Thread (20 messages) 20 messages, 4 authors, 2012-03-20

Re: [PATCH 0/8] net/mlx4_en: DCB QoS support

From: Amir Vadai <hidden>
Date: 2012-03-20 11:29:33

On 03/13/2012 07:21 PM, Amir Vadai wrote:
DCBX version 802.1qaz is supported.
User Priority (UP) is set in QP context instead of in WQE (QP Work Queue
Element), which means that all traffic from a queue will have the same UP.
UP is also set for untagged traffic to be able to classify such traffic too.

Mapping from sk_prio to User Priority is done by sch_mqprio mapping. Although
confusingly sch_mqprio maps sk_prio to something called TC, it is not related
to DCBX's TC, and is interpreted by mlx4_en driver as UP.

The Current HW based QoS mechanism which was introduced in commit 4f57c087de9
"net: implement mechanism for HW based QOS" is in orientation to ETS traffic
class. Patch 7/8 introduces an approach which allow to use this mechanism also
with hardware who has queues per user priority (UP). After the change,
__skb_tx_hash() will direct a flow to a tx ring from a range of tx rings. This
range is defined by the caller function by the specific HW. If TC based queues,
the range is by TC number and for UP based queues, the range is by UP.

Amir Vadai (8):
   net/mlx4_en: Force user priority by QP attribute
   net/mlx4_core: set port QoS attributes
   net/mlx4_en: DCB QoS support
   net/mlx4_en: Set max rate-limit for a TC
   net/mlx4_en: sk_prio<=>  UP for untagged traffic
   IB/rdma_cm: TOS<=>  UP mapping for IBoE
   net: support tx_ring per UP in HW based QoS mechanism
   net/mlx4_en: num cores tx rings for every UP

  drivers/infiniband/core/cma.c                     |   35 ++++-
  drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c   |   11 +-
  drivers/net/ethernet/mellanox/mlx4/Kconfig        |   12 ++
  drivers/net/ethernet/mellanox/mlx4/Makefile       |    1 +
  drivers/net/ethernet/mellanox/mlx4/en_dcb_nl.c    |  215 +++++++++++++++++++++
  drivers/net/ethernet/mellanox/mlx4/en_main.c      |    6 +-
  drivers/net/ethernet/mellanox/mlx4/en_netdev.c    |   64 ++++++-
  drivers/net/ethernet/mellanox/mlx4/en_port.h      |    2 +
  drivers/net/ethernet/mellanox/mlx4/en_resources.c |    6 +-
  drivers/net/ethernet/mellanox/mlx4/en_rx.c        |    4 +-
  drivers/net/ethernet/mellanox/mlx4/en_sysfs.c     |  120 ++++++++++++
  drivers/net/ethernet/mellanox/mlx4/en_tx.c        |   20 +-
  drivers/net/ethernet/mellanox/mlx4/mlx4.h         |   20 ++
  drivers/net/ethernet/mellanox/mlx4/mlx4_en.h      |   38 +++-
  drivers/net/ethernet/mellanox/mlx4/port.c         |   62 ++++++
  include/linux/mlx4/cmd.h                          |    4 +
  include/linux/mlx4/device.h                       |    3 +
  include/linux/mlx4/qp.h                           |    3 +-
  include/linux/netdevice.h                         |   12 +-
  include/linux/skbuff.h                            |    3 +-
  net/core/dev.c                                    |   10 +-
  21 files changed, 615 insertions(+), 36 deletions(-)
  create mode 100644 drivers/net/ethernet/mellanox/mlx4/en_dcb_nl.c
  create mode 100644 drivers/net/ethernet/mellanox/mlx4/en_sysfs.c

Hi Dave,

Patches 7-8 who deal with the interaction between the kernel HW QoS constructs to the queue selection logic are still under discussion with John and some changes might be needed there.
At this point, we ask for patches 1-6 to be pulled in, and continue the discussion from there.

Thanks,
Amir
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help