;dpif-netdev/mfex: Add AVX512 based optimized miniflow
extract")
Fixes: b366fa2f4947 ("dpif-netdev: Call cpuid for x86 isa availability.")
Reported-at: https://bugzilla.redhat.com/2100393
Reported-by: Ales Musil
Co-authored-by: Ales Musil
Signed-off-by: Ales Musil
Signed-off-by: Davi
;dpif-netdev/mfex: Add AVX512 based optimized miniflow
extract")
Fixes: b366fa2f4947 ("dpif-netdev: Call cpuid for x86 isa availability.")
Reported-at: https://bugzilla.redhat.com/2100393
Reported-by: Ales Musil
Co-authored-by: Ales Musil
Signed-off-by: Ales Musil
Signed-off-by: Davi
;dpif-netdev/mfex: Add AVX512 based optimized miniflow
extract")
Fixes: b366fa2f4947 ("dpif-netdev: Call cpuid for x86 isa availability.")
Reported-at: https://bugzilla.redhat.com/2100393
Reported-by: Ales Musil
Co-authored-by: Ales Musil
Signed-off-by: Ales Musil
Signed-off-by: Davi
hread pthread_create.c
>
> SUMMARY: AddressSanitizer: 38656 byte(s) leaked in 2 allocation(s).
>
> Fixes: 9ac84a1a3698 ("dpif-avx512: Add ISA implementation of dpif.")
> Signed-off-by: Ilya Maximets
Reviewed-by: David Marchand
--
David Marchand
_
act")
Fixes: b366fa2f4947 ("dpif-netdev: Call cpuid for x86 isa availability.")
Reported-at: https://bugzilla.redhat.com/2100393
Reported-by: Ales Musil
Co-authored-by: Ales Musil
Signed-off-by: Ales Musil
Signed-off-by: David Marchand
Acked-by: Sunil Pai G
---
Changes since
On Fri, Jun 24, 2022 at 9:30 AM David Marchand
wrote:
>
> As described in the bugzilla below, cpu_has_isa code may be compiled
> with some AVX512 instructions in it, because cpu.c is built as part of
> the libopenvswitchavx512.
> This is a problem when this function (supposed to p
iables when a simple boolean
is enough,
Fixes: b366fa2f4947 ("dpif-netdev: Call cpuid for x86 isa availability.")
Reported-at: https://bugzilla.redhat.com/2100393
Reported-by: Ales Musil
Signed-off-by: David Marchand
---
lib/automake.mk| 4 +--
lib/dpif-netd
-lrte_telemetry -lrte_kvargs -lbsd -lmlx4 -libverbs -lmlx5
-lpcap -lnuma -latomic -lm
For this error on cpu_has_isa to happen, the linker probably judged no
symbol from cpu.c could be used and decided to drop the whole .o.
So we need something to make sure those won't be dropped.
I just
On Wed, Jun 1, 2022 at 6:38 PM Kevin Traynor wrote:
>
> Test that user configured mempool params have been stored.
>
> Signed-off-by: Kevin Traynor
Reviewed-by: David Marchand
> ---
> tests/system-dpdk.at | 34 ++
> 1 file changed, 34 i
On Wed, Jun 1, 2022 at 6:38 PM Kevin Traynor wrote:
>
> Splitting them allows them to be reused separately. This
> is useful for setting some things in ovsdb before vswitchd is
> started or DPDK is initialized.
>
> Signed-off-by: Kevin Traynor
Reviewed-by: David Marchand
mempool based on 6000 MTU
> * MTU 9000, use mempool based on 9000 MTU
> * MTU 9300, use mempool based on 9300 MTU (existing behaviour)
>
> Default behaviour is unchanged and mempools are still only created
> when needed.
>
> Signed-off-by: Kevin Traynor
Reviewed
On Fri, Jun 3, 2022 at 5:16 PM Mike Pattrick wrote:
>
> From: Flavio Leitner
>
> This helps to identify when it is about the flags or
> the packet itself.
>
> Signed-off-by: Flavio Leitner
> Co-authored-by: Mike Pattrick
> Signed-off-by: Mike Pattrick
Reviewed-by:
On Fri, Jun 3, 2022 at 5:16 PM Mike Pattrick wrote:
>
> From: Flavio Leitner
>
> Rename to better represent their flags.
>
> Signed-off-by: Flavio Leitner
> Co-authored-by: Mike Pattrick
> Signed-off-by: Mike Pattrick
Reviewed-by: David Marcha
ddr eth_src,
lib/packets.c:compose_nd_na(struct dp_packet *b,
lib/packets.c:compose_nd_ra(struct dp_packet *b,
lib/packets.c:packet_put_ra_prefix_opt(struct dp_packet *b,
Did you skip those because they were not calling functions from the
dp_packet API?
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
On Fri, Jun 3, 2022 at 5:16 PM Mike Pattrick wrote:
>
> From: Flavio Leitner
>
> The name correlates better with the flag names.
>
> Signed-off-by: Flavio Leitner
> Co-authored-by: Mike Pattrick
> Signed-off-by: Mike Pattrick
Nit: suggest title prefix dp-packet:
Rev
ck
> Signed-off-by: Mike Pattrick
Nit: suggest title:
netdev: Prefix offload flags with NETDEV_OFFLOAD_
Reviewed-by: David Marchand
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
lgtm.
Reviewed-by: David Marchand
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
as not added, it could
> also be using uneeded memory.
>
> This patch delays the creation of the mempool for a vhost port until the
> vhost device is added.
>
> Signed-off-by: Kevin Traynor
> Reviewed-by: David Marchand
Thanks for aligning the behavior, and updating the commi
gned-off-by: Peng He
> Tested-by: David Marchand
> Acked-by: Eelco Chaudron
> ---
> ofproto/ofproto-dpif.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/ofproto/ofproto-dpif.c b/ofproto/ofproto-dpif.c
> index 6601f2346..0d0d985f0 100644
> --- a/ofproto/ofprot
// wait for rcu_cb done
+ * do_destroy_A(); // free A
+ * }
> */
>
> #include "compiler.h"
> @@ -310,4 +323,6 @@ void ovsrcu_synchronize(void);
>
> void ovsrcu_exit(void);
>
> +void ovsrcu_barrier(void);
> +
> #endif /* ovs-rcu.h */
> --
> 2.25.1
>
Reviewed-by: David Marchand
I think we should have a unit test for this new part of the ovs-rcu API.
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
; * MTU 9300, use mempool based on 9300 MTU (existing behaviour)
>
> Default behaviour is unchanged and mempools are still only created
> when needed.
>
> Signed-off-by: Kevin Traynor
This lgtm, I have some small nits below, but otherwise:
Reviewed-by: David Marcha
o ovs-lib.
> Maybe we can create a couple of variables (e.g: OVS_BIN_PATH and
> OVS_SBIN_PATH)
> in ovs-lib and let each script use the one it needs?
Sorry, it fell through the cracks..
Your proposal lgtm.
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
one mempool in numa1
- after patch, once vm in numa0 is started too,
mempool @0x11ffe01e40
mempool @0x17f51dcc0
Reviewed-by: David Marchand
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
> +esac
> +esac
> +done
> +
Seeing how we have similar code in other scripts (ovs-ctl,
ovs-kmod-ctl), would it make sense to move this to ovs-lib?
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
+ Value should be set in the following form:
> +
> +other_config:dpdk-shared-mempory-config=<
> + user-shared-mempool-mtu-list>
> +
> + where
> +
> +
> +
> +<user-shared-mempool-mtu-list> ::=
> +NULL | <non-empty-list>
> +
> +
> +<non-empty-list> ::= <user-mtus> |
> + <user-mtus> ,
> + <non-empty-list>
> +
> +
> +<user-mtus> ::= <mtu-all-socket> |
> + <mtu-socket-pair>
> +
> +
> +<mtu-all-socket> ::= <mtu>
> +
> +
> +<mtu-socket-pair> ::= <mtu> :
> <socket-id>
> +
> +
> +
> +
> + Changing this value requires restarting the daemon if dpdk-init has
> + already been set to true.
> +
> +
> +
>type='{"type": "integer",
> --
> 2.34.1
>
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
lp.
Seeing how PVP gets better, I think the performance drop in the
physical only usecase is acceptable.
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
On Tue, Apr 12, 2022 at 4:21 PM Kevin Traynor wrote:
>
> In some places it is using Markdown syntax and in others
> it is not needed as there is already a code block.
>
> Signed-off-by: Kevin Traynor
Patch lgtm, and rendered documentation is ok.
Reviewed-by: David Marcha
On Tue, Apr 12, 2022 at 4:20 PM Kevin Traynor wrote:
>
> My xml editor keeps autofixing these which means I have to be
> careful during 'git add' for unrelated changes. Might as well
> just fix them.
>
> Signed-off-by: Kevin Traynor
Reviewed-by: David Ma
gt; testsuite:test
> asan: asan
+1
>
> I might be wrong but it seems to me we need something similar for afxdp too.
The Ubuntu 18.04 kernel does not have AF_XDP support iirc, so either
we keep on building a 5.3 kernel and compile ovs with afxdp against
t
nymore, is it?
(since we are back with the code from the common helper under the lock)
>
> Signed-off-by: Flavio Leitner
> Reviewed-by: David Marchand
> Signed-off-by: Mike Pattrick
The robot complains about your sign-off, you should add yourself as co-author.
Good job on ident
Dropped packets were not counted as tx_dropped when a DPDK netdev is
down (like after calling netdev-dpdk/set-admin-state dpdk1 down).
Fixes: 3b1fb0779b87 ("netdev-dpdk: Don't call rte_dev_stop() in
update_flags().")
Signed-off-by: David Marchand
---
lib/netdev-dpdk.c | 3 +++
A lock annotation was left behind after removing the nonpmd mutex.
Remove it.
Fixes: 1166b0d82043 ("netdev-dpdk: Remove useless nonpmd_mempool_mutex.")
Signed-off-by: David Marchand
---
lib/netdev-dpdk.c | 1 -
1 file changed, 1 deletion(-)
diff --git a/lib/netdev-dpdk.c b/lib/net
nough, as the destinations
> are all 32-bit?
For storing/comparing to explicit uint32_t variables, either
(uint32_t)1 or UINT32_C(1) are more natural.
Any reason not to use those?
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
lication, so lowering optimizations to -O1 here.
CFLAGS_ASAN="-O1 -fno-omit-frame-pointer -fno-common -fsanitize=address"
CFLAGS_FOR_OVS="${CFLAGS_FOR_OVS} ${CFLAGS_ASAN}"
Then, those CFLAGS can be passed at ./configure time.
--
David Marchand
__
seq_destroy(ofproto->ams_seq);
+/* Wait for all the meter rules to be destroyed. */
+seq = seq_create();
+seqno = seq_read();
+ovsrcu_synchronize();
+ovsrcu_postpone__((void (*)(void *))seq_change, seq);
+do {
+seq_wait(seq, seqno);
+poll_block(
eq));
+ovsrcu_postpone__((void (*)(void *))seq_change, seq);
+poll_block();
+seq_destroy(seq);
+
close_dpif_backer(ofproto->backer, del);
}
Let's hope it passes the night tests :-).
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
gs.
visible*
>
> Signed-off-by: Kevin Traynor
I have two small comments (see below), but otherwise:
Acked-by: David Marchand
> ---
> lib/dpif-netdev.c | 59 +--
> 1 file changed, 32 insertions(+), 27 deletions(-)
>
> diff --g
here is no user visibile change in behaviour.
visible*
>
> Signed-off-by: Kevin Traynor
Acked-by: David Marchand
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
co Chaudron
This fix is straightforward.
I am a bit surprised that we lived with this bug for so long.
(PMD-related) unit tests looks good too.
Reviewed-by: David Marchand
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
ned-off-by: David Marchand
---
Changes since RFC v1:
- added a note in NEWS,
- waived warning/error logs in system-dpdk for runs with broken drivers
(like net/pcap, reported by AVX512 Intel CI),
---
NEWS | 4 +
lib/dpif-netdev.c | 85 +-
lib/netdev-dpdk.c
On Wed, Mar 23, 2022 at 2:08 PM Mike Pattrick wrote:
>
> On Wed, Mar 23, 2022 at 4:23 AM David Marchand
> wrote:
> >
> > On Tue, Mar 22, 2022 at 6:43 PM Mike Pattrick wrote:
> > >
> > > In glibc 2.32 sys_siglist is no longer exported. The MT-safe functi
On Wed, Mar 23, 2022 at 3:09 PM Kevin Traynor wrote:
>
> Ensure that if there are no local numa PMD thread
> cores available that pmd cores from all other non-local
> numas will be used.
>
> Signed-off-by: Kevin Traynor
Acked-by: David Marchand
and a replacement
for direct access to sys_siglist... ?
--
David marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
On Tue, Mar 22, 2022 at 3:41 AM Harold Huang wrote:
>
> On Fri, Mar 18, 2022 at 11:33 PM David Marchand
> wrote:
> >
> > DPDK based dp-packets points to data buffers that can't be expanded
> > dynamically.
> > Their layout is as follows:
> > - a
So I guess you mean "multi segments" mbuf.
This patch only allows resizing mono segment mbuf.
I have no plan to revive the multi segments work.
Do you see an issue with proposed code change?
--
David Marchand
___
dev mailing l
s
good enough.
Thanks.
>
> Fixes: b4e50218a0f8 ("bond: Add 'primary' interface concept for active-backup
> mode.")
> Fixes: 9e11517e6ca6 ("ovs-router: Fix flushing of local routes.")
> Signed-off-by: Ilya Maximets
Reviewed-by: David Marchand
I
; Signed-off-by: Kevin Traynor
> Acked-by: Mike Pattrick
Acked-by: David Marchand
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
uld check the individual rxqs for pinning before we disallow
> PMD ALB.
>
> Also, update function comments to make it's operation clearer.
>
> Fixes: 6193e03267c1 ("dpif-netdev: Allow pin rxq and non-isolate PMD.")
But, in any case,
d: 0
> +port: p0 queue-id: 1
> +port: p0 queue-id: 2
> +port: p0 queue-id: 3
> +])
As stated in the comment, we only need to make sure we have a number
of pmd polling some rxq.
This can be done with an existing helper (which, btw, could also embed
the filter on AVAIL$ rather than du
n.
>
> Store the last numa used when successfully finding a non-local numa
> with available pmds, so the numa RR state is kept for subsequent rxqs.
>
> Fixes: f577c2d046b2 ("dpif-netdev: Rework rxq scheduling code.")
>
> Signed-off-by: Kevin Tr
empty line between Fixes: and the rest of the tags is something
Thomas insists on, on dpdk side.
But this is not a common practice in other projects (OVS included).
This empty line could be removed.
>
> Signed-off-by: Kevin Traynor
> Acked-by: Mike Pattrick
Acked-by: Da
ary' interface concept for active-backup
> mode.")
> Signed-off-by: Ilya Maximets
Otherwise the patch lgtm.
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
s),
a 128 bytes headroom is too short.
Dynamically allocate buffers in DPDK memory and make use of DPDK
external buffers API (previously used for userspace TSO).
Signed-off-by: David Marchand
---
lib/dp-packet.c | 17 -
lib/netdev-dpdk.c
On Thu, Mar 10, 2022 at 1:51 PM Ilya Maximets wrote:
>
> On 3/10/22 13:12, David Marchand wrote:
> > The hacking flake8 plugin generates a lot of noise in GHA logs about an
> > issue on the Apache license contained in OVS python files:
> > https://github.com/openvswitch/o
On Tue, Mar 8, 2022 at 8:44 PM Ilya Maximets wrote:
>
> On 3/8/22 17:39, David Marchand wrote:
> > On Mon, Mar 7, 2022 at 11:28 PM Ilya Maximets wrote:
> >>
> >> 'ofproto' should still be in place while freeing meter id, but this is
> >> not
from getting shot thanks to refcount, but I don't
understand what prevents backer content from being uninitialised (like
destroying the meter id pool) in close_dpif_backer().
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
/lib/ovs-rcu.c:362:14
#3 0x73a25c in ovsthread_wrapper
/home/runner/work/ovs/ovs/openvswitch-2.17.90/_build/sub/../../lib/ovs-thread.c:422:12
#4 0x7fb98a2fa6da in start_thread
(/lib/x86_64-linux-gnu/libpthread.so.0+0x76da)
#5 0x7fb98987971e in clone (/lib/x86_64-linux-gnu/libc.so.6+0x12171e)
There were previous reports from Ilya and others (me included).
Initial report:
https://mail.openvswitch.org/pipermail/ovs-dev/2021-February/380582.html
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
ned-off-by: David Marchand
---
lib/dpif-netdev.c | 86 +-
lib/netdev-dpdk.c | 199 +++---
lib/netdev-provider.h | 8 ++
lib/netdev.c | 9 ++
lib/netdev.h | 1 +
5 files changed, 272 insertions(+), 31 deletions(-)
di
han 1000+ flows being offloading simultanously.
>
> Wilcarding flow count number as for different systems
> under test the number could vary in the warning log.
>
> Suggested-by: Eelco Chaudron
> Signed-off-by: Kumar Amber
Though I can't reproduce the issue, the patch lgtm.
Revie
ds can already be retrieved from
existing ethdev API without using DPDK interrupt thread/framework.
>
> In any case, some extra engineering will be required to support vhost
> rx interrupts even without dma.
I have a PoC in my branches.
I'll send a RFC on this topic, afte
DPDK 21.11.
+1
>
> [0]
> http://git.dpdk.org/dpdk/commit/?id=295968d1740760337e16b0d7914875c5cac52850
>
> Signed-off-by: Kevin Traynor
>
> ---
>
> N.B. This is not required for 2.17, the existing defines work fine.
LGTM, thanks Kevin.
Acked-by: David Marchand
--
David Marchand
I can't reproduce this warning on a AVX512 system with F34.
That may be a lack of luck on my side, like not enough good packets
generated... ?
In any case, there is nothing tightly binding this test with the 10k
current upcall flow limit.
A wildcard pattern rather than a 1
logs matching.")
Signed-off-by: David Marchand
---
tests/system-dpdk.at | 2 ++
1 file changed, 2 insertions(+)
diff --git a/tests/system-dpdk.at b/tests/system-dpdk.at
index 1dd7aae1b7..9384cf7f0f 100644
--- a/tests/system-dpdk.at
+++ b/tests/system-dpdk.at
@@ -5,8 +5,10 @@ AT_BANNER([OVS
mple, if the payload is a
> TCP packet, most probably the flag DP_PACKET_OL_TX_UDP_CKSUM doesn't
> make sense after decapsulating that.
>
> This patch skips the tunnel tests when the userspace TSO is enabled.
> Fixes: 29bb3093eb8b ("userspace: Enable TSO support for non-DPDK
hide this issue, but the memory allocator still expects mp
to be available, regardless of --in-memory or rte_mp_disable calls.
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
7060-2-david.march...@redhat.com/
I updated my repo and I'll send a fix on the ml.
https://github.com/david-marchand/ovs/commit/system-dpdk
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
On Wed, Jan 12, 2022 at 5:17 PM David Marchand
wrote:
>
> On Wed, Jan 12, 2022 at 4:48 PM Phelan, Michael
> wrote:
> >
> > Hi,
> >
> > During internal testing of the AVX-512 CI, a bug was picked up on the OVS
> > master branch which resulted in the
n list.@d
\@Error: unknown miniflow extract implementation superstudy.@d
\@Error: invalid study_pkt_cnt value: -pmd.@d
+\@Interface p1 does not support MTU configuration, max packet size
supported is 1500.@d
+\@Rx checksum offload is not supported on port 0@d
])")
AT_CLEANUP dnl
dnl
on gre
fields.")
Signed-off-by: David Marchand
---
NEWS | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/NEWS b/NEWS
index bc4a1cfacc..a654e675c0 100644
--- a/NEWS
+++ b/NEWS
@@ -12,7 +12,7 @@ Post-v2.16.0
* Add hardware offload support for
; # additional warnings in alphabetical order
> +'-Wcast-align',
> '-Wcast-qual',
> '-Wdeprecated',
> '-Wformat',
> ---
>
> But I don't know if that will cause any problems.
On Tue, Jan 4, 2022 at 12:34 PM Ilya Maximets wrote:
>
> On 1/3/22 15:15, David Marchand wrote:
> > The DPDK unit test only runs if vfio or igb_uio kernel modules are loaded:
> > on systems with only mlx5, this test is always skipped.
> >
> > Besides, the test
>
> 2. Without dummy-numa real CPU affinity will be applied. In that
>case during the parallel test execution threads from different tests
>will be scheduled on same cores. Dummy-numa fakes the CPU affinity
>allowing the kernel to re-schedule pmd threads t
> > AT_SETUP([OVS-DPDK - MFEX Configuration])
> > AT_KEYWORDS([dpdk])
> > +AT_SKIP_IF([! $PYTHON3 -c "import scapy"], [], [])
> > +
> > OVS_DPDK_START()
> > AT_CHECK([ovs-vsctl --no-wait set Open_vSwitch .
> > other_config:pmd-cpu-mask=0
> enabling hash based packet steering, ensuring packets
> are transmitted on both queues.
>
> Signed-off-by: Maxime Coquelin
Reviewed-by: David Marchand
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
3 deletions(-)
> create mode 100644 Documentation/topics/userspace-tx-steering.rst
I would announce this feature in NEWS.
Below are some comments, but nothing blocking for this feature.
Once fixed, you can add my:
Reviewed-by: David Marchand
>
> diff --git a/Documentation/automake
duce functionnal changes.
>
> Signed-off-by: Maxime Coquelin
Reviewed-by: David Marchand
> @@ -5804,7 +5810,8 @@ reconfigure_datapath(struct dp_netdev *dp)
> seq_change(dp->port_seq);
> port_destroy(port);
> } else {
> -
l "errors" counters in netdev-dummy.
>
> Signed-off-by: Maxime Coquelin
Similar to previous patch, I only have some nits.
The patch lgtm.
Reviewed-by: David Marchand
> @@ -954,6 +966,8 @@ static int
> netdev_dummy_reconfigure(struct netdev *netdev_)
> {
> struc
>
> This patch introduces "bytes", "packets" and "error"
> per-queue custom statistics for Vhost-user ports.
>
> Suggested-by David Marchand
> Signed-off-by: Maxime Coquelin
A couple of nits about a comment and coding style, but otherwise it lgtm
The MFEX code and tests do not depend on DPDK anymore.
We can move the unit tests to dpif-netdev.
Signed-off-by: David Marchand
Reviewed-by: Maxime Coquelin
Acked-by: Eelco Chaudron
Acked-by: Kumar Amber
---
Note: this patch depends on series
https://patchwork.ozlabs.org/project/openvswitch
that comes from EPEL.
Suggested-by: Ilya Maximets
Signed-off-by: David Marchand
Reviewed-by: Maxime Coquelin
Acked-by: Eelco Chaudron
---
Changes since v3:
- dropped documentation update following rebase,
- fixed regression in fuzzy packets generation,
Changes since v2:
- updated documentation,
PCI_ADDR is renamed
to DPDK_PCI_ADDR.
Note: with mlx5 devices, there is now more OVS/DPDK warnings to waive.
Signed-off-by: David Marchand
Reviewed-by: Maxime Coquelin
Acked-by: Eelco Chaudron
---
Changes since v3:
- fixed nit from Maxime,
Changes since v2:
- sorted logs alphabetically
t/?id=c1077933d45b
3: https://git.dpdk.org/dpdk/commit/?id=e9b3d79b0696
4: https://git.dpdk.org/dpdk/commit/?id=c69150679891
Signed-off-by: David Marchand
Reviewed-by: Maxime Coquelin
Acked-by: Eelco Chaudron
---
Changes since v3:
- updated with v21.11 log updates,
- restored telemetry log,
Ch
efault, and can be enabled with:
> $ ovs-vsctl set open_vswitch . other_config:pmd-auto-lb="true"
>
> Signed-off-by: Kevin Traynor
> Acked-by: Sunil Pai G
Acked-by: David Marchand
--
David Marchand
___
dev
On Mon, Dec 20, 2021 at 3:39 PM Kevin Traynor wrote:
>
> Updates to the PMD Auto Load Balance section to make it more readable.
>
> No change to the core content.
>
> Signed-off-by: Kevin Traynor
> Acked-by: Sunil Pai G
Reviewed-by: David Marchand
s: links in ovs doc.
I see that there are more http than https links in the current documentation.
My preferrence would go to https: but I suppose both are acceptable.
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
On Mon, Dec 20, 2021 at 3:39 PM Kevin Traynor wrote:
>
> Some minor spelling and grammar fixes in pmd.rst.
>
> Signed-off-by: Kevin Traynor
> Acked-by: Sunil Pai G
Reviewed-by: David Marchand
--
David Marchand
___
dev
d0dae22a0e ("doc: Add "PMD" topic document")
> Signed-off-by: Kevin Traynor
> Acked-by: Sunil Pai G
Reviewed-by: David Marchand
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
; (with minor
> modifications for assert)
Ilya comments made sense, I was expeecting replies from your side but seen none.
So I understood this as you were ok with Ilya comments, so I only
updated the parts we agreed on previously.
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
cpuid API that comes from the compiler.
Note: this also fixes the check on BMI2 availability: DPDK had a bug
for this isa, see https://git.dpdk.org/dpdk/commit/?id=aae3037ab1e0.
Suggested-by: Ilya Maximets
Signed-off-by: David Marchand
---
Changes since v3:
- added an assert in cpuid check,
Ch
ely on
> Windows. And that never was a problem, AFAIK.
+1
I'll keep the current implementation, but respin with an ovs_assert()
where needed as agreed previously.
More high level logs might be added in dpif in the future.
Thanks.
--
David Marchand
__
nal authorship (I hope that git send-email tracked that
>correctly as I didn't re-send someone else's patches for a while).
> - Minor rebase, typo fix.
> - Re-sending to remind others about the patch and address all the
>
Waiting only on the vhost user port to be ready is not enough since a
tap is also initialized by testpmd and is used to inject/receive packets
in/from the kernel.
Wait on the tap link status.
Fixes: 18db7ec5eb83 ("system-dpdk: Improve vhost-user ping tests reliability.")
Signed-off
356,7 +356,64 @@ The phy test will skip if no compatible physical
> >> device is available.
> >> .. _Configure hugepages:
> >> https://doc.dpdk.org/guides-21.11/linux_gsg/sys_reqs.html
> >>
> >> All the features documented under `Unit Tests`_ are av
On Mon, Dec 6, 2021 at 11:16 PM Ilya Maximets wrote:
>
> On 11/30/21 16:00, David Marchand wrote:
> > net_pcap is not always available in DPDK (like, in a dev
> > environment when you forgot to install the libpcap-devel).
> > On the other hand, OVS already has its own way
me parts of the doc still needs some re-wording, but this change
> provides at least a starting point for improvement setting a better
> structure for the document.
+1.
>
> Signed-off-by: Ilya Maximets
I don't mind if you drop some of my suggestion / nits (see inline ),
this
On Wed, Nov 10, 2021 at 5:54 PM David Marchand
wrote:
>
> Previously in OVS, a PMD thread running on cpu X used lcore X.
> This assumption limited OVS to run PMD threads on physical cpu <
> RTE_MAX_LCORE.
>
> DPDK 20.08 introduced a new API that associates a non-EAL thread to
On Thu, Dec 9, 2021 at 3:51 PM Maxime Coquelin
wrote:
> On 12/7/21 21:37, David Marchand wrote:
> >> +struct netdev_dpdk_vhost_q_stats {
> >> +uint64_t bytes;
> >> +uint64_t packets;
> >> +uint64_t errors;
> >> +};
> >> +
> &
> For credit all authors of the original commits to 'dpdk-latest' with the above
> changes have been added as co-authors for this commit.
>
> Signed-off-by: David Marchand
> Co-authored-by: David Marchand
> Reviewed-by: Maxime Coquelin
> Signed-off-by: Ian Stokes
&
es=2414
tx_dropped=1
tx_packets=20
tx_q0_bytes=2414
tx_q0_packets=20
So we end up with misalignment of "global" stats and per q stats in
this situation.
Can we do something about it?
I am unsure about resetting all stats to 0, but it could be the
simpler solution...
--
David Marchand
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
On Tue, Dec 7, 2021 at 11:11 PM Maxime Coquelin
wrote:
> On 12/7/21 21:37, David Marchand wrote:
> > On Wed, Nov 24, 2021 at 10:24 PM Maxime Coquelin
> > wrote:
> >>
> >> HXPS feature will enable steering Tx packets on transmist
> >
> > transmit*
&g
501 - 600 of 1011 matches
Mail list logo
