netdev - Re: [PATCHv4 bpf-next 0/2] xdp: add dev map multicast support

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <87img8l893.fsf@toke.dk>
Date:   Wed, 03 Jun 2020 13:05:28 +0200
From:   Toke Høiland-Jørgensen <toke@...hat.com>
To:     Hangbin Liu <liuhangbin@...il.com>
Cc:     bpf@...r.kernel.org, netdev@...r.kernel.org,
        Jiri Benc <jbenc@...hat.com>,
        Jesper Dangaard Brouer <brouer@...hat.com>,
        Eelco Chaudron <echaudro@...hat.com>, ast@...nel.org,
        Daniel Borkmann <daniel@...earbox.net>,
        Lorenzo Bianconi <lorenzo.bianconi@...hat.com>
Subject: Re: [PATCHv4 bpf-next 0/2] xdp: add dev map multicast support

Hangbin Liu <liuhangbin@...il.com> writes:

> On Wed, May 27, 2020 at 12:21:54PM +0200, Toke HÃƒÂ¸iland-JÃƒÂ¸rgensen wrote:
>> > The example in patch 2 is functional, but not a lot of effort
>> > has been made on performance optimisation. I did a simple test(pkt size 64)
>> > with pktgen. Here is the test result with BPF_MAP_TYPE_DEVMAP_HASH
>> > arrays:
>> >
>> > bpf_redirect_map() with 1 ingress, 1 egress:
>> > generic path: ~1600k pps
>> > native path: ~980k pps
>> >
>> > bpf_redirect_map_multi() with 1 ingress, 3 egress:
>> > generic path: ~600k pps
>> > native path: ~480k pps
>> >
>> > bpf_redirect_map_multi() with 1 ingress, 9 egress:
>> > generic path: ~125k pps
>> > native path: ~100k pps
>> >
>> > The bpf_redirect_map_multi() is slower than bpf_redirect_map() as we loop
>> > the arrays and do clone skb/xdpf. The native path is slower than generic
>> > path as we send skbs by pktgen. So the result looks reasonable.
>> 
>> How are you running these tests? Still on virtual devices? We really
>> need results from a physical setup in native mode to assess the impact
>> on the native-XDP fast path. The numbers above don't tell much in this
>> regard. I'd also like to see a before/after patch for straight
>> bpf_redirect_map(), since you're messing with the fast path, and we want
>> to make sure it's not causing a performance regression for regular
>> redirect.
>> 
>> Finally, since the overhead seems to be quite substantial: A comparison
>> with a regular network stack bridge might make sense? After all we also
>> want to make sure it's a performance win over that :)
>
> Hi Toke,
>
> Here is the result I tested with 2 i40e 10G ports on physical machine.
> The pktgen pkt_size is 64.

These numbers seem a bit low (I'm getting ~8.5MPPS on my test machine
for a simple redirect). Some of that may just be performance of the
machine, I guess (what are you running this on?), but please check that
you are not limited by pktgen itself - i.e., that pktgen is generating
traffic at a higher rate than what XDP is processing.

> Bridge forwarding(I use sample/bpf/xdp1 to count the PPS, so there are two modes data):
> generic mode: 1.32M PPS
> driver mode: 1.66M PPS

I'm not sure I understand this - what are you measuring here exactly?

> xdp_redirect_map:
> generic mode: 1.88M PPS
> driver mode: 2.74M PPS

Please add numbers without your patch applied as well, for comparison.

> xdp_redirect_map_multi:
> generic mode: 1.38M PPS
> driver mode: 2.73M PPS

I assume this is with a single interface only, right? Could you please
add a test with a second interface (so the packet is cloned) as well?
You can just use a veth as the second target device.

-Toke