On 2021/02/26 10:13, William Tu wrote:
On Thu, Feb 25, 2021 at 7:26 AM Toshiaki Makita
wrote:
On 2021/02/24 3:37, William Tu wrote:
I don't know if this is too much to ask for.
I wonder if you, or we can work together, to add at least a tunnel
support, ex: vxlan?
The current version
vs-ofctl dump-flows
show correct stats. I don't understand why without flow_get, it still shows
correct flow stats...
Interesting...
Toshiaki Makita
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
at.
E.g. The problem happens around loops. Loop unrolling may help mitigating it or
not?
Toshiaki Makita
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
On 2021/02/16 10:49, William Tu wrote:
On Tue, Feb 9, 2021 at 1:39 AM Toshiaki Makita
wrote:
On 2021/02/05 2:36, William Tu wrote:
Hi Toshiaki,
Thanks for the patch. I've been testing it for a couple days.
I liked it a lot! The compile and build process all work without any issues.
big limitation.
BTW there is some progress on TSO in XDP world...
https://patchwork.kernel.org/project/netdevbpf/cover/cover.1611086134.git.lore...@kernel.org/
Toshiaki Makita
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
On 2021/02/05 12:03, William Tu wrote:
On Thu, Feb 4, 2021 at 5:15 PM William Tu wrote:
Hi Toshiaki,
Thanks for the patch. I have some questions inline.
On Thu, Jul 30, 2020 at 7:55 PM Toshiaki Makita
wrote:
This provider offloads classifier to software XDP.
It works only when a custom
il of the
patch set...
On Thu, Jul 30, 2020 at 7:55 PM Toshiaki Makita
wrote:
This patch adds an XDP-based flow cache using the OVS netdev-offload
flow API provider. When an OVS device with XDP offload enabled,
packets first are processed in the XDP flow cache (with parse, and
table l
icated/isolated XSK rings, which is not meant for a mellanox-specific
feature.
https://patchwork.ozlabs.org/project/netdev/cover/20190524093431.20887-1-maxi...@mellanox.com/
https://patchwork.ozlabs.org/project/netdev/cover/20190612155605.22450-1-maxi...@mellanox.com/
Toshiaki Makita
I think addin
On 2021/02/05 18:00, Eli Britstein wrote:
On 2/4/2021 9:18 PM, William Tu wrote:
On Thu, Jul 30, 2020 at 7:55 PM Toshiaki Makita
wrote:
The following commit will introduce another offload driver using XDP.
When using afxdp netdev, both of TC and XDP will be supported, so let's
a
On 2020/11/02 18:37, Numan Siddique wrote:
On Fri, Oct 30, 2020 at 10:49 AM Toshiaki Makita
wrote:
Hi all,
It's about 3 months since I submitted this patch set.
Could someone review this?
Or should I resubmit the patch set on the top of current master?
Since the patches don't app
Hi all,
It's about 3 months since I submitted this patch set.
Could someone review this?
Or should I resubmit the patch set on the top of current master?
Thanks,
Toshiaki Makita
On 2020/08/15 10:54, Toshiaki Makita wrote:
Ping.
Any feedback is welcome.
Thanks,
Toshiaki Makita
On 2020/
Ping.
Any feedback is welcome.
Thanks,
Toshiaki Makita
On 2020/07/31 11:55, Toshiaki Makita wrote:
This patch adds an XDP-based flow cache using the OVS netdev-offload
flow API provider. When an OVS device with XDP offload enabled,
packets first are processed in the XDP flow cache (with parse
On 2020/07/31 13:07, 0-day Robot wrote:
Bleep bloop. Greetings Toshiaki Makita, I am a robot and I have tried out your
patch.
Thanks for your contribution.
I encountered some error that I wasn't expecting. See the details below.
checkpatch:
ERROR: Improper whitespace around control
ore details, refer to the reference BPF program in next commit.
In the future it may be possible to offload classifier to SmartNIC
through XDP, but currently it's not because map-in-map is not supported
by XDP hw-offload.
Signed-off-by: Toshiaki Makita
---
acinclude.m4
etdev supports it.
Signed-off-by: Toshiaki Makita
---
lib/netdev-offload.c | 39 +++
1 file changed, 39 insertions(+)
diff --git a/lib/netdev-offload.c b/lib/netdev-offload.c
index 2da3bc701..f9c0b0438 100644
--- a/lib/netdev-offload.c
+++ b/lib/netdev-offl
This adds a reference program, flowtable_afxdp.o, which can be used to
offload flows to XDP through netdev-offload-xdp.
The program can be compiled with --enable-xdp-offload switch.
Signed-off-by: Toshiaki Makita
---
Makefile.am | 9 +-
acinclude.m4 | 37 ++-
bpf
.
$ ip link set veth1 xdpdrv object /path/to/ovs/bpf/xdp_noop.o section xdp
Signed-off-by: Toshiaki Makita
---
bpf/Makefile.am | 11 +++
bpf/xdp_noop.c | 31 +++
2 files changed, 38 insertions(+), 4 deletions(-)
create mode 100644 bpf/xdp_noop.c
diff --git a
lity to delete subtable entries.
v2:
- Add uninit callback of netdev-offload-xdp.
- Introduce "offload-driver" other_config to specify offload driver.
- Add --enable-bpf (HAVE_BPF) config option to build bpf programs.
- Workaround incorrect UINTPTR_MAX in x64 clang bpf build.
- Fix boot.sh au
d by OVS.
Note: kernel in AF_XDP CI test is updated to 5.5 because libbpf from 5.3
does not have newly used APIs like "bpf_program__get_type".
Signed-off-by: William Tu
Co-Authored-by: Toshiaki Makita
Signed-off-by: Toshiaki Makita
---
.travis.yml | 2 +-
Do
On 2020/07/28 3:14, William Tu wrote:
On Sun, Jul 26, 2020 at 9:55 AM Toshiaki Makita
wrote:
snip
How about doing s.t like:
--enable-afxdp: the current one on master without the xdp offload program
--enable-afxdp-with-bpf: the afxdp one plus your xdp offload program
So that when users
Hi William,
On 2020/07/16 23:09, Toshiaki Makita wrote:
On 2020/07/16 10:30, William Tu wrote:
On Mon, Jun 29, 2020 at 8:30 AM Toshiaki Makita
wrote:
From: William Tu
Now netdev-afxdp always forwards all packets to userspace because
it is using libbpf's default XDP program
Hi William,
On 2020/07/16 23:05, Toshiaki Makita wrote:
Hi William,
Thank you for taking a look!
On 2020/07/16 0:49, William Tu wrote:
Hi Toshiaki,
Thanks for the interesting patch series!
I haven't finished reviewing, but I went through a couple of
steps to make sure it works
William,
On 2020/07/22 17:46, Toshiaki Makita wrote:
On 2020/07/22 3:10, William Tu wrote:
Thanks for the patch. My comments below:
On Mon, Jun 29, 2020 at 8:30 AM Toshiaki Makita
wrote:
...
diff --git a/lib/automake.mk b/lib/automake.mk
index 86940ccd2..1fa1371f3 100644
--- a/lib
On 2020/07/22 3:10, William Tu wrote:
Thanks for the patch. My comments below:
On Mon, Jun 29, 2020 at 8:30 AM Toshiaki Makita
wrote:
This provider offloads classifier to software XDP.
It works only when a custom XDP object is loaded by afxdp netdev.
The BPF program needs to implement
On 2020/07/22 0:34, William Tu wrote:
On Tue, Jul 7, 2020 at 2:07 AM Toshiaki Makita
wrote:
On 2020/06/30 0:30, Toshiaki Makita wrote:
...
int netdev_afxdp_init(void)
{
libbpf_set_print(libbpf_print);
-return 0;
+return netdev_register_flow_api_provider
On 2020/07/22 0:38, Aaron Conole wrote:
William Tu writes:
On Tue, Jun 30, 2020 at 12:11 AM Toshiaki Makita
wrote:
On 2020/06/30 1:17, 0-day Robot wrote:
Bleep bloop. Greetings Toshiaki Makita, I am a robot and I have tried out your
patch.
Thanks for your contribution.
I encountered
On 2020/07/17 1:14, William Tu wrote:
On Mon, Jun 29, 2020 at 8:30 AM Toshiaki Makita
wrote:
The following commit will introduce another offload driver using XDP.
When using afxdp netdev, both of TC and XDP will be supported, so let's
add an other_config to specify which offload driv
On 2020/07/16 10:30, William Tu wrote:
On Mon, Jun 29, 2020 at 8:30 AM Toshiaki Makita
wrote:
From: William Tu
Now netdev-afxdp always forwards all packets to userspace because
it is using libbpf's default XDP program, see 'xsk_load_xdp_prog'.
There are some cases when use
:29AM +0900, Toshiaki Makita wrote:
...
* How to use
1. Install clang/llvm >= 9, libbpf >= 0.0.6 (included in kernel 5.5), and
kernel >= 5.3.
Encounter an error:
lib/netdev-offload-xdp.c: In function ‘probe_meta_info’:
lib/netdev-offload-xdp.c:386:19: error: implicit declarati
On 2020/06/30 0:30, Toshiaki Makita wrote:
This patch adds an XDP-based flow cache using the OVS netdev-offload
flow API provider. When an OVS device with XDP offload enabled,
packets first are processed in the XDP flow cache (with parse, and
table lookup implemented in eBPF) and if hits, the
On 2020/06/30 0:30, Toshiaki Makita wrote:
...
int netdev_afxdp_init(void)
{
libbpf_set_print(libbpf_print);
-return 0;
+return netdev_register_flow_api_provider(&netdev_offload_xdp);
This causes duplicate flow api provider error because afxdp and afxdp-nonpmd
are using
On 2020/06/30 1:17, 0-day Robot wrote:
Bleep bloop. Greetings Toshiaki Makita, I am a robot and I have tried out your
patch.
Thanks for your contribution.
I encountered some error that I wasn't expecting. See the details below.
checkpatch:
WARNING: Comment with 'xxx' marker
d by OVS.
Note: kernel in AF_XDP CI test is updated to 5.5 because libbpf from 5.3
does not have newly used APIs like "bpf_program__get_type".
Signed-off-by: William Tu
Co-Authored-by: Toshiaki Makita
Signed-off-by: Toshiaki Makita
---
.travis.yml | 2 +-
Do
This adds a reference program, flowtable_afxdp.o, which can be used to
offload flows to XDP through netdev-offload-xdp.
The program can be compiled with --enable-bpf switch.
Signed-off-by: Toshiaki Makita
---
Makefile.am | 9 +-
acinclude.m4 | 57
bpf/.gitignore
etdev supports it.
Signed-off-by: Toshiaki Makita
---
lib/netdev-offload.c | 37 +
1 file changed, 29 insertions(+), 8 deletions(-)
diff --git a/lib/netdev-offload.c b/lib/netdev-offload.c
index ab97a292e..669513646 100644
--- a/lib/netdev-offload.c
+++ b/lib/n
e "offload-driver" other_config to specify offload driver.
- Add --enable-bpf (HAVE_BPF) config option to build bpf programs.
- Workaround incorrect UINTPTR_MAX in x64 clang bpf build.
- Fix boot.sh autoconf warning.
Toshiaki Makita (3):
netdev-offload: Add "offload-driver&
ore details, refer to the reference BPF program in next commit.
In the future it may be possible to offload classifier to SmartNIC
through XDP, but currently it's not because map-in-map is not supported
by XDP hw-offload.
Signed-off-by: Toshiaki Makita
---
lib/automake.mk |
Hi William,
On 2020/05/08 0:08, Toshiaki Makita wrote:
On 2020/05/05 23:29, William Tu wrote:
On Tue, Apr 21, 2020 at 11:47:00PM +0900, Toshiaki Makita wrote:
This patch adds an XDP-based flow cache using the OVS netdev-offload
flow API provider. When an OVS device with XDP offload enabled
Hi Tonghao,
On 2020/05/14 13:08, Tonghao Zhang wrote:
On Tue, Apr 21, 2020 at 10:48 PM Toshiaki Makita
wrote:
This adds a reference program, flowtable_afxdp.o, which can be used to
offload flows to XDP through netdev-offload-xdp.
The program will be compiled when using --enable-bpf switch
On 2020/05/09 0:05, William Tu wrote:
On Thu, May 7, 2020 at 7:56 AM Toshiaki Makita
wrote:
On 2020/05/06 0:18, William Tu wrote:
On Tue, May 05, 2020 at 11:43:58AM +0900, Toshiaki Makita wrote:
On 2020/05/04 1:22, William Tu wrote:
On Tue, Apr 21, 2020 at 11:47:04PM +0900, Toshiaki Makita
On 2020/05/05 23:29, William Tu wrote:
On Tue, Apr 21, 2020 at 11:47:00PM +0900, Toshiaki Makita wrote:
This patch adds an XDP-based flow cache using the OVS netdev-offload
flow API provider. When an OVS device with XDP offload enabled,
packets first are processed in the XDP flow cache (with
On 2020/05/05 23:31, William Tu wrote:
On Tue, May 05, 2020 at 11:50:38AM +0900, Toshiaki Makita wrote:
On 2020/05/05 1:24, William Tu wrote:
On Tue, Apr 21, 2020 at 11:47:00PM +0900, Toshiaki Makita wrote:
This patch adds an XDP-based flow cache using the OVS netdev-offload
flow API provider
On 2020/05/06 0:18, William Tu wrote:
On Tue, May 05, 2020 at 11:43:58AM +0900, Toshiaki Makita wrote:
On 2020/05/04 1:22, William Tu wrote:
On Tue, Apr 21, 2020 at 11:47:04PM +0900, Toshiaki Makita wrote:
This adds a reference program, flowtable_afxdp.o, which can be used to
offload flows to
On 2020/05/06 0:37, William Tu wrote:
On Tue, Apr 21, 2020 at 11:47:03PM +0900, Toshiaki Makita wrote:...
+/* Convert odp_port to devmap_idx in output action */
+static int
+convert_port_to_devmap_idx(struct nlattr *actions, size_t actions_len)
+{
+struct nlattr *a;
+unsigned int left
On 2020/05/05 1:24, William Tu wrote:
On Tue, Apr 21, 2020 at 11:47:00PM +0900, Toshiaki Makita wrote:
This patch adds an XDP-based flow cache using the OVS netdev-offload
flow API provider. When an OVS device with XDP offload enabled,
packets first are processed in the XDP flow cache (with
On 2020/05/04 1:22, William Tu wrote:
On Tue, Apr 21, 2020 at 11:47:04PM +0900, Toshiaki Makita wrote:
This adds a reference program, flowtable_afxdp.o, which can be used to
offload flows to XDP through netdev-offload-xdp.
The program will be compiled when using --enable-bpf switch.
Signed-off
Hi William,
On 2020/05/01 0:26, William Tu wrote:
On Tue, Apr 21, 2020 at 11:47:02PM +0900, Toshiaki Makita wrote:
The following commit will introduce another offload driver using XDP.
When using afxdp netdev, either of TC or XDP will be supported, so let's
add an other_config to specify
XDP program when
loading it.
(More explanation is TBD)
In the future it may be possible to offload classifier to SmartNIC
through XDP, but it will require map-in-map support in NIC HW or
TCAM-like map support in BPF and HW which currently does not exist.
Signed-off-by: Toshiaki Makita
---
lib
This adds a reference program, flowtable_afxdp.o, which can be used to
offload flows to XDP through netdev-offload-xdp.
The program will be compiled when using --enable-bpf switch.
Signed-off-by: Toshiaki Makita
---
Makefile.am | 9 +-
acinclude.m4 | 56 +
bpf
by OVS.
Signed-off-by: William Tu
Co-Authored-by: Toshiaki Makita
Signed-off-by: Toshiaki Makita
---
Documentation/intro/install/afxdp.rst | 59 ++
NEWS | 2 +
lib/netdev-afxdp.c| 151 --
lib/netdev-
The following commit will introduce another offload driver using XDP.
When using afxdp netdev, either of TC or XDP will be supported, so let's
add an other_config to specify which offload driver is preferable.
When not specified, TC will be used if netdev supports it.
Signed-off-by: Tos
it, maybe it's time to switch to using libbpf standalone repository?
- Fix a crash bug in patch 1 which has been reported by Eelco Chaudron.
- Add test for XDP offload driver.
- Add documentation.
- Implement more actions like vlan push/pop.
Toshiaki Makita (3):
netdev-offload: Add "offl
ugh the v1.
Thank you for reading :)
If you have v2, please send it.
Now I'm preparing it.
Toshiaki Makita
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
On 2020/03/14 2:48, Ben Pfaff wrote:
On Wed, Mar 11, 2020 at 12:29:06AM +0900, Toshiaki Makita wrote:
This patch adds an XDP-based flow cache using the OVS netdev-offload
flow API provider. When a OVS device with XDP offload enabled,
packets first are processed in the XDP flow cache (with
p.
FIXME: Temporarily using HAVE_AF_XDP switch to enable BPF programs
build.
TODO: Install the XDP program somewhere (maybe /usr/lib/openvswitch/bpf?).
Signed-off-by: Toshiaki Makita
---
Makefile.am | 10 +-
bpf/.gitignore| 4 +
bpf/Makefile.am | 56 +
bpf/bpf_miniflow.h
e to build assertion.
Use OVS_ALIGNED_VAR instead of pad. This works for clang BPF as well as
other compilers and architectures.
Signed-off-by: Toshiaki Makita
---
include/openvswitch/tun-metadata.h | 6 +-
1 file changed, 1 insertion(+), 5 deletions(-)
diff --git a/include/openvswitch/tun-met
*** THIS PATCH IS NOT INTENDED TO BE APPLIED ***
Temporarily comment out TC flow api provider for testing and register
xdp offload provider.
FIXME: Need a way to specify flow api provider from configuration.
Signed-off-by: Toshiaki Makita
---
lib/netdev-afxdp.c| 3 ++-
lib/netdev
tion is
supported by the program. Please see probe_supported_keys() and
probe_supported_actions() in netdev-offload-xdp.c.
In the future it may be possible to offload classifier to SmartNIC
through XDP, but it will require map-in-map support in NIC HW or
TCAM-like map support in BPF and HW whi
ll
consuming 100% cpu. This means we need probably only one pmd for xdp
even when we want to use more cores for multi-flow.
I'll also test afxdp-nonpmd when it is applied.
This patch set is based on top of commit 59e994426 ("datapath: Update
kernel test list, news and FAQ").
[1]
by OVS.
Signed-off-by: William Tu
Co-Authored-by: Toshiaki Makita
Signed-off-by: Toshiaki Makita
---
v7:
- rebase to master
- apply Toshiaki's feedbacks and fixes
- Close BPF objects by using bpf_object__close()
Note that bpf_object__close() closes all program fds
and m
60 matches
Mail list logo
