| Message ID | 20210806054904.534315-3-joel@jms.id.au (mailing list archive) |
|---|---|
| State | Changes Requested |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | net: Add LiteETH network driver | expand |
Hi Joel, Thanks for pushing this upstream (and for writing it to begin with)! Would you mind diff-ing your version of litex_liteeth.c against what is currently in https://github.com/litex-hub/linux/blob/litex-rebase/drivers/net/ethernet/litex/litex_liteeth.c ? Two main differences we should discuss: 1. there's a polling mode (added by Antony Pavlov), and if we decide *not* to keep it around, I want to ensure we do that deliberately, with an explanation as to why; 2. LiteX CSRs are accessed using `litex_[read|write][8|16|32]()` as opposed to simply `[read|write][b|w|l]()`. The former set are defined in `include/linux/litex.h` and are needed to ensure correct accesses regardless of endianness, since by default LiteX registers' endianness mirrors that of the configured CPU. Thanks much, --Gabriel On Fri, Aug 06, 2021 at 03:19:04PM +0930, Joel Stanley wrote: > LiteX is a soft system-on-chip that targets FPGAs. LiteETH is a basic > network device that is commonly used in LiteX designs. > > The driver was first written in 2017 and has been maintained by the > LiteX community in various trees. Thank you to all who have contributed. > > Co-developed-by: Gabriel Somlo <gsomlo@gmail.com> > Co-developed-by: David Shah <dave@ds0.me> > Co-developed-by: Stafford Horne <shorne@gmail.com> > Signed-off-by: Joel Stanley <joel@jms.id.au> > --- > drivers/net/ethernet/Kconfig | 1 + > drivers/net/ethernet/Makefile | 1 + > drivers/net/ethernet/litex/Kconfig | 24 ++ > drivers/net/ethernet/litex/Makefile | 5 + > drivers/net/ethernet/litex/litex_liteeth.c | 340 +++++++++++++++++++++ > 5 files changed, 371 insertions(+) > create mode 100644 drivers/net/ethernet/litex/Kconfig > create mode 100644 drivers/net/ethernet/litex/Makefile > create mode 100644 drivers/net/ethernet/litex/litex_liteeth.c > > diff --git a/drivers/net/ethernet/Kconfig b/drivers/net/ethernet/Kconfig > index 1cdff1dca790..d796684ec9ca 100644 > --- a/drivers/net/ethernet/Kconfig > +++ b/drivers/net/ethernet/Kconfig > @@ -118,6 +118,7 @@ config LANTIQ_XRX200 > Support for the PMAC of the Gigabit switch (GSWIP) inside the > Lantiq / Intel VRX200 VDSL SoC > > +source "drivers/net/ethernet/litex/Kconfig" > source "drivers/net/ethernet/marvell/Kconfig" > source "drivers/net/ethernet/mediatek/Kconfig" > source "drivers/net/ethernet/mellanox/Kconfig" > diff --git a/drivers/net/ethernet/Makefile b/drivers/net/ethernet/Makefile > index cb3f9084a21b..aaa5078cd7d1 100644 > --- a/drivers/net/ethernet/Makefile > +++ b/drivers/net/ethernet/Makefile > @@ -51,6 +51,7 @@ obj-$(CONFIG_JME) += jme.o > obj-$(CONFIG_KORINA) += korina.o > obj-$(CONFIG_LANTIQ_ETOP) += lantiq_etop.o > obj-$(CONFIG_LANTIQ_XRX200) += lantiq_xrx200.o > +obj-$(CONFIG_NET_VENDOR_LITEX) += litex/ > obj-$(CONFIG_NET_VENDOR_MARVELL) += marvell/ > obj-$(CONFIG_NET_VENDOR_MEDIATEK) += mediatek/ > obj-$(CONFIG_NET_VENDOR_MELLANOX) += mellanox/ > diff --git a/drivers/net/ethernet/litex/Kconfig b/drivers/net/ethernet/litex/Kconfig > new file mode 100644 > index 000000000000..0a863e3ff7bf > --- /dev/null > +++ b/drivers/net/ethernet/litex/Kconfig > @@ -0,0 +1,24 @@ > +# > +# LiteX device configuration > +# > + > +config NET_VENDOR_LITEX > + bool "LiteX devices" > + default y > + help > + If you have a network (Ethernet) card belonging to this class, say Y. > + > + Note that the answer to this question doesn't directly affect the > + kernel: saying N will just cause the configurator to skip all > + the questions about LiteX devices. If you say Y, you will be asked > + for your specific card in the following questions. > + > +if NET_VENDOR_LITEX > + > +config LITEX_LITEETH > + tristate "LiteX Ethernet support" > + help > + If you wish to compile a kernel for hardware with a LiteX LiteEth > + device then you should answer Y to this. > + > +endif # NET_VENDOR_LITEX > diff --git a/drivers/net/ethernet/litex/Makefile b/drivers/net/ethernet/litex/Makefile > new file mode 100644 > index 000000000000..9343b73b8e49 > --- /dev/null > +++ b/drivers/net/ethernet/litex/Makefile > @@ -0,0 +1,5 @@ > +# > +# Makefile for the LiteX network device drivers. > +# > + > +obj-$(CONFIG_LITEX_LITEETH) += litex_liteeth.o > diff --git a/drivers/net/ethernet/litex/litex_liteeth.c b/drivers/net/ethernet/litex/litex_liteeth.c > new file mode 100644 > index 000000000000..48b551e6d97b > --- /dev/null > +++ b/drivers/net/ethernet/litex/litex_liteeth.c > @@ -0,0 +1,340 @@ > +// SPDX-License-Identifier: GPL-2.0-or-later > +/* > + * LiteX Liteeth Ethernet > + * > + * Copyright 2017 Joel Stanley <joel@jms.id.au> > + * > + */ > + > +#include <linux/etherdevice.h> > +#include <linux/interrupt.h> > +#include <linux/module.h> > +#include <linux/of.h> > +#include <linux/of_net.h> > +#include <linux/of_address.h> > +#include <linux/phy.h> > +#include <linux/platform_device.h> > +#include <linux/iopoll.h> > + > +#define LITEETH_WRITER_SLOT 0x00 > +#define LITEETH_WRITER_LENGTH 0x04 > +#define LITEETH_WRITER_ERRORS 0x08 > +#define LITEETH_WRITER_EV_STATUS 0x0C > +#define LITEETH_WRITER_EV_PENDING 0x10 > +#define LITEETH_WRITER_EV_ENABLE 0x14 > +#define LITEETH_READER_START 0x18 > +#define LITEETH_READER_READY 0x1C > +#define LITEETH_READER_LEVEL 0x20 > +#define LITEETH_READER_SLOT 0x24 > +#define LITEETH_READER_LENGTH 0x28 > +#define LITEETH_READER_EV_STATUS 0x2C > +#define LITEETH_READER_EV_PENDING 0x30 > +#define LITEETH_READER_EV_ENABLE 0x34 > +#define LITEETH_PREAMBLE_CRC 0x38 > +#define LITEETH_PREAMBLE_ERRORS 0x3C > +#define LITEETH_CRC_ERRORS 0x40 > + > +#define LITEETH_PHY_CRG_RESET 0x00 > +#define LITEETH_MDIO_W 0x04 > +#define LITEETH_MDIO_R 0x0C > + > +#define DRV_NAME "liteeth" > + > +#define LITEETH_BUFFER_SIZE 0x800 > +#define MAX_PKT_SIZE LITEETH_BUFFER_SIZE > + > +struct liteeth { > + void __iomem *base; > + void __iomem *mdio_base; > + struct net_device *netdev; > + struct device *dev; > + struct mii_bus *mii_bus; > + > + /* Link management */ > + int cur_duplex; > + int cur_speed; > + > + /* Tx */ > + int tx_slot; > + int num_tx_slots; > + void __iomem *tx_base; > + > + /* Rx */ > + int rx_slot; > + int num_rx_slots; > + void __iomem *rx_base; > +}; > + > + > +static int liteeth_rx(struct net_device *netdev) > +{ > + struct liteeth *priv = netdev_priv(netdev); > + struct sk_buff *skb; > + unsigned char *data; > + u8 rx_slot; > + int len; > + > + rx_slot = readb(priv->base + LITEETH_WRITER_SLOT); > + len = readl(priv->base + LITEETH_WRITER_LENGTH); > + > + skb = netdev_alloc_skb(netdev, len + NET_IP_ALIGN); > + if (!skb) { > + netdev_err(netdev, "couldn't get memory"); > + netdev->stats.rx_dropped++; > + return NET_RX_DROP; > + } > + > + /* Ensure alignemnt of the ip header within the skb */ > + skb_reserve(skb, NET_IP_ALIGN); > + if (len == 0 || len > 2048) > + return NET_RX_DROP; > + data = skb_put(skb, len); > + memcpy_fromio(data, priv->rx_base + rx_slot * LITEETH_BUFFER_SIZE, len); > + skb->protocol = eth_type_trans(skb, netdev); > + > + netdev->stats.rx_packets++; > + netdev->stats.rx_bytes += len; > + > + return netif_rx(skb); > +} > + > +static irqreturn_t liteeth_interrupt(int irq, void *dev_id) > +{ > + struct net_device *netdev = dev_id; > + struct liteeth *priv = netdev_priv(netdev); > + u8 reg; > + > + reg = readb(priv->base + LITEETH_READER_EV_PENDING); > + if (reg) { > + netdev->stats.tx_packets++; > + writeb(reg, priv->base + LITEETH_READER_EV_PENDING); > + } > + > + reg = readb(priv->base + LITEETH_WRITER_EV_PENDING); > + if (reg) { > + liteeth_rx(netdev); > + writeb(reg, priv->base + LITEETH_WRITER_EV_PENDING); > + } > + > + return IRQ_HANDLED; > +} > + > +static int liteeth_open(struct net_device *netdev) > +{ > + struct liteeth *priv = netdev_priv(netdev); > + int err; > + > + /* Clear pending events */ > + writeb(1, priv->base + LITEETH_WRITER_EV_PENDING); > + writeb(1, priv->base + LITEETH_READER_EV_PENDING); > + > + err = request_irq(netdev->irq, liteeth_interrupt, 0, netdev->name, netdev); > + if (err) { > + netdev_err(netdev, "failed to request irq %d\n", netdev->irq); > + return err; > + } > + > + /* Enable IRQs */ > + writeb(1, priv->base + LITEETH_WRITER_EV_ENABLE); > + writeb(1, priv->base + LITEETH_READER_EV_ENABLE); > + > + /* TODO: Remove these once we have working mdio support */ > + priv->cur_duplex = DUPLEX_FULL; > + priv->cur_speed = SPEED_100; > + netif_carrier_on(netdev); > + > + netif_start_queue(netdev); > + > + return 0; > +} > + > +static int liteeth_stop(struct net_device *netdev) > +{ > + struct liteeth *priv = netdev_priv(netdev); > + > + netif_stop_queue(netdev); > + > + writeb(0, priv->base + LITEETH_WRITER_EV_ENABLE); > + writeb(0, priv->base + LITEETH_READER_EV_ENABLE); > + > + free_irq(netdev->irq, netdev); > + > + return 0; > +} > + > +static int liteeth_start_xmit(struct sk_buff *skb, struct net_device *netdev) > +{ > + struct liteeth *priv = netdev_priv(netdev); > + void __iomem *txbuffer; > + int ret; > + u8 val; > + > + /* Reject oversize packets */ > + if (unlikely(skb->len > MAX_PKT_SIZE)) { > + if (net_ratelimit()) > + netdev_dbg(netdev, "tx packet too big\n"); > + goto drop; > + } > + > + txbuffer = priv->tx_base + priv->tx_slot * LITEETH_BUFFER_SIZE; > + memcpy_toio(txbuffer, skb->data, skb->len); > + writeb(priv->tx_slot, priv->base + LITEETH_READER_SLOT); > + writew(skb->len, priv->base + LITEETH_READER_LENGTH); > + > + ret = readl_poll_timeout_atomic(priv->base + LITEETH_READER_READY, val, val, 5, 1000); > + if (ret == -ETIMEDOUT) { > + netdev_err(netdev, "LITEETH_READER_READY timed out\n"); > + goto drop; > + } > + > + writeb(1, priv->base + LITEETH_READER_START); > + > + netdev->stats.tx_bytes += skb->len; > + > + priv->tx_slot = (priv->tx_slot + 1) % priv->num_tx_slots; > + dev_kfree_skb_any(skb); > + return NETDEV_TX_OK; > +drop: > + /* Drop the packet */ > + dev_kfree_skb_any(skb); > + netdev->stats.tx_dropped++; > + > + return NETDEV_TX_OK; > +} > + > +static const struct net_device_ops liteeth_netdev_ops = { > + .ndo_open = liteeth_open, > + .ndo_stop = liteeth_stop, > + .ndo_start_xmit = liteeth_start_xmit, > +}; > + > +static void liteeth_reset_hw(struct liteeth *priv) > +{ > + /* Reset, twice */ > + writeb(0, priv->base + LITEETH_PHY_CRG_RESET); > + udelay(10); > + writeb(1, priv->base + LITEETH_PHY_CRG_RESET); > + udelay(10); > + writeb(0, priv->base + LITEETH_PHY_CRG_RESET); > + udelay(10); > +} > + > +static int liteeth_probe(struct platform_device *pdev) > +{ > + struct net_device *netdev; > + void __iomem *buf_base; > + struct resource *res; > + struct liteeth *priv; > + int irq, err; > + > + netdev = alloc_etherdev(sizeof(*priv)); > + if (!netdev) > + return -ENOMEM; > + > + priv = netdev_priv(netdev); > + priv->netdev = netdev; > + priv->dev = &pdev->dev; > + > + irq = platform_get_irq(pdev, 0); > + if (irq < 0) { > + dev_err(&pdev->dev, "Failed to get IRQ\n"); > + goto err; > + } > + > + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); > + priv->base = devm_ioremap_resource(&pdev->dev, res); > + if (IS_ERR(priv->base)) { > + err = PTR_ERR(priv->base); > + goto err; > + } > + > + res = platform_get_resource(pdev, IORESOURCE_MEM, 1); > + priv->mdio_base = devm_ioremap_resource(&pdev->dev, res); > + if (IS_ERR(priv->mdio_base)) { > + err = PTR_ERR(priv->mdio_base); > + goto err; > + } > + > + res = platform_get_resource(pdev, IORESOURCE_MEM, 2); > + buf_base = devm_ioremap_resource(&pdev->dev, res); > + if (IS_ERR(buf_base)) { > + err = PTR_ERR(buf_base); > + goto err; > + } > + > + err = of_property_read_u32(pdev->dev.of_node, "rx-fifo-depth", > + &priv->num_rx_slots); > + if (err) { > + dev_err(&pdev->dev, "unable to get rx-fifo-depth\n"); > + goto err; > + } > + > + err = of_property_read_u32(pdev->dev.of_node, "tx-fifo-depth", > + &priv->num_tx_slots); > + if (err) { > + dev_err(&pdev->dev, "unable to get tx-fifo-depth\n"); > + goto err; > + } > + > + /* Rx slots */ > + priv->rx_base = buf_base; > + priv->rx_slot = 0; > + > + /* Tx slots come after Rx slots */ > + priv->tx_base = buf_base + priv->num_rx_slots * LITEETH_BUFFER_SIZE; > + priv->tx_slot = 0; > + > + err = of_get_mac_address(pdev->dev.of_node, netdev->dev_addr); > + if (err) > + eth_hw_addr_random(netdev); > + > + SET_NETDEV_DEV(netdev, &pdev->dev); > + platform_set_drvdata(pdev, netdev); > + > + netdev->netdev_ops = &liteeth_netdev_ops; > + netdev->irq = irq; > + > + liteeth_reset_hw(priv); > + > + err = register_netdev(netdev); > + if (err) { > + dev_err(&pdev->dev, "Failed to register netdev\n"); > + goto err; > + } > + > + netdev_info(netdev, "irq %d, mapped at %px\n", netdev->irq, priv->base); > + > + return 0; > +err: > + free_netdev(netdev); > + return err; > +} > + > +static int liteeth_remove(struct platform_device *pdev) > +{ > + struct net_device *netdev = platform_get_drvdata(pdev); > + > + unregister_netdev(netdev); > + free_netdev(netdev); > + > + return 0; > +} > + > +static const struct of_device_id liteeth_of_match[] = { > + { .compatible = "litex,liteeth" }, > + { } > +}; > +MODULE_DEVICE_TABLE(of, liteeth_of_match); > + > +static struct platform_driver liteeth_driver = { > + .probe = liteeth_probe, > + .remove = liteeth_remove, > + .driver = { > + .name = DRV_NAME, > + .of_match_table = liteeth_of_match, > + }, > +}; > +module_platform_driver(liteeth_driver); > + > +MODULE_AUTHOR("Joel Stanley <joel@jms.id.au>"); > +MODULE_LICENSE("GPL"); > -- > 2.32.0 >
On Fri, 6 Aug 2021 15:19:04 +0930 Joel Stanley wrote: > LiteX is a soft system-on-chip that targets FPGAs. LiteETH is a basic > network device that is commonly used in LiteX designs. > > The driver was first written in 2017 and has been maintained by the > LiteX community in various trees. Thank you to all who have contributed. > +config NET_VENDOR_LITEX > + bool "LiteX devices" > + default y > + help > + If you have a network (Ethernet) card belonging to this class, say Y. > + > + Note that the answer to this question doesn't directly affect the > + kernel: saying N will just cause the configurator to skip all > + the questions about LiteX devices. If you say Y, you will be asked > + for your specific card in the following questions. Maybe mention where the device is usually found (FPGAs) like you did in the commit message, to help folks make a decision here? > +config LITEX_LITEETH > + tristate "LiteX Ethernet support" > + help > + If you wish to compile a kernel for hardware with a LiteX LiteEth > + device then you should answer Y to this. > +struct liteeth { > + void __iomem *base; > + void __iomem *mdio_base; > + struct net_device *netdev; > + struct device *dev; > + struct mii_bus *mii_bus; unused field > + > + /* Link management */ > + int cur_duplex; > + int cur_speed; > + > + /* Tx */ > + int tx_slot; > + int num_tx_slots; > + void __iomem *tx_base; > + > + /* Rx */ > + int rx_slot; > + int num_rx_slots; > + void __iomem *rx_base; > +}; > + > + > +static int liteeth_rx(struct net_device *netdev) > +{ > + struct liteeth *priv = netdev_priv(netdev); > + struct sk_buff *skb; > + unsigned char *data; > + u8 rx_slot; > + int len; > + > + rx_slot = readb(priv->base + LITEETH_WRITER_SLOT); > + len = readl(priv->base + LITEETH_WRITER_LENGTH); > + > + skb = netdev_alloc_skb(netdev, len + NET_IP_ALIGN); netdev_alloc_skb_ip_align() ... > + if (!skb) { > + netdev_err(netdev, "couldn't get memory"); \n at the end? You can skip it but be consistent across messages > + netdev->stats.rx_dropped++; > + return NET_RX_DROP; > + } > + > + /* Ensure alignemnt of the ip header within the skb */ > + skb_reserve(skb, NET_IP_ALIGN); ... then skip this > + if (len == 0 || len > 2048) > + return NET_RX_DROP; Should this be counted somehow? > + data = skb_put(skb, len); > + memcpy_fromio(data, priv->rx_base + rx_slot * LITEETH_BUFFER_SIZE, len); > + skb->protocol = eth_type_trans(skb, netdev); > + > + netdev->stats.rx_packets++; > + netdev->stats.rx_bytes += len; > + > + return netif_rx(skb); > +} > + > +static irqreturn_t liteeth_interrupt(int irq, void *dev_id) > +{ > + struct net_device *netdev = dev_id; > + struct liteeth *priv = netdev_priv(netdev); > + u8 reg; > + > + reg = readb(priv->base + LITEETH_READER_EV_PENDING); > + if (reg) { > + netdev->stats.tx_packets++; > + writeb(reg, priv->base + LITEETH_READER_EV_PENDING); > + } > + > + reg = readb(priv->base + LITEETH_WRITER_EV_PENDING); > + if (reg) { > + liteeth_rx(netdev); > + writeb(reg, priv->base + LITEETH_WRITER_EV_PENDING); > + } > + > + return IRQ_HANDLED; > +} > + > +static int liteeth_open(struct net_device *netdev) > +{ > + struct liteeth *priv = netdev_priv(netdev); > + int err; > + > + /* Clear pending events */ > + writeb(1, priv->base + LITEETH_WRITER_EV_PENDING); > + writeb(1, priv->base + LITEETH_READER_EV_PENDING); > + > + err = request_irq(netdev->irq, liteeth_interrupt, 0, netdev->name, netdev); > + if (err) { > + netdev_err(netdev, "failed to request irq %d\n", netdev->irq); > + return err; > + } > + > + /* Enable IRQs */ > + writeb(1, priv->base + LITEETH_WRITER_EV_ENABLE); > + writeb(1, priv->base + LITEETH_READER_EV_ENABLE); > + > + /* TODO: Remove these once we have working mdio support */ > + priv->cur_duplex = DUPLEX_FULL; > + priv->cur_speed = SPEED_100; please remove the fields until they're actually used > + netif_carrier_on(netdev); > + > + netif_start_queue(netdev); > + > + return 0; > +} > + > +static int liteeth_stop(struct net_device *netdev) > +{ > + struct liteeth *priv = netdev_priv(netdev); carrier_off() for symmetry? > + netif_stop_queue(netdev); > + > + writeb(0, priv->base + LITEETH_WRITER_EV_ENABLE); > + writeb(0, priv->base + LITEETH_READER_EV_ENABLE); > + > + free_irq(netdev->irq, netdev); > + return 0; > +} > + > +static int liteeth_start_xmit(struct sk_buff *skb, struct net_device *netdev) > +{ > + struct liteeth *priv = netdev_priv(netdev); > + void __iomem *txbuffer; > + int ret; > + u8 val; > + > + /* Reject oversize packets */ > + if (unlikely(skb->len > MAX_PKT_SIZE)) { > + if (net_ratelimit()) > + netdev_dbg(netdev, "tx packet too big\n"); > + goto drop; > + } > + > + txbuffer = priv->tx_base + priv->tx_slot * LITEETH_BUFFER_SIZE; > + memcpy_toio(txbuffer, skb->data, skb->len); > + writeb(priv->tx_slot, priv->base + LITEETH_READER_SLOT); > + writew(skb->len, priv->base + LITEETH_READER_LENGTH); > + > + ret = readl_poll_timeout_atomic(priv->base + LITEETH_READER_READY, val, val, 5, 1000); Why the need for poll if there is an interrupt? Why not stop the Tx queue once you're out of slots and restart it when the completion interrupt comes? > + if (ret == -ETIMEDOUT) { > + netdev_err(netdev, "LITEETH_READER_READY timed out\n"); ratelimit this as well, please > + goto drop; > + } > + > + writeb(1, priv->base + LITEETH_READER_START); > + > + netdev->stats.tx_bytes += skb->len; Please count bytes and packets in the same place > + priv->tx_slot = (priv->tx_slot + 1) % priv->num_tx_slots; > + dev_kfree_skb_any(skb); > + return NETDEV_TX_OK; > +drop: > + /* Drop the packet */ > + dev_kfree_skb_any(skb); > + netdev->stats.tx_dropped++; > + > + return NETDEV_TX_OK; > +} > +static int liteeth_probe(struct platform_device *pdev) > +{ > + struct net_device *netdev; > + void __iomem *buf_base; > + struct resource *res; > + struct liteeth *priv; > + int irq, err; > + > + netdev = alloc_etherdev(sizeof(*priv)); > + if (!netdev) > + return -ENOMEM; > + > + priv = netdev_priv(netdev); > + priv->netdev = netdev; > + priv->dev = &pdev->dev; > + > + irq = platform_get_irq(pdev, 0); > + if (irq < 0) { > + dev_err(&pdev->dev, "Failed to get IRQ\n"); > + goto err; `err` variable is not set here, you'd return 0 > + } > + > + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); > + priv->base = devm_ioremap_resource(&pdev->dev, res); > + if (IS_ERR(priv->base)) { > + err = PTR_ERR(priv->base); > + goto err; > + } > + > + res = platform_get_resource(pdev, IORESOURCE_MEM, 1); > + priv->mdio_base = devm_ioremap_resource(&pdev->dev, res); > + if (IS_ERR(priv->mdio_base)) { > + err = PTR_ERR(priv->mdio_base); > + goto err; > + } > + > + res = platform_get_resource(pdev, IORESOURCE_MEM, 2); > + buf_base = devm_ioremap_resource(&pdev->dev, res); > + if (IS_ERR(buf_base)) { > + err = PTR_ERR(buf_base); > + goto err; > + } > + > + err = of_property_read_u32(pdev->dev.of_node, "rx-fifo-depth", > + &priv->num_rx_slots); Please run checkpatch --strict and fix what it points out > + if (err) { > + dev_err(&pdev->dev, "unable to get rx-fifo-depth\n"); > + goto err; > + } > + return 0; > +err: > + free_netdev(netdev); > + return err; > +}
> +static void liteeth_reset_hw(struct liteeth *priv) > +{ > + /* Reset, twice */ > + writeb(0, priv->base + LITEETH_PHY_CRG_RESET); > + udelay(10); > + writeb(1, priv->base + LITEETH_PHY_CRG_RESET); > + udelay(10); > + writeb(0, priv->base + LITEETH_PHY_CRG_RESET); > + udelay(10); What is this actually resetting? > +static int liteeth_probe(struct platform_device *pdev) > +{ > + struct net_device *netdev; > + void __iomem *buf_base; > + struct resource *res; > + struct liteeth *priv; > + int irq, err; > + > + netdev = alloc_etherdev(sizeof(*priv)); > + if (!netdev) > + return -ENOMEM; > + > + priv = netdev_priv(netdev); > + priv->netdev = netdev; > + priv->dev = &pdev->dev; > + > + irq = platform_get_irq(pdev, 0); > + if (irq < 0) { > + dev_err(&pdev->dev, "Failed to get IRQ\n"); > + goto err; > + } > + > + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); > + priv->base = devm_ioremap_resource(&pdev->dev, res); > + if (IS_ERR(priv->base)) { > + err = PTR_ERR(priv->base); > + goto err; > + } > + > + res = platform_get_resource(pdev, IORESOURCE_MEM, 1); > + priv->mdio_base = devm_ioremap_resource(&pdev->dev, res); > + if (IS_ERR(priv->mdio_base)) { > + err = PTR_ERR(priv->mdio_base); > + goto err; > + } So you don't have any PHY handling, or any MDIO bus master code. So i would drop this, until the MDIO architecture question is answered. I also wonder how much use the MAC driver is without any PHY code? Unless you have a good reason, i don't think we should merge this until it makes the needed calls into phylib. It is not much code to add. Andrew
On Sat, 7 Aug 2021 at 19:19, Andrew Lunn <andrew@lunn.ch> wrote: > > > +static void liteeth_reset_hw(struct liteeth *priv) > > +{ > > + /* Reset, twice */ > > + writeb(0, priv->base + LITEETH_PHY_CRG_RESET); > > + udelay(10); > > + writeb(1, priv->base + LITEETH_PHY_CRG_RESET); > > + udelay(10); > > + writeb(0, priv->base + LITEETH_PHY_CRG_RESET); > > + udelay(10); > > What is this actually resetting? This comes from the reference firmware that many (but not all) litex systems run before loading their operating system. I'm not completely sure how necessary it still is; I will drop it for now. > > > +static int liteeth_probe(struct platform_device *pdev) > > +{ > > + struct net_device *netdev; > > + void __iomem *buf_base; > > + struct resource *res; > > + struct liteeth *priv; > > + int irq, err; > > + > > + netdev = alloc_etherdev(sizeof(*priv)); > > + if (!netdev) > > + return -ENOMEM; > > + > > + priv = netdev_priv(netdev); > > + priv->netdev = netdev; > > + priv->dev = &pdev->dev; > > + > > + irq = platform_get_irq(pdev, 0); > > + if (irq < 0) { > > + dev_err(&pdev->dev, "Failed to get IRQ\n"); > > + goto err; > > + } > > + > > + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); > > + priv->base = devm_ioremap_resource(&pdev->dev, res); > > + if (IS_ERR(priv->base)) { > > + err = PTR_ERR(priv->base); > > + goto err; > > + } > > + > > + res = platform_get_resource(pdev, IORESOURCE_MEM, 1); > > + priv->mdio_base = devm_ioremap_resource(&pdev->dev, res); > > + if (IS_ERR(priv->mdio_base)) { > > + err = PTR_ERR(priv->mdio_base); > > + goto err; > > + } > > So you don't have any PHY handling, or any MDIO bus master code. So i > would drop this, until the MDIO architecture question is answered. I > also wonder how much use the MAC driver is without any PHY code? > Unless you have a good reason, i don't think we should merge this > until it makes the needed calls into phylib. It is not much code to > add. You mean I should skip out the parsing of the mdio base until I'm using it? That's reasonable.
Thanks for the review Jakub. On Fri, 6 Aug 2021 at 23:10, Jakub Kicinski <kuba@kernel.org> wrote: > > +static int liteeth_start_xmit(struct sk_buff *skb, struct net_device *netdev) > > +{ > > + struct liteeth *priv = netdev_priv(netdev); > > + void __iomem *txbuffer; > > + int ret; > > + u8 val; > > + > > + /* Reject oversize packets */ > > + if (unlikely(skb->len > MAX_PKT_SIZE)) { > > + if (net_ratelimit()) > > + netdev_dbg(netdev, "tx packet too big\n"); > > + goto drop; > > + } > > + > > + txbuffer = priv->tx_base + priv->tx_slot * LITEETH_BUFFER_SIZE; > > + memcpy_toio(txbuffer, skb->data, skb->len); > > + writeb(priv->tx_slot, priv->base + LITEETH_READER_SLOT); > > + writew(skb->len, priv->base + LITEETH_READER_LENGTH); > > + > > + ret = readl_poll_timeout_atomic(priv->base + LITEETH_READER_READY, val, val, 5, 1000); > > Why the need for poll if there is an interrupt? > Why not stop the Tx queue once you're out of slots and restart > it when the completion interrupt comes? That makes sense. In testing I have not been able to hit the LITEETH_READER_READY not-ready state. I assume it's there to say that the slots are full. > > > + if (ret == -ETIMEDOUT) { > > + netdev_err(netdev, "LITEETH_READER_READY timed out\n"); > > ratelimit this as well, please > > > + goto drop; > > + } > > + > > + writeb(1, priv->base + LITEETH_READER_START); > > + > > + netdev->stats.tx_bytes += skb->len; > > Please count bytes and packets in the same place AFAIK we don't know the length when the interrupt comes in, so we need to count both here in xmit? > > > + priv->tx_slot = (priv->tx_slot + 1) % priv->num_tx_slots; > > + dev_kfree_skb_any(skb); > > + return NETDEV_TX_OK; > > +drop: > > + /* Drop the packet */ > > + dev_kfree_skb_any(skb); > > + netdev->stats.tx_dropped++; > > + > > + return NETDEV_TX_OK; > > +}
On Mon, Aug 09, 2021 at 03:20:57AM +0000, Joel Stanley wrote: > On Sat, 7 Aug 2021 at 19:19, Andrew Lunn <andrew@lunn.ch> wrote: > > > > > +static void liteeth_reset_hw(struct liteeth *priv) > > > +{ > > > + /* Reset, twice */ > > > + writeb(0, priv->base + LITEETH_PHY_CRG_RESET); > > > + udelay(10); > > > + writeb(1, priv->base + LITEETH_PHY_CRG_RESET); > > > + udelay(10); > > > + writeb(0, priv->base + LITEETH_PHY_CRG_RESET); > > > + udelay(10); > > > > What is this actually resetting? > > This comes from the reference firmware that many (but not all) litex > systems run before loading their operating system. > > I'm not completely sure how necessary it still is; I will drop it for now. Which did not answer my question. Once we know what is being reset, we can maybe suggest when/how it should be reset. > > > +static int liteeth_probe(struct platform_device *pdev) > > > +{ > > > + struct net_device *netdev; > > > + void __iomem *buf_base; > > > + struct resource *res; > > > + struct liteeth *priv; > > > + int irq, err; > > > + > > > + netdev = alloc_etherdev(sizeof(*priv)); > > > + if (!netdev) > > > + return -ENOMEM; > > > + > > > + priv = netdev_priv(netdev); > > > + priv->netdev = netdev; > > > + priv->dev = &pdev->dev; > > > + > > > + irq = platform_get_irq(pdev, 0); > > > + if (irq < 0) { > > > + dev_err(&pdev->dev, "Failed to get IRQ\n"); > > > + goto err; > > > + } > > > + > > > + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); > > > + priv->base = devm_ioremap_resource(&pdev->dev, res); > > > + if (IS_ERR(priv->base)) { > > > + err = PTR_ERR(priv->base); > > > + goto err; > > > + } > > > + > > > + res = platform_get_resource(pdev, IORESOURCE_MEM, 1); > > > + priv->mdio_base = devm_ioremap_resource(&pdev->dev, res); > > > + if (IS_ERR(priv->mdio_base)) { > > > + err = PTR_ERR(priv->mdio_base); > > > + goto err; > > > + } > > > > So you don't have any PHY handling, or any MDIO bus master code. So i > > would drop this, until the MDIO architecture question is answered. I > > also wonder how much use the MAC driver is without any PHY code? > > Unless you have a good reason, i don't think we should merge this > > until it makes the needed calls into phylib. It is not much code to > > add. > > You mean I should skip out the parsing of the mdio base until I'm > using it? That's reasonable. It could be we insist you add MDIO and PHY handling. But first we need to understand the architecture. Andrew
On Mon, 9 Aug 2021 12:03:36 +0000 Joel Stanley wrote: > On Fri, 6 Aug 2021 at 23:10, Jakub Kicinski <kuba@kernel.org> wrote: > > > +static int liteeth_start_xmit(struct sk_buff *skb, struct net_device *netdev) > > > +{ > > > + struct liteeth *priv = netdev_priv(netdev); > > > + void __iomem *txbuffer; > > > + int ret; > > > + u8 val; > > > + > > > + /* Reject oversize packets */ > > > + if (unlikely(skb->len > MAX_PKT_SIZE)) { > > > + if (net_ratelimit()) > > > + netdev_dbg(netdev, "tx packet too big\n"); > > > + goto drop; > > > + } > > > + > > > + txbuffer = priv->tx_base + priv->tx_slot * LITEETH_BUFFER_SIZE; > > > + memcpy_toio(txbuffer, skb->data, skb->len); > > > + writeb(priv->tx_slot, priv->base + LITEETH_READER_SLOT); > > > + writew(skb->len, priv->base + LITEETH_READER_LENGTH); > > > + > > > + ret = readl_poll_timeout_atomic(priv->base + LITEETH_READER_READY, val, val, 5, 1000); > > > > Why the need for poll if there is an interrupt? > > Why not stop the Tx queue once you're out of slots and restart > > it when the completion interrupt comes? > > That makes sense. > > In testing I have not been able to hit the LITEETH_READER_READY > not-ready state. I assume it's there to say that the slots are full. In that case it's probably best to stop the Tx queue in the xmit routine once all the lots are used, and restart it from the interrupt. I was guessing maybe the IRQ is not always there, but that doesn't seem to be the case. > > > + if (ret == -ETIMEDOUT) { > > > + netdev_err(netdev, "LITEETH_READER_READY timed out\n"); > > > > ratelimit this as well, please > > > > > + goto drop; > > > + } > > > + > > > + writeb(1, priv->base + LITEETH_READER_START); > > > + > > > + netdev->stats.tx_bytes += skb->len; > > > > Please count bytes and packets in the same place > > AFAIK we don't know the length when the interrupt comes in, so we need > to count both here in xmit? Either that or allocate a small array (num_tx_slots) to save the lengths for IRQ routine to use. > > > + priv->tx_slot = (priv->tx_slot + 1) % priv->num_tx_slots; > > > + dev_kfree_skb_any(skb); > > > + return NETDEV_TX_OK; > > > +drop: > > > + /* Drop the packet */ > > > + dev_kfree_skb_any(skb); > > > + netdev->stats.tx_dropped++; > > > + > > > + return NETDEV_TX_OK; > > > +}
Hi Florent and Gabriel, On Fri, 6 Aug 2021 at 12:27, Florent Kermarrec <florent@enjoy-digital.fr> wrote: > > Hi Gabriel, Joel, > > for 1), the polling mode has been useful recently with Linux-On-LiteX-Vexriscv to bring up the SMP version before enabling the IRQs and can probably be useful in the future for similar purposes. So if it's not too complicated to maintain, this could be useful to keep it. I would prefer to get a basic working driver for the straight forward Linux-on-FPGA use case merged first. We can then work out what should be done next. I would like to see hardware improvements to make it go faster, and we can look at other features like polling if you still need them. > > Le ven. 6 août 2021 à 13:48, Gabriel L. Somlo <gsomlo@gmail.com> a écrit : >> >> Hi Joel, >> >> Thanks for pushing this upstream (and for writing it to begin with)! >> >> Would you mind diff-ing your version of litex_liteeth.c against what >> is currently in >> https://github.com/litex-hub/linux/blob/litex-rebase/drivers/net/ethernet/litex/litex_liteeth.c ? >> >> Two main differences we should discuss: >> >> 1. there's a polling mode (added by Antony Pavlov), and if we >> decide *not* to keep it around, I want to ensure we do that >> deliberately, with an explanation as to why; I assumed this was only for broken systems, or systems that were still being brought up. Haven't the risc-v socs improved to the point where they have working interrupts now? >> >> 2. LiteX CSRs are accessed using `litex_[read|write][8|16|32]()` >> as opposed to simply `[read|write][b|w|l]()`. The former set >> are defined in `include/linux/litex.h` and are needed to >> ensure correct accesses regardless of endianness, since by >> default LiteX registers' endianness mirrors that of the >> configured CPU. I don't like that they get the parameters in the wrong order compared to the rest of the kernel, and that they force 32-bit access no matter the width of the access being performed. They also mean we can't use _releaxed variants of accessors to eg. read a set of registers without having a barrier in between. That said, I have used the csr accessors so we can support the mor1k, which afaict is the only litex CPU that defaults to big endian. If you could help review the patch so we can get it merged this cycle, then we can work on future enhancements. Cheers, Joel
diff --git a/drivers/net/ethernet/Kconfig b/drivers/net/ethernet/Kconfig index 1cdff1dca790..d796684ec9ca 100644 --- a/drivers/net/ethernet/Kconfig +++ b/drivers/net/ethernet/Kconfig @@ -118,6 +118,7 @@ config LANTIQ_XRX200 Support for the PMAC of the Gigabit switch (GSWIP) inside the Lantiq / Intel VRX200 VDSL SoC +source "drivers/net/ethernet/litex/Kconfig" source "drivers/net/ethernet/marvell/Kconfig" source "drivers/net/ethernet/mediatek/Kconfig" source "drivers/net/ethernet/mellanox/Kconfig" diff --git a/drivers/net/ethernet/Makefile b/drivers/net/ethernet/Makefile index cb3f9084a21b..aaa5078cd7d1 100644 --- a/drivers/net/ethernet/Makefile +++ b/drivers/net/ethernet/Makefile @@ -51,6 +51,7 @@ obj-$(CONFIG_JME) += jme.o obj-$(CONFIG_KORINA) += korina.o obj-$(CONFIG_LANTIQ_ETOP) += lantiq_etop.o obj-$(CONFIG_LANTIQ_XRX200) += lantiq_xrx200.o +obj-$(CONFIG_NET_VENDOR_LITEX) += litex/ obj-$(CONFIG_NET_VENDOR_MARVELL) += marvell/ obj-$(CONFIG_NET_VENDOR_MEDIATEK) += mediatek/ obj-$(CONFIG_NET_VENDOR_MELLANOX) += mellanox/ diff --git a/drivers/net/ethernet/litex/Kconfig b/drivers/net/ethernet/litex/Kconfig new file mode 100644 index 000000000000..0a863e3ff7bf --- /dev/null +++ b/drivers/net/ethernet/litex/Kconfig @@ -0,0 +1,24 @@ +# +# LiteX device configuration +# + +config NET_VENDOR_LITEX + bool "LiteX devices" + default y + help + If you have a network (Ethernet) card belonging to this class, say Y. + + Note that the answer to this question doesn't directly affect the + kernel: saying N will just cause the configurator to skip all + the questions about LiteX devices. If you say Y, you will be asked + for your specific card in the following questions. + +if NET_VENDOR_LITEX + +config LITEX_LITEETH + tristate "LiteX Ethernet support" + help + If you wish to compile a kernel for hardware with a LiteX LiteEth + device then you should answer Y to this. + +endif # NET_VENDOR_LITEX diff --git a/drivers/net/ethernet/litex/Makefile b/drivers/net/ethernet/litex/Makefile new file mode 100644 index 000000000000..9343b73b8e49 --- /dev/null +++ b/drivers/net/ethernet/litex/Makefile @@ -0,0 +1,5 @@ +# +# Makefile for the LiteX network device drivers. +# + +obj-$(CONFIG_LITEX_LITEETH) += litex_liteeth.o diff --git a/drivers/net/ethernet/litex/litex_liteeth.c b/drivers/net/ethernet/litex/litex_liteeth.c new file mode 100644 index 000000000000..48b551e6d97b --- /dev/null +++ b/drivers/net/ethernet/litex/litex_liteeth.c @@ -0,0 +1,340 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * LiteX Liteeth Ethernet + * + * Copyright 2017 Joel Stanley <joel@jms.id.au> + * + */ + +#include <linux/etherdevice.h> +#include <linux/interrupt.h> +#include <linux/module.h> +#include <linux/of.h> +#include <linux/of_net.h> +#include <linux/of_address.h> +#include <linux/phy.h> +#include <linux/platform_device.h> +#include <linux/iopoll.h> + +#define LITEETH_WRITER_SLOT 0x00 +#define LITEETH_WRITER_LENGTH 0x04 +#define LITEETH_WRITER_ERRORS 0x08 +#define LITEETH_WRITER_EV_STATUS 0x0C +#define LITEETH_WRITER_EV_PENDING 0x10 +#define LITEETH_WRITER_EV_ENABLE 0x14 +#define LITEETH_READER_START 0x18 +#define LITEETH_READER_READY 0x1C +#define LITEETH_READER_LEVEL 0x20 +#define LITEETH_READER_SLOT 0x24 +#define LITEETH_READER_LENGTH 0x28 +#define LITEETH_READER_EV_STATUS 0x2C +#define LITEETH_READER_EV_PENDING 0x30 +#define LITEETH_READER_EV_ENABLE 0x34 +#define LITEETH_PREAMBLE_CRC 0x38 +#define LITEETH_PREAMBLE_ERRORS 0x3C +#define LITEETH_CRC_ERRORS 0x40 + +#define LITEETH_PHY_CRG_RESET 0x00 +#define LITEETH_MDIO_W 0x04 +#define LITEETH_MDIO_R 0x0C + +#define DRV_NAME "liteeth" + +#define LITEETH_BUFFER_SIZE 0x800 +#define MAX_PKT_SIZE LITEETH_BUFFER_SIZE + +struct liteeth { + void __iomem *base; + void __iomem *mdio_base; + struct net_device *netdev; + struct device *dev; + struct mii_bus *mii_bus; + + /* Link management */ + int cur_duplex; + int cur_speed; + + /* Tx */ + int tx_slot; + int num_tx_slots; + void __iomem *tx_base; + + /* Rx */ + int rx_slot; + int num_rx_slots; + void __iomem *rx_base; +}; + + +static int liteeth_rx(struct net_device *netdev) +{ + struct liteeth *priv = netdev_priv(netdev); + struct sk_buff *skb; + unsigned char *data; + u8 rx_slot; + int len; + + rx_slot = readb(priv->base + LITEETH_WRITER_SLOT); + len = readl(priv->base + LITEETH_WRITER_LENGTH); + + skb = netdev_alloc_skb(netdev, len + NET_IP_ALIGN); + if (!skb) { + netdev_err(netdev, "couldn't get memory"); + netdev->stats.rx_dropped++; + return NET_RX_DROP; + } + + /* Ensure alignemnt of the ip header within the skb */ + skb_reserve(skb, NET_IP_ALIGN); + if (len == 0 || len > 2048) + return NET_RX_DROP; + data = skb_put(skb, len); + memcpy_fromio(data, priv->rx_base + rx_slot * LITEETH_BUFFER_SIZE, len); + skb->protocol = eth_type_trans(skb, netdev); + + netdev->stats.rx_packets++; + netdev->stats.rx_bytes += len; + + return netif_rx(skb); +} + +static irqreturn_t liteeth_interrupt(int irq, void *dev_id) +{ + struct net_device *netdev = dev_id; + struct liteeth *priv = netdev_priv(netdev); + u8 reg; + + reg = readb(priv->base + LITEETH_READER_EV_PENDING); + if (reg) { + netdev->stats.tx_packets++; + writeb(reg, priv->base + LITEETH_READER_EV_PENDING); + } + + reg = readb(priv->base + LITEETH_WRITER_EV_PENDING); + if (reg) { + liteeth_rx(netdev); + writeb(reg, priv->base + LITEETH_WRITER_EV_PENDING); + } + + return IRQ_HANDLED; +} + +static int liteeth_open(struct net_device *netdev) +{ + struct liteeth *priv = netdev_priv(netdev); + int err; + + /* Clear pending events */ + writeb(1, priv->base + LITEETH_WRITER_EV_PENDING); + writeb(1, priv->base + LITEETH_READER_EV_PENDING); + + err = request_irq(netdev->irq, liteeth_interrupt, 0, netdev->name, netdev); + if (err) { + netdev_err(netdev, "failed to request irq %d\n", netdev->irq); + return err; + } + + /* Enable IRQs */ + writeb(1, priv->base + LITEETH_WRITER_EV_ENABLE); + writeb(1, priv->base + LITEETH_READER_EV_ENABLE); + + /* TODO: Remove these once we have working mdio support */ + priv->cur_duplex = DUPLEX_FULL; + priv->cur_speed = SPEED_100; + netif_carrier_on(netdev); + + netif_start_queue(netdev); + + return 0; +} + +static int liteeth_stop(struct net_device *netdev) +{ + struct liteeth *priv = netdev_priv(netdev); + + netif_stop_queue(netdev); + + writeb(0, priv->base + LITEETH_WRITER_EV_ENABLE); + writeb(0, priv->base + LITEETH_READER_EV_ENABLE); + + free_irq(netdev->irq, netdev); + + return 0; +} + +static int liteeth_start_xmit(struct sk_buff *skb, struct net_device *netdev) +{ + struct liteeth *priv = netdev_priv(netdev); + void __iomem *txbuffer; + int ret; + u8 val; + + /* Reject oversize packets */ + if (unlikely(skb->len > MAX_PKT_SIZE)) { + if (net_ratelimit()) + netdev_dbg(netdev, "tx packet too big\n"); + goto drop; + } + + txbuffer = priv->tx_base + priv->tx_slot * LITEETH_BUFFER_SIZE; + memcpy_toio(txbuffer, skb->data, skb->len); + writeb(priv->tx_slot, priv->base + LITEETH_READER_SLOT); + writew(skb->len, priv->base + LITEETH_READER_LENGTH); + + ret = readl_poll_timeout_atomic(priv->base + LITEETH_READER_READY, val, val, 5, 1000); + if (ret == -ETIMEDOUT) { + netdev_err(netdev, "LITEETH_READER_READY timed out\n"); + goto drop; + } + + writeb(1, priv->base + LITEETH_READER_START); + + netdev->stats.tx_bytes += skb->len; + + priv->tx_slot = (priv->tx_slot + 1) % priv->num_tx_slots; + dev_kfree_skb_any(skb); + return NETDEV_TX_OK; +drop: + /* Drop the packet */ + dev_kfree_skb_any(skb); + netdev->stats.tx_dropped++; + + return NETDEV_TX_OK; +} + +static const struct net_device_ops liteeth_netdev_ops = { + .ndo_open = liteeth_open, + .ndo_stop = liteeth_stop, + .ndo_start_xmit = liteeth_start_xmit, +}; + +static void liteeth_reset_hw(struct liteeth *priv) +{ + /* Reset, twice */ + writeb(0, priv->base + LITEETH_PHY_CRG_RESET); + udelay(10); + writeb(1, priv->base + LITEETH_PHY_CRG_RESET); + udelay(10); + writeb(0, priv->base + LITEETH_PHY_CRG_RESET); + udelay(10); +} + +static int liteeth_probe(struct platform_device *pdev) +{ + struct net_device *netdev; + void __iomem *buf_base; + struct resource *res; + struct liteeth *priv; + int irq, err; + + netdev = alloc_etherdev(sizeof(*priv)); + if (!netdev) + return -ENOMEM; + + priv = netdev_priv(netdev); + priv->netdev = netdev; + priv->dev = &pdev->dev; + + irq = platform_get_irq(pdev, 0); + if (irq < 0) { + dev_err(&pdev->dev, "Failed to get IRQ\n"); + goto err; + } + + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); + priv->base = devm_ioremap_resource(&pdev->dev, res); + if (IS_ERR(priv->base)) { + err = PTR_ERR(priv->base); + goto err; + } + + res = platform_get_resource(pdev, IORESOURCE_MEM, 1); + priv->mdio_base = devm_ioremap_resource(&pdev->dev, res); + if (IS_ERR(priv->mdio_base)) { + err = PTR_ERR(priv->mdio_base); + goto err; + } + + res = platform_get_resource(pdev, IORESOURCE_MEM, 2); + buf_base = devm_ioremap_resource(&pdev->dev, res); + if (IS_ERR(buf_base)) { + err = PTR_ERR(buf_base); + goto err; + } + + err = of_property_read_u32(pdev->dev.of_node, "rx-fifo-depth", + &priv->num_rx_slots); + if (err) { + dev_err(&pdev->dev, "unable to get rx-fifo-depth\n"); + goto err; + } + + err = of_property_read_u32(pdev->dev.of_node, "tx-fifo-depth", + &priv->num_tx_slots); + if (err) { + dev_err(&pdev->dev, "unable to get tx-fifo-depth\n"); + goto err; + } + + /* Rx slots */ + priv->rx_base = buf_base; + priv->rx_slot = 0; + + /* Tx slots come after Rx slots */ + priv->tx_base = buf_base + priv->num_rx_slots * LITEETH_BUFFER_SIZE; + priv->tx_slot = 0; + + err = of_get_mac_address(pdev->dev.of_node, netdev->dev_addr); + if (err) + eth_hw_addr_random(netdev); + + SET_NETDEV_DEV(netdev, &pdev->dev); + platform_set_drvdata(pdev, netdev); + + netdev->netdev_ops = &liteeth_netdev_ops; + netdev->irq = irq; + + liteeth_reset_hw(priv); + + err = register_netdev(netdev); + if (err) { + dev_err(&pdev->dev, "Failed to register netdev\n"); + goto err; + } + + netdev_info(netdev, "irq %d, mapped at %px\n", netdev->irq, priv->base); + + return 0; +err: + free_netdev(netdev); + return err; +} + +static int liteeth_remove(struct platform_device *pdev) +{ + struct net_device *netdev = platform_get_drvdata(pdev); + + unregister_netdev(netdev); + free_netdev(netdev); + + return 0; +} + +static const struct of_device_id liteeth_of_match[] = { + { .compatible = "litex,liteeth" }, + { } +}; +MODULE_DEVICE_TABLE(of, liteeth_of_match); + +static struct platform_driver liteeth_driver = { + .probe = liteeth_probe, + .remove = liteeth_remove, + .driver = { + .name = DRV_NAME, + .of_match_table = liteeth_of_match, + }, +}; +module_platform_driver(liteeth_driver); + +MODULE_AUTHOR("Joel Stanley <joel@jms.id.au>"); +MODULE_LICENSE("GPL");
LiteX is a soft system-on-chip that targets FPGAs. LiteETH is a basic network device that is commonly used in LiteX designs. The driver was first written in 2017 and has been maintained by the LiteX community in various trees. Thank you to all who have contributed. Co-developed-by: Gabriel Somlo <gsomlo@gmail.com> Co-developed-by: David Shah <dave@ds0.me> Co-developed-by: Stafford Horne <shorne@gmail.com> Signed-off-by: Joel Stanley <joel@jms.id.au> --- drivers/net/ethernet/Kconfig | 1 + drivers/net/ethernet/Makefile | 1 + drivers/net/ethernet/litex/Kconfig | 24 ++ drivers/net/ethernet/litex/Makefile | 5 + drivers/net/ethernet/litex/litex_liteeth.c | 340 +++++++++++++++++++++ 5 files changed, 371 insertions(+) create mode 100644 drivers/net/ethernet/litex/Kconfig create mode 100644 drivers/net/ethernet/litex/Makefile create mode 100644 drivers/net/ethernet/litex/litex_liteeth.c