PMD tunnel offload object can be referenced by application
tunnel_decap_set or tunnel_match commands, when these commands request
private tunnel items or actions and flow rules related to that tunnel.
PMD keeps tunnel offload object as long as there is an active
reference to that tunnel.
PMD releas
Hi Stephen,
the patch we submitted is safe, it has been verified that it indeed fixes the
bug:
Besides rmem and genmc model checkers, we verified the bugfix with herd7 tool,
which is the official tool ARM provides to check such pieces of code against
their
memory model.
Actually, I think there
On Tue, Nov 24, 2020 at 9:04 PM Ferruh Yigit wrote:
>
> On 10/26/2020 9:46 PM, Ajit Khaparde wrote:
> > Devices belonging to BCM573xx and BCM5740x family will not be supported
> > from the 21.02 release.
> >
> > Signed-off-by: Ajit Khaparde
> > ---
> > doc/guides/rel_notes/deprecation.rst | 6 +
Currently TRUFLOW is supported only on Whitney+ and Stingray devices.
Update the PMD doc with this info.
Signed-off-by: Ajit Khaparde
---
doc/guides/nics/bnxt.rst | 2 ++
1 file changed, 2 insertions(+)
diff --git a/doc/guides/nics/bnxt.rst b/doc/guides/nics/bnxt.rst
index 2f1ecffc6..055f1ff91
On Tue, Nov 24, 2020 at 4:15 PM Timothy Redaelli wrote:
>
> Commit 06c7871dde01 ("eal: restrict default plugin path to shared lib mode")
> introduced a check that enabled shared lib mode when librte_eal.so can
> be loaded, but it can't work, at least, on Fedora/CentOS/RHEL since .so
> symlinks are
On 11/24/2020 9:37 PM, Thomas Monjalon wrote:
24/11/2020 18:45, Stephen Hemminger:
Could we kill the old dpdk-setup.sh script?
Or add for 20.11 a notice "this script will go away, learn to use other ones..."
Yes we should add a deprecation warning in the script,
and plan its removal for DPDK 2
Support outer L3 and inner L3 + L4 Tx checksum offload for GTP-U packets
that contain extension header on FVL.
Support outer L3 + L4 and inner L3 + L4 Rx and Tx checksum offload for
GTP-U packets that contain extension header on CVL.
Murphy Yang (2):
net/i40e: support GTP checksum offload
app/
To support Tx side offload on GTP tunneling packet, driver should
set correct tunneling parameters of 'L4TUNT'.
Signed-off-by: Murphy Yang
---
drivers/net/i40e/i40e_rxtx.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/net/i40e/i40e_rxtx.c b/drivers/net/i40e/i40e_rxtx.c
index 5df9a9
To enable Tx side offload on GTP-U packets that contain extension header,
testpmd should set 'info->l2_len' correctly, this value should contain UDP
header length, GTP-U header length and GTP-U extension header length.
Signed-off-by: Murphy Yang
---
app/test-pmd/csumonly.c | 26 +
PMD did not remove tunnel offload object from tunnels database before
it released the object memory. As the result, the tunnels database
become corrupted and subsequent search operations triggered PMD crash.
The patch removes tunnel offload object from the tunnels database when
the object is not in
From: Alvin Zhang
The Tx buffer may overflow when there is more than one port.
Signed-off-by: Alvin Zhang
Fixes: 002ade70e933 ("app/test: measure cycles per packet in Rx/Tx")
Cc: sta...@dpdk.org
---
app/test/test_pmd_perf.c | 18 --
1 file changed, 8 insertions(+), 10 deletion
On Tue, Nov 24, 2020 at 4:15 PM Timothy Redaelli wrote:
>
> Commit 49b536fc3060 ("eal: load only shared libs from driver plugin
> directories")
> introduced a check that any shared library must ends with .so, but it can't
> work, at least, on Fedora/CentOS/RHEL since .so symlinks are not installe
On 11/24/20 4:14 PM, Timothy Redaelli wrote:
> Currently eal only tries to find shared lib by looking for .so files,
> but on Fedora, CentOS and RHEL the .so file is not installed by dpdk
> package, but it's only installed by dpdk-devel package, since .so files
> should not be necessary in order
when receive packets, the max bunch number of mbuf are allocated
if hardware does not receive the max bunch number packets, it
will free redundancy mbuf, this is low performance
so optimize rx performance, by allocating number of mbuf based on
result of xsk_ring_cons__peek, to avoid to redundancy
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Hideyuki Yamashita
> Sent: Wednesday, November 25, 2020 6:40 AM
>
> Hello,
>
> Following are the work items planned for 21.02 from NTT TechnoCross:
> I will try to post patch set after 20.11 is released.
>
> ---
> 1) Introduce API stats func
On 11/25/2020 10:01 AM, David Marchand wrote:
On Tue, Nov 24, 2020 at 4:15 PM Timothy Redaelli wrote:
Commit 49b536fc3060 ("eal: load only shared libs from driver plugin
directories")
introduced a check that any shared library must ends with .so, but it can't
work, at least, on Fedora/CentOS/
+/* LACP Rate */
+#define LACP_RATE_SLOW (0)
+#define LACP_RATE_FAST (1)
parentheses here can be deleted.
/* Set balance mode transmit policy*/
blank space shoud be added before "*/".
+ if (rte_eth_bond_lacp_rate_set(port_id, lacp_rate)
+
25/11/2020 10:44, Gregory Etelson:
> PMD did not remove tunnel offload object from tunnels database before
> it released the object memory. As the result, the tunnels database
> become corrupted and subsequent search operations triggered PMD crash.
> The patch removes tunnel offload object from the
> > In the function rte_flow_shared_action_destroy(), the errno
> > ETOOMANYREFS has been replaced with EBUSY in the commit dc328d1c555a
> > ("ethdev: rename a flow shared action error code").
> >
> > Another occurrence of ETOOMANYREFS, added later by mistake, is replaced
> > with EBUSY errno.
> >
On Mon, Nov 23, 2020 at 6:49 PM Cristian Dumitrescu
wrote:
> diff --git a/doc/guides/sample_app_ug/pipeline.rst
> b/doc/guides/sample_app_ug/pipeline.rst
> new file mode 100644
> index 0..9b52d8a22
> --- /dev/null
> +++ b/doc/guides/sample_app_ug/pipeline.rst
> @@ -0,0 +1,118 @@
> +.. SP
> Subject: [dpdk-dev] [PATCH][v4] net/af_xdp: optimize RX path by removing
> the unneeded allocation mbuf
>
> when receive packets, the max bunch number of mbuf are allocated
> if hardware does not receive the max bunch number packets, it
> will free redundancy mbuf, this is low performance
>
> s
25/11/2020 03:23, Jiang, YuX:
> Tested-by: Jiang, YuX
Thanks for the test.
Next time, please reply below with this exact line (name):
Tested-by: Yu Jiang
> > -Original Message-
> > From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Stephen
> > Hemminger
> > Sent: Wednesday, November
24/11/2020 18:53, Stephen Hemminger:
> On Tue, 24 Nov 2020 09:46:56 +
> Bruce Richardson wrote:
>
> > The "misc" and "regex" device classes were missing from the list used to
> > check arguments, preventing them from being used with "--status-dev"
> > flag to list only devices of those types.
David Marchand writes:
> With the recent changes in terms of free access to the Travis CI, let's
> offer an alternative with Github Actions.
> Running jobs on ARM is not supported unless using external runners, so
> this commit only adds builds for x86_64 and cross compiling for i386 and
> aarch6
On 11/24/2020 12:23 PM, Ferruh Yigit wrote:
On 11/24/2020 10:27 AM, Thomas Monjalon wrote:
Is it OK to keep this regression?
Ferruh, what do you suggest?
I confirm the '--txpkts' parameter is broken now, I suggest submitting a defect
for it and continue with the regression.
Hi Slava,
C
> -Original Message-
> From: David Marchand
> Sent: Wednesday, November 25, 2020 1:00 PM
> To: Dumitrescu, Cristian
> Cc: dev
> Subject: Re: [PATCH V2] doc: add pipeline example to user guide
>
> On Mon, Nov 23, 2020 at 6:49 PM Cristian Dumitrescu
> wrote:
> > diff --git a/doc/guides
> > The initialization me->locked=1 in lock() must happen before
> > next->locked=0 in unlock(), otherwise a thread may hang forever,
> > waiting me->locked become 0. On weak memory systems (such as ARMv8),
> > the current implementation allows me->locked=1 to be reordered with
> >
On Mon, Nov 23, 2020 at 6:49 PM Cristian Dumitrescu
wrote:
>
> Document the pipeline example in the Sample Application User Guide.
>
> Signed-off-by: Cristian Dumitrescu
Applied, thanks.
--
David Marchand
On Wed, Nov 25, 2020 at 2:45 PM Aaron Conole wrote:
> Thanks for working on this. Sadly, I think we will have to abandon
> Travis soon - given the new changes it is looking very awful. Robot
> already is starved for job time.
>
> Since we don't have ARM test runs, I guess we will have to rely on
This patch adds missing backend features negotiation for
in Vhost-vDPA. Without it, IOTLB messages v2 could be sent
by Virtio-user PMD while not supported by the backend.
Fixes: 6b901437056e ("net/virtio: introduce vhost-vDPA backend")
Cc: sta...@dpdk.org
Signed-off-by: Maxime Coquelin
---
driv
Vhost-vDPA support in Virtio-user PMD was lacking
backend features negotiation and IOTLB batching
support, causing intialization issues with Mellanox
vDPA device managed by Kernel vDPA framework.
Changes in v1:
=
- Add backend features negotiation (David, Jason)
- Ensure Kernel support
This patch fixes an overhead met with mlx5-vdpa Kernel
driver, where for every page in the mapped area, all the
memory tables gets updated. For example, with 2MB hugepages,
a single IOTLB_UPDATE for a 1GB region causes 512 memory
updates on mlx5-vdpa side.
Using batching mode, the mlx5 driver will
> -Original Message-
> From: dev On Behalf Of Timothy Redaelli
> Sent: Tuesday, November 24, 2020 5:14 PM
> To: Anatoly Burakov
> Cc: bruce.richard...@intel.com; dev@dpdk.org; sta...@dpdk.org
> Subject: [dpdk-dev] [PATCH 0/2] Fix shared lib detection on
> Fedora/CentOS/RHEL
>
Tested-by:
The function should return success if called with no active VF's.
It is OK to call this function if the desired outcome is to change
the flag for PF's even if there are no VF's.
Original code may have worked by accidental uninitialized variable;
then this bug was introduced by change to fix compil
This is an effort to update the dpdk-setup.sh script and keep it one
more year until whole functionality of it is replaced.
Other option is to remove it completely, please comment.
Ferruh Yigit (6):
usertools/setup: remove make based build
usertools/setup: remove inserting custom kernel modul
DPDK build system is switched to 'meson' based build, please check the
documentation for details: "doc/guides/linux_gsg/build_dpdk.rst"
Signed-off-by: Ferruh Yigit
---
usertools/dpdk-setup.sh | 74 +++--
1 file changed, 4 insertions(+), 70 deletions(-)
diff -
Custom DPDK kernel module insert support relies on make based build
variable 'RTE_TARGET' to find the location of the kernel modules, which
is not valid anymore.
Also 'igb_uio' kernel module moved to another git repository:
https://git.dpdk.org/dpdk-kmods/
Signed-off-by: Ferruh Yigit
---
userto
Running application relies on make based build variable 'RTE_TARGET' to
find the location of the applications, which is not valid anymore.
Signed-off-by: Ferruh Yigit
---
usertools/dpdk-setup.sh | 52 ++---
1 file changed, 2 insertions(+), 50 deletions(-)
dif
There is a specific tool to setup hugepages, with better support, no
need to have duplication.
The hugepage script is: './usertools/dpdk-hugepages.py'
Please try "./usertools/dpdk-hugepages.py -h" for more information about
the tool.
Signed-off-by: Ferruh Yigit
---
usertools/dpdk-setup.sh | 14
script is checking the existing of the kernel module file, but in some
distros kernel modules are stored compressed, like as 'vfio-pci.ko.xz'.
Add wildcard to cover both compressed and not compressed cases.
Signed-off-by: Ferruh Yigit
---
usertools/dpdk-setup.sh | 2 +-
1 file changed, 1 insert
There are still some functions in the script that not replaced
completely, give more time to complete removal of the tool.
Current functionality supported by the script is:
--
Step 1: Setup linux environment
---
Almost every PMD seems to do something like this (igb in this example):
autoneg = (*speeds & ETH_LINK_SPEED_FIXED) == 0;
But testpmd’s parse_and_check_speed_duplex() doesn’t ever set the FIXED bit. It
either sets AUTONEG (0) or the bit for one of the individual speeds.
Thus when I run a
24/11/2020 07:45, Guoyang Zhou:
> The maximum number of queues on the chip can be any value greater
> than 0, it does not have to be the power of two, if it does not be
> fixed, dpdk initialization may be failed with OVS mode firmware
> and remove the operation of optical modules in the link functi
24/11/2020 16:13, Timothy Redaelli:
> Currently eal only tries to find shared lib by looking for .so files,
> but on Fedora, CentOS and RHEL the .so file is not installed by dpdk
> package, but it's only installed by dpdk-devel package, since .so files
> should not be necessary in order to run a pr
24/11/2020 15:41, Ajit Khaparde:
> On Wed, Nov 18, 2020 at 8:15 AM Gregory Etelson wrote:
> >
> > There is a discrepancy between RTE ETHDEV API and flow rules guide
> > regarding flow rules maintenance after port stop. RTE ETHDEV API in
> > librte_ethdev.h declares that flow rules will not be sto
24/11/2020 14:22, Ferruh Yigit:
> On 11/24/2020 11:49 AM, Viacheslav Ovsiienko wrote:
> > If there was no RTE_NET_I40E configured the static routine
> > str2flowtype() was not used causing compilation warning.
> >
> > The str2flowtype() is moved under #ifdef RTE_NET_I40E block.
> >
> > Fixes: 1be
> >> There was a typo in eCPRI header definition.
> >>
> >> Fixes: d164c609e70b ("ethdev: add eCPRI key fields to flow API")
> >> Cc: sta...@dpdk.org
> >>
> >> Reported-by: Rani Sharoni
> >> Signed-off-by: Viacheslav Ovsiienko
> >
> > Reviewed-by: Bing Zhao
>
> Reviewed-by: Ferruh Yigit
Appl
Hello Morten,
Thanks for your giving me your valuable feedback.
Please see inline tagged with [Hideyuki].
> > From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Hideyuki Yamashita
> > Sent: Wednesday, November 25, 2020 6:40 AM
> >
> > Hello,
> >
> > Following are the work items planned for 21
It looks good to me, thanks.
what scenarios may cause bugs in old ways.
Could you give an example, thanks.
It looks good to me, thanks.
>
> On Wed, Nov 25, 2020 at 2:45 PM Aaron Conole wrote:
> > Thanks for working on this. Sadly, I think we will have to abandon
> > Travis soon - given the new changes it is looking very awful. Robot
> > already is starved for job time.
I am looking at [1], is DPDK not considered as open sourc
https://bugs.dpdk.org/show_bug.cgi?id=584
Bug ID: 584
Summary: testpmd --txpkts command line parameter is ignored
silently
Product: DPDK
Version: 20.11
Hardware: All
OS: All
Status: UNCONFIRMED
The bug: https://bugs.dpdk.org/show_bug.cgi?id=584
Can we pass the nb_segs = 1 always?
One descriptor is minimal basic capability to send, it should be always
supported.
With best regards, Slava
> -Original Message-
> From: Ferruh Yigit
> Sent: Wednesday, November 25, 2020 16:07
> To:
54 matches
Mail list logo
