mbox series

[net-next,0/8,pull,request] ice: use less resources in switchdev

Message ID 20240325202623.1012287-1-anthony.l.nguyen@intel.com (mailing list archive)
Headers show
Series ice: use less resources in switchdev | expand

Message

Tony Nguyen March 25, 2024, 8:26 p.m. UTC 
Michal Swiatkowski says:

Switchdev is using one queue per created port representor. This can
quickly lead to Rx queue shortage, as with subfunction support user
can create high number of PRs.

Save one MSI-X and 'number of PRs' * 1 queues.
Refactor switchdev slow-path to use less resources (even no additional
resources). Do this by removing control plane VSI and move its
functionality to PF VSI. Even with current solution PF is acting like
uplink and can't be used simultaneously for other use cases (adding
filters can break slow-path).

In short, do Tx via PF VSI and Rx packets using PF resources. Rx needs
additional code in interrupt handler to choose correct PR netdev.
Previous solution had to queue filters, it was way more elegant but
needed one queue per PRs. Beside that this refactor mostly simplifies
switchdev configuration.

The following are changes since commit 537c2e91d3549e5d6020bb0576cf9b54a845255f:
  Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
and are available in the git repository at:
  git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue 100GbE

Michal Swiatkowski (8):
  ice: remove eswitch changing queues algorithm
  ice: do Tx through PF netdev in slow-path
  ice: default Tx rule instead of to queue
  ice: control default Tx rule in lag
  ice: remove switchdev control plane VSI
  ice: change repr::id values
  ice: do switchdev slow-path Rx using PF VSI
  ice: count representor stats

 drivers/net/ethernet/intel/ice/ice.h          |   7 -
 drivers/net/ethernet/intel/ice/ice_base.c     |  44 +--
 drivers/net/ethernet/intel/ice/ice_dcb_lib.c  |   4 +-
 drivers/net/ethernet/intel/ice/ice_eswitch.c  | 362 +++---------------
 drivers/net/ethernet/intel/ice/ice_eswitch.h  |  13 +-
 drivers/net/ethernet/intel/ice/ice_lag.c      |  53 ++-
 drivers/net/ethernet/intel/ice/ice_lag.h      |   3 +-
 drivers/net/ethernet/intel/ice/ice_lib.c      |  49 +--
 drivers/net/ethernet/intel/ice/ice_main.c     |  10 +-
 drivers/net/ethernet/intel/ice/ice_repr.c     | 126 +++---
 drivers/net/ethernet/intel/ice/ice_repr.h     |  24 +-
 drivers/net/ethernet/intel/ice/ice_sriov.c    |   3 -
 drivers/net/ethernet/intel/ice/ice_switch.c   |   4 +
 drivers/net/ethernet/intel/ice/ice_switch.h   |   5 +-
 drivers/net/ethernet/intel/ice/ice_txrx.h     |   1 +
 drivers/net/ethernet/intel/ice/ice_txrx_lib.c |  11 +-
 drivers/net/ethernet/intel/ice/ice_type.h     |   1 -
 .../net/ethernet/intel/ice/ice_vsi_vlan_ops.c |   1 -
 18 files changed, 232 insertions(+), 489 deletions(-)

Comments

patchwork-bot+netdevbpf@kernel.org March 29, 2024, 6 a.m. UTC | #1 
Hello:

This series was applied to netdev/net-next.git (main)
by Tony Nguyen <anthony.l.nguyen@intel.com>:

On Mon, 25 Mar 2024 13:26:08 -0700 you wrote:
> Michal Swiatkowski says:
> 
> Switchdev is using one queue per created port representor. This can
> quickly lead to Rx queue shortage, as with subfunction support user
> can create high number of PRs.
> 
> Save one MSI-X and 'number of PRs' * 1 queues.
> Refactor switchdev slow-path to use less resources (even no additional
> resources). Do this by removing control plane VSI and move its
> functionality to PF VSI. Even with current solution PF is acting like
> uplink and can't be used simultaneously for other use cases (adding
> filters can break slow-path).
> 
> [...]

Here is the summary with links:
  - [net-next,1/8] ice: remove eswitch changing queues algorithm
    https://git.kernel.org/netdev/net-next/c/8c67b7a914cc
  - [net-next,2/8] ice: do Tx through PF netdev in slow-path
    https://git.kernel.org/netdev/net-next/c/defd52455aee
  - [net-next,3/8] ice: default Tx rule instead of to queue
    https://git.kernel.org/netdev/net-next/c/50d62022f455
  - [net-next,4/8] ice: control default Tx rule in lag
    https://git.kernel.org/netdev/net-next/c/9cba6e1767bf
  - [net-next,5/8] ice: remove switchdev control plane VSI
    https://git.kernel.org/netdev/net-next/c/33bf1e86231d
  - [net-next,6/8] ice: change repr::id values
    https://git.kernel.org/netdev/net-next/c/6235cb6e5b0d
  - [net-next,7/8] ice: do switchdev slow-path Rx using PF VSI
    https://git.kernel.org/netdev/net-next/c/44ba608db509
  - [net-next,8/8] ice: count representor stats
    https://git.kernel.org/netdev/net-next/c/4498159a5093

You are awesome, thank you!