mbox series

[net-next,v4,0/4] net: lan966x: Add support for FDMA

Message ID 20220408070357.559899-1-horatiu.vultur@microchip.com (mailing list archive)
Headers show
Series net: lan966x: Add support for FDMA | expand

Message

Horatiu Vultur April 8, 2022, 7:03 a.m. UTC
Currently when injecting or extracting a frame from CPU, the frame
is given to the HW each word at a time. There is another way to
inject/extract frames from CPU using FDMA(Frame Direct Memory Access).
In this way the entire frame is given to the HW. This improves both
RX and TX bitrate.

Tested-by: Michael Walle <michael@walle.cc> # on kontron-kswitch-d10

v3->v4:
- fix race conditions when changing the MTU
- build skbs once frames are received and not before
- use NAPI_POLL_WEIGHT instead of FDMA_WEIGHT

v2->v3:
- move skb_tx_timestamp before the IFH is inserted, because in case of PHY
  timestamping, the PHY should not see the IFH.
- use lower/upper_32_bits()
- reimplement the RX path in case of memory pressure.
- use devm_request_irq instead of devm_request_threaded_irq
- add various checks for return values.

v1->v2:
- fix typo in commit message in last patch
- remove first patch as the changes are already there
- make sure that there is space in skb to put the FCS
- move skb_tx_timestamp closer to the handover of the frame to the HW

Horatiu Vultur (4):
  net: lan966x: Add registers that are used for FDMA.
  net: lan966x: Expose functions that are needed by FDMA
  net: lan966x: Add FDMA functionality
  net: lan966x: Update FDMA to change MTU.

 .../net/ethernet/microchip/lan966x/Makefile   |   2 +-
 .../ethernet/microchip/lan966x/lan966x_fdma.c | 842 ++++++++++++++++++
 .../ethernet/microchip/lan966x/lan966x_main.c |  59 +-
 .../ethernet/microchip/lan966x/lan966x_main.h | 117 +++
 .../ethernet/microchip/lan966x/lan966x_port.c |   3 +
 .../ethernet/microchip/lan966x/lan966x_regs.h | 106 +++
 6 files changed, 1117 insertions(+), 12 deletions(-)
 create mode 100644 drivers/net/ethernet/microchip/lan966x/lan966x_fdma.c

Comments

patchwork-bot+netdevbpf@kernel.org April 12, 2022, 4 a.m. UTC | #1
Hello:

This series was applied to netdev/net-next.git (master)
by Jakub Kicinski <kuba@kernel.org>:

On Fri, 8 Apr 2022 09:03:53 +0200 you wrote:
> Currently when injecting or extracting a frame from CPU, the frame
> is given to the HW each word at a time. There is another way to
> inject/extract frames from CPU using FDMA(Frame Direct Memory Access).
> In this way the entire frame is given to the HW. This improves both
> RX and TX bitrate.
> 
> Tested-by: Michael Walle <michael@walle.cc> # on kontron-kswitch-d10
> 
> [...]

Here is the summary with links:
  - [net-next,v4,1/4] net: lan966x: Add registers that are used for FDMA.
    https://git.kernel.org/netdev/net-next/c/fdb2981c00bb
  - [net-next,v4,2/4] net: lan966x: Expose functions that are needed by FDMA
    https://git.kernel.org/netdev/net-next/c/8f2c7d9ad778
  - [net-next,v4,3/4] net: lan966x: Add FDMA functionality
    https://git.kernel.org/netdev/net-next/c/c8349639324a
  - [net-next,v4,4/4] net: lan966x: Update FDMA to change MTU.
    https://git.kernel.org/netdev/net-next/c/2ea1cbac267e

You are awesome, thank you!