u should start a
discussion on that topic on the d...@dpdk.org mailing list. Without noise,
nothing will happen :) As usual contributions are always welcome!
--
Adrien Mazarguil
6WIND
ile doing
Rx, right before returning them to the application to avoid starvation.
I do not think DPDK applications are ready for 2 and 3 yet. Also due to the
above, doing so would require extra work on the PMD side.
--
Adrien Mazarguil
6WIND
g/guides/nics/features.html
[2] http://doc.dpdk.org/guides/nics/overview.html
--
Adrien Mazarguil
6WIND
SS action).
- Rule with priority 1: match everything (i.e. empty pattern) and spread
that into the remaining queues (QUEUE or RSS action depending on the
number of target queues).
--
Adrien Mazarguil
6WIND
t week.
>
> Another question, is ixgbevf also support RTE_FLOW_ITEM_RAW?
The ixgbe driver looks like it does but I'm not sure regarding the VF
variant, Cc'ing maintainers just in case.
--
Adrien Mazarguil
6WIND
ch, this one should also match ICMP traffic inside recognized VLAN,
QinQ and tunnel types (e.g. VXLAN) at no extra cost.
--
Adrien Mazarguil
6WIND
never had a chance to use this pattern item with any PMDs
that support it, so I don't know close their behavior sticks to what I just
described, especially regarding the handling of mask.pattern. Do not
hesitate to report odd or inconsistent behavior in any case.
--
Adrien Mazarguil
6WIND
ght otherwise
dispatch it to a different queue.
Question is which drivers implement this use case. While I'm aware mlx5
supports PF/VF/representor redirections, I don't know if a VF can feed its
own representor like that. A true hairpin action is missing anyway.
--
Adrien Mazarguil
6WIND
rtemap_976
-rw--- 1 root root 2097152 May 14 10:13 rtemap_977
-rw--- 1 root root 2097152 May 14 10:13 rtemap_978
[...]
This means a mere user can't use them. In which case you need to chown/chmod
or remove them altogether.
Have you tried --no-huge?
--
Adrien Mazarguil
6WIND
r PMDs do not typically leave a kernel netdevice around for
external applications to play with.
--
Adrien Mazarguil
6WIND
with --no-huge).
If acceptable for your application, then you may try something like:
root# setcap cap_net_admin,cap_net_raw,cap_ipc_lock=ep ./build/l2fwd
user$ ./build/l2fwd [...]
Note the described approach adds capabilities through extended device
attributes on the file system (those are rarely preserved when copying
files). While finer grained than "chmod +s", you still need to carefully
consider the security implications, trust the program and its users.
Have a look at man capabilities(7).
--
Adrien Mazarguil
6WIND
recommended on instructions page
> "http://dpdk.org/doc/guides/nics/mlx4.html"; and it is interesting to
> bring it to working state.
> /Regards, Timur
rdma-core requires an up-to-date kernel. If this can't be satisfied,
MLNX_OFED is the only alternative (the mlnx-ofa_kerne
the above error message. -1
should be used, -7 is only for specific use cases and should probably not
even be documented in this file.
--
Adrien Mazarguil
6WIND
On Thu, Sep 21, 2017 at 11:17:59AM +0200, Adrien Mazarguil wrote:
> On Wed, Sep 20, 2017 at 06:33:43PM +0100, Kevin Traynor wrote:
> > On 09/08/2017 10:56 AM, Loftus, Ciara wrote:
> > >> Hi,
> > >>
> > >> I have compiled and built ovs-dpdk using DPDK v1
catenation of the associated Verbs device
name and the physical port number, e.g. "mlx4_0 port 0" instead of
":02:42.0".
Referring to a port with this name through the above function should yield
the correct device.
If necessary, the PCI address to Verbs device name conversion
[1]?
The related patches have been submitted and are on track for DPDK 17.11.
While they are not yet on the master tree, you can find them on
dpdk-next-net.
[1] http://dpdk.org/ml/archives/dev/2017-August/073081.html
--
Adrien Mazarguil
6WIND
not get much more than
this with small packets.
> > 在 2017年7月25日,02:20,Adrien Mazarguil 写道:
> >
> > Hi Zhilong,
> >
> > On Sat, Jul 22, 2017 at 12:05:51AM +0800, zhilong zheng wrote:
> >> Hi all,
> >>
> >> I have some problem when generating
ng on packet size and other configuration settings, you may have hit
the maximum packet rate, these devices cannot reach line rate with 64-byte
packets for instance.
[1] http://dpdk.org/doc/guides/nics/mlx4.html#prerequisites
--
Adrien Mazarguil
6WIND
com/page/mlxup_firmware_tool
[2] http://www.mellanox.com/page/firmware_table_ConnectX3EN
[3]
https://forums.servethehome.com/index.php?threads/flash-oem-ibm-mellanox-connectx-3-en-with-stock-firmware.8689/
--
Adrien Mazarguil
6WIND
u likely need to transfer the associated netdevices to
your Docker instance for this to work though. You'll likely need to somehow
transfer the related IB instances as well.
At least that's the theory, I've never personally tried to use Docker with
mlx4/mlx5 VFs.
--
Adrien Mazarguil
6WIND
and attempts to use as it is running in blacklist mode. Perhaps that
device was not configured properly.
Try to white-list the devices you want to use by explicitly providing their
PCI bus addresses through -w arguments instead.
--
Adrien Mazarguil
6WIND
llowing comment:
"the rte_flow_error returned was not very useful for it does return the
error of the last tried filter-type (L2 tunnel in ixgbe), and not the error
of the filter-type that my setup should use (flow director)"
CC'ing Wenzhuo, I think this issue is present in 17.02, has it been fixed
since?
[1] http://dpdk.org/ml/archives/dev/2017-March/059928.html
--
Adrien Mazarguil
6WIND
rpt also
>
>
> From: Lu, Wenzhuo
> Sent: Wednesday, March 8, 2017 4:16 AM
> To: Le Scouarnec Nicolas; d...@dpdk.org; Adrien Mazarguil
> (adrien.mazarg...@6wind.com)
> Cc: Yigit, Ferruh
> Subject: RE: Issues with ixgbe and rte_flow
>
> >> I have been using
ew features will still be missing such as setting interfaces up and down,
updating MTU of the physical link and so on, basically all netdevice
operations that are not queries. Those can be performed as super-user on the
related netdevice directly from a separate process, even before starting the
DPDK application.
--
Adrien Mazarguil
6WIND
x4*.so* on your system, make
sure DPDK links against OFED's version. I suggest uninstalling all previous
versions to be sure.
Check also that you do not have both /usr/include/infiniband and
/usr/local/include/infiniband.
--
Adrien Mazarguil
6WIND
3:1003 librte_pmd_mlx4
> PMD: librte_pmd_mlx4: PCI information matches, using device "mlx4_0" (VF:
> false)
> PMD: librte_pmd_mlx4: 1 port(s) detected
> PMD: librte_pmd_mlx4: ibv_exp_query_device() failed
Make sure ib_uverbs, mlx4_en, mlx4_ib and mlx4_core are all loaded and
comi
26 matches
Mail list logo