summaryrefslogtreecommitdiffstats
path: root/drivers/net/ethernet/nxp
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/net/ethernet/nxp')
-rw-r--r--drivers/net/ethernet/nxp/Kconfig10
-rw-r--r--drivers/net/ethernet/nxp/Makefile2
-rw-r--r--drivers/net/ethernet/nxp/lpc_eth.c1524
3 files changed, 1536 insertions, 0 deletions
diff --git a/drivers/net/ethernet/nxp/Kconfig b/drivers/net/ethernet/nxp/Kconfig
new file mode 100644
index 0000000000..c84997db82
--- /dev/null
+++ b/drivers/net/ethernet/nxp/Kconfig
@@ -0,0 +1,10 @@
+# SPDX-License-Identifier: GPL-2.0-only
+config LPC_ENET
+ tristate "NXP ethernet MAC on LPC devices"
+ depends on ARCH_LPC32XX || COMPILE_TEST
+ select PHYLIB
+ select CRC32
+ help
+ Say Y or M here if you want to use the NXP ethernet MAC included on
+ some NXP LPC devices. You can safely enable this option for LPC32xx
+ SoC. Also available as a module.
diff --git a/drivers/net/ethernet/nxp/Makefile b/drivers/net/ethernet/nxp/Makefile
new file mode 100644
index 0000000000..cba6ddcc39
--- /dev/null
+++ b/drivers/net/ethernet/nxp/Makefile
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0-only
+obj-$(CONFIG_LPC_ENET) += lpc_eth.o
diff --git a/drivers/net/ethernet/nxp/lpc_eth.c b/drivers/net/ethernet/nxp/lpc_eth.c
new file mode 100644
index 0000000000..1a4a272f4c
--- /dev/null
+++ b/drivers/net/ethernet/nxp/lpc_eth.c
@@ -0,0 +1,1524 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * drivers/net/ethernet/nxp/lpc_eth.c
+ *
+ * Author: Kevin Wells <kevin.wells@nxp.com>
+ *
+ * Copyright (C) 2010 NXP Semiconductors
+ * Copyright (C) 2012 Roland Stigge <stigge@antcom.de>
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/clk.h>
+#include <linux/crc32.h>
+#include <linux/etherdevice.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_mdio.h>
+#include <linux/of_net.h>
+#include <linux/phy.h>
+#include <linux/platform_device.h>
+#include <linux/spinlock.h>
+#include <linux/soc/nxp/lpc32xx-misc.h>
+
+#define MODNAME "lpc-eth"
+#define DRV_VERSION "1.00"
+
+#define ENET_MAXF_SIZE 1536
+#define ENET_RX_DESC 48
+#define ENET_TX_DESC 16
+
+#define NAPI_WEIGHT 16
+
+/*
+ * Ethernet MAC controller Register offsets
+ */
+#define LPC_ENET_MAC1(x) (x + 0x000)
+#define LPC_ENET_MAC2(x) (x + 0x004)
+#define LPC_ENET_IPGT(x) (x + 0x008)
+#define LPC_ENET_IPGR(x) (x + 0x00C)
+#define LPC_ENET_CLRT(x) (x + 0x010)
+#define LPC_ENET_MAXF(x) (x + 0x014)
+#define LPC_ENET_SUPP(x) (x + 0x018)
+#define LPC_ENET_TEST(x) (x + 0x01C)
+#define LPC_ENET_MCFG(x) (x + 0x020)
+#define LPC_ENET_MCMD(x) (x + 0x024)
+#define LPC_ENET_MADR(x) (x + 0x028)
+#define LPC_ENET_MWTD(x) (x + 0x02C)
+#define LPC_ENET_MRDD(x) (x + 0x030)
+#define LPC_ENET_MIND(x) (x + 0x034)
+#define LPC_ENET_SA0(x) (x + 0x040)
+#define LPC_ENET_SA1(x) (x + 0x044)
+#define LPC_ENET_SA2(x) (x + 0x048)
+#define LPC_ENET_COMMAND(x) (x + 0x100)
+#define LPC_ENET_STATUS(x) (x + 0x104)
+#define LPC_ENET_RXDESCRIPTOR(x) (x + 0x108)
+#define LPC_ENET_RXSTATUS(x) (x + 0x10C)
+#define LPC_ENET_RXDESCRIPTORNUMBER(x) (x + 0x110)
+#define LPC_ENET_RXPRODUCEINDEX(x) (x + 0x114)
+#define LPC_ENET_RXCONSUMEINDEX(x) (x + 0x118)
+#define LPC_ENET_TXDESCRIPTOR(x) (x + 0x11C)
+#define LPC_ENET_TXSTATUS(x) (x + 0x120)
+#define LPC_ENET_TXDESCRIPTORNUMBER(x) (x + 0x124)
+#define LPC_ENET_TXPRODUCEINDEX(x) (x + 0x128)
+#define LPC_ENET_TXCONSUMEINDEX(x) (x + 0x12C)
+#define LPC_ENET_TSV0(x) (x + 0x158)
+#define LPC_ENET_TSV1(x) (x + 0x15C)
+#define LPC_ENET_RSV(x) (x + 0x160)
+#define LPC_ENET_FLOWCONTROLCOUNTER(x) (x + 0x170)
+#define LPC_ENET_FLOWCONTROLSTATUS(x) (x + 0x174)
+#define LPC_ENET_RXFILTER_CTRL(x) (x + 0x200)
+#define LPC_ENET_RXFILTERWOLSTATUS(x) (x + 0x204)
+#define LPC_ENET_RXFILTERWOLCLEAR(x) (x + 0x208)
+#define LPC_ENET_HASHFILTERL(x) (x + 0x210)
+#define LPC_ENET_HASHFILTERH(x) (x + 0x214)
+#define LPC_ENET_INTSTATUS(x) (x + 0xFE0)
+#define LPC_ENET_INTENABLE(x) (x + 0xFE4)
+#define LPC_ENET_INTCLEAR(x) (x + 0xFE8)
+#define LPC_ENET_INTSET(x) (x + 0xFEC)
+#define LPC_ENET_POWERDOWN(x) (x + 0xFF4)
+
+/*
+ * mac1 register definitions
+ */
+#define LPC_MAC1_RECV_ENABLE (1 << 0)
+#define LPC_MAC1_PASS_ALL_RX_FRAMES (1 << 1)
+#define LPC_MAC1_RX_FLOW_CONTROL (1 << 2)
+#define LPC_MAC1_TX_FLOW_CONTROL (1 << 3)
+#define LPC_MAC1_LOOPBACK (1 << 4)
+#define LPC_MAC1_RESET_TX (1 << 8)
+#define LPC_MAC1_RESET_MCS_TX (1 << 9)
+#define LPC_MAC1_RESET_RX (1 << 10)
+#define LPC_MAC1_RESET_MCS_RX (1 << 11)
+#define LPC_MAC1_SIMULATION_RESET (1 << 14)
+#define LPC_MAC1_SOFT_RESET (1 << 15)
+
+/*
+ * mac2 register definitions
+ */
+#define LPC_MAC2_FULL_DUPLEX (1 << 0)
+#define LPC_MAC2_FRAME_LENGTH_CHECKING (1 << 1)
+#define LPC_MAC2_HUGH_LENGTH_CHECKING (1 << 2)
+#define LPC_MAC2_DELAYED_CRC (1 << 3)
+#define LPC_MAC2_CRC_ENABLE (1 << 4)
+#define LPC_MAC2_PAD_CRC_ENABLE (1 << 5)
+#define LPC_MAC2_VLAN_PAD_ENABLE (1 << 6)
+#define LPC_MAC2_AUTO_DETECT_PAD_ENABLE (1 << 7)
+#define LPC_MAC2_PURE_PREAMBLE_ENFORCEMENT (1 << 8)
+#define LPC_MAC2_LONG_PREAMBLE_ENFORCEMENT (1 << 9)
+#define LPC_MAC2_NO_BACKOFF (1 << 12)
+#define LPC_MAC2_BACK_PRESSURE (1 << 13)
+#define LPC_MAC2_EXCESS_DEFER (1 << 14)
+
+/*
+ * ipgt register definitions
+ */
+#define LPC_IPGT_LOAD(n) ((n) & 0x7F)
+
+/*
+ * ipgr register definitions
+ */
+#define LPC_IPGR_LOAD_PART2(n) ((n) & 0x7F)
+#define LPC_IPGR_LOAD_PART1(n) (((n) & 0x7F) << 8)
+
+/*
+ * clrt register definitions
+ */
+#define LPC_CLRT_LOAD_RETRY_MAX(n) ((n) & 0xF)
+#define LPC_CLRT_LOAD_COLLISION_WINDOW(n) (((n) & 0x3F) << 8)
+
+/*
+ * maxf register definitions
+ */
+#define LPC_MAXF_LOAD_MAX_FRAME_LEN(n) ((n) & 0xFFFF)
+
+/*
+ * supp register definitions
+ */
+#define LPC_SUPP_SPEED (1 << 8)
+#define LPC_SUPP_RESET_RMII (1 << 11)
+
+/*
+ * test register definitions
+ */
+#define LPC_TEST_SHORTCUT_PAUSE_QUANTA (1 << 0)
+#define LPC_TEST_PAUSE (1 << 1)
+#define LPC_TEST_BACKPRESSURE (1 << 2)
+
+/*
+ * mcfg register definitions
+ */
+#define LPC_MCFG_SCAN_INCREMENT (1 << 0)
+#define LPC_MCFG_SUPPRESS_PREAMBLE (1 << 1)
+#define LPC_MCFG_CLOCK_SELECT(n) (((n) & 0x7) << 2)
+#define LPC_MCFG_CLOCK_HOST_DIV_4 0
+#define LPC_MCFG_CLOCK_HOST_DIV_6 2
+#define LPC_MCFG_CLOCK_HOST_DIV_8 3
+#define LPC_MCFG_CLOCK_HOST_DIV_10 4
+#define LPC_MCFG_CLOCK_HOST_DIV_14 5
+#define LPC_MCFG_CLOCK_HOST_DIV_20 6
+#define LPC_MCFG_CLOCK_HOST_DIV_28 7
+#define LPC_MCFG_RESET_MII_MGMT (1 << 15)
+
+/*
+ * mcmd register definitions
+ */
+#define LPC_MCMD_READ (1 << 0)
+#define LPC_MCMD_SCAN (1 << 1)
+
+/*
+ * madr register definitions
+ */
+#define LPC_MADR_REGISTER_ADDRESS(n) ((n) & 0x1F)
+#define LPC_MADR_PHY_0ADDRESS(n) (((n) & 0x1F) << 8)
+
+/*
+ * mwtd register definitions
+ */
+#define LPC_MWDT_WRITE(n) ((n) & 0xFFFF)
+
+/*
+ * mrdd register definitions
+ */
+#define LPC_MRDD_READ_MASK 0xFFFF
+
+/*
+ * mind register definitions
+ */
+#define LPC_MIND_BUSY (1 << 0)
+#define LPC_MIND_SCANNING (1 << 1)
+#define LPC_MIND_NOT_VALID (1 << 2)
+#define LPC_MIND_MII_LINK_FAIL (1 << 3)
+
+/*
+ * command register definitions
+ */
+#define LPC_COMMAND_RXENABLE (1 << 0)
+#define LPC_COMMAND_TXENABLE (1 << 1)
+#define LPC_COMMAND_REG_RESET (1 << 3)
+#define LPC_COMMAND_TXRESET (1 << 4)
+#define LPC_COMMAND_RXRESET (1 << 5)
+#define LPC_COMMAND_PASSRUNTFRAME (1 << 6)
+#define LPC_COMMAND_PASSRXFILTER (1 << 7)
+#define LPC_COMMAND_TXFLOWCONTROL (1 << 8)
+#define LPC_COMMAND_RMII (1 << 9)
+#define LPC_COMMAND_FULLDUPLEX (1 << 10)
+
+/*
+ * status register definitions
+ */
+#define LPC_STATUS_RXACTIVE (1 << 0)
+#define LPC_STATUS_TXACTIVE (1 << 1)
+
+/*
+ * tsv0 register definitions
+ */
+#define LPC_TSV0_CRC_ERROR (1 << 0)
+#define LPC_TSV0_LENGTH_CHECK_ERROR (1 << 1)
+#define LPC_TSV0_LENGTH_OUT_OF_RANGE (1 << 2)
+#define LPC_TSV0_DONE (1 << 3)
+#define LPC_TSV0_MULTICAST (1 << 4)
+#define LPC_TSV0_BROADCAST (1 << 5)
+#define LPC_TSV0_PACKET_DEFER (1 << 6)
+#define LPC_TSV0_ESCESSIVE_DEFER (1 << 7)
+#define LPC_TSV0_ESCESSIVE_COLLISION (1 << 8)
+#define LPC_TSV0_LATE_COLLISION (1 << 9)
+#define LPC_TSV0_GIANT (1 << 10)
+#define LPC_TSV0_UNDERRUN (1 << 11)
+#define LPC_TSV0_TOTAL_BYTES(n) (((n) >> 12) & 0xFFFF)
+#define LPC_TSV0_CONTROL_FRAME (1 << 28)
+#define LPC_TSV0_PAUSE (1 << 29)
+#define LPC_TSV0_BACKPRESSURE (1 << 30)
+#define LPC_TSV0_VLAN (1 << 31)
+
+/*
+ * tsv1 register definitions
+ */
+#define LPC_TSV1_TRANSMIT_BYTE_COUNT(n) ((n) & 0xFFFF)
+#define LPC_TSV1_COLLISION_COUNT(n) (((n) >> 16) & 0xF)
+
+/*
+ * rsv register definitions
+ */
+#define LPC_RSV_RECEIVED_BYTE_COUNT(n) ((n) & 0xFFFF)
+#define LPC_RSV_RXDV_EVENT_IGNORED (1 << 16)
+#define LPC_RSV_RXDV_EVENT_PREVIOUSLY_SEEN (1 << 17)
+#define LPC_RSV_CARRIER_EVNT_PREVIOUS_SEEN (1 << 18)
+#define LPC_RSV_RECEIVE_CODE_VIOLATION (1 << 19)
+#define LPC_RSV_CRC_ERROR (1 << 20)
+#define LPC_RSV_LENGTH_CHECK_ERROR (1 << 21)
+#define LPC_RSV_LENGTH_OUT_OF_RANGE (1 << 22)
+#define LPC_RSV_RECEIVE_OK (1 << 23)
+#define LPC_RSV_MULTICAST (1 << 24)
+#define LPC_RSV_BROADCAST (1 << 25)
+#define LPC_RSV_DRIBBLE_NIBBLE (1 << 26)
+#define LPC_RSV_CONTROL_FRAME (1 << 27)
+#define LPC_RSV_PAUSE (1 << 28)
+#define LPC_RSV_UNSUPPORTED_OPCODE (1 << 29)
+#define LPC_RSV_VLAN (1 << 30)
+
+/*
+ * flowcontrolcounter register definitions
+ */
+#define LPC_FCCR_MIRRORCOUNTER(n) ((n) & 0xFFFF)
+#define LPC_FCCR_PAUSETIMER(n) (((n) >> 16) & 0xFFFF)
+
+/*
+ * flowcontrolstatus register definitions
+ */
+#define LPC_FCCR_MIRRORCOUNTERCURRENT(n) ((n) & 0xFFFF)
+
+/*
+ * rxfilterctrl, rxfilterwolstatus, and rxfilterwolclear shared
+ * register definitions
+ */
+#define LPC_RXFLTRW_ACCEPTUNICAST (1 << 0)
+#define LPC_RXFLTRW_ACCEPTUBROADCAST (1 << 1)
+#define LPC_RXFLTRW_ACCEPTUMULTICAST (1 << 2)
+#define LPC_RXFLTRW_ACCEPTUNICASTHASH (1 << 3)
+#define LPC_RXFLTRW_ACCEPTUMULTICASTHASH (1 << 4)
+#define LPC_RXFLTRW_ACCEPTPERFECT (1 << 5)
+
+/*
+ * rxfilterctrl register definitions
+ */
+#define LPC_RXFLTRWSTS_MAGICPACKETENWOL (1 << 12)
+#define LPC_RXFLTRWSTS_RXFILTERENWOL (1 << 13)
+
+/*
+ * rxfilterwolstatus/rxfilterwolclear register definitions
+ */
+#define LPC_RXFLTRWSTS_RXFILTERWOL (1 << 7)
+#define LPC_RXFLTRWSTS_MAGICPACKETWOL (1 << 8)
+
+/*
+ * intstatus, intenable, intclear, and Intset shared register
+ * definitions
+ */
+#define LPC_MACINT_RXOVERRUNINTEN (1 << 0)
+#define LPC_MACINT_RXERRORONINT (1 << 1)
+#define LPC_MACINT_RXFINISHEDINTEN (1 << 2)
+#define LPC_MACINT_RXDONEINTEN (1 << 3)
+#define LPC_MACINT_TXUNDERRUNINTEN (1 << 4)
+#define LPC_MACINT_TXERRORINTEN (1 << 5)
+#define LPC_MACINT_TXFINISHEDINTEN (1 << 6)
+#define LPC_MACINT_TXDONEINTEN (1 << 7)
+#define LPC_MACINT_SOFTINTEN (1 << 12)
+#define LPC_MACINT_WAKEUPINTEN (1 << 13)
+
+/*
+ * powerdown register definitions
+ */
+#define LPC_POWERDOWN_MACAHB (1 << 31)
+
+static phy_interface_t lpc_phy_interface_mode(struct device *dev)
+{
+ if (dev && dev->of_node) {
+ const char *mode = of_get_property(dev->of_node,
+ "phy-mode", NULL);
+ if (mode && !strcmp(mode, "mii"))
+ return PHY_INTERFACE_MODE_MII;
+ }
+ return PHY_INTERFACE_MODE_RMII;
+}
+
+static bool use_iram_for_net(struct device *dev)
+{
+ if (dev && dev->of_node)
+ return of_property_read_bool(dev->of_node, "use-iram");
+ return false;
+}
+
+/* Receive Status information word */
+#define RXSTATUS_SIZE 0x000007FF
+#define RXSTATUS_CONTROL (1 << 18)
+#define RXSTATUS_VLAN (1 << 19)
+#define RXSTATUS_FILTER (1 << 20)
+#define RXSTATUS_MULTICAST (1 << 21)
+#define RXSTATUS_BROADCAST (1 << 22)
+#define RXSTATUS_CRC (1 << 23)
+#define RXSTATUS_SYMBOL (1 << 24)
+#define RXSTATUS_LENGTH (1 << 25)
+#define RXSTATUS_RANGE (1 << 26)
+#define RXSTATUS_ALIGN (1 << 27)
+#define RXSTATUS_OVERRUN (1 << 28)
+#define RXSTATUS_NODESC (1 << 29)
+#define RXSTATUS_LAST (1 << 30)
+#define RXSTATUS_ERROR (1 << 31)
+
+#define RXSTATUS_STATUS_ERROR \
+ (RXSTATUS_NODESC | RXSTATUS_OVERRUN | RXSTATUS_ALIGN | \
+ RXSTATUS_RANGE | RXSTATUS_LENGTH | RXSTATUS_SYMBOL | RXSTATUS_CRC)
+
+/* Receive Descriptor control word */
+#define RXDESC_CONTROL_SIZE 0x000007FF
+#define RXDESC_CONTROL_INT (1 << 31)
+
+/* Transmit Status information word */
+#define TXSTATUS_COLLISIONS_GET(x) (((x) >> 21) & 0xF)
+#define TXSTATUS_DEFER (1 << 25)
+#define TXSTATUS_EXCESSDEFER (1 << 26)
+#define TXSTATUS_EXCESSCOLL (1 << 27)
+#define TXSTATUS_LATECOLL (1 << 28)
+#define TXSTATUS_UNDERRUN (1 << 29)
+#define TXSTATUS_NODESC (1 << 30)
+#define TXSTATUS_ERROR (1 << 31)
+
+/* Transmit Descriptor control word */
+#define TXDESC_CONTROL_SIZE 0x000007FF
+#define TXDESC_CONTROL_OVERRIDE (1 << 26)
+#define TXDESC_CONTROL_HUGE (1 << 27)
+#define TXDESC_CONTROL_PAD (1 << 28)
+#define TXDESC_CONTROL_CRC (1 << 29)
+#define TXDESC_CONTROL_LAST (1 << 30)
+#define TXDESC_CONTROL_INT (1 << 31)
+
+/*
+ * Structure of a TX/RX descriptors and RX status
+ */
+struct txrx_desc_t {
+ __le32 packet;
+ __le32 control;
+};
+struct rx_status_t {
+ __le32 statusinfo;
+ __le32 statushashcrc;
+};
+
+/*
+ * Device driver data structure
+ */
+struct netdata_local {
+ struct platform_device *pdev;
+ struct net_device *ndev;
+ struct device_node *phy_node;
+ spinlock_t lock;
+ void __iomem *net_base;
+ u32 msg_enable;
+ unsigned int skblen[ENET_TX_DESC];
+ unsigned int last_tx_idx;
+ unsigned int num_used_tx_buffs;
+ struct mii_bus *mii_bus;
+ struct clk *clk;
+ dma_addr_t dma_buff_base_p;
+ void *dma_buff_base_v;
+ size_t dma_buff_size;
+ struct txrx_desc_t *tx_desc_v;
+ u32 *tx_stat_v;
+ void *tx_buff_v;
+ struct txrx_desc_t *rx_desc_v;
+ struct rx_status_t *rx_stat_v;
+ void *rx_buff_v;
+ int link;
+ int speed;
+ int duplex;
+ struct napi_struct napi;
+};
+
+/*
+ * MAC support functions
+ */
+static void __lpc_set_mac(struct netdata_local *pldat, const u8 *mac)
+{
+ u32 tmp;
+
+ /* Set station address */
+ tmp = mac[0] | ((u32)mac[1] << 8);
+ writel(tmp, LPC_ENET_SA2(pldat->net_base));
+ tmp = mac[2] | ((u32)mac[3] << 8);
+ writel(tmp, LPC_ENET_SA1(pldat->net_base));
+ tmp = mac[4] | ((u32)mac[5] << 8);
+ writel(tmp, LPC_ENET_SA0(pldat->net_base));
+
+ netdev_dbg(pldat->ndev, "Ethernet MAC address %pM\n", mac);
+}
+
+static void __lpc_get_mac(struct netdata_local *pldat, u8 *mac)
+{
+ u32 tmp;
+
+ /* Get station address */
+ tmp = readl(LPC_ENET_SA2(pldat->net_base));
+ mac[0] = tmp & 0xFF;
+ mac[1] = tmp >> 8;
+ tmp = readl(LPC_ENET_SA1(pldat->net_base));
+ mac[2] = tmp & 0xFF;
+ mac[3] = tmp >> 8;
+ tmp = readl(LPC_ENET_SA0(pldat->net_base));
+ mac[4] = tmp & 0xFF;
+ mac[5] = tmp >> 8;
+}
+
+static void __lpc_params_setup(struct netdata_local *pldat)
+{
+ u32 tmp;
+
+ if (pldat->duplex == DUPLEX_FULL) {
+ tmp = readl(LPC_ENET_MAC2(pldat->net_base));
+ tmp |= LPC_MAC2_FULL_DUPLEX;
+ writel(tmp, LPC_ENET_MAC2(pldat->net_base));
+ tmp = readl(LPC_ENET_COMMAND(pldat->net_base));
+ tmp |= LPC_COMMAND_FULLDUPLEX;
+ writel(tmp, LPC_ENET_COMMAND(pldat->net_base));
+ writel(LPC_IPGT_LOAD(0x15), LPC_ENET_IPGT(pldat->net_base));
+ } else {
+ tmp = readl(LPC_ENET_MAC2(pldat->net_base));
+ tmp &= ~LPC_MAC2_FULL_DUPLEX;
+ writel(tmp, LPC_ENET_MAC2(pldat->net_base));
+ tmp = readl(LPC_ENET_COMMAND(pldat->net_base));
+ tmp &= ~LPC_COMMAND_FULLDUPLEX;
+ writel(tmp, LPC_ENET_COMMAND(pldat->net_base));
+ writel(LPC_IPGT_LOAD(0x12), LPC_ENET_IPGT(pldat->net_base));
+ }
+
+ if (pldat->speed == SPEED_100)
+ writel(LPC_SUPP_SPEED, LPC_ENET_SUPP(pldat->net_base));
+ else
+ writel(0, LPC_ENET_SUPP(pldat->net_base));
+}
+
+static void __lpc_eth_reset(struct netdata_local *pldat)
+{
+ /* Reset all MAC logic */
+ writel((LPC_MAC1_RESET_TX | LPC_MAC1_RESET_MCS_TX | LPC_MAC1_RESET_RX |
+ LPC_MAC1_RESET_MCS_RX | LPC_MAC1_SIMULATION_RESET |
+ LPC_MAC1_SOFT_RESET), LPC_ENET_MAC1(pldat->net_base));
+ writel((LPC_COMMAND_REG_RESET | LPC_COMMAND_TXRESET |
+ LPC_COMMAND_RXRESET), LPC_ENET_COMMAND(pldat->net_base));
+}
+
+static int __lpc_mii_mngt_reset(struct netdata_local *pldat)
+{
+ /* Reset MII management hardware */
+ writel(LPC_MCFG_RESET_MII_MGMT, LPC_ENET_MCFG(pldat->net_base));
+
+ /* Setup MII clock to slowest rate with a /28 divider */
+ writel(LPC_MCFG_CLOCK_SELECT(LPC_MCFG_CLOCK_HOST_DIV_28),
+ LPC_ENET_MCFG(pldat->net_base));
+
+ return 0;
+}
+
+static inline phys_addr_t __va_to_pa(void *addr, struct netdata_local *pldat)
+{
+ phys_addr_t phaddr;
+
+ phaddr = addr - pldat->dma_buff_base_v;
+ phaddr += pldat->dma_buff_base_p;
+
+ return phaddr;
+}
+
+static void lpc_eth_enable_int(void __iomem *regbase)
+{
+ writel((LPC_MACINT_RXDONEINTEN | LPC_MACINT_TXDONEINTEN),
+ LPC_ENET_INTENABLE(regbase));
+}
+
+static void lpc_eth_disable_int(void __iomem *regbase)
+{
+ writel(0, LPC_ENET_INTENABLE(regbase));
+}
+
+/* Setup TX/RX descriptors */
+static void __lpc_txrx_desc_setup(struct netdata_local *pldat)
+{
+ u32 *ptxstat;
+ void *tbuff;
+ int i;
+ struct txrx_desc_t *ptxrxdesc;
+ struct rx_status_t *prxstat;
+
+ tbuff = PTR_ALIGN(pldat->dma_buff_base_v, 16);
+
+ /* Setup TX descriptors, status, and buffers */
+ pldat->tx_desc_v = tbuff;
+ tbuff += sizeof(struct txrx_desc_t) * ENET_TX_DESC;
+
+ pldat->tx_stat_v = tbuff;
+ tbuff += sizeof(u32) * ENET_TX_DESC;
+
+ tbuff = PTR_ALIGN(tbuff, 16);
+ pldat->tx_buff_v = tbuff;
+ tbuff += ENET_MAXF_SIZE * ENET_TX_DESC;
+
+ /* Setup RX descriptors, status, and buffers */
+ pldat->rx_desc_v = tbuff;
+ tbuff += sizeof(struct txrx_desc_t) * ENET_RX_DESC;
+
+ tbuff = PTR_ALIGN(tbuff, 16);
+ pldat->rx_stat_v = tbuff;
+ tbuff += sizeof(struct rx_status_t) * ENET_RX_DESC;
+
+ tbuff = PTR_ALIGN(tbuff, 16);
+ pldat->rx_buff_v = tbuff;
+ tbuff += ENET_MAXF_SIZE * ENET_RX_DESC;
+
+ /* Map the TX descriptors to the TX buffers in hardware */
+ for (i = 0; i < ENET_TX_DESC; i++) {
+ ptxstat = &pldat->tx_stat_v[i];
+ ptxrxdesc = &pldat->tx_desc_v[i];
+
+ ptxrxdesc->packet = __va_to_pa(
+ pldat->tx_buff_v + i * ENET_MAXF_SIZE, pldat);
+ ptxrxdesc->control = 0;
+ *ptxstat = 0;
+ }
+
+ /* Map the RX descriptors to the RX buffers in hardware */
+ for (i = 0; i < ENET_RX_DESC; i++) {
+ prxstat = &pldat->rx_stat_v[i];
+ ptxrxdesc = &pldat->rx_desc_v[i];
+
+ ptxrxdesc->packet = __va_to_pa(
+ pldat->rx_buff_v + i * ENET_MAXF_SIZE, pldat);
+ ptxrxdesc->control = RXDESC_CONTROL_INT | (ENET_MAXF_SIZE - 1);
+ prxstat->statusinfo = 0;
+ prxstat->statushashcrc = 0;
+ }
+
+ /* Setup base addresses in hardware to point to buffers and
+ * descriptors
+ */
+ writel((ENET_TX_DESC - 1),
+ LPC_ENET_TXDESCRIPTORNUMBER(pldat->net_base));
+ writel(__va_to_pa(pldat->tx_desc_v, pldat),
+ LPC_ENET_TXDESCRIPTOR(pldat->net_base));
+ writel(__va_to_pa(pldat->tx_stat_v, pldat),
+ LPC_ENET_TXSTATUS(pldat->net_base));
+ writel((ENET_RX_DESC - 1),
+ LPC_ENET_RXDESCRIPTORNUMBER(pldat->net_base));
+ writel(__va_to_pa(pldat->rx_desc_v, pldat),
+ LPC_ENET_RXDESCRIPTOR(pldat->net_base));
+ writel(__va_to_pa(pldat->rx_stat_v, pldat),
+ LPC_ENET_RXSTATUS(pldat->net_base));
+}
+
+static void __lpc_eth_init(struct netdata_local *pldat)
+{
+ u32 tmp;
+
+ /* Disable controller and reset */
+ tmp = readl(LPC_ENET_COMMAND(pldat->net_base));
+ tmp &= ~LPC_COMMAND_RXENABLE | LPC_COMMAND_TXENABLE;
+ writel(tmp, LPC_ENET_COMMAND(pldat->net_base));
+ tmp = readl(LPC_ENET_MAC1(pldat->net_base));
+ tmp &= ~LPC_MAC1_RECV_ENABLE;
+ writel(tmp, LPC_ENET_MAC1(pldat->net_base));
+
+ /* Initial MAC setup */
+ writel(LPC_MAC1_PASS_ALL_RX_FRAMES, LPC_ENET_MAC1(pldat->net_base));
+ writel((LPC_MAC2_PAD_CRC_ENABLE | LPC_MAC2_CRC_ENABLE),
+ LPC_ENET_MAC2(pldat->net_base));
+ writel(ENET_MAXF_SIZE, LPC_ENET_MAXF(pldat->net_base));
+
+ /* Collision window, gap */
+ writel((LPC_CLRT_LOAD_RETRY_MAX(0xF) |
+ LPC_CLRT_LOAD_COLLISION_WINDOW(0x37)),
+ LPC_ENET_CLRT(pldat->net_base));
+ writel(LPC_IPGR_LOAD_PART2(0x12), LPC_ENET_IPGR(pldat->net_base));
+
+ if (lpc_phy_interface_mode(&pldat->pdev->dev) == PHY_INTERFACE_MODE_MII)
+ writel(LPC_COMMAND_PASSRUNTFRAME,
+ LPC_ENET_COMMAND(pldat->net_base));
+ else {
+ writel((LPC_COMMAND_PASSRUNTFRAME | LPC_COMMAND_RMII),
+ LPC_ENET_COMMAND(pldat->net_base));
+ writel(LPC_SUPP_RESET_RMII, LPC_ENET_SUPP(pldat->net_base));
+ }
+
+ __lpc_params_setup(pldat);
+
+ /* Setup TX and RX descriptors */
+ __lpc_txrx_desc_setup(pldat);
+
+ /* Setup packet filtering */
+ writel((LPC_RXFLTRW_ACCEPTUBROADCAST | LPC_RXFLTRW_ACCEPTPERFECT),
+ LPC_ENET_RXFILTER_CTRL(pldat->net_base));
+
+ /* Get the next TX buffer output index */
+ pldat->num_used_tx_buffs = 0;
+ pldat->last_tx_idx =
+ readl(LPC_ENET_TXCONSUMEINDEX(pldat->net_base));
+
+ /* Clear and enable interrupts */
+ writel(0xFFFF, LPC_ENET_INTCLEAR(pldat->net_base));
+ smp_wmb();
+ lpc_eth_enable_int(pldat->net_base);
+
+ /* Enable controller */
+ tmp = readl(LPC_ENET_COMMAND(pldat->net_base));
+ tmp |= LPC_COMMAND_RXENABLE | LPC_COMMAND_TXENABLE;
+ writel(tmp, LPC_ENET_COMMAND(pldat->net_base));
+ tmp = readl(LPC_ENET_MAC1(pldat->net_base));
+ tmp |= LPC_MAC1_RECV_ENABLE;
+ writel(tmp, LPC_ENET_MAC1(pldat->net_base));
+}
+
+static void __lpc_eth_shutdown(struct netdata_local *pldat)
+{
+ /* Reset ethernet and power down PHY */
+ __lpc_eth_reset(pldat);
+ writel(0, LPC_ENET_MAC1(pldat->net_base));
+ writel(0, LPC_ENET_MAC2(pldat->net_base));
+}
+
+/*
+ * MAC<--->PHY support functions
+ */
+static int lpc_mdio_read(struct mii_bus *bus, int phy_id, int phyreg)
+{
+ struct netdata_local *pldat = bus->priv;
+ unsigned long timeout = jiffies + msecs_to_jiffies(100);
+ int lps;
+
+ writel(((phy_id << 8) | phyreg), LPC_ENET_MADR(pldat->net_base));
+ writel(LPC_MCMD_READ, LPC_ENET_MCMD(pldat->net_base));
+
+ /* Wait for unbusy status */
+ while (readl(LPC_ENET_MIND(pldat->net_base)) & LPC_MIND_BUSY) {
+ if (time_after(jiffies, timeout))
+ return -EIO;
+ cpu_relax();
+ }
+
+ lps = readl(LPC_ENET_MRDD(pldat->net_base));
+ writel(0, LPC_ENET_MCMD(pldat->net_base));
+
+ return lps;
+}
+
+static int lpc_mdio_write(struct mii_bus *bus, int phy_id, int phyreg,
+ u16 phydata)
+{
+ struct netdata_local *pldat = bus->priv;
+ unsigned long timeout = jiffies + msecs_to_jiffies(100);
+
+ writel(((phy_id << 8) | phyreg), LPC_ENET_MADR(pldat->net_base));
+ writel(phydata, LPC_ENET_MWTD(pldat->net_base));
+
+ /* Wait for completion */
+ while (readl(LPC_ENET_MIND(pldat->net_base)) & LPC_MIND_BUSY) {
+ if (time_after(jiffies, timeout))
+ return -EIO;
+ cpu_relax();
+ }
+
+ return 0;
+}
+
+static int lpc_mdio_reset(struct mii_bus *bus)
+{
+ return __lpc_mii_mngt_reset((struct netdata_local *)bus->priv);
+}
+
+static void lpc_handle_link_change(struct net_device *ndev)
+{
+ struct netdata_local *pldat = netdev_priv(ndev);
+ struct phy_device *phydev = ndev->phydev;
+ unsigned long flags;
+
+ bool status_change = false;
+
+ spin_lock_irqsave(&pldat->lock, flags);
+
+ if (phydev->link) {
+ if ((pldat->speed != phydev->speed) ||
+ (pldat->duplex != phydev->duplex)) {
+ pldat->speed = phydev->speed;
+ pldat->duplex = phydev->duplex;
+ status_change = true;
+ }
+ }
+
+ if (phydev->link != pldat->link) {
+ if (!phydev->link) {
+ pldat->speed = 0;
+ pldat->duplex = -1;
+ }
+ pldat->link = phydev->link;
+
+ status_change = true;
+ }
+
+ spin_unlock_irqrestore(&pldat->lock, flags);
+
+ if (status_change)
+ __lpc_params_setup(pldat);
+}
+
+static int lpc_mii_probe(struct net_device *ndev)
+{
+ struct netdata_local *pldat = netdev_priv(ndev);
+ struct phy_device *phydev;
+
+ /* Attach to the PHY */
+ if (lpc_phy_interface_mode(&pldat->pdev->dev) == PHY_INTERFACE_MODE_MII)
+ netdev_info(ndev, "using MII interface\n");
+ else
+ netdev_info(ndev, "using RMII interface\n");
+
+ if (pldat->phy_node)
+ phydev = of_phy_find_device(pldat->phy_node);
+ else
+ phydev = phy_find_first(pldat->mii_bus);
+ if (!phydev) {
+ netdev_err(ndev, "no PHY found\n");
+ return -ENODEV;
+ }
+
+ phydev = phy_connect(ndev, phydev_name(phydev),
+ &lpc_handle_link_change,
+ lpc_phy_interface_mode(&pldat->pdev->dev));
+ if (IS_ERR(phydev)) {
+ netdev_err(ndev, "Could not attach to PHY\n");
+ return PTR_ERR(phydev);
+ }
+
+ phy_set_max_speed(phydev, SPEED_100);
+
+ pldat->link = 0;
+ pldat->speed = 0;
+ pldat->duplex = -1;
+
+ phy_attached_info(phydev);
+
+ return 0;
+}
+
+static int lpc_mii_init(struct netdata_local *pldat)
+{
+ struct device_node *node;
+ int err = -ENXIO;
+
+ pldat->mii_bus = mdiobus_alloc();
+ if (!pldat->mii_bus) {
+ err = -ENOMEM;
+ goto err_out;
+ }
+
+ /* Setup MII mode */
+ if (lpc_phy_interface_mode(&pldat->pdev->dev) == PHY_INTERFACE_MODE_MII)
+ writel(LPC_COMMAND_PASSRUNTFRAME,
+ LPC_ENET_COMMAND(pldat->net_base));
+ else {
+ writel((LPC_COMMAND_PASSRUNTFRAME | LPC_COMMAND_RMII),
+ LPC_ENET_COMMAND(pldat->net_base));
+ writel(LPC_SUPP_RESET_RMII, LPC_ENET_SUPP(pldat->net_base));
+ }
+
+ pldat->mii_bus->name = "lpc_mii_bus";
+ pldat->mii_bus->read = &lpc_mdio_read;
+ pldat->mii_bus->write = &lpc_mdio_write;
+ pldat->mii_bus->reset = &lpc_mdio_reset;
+ snprintf(pldat->mii_bus->id, MII_BUS_ID_SIZE, "%s-%x",
+ pldat->pdev->name, pldat->pdev->id);
+ pldat->mii_bus->priv = pldat;
+ pldat->mii_bus->parent = &pldat->pdev->dev;
+
+ node = of_get_child_by_name(pldat->pdev->dev.of_node, "mdio");
+ err = of_mdiobus_register(pldat->mii_bus, node);
+ of_node_put(node);
+ if (err)
+ goto err_out_unregister_bus;
+
+ err = lpc_mii_probe(pldat->ndev);
+ if (err)
+ goto err_out_unregister_bus;
+
+ return 0;
+
+err_out_unregister_bus:
+ mdiobus_unregister(pldat->mii_bus);
+ mdiobus_free(pldat->mii_bus);
+err_out:
+ return err;
+}
+
+static void __lpc_handle_xmit(struct net_device *ndev)
+{
+ struct netdata_local *pldat = netdev_priv(ndev);
+ u32 txcidx, *ptxstat, txstat;
+
+ txcidx = readl(LPC_ENET_TXCONSUMEINDEX(pldat->net_base));
+ while (pldat->last_tx_idx != txcidx) {
+ unsigned int skblen = pldat->skblen[pldat->last_tx_idx];
+
+ /* A buffer is available, get buffer status */
+ ptxstat = &pldat->tx_stat_v[pldat->last_tx_idx];
+ txstat = *ptxstat;
+
+ /* Next buffer and decrement used buffer counter */
+ pldat->num_used_tx_buffs--;
+ pldat->last_tx_idx++;
+ if (pldat->last_tx_idx >= ENET_TX_DESC)
+ pldat->last_tx_idx = 0;
+
+ /* Update collision counter */
+ ndev->stats.collisions += TXSTATUS_COLLISIONS_GET(txstat);
+
+ /* Any errors occurred? */
+ if (txstat & TXSTATUS_ERROR) {
+ if (txstat & TXSTATUS_UNDERRUN) {
+ /* FIFO underrun */
+ ndev->stats.tx_fifo_errors++;
+ }
+ if (txstat & TXSTATUS_LATECOLL) {
+ /* Late collision */
+ ndev->stats.tx_aborted_errors++;
+ }
+ if (txstat & TXSTATUS_EXCESSCOLL) {
+ /* Excessive collision */
+ ndev->stats.tx_aborted_errors++;
+ }
+ if (txstat & TXSTATUS_EXCESSDEFER) {
+ /* Defer limit */
+ ndev->stats.tx_aborted_errors++;
+ }
+ ndev->stats.tx_errors++;
+ } else {
+ /* Update stats */
+ ndev->stats.tx_packets++;
+ ndev->stats.tx_bytes += skblen;
+ }
+
+ txcidx = readl(LPC_ENET_TXCONSUMEINDEX(pldat->net_base));
+ }
+
+ if (pldat->num_used_tx_buffs <= ENET_TX_DESC/2) {
+ if (netif_queue_stopped(ndev))
+ netif_wake_queue(ndev);
+ }
+}
+
+static int __lpc_handle_recv(struct net_device *ndev, int budget)
+{
+ struct netdata_local *pldat = netdev_priv(ndev);
+ struct sk_buff *skb;
+ u32 rxconsidx, len, ethst;
+ struct rx_status_t *prxstat;
+ int rx_done = 0;
+
+ /* Get the current RX buffer indexes */
+ rxconsidx = readl(LPC_ENET_RXCONSUMEINDEX(pldat->net_base));
+ while (rx_done < budget && rxconsidx !=
+ readl(LPC_ENET_RXPRODUCEINDEX(pldat->net_base))) {
+ /* Get pointer to receive status */
+ prxstat = &pldat->rx_stat_v[rxconsidx];
+ len = (prxstat->statusinfo & RXSTATUS_SIZE) + 1;
+
+ /* Status error? */
+ ethst = prxstat->statusinfo;
+ if ((ethst & (RXSTATUS_ERROR | RXSTATUS_STATUS_ERROR)) ==
+ (RXSTATUS_ERROR | RXSTATUS_RANGE))
+ ethst &= ~RXSTATUS_ERROR;
+
+ if (ethst & RXSTATUS_ERROR) {
+ int si = prxstat->statusinfo;
+ /* Check statuses */
+ if (si & RXSTATUS_OVERRUN) {
+ /* Overrun error */
+ ndev->stats.rx_fifo_errors++;
+ } else if (si & RXSTATUS_CRC) {
+ /* CRC error */
+ ndev->stats.rx_crc_errors++;
+ } else if (si & RXSTATUS_LENGTH) {
+ /* Length error */
+ ndev->stats.rx_length_errors++;
+ } else if (si & RXSTATUS_ERROR) {
+ /* Other error */
+ ndev->stats.rx_length_errors++;
+ }
+ ndev->stats.rx_errors++;
+ } else {
+ /* Packet is good */
+ skb = dev_alloc_skb(len);
+ if (!skb) {
+ ndev->stats.rx_dropped++;
+ } else {
+ /* Copy packet from buffer */
+ skb_put_data(skb,
+ pldat->rx_buff_v + rxconsidx * ENET_MAXF_SIZE,
+ len);
+
+ /* Pass to upper layer */
+ skb->protocol = eth_type_trans(skb, ndev);
+ netif_receive_skb(skb);
+ ndev->stats.rx_packets++;
+ ndev->stats.rx_bytes += len;
+ }
+ }
+
+ /* Increment consume index */
+ rxconsidx = rxconsidx + 1;
+ if (rxconsidx >= ENET_RX_DESC)
+ rxconsidx = 0;
+ writel(rxconsidx,
+ LPC_ENET_RXCONSUMEINDEX(pldat->net_base));
+ rx_done++;
+ }
+
+ return rx_done;
+}
+
+static int lpc_eth_poll(struct napi_struct *napi, int budget)
+{
+ struct netdata_local *pldat = container_of(napi,
+ struct netdata_local, napi);
+ struct net_device *ndev = pldat->ndev;
+ int rx_done = 0;
+ struct netdev_queue *txq = netdev_get_tx_queue(ndev, 0);
+
+ __netif_tx_lock(txq, smp_processor_id());
+ __lpc_handle_xmit(ndev);
+ __netif_tx_unlock(txq);
+ rx_done = __lpc_handle_recv(ndev, budget);
+
+ if (rx_done < budget) {
+ napi_complete_done(napi, rx_done);
+ lpc_eth_enable_int(pldat->net_base);
+ }
+
+ return rx_done;
+}
+
+static irqreturn_t __lpc_eth_interrupt(int irq, void *dev_id)
+{
+ struct net_device *ndev = dev_id;
+ struct netdata_local *pldat = netdev_priv(ndev);
+ u32 tmp;
+
+ spin_lock(&pldat->lock);
+
+ tmp = readl(LPC_ENET_INTSTATUS(pldat->net_base));
+ /* Clear interrupts */
+ writel(tmp, LPC_ENET_INTCLEAR(pldat->net_base));
+
+ lpc_eth_disable_int(pldat->net_base);
+ if (likely(napi_schedule_prep(&pldat->napi)))
+ __napi_schedule(&pldat->napi);
+
+ spin_unlock(&pldat->lock);
+
+ return IRQ_HANDLED;
+}
+
+static int lpc_eth_close(struct net_device *ndev)
+{
+ unsigned long flags;
+ struct netdata_local *pldat = netdev_priv(ndev);
+
+ if (netif_msg_ifdown(pldat))
+ dev_dbg(&pldat->pdev->dev, "shutting down %s\n", ndev->name);
+
+ napi_disable(&pldat->napi);
+ netif_stop_queue(ndev);
+
+ spin_lock_irqsave(&pldat->lock, flags);
+ __lpc_eth_reset(pldat);
+ netif_carrier_off(ndev);
+ writel(0, LPC_ENET_MAC1(pldat->net_base));
+ writel(0, LPC_ENET_MAC2(pldat->net_base));
+ spin_unlock_irqrestore(&pldat->lock, flags);
+
+ if (ndev->phydev)
+ phy_stop(ndev->phydev);
+ clk_disable_unprepare(pldat->clk);
+
+ return 0;
+}
+
+static netdev_tx_t lpc_eth_hard_start_xmit(struct sk_buff *skb,
+ struct net_device *ndev)
+{
+ struct netdata_local *pldat = netdev_priv(ndev);
+ u32 len, txidx;
+ u32 *ptxstat;
+ struct txrx_desc_t *ptxrxdesc;
+
+ len = skb->len;
+
+ spin_lock_irq(&pldat->lock);
+
+ if (pldat->num_used_tx_buffs >= (ENET_TX_DESC - 1)) {
+ /* This function should never be called when there are no
+ * buffers
+ */
+ netif_stop_queue(ndev);
+ spin_unlock_irq(&pldat->lock);
+ WARN(1, "BUG! TX request when no free TX buffers!\n");
+ return NETDEV_TX_BUSY;
+ }
+
+ /* Get the next TX descriptor index */
+ txidx = readl(LPC_ENET_TXPRODUCEINDEX(pldat->net_base));
+
+ /* Setup control for the transfer */
+ ptxstat = &pldat->tx_stat_v[txidx];
+ *ptxstat = 0;
+ ptxrxdesc = &pldat->tx_desc_v[txidx];
+ ptxrxdesc->control =
+ (len - 1) | TXDESC_CONTROL_LAST | TXDESC_CONTROL_INT;
+
+ /* Copy data to the DMA buffer */
+ memcpy(pldat->tx_buff_v + txidx * ENET_MAXF_SIZE, skb->data, len);
+
+ /* Save the buffer and increment the buffer counter */
+ pldat->skblen[txidx] = len;
+ pldat->num_used_tx_buffs++;
+
+ /* Start transmit */
+ txidx++;
+ if (txidx >= ENET_TX_DESC)
+ txidx = 0;
+ writel(txidx, LPC_ENET_TXPRODUCEINDEX(pldat->net_base));
+
+ /* Stop queue if no more TX buffers */
+ if (pldat->num_used_tx_buffs >= (ENET_TX_DESC - 1))
+ netif_stop_queue(ndev);
+
+ spin_unlock_irq(&pldat->lock);
+
+ dev_kfree_skb(skb);
+ return NETDEV_TX_OK;
+}
+
+static int lpc_set_mac_address(struct net_device *ndev, void *p)
+{
+ struct sockaddr *addr = p;
+ struct netdata_local *pldat = netdev_priv(ndev);
+ unsigned long flags;
+
+ if (!is_valid_ether_addr(addr->sa_data))
+ return -EADDRNOTAVAIL;
+ eth_hw_addr_set(ndev, addr->sa_data);
+
+ spin_lock_irqsave(&pldat->lock, flags);
+
+ /* Set station address */
+ __lpc_set_mac(pldat, ndev->dev_addr);
+
+ spin_unlock_irqrestore(&pldat->lock, flags);
+
+ return 0;
+}
+
+static void lpc_eth_set_multicast_list(struct net_device *ndev)
+{
+ struct netdata_local *pldat = netdev_priv(ndev);
+ struct netdev_hw_addr_list *mcptr = &ndev->mc;
+ struct netdev_hw_addr *ha;
+ u32 tmp32, hash_val, hashlo, hashhi;
+ unsigned long flags;
+
+ spin_lock_irqsave(&pldat->lock, flags);
+
+ /* Set station address */
+ __lpc_set_mac(pldat, ndev->dev_addr);
+
+ tmp32 = LPC_RXFLTRW_ACCEPTUBROADCAST | LPC_RXFLTRW_ACCEPTPERFECT;
+
+ if (ndev->flags & IFF_PROMISC)
+ tmp32 |= LPC_RXFLTRW_ACCEPTUNICAST |
+ LPC_RXFLTRW_ACCEPTUMULTICAST;
+ if (ndev->flags & IFF_ALLMULTI)
+ tmp32 |= LPC_RXFLTRW_ACCEPTUMULTICAST;
+
+ if (netdev_hw_addr_list_count(mcptr))
+ tmp32 |= LPC_RXFLTRW_ACCEPTUMULTICASTHASH;
+
+ writel(tmp32, LPC_ENET_RXFILTER_CTRL(pldat->net_base));
+
+
+ /* Set initial hash table */
+ hashlo = 0x0;
+ hashhi = 0x0;
+
+ /* 64 bits : multicast address in hash table */
+ netdev_hw_addr_list_for_each(ha, mcptr) {
+ hash_val = (ether_crc(6, ha->addr) >> 23) & 0x3F;
+
+ if (hash_val >= 32)
+ hashhi |= 1 << (hash_val - 32);
+ else
+ hashlo |= 1 << hash_val;
+ }
+
+ writel(hashlo, LPC_ENET_HASHFILTERL(pldat->net_base));
+ writel(hashhi, LPC_ENET_HASHFILTERH(pldat->net_base));
+
+ spin_unlock_irqrestore(&pldat->lock, flags);
+}
+
+static int lpc_eth_open(struct net_device *ndev)
+{
+ struct netdata_local *pldat = netdev_priv(ndev);
+ int ret;
+
+ if (netif_msg_ifup(pldat))
+ dev_dbg(&pldat->pdev->dev, "enabling %s\n", ndev->name);
+
+ ret = clk_prepare_enable(pldat->clk);
+ if (ret)
+ return ret;
+
+ /* Suspended PHY makes LPC ethernet core block, so resume now */
+ phy_resume(ndev->phydev);
+
+ /* Reset and initialize */
+ __lpc_eth_reset(pldat);
+ __lpc_eth_init(pldat);
+
+ /* schedule a link state check */
+ phy_start(ndev->phydev);
+ netif_start_queue(ndev);
+ napi_enable(&pldat->napi);
+
+ return 0;
+}
+
+/*
+ * Ethtool ops
+ */
+static void lpc_eth_ethtool_getdrvinfo(struct net_device *ndev,
+ struct ethtool_drvinfo *info)
+{
+ strscpy(info->driver, MODNAME, sizeof(info->driver));
+ strscpy(info->version, DRV_VERSION, sizeof(info->version));
+ strscpy(info->bus_info, dev_name(ndev->dev.parent),
+ sizeof(info->bus_info));
+}
+
+static u32 lpc_eth_ethtool_getmsglevel(struct net_device *ndev)
+{
+ struct netdata_local *pldat = netdev_priv(ndev);
+
+ return pldat->msg_enable;
+}
+
+static void lpc_eth_ethtool_setmsglevel(struct net_device *ndev, u32 level)
+{
+ struct netdata_local *pldat = netdev_priv(ndev);
+
+ pldat->msg_enable = level;
+}
+
+static const struct ethtool_ops lpc_eth_ethtool_ops = {
+ .get_drvinfo = lpc_eth_ethtool_getdrvinfo,
+ .get_msglevel = lpc_eth_ethtool_getmsglevel,
+ .set_msglevel = lpc_eth_ethtool_setmsglevel,
+ .get_link = ethtool_op_get_link,
+ .get_link_ksettings = phy_ethtool_get_link_ksettings,
+ .set_link_ksettings = phy_ethtool_set_link_ksettings,
+};
+
+static const struct net_device_ops lpc_netdev_ops = {
+ .ndo_open = lpc_eth_open,
+ .ndo_stop = lpc_eth_close,
+ .ndo_start_xmit = lpc_eth_hard_start_xmit,
+ .ndo_set_rx_mode = lpc_eth_set_multicast_list,
+ .ndo_eth_ioctl = phy_do_ioctl_running,
+ .ndo_set_mac_address = lpc_set_mac_address,
+ .ndo_validate_addr = eth_validate_addr,
+};
+
+static int lpc_eth_drv_probe(struct platform_device *pdev)
+{
+ struct device *dev = &pdev->dev;
+ struct device_node *np = dev->of_node;
+ struct netdata_local *pldat;
+ struct net_device *ndev;
+ dma_addr_t dma_handle;
+ struct resource *res;
+ u8 addr[ETH_ALEN];
+ int irq, ret;
+
+ /* Setup network interface for RMII or MII mode */
+ lpc32xx_set_phy_interface_mode(lpc_phy_interface_mode(dev));
+
+ /* Get platform resources */
+ res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+ irq = platform_get_irq(pdev, 0);
+ if (!res || irq < 0) {
+ dev_err(dev, "error getting resources.\n");
+ ret = -ENXIO;
+ goto err_exit;
+ }
+
+ /* Allocate net driver data structure */
+ ndev = alloc_etherdev(sizeof(struct netdata_local));
+ if (!ndev) {
+ dev_err(dev, "could not allocate device.\n");
+ ret = -ENOMEM;
+ goto err_exit;
+ }
+
+ SET_NETDEV_DEV(ndev, dev);
+
+ pldat = netdev_priv(ndev);
+ pldat->pdev = pdev;
+ pldat->ndev = ndev;
+
+ spin_lock_init(&pldat->lock);
+
+ /* Save resources */
+ ndev->irq = irq;
+
+ /* Get clock for the device */
+ pldat->clk = clk_get(dev, NULL);
+ if (IS_ERR(pldat->clk)) {
+ dev_err(dev, "error getting clock.\n");
+ ret = PTR_ERR(pldat->clk);
+ goto err_out_free_dev;
+ }
+
+ /* Enable network clock */
+ ret = clk_prepare_enable(pldat->clk);
+ if (ret)
+ goto err_out_clk_put;
+
+ /* Map IO space */
+ pldat->net_base = ioremap(res->start, resource_size(res));
+ if (!pldat->net_base) {
+ dev_err(dev, "failed to map registers\n");
+ ret = -ENOMEM;
+ goto err_out_disable_clocks;
+ }
+ ret = request_irq(ndev->irq, __lpc_eth_interrupt, 0,
+ ndev->name, ndev);
+ if (ret) {
+ dev_err(dev, "error requesting interrupt.\n");
+ goto err_out_iounmap;
+ }
+
+ /* Setup driver functions */
+ ndev->netdev_ops = &lpc_netdev_ops;
+ ndev->ethtool_ops = &lpc_eth_ethtool_ops;
+ ndev->watchdog_timeo = msecs_to_jiffies(2500);
+
+ /* Get size of DMA buffers/descriptors region */
+ pldat->dma_buff_size = (ENET_TX_DESC + ENET_RX_DESC) * (ENET_MAXF_SIZE +
+ sizeof(struct txrx_desc_t) + sizeof(struct rx_status_t));
+
+ if (use_iram_for_net(dev)) {
+ if (pldat->dma_buff_size >
+ lpc32xx_return_iram(&pldat->dma_buff_base_v, &dma_handle)) {
+ pldat->dma_buff_base_v = NULL;
+ pldat->dma_buff_size = 0;
+ netdev_err(ndev,
+ "IRAM not big enough for net buffers, using SDRAM instead.\n");
+ }
+ }
+
+ if (pldat->dma_buff_base_v == NULL) {
+ ret = dma_coerce_mask_and_coherent(dev, DMA_BIT_MASK(32));
+ if (ret)
+ goto err_out_free_irq;
+
+ pldat->dma_buff_size = PAGE_ALIGN(pldat->dma_buff_size);
+
+ /* Allocate a chunk of memory for the DMA ethernet buffers
+ * and descriptors
+ */
+ pldat->dma_buff_base_v =
+ dma_alloc_coherent(dev,
+ pldat->dma_buff_size, &dma_handle,
+ GFP_KERNEL);
+ if (pldat->dma_buff_base_v == NULL) {
+ ret = -ENOMEM;
+ goto err_out_free_irq;
+ }
+ }
+ pldat->dma_buff_base_p = dma_handle;
+
+ netdev_dbg(ndev, "IO address space :%pR\n", res);
+ netdev_dbg(ndev, "IO address size :%zd\n",
+ (size_t)resource_size(res));
+ netdev_dbg(ndev, "IO address (mapped) :0x%p\n",
+ pldat->net_base);
+ netdev_dbg(ndev, "IRQ number :%d\n", ndev->irq);
+ netdev_dbg(ndev, "DMA buffer size :%zd\n", pldat->dma_buff_size);
+ netdev_dbg(ndev, "DMA buffer P address :%pad\n",
+ &pldat->dma_buff_base_p);
+ netdev_dbg(ndev, "DMA buffer V address :0x%p\n",
+ pldat->dma_buff_base_v);
+
+ pldat->phy_node = of_parse_phandle(np, "phy-handle", 0);
+
+ /* Get MAC address from current HW setting (POR state is all zeros) */
+ __lpc_get_mac(pldat, addr);
+ eth_hw_addr_set(ndev, addr);
+
+ if (!is_valid_ether_addr(ndev->dev_addr)) {
+ of_get_ethdev_address(np, ndev);
+ }
+ if (!is_valid_ether_addr(ndev->dev_addr))
+ eth_hw_addr_random(ndev);
+
+ /* then shut everything down to save power */
+ __lpc_eth_shutdown(pldat);
+
+ /* Set default parameters */
+ pldat->msg_enable = NETIF_MSG_LINK;
+
+ /* Force an MII interface reset and clock setup */
+ __lpc_mii_mngt_reset(pldat);
+
+ /* Force default PHY interface setup in chip, this will probably be
+ * changed by the PHY driver
+ */
+ pldat->link = 0;
+ pldat->speed = 100;
+ pldat->duplex = DUPLEX_FULL;
+ __lpc_params_setup(pldat);
+
+ netif_napi_add_weight(ndev, &pldat->napi, lpc_eth_poll, NAPI_WEIGHT);
+
+ ret = register_netdev(ndev);
+ if (ret) {
+ dev_err(dev, "Cannot register net device, aborting.\n");
+ goto err_out_dma_unmap;
+ }
+ platform_set_drvdata(pdev, ndev);
+
+ ret = lpc_mii_init(pldat);
+ if (ret)
+ goto err_out_unregister_netdev;
+
+ netdev_info(ndev, "LPC mac at 0x%08lx irq %d\n",
+ (unsigned long)res->start, ndev->irq);
+
+ device_init_wakeup(dev, 1);
+ device_set_wakeup_enable(dev, 0);
+
+ return 0;
+
+err_out_unregister_netdev:
+ unregister_netdev(ndev);
+err_out_dma_unmap:
+ if (!use_iram_for_net(dev) ||
+ pldat->dma_buff_size > lpc32xx_return_iram(NULL, NULL))
+ dma_free_coherent(dev, pldat->dma_buff_size,
+ pldat->dma_buff_base_v,
+ pldat->dma_buff_base_p);
+err_out_free_irq:
+ free_irq(ndev->irq, ndev);
+err_out_iounmap:
+ iounmap(pldat->net_base);
+err_out_disable_clocks:
+ clk_disable_unprepare(pldat->clk);
+err_out_clk_put:
+ clk_put(pldat->clk);
+err_out_free_dev:
+ free_netdev(ndev);
+err_exit:
+ pr_err("%s: not found (%d).\n", MODNAME, ret);
+ return ret;
+}
+
+static int lpc_eth_drv_remove(struct platform_device *pdev)
+{
+ struct net_device *ndev = platform_get_drvdata(pdev);
+ struct netdata_local *pldat = netdev_priv(ndev);
+
+ unregister_netdev(ndev);
+
+ if (!use_iram_for_net(&pldat->pdev->dev) ||
+ pldat->dma_buff_size > lpc32xx_return_iram(NULL, NULL))
+ dma_free_coherent(&pldat->pdev->dev, pldat->dma_buff_size,
+ pldat->dma_buff_base_v,
+ pldat->dma_buff_base_p);
+ free_irq(ndev->irq, ndev);
+ iounmap(pldat->net_base);
+ mdiobus_unregister(pldat->mii_bus);
+ mdiobus_free(pldat->mii_bus);
+ clk_disable_unprepare(pldat->clk);
+ clk_put(pldat->clk);
+ free_netdev(ndev);
+
+ return 0;
+}
+
+#ifdef CONFIG_PM
+static int lpc_eth_drv_suspend(struct platform_device *pdev,
+ pm_message_t state)
+{
+ struct net_device *ndev = platform_get_drvdata(pdev);
+ struct netdata_local *pldat = netdev_priv(ndev);
+
+ if (device_may_wakeup(&pdev->dev))
+ enable_irq_wake(ndev->irq);
+
+ if (ndev) {
+ if (netif_running(ndev)) {
+ netif_device_detach(ndev);
+ __lpc_eth_shutdown(pldat);
+ clk_disable_unprepare(pldat->clk);
+
+ /*
+ * Reset again now clock is disable to be sure
+ * EMC_MDC is down
+ */
+ __lpc_eth_reset(pldat);
+ }
+ }
+
+ return 0;
+}
+
+static int lpc_eth_drv_resume(struct platform_device *pdev)
+{
+ struct net_device *ndev = platform_get_drvdata(pdev);
+ struct netdata_local *pldat;
+ int ret;
+
+ if (device_may_wakeup(&pdev->dev))
+ disable_irq_wake(ndev->irq);
+
+ if (ndev) {
+ if (netif_running(ndev)) {
+ pldat = netdev_priv(ndev);
+
+ /* Enable interface clock */
+ ret = clk_enable(pldat->clk);
+ if (ret)
+ return ret;
+
+ /* Reset and initialize */
+ __lpc_eth_reset(pldat);
+ __lpc_eth_init(pldat);
+
+ netif_device_attach(ndev);
+ }
+ }
+
+ return 0;
+}
+#endif
+
+static const struct of_device_id lpc_eth_match[] = {
+ { .compatible = "nxp,lpc-eth" },
+ { }
+};
+MODULE_DEVICE_TABLE(of, lpc_eth_match);
+
+static struct platform_driver lpc_eth_driver = {
+ .probe = lpc_eth_drv_probe,
+ .remove = lpc_eth_drv_remove,
+#ifdef CONFIG_PM
+ .suspend = lpc_eth_drv_suspend,
+ .resume = lpc_eth_drv_resume,
+#endif
+ .driver = {
+ .name = MODNAME,
+ .of_match_table = lpc_eth_match,
+ },
+};
+
+module_platform_driver(lpc_eth_driver);
+
+MODULE_AUTHOR("Kevin Wells <kevin.wells@nxp.com>");
+MODULE_AUTHOR("Roland Stigge <stigge@antcom.de>");
+MODULE_DESCRIPTION("LPC Ethernet Driver");
+MODULE_LICENSE("GPL");