[PATCH 00/46] Support AMD Solarflare X45xx adaptors

[Stephen Hemminger]

On Wed, 16 Apr 2025 21:37:09 +0400 (+04)
Ivan Malov <ivan.malov at arknetworks.am> wrote:

> On Wed, 16 Apr 2025, Stephen Hemminger wrote:
> 
> > On Wed, 16 Apr 2025 19:38:58 +0400 (+04)
> > Ivan Malov <ivan.malov at arknetworks.am> wrote:
> >  
> >> On Wed, 16 Apr 2025, Stephen Hemminger wrote:
> >>  
> >>> On Wed, 16 Apr 2025 17:59:30 +0400
> >>> Ivan Malov <ivan.malov at arknetworks.am> wrote:
> >>>  
> >>>> New X4522 (dual port SFP56) and X4542 (dual port QSFP56) adaptors are
> >>>> Medford4 (X4) chips that are based on EF10 architecture. An X4 NIC
> >>>> supports multiple network engine types. This series provides support
> >>>> only for the Medford2-alike, 'full-feature' (FF) network engine. This
> >>>> shall not be confused with the concept of 'datapath FW variants': the
> >>>> FF network engine supports both 'full-feature' and 'ultra-low-latency'
> >>>> datapath FW variants, with corresponding Medford2-alike feature sets.
> >>>>
> >>>> The first part of the series provides general support for the adaptors,
> >>>> whilst the second one adds support for the new management controller
> >>>> interface for configuration of network port features (netport MCDI).
> >>>>
> >>>> For now, only support for physical functions (PFs) is concerned. There
> >>>> is a small number of TODO and FIXME markings in the code. Those are
> >>>> normal at this development stage and will be removed by future patches
> >>>> when VF support has fleshed out.
> >>>>
> >>>>
> >>>> Andy Moreton (3):
> >>>>   common/sfc_efx/base: update X4 BAR layout and PCI IDs
> >>>>   net/sfc: add Medford4 with only full feature datapath engine
> >>>>   common/sfc_efx/base: add port mode for 8 port hardware
> >>>>
> >>>> Denis Pryazhennikov (15):
> >>>>   common/sfc_efx/base: add Medford4 PCI IDs to common code
> >>>>   common/sfc_efx/base: add efsys option for Medford4
> >>>>   common/sfc_efx/base: add Medford4 support to NIC module
> >>>>   common/sfc_efx/base: add Medford4 support to EV module
> >>>>   common/sfc_efx/base: add Medford4 support to FILTER module
> >>>>   common/sfc_efx/base: add Medford4 support to INTR module
> >>>>   common/sfc_efx/base: add Medford4 support to MAC module
> >>>>   common/sfc_efx/base: add Medford4 support to PHY module
> >>>>   common/sfc_efx/base: add Medford4 support to TUNNEL module
> >>>>   common/sfc_efx/base: add Medford4 support to MCDI module
> >>>>   common/sfc_efx/base: add Medford4 support to Rx module
> >>>>   common/sfc_efx/base: add Medford4 support to Tx module
> >>>>   drivers: enable support for AMD Solarflare X4 adapter family
> >>>>   common/sfc_efx/base: add new X4 port mode
> >>>>   common/sfc_efx/base: extend list of supported X4 port modes
> >>>>
> >>>> Ivan Malov (28):
> >>>>   common/sfc_efx/base: update MCDI headers
> >>>>   common/sfc_efx/base: provide a stub for basic netport attach
> >>>>   common/sfc_efx/base: provide defaults on netport attach path
> >>>>   common/sfc_efx/base: obtain assigned netport handle from NIC
> >>>>   common/sfc_efx/base: allow for const in MCDI struct accessor
> >>>>   common/sfc_efx/base: get netport fixed capabilities on probe
> >>>>   common/sfc_efx/base: decode netport link state on probe path
> >>>>   common/sfc_efx/base: fill in loopback modes on netport probe
> >>>>   common/sfc_efx/base: introduce Medford4 stub for PHY methods
> >>>>   common/sfc_efx/base: refactor EF10 link mode decoding helper
> >>>>   common/sfc_efx/base: provide PHY link get method on Medford4
> >>>>   common/sfc_efx/base: implement PHY link control for Medford4
> >>>>   common/sfc_efx/base: introduce Medford4 stub for MAC methods
> >>>>   common/sfc_efx/base: add MAC reconfigure method for Medford4
> >>>>   common/sfc_efx/base: fill in software LUT for MAC statistics
> >>>>   common/sfc_efx/base: fill in MAC statistics mask on Medford4
> >>>>   common/sfc_efx/base: support MAC statistics on Medford4 NICs
> >>>>   common/sfc_efx/base: implement MAC PDU controls for Medford4
> >>>>   common/sfc_efx/base: correct MAC PDU calculation on Medford4
> >>>>   net/sfc: make use of generic EFX MAC PDU calculation helpers
> >>>>   common/sfc_efx/base: ignore legacy link events on new boards
> >>>>   common/sfc_efx/base: add link event processing on new boards
> >>>>   net/sfc: query link status on link change events on new NICs
> >>>>   common/sfc_efx/base: subscribe to netport link change events
> >>>>   net/sfc: offer support for 200G link ability on new adaptors
> >>>>   common/sfc_efx/base: support controls for netport lane count
> >>>>   net/sfc: add support for control of physical port lane count
> >>>>   doc: advertise support for AMD Solarflare X45xx adapters
> >>>>
> >>>>  .mailmap                                      |    3 +-
> >>>>  doc/guides/nics/sfc_efx.rst                   |    9 +-
> >>>>  doc/guides/rel_notes/release_25_07.rst        |    4 +
> >>>>  drivers/common/sfc_efx/base/ef10_ev.c         |   39 +
> >>>>  drivers/common/sfc_efx/base/ef10_impl.h       |   19 +
> >>>>  drivers/common/sfc_efx/base/ef10_nic.c        |   98 +-
> >>>>  drivers/common/sfc_efx/base/ef10_phy.c        |   43 +-
> >>>>  drivers/common/sfc_efx/base/ef10_tlv_layout.h |    9 +-
> >>>>  drivers/common/sfc_efx/base/efx.h             |   98 +-
> >>>>  drivers/common/sfc_efx/base/efx_check.h       |   24 +-
> >>>>  drivers/common/sfc_efx/base/efx_ev.c          |    6 +
> >>>>  drivers/common/sfc_efx/base/efx_filter.c      |    6 +
> >>>>  drivers/common/sfc_efx/base/efx_impl.h        |  115 +-
> >>>>  drivers/common/sfc_efx/base/efx_intr.c        |    6 +
> >>>>  drivers/common/sfc_efx/base/efx_mac.c         |   56 +-
> >>>>  drivers/common/sfc_efx/base/efx_mcdi.c        |   18 +-
> >>>>  drivers/common/sfc_efx/base/efx_mcdi.h        |    2 +-
> >>>>  drivers/common/sfc_efx/base/efx_nic.c         |   60 +
> >>>>  drivers/common/sfc_efx/base/efx_np.c          | 1625 +++++
> >>>>  drivers/common/sfc_efx/base/efx_phy.c         |   88 +-
> >>>>  drivers/common/sfc_efx/base/efx_port.c        |    1 +
> >>>>  drivers/common/sfc_efx/base/efx_regs_mcdi.h   | 5868 ++++++++++++++++-
> >>>>  drivers/common/sfc_efx/base/efx_rx.c          |    6 +
> >>>>  drivers/common/sfc_efx/base/efx_tunnel.c      |   18 +-
> >>>>  drivers/common/sfc_efx/base/efx_tx.c          |   33 +
> >>>>  drivers/common/sfc_efx/base/medford4_impl.h   |  110 +
> >>>>  drivers/common/sfc_efx/base/medford4_mac.c    |  299 +
> >>>>  drivers/common/sfc_efx/base/medford4_phy.c    |  156 +
> >>>>  drivers/common/sfc_efx/base/meson.build       |    3 +
> >>>>  drivers/common/sfc_efx/efsys.h                |    2 +
> >>>>  drivers/common/sfc_efx/sfc_base_symbols.c     |    2 +
> >>>>  drivers/net/sfc/sfc.c                         |    5 +-
> >>>>  drivers/net/sfc/sfc.h                         |    4 +
> >>>>  drivers/net/sfc/sfc_dp_tx.h                   |    3 +
> >>>>  drivers/net/sfc/sfc_ef10_tx.c                 |   13 +-
> >>>>  drivers/net/sfc/sfc_ethdev.c                  |  186 +-
> >>>>  drivers/net/sfc/sfc_ev.c                      |   51 +-
> >>>>  drivers/net/sfc/sfc_port.c                    |   27 +-
> >>>>  drivers/net/sfc/sfc_repr.c                    |    7 +-
> >>>>  drivers/net/sfc/sfc_repr.h                    |    1 +
> >>>>  drivers/net/sfc/sfc_tx.c                      |    2 +
> >>>>  41 files changed, 9000 insertions(+), 125 deletions(-)
> >>>>  create mode 100644 drivers/common/sfc_efx/base/efx_np.c
> >>>>  create mode 100644 drivers/common/sfc_efx/base/medford4_impl.h
> >>>>  create mode 100644 drivers/common/sfc_efx/base/medford4_mac.c
> >>>>  create mode 100644 drivers/common/sfc_efx/base/medford4_phy.c
> >>>>  
> >>>
> >>> Overall looks good but:
> >>>  - need to fix build on FreeBSD, you are using an errno not available there.
> >>>  - can you address some of the checkpatch warnings in the base code.
> >>>  
> >>
> >> I can fix the errno, yes. Thanks for catching this.
> >>
> >> As for the checkpatch warnings in the base code -- unfortunately, that part does
> >> not follow DPDK style in general and has always triggered such warnings in DPDK.
> >> Should I try to fix some anyway? Which, for instance?  
> >
> > Mostly it is the use of BSD style parens on returns that clutters the warnings.
> > I.e
> > 	return (0);  
> 
> I see. Unfortunately, this, as well as space character in 'sizeof ()', follows
> the base driver's own style and it has been so for a long time. May be it is
> acceptable to retain it. But I do care to comply with DPDK conventions in the
> SFC driver itself. So shall I fix ENODATA->ENODEV and send v2?

As long as isn't too weird, base code can follow a slightly different set
of rules. Please send v2 and do a test build on FreeBSD (in a vm is fine).