Address review comments from v1. I missed the intel-wired-lan list on the first submission, so check out the netdev list archive for the first rev of the patch.
I think it's still up in the air on how best to register a single bus that's shared among up to four MACs. This code works for me, but there might be a better way to do this. Same caveats about testing still apply. My test setup is a Marvell Peridot switch hanging off of a Intel C3000 SoC. Clause 45 devices and other ixgbe devices have not been tested. Changes from v1 -> v2 [PATCH 1/2] ixgbe: register a mdiobus * Add intel-wired-...@lists.osuosl.org to CC list, see * select MII in Kconfig (thanks to the kbuild bot) * Only call mdiobus_regsiter for single x500em_a device * Use readx_poll_timeout() in ixgbe_msca_cmd() * Register different bus->read/write callbacks in ixgbe_mii_bus_init() so there's device id check on every access * Use device pci_name() in bus->id instead of parent bridge's name [PATCH 2/2] ixgbe: use mii_bus to handle MII related ioctls * Only use mdiobus_read/write for adapters that registered a mdiobus Stephen Douthit (2): ixgbe: register a mdiobus ixgbe: use mii_bus to handle MII related ioctls drivers/net/ethernet/intel/Kconfig | 1 + drivers/net/ethernet/intel/ixgbe/ixgbe.h | 2 + drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 23 ++ drivers/net/ethernet/intel/ixgbe/ixgbe_phy.c | 307 ++++++++++++++++++ drivers/net/ethernet/intel/ixgbe/ixgbe_phy.h | 2 + 5 files changed, 335 insertions(+) -- 2.17.2
