mbox series

[net-next,v2,0/7] net: stmmac: FPE via ethtool + tc

Message ID cover.1723548320.git.0x1207@gmail.com (mailing list archive)
Headers show
Series net: stmmac: FPE via ethtool + tc | expand

Message

Furong Xu Aug. 13, 2024, 11:47 a.m. UTC
Move the Frame Preemption(FPE) over to the new standard API which uses
ethtool-mm/tc-mqprio/tc-taprio.

Changes in v2:
  1. refactor FPE verification processe
  2. suspend/resume and kselftest-ethtool_mm, all test cases passed
  3. handle TC:TXQ remapping for DWMAC CORE4+

Furong Xu (7):
  net: stmmac: move stmmac_fpe_cfg to stmmac_priv data
  net: stmmac: drop stmmac_fpe_handshake
  net: stmmac: refactor FPE verification processe
  net: stmmac: configure FPE via ethtool-mm
  net: stmmac: support fp parameter of tc-mqprio
  net: stmmac: support fp parameter of tc-taprio
  net: stmmac: silence FPE kernel logs

 .../net/ethernet/stmicro/stmmac/dwmac4_core.c |  10 +-
 drivers/net/ethernet/stmicro/stmmac/dwmac5.c  |  96 +++++++++-
 drivers/net/ethernet/stmicro/stmmac/dwmac5.h  |  11 +-
 .../ethernet/stmicro/stmmac/dwxgmac2_core.c   |   4 +-
 drivers/net/ethernet/stmicro/stmmac/hwif.h    |  20 +-
 drivers/net/ethernet/stmicro/stmmac/stmmac.h  |  22 +++
 .../ethernet/stmicro/stmmac/stmmac_ethtool.c  | 103 +++++++++++
 .../net/ethernet/stmicro/stmmac/stmmac_main.c | 173 +++++++++---------
 .../net/ethernet/stmicro/stmmac/stmmac_tc.c   | 115 ++++++++----
 include/linux/stmmac.h                        |  28 ---
 10 files changed, 416 insertions(+), 166 deletions(-)

Comments

Vladimir Oltean Aug. 16, 2024, 11:48 a.m. UTC | #1
On Tue, Aug 13, 2024 at 07:47:26PM +0800, Furong Xu wrote:
> Move the Frame Preemption(FPE) over to the new standard API which uses
> ethtool-mm/tc-mqprio/tc-taprio.
> 
> Changes in v2:
>   1. refactor FPE verification processe
>   2. suspend/resume and kselftest-ethtool_mm, all test cases passed
>   3. handle TC:TXQ remapping for DWMAC CORE4+

This is starting to look better and better. I wouldn't be sad if it was
merged like this, but the locking still looks a little bit wacky to me,
and it's not 100% clear what priv->mm_lock protects and what it doesn't.

I can make a breakdown of how each member of fpe_cfg is used, and thus
understand exactly what are the locking needs and how they are addressed
in this version, but most likely not today.