Re: [dpdk-dev] 17.11.10 (LTS) patches review and test

[Luca Boccassi <bluca@debian.org>]

On Sun, 2020-01-26 at 14:58 +0000, Ali Alnubani wrote:
> Hi Luca,
> Sorry for the late response 😊
> 
> > -----Original Message-----
> > From: Luca Boccassi <
> > bluca@debian.org
> > >
> > Sent: Thursday, January 23, 2020 1:06 PM
> > To: Ali Alnubani <
> > alialnu@mellanox.com
> > >; dpdk stable <
> > stable@dpdk.org
> > >
> > Cc: 
> > dev@dpdk.org
> > ; Raslan Darawsheh <
> > rasland@mellanox.com
> > >; Thomas
> > Monjalon <
> > thomas@monjalon.net
> > >; Slava Ovsiienko
> > <
> > viacheslavo@mellanox.com
> > >; Matan Azrad <
> > matan@mellanox.com
> > >; Phil
> > Yang <
> > phil.yang@arm.com
> > >; David Marchand
> > <
> > david.marchand@redhat.com
> > >
> > Subject: Re: 17.11.10 (LTS) patches review and test
> > 
> > On Wed, 2020-01-22 at 16:34 +0000, Luca Boccassi wrote:
> > > On Wed, 2020-01-22 at 16:06 +0000, Ali Alnubani wrote:
> > > > Hi,
> > > > 
> > > > > -----Original Message-----
> > > > > From: Luca Boccassi <
> > > > > bluca@debian.org
> > > > > 
> > > > > 
> > > > > 
> > > > > Sent: Monday, January 13, 2020 2:47 PM
> > > > > To: dpdk stable <
> > > > > stable@dpdk.org
> > > > > 
> > > > > 
> > > > > 
> > > > > Cc:
> > > > > dev@dpdk.org
> > > > > 
> > > > > 
> > > > > ; Akhil Goyal <
> > > > > akhil.goyal@nxp.com
> > > > > 
> > > > > 
> > > > > > ; Ali Alnubani
> > > > > 
> > > > > <
> > > > > alialnu@mellanox.com
> > > > > 
> > > > > 
> > > > > > ;
> > > > > 
> > > > > benjamin.walker@intel.com
> > > > > 
> > > > > 
> > > > > ; David Christensen
> > > > > <
> > > > > drc@linux.vnet.ibm.com
> > > > > 
> > > > > 
> > > > > > ; Hemant Agrawal <
> > > > > 
> > > > > hemant.agrawal@nxp.com
> > > > > 
> > > > > 
> > > > > > ;
> > > > > 
> > > > > Ian Stokes <
> > > > > ian.stokes@intel.com
> > > > > 
> > > > > 
> > > > > > ; Jerin Jacob <
> > > > > 
> > > > > jerinj@marvell.com
> > > > > 
> > > > > 
> > > > > > ; John
> > > > > 
> > > > > McNamara <
> > > > > john.mcnamara@intel.com
> > > > > 
> > > > > 
> > > > > > ; Kevin Traynor
> > > > > 
> > > > > <
> > > > > ktraynor@redhat.com
> > > > > 
> > > > > 
> > > > > > ; Pei Zhang <
> > > > > 
> > > > > pezhang@redhat.com
> > > > > 
> > > > > 
> > > > > > ;
> > > > > 
> > > > > pingx.yu@intel.com
> > > > > 
> > > > > 
> > > > > ;
> > > > > qian.q.xu@intel.com
> > > > > 
> > > > > 
> > > > > ; Raslan Darawsheh
> > > > > <
> > > > > rasland@mellanox.com
> > > > > 
> > > > > 
> > > > > > ; Thomas Monjalon <
> > > > > 
> > > > > thomas@monjalon.net
> > > > > 
> > > > > 
> > > > > > ;
> > > > > 
> > > > > yuan.peng@intel.com
> > > > > 
> > > > > 
> > > > > ; Chen, Zhaoyan <
> > > > > zhaoyan.chen@intel.com
> > > > > 
> > > > > 
> > > > > 
> > > > > Subject: 17.11.10 (LTS) patches review and test
> > > > > 
> > > > > Hi all,
> > > > > 
> > > > > Here is a list of patches targeted for LTS release 17.11.10.
> > > > > This will be the last 17.11 release.
> > > > > 
> > > > > The planned date for the final release is the 30th of
> > > > > January.
> > > > > 
> > > > > Please help with testing and validation of your use cases and
> > > > > report any issues/results with reply-all to this mail. For
> > > > > the
> > > > > final release the fixes and reported validations will be
> > > > > added to
> > > > > the release notes.
> > > > > 
> > > > 
> > > > The following cases were tested on Mellanox devices:
> > > > - Verify sending and receiving multiple types of traffic.
> > > > - testpmd xstats counter tests.
> > > > - testpmd timestamp tests.
> > > > - Changing/checking link status through testpmd.
> > > > - RTE flow and flow_director tests.
> > > > - Some RSS tests.
> > > > - VLAN stripping and insertion tests.
> > > > - Checksum and TSO tests.
> > > > - ptype tests.
> > > > - Multi-process tests.
> > > > 
> > > > Testing matrix:
> > > > - NIC: ConnectX-4 Lx / OS: RHEL7.4 / Driver: MLNX_OFED_LINUX-
> > > > 4.7-
> > > > 3.2.9.0 / Firmware: 14.26.4012
> > > > - NIC: ConnectX-5 / OS: RHEL7.4 / Driver: MLNX_OFED_LINUX-4.7-
> > > > 3.2.9.0
> > > > / Firmware: 16.26.4012
> > > > 
> > > > Current issues:
> > > > - The message “PMD: Failed to send netlink message: Bad file
> > > > descriptor (9)” is shown at testpmd startup. Still didn’t debug
> > > > its
> > > > effect on functionality.
> > > > - ibv_icmd_create_cq () from /lib64/libibverbs.so.1 segfaults
> > > > after
> > > > setting the number of queues and restarting the ports.
> > > > - Enabling MLX4_PMD causes a build failure with OFED-4.7-
> > > > 3.2.9.0
> > > > (“error: unknown type name ‘off_t’”).
> > > > - Count action isn’t supported in rte_flow.
> > > > - Error creating rte_flow rules with vlan items.
> > > > 
> > > > 
> > > > Regards,
> > > > Ali
> > > 
> > > Thanks - are these new regressions since 17.11.9?
> 
> Only the segfault is caused by new changes. It is caused by commit:
> https://git.dpdk.org/dpdk-stable/commit/?h=17.11&id=a10a7083d164fb48cc3978067820c3d256899f3e
>  (Added Xueming Li to CC list.)
> I should've reported the rest of the issues earlier, my bad.

No problem! Let's wait and see if Xueming can help us.

> > It was mentioned in the release meeting call that they are new
> > regressions
> > since 17.11.10-rc1 - these are the commits that were backported
> > that
> > changed drivers/mlx*:
> > 
> 
> <removed>
> > They seem quite innocous to me, but I might be wrong. Authors and
> > reviewers CC'ed.
> > 
> > - there's no "off_t" reference in MLX4
> 
> This started reproducing since driver (MLNX_OFED_LINUX) version 4.6-
> 1.0.1.1. Full error:
> """
> In file included from /path/to/drivers/net/mlx4/mlx4_rxq.c:49:0:
> /usr/include/infiniband/mlx4dv.h:176:2: error: unknown type name
> ‘off_t’
>   off_t   uar_mmap_offset;
>   ^
> make[6]: *** [mlx4_rxq.o] Error 1
> make[6]: *** Waiting for unfinished jobs....
> """

If I am understanding this correctly, newer OFED versions aren't
compatible with this LTS? Is that it? Is there anything we can do about
it?

> > - the only flow change was this commit and it seems unrelated:
> > 
> 
> <removed>
> > So I'd imagine it could be some unrelated change, in the core
> > libraries, that
> > causes these issues. Would you be able to bisect the tree to see
> > exactly
> > which commits cause issues?
> 
> Regarding issues:
> - Count action isn’t supported in rte_flow.
> - Error creating rte_flow rules with vlan items.
> These are related to support in the OFED driver, and we still need to
> do more investigation.

Thank you, I will wait for an update.

-- 
Kind regards,
Luca Boccassi