From: John Fastabend <firstname.lastname@example.org> To: Toshiaki Makita <email@example.com>, Alexei Starovoitov <firstname.lastname@example.org>, Daniel Borkmann <email@example.com>, Martin KaFai Lau <firstname.lastname@example.org>, Song Liu <email@example.com>, Yonghong Song <firstname.lastname@example.org>, "David S. Miller" <email@example.com>, Jakub Kicinski <firstname.lastname@example.org>, Jesper Dangaard Brouer <email@example.com>, John Fastabend <firstname.lastname@example.org>, Jamal Hadi Salim <email@example.com>, Cong Wang <firstname.lastname@example.org>, Jiri Pirko <email@example.com>, Pablo Neira Ayuso <firstname.lastname@example.org>, Jozsef Kadlecsik <email@example.com>, Florian Westphal <firstname.lastname@example.org>, Pravin B Shelar <email@example.com> Cc: Toshiaki Makita <firstname.lastname@example.org>, email@example.com, firstname.lastname@example.org, William Tu <email@example.com>, Stanislav Fomichev <firstname.lastname@example.org> Subject: RE: [RFC PATCH v2 bpf-next 00/15] xdp_flow: Flow offload to XDP Date: Fri, 18 Oct 2019 08:22:41 -0700 [thread overview] Message-ID: <5da9d8c125fd4_31cf2adc704105c456@john-XPS-13-9370.notmuch> (raw) In-Reply-To: <email@example.com> Toshiaki Makita wrote: > This is a PoC for an idea to offload flow, i.e. TC flower and nftables, > to XDP. > I've only read the cover letter so far but... > * Motivation > > The purpose is to speed up flow based network features like TC flower and > nftables by making use of XDP. > > I chose flow feature because my current interest is in OVS. OVS uses TC > flower to offload flow tables to hardware, so if TC can offload flows to > XDP, OVS also can be offloaded to XDP. This adds a non-trivial amount of code and complexity so I'm critical of the usefulness of being able to offload TC flower to XDP when userspace can simply load an XDP program. Why does OVS use tc flower at all if XDP is about 5x faster using your measurements below? Rather than spend energy adding code to a use case that as far as I can tell is narrowly focused on offload support can we enumerate what is missing on XDP side that blocks OVS from using it directly? Additionally for hardware that can do XDP/BPF offload you will get the hardware offload for free. Yes I know XDP is bytecode and you can't "offload" bytecode into a flow based interface likely backed by a tcam but IMO that doesn't mean we should leak complexity into the kernel network stack to fix this. Use the tc-flower for offload only (it has support for this) if you must and use the best (in terms of Mpps) software interface for your software bits. And if you want auto-magic offload support build hardware with BPF offload support. In addition by using XDP natively any extra latency overhead from bouncing calls through multiple layers would be removed. > > When TC flower filter is offloaded to XDP, the received packets are > handled by XDP first, and if their protocol or something is not > supported by the eBPF program, the program returns XDP_PASS and packets > are passed to upper layer TC. > > The packet processing flow will be like this when this mechanism, > xdp_flow, is used with OVS. Same as obove just cross out the 'TC flower' box and add support for your missing features to 'XDP prog' box. Now you have less code to maintain and less bugs and aren't pushing packets through multiple hops in a call chain. > > +-------------+ > | openvswitch | > | kmod | > +-------------+ > ^ > | if not match in filters (flow key or action not supported by TC) > +-------------+ > | TC flower | > +-------------+ > ^ > | if not match in flow tables (flow key or action not supported by XDP) > +-------------+ > | XDP prog | > +-------------+ > ^ > | incoming packets > > Of course we can directly use TC flower without OVS to speed up TC. huh? TC flower is part of TC so not sure what 'speed up TC' means. I guess this means using tc flower offload to xdp prog would speed up general tc flower usage as well? But again if we are concerned about Mpps metrics just write the XDP program directly. > > This is useful especially when the device does not support HW-offload. > Such interfaces include virtual interfaces like veth. I disagree, use XDP directly. > > > * How to use [...] > * Performance [...] > Tested single core/single flow with 3 kinds of configurations. > (spectre_v2 disabled) > - xdp_flow: hw-offload=true, flow-offload-xdp on > - TC: hw-offload=true, flow-offload-xdp off (software TC) > - ovs kmod: hw-offload=false > > xdp_flow TC ovs kmod > -------- -------- -------- > 5.2 Mpps 1.2 Mpps 1.1 Mpps > > So xdp_flow drop rate is roughly 4x-5x faster than software TC or ovs kmod. +1 yep the main point of using XDP ;) > > OTOH the time to add a flow increases with xdp_flow. > > ping latency of first packet when veth1 does XDP_PASS instead of DROP: > > xdp_flow TC ovs kmod > -------- -------- -------- > 22ms 6ms 0.6ms > > xdp_flow does a lot of work to emulate TC behavior including UMH > transaction and multiple bpf map update from UMH which I think increases > the latency. And this is IMO sinks why we would adopt this. A native XDP solution would as far as I can tell not suffer this latency. So in short, we add lots of code that needs to be maintained, in my opinion it adds complexity, and finally I can't see what XDP is missing today (with the code we already have upstream!) to block doing an implementation without any changes. > > > * Implementation > [...] > > * About OVS AF_XDP netdev [...] > * About alternative userland (ovs-vswitchd etc.) implementation > > Maybe a similar logic can be implemented in ovs-vswitchd offload > mechanism, instead of adding code to kernel. I just thought offloading > TC is more generic and allows wider usage with direct TC command. > > For example, considering that OVS inserts a flow to kernel only when > flow miss happens in kernel, we can in advance add offloaded flows via > tc filter to avoid flow insertion latency for certain sensitive flows. > TC flower usage without using OVS is also possible. I argue to cut tc filter out entirely and then I think non of this is needed. > > Also as written above nftables can be offloaded to XDP with this > mechanism as well. Or same argument use XDP directly. > > Another way to achieve this from userland is to add notifications in > flow_offload kernel code to inform userspace of flow addition and > deletion events, and listen them by a deamon which in turn loads eBPF > programs, attach them to XDP, and modify eBPF maps. Although this may > open up more use cases, I'm not thinking this is the best solution > because it requires emulation of kernel behavior as an offload engine > but flow related code is heavily changing which is difficult to follow > from out of tree. So if everything was already in XDP why would we need these notifications? I think a way to poll on a map from user space would be a great idea e.g. everytime my XDP program adds a flow to my hash map wake up my userspace agent with some ctx on what was added or deleted so I can do some control plane logic. [...] Lots of code churn... > 24 files changed, 2864 insertions(+), 30 deletions(-) Thanks, John
next prev parent reply other threads:[~2019-10-18 15:22 UTC|newest] Thread overview: 58+ messages / expand[flat|nested] mbox.gz Atom feed top 2019-10-18 4:07 Toshiaki Makita 2019-10-18 4:07 ` [RFC PATCH v2 bpf-next 01/15] xdp_flow: Add skeleton of XDP based flow offload driver Toshiaki Makita 2019-10-18 4:07 ` [RFC PATCH v2 bpf-next 02/15] xdp_flow: Add skeleton bpf program for XDP Toshiaki Makita 2019-10-18 4:07 ` [RFC PATCH v2 bpf-next 03/15] bpf: Add API to get program from id Toshiaki Makita 2019-10-18 4:07 ` [RFC PATCH v2 bpf-next 04/15] xdp: Export dev_check_xdp and dev_change_xdp Toshiaki Makita 2019-10-18 4:07 ` [RFC PATCH v2 bpf-next 05/15] xdp_flow: Attach bpf prog to XDP in kernel after UMH loaded program Toshiaki Makita 2019-10-18 4:07 ` [RFC PATCH v2 bpf-next 06/15] xdp_flow: Prepare flow tables in bpf Toshiaki Makita 2019-10-18 4:07 ` [RFC PATCH v2 bpf-next 07/15] xdp_flow: Add flow entry insertion/deletion logic in UMH Toshiaki Makita 2019-10-18 4:07 ` [RFC PATCH v2 bpf-next 08/15] xdp_flow: Add flow handling and basic actions in bpf prog Toshiaki Makita 2019-10-18 4:07 ` [RFC PATCH v2 bpf-next 09/15] xdp_flow: Implement flow replacement/deletion logic in xdp_flow kmod Toshiaki Makita 2019-10-18 4:07 ` [RFC PATCH v2 bpf-next 10/15] xdp_flow: Add netdev feature for enabling flow offload to XDP Toshiaki Makita 2019-10-18 4:07 ` [RFC PATCH v2 bpf-next 11/15] xdp_flow: Implement redirect action Toshiaki Makita 2019-10-18 4:07 ` [RFC PATCH v2 bpf-next 12/15] xdp_flow: Implement vlan_push action Toshiaki Makita 2019-10-18 4:07 ` [RFC PATCH v2 bpf-next 13/15] bpf, selftest: Add test for xdp_flow Toshiaki Makita 2019-10-18 4:07 ` [RFC PATCH v2 bpf-next 14/15] i40e: prefetch xdp->data before running XDP prog Toshiaki Makita 2019-10-18 4:07 ` [RFC PATCH v2 bpf-next 15/15] bpf, hashtab: Compare keys in long Toshiaki Makita 2019-10-18 15:22 ` John Fastabend [this message] 2019-10-21 7:31 ` [RFC PATCH v2 bpf-next 00/15] xdp_flow: Flow offload to XDP Toshiaki Makita 2019-10-22 16:54 ` John Fastabend 2019-10-22 17:45 ` Toke Høiland-Jørgensen 2019-10-24 4:27 ` John Fastabend 2019-10-24 10:13 ` Toke Høiland-Jørgensen 2019-10-27 13:19 ` Toshiaki Makita 2019-10-27 15:21 ` Toke Høiland-Jørgensen 2019-10-28 3:16 ` David Ahern 2019-10-28 8:36 ` Toke Høiland-Jørgensen 2019-10-28 10:08 ` Jesper Dangaard Brouer 2019-10-28 19:07 ` David Ahern 2019-10-28 19:05 ` David Ahern 2019-10-31 0:18 ` Toshiaki Makita 2019-10-31 12:12 ` Toke Høiland-Jørgensen 2019-11-11 7:32 ` Toshiaki Makita 2019-11-12 16:53 ` Toke Høiland-Jørgensen 2019-11-14 10:11 ` Toshiaki Makita 2019-11-14 12:41 ` Toke Høiland-Jørgensen 2019-11-18 6:41 ` Toshiaki Makita 2019-11-18 10:20 ` Toke Høiland-Jørgensen 2019-11-22 5:42 ` Toshiaki Makita 2019-11-22 11:54 ` Toke Høiland-Jørgensen 2019-11-25 10:18 ` Toshiaki Makita 2019-11-25 13:03 ` Toke Høiland-Jørgensen 2019-11-18 10:28 ` Toke Høiland-Jørgensen 2019-10-27 13:13 ` Toshiaki Makita 2019-10-27 15:24 ` Toke Høiland-Jørgensen 2019-10-27 19:17 ` David Miller 2019-10-31 0:32 ` Toshiaki Makita 2019-11-12 17:50 ` William Tu 2019-11-14 10:06 ` Toshiaki Makita 2019-11-14 17:09 ` William Tu 2019-11-15 13:16 ` Toke Høiland-Jørgensen 2019-11-12 17:38 ` William Tu 2019-10-23 14:11 ` Jamal Hadi Salim 2019-10-24 4:38 ` John Fastabend 2019-10-24 17:05 ` Jamal Hadi Salim 2019-10-27 13:27 ` Toshiaki Makita 2019-10-27 13:06 ` Toshiaki Makita 2019-10-21 11:23 ` Björn Töpel 2019-10-21 11:47 ` Toshiaki Makita
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=5da9d8c125fd4_31cf2adc704105c456@john-XPS-13-9370.notmuch \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --subject='RE: [RFC PATCH v2 bpf-next 00/15] xdp_flow: Flow offload to XDP' \ /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
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).