[dpdk-dev] [PATCH v6 0/9] Dynamic RSS Configuration for Bonding

[Declan Doherty]

On 16/10/15 11:00, Tomasz Kulasek wrote:
> OVERVIEW
> --------
> 1) Setting .rxmode.mq_mode for bonding device to ETH_MQ_RX_RSS makes bonding
> device fully RSS-capable, so all slaves are synchronized with its configuration.
> This mode is intended to provide RSS configuration as known from "dynamic RSS
> configuration for one port" and made slaves transparent for client application
> implementation.
>
> 2) If .rxmode.mq_mode for bonding device isn't ETH_MQ_RX_RSS, slaves are not
> synchronized. That provides an ability to configure them manually. This mode may
> be useful when application wants to manage RSS in an unusual way and the
> consistency of RSS configuration for slaves isn't required.
>
> Turning on/off RSS mode for slaves when bonding is started is not possible.
> Other RSS configuration is propagated over slaves, when bonding device API is
> used to do it.
>
> v6 changes:
>   - patchset reordered
>   - fixed forward dependency between patch 4/9 and 5/9
>
> v5 changes:
>   - updated to DPDK 2.2
>   - removed copyright change from null device source
>   - removed queue_stats_mapping_set from eth_dev_ops of bonding device
>   - null pmd cleanups (removed unnecessary malloc, replaced memcpy with
>     rte_memcpy)
>   - fixed queues number configuration in null pmd
>
> v4 changes:
>   - fixed copy-paste error,
>   - removed example application as too complex and introducing a new
>     dependency,
>   - addapted null pmd to be used as testing device for dynamic RSS configuration,
>   - addapted test units to use null pmd instead of ring pmd,
>   - ring pmd is not used and changed in this patchset
>
> v3 changes:
>   - checkpatch cleanups
>
> v2 changes:
>   - added support for keys other than 40 bytes long,
>   - now, if RSS key is not set for bonding, it is not set also for slaves,
>   - fix - full initial RSS configuration before any slave is added was not
>     possible due to the initially zeroed flow_type_rss_offloads for bonding,
>   - fix - changed error to warning when slave is synchronizing due to the
>     bonding's initial configuration (to allow use slaves' drivers not supporting
>     dynamic RSS configuration in bonding),
>   - some code cleanups,
>   - updated documentation,
>
> Tomasz Kulasek (9):
>    bonding: rss dynamic configuration
>    null: fix segfault when null_pmd added to bonding
>    null: extend number of virtual queues
>    null: export eth_dev_null_create
>    null: virtual dynamic rss configuration
>    test: dynamic rss configuration
>    bonding: per queue stats
>    doc: fixed spellings and typos
>    doc: dynamic rss configuration for bonding
>
>   app/test/Makefile                                  |    8 +
>   app/test/test_link_bonding_rssconf.c               |  679 ++++++++++++++++++++
>   .../prog_guide/link_bonding_poll_mode_drv_lib.rst  |   42 +-
>   drivers/net/bonding/rte_eth_bond_api.c             |   28 +
>   drivers/net/bonding/rte_eth_bond_pmd.c             |  216 ++++++-
>   drivers/net/bonding/rte_eth_bond_private.h         |   12 +
>   drivers/net/null/Makefile                          |    2 +-
>   drivers/net/null/rte_eth_null.c                    |  149 ++++-
>   drivers/net/null/rte_eth_null.h                    |   40 ++
>   drivers/net/null/rte_pmd_null_version.map          |    7 +
>   10 files changed, 1151 insertions(+), 32 deletions(-)
>   create mode 100644 app/test/test_link_bonding_rssconf.c
>   create mode 100644 drivers/net/null/rte_eth_null.h
>

Series Acked-by: Declan Doherty <declan.doherty at intel.com>