diff mbox series

[v2,net-next,2/2] net: stmmac: PCI driver for BCM8958X SoC

Message ID 20240511015924.41457-1-jitendra.vegiraju@broadcom.com (mailing list archive)
State Not Applicable
Delegated to: Netdev Maintainers
Headers show
Series None | expand

Commit Message

Jitendra Vegiraju May 11, 2024, 1:59 a.m. UTC
Broadcom BCM8958X SoCs use Synopsys XGMAC design, which is similar to
dwxgmac2 core implementation in stmmac driver. The existing dwxgmac2 dma
operation functions have some conflicting differences with BCM8958X.
This glue driver attempts to reuse dwxgmac2 implementation wherever
possible, adding alternative implementations where necessary.

v2: code cleanup to address patchwork reports.

Signed-off-by: Jitendra Vegiraju <jitendra.vegiraju@broadcom.com>
---
 MAINTAINERS                                   |   7 +
 drivers/net/ethernet/stmicro/stmmac/Kconfig   |  11 +
 drivers/net/ethernet/stmicro/stmmac/Makefile  |   1 +
 .../net/ethernet/stmicro/stmmac/dwmac-brcm.c  | 638 ++++++++++++++++++
 4 files changed, 657 insertions(+)
 create mode 100644 drivers/net/ethernet/stmicro/stmmac/dwmac-brcm.c

Comments

Andrew Lunn May 11, 2024, 4:16 p.m. UTC | #1
> +	/* This device interface is directly attached to the switch chip on
> +	 *  the SoC. Since no MDIO is present, register fixed_phy.
> +	 */
> +	brcm_priv->phy_dev =
> +		 fixed_phy_register(PHY_POLL,
> +				    &dwxgmac_brcm_fixed_phy_status, NULL);
> +	if (IS_ERR(brcm_priv->phy_dev)) {
> +		dev_err(&pdev->dev, "%s\tNo PHY/fixed_PHY found\n", __func__);
> +		return -ENODEV;
> +	}
> +	phy_attached_info(brcm_priv->phy_dev);

What switch is it? Will there be patches to extend SF2?

	Andrew
Russell King (Oracle) May 11, 2024, 5:12 p.m. UTC | #2
On Sat, May 11, 2024 at 06:16:52PM +0200, Andrew Lunn wrote:
> > +	/* This device interface is directly attached to the switch chip on
> > +	 *  the SoC. Since no MDIO is present, register fixed_phy.
> > +	 */
> > +	brcm_priv->phy_dev =
> > +		 fixed_phy_register(PHY_POLL,
> > +				    &dwxgmac_brcm_fixed_phy_status, NULL);
> > +	if (IS_ERR(brcm_priv->phy_dev)) {
> > +		dev_err(&pdev->dev, "%s\tNo PHY/fixed_PHY found\n", __func__);
> > +		return -ENODEV;
> > +	}
> > +	phy_attached_info(brcm_priv->phy_dev);
> 
> What switch is it? Will there be patches to extend SF2?

... and why is this legacy fixed_phy even necessary when stmmac uses
phylink which supports fixed links, including with custom fixed status?
Andrew Lunn May 11, 2024, 5:19 p.m. UTC | #3
On Sat, May 11, 2024 at 06:12:38PM +0100, Russell King (Oracle) wrote:
> On Sat, May 11, 2024 at 06:16:52PM +0200, Andrew Lunn wrote:
> > > +	/* This device interface is directly attached to the switch chip on
> > > +	 *  the SoC. Since no MDIO is present, register fixed_phy.
> > > +	 */
> > > +	brcm_priv->phy_dev =
> > > +		 fixed_phy_register(PHY_POLL,
> > > +				    &dwxgmac_brcm_fixed_phy_status, NULL);
> > > +	if (IS_ERR(brcm_priv->phy_dev)) {
> > > +		dev_err(&pdev->dev, "%s\tNo PHY/fixed_PHY found\n", __func__);
> > > +		return -ENODEV;
> > > +	}
> > > +	phy_attached_info(brcm_priv->phy_dev);
> > 
> > What switch is it? Will there be patches to extend SF2?
> 
> ... and why is this legacy fixed_phy even necessary when stmmac uses
> phylink which supports fixed links, including with custom fixed status?

It might be because it is a PCI device, and they are trying to avoid
DT? Maybe because they have not figured out how to add DT properties
to a PCI device. It is possible.

	Andrew
Andrew Lunn May 11, 2024, 5:50 p.m. UTC | #4
On Sat, May 11, 2024 at 06:12:38PM +0100, Russell King (Oracle) wrote:
> On Sat, May 11, 2024 at 06:16:52PM +0200, Andrew Lunn wrote:
> > > +	/* This device interface is directly attached to the switch chip on
> > > +	 *  the SoC. Since no MDIO is present, register fixed_phy.
> > > +	 */
> > > +	brcm_priv->phy_dev =
> > > +		 fixed_phy_register(PHY_POLL,
> > > +				    &dwxgmac_brcm_fixed_phy_status, NULL);
> > > +	if (IS_ERR(brcm_priv->phy_dev)) {
> > > +		dev_err(&pdev->dev, "%s\tNo PHY/fixed_PHY found\n", __func__);
> > > +		return -ENODEV;
> > > +	}
> > > +	phy_attached_info(brcm_priv->phy_dev);
> > 
> > What switch is it? Will there be patches to extend SF2?
> 
> ... and why is this legacy fixed_phy even necessary when stmmac uses
> phylink which supports fixed links, including with custom fixed status?

And now you mentions legacy Fixed link:

+MODULE_DESCRIPTION("Broadcom 10G Automotive Ethernet PCIe driver");

This claims it is a 10G device. You cannot represent 10G using legacy
fixed link.

Does this MAC directly connect to the switch within the SoC? There is
no external MII interface? Realtek have been posting a MAC driver for
something similar were the MAC is directly connected to the switch
within the SoC. The MAC is fixed at 5G, there is no phylink/phylib
support, set_link_ksetting return -EOPNOTSUPP and get_link_ksettings
returns hard coded 5G.

We need a better understanding of the architecture here, before we can
advise the correct way to do this.

      Andrew
Russell King (Oracle) May 11, 2024, 7:34 p.m. UTC | #5
Hi,

Thanks for the patch,. but there are things that need some improvement.

On Fri, May 10, 2024 at 06:59:24PM -0700, Jitendra Vegiraju wrote:
> +static void dwxgmac_brcm_dma_init_tx_chan(struct stmmac_priv *priv,
> +					  void __iomem *ioaddr,
> +					  struct stmmac_dma_cfg *dma_cfg,
> +					  dma_addr_t phy, u32 chan)
> +{
> +	u32 value;
> +
> +	value = readl(ioaddr + XGMAC_DMA_CH_TX_CONTROL(chan));
> +	value &= ~XGMAC_TxPBL;
> +	value &= ~GENMASK(6, 4);
> +	writel(value, ioaddr + XGMAC_DMA_CH_TX_CONTROL(chan));
> +
> +	writel(upper_32_bits(phy), ioaddr + XGMAC_DMA_CH_TxDESC_HADDR(chan));
> +	writel(lower_32_bits(phy), ioaddr + XGMAC_DMA_CH_TxDESC_LADDR(chan));

Please use "dma_addr" not "phy" here. "phy" could mean ethernet phy.
I personally dislike "physical address" for DMA stuff because if
there's an IOMMU or other translation layer present, what you have
here is *not* a physical address.

> +static void dwxgmac_brcm_dma_init_rx_chan(struct stmmac_priv *priv,
> +					  void __iomem *ioaddr,
> +					  struct stmmac_dma_cfg *dma_cfg,
> +					  dma_addr_t phy, u32 chan)
> +{
> +	u32 value;
> +
> +	value = readl(ioaddr + XGMAC_DMA_CH_RX_CONTROL(chan));
> +	value &= ~XGMAC_RxPBL;
> +	writel(value, ioaddr + XGMAC_DMA_CH_RX_CONTROL(chan));
> +
> +	writel(upper_32_bits(phy), ioaddr + XGMAC_DMA_CH_RxDESC_HADDR(chan));
> +	writel(lower_32_bits(phy), ioaddr + XGMAC_DMA_CH_RxDESC_LADDR(chan));

Ditto.

...

> +static void dwxgmac_brcm_fix_speed(void *priv, unsigned int speed,
> +				   unsigned int mode)
> +{
> +}

If this is empty, do you really need it? The method is optional.

...

> +static int dwxgmac_brcm_pci_probe(struct pci_dev *pdev,
> +				  const struct pci_device_id *id)
> +{
...
> +	/* This device interface is directly attached to the switch chip on
> +	 *  the SoC. Since no MDIO is present, register fixed_phy.
> +	 */
> +	brcm_priv->phy_dev =
> +		 fixed_phy_register(PHY_POLL,
> +				    &dwxgmac_brcm_fixed_phy_status, NULL);
> +	if (IS_ERR(brcm_priv->phy_dev)) {
> +		dev_err(&pdev->dev, "%s\tNo PHY/fixed_PHY found\n", __func__);
> +		return -ENODEV;
> +	}
> +	phy_attached_info(brcm_priv->phy_dev);

As pointed out in the other sub-thread, you don't need this. If you need
a fixed-link and you don't have a firmware description of it, you can
provide a swnode based description through plat->port_node that will be
passed to phylink. Through that, you can tell phylink to create a
fixed link.

> +	ret = stmmac_dvr_probe(&pdev->dev, plat, &res);
> +	if (ret)
> +		goto err_disable_msi;
> +
> +	/* The stmmac core driver doesn't have the infrastructure to
> +	 * support fixed-phy mdio bus for non-platform bus drivers.
> +	 * Until a better solution is implemented, initialize the
> +	 * following entries after priv structure is populated.
> +	 */
> +	ndev = dev_get_drvdata(&pdev->dev);
> +	priv = netdev_priv(ndev);
> +	priv->mii = mdio_find_bus("fixed-0");
> +
> +	ndev->hw_features &= ~NETIF_F_HW_VLAN_CTAG_RX;
> +	priv->hw->hw_vlan_en = false;

Basically... no. Do not do any setup after stmmac_dvr_probe(), because
the network device has already been registered and published to
userspace, and userspace may have already opened the network device.
Russell King (Oracle) May 11, 2024, 7:35 p.m. UTC | #6
On Sat, May 11, 2024 at 07:19:07PM +0200, Andrew Lunn wrote:
> It might be because it is a PCI device, and they are trying to avoid
> DT? Maybe because they have not figured out how to add DT properties
> to a PCI device. It is possible.

swnodes.
Russell King (Oracle) May 11, 2024, 7:36 p.m. UTC | #7
On Sat, May 11, 2024 at 07:50:03PM +0200, Andrew Lunn wrote:
> And now you mentions legacy Fixed link:
> 
> +MODULE_DESCRIPTION("Broadcom 10G Automotive Ethernet PCIe driver");
> 
> This claims it is a 10G device. You cannot represent 10G using legacy
> fixed link.

While it may be a 10G device, it seems the fixed-link specification
in the driver is set to 1G !
Simon Horman May 12, 2024, 8:35 a.m. UTC | #8
On Fri, May 10, 2024 at 06:59:24PM -0700, Jitendra Vegiraju wrote:
> Broadcom BCM8958X SoCs use Synopsys XGMAC design, which is similar to
> dwxgmac2 core implementation in stmmac driver. The existing dwxgmac2 dma
> operation functions have some conflicting differences with BCM8958X.
> This glue driver attempts to reuse dwxgmac2 implementation wherever
> possible, adding alternative implementations where necessary.
> 
> v2: code cleanup to address patchwork reports.
> 
> Signed-off-by: Jitendra Vegiraju <jitendra.vegiraju@broadcom.com>

...

> diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac-brcm.c b/drivers/net/ethernet/stmicro/stmmac/dwmac-brcm.c

...

> +static struct mac_device_info *dwxgmac_brcm_setup(void *ppriv)
> +{
> +	struct mac_device_info *mac;
> +	struct stmmac_priv *priv = ppriv;

Hi, Jitendra,

A minor nit from my side.

Please consider using reverse xmas tree order - longest line to shortest -
for new Networking code.

This tool can be of assistance: https://github.com/ecree-solarflare/xmastree

...
Jitendra Vegiraju May 13, 2024, 5:32 p.m. UTC | #9
On Sat, May 11, 2024 at 10:50 AM Andrew Lunn <andrew@lunn.ch> wrote:
>
> On Sat, May 11, 2024 at 06:12:38PM +0100, Russell King (Oracle) wrote:
> > On Sat, May 11, 2024 at 06:16:52PM +0200, Andrew Lunn wrote:
> > > > + /* This device interface is directly attached to the switch chip on
> > > > +  *  the SoC. Since no MDIO is present, register fixed_phy.
> > > > +  */
> > > > + brcm_priv->phy_dev =
> > > > +          fixed_phy_register(PHY_POLL,
> > > > +                             &dwxgmac_brcm_fixed_phy_status, NULL);
> > > > + if (IS_ERR(brcm_priv->phy_dev)) {
> > > > +         dev_err(&pdev->dev, "%s\tNo PHY/fixed_PHY found\n", __func__);
> > > > +         return -ENODEV;
> > > > + }
> > > > + phy_attached_info(brcm_priv->phy_dev);
> > >
> > > What switch is it? Will there be patches to extend SF2?
> >
> > ... and why is this legacy fixed_phy even necessary when stmmac uses
> > phylink which supports fixed links, including with custom fixed status?
>
> And now you mentions legacy Fixed link:
>
> +MODULE_DESCRIPTION("Broadcom 10G Automotive Ethernet PCIe driver");
>
> This claims it is a 10G device. You cannot represent 10G using legacy
> fixed link.
>
> Does this MAC directly connect to the switch within the SoC? There is
> no external MII interface? Realtek have been posting a MAC driver for
> something similar were the MAC is directly connected to the switch
> within the SoC. The MAC is fixed at 5G, there is no phylink/phylib
> support, set_link_ksetting return -EOPNOTSUPP and get_link_ksettings
> returns hard coded 5G.
>
> We need a better understanding of the architecture here, before we can
> advise the correct way to do this.
>
Yes, the MAC directly connects to switch within the SoC with no external MII.
The SoC is BCM89586M/BCM89587 automotive ethernet switch.
The SOC presents PCIE interfaces on BCM89586M/BCM89587 automotive
ethernet switch.
The switch supports many ethernet interfaces out of which one or two
interfaces are presented as PCIE endpoints to the host connected on
the PCIE bus.
The MAC connects to switch using XGMII interface internal to the SOC.
The high level diagram is shown below:

+==================================================+
   +--------+                     |                     BCM8958X
switch SoC               +----------------+         |
   | Host   |                      |  +----------------+
    +-------+                 |                     |         | ===
more ethernet IFs
   | CPU   | ===PCIE===| PCIE endpoint |==DMA==| MAC |==XGMII==|
switch fabric |         | === more ethernet IFs
   |Linux   |                      | +----------------+
   +-------+                 |                      |         |
   +-------+                       |
                                      +-----------------+        |

+==================================================+
Since the legacy fixed link cannot support 10G, we are initializing to
fixed speed 1G.
>       Andrew
Jitendra Vegiraju May 13, 2024, 5:38 p.m. UTC | #10
Thanks for reviewing the patch.
On Sat, May 11, 2024 at 12:34 PM Russell King (Oracle)
<linux@armlinux.org.uk> wrote:
>
> Hi,
>
> Thanks for the patch,. but there are things that need some improvement.
>
> On Fri, May 10, 2024 at 06:59:24PM -0700, Jitendra Vegiraju wrote:
> > +static void dwxgmac_brcm_dma_init_tx_chan(struct stmmac_priv *priv,
> > +                                       void __iomem *ioaddr,
> > +                                       struct stmmac_dma_cfg *dma_cfg,
> > +                                       dma_addr_t phy, u32 chan)
> > +{
> > +     u32 value;
> > +
> > +     value = readl(ioaddr + XGMAC_DMA_CH_TX_CONTROL(chan));
> > +     value &= ~XGMAC_TxPBL;
> > +     value &= ~GENMASK(6, 4);
> > +     writel(value, ioaddr + XGMAC_DMA_CH_TX_CONTROL(chan));
> > +
> > +     writel(upper_32_bits(phy), ioaddr + XGMAC_DMA_CH_TxDESC_HADDR(chan));
> > +     writel(lower_32_bits(phy), ioaddr + XGMAC_DMA_CH_TxDESC_LADDR(chan));
>
> Please use "dma_addr" not "phy" here. "phy" could mean ethernet phy.
> I personally dislike "physical address" for DMA stuff because if
> there's an IOMMU or other translation layer present, what you have
> here is *not* a physical address.
>
Yes, we will address it.
> > +static void dwxgmac_brcm_dma_init_rx_chan(struct stmmac_priv *priv,
> > +                                       void __iomem *ioaddr,
> > +                                       struct stmmac_dma_cfg *dma_cfg,
> > +                                       dma_addr_t phy, u32 chan)
> > +{
> > +     u32 value;
> > +
> > +     value = readl(ioaddr + XGMAC_DMA_CH_RX_CONTROL(chan));
> > +     value &= ~XGMAC_RxPBL;
> > +     writel(value, ioaddr + XGMAC_DMA_CH_RX_CONTROL(chan));
> > +
> > +     writel(upper_32_bits(phy), ioaddr + XGMAC_DMA_CH_RxDESC_HADDR(chan));
> > +     writel(lower_32_bits(phy), ioaddr + XGMAC_DMA_CH_RxDESC_LADDR(chan));
>
> Ditto.
>
Noted.
> ...
>
> > +static void dwxgmac_brcm_fix_speed(void *priv, unsigned int speed,
> > +                                unsigned int mode)
> > +{
> > +}
>
> If this is empty, do you really need it? The method is optional.
>
> ...
>
> > +static int dwxgmac_brcm_pci_probe(struct pci_dev *pdev,
> > +                               const struct pci_device_id *id)
> > +{
> ...
> > +     /* This device interface is directly attached to the switch chip on
> > +      *  the SoC. Since no MDIO is present, register fixed_phy.
> > +      */
> > +     brcm_priv->phy_dev =
> > +              fixed_phy_register(PHY_POLL,
> > +                                 &dwxgmac_brcm_fixed_phy_status, NULL);
> > +     if (IS_ERR(brcm_priv->phy_dev)) {
> > +             dev_err(&pdev->dev, "%s\tNo PHY/fixed_PHY found\n", __func__);
> > +             return -ENODEV;
> > +     }
> > +     phy_attached_info(brcm_priv->phy_dev);
>
> As pointed out in the other sub-thread, you don't need this. If you need
> a fixed-link and you don't have a firmware description of it, you can
> provide a swnode based description through plat->port_node that will be
> passed to phylink. Through that, you can tell phylink to create a
> fixed link.
>
Thank you for the pointers or software node support.
Since the driver is initially targetted for X86/_64, we were not sure
how to deal with lack of OF support.
We will try out the software node facility.
> > +     ret = stmmac_dvr_probe(&pdev->dev, plat, &res);
> > +     if (ret)
> > +             goto err_disable_msi;
> > +
> > +     /* The stmmac core driver doesn't have the infrastructure to
> > +      * support fixed-phy mdio bus for non-platform bus drivers.
> > +      * Until a better solution is implemented, initialize the
> > +      * following entries after priv structure is populated.
> > +      */
> > +     ndev = dev_get_drvdata(&pdev->dev);
> > +     priv = netdev_priv(ndev);
> > +     priv->mii = mdio_find_bus("fixed-0");
> > +
> > +     ndev->hw_features &= ~NETIF_F_HW_VLAN_CTAG_RX;
> > +     priv->hw->hw_vlan_en = false;
>
> Basically... no. Do not do any setup after stmmac_dvr_probe(), because
> the network device has already been registered and published to
> userspace, and userspace may have already opened the network device.
>
This will be tied to the above topic.
> --
> RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
> FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!
Russell King (Oracle) May 13, 2024, 5:41 p.m. UTC | #11
On Mon, May 13, 2024 at 10:38:46AM -0700, Jitendra Vegiraju wrote:
> Thanks for reviewing the patch.
> On Sat, May 11, 2024 at 12:34 PM Russell King (Oracle)
> <linux@armlinux.org.uk> wrote:
> > As pointed out in the other sub-thread, you don't need this. If you need
> > a fixed-link and you don't have a firmware description of it, you can
> > provide a swnode based description through plat->port_node that will be
> > passed to phylink. Through that, you can tell phylink to create a
> > fixed link.
> >
> Thank you for the pointers or software node support.
> Since the driver is initially targetted for X86/_64, we were not sure
> how to deal with lack of OF support.
> We will try out the software node facility.

You may wish to have a look at drivers/net/ethernet/wangxun/ which
also creates software nodes for phylink.

Thanks.
Andrew Lunn May 13, 2024, 6:07 p.m. UTC | #12
> Yes, the MAC directly connects to switch within the SoC with no external MII.
> The SoC is BCM89586M/BCM89587 automotive ethernet switch.
> The SOC presents PCIE interfaces on BCM89586M/BCM89587 automotive
> ethernet switch.
> The switch supports many ethernet interfaces out of which one or two
> interfaces are presented as PCIE endpoints to the host connected on
> the PCIE bus.
> The MAC connects to switch using XGMII interface internal to the SOC.
> The high level diagram is shown below:
> 
> +==================================================+
>    +--------+                     |                     BCM8958X
> switch SoC               +----------------+         |
>    | Host   |                      |  +----------------+
>     +-------+                 |                     |         | ===
> more ethernet IFs
>    | CPU   | ===PCIE===| PCIE endpoint |==DMA==| MAC |==XGMII==|
> switch fabric |         | === more ethernet IFs
>    |Linux   |                      | +----------------+
>    +-------+                 |                      |         |
>    +-------+                       |
>                                       +-----------------+        |
> 
> +==================================================+
> Since the legacy fixed link cannot support 10G, we are initializing to
> fixed speed 1G.

You ASCII art is broken, probably because you are not using a fixed
width font.

So the interface between the MAC and the switch is fixed at XGMII. Is
the MAC actually capable of anything other than XGMII? If XGMII is all
it can do, then there is no need for a fixed link. You use a
fixed-link when you have a conventional off the shelf MAC which can do
10BaseT_Half through to 10GBaseT. Normally there would be a PHY
connected to the MAC and phylib/phylink will determine the line rate
and tell the MAC what speed to operate at. However, if this device
only supports XGMII, it is impossible to connect to a PHY because
there is no external MII interface, then skip all the phylib/phylink
support and hard code it. Look at the patches on the netdev list for
the RealTek automotive driver which seems to be pretty similar.

	   Andrew
Andrew Lunn May 13, 2024, 7:52 p.m. UTC | #13
On Mon, May 13, 2024 at 06:41:35PM +0100, Russell King (Oracle) wrote:
> On Mon, May 13, 2024 at 10:38:46AM -0700, Jitendra Vegiraju wrote:
> > Thanks for reviewing the patch.
> > On Sat, May 11, 2024 at 12:34 PM Russell King (Oracle)
> > <linux@armlinux.org.uk> wrote:
> > > As pointed out in the other sub-thread, you don't need this. If you need
> > > a fixed-link and you don't have a firmware description of it, you can
> > > provide a swnode based description through plat->port_node that will be
> > > passed to phylink. Through that, you can tell phylink to create a
> > > fixed link.
> > >
> > Thank you for the pointers or software node support.
> > Since the driver is initially targetted for X86/_64, we were not sure
> > how to deal with lack of OF support.
> > We will try out the software node facility.
> 
> You may wish to have a look at drivers/net/ethernet/wangxun/ which
> also creates software nodes for phylink.

How complex is the switch configuration? So far, you have not said
anything about it. Is it derived from b53/SF2?

There is an alternative route you can take. Work with bootlin and use
DT overlays.

https://lore.kernel.org/linux-pci/20240430083730.134918-1-herve.codina@bootlin.com/

Looking at the product brief, the BCM89586M has MDIO busses, SPI
busses, GPIO, etc. It is unclear if these are available on the PCIe
interface, or are only connected to the Cortex-M7? I would guess the
QSPI, DEBUG/JTAG and the UART go to the M7, for its boot media and
console. But the other interfaces could be for Linux to control over
the PCIe. Additionally, the PHY-less ports doing XFI, 5G, 2.5G SGMII
etc, would have either an SFP or multi-gigi PHY connected, hanging of
one of the MDIO busses, GPIOs used for SFP LOS, TX-enable etc. Oddly
there is no I2C for the SPF, but i suppose you could do SPI->I2C.
Anyway, all that is going to need a complex configuration, so maybe DT
overlays make sense, because once the initial work getting Bootlins
patches merged is complete, you get the rest pretty much for free.

	Andrew
Russell King (Oracle) May 14, 2024, 8:19 a.m. UTC | #14
On Mon, May 13, 2024 at 10:32:19AM -0700, Jitendra Vegiraju wrote:
> +==================================================+
> Since the legacy fixed link cannot support 10G, we are initializing to
> fixed speed 1G.

Or to put it a different way... "I can't represent my hardware so I'm
going to hack around with the kernel in a way that lies to the kernel
about what the hardware is doing but it'll work for me!"

Sorry, but no, this isn't some hacky github project, this is the kernel
where we engineer proper solutions.

I think I've just lost all interest in this...
Florian Fainelli May 14, 2024, 4:18 p.m. UTC | #15
On 5/14/24 01:19, Russell King (Oracle) wrote:
> On Mon, May 13, 2024 at 10:32:19AM -0700, Jitendra Vegiraju wrote:
>> +==================================================+
>> Since the legacy fixed link cannot support 10G, we are initializing to
>> fixed speed 1G.
> 
> Or to put it a different way... "I can't represent my hardware so I'm
> going to hack around with the kernel in a way that lies to the kernel
> about what the hardware is doing but it'll work for me!"
> 
> Sorry, but no, this isn't some hacky github project, this is the kernel
> where we engineer proper solutions.

You are painting a picture of someone who is a first time contributor to 
Linux, there should not be any ill intentions at that point, just a 
knowledge gap that needs to be filled.

When I reviewed the patches internally the topic of fixed-link versus 
using PHYLINK did come up, and I should have dug a lot more and asked 
more questions to understand the full picture. Since the bulk of the 
changes had to do with hooking up a different DMA engine and 
configuration, I focused on that part and let the PHY connectivity 
aspect slip.

I will work with Jitendra to bring him up to speed with the software 
nodes, the work that Herve is doing on supporting OF overlays with PCIe 
root complexes and when net-next re-opens, we should have a different 
submission for you to look at then.

Thanks for your patience.
diff mbox series

Patch

diff --git a/MAINTAINERS b/MAINTAINERS
index b81b2be60b77..1eaf52047810 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -4115,6 +4115,13 @@  N:	brcmstb
 N:	bcm7038
 N:	bcm7120
 
+BROADCOM BCM8958X ETHERNET DRIVER
+M:	Jitendra Vegiraju <jitendra.vegiraju@broadcom.com>
+R:	Broadcom internal kernel review list <bcm-kernel-feedback-list@broadcom.com>
+L:	netdev@vger.kernel.org
+S:	Maintained
+F:	drivers/net/ethernet/stmicro/stmmac/dwmac-brcm.c
+
 BROADCOM BCMBCA ARM ARCHITECTURE
 M:	William Zhang <william.zhang@broadcom.com>
 M:	Anand Gore <anand.gore@broadcom.com>
diff --git a/drivers/net/ethernet/stmicro/stmmac/Kconfig b/drivers/net/ethernet/stmicro/stmmac/Kconfig
index 4ec61f1ee71a..6c06149712c8 100644
--- a/drivers/net/ethernet/stmicro/stmmac/Kconfig
+++ b/drivers/net/ethernet/stmicro/stmmac/Kconfig
@@ -286,6 +286,17 @@  config DWMAC_LOONGSON
 	  This selects the LOONGSON PCI bus support for the stmmac driver,
 	  Support for ethernet controller on Loongson-2K1000 SoC and LS7A1000 bridge.
 
+config DWMAC_BRCM
+	tristate "Broadcom XGMAC support"
+	depends on STMMAC_ETH && PCI
+	depends on COMMON_CLK
+	help
+	  Support for ethernet controllers on Broadcom BCM8958x SoCs
+
+	  This selects Broadcom XGMAC specific PCI bus support for the
+	  stmmac driver. This driver provides the glue layer on top of the
+	  stmmac driver required for the Broadcom BCM8958x SoC devices.
+
 config STMMAC_PCI
 	tristate "STMMAC PCI bus support"
 	depends on STMMAC_ETH && PCI
diff --git a/drivers/net/ethernet/stmicro/stmmac/Makefile b/drivers/net/ethernet/stmicro/stmmac/Makefile
index 26cad4344701..1cd0f508bafb 100644
--- a/drivers/net/ethernet/stmicro/stmmac/Makefile
+++ b/drivers/net/ethernet/stmicro/stmmac/Makefile
@@ -40,4 +40,5 @@  dwmac-altr-socfpga-objs := dwmac-socfpga.o
 obj-$(CONFIG_STMMAC_PCI)	+= stmmac-pci.o
 obj-$(CONFIG_DWMAC_INTEL)	+= dwmac-intel.o
 obj-$(CONFIG_DWMAC_LOONGSON)	+= dwmac-loongson.o
+obj-$(CONFIG_DWMAC_BRCM)	+= dwmac-brcm.o
 stmmac-pci-objs:= stmmac_pci.o
diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac-brcm.c b/drivers/net/ethernet/stmicro/stmmac/dwmac-brcm.c
new file mode 100644
index 000000000000..74f8e137d823
--- /dev/null
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac-brcm.c
@@ -0,0 +1,638 @@ 
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (c) 2024 Broadcom Corporation
+ * This file contains the functions to handle the Broadcom XGMAC PCI driver.
+ *
+ */
+
+#include <linux/clk-provider.h>
+#include <linux/dmi.h>
+#include <linux/pci.h>
+#include <linux/phy.h>
+#include <linux/phy_fixed.h>
+
+#include "stmmac.h"
+#include "dwxgmac2.h"
+
+#define PCI_DEVICE_ID_BROADCOM_BCM8958X		0xa00d
+#define BRCM_MAX_MTU				1500
+#define READ_POLL_DELAY_US			100
+#define READ_POLL_TIMEOUT_US			10000
+#define DWMAC_125MHZ				125000000
+#define DWMAC_250MHZ				250000000
+
+/* TX and RX Queue couts */
+#define BRCM_TX_Q_COUNT				4
+#define BRCM_RX_Q_COUNT				1
+
+/* PDMA Channel counts */
+#define PDMA_TX_CH_COUNT			8
+#define PDMA_RX_CH_COUNT			10
+
+/* PDMA register type */
+#define PDMA_CH_TX_EXT_CFGR			0
+#define PDMA_CH_RX_EXT_CFGR			1
+#define PDMA_CH_TX_DBG_STSR			2
+#define PDMA_CH_RX_DBG_STSR			3
+
+/* VDMA register type */
+#define VDMA_CH_TX_DESC_CTRLR			4
+#define VDMA_CH_RX_DESC_CTRLR			5
+
+/* VDMA channel count */
+#define VDMA_TOTAL_CH_COUNT			32
+
+#define DMA_CH_IND_CTRLR			0x3080
+#define DMA_CH_IND_DATAR			0x3084
+
+#define BRCM_XGMAC_RX_CFG			0x2000
+#define BRCM_XGMAC_RXQ_CTRL1_CFG		0x8000
+
+#define BRCM_XGMAC_DMA_TX_SIZE			4096
+#define BRCM_XGMAC_DMA_RX_SIZE			4096
+#define BRCM_XGMAC_BAR0_MASK			BIT(0)
+
+#define BRCM_XGMAC_IOMEM_MISC_REG_OFFSET	0x0
+#define BRCM_XGMAC_IOMEM_MBOX_REG_OFFSET	0x1000
+#define BRCM_XGMAC_IOMEM_CFG_REG_OFFSET		0x3000
+
+#define XGMAC_PCIE_CFG_MSIX_ADDR_MATCH_LOW	0x940
+#define XGMAC_PCIE_CFG_MSIX_ADDR_MATCH_LO_VALUE	0x00000001
+#define XGMAC_PCIE_CFG_MSIX_ADDR_MATCH_HIGH	0x944
+#define XGMAC_PCIE_CFG_MSIX_ADDR_MATCH_HI_VALUE	0x88000000
+
+#define XGMAC_PCIE_MISC_MII_CTRL			0x4
+#define XGMAC_PCIE_MISC_MII_CTRL_VALUE			0x7
+#define XGMAC_PCIE_MISC_PCIESS_CTRL			0x8
+#define XGMAC_PCIE_MISC_PCIESS_CTRL_VALUE		0x200
+#define XGMAC_PCIE_MISC_MSIX_ADDR_MATCH_LO		0x90
+#define XGMAC_PCIE_MISC_MSIX_ADDR_MATCH_LO_VALUE	0x00000001
+#define XGMAC_PCIE_MISC_MSIX_ADDR_MATCH_HI		0x94
+#define XGMAC_PCIE_MISC_MSIX_ADDR_MATCH_HI_VALUE	0x88000000
+#define XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_EP2HOST0	0x700
+#define XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_EP2HOST0_VALUE	1
+#define XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_EP2HOST1	0x704
+#define XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_EP2HOST1_VALUE	1
+#define XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_EP2HOST_DBELL	0x728
+#define XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_EP2HOST_DBELL_VALUE	1
+#define XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_SBD_ALL		0x740
+#define XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_SBD_ALL_VALUE	0
+
+#define DMA_CH_IND_CTRLR_MSEL_OFF		24
+#define DMA_CH_IND_CTRLR_MSEL_MASK		GENMASK(27, 24)
+#define DMA_CH_IND_CTRLR_AOFF_OFF		8
+#define DMA_CH_IND_CTRLR_AOFF_MASK		GENMASK(14, 8)
+#define DMA_CH_IND_CTRLR_AUTO_OFF		4
+#define DMA_CH_IND_CTRLR_AUTO_MASK		GENMASK(5, 4)
+#define DMA_CH_IND_CTRLR_CT_OFF			1
+#define DMA_CH_IND_CTRLR_CT_MASK		BIT(1)
+#define DMA_CH_IND_CTRLR_OB_OFF			0
+#define DMA_CH_IND_CTRLR_OB_MASK		BIT(0)
+
+/* DMA Descriptor configuration */
+#define BRCM_PDMA_DESC_CTRL_CFG_VALUE		0x1B
+
+#define XGMAC_PCIE_MISC_FUNC_RESOURCES_PF0	0x804
+
+/* MSIX Vector map register starting offsets */
+#define XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_RX0_PF0	0x840
+#define XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_TX0_PF0	0x890
+#define BRCM_MAX_DMA_CHANNEL_PAIRS		4
+
+#define BRCM_XGMAC_MSI_MAC_VECTOR		0
+#define BRCM_XGMAC_MSI_RX_VECTOR_START		9
+#define BRCM_XGMAC_MSI_TX_VECTOR_START		10
+
+static int num_instances;
+
+struct brcm_priv_data {
+	struct phy_device *phy_dev;
+	void __iomem *mbox_regs;    /* MBOX */
+	void __iomem *misc_regs;    /* MISC_cfg */
+	u16	dev_id;
+	u16	phy_addr;
+};
+
+static struct fixed_phy_status dwxgmac_brcm_fixed_phy_status = {
+	.link	= 1,
+	.speed	= SPEED_1000,
+	.duplex	= DUPLEX_FULL,
+};
+
+struct dwxgmac_brcm_pci_info {
+	int (*setup)(struct pci_dev *pdev, struct plat_stmmacenet_data *plat);
+};
+
+static void misc_iowrite(struct brcm_priv_data *brcm_priv,
+			 u32 reg, u32 val)
+{
+	iowrite32(val, brcm_priv->misc_regs + reg);
+}
+
+static void dwxgmac_brcm_pdma_set(void __iomem *ioaddr, u32 type, u32 chan,
+				  u32 val)
+{
+	u32 var = 0;
+
+	var |= FIELD_PREP(DMA_CH_IND_CTRLR_MSEL_MASK, type);
+	var |= FIELD_PREP(DMA_CH_IND_CTRLR_AOFF_MASK, chan);
+	var |= FIELD_PREP(DMA_CH_IND_CTRLR_CT_MASK, 0);
+	var |= FIELD_PREP(DMA_CH_IND_CTRLR_OB_MASK, 1);
+
+	if (!FIELD_GET(DMA_CH_IND_CTRLR_OB_MASK,
+		       readl(ioaddr + DMA_CH_IND_CTRLR))) {
+		writel(0x0, (ioaddr + DMA_CH_IND_CTRLR));
+		writel(val, (ioaddr + DMA_CH_IND_DATAR));
+	}
+
+	writel(var, (ioaddr + DMA_CH_IND_CTRLR));
+	readl_poll_timeout(ioaddr + DMA_CH_IND_CTRLR, var,
+			   !(var & XGMAC_OB), READ_POLL_TIMEOUT_US,
+			   READ_POLL_TIMEOUT_US);
+}
+
+static void dwxgmac_brcm_dma_init(void __iomem *ioaddr,
+				  struct stmmac_dma_cfg *dma_cfg, int atds)
+{
+	u32 val = dma_cfg->pbl << 24;
+	u32 i;
+
+	if (dma_cfg->pblx8)
+		val |= (1 << 19);
+
+	dwxgmac2_dma_init(ioaddr, dma_cfg, atds);
+
+	for (i = 0; i < PDMA_TX_CH_COUNT; i++)
+		dwxgmac_brcm_pdma_set(ioaddr, PDMA_CH_TX_EXT_CFGR, i, val);
+
+	for (i = 0; i < PDMA_RX_CH_COUNT; i++)
+		dwxgmac_brcm_pdma_set(ioaddr, PDMA_CH_RX_EXT_CFGR, i, val);
+
+	for (i = 0; i < VDMA_TOTAL_CH_COUNT; i++) {
+		dwxgmac_brcm_pdma_set(ioaddr, VDMA_CH_TX_DESC_CTRLR, i,
+				      BRCM_PDMA_DESC_CTRL_CFG_VALUE);
+		dwxgmac_brcm_pdma_set(ioaddr, VDMA_CH_RX_DESC_CTRLR, i,
+				      BRCM_PDMA_DESC_CTRL_CFG_VALUE);
+	}
+}
+
+static void dwxgmac_brcm_dma_init_tx_chan(struct stmmac_priv *priv,
+					  void __iomem *ioaddr,
+					  struct stmmac_dma_cfg *dma_cfg,
+					  dma_addr_t phy, u32 chan)
+{
+	u32 value;
+
+	value = readl(ioaddr + XGMAC_DMA_CH_TX_CONTROL(chan));
+	value &= ~XGMAC_TxPBL;
+	value &= ~GENMASK(6, 4);
+	writel(value, ioaddr + XGMAC_DMA_CH_TX_CONTROL(chan));
+
+	writel(upper_32_bits(phy), ioaddr + XGMAC_DMA_CH_TxDESC_HADDR(chan));
+	writel(lower_32_bits(phy), ioaddr + XGMAC_DMA_CH_TxDESC_LADDR(chan));
+}
+
+static void dwxgmac_brcm_dma_init_rx_chan(struct stmmac_priv *priv,
+					  void __iomem *ioaddr,
+					  struct stmmac_dma_cfg *dma_cfg,
+					  dma_addr_t phy, u32 chan)
+{
+	u32 value;
+
+	value = readl(ioaddr + XGMAC_DMA_CH_RX_CONTROL(chan));
+	value &= ~XGMAC_RxPBL;
+	writel(value, ioaddr + XGMAC_DMA_CH_RX_CONTROL(chan));
+
+	writel(upper_32_bits(phy), ioaddr + XGMAC_DMA_CH_RxDESC_HADDR(chan));
+	writel(lower_32_bits(phy), ioaddr + XGMAC_DMA_CH_RxDESC_LADDR(chan));
+}
+
+static const struct stmmac_dma_ops dwxgmac_brcm_dma_ops = {
+	.reset = dwxgmac2_dma_reset,
+	.init = dwxgmac_brcm_dma_init,
+	.init_chan = dwxgmac2_dma_init_chan,
+	.init_rx_chan = dwxgmac_brcm_dma_init_rx_chan,
+	.init_tx_chan = dwxgmac_brcm_dma_init_tx_chan,
+	.axi = dwxgmac2_dma_axi,
+	.dump_regs = dwxgmac2_dma_dump_regs,
+	.dma_rx_mode = dwxgmac2_dma_rx_mode,
+	.dma_tx_mode = dwxgmac2_dma_tx_mode,
+	.enable_dma_irq = dwxgmac2_enable_dma_irq,
+	.disable_dma_irq = dwxgmac2_disable_dma_irq,
+	.start_tx = dwxgmac2_dma_start_tx,
+	.stop_tx = dwxgmac2_dma_stop_tx,
+	.start_rx = dwxgmac2_dma_start_rx,
+	.stop_rx = dwxgmac2_dma_stop_rx,
+	.dma_interrupt = dwxgmac2_dma_interrupt,
+	.get_hw_feature = dwxgmac2_get_hw_feature,
+	.rx_watchdog = dwxgmac2_rx_watchdog,
+	.set_rx_ring_len = dwxgmac2_set_rx_ring_len,
+	.set_tx_ring_len = dwxgmac2_set_tx_ring_len,
+	.set_rx_tail_ptr = dwxgmac2_set_rx_tail_ptr,
+	.set_tx_tail_ptr = dwxgmac2_set_tx_tail_ptr,
+	.enable_tso = dwxgmac2_enable_tso,
+	.qmode = dwxgmac2_qmode,
+	.set_bfsize = dwxgmac2_set_bfsize,
+	.enable_sph = dwxgmac2_enable_sph,
+	.enable_tbs = dwxgmac2_enable_tbs,
+};
+
+static void dwxgmac_brcm_fix_speed(void *priv, unsigned int speed,
+				   unsigned int mode)
+{
+}
+
+static struct mac_device_info *dwxgmac_brcm_setup(void *ppriv)
+{
+	struct mac_device_info *mac;
+	struct stmmac_priv *priv = ppriv;
+
+	mac = devm_kzalloc(priv->device, sizeof(*mac), GFP_KERNEL);
+	if (!mac)
+		return NULL;
+
+	mac->dma = &dwxgmac_brcm_dma_ops;
+
+	priv->dma_conf.dma_tx_size = BRCM_XGMAC_DMA_TX_SIZE;
+	priv->dma_conf.dma_rx_size = BRCM_XGMAC_DMA_RX_SIZE;
+	priv->plat->rss_en = 1;
+	mac->pcsr = priv->ioaddr;
+	priv->dev->priv_flags |= IFF_UNICAST_FLT;
+	mac->multicast_filter_bins = priv->plat->multicast_filter_bins;
+	mac->unicast_filter_entries = priv->plat->unicast_filter_entries;
+	mac->mcast_bits_log2 = 0;
+
+	if (mac->multicast_filter_bins)
+		mac->mcast_bits_log2 = ilog2(mac->multicast_filter_bins);
+
+	mac->link.duplex = 0;
+	mac->link.xgmii.speed10000 = XGMAC_CONFIG_SS_10000;
+	mac->link.speed_mask = XGMAC_CONFIG_SS_MASK;
+	return mac;
+}
+
+static void dwxgmac_brcm_common_default_data(struct plat_stmmacenet_data *plat)
+{
+	int i;
+
+	plat->has_xgmac = 1;
+	plat->force_sf_dma_mode = 1;
+	plat->mac_port_sel_speed = SPEED_10000;
+	plat->clk_ptp_rate = DWMAC_125MHZ;
+	plat->clk_ref_rate = DWMAC_250MHZ;
+	plat->setup = dwxgmac_brcm_setup;
+	plat->tx_coe = 1;
+	plat->rx_coe = 1;
+	plat->max_speed = SPEED_10000;
+	plat->fix_mac_speed = dwxgmac_brcm_fix_speed;
+
+	/* Set default value for multicast hash bins */
+	plat->multicast_filter_bins = HASH_TABLE_SIZE;
+
+	/* Set default value for unicast filter entries */
+	plat->unicast_filter_entries = 1;
+
+	/* Set the maxmtu to device's default */
+	plat->maxmtu = BRCM_MAX_MTU;
+
+	/* Set default number of RX and TX queues to use */
+	plat->tx_queues_to_use = BRCM_TX_Q_COUNT;
+	plat->rx_queues_to_use = BRCM_RX_Q_COUNT;
+
+	plat->tx_sched_algorithm = MTL_TX_ALGORITHM_SP;
+	for (i = 0; i < plat->tx_queues_to_use; i++) {
+		plat->tx_queues_cfg[i].use_prio = false;
+		plat->tx_queues_cfg[i].prio = 0;
+		plat->tx_queues_cfg[i].mode_to_use = MTL_QUEUE_DCB;
+	}
+
+	plat->rx_sched_algorithm = MTL_RX_ALGORITHM_SP;
+	for (i = 0; i < plat->rx_queues_to_use; i++) {
+		plat->rx_queues_cfg[i].use_prio = false;
+		plat->rx_queues_cfg[i].mode_to_use = MTL_QUEUE_DCB;
+		plat->rx_queues_cfg[i].pkt_route = 0x0;
+		plat->rx_queues_cfg[i].chan = i;
+	}
+}
+
+static int dwxgmac_brcm_default_data(struct pci_dev *pdev,
+				     struct plat_stmmacenet_data *plat)
+{
+	struct brcm_priv_data *brcm_priv = plat->bsp_priv;
+
+	/* Set common default data first */
+	dwxgmac_brcm_common_default_data(plat);
+
+	plat->bus_id = 0;
+	plat->phy_addr = num_instances++;
+	brcm_priv->phy_addr = num_instances;
+	plat->phy_interface = PHY_INTERFACE_MODE_USXGMII;
+
+	plat->dma_cfg->pbl = 32;
+	plat->dma_cfg->pblx8 = 0;
+	plat->dma_cfg->aal = 0;
+	plat->dma_cfg->eame = 1;
+
+	plat->axi->axi_wr_osr_lmt = 31;
+	plat->axi->axi_rd_osr_lmt = 31;
+	plat->axi->axi_fb = 0;
+	plat->axi->axi_blen[0] = 4;
+	plat->axi->axi_blen[1] = 8;
+	plat->axi->axi_blen[2] = 16;
+	plat->axi->axi_blen[3] = 32;
+	plat->axi->axi_blen[4] = 64;
+	plat->axi->axi_blen[5] = 128;
+	plat->axi->axi_blen[6] = 256;
+
+	plat->msi_mac_vec = BRCM_XGMAC_MSI_MAC_VECTOR;
+	plat->msi_rx_base_vec = BRCM_XGMAC_MSI_RX_VECTOR_START;
+	plat->msi_tx_base_vec = BRCM_XGMAC_MSI_TX_VECTOR_START;
+
+	return 0;
+}
+
+static struct dwxgmac_brcm_pci_info dwxgmac_brcm_pci_info = {
+	.setup = dwxgmac_brcm_default_data,
+};
+
+static int brcm_config_multi_msi(struct pci_dev *pdev,
+				 struct plat_stmmacenet_data *plat,
+				 struct stmmac_resources *res)
+{
+	int ret;
+	int i;
+
+	if (plat->msi_rx_base_vec >= STMMAC_MSI_VEC_MAX ||
+	    plat->msi_tx_base_vec >= STMMAC_MSI_VEC_MAX) {
+		dev_err(&pdev->dev, "%s: Invalid RX & TX vector defined\n",
+			__func__);
+		return -EINVAL;
+	}
+
+	ret = pci_alloc_irq_vectors(pdev, 2, STMMAC_MSI_VEC_MAX,
+				    PCI_IRQ_MSI | PCI_IRQ_MSIX);
+	if (ret < 0) {
+		dev_err(&pdev->dev, "%s: multi MSI enablement failed\n",
+			__func__);
+		return ret;
+	}
+
+	/* For RX MSI */
+	for (i = 0; i < plat->rx_queues_to_use; i++)
+		res->rx_irq[i] = pci_irq_vector(pdev,
+						plat->msi_rx_base_vec + i * 2);
+
+	/* For TX MSI */
+	for (i = 0; i < plat->tx_queues_to_use; i++)
+		res->tx_irq[i] = pci_irq_vector(pdev,
+						plat->msi_tx_base_vec + i * 2);
+
+	if (plat->msi_mac_vec < STMMAC_MSI_VEC_MAX)
+		res->irq = pci_irq_vector(pdev, plat->msi_mac_vec);
+
+	plat->flags |= STMMAC_FLAG_MULTI_MSI_EN;
+	plat->flags |= STMMAC_FLAG_TSO_EN;
+
+	return 0;
+}
+
+static int dwxgmac_brcm_pci_probe(struct pci_dev *pdev,
+				  const struct pci_device_id *id)
+{
+	struct dwxgmac_brcm_pci_info *info =
+		(struct dwxgmac_brcm_pci_info *)id->driver_data;
+	struct plat_stmmacenet_data *plat;
+	struct brcm_priv_data *brcm_priv;
+	struct stmmac_resources res;
+	struct net_device *ndev;
+	struct stmmac_priv *priv;
+	int rx_offset;
+	int tx_offset;
+	int vector;
+	int ret;
+
+	brcm_priv = devm_kzalloc(&pdev->dev, sizeof(*brcm_priv), GFP_KERNEL);
+	if (!brcm_priv)
+		return -ENOMEM;
+
+	plat = devm_kzalloc(&pdev->dev, sizeof(*plat), GFP_KERNEL);
+	if (!plat)
+		return -ENOMEM;
+
+	plat->dma_cfg = devm_kzalloc(&pdev->dev, sizeof(*plat->dma_cfg),
+				     GFP_KERNEL);
+	if (!plat->dma_cfg)
+		return -ENOMEM;
+
+	plat->axi = devm_kzalloc(&pdev->dev, sizeof(*plat->axi), GFP_KERNEL);
+	if (!plat->axi)
+		return -ENOMEM;
+
+	pci_read_config_word(pdev, 2, &brcm_priv->dev_id);
+
+	/* This device interface is directly attached to the switch chip on
+	 *  the SoC. Since no MDIO is present, register fixed_phy.
+	 */
+	brcm_priv->phy_dev =
+		 fixed_phy_register(PHY_POLL,
+				    &dwxgmac_brcm_fixed_phy_status, NULL);
+	if (IS_ERR(brcm_priv->phy_dev)) {
+		dev_err(&pdev->dev, "%s\tNo PHY/fixed_PHY found\n", __func__);
+		return -ENODEV;
+	}
+	phy_attached_info(brcm_priv->phy_dev);
+
+	/* Disable D3COLD as our device does not support it */
+	pci_d3cold_disable(pdev);
+
+	/* Enable PCI device */
+	ret = pcim_enable_device(pdev);
+	if (ret) {
+		dev_err(&pdev->dev, "%s: ERROR: failed to enable device\n",
+			__func__);
+		return ret;
+	}
+
+	/* Get the base address of device */
+	ret = pcim_iomap_regions(pdev, BRCM_XGMAC_BAR0_MASK, pci_name(pdev));
+	if (ret)
+		goto err_disable_device;
+	pci_set_master(pdev);
+
+	memset(&res, 0, sizeof(res));
+	res.addr = pcim_iomap_table(pdev)[0];
+	/* MISC Regs */
+	brcm_priv->misc_regs = res.addr + BRCM_XGMAC_IOMEM_MISC_REG_OFFSET;
+	/* MBOX Regs */
+	brcm_priv->mbox_regs = res.addr + BRCM_XGMAC_IOMEM_MBOX_REG_OFFSET;
+	/* XGMAC config Regs */
+	res.addr += BRCM_XGMAC_IOMEM_CFG_REG_OFFSET;
+
+	plat->bsp_priv = brcm_priv;
+
+	/* Initialize all MSI vectors to invalid so that it can be set
+	 * according to platform data settings below.
+	 * Note: MSI vector takes value from 0 up to 31 (STMMAC_MSI_VEC_MAX)
+	 */
+	plat->msi_mac_vec = STMMAC_MSI_VEC_MAX;
+	plat->msi_wol_vec = STMMAC_MSI_VEC_MAX;
+	plat->msi_lpi_vec = STMMAC_MSI_VEC_MAX;
+	plat->msi_sfty_ce_vec = STMMAC_MSI_VEC_MAX;
+	plat->msi_sfty_ue_vec = STMMAC_MSI_VEC_MAX;
+	plat->msi_rx_base_vec = STMMAC_MSI_VEC_MAX;
+	plat->msi_tx_base_vec = STMMAC_MSI_VEC_MAX;
+
+	ret = info->setup(pdev, plat);
+	if (ret)
+		goto err_disable_device;
+
+	pci_write_config_dword(pdev, XGMAC_PCIE_CFG_MSIX_ADDR_MATCH_LOW,
+			       XGMAC_PCIE_CFG_MSIX_ADDR_MATCH_LO_VALUE);
+	pci_write_config_dword(pdev, XGMAC_PCIE_CFG_MSIX_ADDR_MATCH_HIGH,
+			       XGMAC_PCIE_CFG_MSIX_ADDR_MATCH_HI_VALUE);
+
+	misc_iowrite(brcm_priv, XGMAC_PCIE_MISC_MSIX_ADDR_MATCH_LO,
+		     XGMAC_PCIE_MISC_MSIX_ADDR_MATCH_LO_VALUE);
+	misc_iowrite(brcm_priv, XGMAC_PCIE_MISC_MSIX_ADDR_MATCH_HI,
+		     XGMAC_PCIE_MISC_MSIX_ADDR_MATCH_HI_VALUE);
+
+	/* SBD Interrupt */
+	misc_iowrite(brcm_priv, XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_SBD_ALL,
+		     XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_SBD_ALL_VALUE);
+	/* EP_DOORBELL Interrupt */
+	misc_iowrite(brcm_priv, XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_EP2HOST_DBELL,
+		     XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_EP2HOST_DBELL_VALUE);
+	/* EP_H0 Interrupt */
+	misc_iowrite(brcm_priv, XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_EP2HOST0,
+		     XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_EP2HOST0_VALUE);
+	/* EP_H1 Interrupt */
+	misc_iowrite(brcm_priv, XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_EP2HOST1,
+		     XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_EP2HOST1_VALUE);
+
+	rx_offset = XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_RX0_PF0;
+	tx_offset = XGMAC_PCIE_MISC_MSIX_VECTOR_MAP_TX0_PF0;
+	vector = BRCM_XGMAC_MSI_RX_VECTOR_START;
+	for (int i = 0; i < BRCM_MAX_DMA_CHANNEL_PAIRS; i++) {
+		/* RX Interrupt */
+		misc_iowrite(brcm_priv, rx_offset, vector++);
+		/* TX Interrupt */
+		misc_iowrite(brcm_priv, tx_offset, vector++);
+		rx_offset += 4;
+		tx_offset += 4;
+	}
+
+	/* Enable Switch Link */
+	misc_iowrite(brcm_priv, XGMAC_PCIE_MISC_MII_CTRL,
+		     XGMAC_PCIE_MISC_MII_CTRL_VALUE);
+	/* Enable MSI-X */
+	misc_iowrite(brcm_priv, XGMAC_PCIE_MISC_PCIESS_CTRL,
+		     XGMAC_PCIE_MISC_PCIESS_CTRL_VALUE);
+
+	ret = brcm_config_multi_msi(pdev, plat, &res);
+	if (ret) {
+		dev_err(&pdev->dev,
+			"%s: ERROR: failed to enable IRQ\n", __func__);
+		return ret;
+	}
+
+	ret = stmmac_dvr_probe(&pdev->dev, plat, &res);
+	if (ret)
+		goto err_disable_msi;
+
+	/* The stmmac core driver doesn't have the infrastructure to
+	 * support fixed-phy mdio bus for non-platform bus drivers.
+	 * Until a better solution is implemented, initialize the
+	 * following entries after priv structure is populated.
+	 */
+	ndev = dev_get_drvdata(&pdev->dev);
+	priv = netdev_priv(ndev);
+	priv->mii = mdio_find_bus("fixed-0");
+
+	ndev->hw_features &= ~NETIF_F_HW_VLAN_CTAG_RX;
+	priv->hw->hw_vlan_en = false;
+
+	dev_info(&pdev->dev, "%s\tComplete\n", __func__);
+
+	return ret;
+
+err_disable_msi:
+	pci_disable_msi(pdev);
+err_disable_device:
+	pci_disable_device(pdev);
+	return ret;
+}
+
+static void dwxgmac_brcm_pci_remove(struct pci_dev *pdev)
+{
+	struct net_device *ndev = dev_get_drvdata(&pdev->dev);
+	struct stmmac_priv *priv = netdev_priv(ndev);
+	struct brcm_priv_data *brcm_priv = priv->plat->bsp_priv;
+	struct phy_device *phydev = brcm_priv->phy_dev;
+
+	priv->mii = NULL;
+	stmmac_dvr_remove(&pdev->dev);
+	pcim_iounmap_regions(pdev, BRCM_XGMAC_BAR0_MASK);
+	pci_clear_master(pdev);
+	fixed_phy_unregister(phydev);
+}
+
+static int __maybe_unused dwxgmac_brcm_pci_suspend(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+	int ret;
+
+	ret = stmmac_suspend(dev);
+	if (ret)
+		return ret;
+
+	ret = pci_save_state(pdev);
+	if (ret)
+		return ret;
+
+	pci_disable_device(pdev);
+	pci_wake_from_d3(pdev, true);
+	return 0;
+}
+
+static int __maybe_unused dwxgmac_brcm_pci_resume(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+	int ret;
+
+	pci_restore_state(pdev);
+	pci_set_power_state(pdev, PCI_D0);
+
+	ret = pci_enable_device(pdev);
+	if (ret)
+		return ret;
+
+	pci_set_master(pdev);
+
+	return stmmac_resume(dev);
+}
+
+static SIMPLE_DEV_PM_OPS(dwxgmac_brcm_pm_ops,
+			 dwxgmac_brcm_pci_suspend,
+			 dwxgmac_brcm_pci_resume);
+
+static const struct pci_device_id dwxgmac_brcm_id_table[] = {
+	{ PCI_DEVICE_DATA(BROADCOM, BCM8958X, &dwxgmac_brcm_pci_info) },
+	{}
+};
+
+MODULE_DEVICE_TABLE(pci, dwxgmac_brcm_id_table);
+
+static struct pci_driver dwxgmac_brcm_pci_driver = {
+	.name = "brcm-bcm8958x",
+	.id_table = dwxgmac_brcm_id_table,
+	.probe	= dwxgmac_brcm_pci_probe,
+	.remove = dwxgmac_brcm_pci_remove,
+	.driver = {
+		.pm = &dwxgmac_brcm_pm_ops,
+	},
+};
+
+module_pci_driver(dwxgmac_brcm_pci_driver);
+
+MODULE_DESCRIPTION("Broadcom 10G Automotive Ethernet PCIe driver");
+MODULE_LICENSE("GPL");