[Kernel-packages] [Bug 1872726] Comment bridged from LTC Bugzilla

2020-05-20 Thread bugproxy
--- Comment From heinz-werner_se...@de.ibm.com 2020-05-20 03:28 EDT---
Fix already verified by IBM. Works as expected.

-- 
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/1872726

Title:
  [Ubuntu 20.04] net/mlx5e: Fix endianness handling in pedit mask

Status in Ubuntu on IBM z Systems:
  Fix Committed
Status in linux package in Ubuntu:
  In Progress
Status in linux source package in Focal:
  Fix Committed
Status in linux source package in Groovy:
  In Progress

Bug description:
  SRU Justification:
  ==

  [Impact]

  * An issue with the endianess handling in the Mellanox mlx5 driver was
  found.

  * The mask value is provided as 64 bit and has to be casted in either
  32 or 16 bit.

  * On big endian systems the wrong half was casted which resulted in an
  all zero mask.

  [Fix]

  * Backport: https://launchpadlibrarian.net/476243025/0001-net-mlx5
  -fix-endianness-handling-in-pedit-mask.patch

  [Test Case]

  * An s390x system with RoCE Express 2(.1) system is needed and the
  driver loaded.

  * Check whether the mask value stays zero, or if it also get's non-
  zero values.

  * Functional testing is currently only doable by IBM, since we only
  have RoCE (1) hardware that uses the mlx4 driver.

  [Regression Potential]

  * There is regression potential is moderate, since:

  * the RoCE 2(.1) cards are pretty new and not very wide spread, yet

  * the fix got already upstream accepted with 5.6

  * However, at the end the patch modifies Mellanox common code
  (drivers/net/ethernet/mellanox/mlx5/core/en_tc.c) to make the driver
  work correctly on s390x.

  * but the changes were reviewed, signed off by Mellanox engineers and
  are very limited.

  [Other Info]

  * The above backport (patch-file) is based on commit
  404402abd5f90aa90a134eb9604b1750c1941529 404402abd5f9 "net/mlx5e: Fix
  endianness handling in pedit mask" - the backport was needed for
  getting it applied to focal master-next.

  * The commit itself got upstream accepted with kernel v5.6, hence
  should automatically land in 'gorilla', but since gorilla is still
  based on 5.4, I'm adding 'G' to this SRU.

  __

  Issue found in the Mellanox mlx5 device driver:

  The mask value is provided as 64 bit and has to be casted in
  either 32 or 16 bit. On big endian systems the wrong half was
  casted which resulted in an all zero mask.

  We need to get the upstream commit picked up for the Ubuntu 20.04
  kernel.

  
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=404402abd5f90aa90a134eb9604b1750c1941529

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu-z-systems/+bug/1872726/+subscriptions

-- 
Mailing list: https://launchpad.net/~kernel-packages
Post to : kernel-packages@lists.launchpad.net
Unsubscribe : https://launchpad.net/~kernel-packages
More help   : https://help.launchpad.net/ListHelp


[Kernel-packages] [Bug 1872726] Comment bridged from LTC Bugzilla

2020-04-23 Thread bugproxy
--- Comment From niklas.schne...@ibm.com 2020-04-23 08:56 EDT---
Sebastian has supplied me with the following information for reproducing the 
issue and is currently working on a backport:

echo 1 > /sys/class/net/p0/device/sriov_numvfs
echo 0101:00:00.0 > /sys/bus/pci/drivers/mlx5_core/unbind
devlink dev eswitch set pci/0100:00:00.0 mode switchdev
ethtool -K p0 hw-tc-offload on
echo 0101:00:00.0 > /sys/bus/pci/drivers/mlx5_core/bind
tc qdisc add dev p0 ingress
tc qdisc add dev p0v0_r ingress
tc filter add dev p0v0_r protocol ip parent :0 prio 1 flower skip_sw action 
pedit ex munge eth dst set 11:22:33:44:55:66 action goto chain 1

Without the patch the last command would fail with the following output
in dmesg

[ 548.966006] mlx5_core 0100:00:00.0: mlx5_cmd_check:711:(pid 5220):
SET_FLOW_TABLE_ENTRY(0x936) op_mod(0x0) failed, status bad
parameter(0x3), syndrome (0x144b7a)

This was only tested on Z but we suspect that the same problem should
also occur on other Big Endian architectures

-- 
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/1872726

Title:
  [Ubuntu 20.04] net/mlx5e: Fix endianness handling in pedit mask

Status in Ubuntu on IBM z Systems:
  Incomplete
Status in linux package in Ubuntu:
  New

Bug description:
  Issue found in the Mellanox mlx5 device driver:

  The mask value is provided as 64 bit and has to be casted in
  either 32 or 16 bit. On big endian systems the wrong half was
  casted which resulted in an all zero mask.

  We need to get the upstream commit picked up for the Ubuntu 20.04
  kernel.

  
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=404402abd5f90aa90a134eb9604b1750c1941529

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu-z-systems/+bug/1872726/+subscriptions

-- 
Mailing list: https://launchpad.net/~kernel-packages
Post to : kernel-packages@lists.launchpad.net
Unsubscribe : https://launchpad.net/~kernel-packages
More help   : https://help.launchpad.net/ListHelp