From: Christian Deacon <gamemann@gflclan.com>
To: xdp-newbies@vger.kernel.org
Subject: AF_XDP Sockets Not Receiving Packets In XDP DRV Mode
Date: Mon, 7 Jun 2021 17:23:04 -0500 [thread overview]
Message-ID: <705dc02f-9e53-4cea-f100-43145d6d8684@gflclan.com> (raw)
Hey everyone,
I believe I made a thread regarding this issue around a year ago or so,
but I thought it was related to the `virtio_net` driver. This time,
however, I am testing this on a dedicated server with a dedicated NIC
(see information at bottom) and have the same issue.
I am trying to create an AF_XDP program and in this program, I currently
spin up n AF_XDP sockets where n represents the RX queue count. Each
AF_XDP socket is bound to a specific RX queue. From what I've seen, most
examples recommend binding the socket to one RX queue and using flow
steering via `ethtool` to steer all packets destined for the AF_XDP
socket to the RX queue the AF_XDP socket is bound to. The two problems I
see here are:
1. To my understanding, this NIC only supports matching on layer 2/3/4
headers and with my AF_XDP program, I want to redirect packets to the
AF_XDP sockets depending on the first 4 - 5 bytes of the payload.
2. It would be better for performance to spread these packets out to
individual RX queues, even if they're destined for the AF_XDP sockets
(please correct me if I am wrong here).
In the program, I create a separate thread for each AF_XDP socket and
this thread polls that specific AF_XDP socket.
Here is the program's source code:
https://github.com/gamemann/AF_XDP-Test
When loading the XDP program in DRV mode, the packets destined for the
AF_XDP sockets never reach the socket. However, when loaded in SKB mode,
the packets do arrive. The return code of the `bpf_redirect_map()`
function is the same for each mode (`4` which is successful I believe,
XDP_REDIRECT). The AF_XDP sockets are also created successfully with
each mode and I confirmed I receive a file descriptor (>0) from each
AF_XDP socket.
The AF_XDP sockets do not share the same UMEM as well. Individual UMEMs
are assigned to each AF_XDP socket.
Here is information regarding the Linux kernel and NIC.
Kernel:
```
cdeacon@xxxxxx:~/AF_XDP-Test$ uname -r
5.12.5-051205-generic
```
Note - I've had this same issue on other Linux kernels from 5.4 to 5.10.
Though, that was on a VM with the `virtio_net` driver.
NIC information:
```
*-network
description: Ethernet interface
physical id: 1
bus info: pci@0000:04:00.0
logical name: yyyyyyyy
serial: xx:xx:xx:xx:xx
capabilities: ethernet physical fibre
configuration: autonegotiation=on broadcast=yes driver=nfp
driverversion=5.12.5-051205-generic duplex=full firmware=0.0.3.5 0.22
nic-2.1.16.1 nic ip=xxx.xxx.xxx.xxx link=yes multicast=yes port=fibre
```
NIC queues:
```
cdeacon@xxxxxx:~/AF_XDP-Test$ sudo ethtool -l yyyyyyyy
Channel parameters for yyyyyyyy:
Pre-set maximums:
RX: 16
TX: 16
Other: 2
Combined: 16
Current hardware settings:
RX: 0
TX: 8
Other: 2
Combined: 8
```
NIC settings/features:
```
cdeacon@xxxxxx:~/AF_XDP-Test$ sudo ethtool -k yyyyyyyy
Features for yyyyyyyy:
rx-checksumming: on
tx-checksumming: on
tx-checksum-ipv4: on
tx-checksum-ip-generic: off [fixed]
tx-checksum-ipv6: on
tx-checksum-fcoe-crc: off [fixed]
tx-checksum-sctp: off [fixed]
scatter-gather: on
tx-scatter-gather: on
tx-scatter-gather-fraglist: off [fixed]
tcp-segmentation-offload: off
tx-tcp-segmentation: off
tx-tcp-ecn-segmentation: off [fixed]
tx-tcp-mangleid-segmentation: off
tx-tcp6-segmentation: off
generic-segmentation-offload: on
generic-receive-offload: on
large-receive-offload: off [fixed]
rx-vlan-offload: off [fixed]
tx-vlan-offload: off [fixed]
ntuple-filters: off [fixed]
receive-hashing: on
highdma: on
rx-vlan-filter: off [fixed]
vlan-challenged: off [fixed]
tx-lockless: off [fixed]
netns-local: off [fixed]
tx-gso-robust: off [fixed]
tx-fcoe-segmentation: off [fixed]
tx-gre-segmentation: on
tx-gre-csum-segmentation: off [fixed]
tx-ipxip4-segmentation: off [fixed]
tx-ipxip6-segmentation: off [fixed]
tx-udp_tnl-segmentation: on
tx-udp_tnl-csum-segmentation: off [fixed]
tx-gso-partial: off [fixed]
tx-tunnel-remcsum-segmentation: off [fixed]
tx-sctp-segmentation: off [fixed]
tx-esp-segmentation: off [fixed]
tx-udp-segmentation: off [fixed]
tx-gso-list: off [fixed]
fcoe-mtu: off [fixed]
tx-nocache-copy: off
loopback: off [fixed]
rx-fcs: off [fixed]
rx-all: off [fixed]
tx-vlan-stag-hw-insert: off [fixed]
rx-vlan-stag-hw-parse: off [fixed]
rx-vlan-stag-filter: off [fixed]
l2-fwd-offload: off [fixed]
hw-tc-offload: off [fixed]
esp-hw-offload: off [fixed]
esp-tx-csum-hw-offload: off [fixed]
rx-udp_tunnel-port-offload: on
tls-hw-tx-offload: off [fixed]
tls-hw-rx-offload: off [fixed]
rx-gro-hw: off [fixed]
tls-hw-record: off [fixed]
rx-gro-list: off
macsec-hw-offload: off [fixed]
rx-udp-gro-forwarding: off
hsr-tag-ins-offload: off [fixed]
hsr-tag-rm-offload: off [fixed]
hsr-fwd-offload: off [fixed]
hsr-dup-offload: off [fixed]
```
From what I've read below, setting up individual AF_XDP sockets on
specific RX queues is supported, even with XDP DRV mode.
https://github.com/xdp-project/xdp-tutorial/tree/master/advanced03-AF_XDP#gotcha-by-rx-queue-id-binding
The AF_XDP test program I made is also based off of the AF_XDP lesson in
the `xdp-tutorial` repository.
One last note I want to add is I've tried adding the AF_XDP socket's FD
to the XSK map inside the userspace via `bpf_map_update_elem(xskmapfd,
&rxqueue, &fd, BPF_ANY)` with no luck (also confirmed the map updated
successfully).
Any help is highly appreciated and if you need any additional
information, please let me know!
Thank you for your time.
next reply other threads:[~2021-06-07 22:28 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-06-07 22:23 Christian Deacon [this message]
2021-06-08 11:26 ` AF_XDP Sockets Not Receiving Packets In XDP DRV Mode Magnus Karlsson
2021-06-08 12:06 ` Maciej Fijalkowski
2021-06-11 20:47 ` Christian Deacon
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=705dc02f-9e53-4cea-f100-43145d6d8684@gflclan.com \
--to=gamemann@gflclan.com \
--cc=xdp-newbies@vger.kernel.org \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.