[U-Boot] [PATCH v2] net: Add MT76xx ethernet driver
Joe Hershberger
joe.hershberger at ni.com
Mon Oct 22 19:36:10 UTC 2018
On Mon, Oct 22, 2018 at 9:31 AM Stefan Roese <sr at denx.de> wrote:
>
> Hi Joe,
>
> On 15.10.18 23:45, Joe Hershberger wrote:
> > On Mon, Oct 8, 2018 at 8:50 AM Stefan Roese <sr at denx.de> wrote:
> >>
> >> This patch adds ethernet support for the Mediatek MT76xx SoC, including
> >> a minimum setup of the integrated switch. This driver is loosly based on
> >> the driver version included in this MediaTek github repository:
> >>
> >> https://github.com/MediaTek-Labs/linkit-smart-uboot.git
> >>
> >> Tested on the MT7688 LinkIt smart-gateway and on the
> >> Gardena-smart-gateway.
> >>
> >> Signed-off-by: Stefan Roese <sr at denx.de>
> >> Cc: Daniel Schwierzeck <daniel.schwierzeck at gmail.com>
> >> Cc: Joe Hershberger <joe.hershberger at ni.com>
> >> ---
> >> v2:
> >> - Use wait_for_bit_le32() in more places
> >> - Use syscon / regmap to access the system-controller registers
> >> needed in this driver instead of finding and mapping this
> >> area
> >>
> >> drivers/net/Kconfig | 7 +
> >> drivers/net/Makefile | 1 +
> >> drivers/net/mt76xx-eth.c | 638 +++++++++++++++++++++++++++++++++++++++
> >> 3 files changed, 646 insertions(+)
> >> create mode 100644 drivers/net/mt76xx-eth.c
> >>
> >> diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
> >> index 5441da47d1..ed73f21a59 100644
> >> --- a/drivers/net/Kconfig
> >> +++ b/drivers/net/Kconfig
> >> @@ -227,6 +227,13 @@ config MACB_ZYNQ
> >> The Cadence MACB ethernet interface was used on Zynq platform.
> >> Say Y to enable support for the MACB/GEM in Zynq chip.
> >>
> >> +config MT76XX_ETH
> >> + bool "MediaTek MT76xx Ethernet Interface"
> >> + depends on ARCH_MT7620
> >> + help
> >> + The MediaTek MT76xx ethernet interface is used on MT7688
> >> + based boards.
> >> +
> >> config PCH_GBE
> >> bool "Intel Platform Controller Hub EG20T GMAC driver"
> >> depends on DM_ETH && DM_PCI
> >> diff --git a/drivers/net/Makefile b/drivers/net/Makefile
> >> index 48a2878071..05189514b7 100644
> >> --- a/drivers/net/Makefile
> >> +++ b/drivers/net/Makefile
> >> @@ -37,6 +37,7 @@ obj-$(CONFIG_LPC32XX_ETH) += lpc32xx_eth.o
> >> obj-$(CONFIG_MACB) += macb.o
> >> obj-$(CONFIG_MCFFEC) += mcffec.o mcfmii.o
> >> obj-$(CONFIG_MPC8XX_FEC) += mpc8xx_fec.o
> >> +obj-$(CONFIG_MT76XX_ETH) += mt76xx-eth.o
> >> obj-$(CONFIG_MVGBE) += mvgbe.o
> >> obj-$(CONFIG_MVNETA) += mvneta.o
> >> obj-$(CONFIG_MVPP2) += mvpp2.o
> >> diff --git a/drivers/net/mt76xx-eth.c b/drivers/net/mt76xx-eth.c
> >> new file mode 100644
> >> index 0000000000..a623a475af
> >> --- /dev/null
> >> +++ b/drivers/net/mt76xx-eth.c
> >> @@ -0,0 +1,638 @@
> >> +// SPDX-License-Identifier: GPL-2.0+
> >> +/*
> >> + * MediaTek ethernet IP driver for U-Boot
> >> + *
> >> + * Copyright (C) 2018 Stefan Roese <sr at denx.de>
> >> + *
> >> + * This code is mostly based on the code extracted from this MediaTek
> >> + * github repository:
> >> + *
> >> + * https://github.com/MediaTek-Labs/linkit-smart-uboot.git
> >> + *
> >> + * I was not able to find a specific license or other developers
> >> + * copyrights here, so I can't add them here.
> >> + */
> >> +
> >> +#include <common.h>
> >> +#include <dm.h>
> >> +#include <malloc.h>
> >> +#include <miiphy.h>
> >> +#include <net.h>
> >> +#include <regmap.h>
> >> +#include <syscon.h>
> >> +#include <wait_bit.h>
> >> +#include <asm/io.h>
> >> +#include <linux/err.h>
> >> +
> >> +/* System controller register */
> >> +#define MT76XX_RSTCTRL_REG 0x34
> >> +#define RSTCTRL_EPHY_RST BIT(24)
> >> +
> >> +#define MT76XX_AGPIO_CFG_REG 0x3c
> >> +#define MT7628_EPHY_GPIO_AIO_EN GENMASK(20, 17)
> >> +#define MT7628_EPHY_P0_DIS BIT(16)
> >> +
> >> +#define MT76XX_GPIO2_MODE_REG 0x64
> >> +
> >> +/* Ethernet frame engine register */
> >> +#define PDMA_RELATED 0x0800
> >> +
> >> +#define TX_BASE_PTR0 (PDMA_RELATED + 0x000)
> >> +#define TX_MAX_CNT0 (PDMA_RELATED + 0x004)
> >> +#define TX_CTX_IDX0 (PDMA_RELATED + 0x008)
> >> +#define TX_DTX_IDX0 (PDMA_RELATED + 0x00c)
> >> +
> >> +#define RX_BASE_PTR0 (PDMA_RELATED + 0x100)
> >> +#define RX_MAX_CNT0 (PDMA_RELATED + 0x104)
> >> +#define RX_CALC_IDX0 (PDMA_RELATED + 0x108)
> >> +
> >> +#define PDMA_GLO_CFG (PDMA_RELATED + 0x204)
> >> +#define PDMA_RST_IDX (PDMA_RELATED + 0x208)
> >> +#define DLY_INT_CFG (PDMA_RELATED + 0x20c)
> >> +
> >> +#define SDM_RELATED 0x0c00
> >> +
> >> +#define SDM_MAC_ADRL (SDM_RELATED + 0x0c) /* MAC address LSB */
> >> +#define SDM_MAC_ADRH (SDM_RELATED + 0x10) /* MAC Address MSB */
> >> +
> >> +#define RST_DTX_IDX0 BIT(0)
> >> +#define RST_DRX_IDX0 BIT(16)
> >> +
> >> +#define TX_DMA_EN BIT(0)
> >> +#define TX_DMA_BUSY BIT(1)
> >> +#define RX_DMA_EN BIT(2)
> >> +#define RX_DMA_BUSY BIT(3)
> >> +#define TX_WB_DDONE BIT(6)
> >> +
> >> +/* Ethernet switch register */
> >> +#define MT76XX_SWITCH_FCT0 0x0008
> >> +#define MT76XX_SWITCH_PFC1 0x0014
> >> +#define MT76XX_SWITCH_FPA 0x0084
> >> +#define MT76XX_SWITCH_SOCPC 0x008c
> >> +#define MT76XX_SWITCH_POC0 0x0090
> >> +#define MT76XX_SWITCH_POC2 0x0098
> >> +#define MT76XX_SWITCH_SGC 0x009c
> >> +#define MT76XX_SWITCH_PCR0 0x00c0
> >> +#define MT76XX_SWITCH_PCR1 0x00c4
> >> +#define MT76XX_SWITCH_FPA1 0x00c8
> >> +#define MT76XX_SWITCH_FCT2 0x00cc
> >> +#define MT76XX_SWITCH_SGC2 0x00e4
> >> +#define MT76XX_SWITCH_BMU_CTRL 0x0110
> >> +
> >> +/* rxd2 */
> >> +#define RX_DMA_DONE BIT(31)
> >> +#define RX_DMA_LSO BIT(30)
> >> +#define RX_DMA_PLEN0(_x) (((_x) & 0x3fff) << 16)
> >
> > For such things, please use FIELD_PREP(GENMASK(x,y),z) from
> > include/linux/bitfield.h
>
> Sure, I'll change this in the next patch version.
>
> >> +#define RX_DMA_GET_PLEN0(_x) (((_x) >> 16) & 0x3fff)
> >
> > For such things, please use FIELD_GET(GENMASK(x,y),z) from
> > include/linux/bitfield.h
>
> Ack.
>
> >> +#define RX_DMA_TAG BIT(15)
> >> +/* rxd3 */
> >> +#define RX_DMA_TPID(_x) (((_x) >> 16) & 0xffff)
> >> +#define RX_DMA_VID(_x) ((_x) & 0xffff)
> >> +/* rxd4 */
> >> +#define RX_DMA_L4VALID BIT(30)
> >> +
> >> +struct fe_rx_dma {
> >> + unsigned int rxd1;
> >> + unsigned int rxd2;
> >> + unsigned int rxd3;
> >> + unsigned int rxd4;
> >> +} __packed __aligned(4);
> >> +
> >> +#define TX_DMA_BUF_LEN 0x3fff
> >> +#define TX_DMA_PLEN0_MASK (TX_DMA_BUF_LEN << 16)
> >> +#define TX_DMA_PLEN0(_x) (((_x) & TX_DMA_BUF_LEN) << 16)
> >> +#define TX_DMA_PLEN1(_x) ((_x) & TX_DMA_BUF_LEN)
> >> +#define TX_DMA_GET_PLEN0(_x) (((_x) >> 16) & TX_DMA_BUF_LEN)
> >> +#define TX_DMA_GET_PLEN1(_x) ((_x) & TX_DMA_BUF_LEN)
> >> +#define TX_DMA_LS1 BIT(14)
> >> +#define TX_DMA_LS0 BIT(30)
> >> +#define TX_DMA_DONE BIT(31)
> >> +
> >> +#define TX_DMA_INS_VLAN_MT7621 BIT(16)
> >> +#define TX_DMA_INS_VLAN BIT(7)
> >> +#define TX_DMA_INS_PPPOE BIT(12)
> >> +#define TX_DMA_QN(_x) ((_x) << 16)
> >> +#define TX_DMA_PN(_x) ((_x) << 24)
> >> +#define TX_DMA_QN_MASK TX_DMA_QN(0x7)
> >> +#define TX_DMA_PN_MASK TX_DMA_PN(0x7)
> >> +#define TX_DMA_UDF BIT(20)
> >> +#define TX_DMA_CHKSUM (0x7 << 29)
> >> +#define TX_DMA_TSO BIT(28)
> >> +
> >> +struct fe_tx_dma {
> >> + unsigned int txd1;
> >> + unsigned int txd2;
> >> + unsigned int txd3;
> >> + unsigned int txd4;
> >> +} __packed __aligned(4);
> >> +
> >> +#define NUM_RX_DESC 256
> >> +#define NUM_TX_DESC 4
> >> +
> >> +#define PADDING_LENGTH 60
> >> +
> >> +#define MTK_QDMA_PAGE_SIZE 2048
> >> +
> >> +#define CONFIG_MDIO_TIMEOUT 100
> >> +#define CONFIG_DMA_STOP_TIMEOUT 100
> >> +#define CONFIG_TX_DMA_TIMEOUT 100
> >> +
> >> +#define LINK_DELAY_TIME 500 /* 500 ms */
> >> +#define LINK_TIMEOUT 10000 /* 10 seconds */
> >> +
> >> +struct mt76xx_eth_dev {
> >> + void __iomem *base; /* frame engine base address */
> >> + void __iomem *eth_sw_base; /* switch base address */
> >> + struct regmap *sysctrl_regmap; /* system-controller reg-map */
> >> +
> >> + struct mii_dev *bus;
> >> +
> >> + struct fe_tx_dma *tx_ring;
> >> + struct fe_rx_dma *rx_ring;
> >> +
> >> + u8 *rx_buf[NUM_RX_DESC];
> >> +
> >> + /* Point to the next RXD DMA wants to use in RXD Ring0 */
> >> + int rx_dma_idx;
> >> + /* Point to the next TXD in TXD Ring0 CPU wants to use */
> >> + int tx_dma_idx;
> >> +};
> >> +
> >> +static int mdio_wait_read(struct mt76xx_eth_dev *priv, u32 mask, bool mask_set)
> >> +{
> >> + void __iomem *base = priv->eth_sw_base;
> >> + int ret;
> >> +
> >> + ret = wait_for_bit_le32(base + MT76XX_SWITCH_PCR1, mask, mask_set,
> >> + CONFIG_MDIO_TIMEOUT, false);
> >> + if (ret) {
> >> + printf("MDIO operation timeout!\n");
> >> + return -ETIMEDOUT;
> >> + }
> >> +
> >> + return 0;
> >> +}
> >> +
> >> +static int mii_mgr_read(struct mt76xx_eth_dev *priv,
> >> + u32 phy_addr, u32 phy_register, u32 *read_data)
> >> +{
> >> + void __iomem *base = priv->eth_sw_base;
> >> + u32 status = 0;
> >> + u32 ret;
> >> +
> >> + *read_data = 0xffff;
> >> + /* Make sure previous read operation is complete */
> >> + ret = mdio_wait_read(priv, BIT(1), false);
> >
> > What is BIT(1) here? Should probably be a #define.
>
> Please note that the documentation is quite "sparse" here for this
> chip. As mentioned in the commit message, this driver version is
> based on the original MediaTek U-Boot driver, which frankly is in
> very poor shape. And of course these are no macros there as well.
> I'll try to find some definitions in the manual that I have access
> to. But please note, that I can't promise to change all "magic
> numbers" into correct macros 7 defines. I hope you understand.
Yes, sounds reasonable.
> >> + if (ret)
> >> + return ret;
> >> +
> >> + writel(BIT(14) | (phy_register << 8) | phy_addr,
> >
> > Magic numbers. #define?
>
> See above.
>
> >> + base + MT76XX_SWITCH_PCR0);
> >> +
> >> + /* Make sure previous read operation is complete */
> >> + ret = mdio_wait_read(priv, BIT(1), true);
> >> + if (ret)
> >> + return ret;
> >> +
> >> + status = readl(base + MT76XX_SWITCH_PCR1);
> >> + *read_data = (u32)(status >> 16);
> >
> > Again, this would be something to define a field for and then use FIELD_GET().
>
> Will do in the next version.
>
> > Please update throughout.
> >
> >> +
> >> + return 0;
> >> +}
> >> +
> >> +static int mii_mgr_write(struct mt76xx_eth_dev *priv,
> >> + u32 phy_addr, u32 phy_register, u32 write_data)
> >> +{
> >> + void __iomem *base = priv->eth_sw_base;
> >> + u32 data;
> >> + int ret;
> >> +
> >> + /* Make sure previous write operation is complete */
> >> + ret = mdio_wait_read(priv, BIT(0), false);
> >> + if (ret)
> >> + return ret;
> >> +
> >> + data = (write_data & 0xffff) << 16;
> >> + data |= (phy_register << 8) | phy_addr;
> >> + data |= BIT(13);
> >> + writel(data, base + MT76XX_SWITCH_PCR0);
> >> +
> >> + return mdio_wait_read(priv, BIT(0), true);
> >> +}
> >> +
> >> +static int mt76xx_mdio_read(struct mii_dev *bus, int addr, int devad, int reg)
> >> +{
> >> + u32 val;
> >> + int ret;
> >> +
> >> + ret = mii_mgr_read(bus->priv, addr, reg, &val);
> >> + if (ret)
> >> + return ret;
> >> +
> >> + return val;
> >> +}
> >> +
> >> +static int mt76xx_mdio_write(struct mii_dev *bus, int addr, int devad, int reg,
> >> + u16 value)
> >> +{
> >> + return mii_mgr_write(bus->priv, addr, reg, value);
> >> +}
> >> +
> >> +static void mt7628_ephy_init(struct mt76xx_eth_dev *priv)
> >> +{
> >> + int i;
> >> +
> >
> > Is there some datasheet or documentation you can reference here? Even
> > if not public, people who have NDA for the part will know where to
> > look.
>
> Not that I know of. My hope is that now that this mainline U-Boot
> port for the MediaTek MT76xx MIPS exisits, other developers / users
> will "jump in" in this development. Perhaps others have more internal
> informantion here.
That would be great.
>
> >> + mii_mgr_write(priv, 0, 31, 0x2000); /* change G2 page */
> >> + mii_mgr_write(priv, 0, 26, 0x0000);
> >> +
> >> + for (i = 0; i < 5; i++) {
> >> + mii_mgr_write(priv, i, 31, 0x8000); /* change L0 page */
> >> + mii_mgr_write(priv, i, 0, 0x3100);
> >> +
> >> + /* EEE disable */
> >> + mii_mgr_write(priv, i, 30, 0xa000);
> >> + mii_mgr_write(priv, i, 31, 0xa000); /* change L2 page */
> >> + mii_mgr_write(priv, i, 16, 0x0606);
> >> + mii_mgr_write(priv, i, 23, 0x0f0e);
> >> + mii_mgr_write(priv, i, 24, 0x1610);
> >> + mii_mgr_write(priv, i, 30, 0x1f15);
> >> + mii_mgr_write(priv, i, 28, 0x6111);
> >> + }
> >> +
> >> + /* 100Base AOI setting */
> >> + mii_mgr_write(priv, 0, 31, 0x5000); /* change G5 page */
> >> + mii_mgr_write(priv, 0, 19, 0x004a);
> >> + mii_mgr_write(priv, 0, 20, 0x015a);
> >> + mii_mgr_write(priv, 0, 21, 0x00ee);
> >> + mii_mgr_write(priv, 0, 22, 0x0033);
> >> + mii_mgr_write(priv, 0, 23, 0x020a);
> >> + mii_mgr_write(priv, 0, 24, 0x0000);
> >> + mii_mgr_write(priv, 0, 25, 0x024a);
> >> + mii_mgr_write(priv, 0, 26, 0x035a);
> >> + mii_mgr_write(priv, 0, 27, 0x02ee);
> >> + mii_mgr_write(priv, 0, 28, 0x0233);
> >> + mii_mgr_write(priv, 0, 29, 0x000a);
> >> + mii_mgr_write(priv, 0, 30, 0x0000);
> >> +
> >> + /* Fix EPHY idle state abnormal behavior */
> >> + mii_mgr_write(priv, 0, 31, 0x4000); /* change G4 page */
> >> + mii_mgr_write(priv, 0, 29, 0x000d);
> >> + mii_mgr_write(priv, 0, 30, 0x0500);
> >> +}
> >> +
> >> +static void rt305x_esw_init(struct mt76xx_eth_dev *priv)
> >> +{
> >> + void __iomem *base = priv->eth_sw_base;
> >> +
> >> + /*
> >> + * FC_RLS_TH=200, FC_SET_TH=160
> >> + * DROP_RLS=120, DROP_SET_TH=80
> >> + */
> >
> > Man, this driver is full of magic.
>
> Yes, sorry for that. But again, compared to the original driver
> its a big improvement, IMHO. Feel free to take a quick glance:
>
> https://github.com/MediaTek-Labs/linkit-smart-7688-uboot/blob/master/drivers/rt2880_eth.c
Yes, it is certainly a big improvement.
>
> >> + writel(0xc8a07850, base + MT76XX_SWITCH_FCT0);
> >> + writel(0x00000000, base + MT76XX_SWITCH_SGC2);
> >> + writel(0x00405555, base + MT76XX_SWITCH_PFC1);
> >> + writel(0x00007f7f, base + MT76XX_SWITCH_POC0);
> >> + writel(0x00007f7f, base + MT76XX_SWITCH_POC2); /* disable VLAN */
> >> + writel(0x0002500c, base + MT76XX_SWITCH_FCT2);
> >> + /* hashing algorithm=XOR48, aging interval=300sec */
> >> + writel(0x0008a301, base + MT76XX_SWITCH_SGC);
> >> + writel(0x02404040, base + MT76XX_SWITCH_SOCPC);
> >> +
> >> + /* Ext PHY Addr=0x1f */
> >> + writel(0x3f502b28, base + MT76XX_SWITCH_FPA1);
> >> + writel(0x00000000, base + MT76XX_SWITCH_FPA);
> >> + /* 1us cycle number=125 (FE's clock=125Mhz) */
> >> + writel(0x7d000000, base + MT76XX_SWITCH_BMU_CTRL);
> >> +
> >> + /* Configure analog GPIO setup */
> >> + regmap_update_bits(priv->sysctrl_regmap, MT76XX_AGPIO_CFG_REG,
> >> + MT7628_EPHY_P0_DIS, MT7628_EPHY_GPIO_AIO_EN);
> >> +
> >> + /* Reset PHY */
> >> + regmap_update_bits(priv->sysctrl_regmap, MT76XX_RSTCTRL_REG,
> >> + 0, RSTCTRL_EPHY_RST);
> >> + regmap_update_bits(priv->sysctrl_regmap, MT76XX_RSTCTRL_REG,
> >> + RSTCTRL_EPHY_RST, 0);
> >> + mdelay(10);
> >> +
> >> + /* Set P0 EPHY LED mode */
> >> + regmap_update_bits(priv->sysctrl_regmap, MT76XX_GPIO2_MODE_REG,
> >> + 0x0ffc0ffc, 0x05540554);
> >> + mdelay(10);
> >> +
> >> + mt7628_ephy_init(priv);
> >> +}
> >> +
> >> +static void eth_dma_start(struct mt76xx_eth_dev *priv)
> >> +{
> >> + void __iomem *base = priv->base;
> >> +
> >> + setbits_le32(base + PDMA_GLO_CFG, TX_WB_DDONE | RX_DMA_EN | TX_DMA_EN);
> >> +}
> >> +
> >> +static void eth_dma_stop(struct mt76xx_eth_dev *priv)
> >> +{
> >> + void __iomem *base = priv->base;
> >> + int ret;
> >> +
> >> + clrbits_le32(base + PDMA_GLO_CFG, TX_WB_DDONE | RX_DMA_EN | TX_DMA_EN);
> >> +
> >> + /* Wait for DMA to stop */
> >> + ret = wait_for_bit_le32(base + PDMA_GLO_CFG,
> >> + RX_DMA_BUSY | TX_DMA_BUSY, false,
> >> + CONFIG_DMA_STOP_TIMEOUT, false);
> >> + if (ret)
> >> + printf("DMA stop timeout error!\n");
> >> +}
> >> +
> >> +static int mt76xx_eth_write_hwaddr(struct udevice *dev)
> >> +{
> >> + struct mt76xx_eth_dev *priv = dev_get_priv(dev);
> >> + void __iomem *base = priv->base;
> >> + u8 *addr = ((struct eth_pdata *)dev_get_platdata(dev))->enetaddr;
> >> + u32 val;
> >> + u16 tmp;
> >> +
> >> + /* Set MAC address. */
> >> + tmp = (u16)addr[0];
> >
> > Is the cast on these lines actually necessary?
>
> I'll check and will update if possible.
>
> >> + val = (tmp << 8) | addr[1];
> >> +
> >> + writel(val, base + SDM_MAC_ADRH);
> >> +
> >> + tmp = (u16)addr[2];
> >> + val = (tmp << 8) | addr[3];
> >> + val = val << 16;
> >> + tmp = (u16)addr[4];
> >> + val |= (tmp << 8) | addr[5];
> >> + writel(val, base + SDM_MAC_ADRL);
> >> +
> >> + return 0;
> >> +}
> >> +
> >> +static int mt76xx_eth_send(struct udevice *dev, void *packet, int length)
> >> +{
> >> + struct mt76xx_eth_dev *priv = dev_get_priv(dev);
> >> + void __iomem *base = priv->base;
> >> + int ret;
> >> + int idx;
> >> + int i;
> >> +
> >> + idx = priv->tx_dma_idx;
> >> +
> >> + /* Pad message to a minimum length */
> >> + if (length < PADDING_LENGTH) {
> >> + char *p = (char *)packet;
> >> +
> >> + for (i = 0; i < PADDING_LENGTH - length; i++)
> >> + p[length + i] = 0;
> >> + length = PADDING_LENGTH;
> >> + }
> >> +
> >> + /* Check if buffer is ready for next TX DMA */
> >> + ret = wait_for_bit_le32(&priv->tx_ring[idx].txd2, TX_DMA_DONE, true,
> >> + CONFIG_TX_DMA_TIMEOUT, false);
> >> + if (ret) {
> >> + printf("TX: DMA still busy on buffer %d\n", idx);
> >> + return ret;
> >> + }
> >> +
> >> + flush_dcache_range((u32)packet, (u32)packet + length);
> >> +
> >> + priv->tx_ring[idx].txd1 = CPHYSADDR(packet);
> >> + priv->tx_ring[idx].txd2 |= TX_DMA_PLEN0(length);
> >> + priv->tx_ring[idx].txd2 &= ~TX_DMA_DONE;
> >> +
> >> + idx = (idx + 1) % NUM_TX_DESC;
> >> +
> >> + /* Make sure the writes exceuted at this place */
> >
> > exceuted -> executed
>
> Thx.
>
> >> + wmb();
> >> + writel(idx, base + TX_CTX_IDX0);
> >> +
> >> + priv->tx_dma_idx = idx;
> >> +
> >> + return 0;
> >> +}
> >> +
> >> +static int mt76xx_eth_recv(struct udevice *dev, int flags, uchar **packetp)
> >> +{
> >> + struct mt76xx_eth_dev *priv = dev_get_priv(dev);
> >> + void __iomem *base = priv->base;
> >> + u32 rxd_info;
> >> + int length;
> >> + int idx;
> >> +
> >> + idx = priv->rx_dma_idx;
> >> +
> >> + rxd_info = priv->rx_ring[idx].rxd2;
> >> + if ((rxd_info & RX_DMA_DONE) == 0)
> >> + return -EAGAIN;
> >> +
> >> + length = RX_DMA_GET_PLEN0(priv->rx_ring[idx].rxd2);
> >> + if (length == 0 || length > MTK_QDMA_PAGE_SIZE) {
> >> + printf("%s: invalid length (%d bytes)\n", __func__, length);
> >> + return -EIO;
> >> + }
> >> +
> >> + *packetp = priv->rx_buf[idx];
> >> + invalidate_dcache_range((u32)*packetp, (u32)*packetp + length);
> >> +
> >> + priv->rx_ring[idx].rxd4 = 0;
> >> + priv->rx_ring[idx].rxd2 = RX_DMA_LSO;
> >> +
> >> + /* Make sure the writes exceuted at this place */
> >
> > exceuted -> executed
>
> Thx.
>
> >> + wmb();
> >> +
> >> + /* Move point to next RXD which wants to alloc */
> >> + writel(idx, base + RX_CALC_IDX0);
> >> +
> >> + /* Update to Next packet point that was received */
> >> + idx = (idx + 1) % NUM_RX_DESC;
> >> +
> >> + priv->rx_dma_idx = idx;
> >> +
> >> + return length;
> >> +}
> >> +
> >> +static int phy_link_up(struct mt76xx_eth_dev *priv)
> >> +{
> >> + u32 val;
> >> +
> >> + mii_mgr_read(priv, 0x00, MII_BMSR, &val);
> >> + return !!(val & BMSR_LSTATUS);
> >> +}
> >> +
> >> +static int mt76xx_eth_start(struct udevice *dev)
> >> +{
> >> + struct mt76xx_eth_dev *priv = dev_get_priv(dev);
> >> + void __iomem *base = priv->base;
> >> + uchar packet[MTK_QDMA_PAGE_SIZE];
> >> + uchar *packetp;
> >> + int i;
> >> +
> >> + for (i = 0; i < NUM_RX_DESC; i++) {
> >> + memset((void *)&priv->rx_ring[i], 0, sizeof(priv->rx_ring[0]));
> >> + priv->rx_ring[i].rxd2 |= RX_DMA_LSO;
> >> + priv->rx_ring[i].rxd1 = CPHYSADDR(priv->rx_buf[i]);
> >> + }
> >> +
> >> + for (i = 0; i < NUM_TX_DESC; i++) {
> >> + memset((void *)&priv->tx_ring[i], 0, sizeof(priv->tx_ring[0]));
> >> + priv->tx_ring[i].txd2 = TX_DMA_LS0 | TX_DMA_DONE;
> >> + priv->tx_ring[i].txd4 = TX_DMA_PN(1);
> >> + }
> >> +
> >> + priv->rx_dma_idx = 0;
> >> + priv->tx_dma_idx = 0;
> >> +
> >> + /* Make sure the writes exceuted at this place */
> >
> > exceuted -> executed
> >
> >> + wmb();
> >> +
> >> + /* disable delay interrupt */
> >> + writel(0, base + DLY_INT_CFG);
> >> +
> >> + clrbits_le32(base + PDMA_GLO_CFG, 0xffff0000);
> >> +
> >> + /* Tell the adapter where the TX/RX rings are located. */
> >> + writel(CPHYSADDR(&priv->rx_ring[0]), base + RX_BASE_PTR0);
> >> + writel(CPHYSADDR((u32)&priv->tx_ring[0]), base + TX_BASE_PTR0);
> >> +
> >> + writel(NUM_RX_DESC, base + RX_MAX_CNT0);
> >> + writel(NUM_TX_DESC, base + TX_MAX_CNT0);
> >> +
> >> + writel(priv->tx_dma_idx, base + TX_CTX_IDX0);
> >> + writel(RST_DTX_IDX0, base + PDMA_RST_IDX);
> >> +
> >> + writel(NUM_RX_DESC - 1, base + RX_CALC_IDX0);
> >> + writel(RST_DRX_IDX0, base + PDMA_RST_IDX);
> >> +
> >> + /* Make sure the writes exceuted at this place */
> >
> > exceuted -> executed
> >
> >> + wmb();
> >> + eth_dma_start(priv);
> >> +
> >> + /* Check if link is not up yet */
> >> + if (!phy_link_up(priv)) {
> >> + /* Wait for link to come up */
> >> +
> >> + printf("Waiting for link to come up .");
> >> + for (i = 0; i < (LINK_TIMEOUT / LINK_DELAY_TIME); i++) {
> >> + mdelay(LINK_DELAY_TIME);
> >> + if (phy_link_up(priv)) {
> >> + mdelay(100); /* Ensure all is ready */
> >> + break;
> >> + }
> >> +
> >> + printf(".");
> >> + }
> >> +
> >> + if (phy_link_up(priv))
> >> + printf(" done\n");
> >> + else
> >> + printf(" timeout! Trying anyways\n");
> >> + }
> >> +
> >> + /*
> >> + * The integrated switch seems to queue some received ethernet
> >
> > Ethernet
>
> Really? "Ethernet" captilized and "switch" not?
Ethernet is a proper name, switch is not.
> >> + * packets in some FIFO. Lets read the already queued packets
> >> + * out by using the reveice routine, so that these old messages
> >
> > reveice -> receive
> >
> >> + * are dropped before the new xfer starts.
> >> + */
> >> + packetp = &packet[0];
> >> + while (mt76xx_eth_recv(dev, 0, &packetp) != -EAGAIN)
> >> + ;
> >> +
> >> + return 0;
> >> +}
> >> +
> >> +static void mt76xx_eth_stop(struct udevice *dev)
> >> +{
> >> + struct mt76xx_eth_dev *priv = dev_get_priv(dev);
> >> +
> >> + eth_dma_stop(priv);
> >> +}
> >> +
> >> +static int mt76xx_eth_probe(struct udevice *dev)
> >> +{
> >> + struct mt76xx_eth_dev *priv = dev_get_priv(dev);
> >> + struct udevice *syscon;
> >> + struct mii_dev *bus;
> >> + int ret;
> >> + int i;
> >> +
> >> + /* Save frame-engine base address for later use */
> >> + priv->base = dev_remap_addr_index(dev, 0);
> >> + if (IS_ERR(priv->base))
> >> + return PTR_ERR(priv->base);
> >> +
> >> + /* Save switch base address for later use */
> >> + priv->eth_sw_base = dev_remap_addr_index(dev, 1);
> >> + if (IS_ERR(priv->eth_sw_base))
> >> + return PTR_ERR(priv->eth_sw_base);
> >> +
> >> + /* Get system controller regmap */
> >> + ret = uclass_get_device_by_phandle(UCLASS_SYSCON, dev,
> >> + "syscon", &syscon);
> >> + if (ret) {
> >> + pr_err("unable to find syscon device\n");
> >> + return ret;
> >> + }
> >> +
> >> + priv->sysctrl_regmap = syscon_get_regmap(syscon);
> >> + if (!priv->sysctrl_regmap) {
> >> + pr_err("unable to find regmap\n");
> >> + return -ENODEV;
> >> + }
> >> +
> >> + /* Put rx and tx rings into KSEG1 area (uncached) */
> >> + priv->tx_ring = (struct fe_tx_dma *)
> >> + KSEG1ADDR(memalign(ARCH_DMA_MINALIGN,
> >> + sizeof(*priv->tx_ring) * NUM_TX_DESC));
> >> + priv->rx_ring = (struct fe_rx_dma *)
> >> + KSEG1ADDR(memalign(ARCH_DMA_MINALIGN,
> >> + sizeof(*priv->rx_ring) * NUM_RX_DESC));
> >> +
> >> + for (i = 0; i < NUM_RX_DESC; i++)
> >> + priv->rx_buf[i] = memalign(PKTALIGN, MTK_QDMA_PAGE_SIZE);
> >> +
> >> + bus = mdio_alloc();
> >> + if (!bus) {
> >> + printf("Failed to allocate MDIO bus\n");
> >> + return -ENOMEM;
> >> + }
> >> +
> >> + bus->read = mt76xx_mdio_read;
> >> + bus->write = mt76xx_mdio_write;
> >> + snprintf(bus->name, sizeof(bus->name), dev->name);
> >> + bus->priv = (void *)priv;
> >> +
> >> + ret = mdio_register(bus);
> >> + if (ret)
> >> + return ret;
> >> +
> >> + /* Switch configuration */
> >> + rt305x_esw_init(priv);
> >> +
> >> + return 0;
> >> +}
> >> +
> >> +static const struct eth_ops mt76xx_eth_ops = {
> >> + .start = mt76xx_eth_start,
> >> + .send = mt76xx_eth_send,
> >> + .recv = mt76xx_eth_recv,
> >> + .stop = mt76xx_eth_stop,
> >> + .write_hwaddr = mt76xx_eth_write_hwaddr,
> >> +};
> >> +
> >> +static const struct udevice_id mt76xx_eth_ids[] = {
> >> + { .compatible = "mediatek,mt7622-eth" },
> >
> > Please add a file under doc/device-tree-bindings/net/ for this adapter.
>
> Okay, will do.
>
> Thanks for the review!
Of course, thanks for all the work.
> Thanks,
> Stefan
> _______________________________________________
> U-Boot mailing list
> U-Boot at lists.denx.de
> https://lists.denx.de/listinfo/u-boot
More information about the U-Boot
mailing list