On 1/3/19 4:35 PM, Jeff Kirsher wrote:
> The new ability added to the driver to use mii_bus to handle MII related
> ioctls is causing compile issues when the driver is compiled into the
> kernel (i.e. not a module).
>
> The simple solution of requiring the driver to be compiled as a module when
> MDIO_BUS is a module, causes a recursion Kconfig issue due to IPSec
> dependencies.
>
> So created another Kconfig option for ixgbe, to enable mdio_bus support for
> DSA devices. This solution solves the problem when the ixgbe driver is
> compiled into the kernel and MDIO_BUS is compiled as a module. In this
> case, the IXGBE_MDIO option is disabled and the code is not compiled
> into the driver.
>
> CC: Dave Jones <da...@codemonkey.org.uk>
> CC: Steve Douthit <steph...@silicom-usa.com>
> CC: Florian Fainelli <f.faine...@gmail.com>
> Signed-off-by: Jeff Kirsher <jeffrey.t.kirs...@intel.com>
> ---
> drivers/net/ethernet/intel/Kconfig | 11 ++++++++++-
> drivers/net/ethernet/intel/ixgbe/ixgbe.h | 4 ++++
> drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 12 ++++++++++--
> drivers/net/ethernet/intel/ixgbe/ixgbe_phy.c | 4 ++++
> drivers/net/ethernet/intel/ixgbe/ixgbe_phy.h | 3 ++-
> 5 files changed, 30 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/net/ethernet/intel/Kconfig
> b/drivers/net/ethernet/intel/Kconfig
> index 31fb76ee9d82..35317ecdd0c3 100644
> --- a/drivers/net/ethernet/intel/Kconfig
> +++ b/drivers/net/ethernet/intel/Kconfig
> @@ -159,7 +159,6 @@ config IXGBE
> tristate "Intel(R) 10GbE PCI Express adapters support"
> depends on PCI
> select MDIO
> - select MDIO_DEVICE
> imply PTP_1588_CLOCK
> ---help---
> This driver supports Intel(R) 10GbE PCI Express family of
> @@ -210,6 +209,16 @@ config IXGBE_IPSEC
> ---help---
> Enable support for IPSec offload in ixgbe.ko
>
> +config IXGBE_MDIO
> + bool "MDIO Support for DSA devices"
> + default n
> + depends on IXGBE && MDIO_BUS && !(IXGBE=y && MDIO_BUS=m)
> + ---help---
> + Say Y here if you want MDIO_BUS support for DSA devices in the
> + driver.
This is also useful for accessing the entire clause 22/45 address space,
which may be worth a mention in the help text.
> + If unsure, say N.
> +
> config IXGBEVF
> tristate "Intel(R) 10GbE PCI Express Virtual Function Ethernet support"
> depends on PCI_MSI
> diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe.h
> b/drivers/net/ethernet/intel/ixgbe/ixgbe.h
> index 08d85e336bd4..9d7496508ee0 100644
> --- a/drivers/net/ethernet/intel/ixgbe/ixgbe.h
> +++ b/drivers/net/ethernet/intel/ixgbe/ixgbe.h
> @@ -12,7 +12,9 @@
> #include <linux/aer.h>
> #include <linux/if_vlan.h>
> #include <linux/jiffies.h>
> +#ifdef CONFIG_IXGBE_MDIO
> #include <linux/phy.h>
> +#endif
>
> #include <linux/timecounter.h>
> #include <linux/net_tstamp.h>
> @@ -562,7 +564,9 @@ struct ixgbe_adapter {
> struct net_device *netdev;
> struct bpf_prog *xdp_prog;
> struct pci_dev *pdev;
> +#ifdef CONFIG_IXGBE_MDIO
> struct mii_bus *mii_bus;
> +#endif
>
> unsigned long state;
>
> diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> index d2ce7f0bc32d..afa0337f7ba0 100644
> --- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> +++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> @@ -39,7 +39,9 @@
> #include "ixgbe.h"
> #include "ixgbe_common.h"
> #include "ixgbe_dcb_82599.h"
> +#ifdef CONFIG_IXGBE_MDIO
> #include "ixgbe_phy.h"
> +#endif
> #include "ixgbe_sriov.h"
> #include "ixgbe_model.h"
> #include "ixgbe_txrx_common.h"
> @@ -8791,6 +8793,7 @@ ixgbe_mdio_read(struct net_device *netdev, int prtad,
> int devad, u16 addr)
> u16 value;
> int rc;
>
> +#ifdef CONFIG_IXGBE_MDIO
> if (adapter->mii_bus) {
> int regnum = addr;
>
> @@ -8799,7 +8802,7 @@ ixgbe_mdio_read(struct net_device *netdev, int prtad,
> int devad, u16 addr)
>
> return mdiobus_read(adapter->mii_bus, prtad, regnum);
> }
> -
> +#endif
> if (prtad != hw->phy.mdio.prtad)
> return -EINVAL;
> rc = hw->phy.ops.read_reg(hw, addr, devad, &value);
> @@ -8814,6 +8817,7 @@ static int ixgbe_mdio_write(struct net_device *netdev,
> int prtad, int devad,
> struct ixgbe_adapter *adapter = netdev_priv(netdev);
> struct ixgbe_hw *hw = &adapter->hw;
>
> +#ifdef CONFIG_IXGBE_MDIO
> if (adapter->mii_bus) {
> int regnum = addr;
>
> @@ -8822,7 +8826,7 @@ static int ixgbe_mdio_write(struct net_device *netdev,
> int prtad, int devad,
>
> return mdiobus_write(adapter->mii_bus, prtad, regnum, value);
> }
> -
> +#endif
> if (prtad != hw->phy.mdio.prtad)
> return -EINVAL;
> return hw->phy.ops.write_reg(hw, addr, devad, value);
> @@ -11141,7 +11145,9 @@ static int ixgbe_probe(struct pci_dev *pdev, const
> struct pci_device_id *ent)
> IXGBE_LINK_SPEED_10GB_FULL | IXGBE_LINK_SPEED_1GB_FULL,
> true);
>
> +#ifdef CONFIG_IXGBE_MDIO
> ixgbe_mii_bus_init(hw);
> +#endif
>
> return 0;
>
> @@ -11193,8 +11199,10 @@ static void ixgbe_remove(struct pci_dev *pdev)
> set_bit(__IXGBE_REMOVING, &adapter->state);
> cancel_work_sync(&adapter->service_task);
>
> +#ifdef CONFIG_IXGBE_MDIO
> if (adapter->mii_bus)
> mdiobus_unregister(adapter->mii_bus);
> +#endif
>
> #ifdef CONFIG_IXGBE_DCA
> if (adapter->flags & IXGBE_FLAG_DCA_ENABLED) {
> diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_phy.c
> b/drivers/net/ethernet/intel/ixgbe/ixgbe_phy.c
> index cc4907f9ff02..05da21920863 100644
> --- a/drivers/net/ethernet/intel/ixgbe/ixgbe_phy.c
> +++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_phy.c
> @@ -3,7 +3,9 @@
>
> #include <linux/pci.h>
> #include <linux/delay.h>
> +#ifdef CONFIG_IXGBE_MDIO
> #include <linux/iopoll.h>
> +#endif
> #include <linux/sched.h>
>
> #include "ixgbe.h"
> @@ -659,6 +661,7 @@ s32 ixgbe_write_phy_reg_generic(struct ixgbe_hw *hw, u32
> reg_addr,
> return status;
> }
>
> +#ifdef CONFIG_IXGBE_MDIO
> #define IXGBE_HW_READ_REG(addr) IXGBE_READ_REG(hw, addr)
>
> /**
> @@ -956,6 +959,7 @@ s32 ixgbe_mii_bus_init(struct ixgbe_hw *hw)
> adapter->mii_bus = NULL;
> return -ENODEV;
> }
> +#endif
>
> /**
> * ixgbe_setup_phy_link_generic - Set and restart autoneg
> diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_phy.h
> b/drivers/net/ethernet/intel/ixgbe/ixgbe_phy.h
> index 214b01085718..88b851178d7e 100644
> --- a/drivers/net/ethernet/intel/ixgbe/ixgbe_phy.h
> +++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_phy.h
> @@ -120,8 +120,9 @@
> /* SFP+ SFF-8472 Compliance code */
> #define IXGBE_SFF_SFF_8472_UNSUP 0x00
>
> +#ifdef CONFIG_IXGBE_MDIO
> s32 ixgbe_mii_bus_init(struct ixgbe_hw *hw);
> -
> +#endif
> s32 ixgbe_identify_phy_generic(struct ixgbe_hw *hw);
> s32 ixgbe_reset_phy_generic(struct ixgbe_hw *hw);
> s32 ixgbe_read_phy_reg_generic(struct ixgbe_hw *hw, u32 reg_addr,
>
Just tested this against a DSA dev in hardware and it works, but I
still think the simpler solution would be to just select PHYLIB:
diff --git a/drivers/net/ethernet/intel/Kconfig
b/drivers/net/ethernet/intel/Kconfig
index 31fb76ee9d82..a1246e89aad4 100644
--- a/drivers/net/ethernet/intel/Kconfig
+++ b/drivers/net/ethernet/intel/Kconfig
@@ -159,7 +159,7 @@ config IXGBE
tristate "Intel(R) 10GbE PCI Express adapters support"
depends on PCI
select MDIO
- select MDIO_DEVICE
+ select PHYLIB
imply PTP_1588_CLOCK
---help---
This driver supports Intel(R) 10GbE PCI Express family of
That seems to be the pattern for most of the other ethernet adapters:
$ grep -r 'select PHYLIB' drivers/net/ethernet/ | wc -l
71
vs checking MDIO_BUS (not counting this patch):
$ grep -r 'depends on.*MDIO_BUS' drivers/net/ethernet/ | wc -l
0
AFAICT in the Kconfig language a || between two dependency expressions
returns the max() of the two expr where (n, m, y) = (0, 1, 2). So in
the case of multiple drivers selecting the same symbol if any are
built-in then the selected symbol must also be built-in. This has held
true for the little bit of poking around I've done via menuconfig, but
I'm definitely not a Kconfig expert.
In any case you can have my:
Tested-by: Stephen Douthit <steph...@silicom-usa.com>
