. Going to send v3

> Tested with net/fib_tests.sh and forwarding/router_multipath.sh
> 
>>
>>> +			score = nh_score;
>>>   		}
>>>   
>>> -		if (hash > nh_upper_bound)
>>> -			continue;
>>> -
>>> -		if (!saddr || nexthop_nh->nh_saddr == saddr) {
>>> -			res->nh_sel = nhsel;
>>> -			res->nhc = &nexthop_nh->nh_common;
>>> -			return;
>>> -		}
>>> -
>>> -		if (found)
>>> -			return;
>>> -
>>>   	} endfor_nexthops(fi);
>>>   }
>>>   #endif
>>> -- 
>>> 2.47.3
>>>
>>
>>


      reply	other threads:[~2025-12-21 18:50 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-12-20  3:23 [PATCH net v2 1/2] net: fib: restore ECMP balance from loopback Vadim Fedorenko
2025-12-20  3:23 ` [PATCH net v2 2/2] selftests: fib_test: Add test case for ipv4 multi nexthops Vadim Fedorenko
2025-12-21 15:59   ` Willem de Bruijn
2025-12-21 15:55 ` [PATCH net v2 1/2] net: fib: restore ECMP balance from loopback Willem de Bruijn
2025-12-21 16:58   ` Ido Schimmel
2025-12-21 18:49     ` Vadim Fedorenko [this message]

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=1c6ba073-79e5-461b-ae76-4ef22fe04632@linux.dev \
    --to=vadim.fedorenko@linux.dev \
    --cc=davem@davemloft.net \
    --cc=dsahern@kernel.org \
    --cc=edumazet@google.com \
    --cc=horms@kernel.org \
    --cc=idosch@nvidia.com \
    --cc=kuba@kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    --cc=shuah@kernel.org \
    --cc=willemb@google.com \
    --cc=willemdebruijn.kernel@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;
as well as URLs for NNTP newsgroup(s).