| /* |
| * QLogic QLA3xxx NIC HBA Driver |
| * Copyright (c) 2003-2006 QLogic Corporation |
| * |
| * See LICENSE.qla3xxx for copyright and licensing details. |
| */ |
| |
| #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt |
| |
| #include <linux/kernel.h> |
| #include <linux/init.h> |
| #include <linux/types.h> |
| #include <linux/module.h> |
| #include <linux/list.h> |
| #include <linux/pci.h> |
| #include <linux/dma-mapping.h> |
| #include <linux/sched.h> |
| #include <linux/slab.h> |
| #include <linux/dmapool.h> |
| #include <linux/mempool.h> |
| #include <linux/spinlock.h> |
| #include <linux/kthread.h> |
| #include <linux/interrupt.h> |
| #include <linux/errno.h> |
| #include <linux/ioport.h> |
| #include <linux/ip.h> |
| #include <linux/in.h> |
| #include <linux/if_arp.h> |
| #include <linux/if_ether.h> |
| #include <linux/netdevice.h> |
| #include <linux/etherdevice.h> |
| #include <linux/ethtool.h> |
| #include <linux/skbuff.h> |
| #include <linux/rtnetlink.h> |
| #include <linux/if_vlan.h> |
| #include <linux/delay.h> |
| #include <linux/mm.h> |
| |
| #include "qla3xxx.h" |
| |
| #define DRV_NAME "qla3xxx" |
| #define DRV_STRING "QLogic ISP3XXX Network Driver" |
| #define DRV_VERSION "v2.03.00-k5" |
| |
| static const char ql3xxx_driver_name[] = DRV_NAME; |
| static const char ql3xxx_driver_version[] = DRV_VERSION; |
| |
| #define TIMED_OUT_MSG \ |
| "Timed out waiting for management port to get free before issuing command\n" |
| |
| MODULE_AUTHOR("QLogic Corporation"); |
| MODULE_DESCRIPTION("QLogic ISP3XXX Network Driver " DRV_VERSION " "); |
| MODULE_LICENSE("GPL"); |
| MODULE_VERSION(DRV_VERSION); |
| |
| static const u32 default_msg |
| = NETIF_MSG_DRV | NETIF_MSG_PROBE | NETIF_MSG_LINK |
| | NETIF_MSG_IFUP | NETIF_MSG_IFDOWN; |
| |
| static int debug = -1; /* defaults above */ |
| module_param(debug, int, 0); |
| MODULE_PARM_DESC(debug, "Debug level (0=none,...,16=all)"); |
| |
| static int msi; |
| module_param(msi, int, 0); |
| MODULE_PARM_DESC(msi, "Turn on Message Signaled Interrupts."); |
| |
| static DEFINE_PCI_DEVICE_TABLE(ql3xxx_pci_tbl) = { |
| {PCI_DEVICE(PCI_VENDOR_ID_QLOGIC, QL3022_DEVICE_ID)}, |
| {PCI_DEVICE(PCI_VENDOR_ID_QLOGIC, QL3032_DEVICE_ID)}, |
| /* required last entry */ |
| {0,} |
| }; |
| |
| MODULE_DEVICE_TABLE(pci, ql3xxx_pci_tbl); |
| |
| /* |
| * These are the known PHY's which are used |
| */ |
| enum PHY_DEVICE_TYPE { |
| PHY_TYPE_UNKNOWN = 0, |
| PHY_VITESSE_VSC8211, |
| PHY_AGERE_ET1011C, |
| MAX_PHY_DEV_TYPES |
| }; |
| |
| struct PHY_DEVICE_INFO { |
| const enum PHY_DEVICE_TYPE phyDevice; |
| const u32 phyIdOUI; |
| const u16 phyIdModel; |
| const char *name; |
| }; |
| |
| static const struct PHY_DEVICE_INFO PHY_DEVICES[] = { |
| {PHY_TYPE_UNKNOWN, 0x000000, 0x0, "PHY_TYPE_UNKNOWN"}, |
| {PHY_VITESSE_VSC8211, 0x0003f1, 0xb, "PHY_VITESSE_VSC8211"}, |
| {PHY_AGERE_ET1011C, 0x00a0bc, 0x1, "PHY_AGERE_ET1011C"}, |
| }; |
| |
| |
| /* |
| * Caller must take hw_lock. |
| */ |
| static int ql_sem_spinlock(struct ql3_adapter *qdev, |
| u32 sem_mask, u32 sem_bits) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 value; |
| unsigned int seconds = 3; |
| |
| do { |
| writel((sem_mask | sem_bits), |
| &port_regs->CommonRegs.semaphoreReg); |
| value = readl(&port_regs->CommonRegs.semaphoreReg); |
| if ((value & (sem_mask >> 16)) == sem_bits) |
| return 0; |
| ssleep(1); |
| } while (--seconds); |
| return -1; |
| } |
| |
| static void ql_sem_unlock(struct ql3_adapter *qdev, u32 sem_mask) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| writel(sem_mask, &port_regs->CommonRegs.semaphoreReg); |
| readl(&port_regs->CommonRegs.semaphoreReg); |
| } |
| |
| static int ql_sem_lock(struct ql3_adapter *qdev, u32 sem_mask, u32 sem_bits) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 value; |
| |
| writel((sem_mask | sem_bits), &port_regs->CommonRegs.semaphoreReg); |
| value = readl(&port_regs->CommonRegs.semaphoreReg); |
| return ((value & (sem_mask >> 16)) == sem_bits); |
| } |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static int ql_wait_for_drvr_lock(struct ql3_adapter *qdev) |
| { |
| int i = 0; |
| |
| while (i < 10) { |
| if (i) |
| ssleep(1); |
| |
| if (ql_sem_lock(qdev, |
| QL_DRVR_SEM_MASK, |
| (QL_RESOURCE_BITS_BASE_CODE | (qdev->mac_index) |
| * 2) << 1)) { |
| netdev_printk(KERN_DEBUG, qdev->ndev, |
| "driver lock acquired\n"); |
| return 1; |
| } |
| } |
| |
| netdev_err(qdev->ndev, "Timed out waiting for driver lock...\n"); |
| return 0; |
| } |
| |
| static void ql_set_register_page(struct ql3_adapter *qdev, u32 page) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| |
| writel(((ISP_CONTROL_NP_MASK << 16) | page), |
| &port_regs->CommonRegs.ispControlStatus); |
| readl(&port_regs->CommonRegs.ispControlStatus); |
| qdev->current_page = page; |
| } |
| |
| static u32 ql_read_common_reg_l(struct ql3_adapter *qdev, u32 __iomem *reg) |
| { |
| u32 value; |
| unsigned long hw_flags; |
| |
| spin_lock_irqsave(&qdev->hw_lock, hw_flags); |
| value = readl(reg); |
| spin_unlock_irqrestore(&qdev->hw_lock, hw_flags); |
| |
| return value; |
| } |
| |
| static u32 ql_read_common_reg(struct ql3_adapter *qdev, u32 __iomem *reg) |
| { |
| return readl(reg); |
| } |
| |
| static u32 ql_read_page0_reg_l(struct ql3_adapter *qdev, u32 __iomem *reg) |
| { |
| u32 value; |
| unsigned long hw_flags; |
| |
| spin_lock_irqsave(&qdev->hw_lock, hw_flags); |
| |
| if (qdev->current_page != 0) |
| ql_set_register_page(qdev, 0); |
| value = readl(reg); |
| |
| spin_unlock_irqrestore(&qdev->hw_lock, hw_flags); |
| return value; |
| } |
| |
| static u32 ql_read_page0_reg(struct ql3_adapter *qdev, u32 __iomem *reg) |
| { |
| if (qdev->current_page != 0) |
| ql_set_register_page(qdev, 0); |
| return readl(reg); |
| } |
| |
| static void ql_write_common_reg_l(struct ql3_adapter *qdev, |
| u32 __iomem *reg, u32 value) |
| { |
| unsigned long hw_flags; |
| |
| spin_lock_irqsave(&qdev->hw_lock, hw_flags); |
| writel(value, reg); |
| readl(reg); |
| spin_unlock_irqrestore(&qdev->hw_lock, hw_flags); |
| } |
| |
| static void ql_write_common_reg(struct ql3_adapter *qdev, |
| u32 __iomem *reg, u32 value) |
| { |
| writel(value, reg); |
| readl(reg); |
| } |
| |
| static void ql_write_nvram_reg(struct ql3_adapter *qdev, |
| u32 __iomem *reg, u32 value) |
| { |
| writel(value, reg); |
| readl(reg); |
| udelay(1); |
| } |
| |
| static void ql_write_page0_reg(struct ql3_adapter *qdev, |
| u32 __iomem *reg, u32 value) |
| { |
| if (qdev->current_page != 0) |
| ql_set_register_page(qdev, 0); |
| writel(value, reg); |
| readl(reg); |
| } |
| |
| /* |
| * Caller holds hw_lock. Only called during init. |
| */ |
| static void ql_write_page1_reg(struct ql3_adapter *qdev, |
| u32 __iomem *reg, u32 value) |
| { |
| if (qdev->current_page != 1) |
| ql_set_register_page(qdev, 1); |
| writel(value, reg); |
| readl(reg); |
| } |
| |
| /* |
| * Caller holds hw_lock. Only called during init. |
| */ |
| static void ql_write_page2_reg(struct ql3_adapter *qdev, |
| u32 __iomem *reg, u32 value) |
| { |
| if (qdev->current_page != 2) |
| ql_set_register_page(qdev, 2); |
| writel(value, reg); |
| readl(reg); |
| } |
| |
| static void ql_disable_interrupts(struct ql3_adapter *qdev) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| |
| ql_write_common_reg_l(qdev, &port_regs->CommonRegs.ispInterruptMaskReg, |
| (ISP_IMR_ENABLE_INT << 16)); |
| |
| } |
| |
| static void ql_enable_interrupts(struct ql3_adapter *qdev) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| |
| ql_write_common_reg_l(qdev, &port_regs->CommonRegs.ispInterruptMaskReg, |
| ((0xff << 16) | ISP_IMR_ENABLE_INT)); |
| |
| } |
| |
| static void ql_release_to_lrg_buf_free_list(struct ql3_adapter *qdev, |
| struct ql_rcv_buf_cb *lrg_buf_cb) |
| { |
| dma_addr_t map; |
| int err; |
| lrg_buf_cb->next = NULL; |
| |
| if (qdev->lrg_buf_free_tail == NULL) { /* The list is empty */ |
| qdev->lrg_buf_free_head = qdev->lrg_buf_free_tail = lrg_buf_cb; |
| } else { |
| qdev->lrg_buf_free_tail->next = lrg_buf_cb; |
| qdev->lrg_buf_free_tail = lrg_buf_cb; |
| } |
| |
| if (!lrg_buf_cb->skb) { |
| lrg_buf_cb->skb = netdev_alloc_skb(qdev->ndev, |
| qdev->lrg_buffer_len); |
| if (unlikely(!lrg_buf_cb->skb)) { |
| netdev_err(qdev->ndev, "failed netdev_alloc_skb()\n"); |
| qdev->lrg_buf_skb_check++; |
| } else { |
| /* |
| * We save some space to copy the ethhdr from first |
| * buffer |
| */ |
| skb_reserve(lrg_buf_cb->skb, QL_HEADER_SPACE); |
| map = pci_map_single(qdev->pdev, |
| lrg_buf_cb->skb->data, |
| qdev->lrg_buffer_len - |
| QL_HEADER_SPACE, |
| PCI_DMA_FROMDEVICE); |
| err = pci_dma_mapping_error(qdev->pdev, map); |
| if (err) { |
| netdev_err(qdev->ndev, |
| "PCI mapping failed with error: %d\n", |
| err); |
| dev_kfree_skb(lrg_buf_cb->skb); |
| lrg_buf_cb->skb = NULL; |
| |
| qdev->lrg_buf_skb_check++; |
| return; |
| } |
| |
| lrg_buf_cb->buf_phy_addr_low = |
| cpu_to_le32(LS_64BITS(map)); |
| lrg_buf_cb->buf_phy_addr_high = |
| cpu_to_le32(MS_64BITS(map)); |
| dma_unmap_addr_set(lrg_buf_cb, mapaddr, map); |
| dma_unmap_len_set(lrg_buf_cb, maplen, |
| qdev->lrg_buffer_len - |
| QL_HEADER_SPACE); |
| } |
| } |
| |
| qdev->lrg_buf_free_count++; |
| } |
| |
| static struct ql_rcv_buf_cb *ql_get_from_lrg_buf_free_list(struct ql3_adapter |
| *qdev) |
| { |
| struct ql_rcv_buf_cb *lrg_buf_cb = qdev->lrg_buf_free_head; |
| |
| if (lrg_buf_cb != NULL) { |
| qdev->lrg_buf_free_head = lrg_buf_cb->next; |
| if (qdev->lrg_buf_free_head == NULL) |
| qdev->lrg_buf_free_tail = NULL; |
| qdev->lrg_buf_free_count--; |
| } |
| |
| return lrg_buf_cb; |
| } |
| |
| static u32 addrBits = EEPROM_NO_ADDR_BITS; |
| static u32 dataBits = EEPROM_NO_DATA_BITS; |
| |
| static void fm93c56a_deselect(struct ql3_adapter *qdev); |
| static void eeprom_readword(struct ql3_adapter *qdev, u32 eepromAddr, |
| unsigned short *value); |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static void fm93c56a_select(struct ql3_adapter *qdev) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 *spir = &port_regs->CommonRegs.serialPortInterfaceReg; |
| |
| qdev->eeprom_cmd_data = AUBURN_EEPROM_CS_1; |
| ql_write_nvram_reg(qdev, spir, ISP_NVRAM_MASK | qdev->eeprom_cmd_data); |
| ql_write_nvram_reg(qdev, spir, |
| ((ISP_NVRAM_MASK << 16) | qdev->eeprom_cmd_data)); |
| } |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static void fm93c56a_cmd(struct ql3_adapter *qdev, u32 cmd, u32 eepromAddr) |
| { |
| int i; |
| u32 mask; |
| u32 dataBit; |
| u32 previousBit; |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 *spir = &port_regs->CommonRegs.serialPortInterfaceReg; |
| |
| /* Clock in a zero, then do the start bit */ |
| ql_write_nvram_reg(qdev, spir, |
| (ISP_NVRAM_MASK | qdev->eeprom_cmd_data | |
| AUBURN_EEPROM_DO_1)); |
| ql_write_nvram_reg(qdev, spir, |
| (ISP_NVRAM_MASK | qdev->eeprom_cmd_data | |
| AUBURN_EEPROM_DO_1 | AUBURN_EEPROM_CLK_RISE)); |
| ql_write_nvram_reg(qdev, spir, |
| (ISP_NVRAM_MASK | qdev->eeprom_cmd_data | |
| AUBURN_EEPROM_DO_1 | AUBURN_EEPROM_CLK_FALL)); |
| |
| mask = 1 << (FM93C56A_CMD_BITS - 1); |
| /* Force the previous data bit to be different */ |
| previousBit = 0xffff; |
| for (i = 0; i < FM93C56A_CMD_BITS; i++) { |
| dataBit = (cmd & mask) |
| ? AUBURN_EEPROM_DO_1 |
| : AUBURN_EEPROM_DO_0; |
| if (previousBit != dataBit) { |
| /* If the bit changed, change the DO state to match */ |
| ql_write_nvram_reg(qdev, spir, |
| (ISP_NVRAM_MASK | |
| qdev->eeprom_cmd_data | dataBit)); |
| previousBit = dataBit; |
| } |
| ql_write_nvram_reg(qdev, spir, |
| (ISP_NVRAM_MASK | qdev->eeprom_cmd_data | |
| dataBit | AUBURN_EEPROM_CLK_RISE)); |
| ql_write_nvram_reg(qdev, spir, |
| (ISP_NVRAM_MASK | qdev->eeprom_cmd_data | |
| dataBit | AUBURN_EEPROM_CLK_FALL)); |
| cmd = cmd << 1; |
| } |
| |
| mask = 1 << (addrBits - 1); |
| /* Force the previous data bit to be different */ |
| previousBit = 0xffff; |
| for (i = 0; i < addrBits; i++) { |
| dataBit = (eepromAddr & mask) ? AUBURN_EEPROM_DO_1 |
| : AUBURN_EEPROM_DO_0; |
| if (previousBit != dataBit) { |
| /* |
| * If the bit changed, then change the DO state to |
| * match |
| */ |
| ql_write_nvram_reg(qdev, spir, |
| (ISP_NVRAM_MASK | |
| qdev->eeprom_cmd_data | dataBit)); |
| previousBit = dataBit; |
| } |
| ql_write_nvram_reg(qdev, spir, |
| (ISP_NVRAM_MASK | qdev->eeprom_cmd_data | |
| dataBit | AUBURN_EEPROM_CLK_RISE)); |
| ql_write_nvram_reg(qdev, spir, |
| (ISP_NVRAM_MASK | qdev->eeprom_cmd_data | |
| dataBit | AUBURN_EEPROM_CLK_FALL)); |
| eepromAddr = eepromAddr << 1; |
| } |
| } |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static void fm93c56a_deselect(struct ql3_adapter *qdev) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 *spir = &port_regs->CommonRegs.serialPortInterfaceReg; |
| |
| qdev->eeprom_cmd_data = AUBURN_EEPROM_CS_0; |
| ql_write_nvram_reg(qdev, spir, ISP_NVRAM_MASK | qdev->eeprom_cmd_data); |
| } |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static void fm93c56a_datain(struct ql3_adapter *qdev, unsigned short *value) |
| { |
| int i; |
| u32 data = 0; |
| u32 dataBit; |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 *spir = &port_regs->CommonRegs.serialPortInterfaceReg; |
| |
| /* Read the data bits */ |
| /* The first bit is a dummy. Clock right over it. */ |
| for (i = 0; i < dataBits; i++) { |
| ql_write_nvram_reg(qdev, spir, |
| ISP_NVRAM_MASK | qdev->eeprom_cmd_data | |
| AUBURN_EEPROM_CLK_RISE); |
| ql_write_nvram_reg(qdev, spir, |
| ISP_NVRAM_MASK | qdev->eeprom_cmd_data | |
| AUBURN_EEPROM_CLK_FALL); |
| dataBit = (ql_read_common_reg(qdev, spir) & |
| AUBURN_EEPROM_DI_1) ? 1 : 0; |
| data = (data << 1) | dataBit; |
| } |
| *value = (u16)data; |
| } |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static void eeprom_readword(struct ql3_adapter *qdev, |
| u32 eepromAddr, unsigned short *value) |
| { |
| fm93c56a_select(qdev); |
| fm93c56a_cmd(qdev, (int)FM93C56A_READ, eepromAddr); |
| fm93c56a_datain(qdev, value); |
| fm93c56a_deselect(qdev); |
| } |
| |
| static void ql_set_mac_addr(struct net_device *ndev, u16 *addr) |
| { |
| __le16 *p = (__le16 *)ndev->dev_addr; |
| p[0] = cpu_to_le16(addr[0]); |
| p[1] = cpu_to_le16(addr[1]); |
| p[2] = cpu_to_le16(addr[2]); |
| } |
| |
| static int ql_get_nvram_params(struct ql3_adapter *qdev) |
| { |
| u16 *pEEPROMData; |
| u16 checksum = 0; |
| u32 index; |
| unsigned long hw_flags; |
| |
| spin_lock_irqsave(&qdev->hw_lock, hw_flags); |
| |
| pEEPROMData = (u16 *)&qdev->nvram_data; |
| qdev->eeprom_cmd_data = 0; |
| if (ql_sem_spinlock(qdev, QL_NVRAM_SEM_MASK, |
| (QL_RESOURCE_BITS_BASE_CODE | (qdev->mac_index) * |
| 2) << 10)) { |
| pr_err("%s: Failed ql_sem_spinlock()\n", __func__); |
| spin_unlock_irqrestore(&qdev->hw_lock, hw_flags); |
| return -1; |
| } |
| |
| for (index = 0; index < EEPROM_SIZE; index++) { |
| eeprom_readword(qdev, index, pEEPROMData); |
| checksum += *pEEPROMData; |
| pEEPROMData++; |
| } |
| ql_sem_unlock(qdev, QL_NVRAM_SEM_MASK); |
| |
| if (checksum != 0) { |
| netdev_err(qdev->ndev, "checksum should be zero, is %x!!\n", |
| checksum); |
| spin_unlock_irqrestore(&qdev->hw_lock, hw_flags); |
| return -1; |
| } |
| |
| spin_unlock_irqrestore(&qdev->hw_lock, hw_flags); |
| return checksum; |
| } |
| |
| static const u32 PHYAddr[2] = { |
| PORT0_PHY_ADDRESS, PORT1_PHY_ADDRESS |
| }; |
| |
| static int ql_wait_for_mii_ready(struct ql3_adapter *qdev) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 temp; |
| int count = 1000; |
| |
| while (count) { |
| temp = ql_read_page0_reg(qdev, &port_regs->macMIIStatusReg); |
| if (!(temp & MAC_MII_STATUS_BSY)) |
| return 0; |
| udelay(10); |
| count--; |
| } |
| return -1; |
| } |
| |
| static void ql_mii_enable_scan_mode(struct ql3_adapter *qdev) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 scanControl; |
| |
| if (qdev->numPorts > 1) { |
| /* Auto scan will cycle through multiple ports */ |
| scanControl = MAC_MII_CONTROL_AS | MAC_MII_CONTROL_SC; |
| } else { |
| scanControl = MAC_MII_CONTROL_SC; |
| } |
| |
| /* |
| * Scan register 1 of PHY/PETBI, |
| * Set up to scan both devices |
| * The autoscan starts from the first register, completes |
| * the last one before rolling over to the first |
| */ |
| ql_write_page0_reg(qdev, &port_regs->macMIIMgmtAddrReg, |
| PHYAddr[0] | MII_SCAN_REGISTER); |
| |
| ql_write_page0_reg(qdev, &port_regs->macMIIMgmtControlReg, |
| (scanControl) | |
| ((MAC_MII_CONTROL_SC | MAC_MII_CONTROL_AS) << 16)); |
| } |
| |
| static u8 ql_mii_disable_scan_mode(struct ql3_adapter *qdev) |
| { |
| u8 ret; |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| |
| /* See if scan mode is enabled before we turn it off */ |
| if (ql_read_page0_reg(qdev, &port_regs->macMIIMgmtControlReg) & |
| (MAC_MII_CONTROL_AS | MAC_MII_CONTROL_SC)) { |
| /* Scan is enabled */ |
| ret = 1; |
| } else { |
| /* Scan is disabled */ |
| ret = 0; |
| } |
| |
| /* |
| * When disabling scan mode you must first change the MII register |
| * address |
| */ |
| ql_write_page0_reg(qdev, &port_regs->macMIIMgmtAddrReg, |
| PHYAddr[0] | MII_SCAN_REGISTER); |
| |
| ql_write_page0_reg(qdev, &port_regs->macMIIMgmtControlReg, |
| ((MAC_MII_CONTROL_SC | MAC_MII_CONTROL_AS | |
| MAC_MII_CONTROL_RC) << 16)); |
| |
| return ret; |
| } |
| |
| static int ql_mii_write_reg_ex(struct ql3_adapter *qdev, |
| u16 regAddr, u16 value, u32 phyAddr) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u8 scanWasEnabled; |
| |
| scanWasEnabled = ql_mii_disable_scan_mode(qdev); |
| |
| if (ql_wait_for_mii_ready(qdev)) { |
| netif_warn(qdev, link, qdev->ndev, TIMED_OUT_MSG); |
| return -1; |
| } |
| |
| ql_write_page0_reg(qdev, &port_regs->macMIIMgmtAddrReg, |
| phyAddr | regAddr); |
| |
| ql_write_page0_reg(qdev, &port_regs->macMIIMgmtDataReg, value); |
| |
| /* Wait for write to complete 9/10/04 SJP */ |
| if (ql_wait_for_mii_ready(qdev)) { |
| netif_warn(qdev, link, qdev->ndev, TIMED_OUT_MSG); |
| return -1; |
| } |
| |
| if (scanWasEnabled) |
| ql_mii_enable_scan_mode(qdev); |
| |
| return 0; |
| } |
| |
| static int ql_mii_read_reg_ex(struct ql3_adapter *qdev, u16 regAddr, |
| u16 *value, u32 phyAddr) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u8 scanWasEnabled; |
| u32 temp; |
| |
| scanWasEnabled = ql_mii_disable_scan_mode(qdev); |
| |
| if (ql_wait_for_mii_ready(qdev)) { |
| netif_warn(qdev, link, qdev->ndev, TIMED_OUT_MSG); |
| return -1; |
| } |
| |
| ql_write_page0_reg(qdev, &port_regs->macMIIMgmtAddrReg, |
| phyAddr | regAddr); |
| |
| ql_write_page0_reg(qdev, &port_regs->macMIIMgmtControlReg, |
| (MAC_MII_CONTROL_RC << 16)); |
| |
| ql_write_page0_reg(qdev, &port_regs->macMIIMgmtControlReg, |
| (MAC_MII_CONTROL_RC << 16) | MAC_MII_CONTROL_RC); |
| |
| /* Wait for the read to complete */ |
| if (ql_wait_for_mii_ready(qdev)) { |
| netif_warn(qdev, link, qdev->ndev, TIMED_OUT_MSG); |
| return -1; |
| } |
| |
| temp = ql_read_page0_reg(qdev, &port_regs->macMIIMgmtDataReg); |
| *value = (u16) temp; |
| |
| if (scanWasEnabled) |
| ql_mii_enable_scan_mode(qdev); |
| |
| return 0; |
| } |
| |
| static int ql_mii_write_reg(struct ql3_adapter *qdev, u16 regAddr, u16 value) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| |
| ql_mii_disable_scan_mode(qdev); |
| |
| if (ql_wait_for_mii_ready(qdev)) { |
| netif_warn(qdev, link, qdev->ndev, TIMED_OUT_MSG); |
| return -1; |
| } |
| |
| ql_write_page0_reg(qdev, &port_regs->macMIIMgmtAddrReg, |
| qdev->PHYAddr | regAddr); |
| |
| ql_write_page0_reg(qdev, &port_regs->macMIIMgmtDataReg, value); |
| |
| /* Wait for write to complete. */ |
| if (ql_wait_for_mii_ready(qdev)) { |
| netif_warn(qdev, link, qdev->ndev, TIMED_OUT_MSG); |
| return -1; |
| } |
| |
| ql_mii_enable_scan_mode(qdev); |
| |
| return 0; |
| } |
| |
| static int ql_mii_read_reg(struct ql3_adapter *qdev, u16 regAddr, u16 *value) |
| { |
| u32 temp; |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| |
| ql_mii_disable_scan_mode(qdev); |
| |
| if (ql_wait_for_mii_ready(qdev)) { |
| netif_warn(qdev, link, qdev->ndev, TIMED_OUT_MSG); |
| return -1; |
| } |
| |
| ql_write_page0_reg(qdev, &port_regs->macMIIMgmtAddrReg, |
| qdev->PHYAddr | regAddr); |
| |
| ql_write_page0_reg(qdev, &port_regs->macMIIMgmtControlReg, |
| (MAC_MII_CONTROL_RC << 16)); |
| |
| ql_write_page0_reg(qdev, &port_regs->macMIIMgmtControlReg, |
| (MAC_MII_CONTROL_RC << 16) | MAC_MII_CONTROL_RC); |
| |
| /* Wait for the read to complete */ |
| if (ql_wait_for_mii_ready(qdev)) { |
| netif_warn(qdev, link, qdev->ndev, TIMED_OUT_MSG); |
| return -1; |
| } |
| |
| temp = ql_read_page0_reg(qdev, &port_regs->macMIIMgmtDataReg); |
| *value = (u16) temp; |
| |
| ql_mii_enable_scan_mode(qdev); |
| |
| return 0; |
| } |
| |
| static void ql_petbi_reset(struct ql3_adapter *qdev) |
| { |
| ql_mii_write_reg(qdev, PETBI_CONTROL_REG, PETBI_CTRL_SOFT_RESET); |
| } |
| |
| static void ql_petbi_start_neg(struct ql3_adapter *qdev) |
| { |
| u16 reg; |
| |
| /* Enable Auto-negotiation sense */ |
| ql_mii_read_reg(qdev, PETBI_TBI_CTRL, ®); |
| reg |= PETBI_TBI_AUTO_SENSE; |
| ql_mii_write_reg(qdev, PETBI_TBI_CTRL, reg); |
| |
| ql_mii_write_reg(qdev, PETBI_NEG_ADVER, |
| PETBI_NEG_PAUSE | PETBI_NEG_DUPLEX); |
| |
| ql_mii_write_reg(qdev, PETBI_CONTROL_REG, |
| PETBI_CTRL_AUTO_NEG | PETBI_CTRL_RESTART_NEG | |
| PETBI_CTRL_FULL_DUPLEX | PETBI_CTRL_SPEED_1000); |
| |
| } |
| |
| static void ql_petbi_reset_ex(struct ql3_adapter *qdev) |
| { |
| ql_mii_write_reg_ex(qdev, PETBI_CONTROL_REG, PETBI_CTRL_SOFT_RESET, |
| PHYAddr[qdev->mac_index]); |
| } |
| |
| static void ql_petbi_start_neg_ex(struct ql3_adapter *qdev) |
| { |
| u16 reg; |
| |
| /* Enable Auto-negotiation sense */ |
| ql_mii_read_reg_ex(qdev, PETBI_TBI_CTRL, ®, |
| PHYAddr[qdev->mac_index]); |
| reg |= PETBI_TBI_AUTO_SENSE; |
| ql_mii_write_reg_ex(qdev, PETBI_TBI_CTRL, reg, |
| PHYAddr[qdev->mac_index]); |
| |
| ql_mii_write_reg_ex(qdev, PETBI_NEG_ADVER, |
| PETBI_NEG_PAUSE | PETBI_NEG_DUPLEX, |
| PHYAddr[qdev->mac_index]); |
| |
| ql_mii_write_reg_ex(qdev, PETBI_CONTROL_REG, |
| PETBI_CTRL_AUTO_NEG | PETBI_CTRL_RESTART_NEG | |
| PETBI_CTRL_FULL_DUPLEX | PETBI_CTRL_SPEED_1000, |
| PHYAddr[qdev->mac_index]); |
| } |
| |
| static void ql_petbi_init(struct ql3_adapter *qdev) |
| { |
| ql_petbi_reset(qdev); |
| ql_petbi_start_neg(qdev); |
| } |
| |
| static void ql_petbi_init_ex(struct ql3_adapter *qdev) |
| { |
| ql_petbi_reset_ex(qdev); |
| ql_petbi_start_neg_ex(qdev); |
| } |
| |
| static int ql_is_petbi_neg_pause(struct ql3_adapter *qdev) |
| { |
| u16 reg; |
| |
| if (ql_mii_read_reg(qdev, PETBI_NEG_PARTNER, ®) < 0) |
| return 0; |
| |
| return (reg & PETBI_NEG_PAUSE_MASK) == PETBI_NEG_PAUSE; |
| } |
| |
| static void phyAgereSpecificInit(struct ql3_adapter *qdev, u32 miiAddr) |
| { |
| netdev_info(qdev->ndev, "enabling Agere specific PHY\n"); |
| /* power down device bit 11 = 1 */ |
| ql_mii_write_reg_ex(qdev, 0x00, 0x1940, miiAddr); |
| /* enable diagnostic mode bit 2 = 1 */ |
| ql_mii_write_reg_ex(qdev, 0x12, 0x840e, miiAddr); |
| /* 1000MB amplitude adjust (see Agere errata) */ |
| ql_mii_write_reg_ex(qdev, 0x10, 0x8805, miiAddr); |
| /* 1000MB amplitude adjust (see Agere errata) */ |
| ql_mii_write_reg_ex(qdev, 0x11, 0xf03e, miiAddr); |
| /* 100MB amplitude adjust (see Agere errata) */ |
| ql_mii_write_reg_ex(qdev, 0x10, 0x8806, miiAddr); |
| /* 100MB amplitude adjust (see Agere errata) */ |
| ql_mii_write_reg_ex(qdev, 0x11, 0x003e, miiAddr); |
| /* 10MB amplitude adjust (see Agere errata) */ |
| ql_mii_write_reg_ex(qdev, 0x10, 0x8807, miiAddr); |
| /* 10MB amplitude adjust (see Agere errata) */ |
| ql_mii_write_reg_ex(qdev, 0x11, 0x1f00, miiAddr); |
| /* point to hidden reg 0x2806 */ |
| ql_mii_write_reg_ex(qdev, 0x10, 0x2806, miiAddr); |
| /* Write new PHYAD w/bit 5 set */ |
| ql_mii_write_reg_ex(qdev, 0x11, |
| 0x0020 | (PHYAddr[qdev->mac_index] >> 8), miiAddr); |
| /* |
| * Disable diagnostic mode bit 2 = 0 |
| * Power up device bit 11 = 0 |
| * Link up (on) and activity (blink) |
| */ |
| ql_mii_write_reg(qdev, 0x12, 0x840a); |
| ql_mii_write_reg(qdev, 0x00, 0x1140); |
| ql_mii_write_reg(qdev, 0x1c, 0xfaf0); |
| } |
| |
| static enum PHY_DEVICE_TYPE getPhyType(struct ql3_adapter *qdev, |
| u16 phyIdReg0, u16 phyIdReg1) |
| { |
| enum PHY_DEVICE_TYPE result = PHY_TYPE_UNKNOWN; |
| u32 oui; |
| u16 model; |
| int i; |
| |
| if (phyIdReg0 == 0xffff) |
| return result; |
| |
| if (phyIdReg1 == 0xffff) |
| return result; |
| |
| /* oui is split between two registers */ |
| oui = (phyIdReg0 << 6) | ((phyIdReg1 & PHY_OUI_1_MASK) >> 10); |
| |
| model = (phyIdReg1 & PHY_MODEL_MASK) >> 4; |
| |
| /* Scan table for this PHY */ |
| for (i = 0; i < MAX_PHY_DEV_TYPES; i++) { |
| if ((oui == PHY_DEVICES[i].phyIdOUI) && |
| (model == PHY_DEVICES[i].phyIdModel)) { |
| netdev_info(qdev->ndev, "Phy: %s\n", |
| PHY_DEVICES[i].name); |
| result = PHY_DEVICES[i].phyDevice; |
| break; |
| } |
| } |
| |
| return result; |
| } |
| |
| static int ql_phy_get_speed(struct ql3_adapter *qdev) |
| { |
| u16 reg; |
| |
| switch (qdev->phyType) { |
| case PHY_AGERE_ET1011C: { |
| if (ql_mii_read_reg(qdev, 0x1A, ®) < 0) |
| return 0; |
| |
| reg = (reg >> 8) & 3; |
| break; |
| } |
| default: |
| if (ql_mii_read_reg(qdev, AUX_CONTROL_STATUS, ®) < 0) |
| return 0; |
| |
| reg = (((reg & 0x18) >> 3) & 3); |
| } |
| |
| switch (reg) { |
| case 2: |
| return SPEED_1000; |
| case 1: |
| return SPEED_100; |
| case 0: |
| return SPEED_10; |
| default: |
| return -1; |
| } |
| } |
| |
| static int ql_is_full_dup(struct ql3_adapter *qdev) |
| { |
| u16 reg; |
| |
| switch (qdev->phyType) { |
| case PHY_AGERE_ET1011C: { |
| if (ql_mii_read_reg(qdev, 0x1A, ®)) |
| return 0; |
| |
| return ((reg & 0x0080) && (reg & 0x1000)) != 0; |
| } |
| case PHY_VITESSE_VSC8211: |
| default: { |
| if (ql_mii_read_reg(qdev, AUX_CONTROL_STATUS, ®) < 0) |
| return 0; |
| return (reg & PHY_AUX_DUPLEX_STAT) != 0; |
| } |
| } |
| } |
| |
| static int ql_is_phy_neg_pause(struct ql3_adapter *qdev) |
| { |
| u16 reg; |
| |
| if (ql_mii_read_reg(qdev, PHY_NEG_PARTNER, ®) < 0) |
| return 0; |
| |
| return (reg & PHY_NEG_PAUSE) != 0; |
| } |
| |
| static int PHY_Setup(struct ql3_adapter *qdev) |
| { |
| u16 reg1; |
| u16 reg2; |
| bool agereAddrChangeNeeded = false; |
| u32 miiAddr = 0; |
| int err; |
| |
| /* Determine the PHY we are using by reading the ID's */ |
| err = ql_mii_read_reg(qdev, PHY_ID_0_REG, ®1); |
| if (err != 0) { |
| netdev_err(qdev->ndev, "Could not read from reg PHY_ID_0_REG\n"); |
| return err; |
| } |
| |
| err = ql_mii_read_reg(qdev, PHY_ID_1_REG, ®2); |
| if (err != 0) { |
| netdev_err(qdev->ndev, "Could not read from reg PHY_ID_1_REG\n"); |
| return err; |
| } |
| |
| /* Check if we have a Agere PHY */ |
| if ((reg1 == 0xffff) || (reg2 == 0xffff)) { |
| |
| /* Determine which MII address we should be using |
| determined by the index of the card */ |
| if (qdev->mac_index == 0) |
| miiAddr = MII_AGERE_ADDR_1; |
| else |
| miiAddr = MII_AGERE_ADDR_2; |
| |
| err = ql_mii_read_reg_ex(qdev, PHY_ID_0_REG, ®1, miiAddr); |
| if (err != 0) { |
| netdev_err(qdev->ndev, |
| "Could not read from reg PHY_ID_0_REG after Agere detected\n"); |
| return err; |
| } |
| |
| err = ql_mii_read_reg_ex(qdev, PHY_ID_1_REG, ®2, miiAddr); |
| if (err != 0) { |
| netdev_err(qdev->ndev, "Could not read from reg PHY_ID_1_REG after Agere detected\n"); |
| return err; |
| } |
| |
| /* We need to remember to initialize the Agere PHY */ |
| agereAddrChangeNeeded = true; |
| } |
| |
| /* Determine the particular PHY we have on board to apply |
| PHY specific initializations */ |
| qdev->phyType = getPhyType(qdev, reg1, reg2); |
| |
| if ((qdev->phyType == PHY_AGERE_ET1011C) && agereAddrChangeNeeded) { |
| /* need this here so address gets changed */ |
| phyAgereSpecificInit(qdev, miiAddr); |
| } else if (qdev->phyType == PHY_TYPE_UNKNOWN) { |
| netdev_err(qdev->ndev, "PHY is unknown\n"); |
| return -EIO; |
| } |
| |
| return 0; |
| } |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static void ql_mac_enable(struct ql3_adapter *qdev, u32 enable) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 value; |
| |
| if (enable) |
| value = (MAC_CONFIG_REG_PE | (MAC_CONFIG_REG_PE << 16)); |
| else |
| value = (MAC_CONFIG_REG_PE << 16); |
| |
| if (qdev->mac_index) |
| ql_write_page0_reg(qdev, &port_regs->mac1ConfigReg, value); |
| else |
| ql_write_page0_reg(qdev, &port_regs->mac0ConfigReg, value); |
| } |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static void ql_mac_cfg_soft_reset(struct ql3_adapter *qdev, u32 enable) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 value; |
| |
| if (enable) |
| value = (MAC_CONFIG_REG_SR | (MAC_CONFIG_REG_SR << 16)); |
| else |
| value = (MAC_CONFIG_REG_SR << 16); |
| |
| if (qdev->mac_index) |
| ql_write_page0_reg(qdev, &port_regs->mac1ConfigReg, value); |
| else |
| ql_write_page0_reg(qdev, &port_regs->mac0ConfigReg, value); |
| } |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static void ql_mac_cfg_gig(struct ql3_adapter *qdev, u32 enable) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 value; |
| |
| if (enable) |
| value = (MAC_CONFIG_REG_GM | (MAC_CONFIG_REG_GM << 16)); |
| else |
| value = (MAC_CONFIG_REG_GM << 16); |
| |
| if (qdev->mac_index) |
| ql_write_page0_reg(qdev, &port_regs->mac1ConfigReg, value); |
| else |
| ql_write_page0_reg(qdev, &port_regs->mac0ConfigReg, value); |
| } |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static void ql_mac_cfg_full_dup(struct ql3_adapter *qdev, u32 enable) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 value; |
| |
| if (enable) |
| value = (MAC_CONFIG_REG_FD | (MAC_CONFIG_REG_FD << 16)); |
| else |
| value = (MAC_CONFIG_REG_FD << 16); |
| |
| if (qdev->mac_index) |
| ql_write_page0_reg(qdev, &port_regs->mac1ConfigReg, value); |
| else |
| ql_write_page0_reg(qdev, &port_regs->mac0ConfigReg, value); |
| } |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static void ql_mac_cfg_pause(struct ql3_adapter *qdev, u32 enable) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 value; |
| |
| if (enable) |
| value = |
| ((MAC_CONFIG_REG_TF | MAC_CONFIG_REG_RF) | |
| ((MAC_CONFIG_REG_TF | MAC_CONFIG_REG_RF) << 16)); |
| else |
| value = ((MAC_CONFIG_REG_TF | MAC_CONFIG_REG_RF) << 16); |
| |
| if (qdev->mac_index) |
| ql_write_page0_reg(qdev, &port_regs->mac1ConfigReg, value); |
| else |
| ql_write_page0_reg(qdev, &port_regs->mac0ConfigReg, value); |
| } |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static int ql_is_fiber(struct ql3_adapter *qdev) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 bitToCheck = 0; |
| u32 temp; |
| |
| switch (qdev->mac_index) { |
| case 0: |
| bitToCheck = PORT_STATUS_SM0; |
| break; |
| case 1: |
| bitToCheck = PORT_STATUS_SM1; |
| break; |
| } |
| |
| temp = ql_read_page0_reg(qdev, &port_regs->portStatus); |
| return (temp & bitToCheck) != 0; |
| } |
| |
| static int ql_is_auto_cfg(struct ql3_adapter *qdev) |
| { |
| u16 reg; |
| ql_mii_read_reg(qdev, 0x00, ®); |
| return (reg & 0x1000) != 0; |
| } |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static int ql_is_auto_neg_complete(struct ql3_adapter *qdev) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 bitToCheck = 0; |
| u32 temp; |
| |
| switch (qdev->mac_index) { |
| case 0: |
| bitToCheck = PORT_STATUS_AC0; |
| break; |
| case 1: |
| bitToCheck = PORT_STATUS_AC1; |
| break; |
| } |
| |
| temp = ql_read_page0_reg(qdev, &port_regs->portStatus); |
| if (temp & bitToCheck) { |
| netif_info(qdev, link, qdev->ndev, "Auto-Negotiate complete\n"); |
| return 1; |
| } |
| netif_info(qdev, link, qdev->ndev, "Auto-Negotiate incomplete\n"); |
| return 0; |
| } |
| |
| /* |
| * ql_is_neg_pause() returns 1 if pause was negotiated to be on |
| */ |
| static int ql_is_neg_pause(struct ql3_adapter *qdev) |
| { |
| if (ql_is_fiber(qdev)) |
| return ql_is_petbi_neg_pause(qdev); |
| else |
| return ql_is_phy_neg_pause(qdev); |
| } |
| |
| static int ql_auto_neg_error(struct ql3_adapter *qdev) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 bitToCheck = 0; |
| u32 temp; |
| |
| switch (qdev->mac_index) { |
| case 0: |
| bitToCheck = PORT_STATUS_AE0; |
| break; |
| case 1: |
| bitToCheck = PORT_STATUS_AE1; |
| break; |
| } |
| temp = ql_read_page0_reg(qdev, &port_regs->portStatus); |
| return (temp & bitToCheck) != 0; |
| } |
| |
| static u32 ql_get_link_speed(struct ql3_adapter *qdev) |
| { |
| if (ql_is_fiber(qdev)) |
| return SPEED_1000; |
| else |
| return ql_phy_get_speed(qdev); |
| } |
| |
| static int ql_is_link_full_dup(struct ql3_adapter *qdev) |
| { |
| if (ql_is_fiber(qdev)) |
| return 1; |
| else |
| return ql_is_full_dup(qdev); |
| } |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static int ql_link_down_detect(struct ql3_adapter *qdev) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 bitToCheck = 0; |
| u32 temp; |
| |
| switch (qdev->mac_index) { |
| case 0: |
| bitToCheck = ISP_CONTROL_LINK_DN_0; |
| break; |
| case 1: |
| bitToCheck = ISP_CONTROL_LINK_DN_1; |
| break; |
| } |
| |
| temp = |
| ql_read_common_reg(qdev, &port_regs->CommonRegs.ispControlStatus); |
| return (temp & bitToCheck) != 0; |
| } |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static int ql_link_down_detect_clear(struct ql3_adapter *qdev) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| |
| switch (qdev->mac_index) { |
| case 0: |
| ql_write_common_reg(qdev, |
| &port_regs->CommonRegs.ispControlStatus, |
| (ISP_CONTROL_LINK_DN_0) | |
| (ISP_CONTROL_LINK_DN_0 << 16)); |
| break; |
| |
| case 1: |
| ql_write_common_reg(qdev, |
| &port_regs->CommonRegs.ispControlStatus, |
| (ISP_CONTROL_LINK_DN_1) | |
| (ISP_CONTROL_LINK_DN_1 << 16)); |
| break; |
| |
| default: |
| return 1; |
| } |
| |
| return 0; |
| } |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static int ql_this_adapter_controls_port(struct ql3_adapter *qdev) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 bitToCheck = 0; |
| u32 temp; |
| |
| switch (qdev->mac_index) { |
| case 0: |
| bitToCheck = PORT_STATUS_F1_ENABLED; |
| break; |
| case 1: |
| bitToCheck = PORT_STATUS_F3_ENABLED; |
| break; |
| default: |
| break; |
| } |
| |
| temp = ql_read_page0_reg(qdev, &port_regs->portStatus); |
| if (temp & bitToCheck) { |
| netif_printk(qdev, link, KERN_DEBUG, qdev->ndev, |
| "not link master\n"); |
| return 0; |
| } |
| |
| netif_printk(qdev, link, KERN_DEBUG, qdev->ndev, "link master\n"); |
| return 1; |
| } |
| |
| static void ql_phy_reset_ex(struct ql3_adapter *qdev) |
| { |
| ql_mii_write_reg_ex(qdev, CONTROL_REG, PHY_CTRL_SOFT_RESET, |
| PHYAddr[qdev->mac_index]); |
| } |
| |
| static void ql_phy_start_neg_ex(struct ql3_adapter *qdev) |
| { |
| u16 reg; |
| u16 portConfiguration; |
| |
| if (qdev->phyType == PHY_AGERE_ET1011C) |
| ql_mii_write_reg(qdev, 0x13, 0x0000); |
| /* turn off external loopback */ |
| |
| if (qdev->mac_index == 0) |
| portConfiguration = |
| qdev->nvram_data.macCfg_port0.portConfiguration; |
| else |
| portConfiguration = |
| qdev->nvram_data.macCfg_port1.portConfiguration; |
| |
| /* Some HBA's in the field are set to 0 and they need to |
| be reinterpreted with a default value */ |
| if (portConfiguration == 0) |
| portConfiguration = PORT_CONFIG_DEFAULT; |
| |
| /* Set the 1000 advertisements */ |
| ql_mii_read_reg_ex(qdev, PHY_GIG_CONTROL, ®, |
| PHYAddr[qdev->mac_index]); |
| reg &= ~PHY_GIG_ALL_PARAMS; |
| |
| if (portConfiguration & PORT_CONFIG_1000MB_SPEED) { |
| if (portConfiguration & PORT_CONFIG_FULL_DUPLEX_ENABLED) |
| reg |= PHY_GIG_ADV_1000F; |
| else |
| reg |= PHY_GIG_ADV_1000H; |
| } |
| |
| ql_mii_write_reg_ex(qdev, PHY_GIG_CONTROL, reg, |
| PHYAddr[qdev->mac_index]); |
| |
| /* Set the 10/100 & pause negotiation advertisements */ |
| ql_mii_read_reg_ex(qdev, PHY_NEG_ADVER, ®, |
| PHYAddr[qdev->mac_index]); |
| reg &= ~PHY_NEG_ALL_PARAMS; |
| |
| if (portConfiguration & PORT_CONFIG_SYM_PAUSE_ENABLED) |
| reg |= PHY_NEG_ASY_PAUSE | PHY_NEG_SYM_PAUSE; |
| |
| if (portConfiguration & PORT_CONFIG_FULL_DUPLEX_ENABLED) { |
| if (portConfiguration & PORT_CONFIG_100MB_SPEED) |
| reg |= PHY_NEG_ADV_100F; |
| |
| if (portConfiguration & PORT_CONFIG_10MB_SPEED) |
| reg |= PHY_NEG_ADV_10F; |
| } |
| |
| if (portConfiguration & PORT_CONFIG_HALF_DUPLEX_ENABLED) { |
| if (portConfiguration & PORT_CONFIG_100MB_SPEED) |
| reg |= PHY_NEG_ADV_100H; |
| |
| if (portConfiguration & PORT_CONFIG_10MB_SPEED) |
| reg |= PHY_NEG_ADV_10H; |
| } |
| |
| if (portConfiguration & PORT_CONFIG_1000MB_SPEED) |
| reg |= 1; |
| |
| ql_mii_write_reg_ex(qdev, PHY_NEG_ADVER, reg, |
| PHYAddr[qdev->mac_index]); |
| |
| ql_mii_read_reg_ex(qdev, CONTROL_REG, ®, PHYAddr[qdev->mac_index]); |
| |
| ql_mii_write_reg_ex(qdev, CONTROL_REG, |
| reg | PHY_CTRL_RESTART_NEG | PHY_CTRL_AUTO_NEG, |
| PHYAddr[qdev->mac_index]); |
| } |
| |
| static void ql_phy_init_ex(struct ql3_adapter *qdev) |
| { |
| ql_phy_reset_ex(qdev); |
| PHY_Setup(qdev); |
| ql_phy_start_neg_ex(qdev); |
| } |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static u32 ql_get_link_state(struct ql3_adapter *qdev) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 bitToCheck = 0; |
| u32 temp, linkState; |
| |
| switch (qdev->mac_index) { |
| case 0: |
| bitToCheck = PORT_STATUS_UP0; |
| break; |
| case 1: |
| bitToCheck = PORT_STATUS_UP1; |
| break; |
| } |
| |
| temp = ql_read_page0_reg(qdev, &port_regs->portStatus); |
| if (temp & bitToCheck) |
| linkState = LS_UP; |
| else |
| linkState = LS_DOWN; |
| |
| return linkState; |
| } |
| |
| static int ql_port_start(struct ql3_adapter *qdev) |
| { |
| if (ql_sem_spinlock(qdev, QL_PHY_GIO_SEM_MASK, |
| (QL_RESOURCE_BITS_BASE_CODE | (qdev->mac_index) * |
| 2) << 7)) { |
| netdev_err(qdev->ndev, "Could not get hw lock for GIO\n"); |
| return -1; |
| } |
| |
| if (ql_is_fiber(qdev)) { |
| ql_petbi_init(qdev); |
| } else { |
| /* Copper port */ |
| ql_phy_init_ex(qdev); |
| } |
| |
| ql_sem_unlock(qdev, QL_PHY_GIO_SEM_MASK); |
| return 0; |
| } |
| |
| static int ql_finish_auto_neg(struct ql3_adapter *qdev) |
| { |
| |
| if (ql_sem_spinlock(qdev, QL_PHY_GIO_SEM_MASK, |
| (QL_RESOURCE_BITS_BASE_CODE | (qdev->mac_index) * |
| 2) << 7)) |
| return -1; |
| |
| if (!ql_auto_neg_error(qdev)) { |
| if (test_bit(QL_LINK_MASTER, &qdev->flags)) { |
| /* configure the MAC */ |
| netif_printk(qdev, link, KERN_DEBUG, qdev->ndev, |
| "Configuring link\n"); |
| ql_mac_cfg_soft_reset(qdev, 1); |
| ql_mac_cfg_gig(qdev, |
| (ql_get_link_speed |
| (qdev) == |
| SPEED_1000)); |
| ql_mac_cfg_full_dup(qdev, |
| ql_is_link_full_dup |
| (qdev)); |
| ql_mac_cfg_pause(qdev, |
| ql_is_neg_pause |
| (qdev)); |
| ql_mac_cfg_soft_reset(qdev, 0); |
| |
| /* enable the MAC */ |
| netif_printk(qdev, link, KERN_DEBUG, qdev->ndev, |
| "Enabling mac\n"); |
| ql_mac_enable(qdev, 1); |
| } |
| |
| qdev->port_link_state = LS_UP; |
| netif_start_queue(qdev->ndev); |
| netif_carrier_on(qdev->ndev); |
| netif_info(qdev, link, qdev->ndev, |
| "Link is up at %d Mbps, %s duplex\n", |
| ql_get_link_speed(qdev), |
| ql_is_link_full_dup(qdev) ? "full" : "half"); |
| |
| } else { /* Remote error detected */ |
| |
| if (test_bit(QL_LINK_MASTER, &qdev->flags)) { |
| netif_printk(qdev, link, KERN_DEBUG, qdev->ndev, |
| "Remote error detected. Calling ql_port_start()\n"); |
| /* |
| * ql_port_start() is shared code and needs |
| * to lock the PHY on it's own. |
| */ |
| ql_sem_unlock(qdev, QL_PHY_GIO_SEM_MASK); |
| if (ql_port_start(qdev)) /* Restart port */ |
| return -1; |
| return 0; |
| } |
| } |
| ql_sem_unlock(qdev, QL_PHY_GIO_SEM_MASK); |
| return 0; |
| } |
| |
| static void ql_link_state_machine_work(struct work_struct *work) |
| { |
| struct ql3_adapter *qdev = |
| container_of(work, struct ql3_adapter, link_state_work.work); |
| |
| u32 curr_link_state; |
| unsigned long hw_flags; |
| |
| spin_lock_irqsave(&qdev->hw_lock, hw_flags); |
| |
| curr_link_state = ql_get_link_state(qdev); |
| |
| if (test_bit(QL_RESET_ACTIVE, &qdev->flags)) { |
| netif_info(qdev, link, qdev->ndev, |
| "Reset in progress, skip processing link state\n"); |
| |
| spin_unlock_irqrestore(&qdev->hw_lock, hw_flags); |
| |
| /* Restart timer on 2 second interval. */ |
| mod_timer(&qdev->adapter_timer, jiffies + HZ * 1); |
| |
| return; |
| } |
| |
| switch (qdev->port_link_state) { |
| default: |
| if (test_bit(QL_LINK_MASTER, &qdev->flags)) |
| ql_port_start(qdev); |
| qdev->port_link_state = LS_DOWN; |
| /* Fall Through */ |
| |
| case LS_DOWN: |
| if (curr_link_state == LS_UP) { |
| netif_info(qdev, link, qdev->ndev, "Link is up\n"); |
| if (ql_is_auto_neg_complete(qdev)) |
| ql_finish_auto_neg(qdev); |
| |
| if (qdev->port_link_state == LS_UP) |
| ql_link_down_detect_clear(qdev); |
| |
| qdev->port_link_state = LS_UP; |
| } |
| break; |
| |
| case LS_UP: |
| /* |
| * See if the link is currently down or went down and came |
| * back up |
| */ |
| if (curr_link_state == LS_DOWN) { |
| netif_info(qdev, link, qdev->ndev, "Link is down\n"); |
| qdev->port_link_state = LS_DOWN; |
| } |
| if (ql_link_down_detect(qdev)) |
| qdev->port_link_state = LS_DOWN; |
| break; |
| } |
| spin_unlock_irqrestore(&qdev->hw_lock, hw_flags); |
| |
| /* Restart timer on 2 second interval. */ |
| mod_timer(&qdev->adapter_timer, jiffies + HZ * 1); |
| } |
| |
| /* |
| * Caller must take hw_lock and QL_PHY_GIO_SEM. |
| */ |
| static void ql_get_phy_owner(struct ql3_adapter *qdev) |
| { |
| if (ql_this_adapter_controls_port(qdev)) |
| set_bit(QL_LINK_MASTER, &qdev->flags); |
| else |
| clear_bit(QL_LINK_MASTER, &qdev->flags); |
| } |
| |
| /* |
| * Caller must take hw_lock and QL_PHY_GIO_SEM. |
| */ |
| static void ql_init_scan_mode(struct ql3_adapter *qdev) |
| { |
| ql_mii_enable_scan_mode(qdev); |
| |
| if (test_bit(QL_LINK_OPTICAL, &qdev->flags)) { |
| if (ql_this_adapter_controls_port(qdev)) |
| ql_petbi_init_ex(qdev); |
| } else { |
| if (ql_this_adapter_controls_port(qdev)) |
| ql_phy_init_ex(qdev); |
| } |
| } |
| |
| /* |
| * MII_Setup needs to be called before taking the PHY out of reset |
| * so that the management interface clock speed can be set properly. |
| * It would be better if we had a way to disable MDC until after the |
| * PHY is out of reset, but we don't have that capability. |
| */ |
| static int ql_mii_setup(struct ql3_adapter *qdev) |
| { |
| u32 reg; |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| |
| if (ql_sem_spinlock(qdev, QL_PHY_GIO_SEM_MASK, |
| (QL_RESOURCE_BITS_BASE_CODE | (qdev->mac_index) * |
| 2) << 7)) |
| return -1; |
| |
| if (qdev->device_id == QL3032_DEVICE_ID) |
| ql_write_page0_reg(qdev, |
| &port_regs->macMIIMgmtControlReg, 0x0f00000); |
| |
| /* Divide 125MHz clock by 28 to meet PHY timing requirements */ |
| reg = MAC_MII_CONTROL_CLK_SEL_DIV28; |
| |
| ql_write_page0_reg(qdev, &port_regs->macMIIMgmtControlReg, |
| reg | ((MAC_MII_CONTROL_CLK_SEL_MASK) << 16)); |
| |
| ql_sem_unlock(qdev, QL_PHY_GIO_SEM_MASK); |
| return 0; |
| } |
| |
| #define SUPPORTED_OPTICAL_MODES (SUPPORTED_1000baseT_Full | \ |
| SUPPORTED_FIBRE | \ |
| SUPPORTED_Autoneg) |
| #define SUPPORTED_TP_MODES (SUPPORTED_10baseT_Half | \ |
| SUPPORTED_10baseT_Full | \ |
| SUPPORTED_100baseT_Half | \ |
| SUPPORTED_100baseT_Full | \ |
| SUPPORTED_1000baseT_Half | \ |
| SUPPORTED_1000baseT_Full | \ |
| SUPPORTED_Autoneg | \ |
| SUPPORTED_TP); \ |
| |
| static u32 ql_supported_modes(struct ql3_adapter *qdev) |
| { |
| if (test_bit(QL_LINK_OPTICAL, &qdev->flags)) |
| return SUPPORTED_OPTICAL_MODES; |
| |
| return SUPPORTED_TP_MODES; |
| } |
| |
| static int ql_get_auto_cfg_status(struct ql3_adapter *qdev) |
| { |
| int status; |
| unsigned long hw_flags; |
| spin_lock_irqsave(&qdev->hw_lock, hw_flags); |
| if (ql_sem_spinlock(qdev, QL_PHY_GIO_SEM_MASK, |
| (QL_RESOURCE_BITS_BASE_CODE | |
| (qdev->mac_index) * 2) << 7)) { |
| spin_unlock_irqrestore(&qdev->hw_lock, hw_flags); |
| return 0; |
| } |
| status = ql_is_auto_cfg(qdev); |
| ql_sem_unlock(qdev, QL_PHY_GIO_SEM_MASK); |
| spin_unlock_irqrestore(&qdev->hw_lock, hw_flags); |
| return status; |
| } |
| |
| static u32 ql_get_speed(struct ql3_adapter *qdev) |
| { |
| u32 status; |
| unsigned long hw_flags; |
| spin_lock_irqsave(&qdev->hw_lock, hw_flags); |
| if (ql_sem_spinlock(qdev, QL_PHY_GIO_SEM_MASK, |
| (QL_RESOURCE_BITS_BASE_CODE | |
| (qdev->mac_index) * 2) << 7)) { |
| spin_unlock_irqrestore(&qdev->hw_lock, hw_flags); |
| return 0; |
| } |
| status = ql_get_link_speed(qdev); |
| ql_sem_unlock(qdev, QL_PHY_GIO_SEM_MASK); |
| spin_unlock_irqrestore(&qdev->hw_lock, hw_flags); |
| return status; |
| } |
| |
| static int ql_get_full_dup(struct ql3_adapter *qdev) |
| { |
| int status; |
| unsigned long hw_flags; |
| spin_lock_irqsave(&qdev->hw_lock, hw_flags); |
| if (ql_sem_spinlock(qdev, QL_PHY_GIO_SEM_MASK, |
| (QL_RESOURCE_BITS_BASE_CODE | |
| (qdev->mac_index) * 2) << 7)) { |
| spin_unlock_irqrestore(&qdev->hw_lock, hw_flags); |
| return 0; |
| } |
| status = ql_is_link_full_dup(qdev); |
| ql_sem_unlock(qdev, QL_PHY_GIO_SEM_MASK); |
| spin_unlock_irqrestore(&qdev->hw_lock, hw_flags); |
| return status; |
| } |
| |
| static int ql_get_settings(struct net_device *ndev, struct ethtool_cmd *ecmd) |
| { |
| struct ql3_adapter *qdev = netdev_priv(ndev); |
| |
| ecmd->transceiver = XCVR_INTERNAL; |
| ecmd->supported = ql_supported_modes(qdev); |
| |
| if (test_bit(QL_LINK_OPTICAL, &qdev->flags)) { |
| ecmd->port = PORT_FIBRE; |
| } else { |
| ecmd->port = PORT_TP; |
| ecmd->phy_address = qdev->PHYAddr; |
| } |
| ecmd->advertising = ql_supported_modes(qdev); |
| ecmd->autoneg = ql_get_auto_cfg_status(qdev); |
| ecmd->speed = ql_get_speed(qdev); |
| ecmd->duplex = ql_get_full_dup(qdev); |
| return 0; |
| } |
| |
| static void ql_get_drvinfo(struct net_device *ndev, |
| struct ethtool_drvinfo *drvinfo) |
| { |
| struct ql3_adapter *qdev = netdev_priv(ndev); |
| strncpy(drvinfo->driver, ql3xxx_driver_name, 32); |
| strncpy(drvinfo->version, ql3xxx_driver_version, 32); |
| strncpy(drvinfo->fw_version, "N/A", 32); |
| strncpy(drvinfo->bus_info, pci_name(qdev->pdev), 32); |
| drvinfo->regdump_len = 0; |
| drvinfo->eedump_len = 0; |
| } |
| |
| static u32 ql_get_msglevel(struct net_device *ndev) |
| { |
| struct ql3_adapter *qdev = netdev_priv(ndev); |
| return qdev->msg_enable; |
| } |
| |
| static void ql_set_msglevel(struct net_device *ndev, u32 value) |
| { |
| struct ql3_adapter *qdev = netdev_priv(ndev); |
| qdev->msg_enable = value; |
| } |
| |
| static void ql_get_pauseparam(struct net_device *ndev, |
| struct ethtool_pauseparam *pause) |
| { |
| struct ql3_adapter *qdev = netdev_priv(ndev); |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| |
| u32 reg; |
| if (qdev->mac_index == 0) |
| reg = ql_read_page0_reg(qdev, &port_regs->mac0ConfigReg); |
| else |
| reg = ql_read_page0_reg(qdev, &port_regs->mac1ConfigReg); |
| |
| pause->autoneg = ql_get_auto_cfg_status(qdev); |
| pause->rx_pause = (reg & MAC_CONFIG_REG_RF) >> 2; |
| pause->tx_pause = (reg & MAC_CONFIG_REG_TF) >> 1; |
| } |
| |
| static const struct ethtool_ops ql3xxx_ethtool_ops = { |
| .get_settings = ql_get_settings, |
| .get_drvinfo = ql_get_drvinfo, |
| .get_link = ethtool_op_get_link, |
| .get_msglevel = ql_get_msglevel, |
| .set_msglevel = ql_set_msglevel, |
| .get_pauseparam = ql_get_pauseparam, |
| }; |
| |
| static int ql_populate_free_queue(struct ql3_adapter *qdev) |
| { |
| struct ql_rcv_buf_cb *lrg_buf_cb = qdev->lrg_buf_free_head; |
| dma_addr_t map; |
| int err; |
| |
| while (lrg_buf_cb) { |
| if (!lrg_buf_cb->skb) { |
| lrg_buf_cb->skb = |
| netdev_alloc_skb(qdev->ndev, |
| qdev->lrg_buffer_len); |
| if (unlikely(!lrg_buf_cb->skb)) { |
| netdev_printk(KERN_DEBUG, qdev->ndev, |
| "Failed netdev_alloc_skb()\n"); |
| break; |
| } else { |
| /* |
| * We save some space to copy the ethhdr from |
| * first buffer |
| */ |
| skb_reserve(lrg_buf_cb->skb, QL_HEADER_SPACE); |
| map = pci_map_single(qdev->pdev, |
| lrg_buf_cb->skb->data, |
| qdev->lrg_buffer_len - |
| QL_HEADER_SPACE, |
| PCI_DMA_FROMDEVICE); |
| |
| err = pci_dma_mapping_error(qdev->pdev, map); |
| if (err) { |
| netdev_err(qdev->ndev, |
| "PCI mapping failed with error: %d\n", |
| err); |
| dev_kfree_skb(lrg_buf_cb->skb); |
| lrg_buf_cb->skb = NULL; |
| break; |
| } |
| |
| |
| lrg_buf_cb->buf_phy_addr_low = |
| cpu_to_le32(LS_64BITS(map)); |
| lrg_buf_cb->buf_phy_addr_high = |
| cpu_to_le32(MS_64BITS(map)); |
| dma_unmap_addr_set(lrg_buf_cb, mapaddr, map); |
| dma_unmap_len_set(lrg_buf_cb, maplen, |
| qdev->lrg_buffer_len - |
| QL_HEADER_SPACE); |
| --qdev->lrg_buf_skb_check; |
| if (!qdev->lrg_buf_skb_check) |
| return 1; |
| } |
| } |
| lrg_buf_cb = lrg_buf_cb->next; |
| } |
| return 0; |
| } |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static void ql_update_small_bufq_prod_index(struct ql3_adapter *qdev) |
| { |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| |
| if (qdev->small_buf_release_cnt >= 16) { |
| while (qdev->small_buf_release_cnt >= 16) { |
| qdev->small_buf_q_producer_index++; |
| |
| if (qdev->small_buf_q_producer_index == |
| NUM_SBUFQ_ENTRIES) |
| qdev->small_buf_q_producer_index = 0; |
| qdev->small_buf_release_cnt -= 8; |
| } |
| wmb(); |
| writel(qdev->small_buf_q_producer_index, |
| &port_regs->CommonRegs.rxSmallQProducerIndex); |
| } |
| } |
| |
| /* |
| * Caller holds hw_lock. |
| */ |
| static void ql_update_lrg_bufq_prod_index(struct ql3_adapter *qdev) |
| { |
| struct bufq_addr_element *lrg_buf_q_ele; |
| int i; |
| struct ql_rcv_buf_cb *lrg_buf_cb; |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| |
| if ((qdev->lrg_buf_free_count >= 8) && |
| (qdev->lrg_buf_release_cnt >= 16)) { |
| |
| if (qdev->lrg_buf_skb_check) |
| if (!ql_populate_free_queue(qdev)) |
| return; |
| |
| lrg_buf_q_ele = qdev->lrg_buf_next_free; |
| |
| while ((qdev->lrg_buf_release_cnt >= 16) && |
| (qdev->lrg_buf_free_count >= 8)) { |
| |
| for (i = 0; i < 8; i++) { |
| lrg_buf_cb = |
| ql_get_from_lrg_buf_free_list(qdev); |
| lrg_buf_q_ele->addr_high = |
| lrg_buf_cb->buf_phy_addr_high; |
| lrg_buf_q_ele->addr_low = |
| lrg_buf_cb->buf_phy_addr_low; |
| lrg_buf_q_ele++; |
| |
| qdev->lrg_buf_release_cnt--; |
| } |
| |
| qdev->lrg_buf_q_producer_index++; |
| |
| if (qdev->lrg_buf_q_producer_index == |
| qdev->num_lbufq_entries) |
| qdev->lrg_buf_q_producer_index = 0; |
| |
| if (qdev->lrg_buf_q_producer_index == |
| (qdev->num_lbufq_entries - 1)) { |
| lrg_buf_q_ele = qdev->lrg_buf_q_virt_addr; |
| } |
| } |
| wmb(); |
| qdev->lrg_buf_next_free = lrg_buf_q_ele; |
| writel(qdev->lrg_buf_q_producer_index, |
| &port_regs->CommonRegs.rxLargeQProducerIndex); |
| } |
| } |
| |
| static void ql_process_mac_tx_intr(struct ql3_adapter *qdev, |
| struct ob_mac_iocb_rsp *mac_rsp) |
| { |
| struct ql_tx_buf_cb *tx_cb; |
| int i; |
| int retval = 0; |
| |
| if (mac_rsp->flags & OB_MAC_IOCB_RSP_S) { |
| netdev_warn(qdev->ndev, |
| "Frame too short but it was padded and sent\n"); |
| } |
| |
| tx_cb = &qdev->tx_buf[mac_rsp->transaction_id]; |
| |
| /* Check the transmit response flags for any errors */ |
| if (mac_rsp->flags & OB_MAC_IOCB_RSP_S) { |
| netdev_err(qdev->ndev, |
| "Frame too short to be legal, frame not sent\n"); |
| |
| qdev->ndev->stats.tx_errors++; |
| retval = -EIO; |
| goto frame_not_sent; |
| } |
| |
| if (tx_cb->seg_count == 0) { |
| netdev_err(qdev->ndev, "tx_cb->seg_count == 0: %d\n", |
| mac_rsp->transaction_id); |
| |
| qdev->ndev->stats.tx_errors++; |
| retval = -EIO; |
| goto invalid_seg_count; |
| } |
| |
| pci_unmap_single(qdev->pdev, |
| dma_unmap_addr(&tx_cb->map[0], mapaddr), |
| dma_unmap_len(&tx_cb->map[0], maplen), |
| PCI_DMA_TODEVICE); |
| tx_cb->seg_count--; |
| if (tx_cb->seg_count) { |
| for (i = 1; i < tx_cb->seg_count; i++) { |
| pci_unmap_page(qdev->pdev, |
| dma_unmap_addr(&tx_cb->map[i], |
| mapaddr), |
| dma_unmap_len(&tx_cb->map[i], maplen), |
| PCI_DMA_TODEVICE); |
| } |
| } |
| qdev->ndev->stats.tx_packets++; |
| qdev->ndev->stats.tx_bytes += tx_cb->skb->len; |
| |
| frame_not_sent: |
| dev_kfree_skb_irq(tx_cb->skb); |
| tx_cb->skb = NULL; |
| |
| invalid_seg_count: |
| atomic_inc(&qdev->tx_count); |
| } |
| |
| static void ql_get_sbuf(struct ql3_adapter *qdev) |
| { |
| if (++qdev->small_buf_index == NUM_SMALL_BUFFERS) |
| qdev->small_buf_index = 0; |
| qdev->small_buf_release_cnt++; |
| } |
| |
| static struct ql_rcv_buf_cb *ql_get_lbuf(struct ql3_adapter *qdev) |
| { |
| struct ql_rcv_buf_cb *lrg_buf_cb = NULL; |
| lrg_buf_cb = &qdev->lrg_buf[qdev->lrg_buf_index]; |
| qdev->lrg_buf_release_cnt++; |
| if (++qdev->lrg_buf_index == qdev->num_large_buffers) |
| qdev->lrg_buf_index = 0; |
| return lrg_buf_cb; |
| } |
| |
| /* |
| * The difference between 3022 and 3032 for inbound completions: |
| * 3022 uses two buffers per completion. The first buffer contains |
| * (some) header info, the second the remainder of the headers plus |
| * the data. For this chip we reserve some space at the top of the |
| * receive buffer so that the header info in buffer one can be |
| * prepended to the buffer two. Buffer two is the sent up while |
| * buffer one is returned to the hardware to be reused. |
| * 3032 receives all of it's data and headers in one buffer for a |
| * simpler process. 3032 also supports checksum verification as |
| * can be seen in ql_process_macip_rx_intr(). |
| */ |
| static void ql_process_mac_rx_intr(struct ql3_adapter *qdev, |
| struct ib_mac_iocb_rsp *ib_mac_rsp_ptr) |
| { |
| struct ql_rcv_buf_cb *lrg_buf_cb1 = NULL; |
| struct ql_rcv_buf_cb *lrg_buf_cb2 = NULL; |
| struct sk_buff *skb; |
| u16 length = le16_to_cpu(ib_mac_rsp_ptr->length); |
| |
| /* |
| * Get the inbound address list (small buffer). |
| */ |
| ql_get_sbuf(qdev); |
| |
| if (qdev->device_id == QL3022_DEVICE_ID) |
| lrg_buf_cb1 = ql_get_lbuf(qdev); |
| |
| /* start of second buffer */ |
| lrg_buf_cb2 = ql_get_lbuf(qdev); |
| skb = lrg_buf_cb2->skb; |
| |
| qdev->ndev->stats.rx_packets++; |
| qdev->ndev->stats.rx_bytes += length; |
| |
| skb_put(skb, length); |
| pci_unmap_single(qdev->pdev, |
| dma_unmap_addr(lrg_buf_cb2, mapaddr), |
| dma_unmap_len(lrg_buf_cb2, maplen), |
| PCI_DMA_FROMDEVICE); |
| prefetch(skb->data); |
| skb_checksum_none_assert(skb); |
| skb->protocol = eth_type_trans(skb, qdev->ndev); |
| |
| netif_receive_skb(skb); |
| lrg_buf_cb2->skb = NULL; |
| |
| if (qdev->device_id == QL3022_DEVICE_ID) |
| ql_release_to_lrg_buf_free_list(qdev, lrg_buf_cb1); |
| ql_release_to_lrg_buf_free_list(qdev, lrg_buf_cb2); |
| } |
| |
| static void ql_process_macip_rx_intr(struct ql3_adapter *qdev, |
| struct ib_ip_iocb_rsp *ib_ip_rsp_ptr) |
| { |
| struct ql_rcv_buf_cb *lrg_buf_cb1 = NULL; |
| struct ql_rcv_buf_cb *lrg_buf_cb2 = NULL; |
| struct sk_buff *skb1 = NULL, *skb2; |
| struct net_device *ndev = qdev->ndev; |
| u16 length = le16_to_cpu(ib_ip_rsp_ptr->length); |
| u16 size = 0; |
| |
| /* |
| * Get the inbound address list (small buffer). |
| */ |
| |
| ql_get_sbuf(qdev); |
| |
| if (qdev->device_id == QL3022_DEVICE_ID) { |
| /* start of first buffer on 3022 */ |
| lrg_buf_cb1 = ql_get_lbuf(qdev); |
| skb1 = lrg_buf_cb1->skb; |
| size = ETH_HLEN; |
| if (*((u16 *) skb1->data) != 0xFFFF) |
| size += VLAN_ETH_HLEN - ETH_HLEN; |
| } |
| |
| /* start of second buffer */ |
| lrg_buf_cb2 = ql_get_lbuf(qdev); |
| skb2 = lrg_buf_cb2->skb; |
| |
| skb_put(skb2, length); /* Just the second buffer length here. */ |
| pci_unmap_single(qdev->pdev, |
| dma_unmap_addr(lrg_buf_cb2, mapaddr), |
| dma_unmap_len(lrg_buf_cb2, maplen), |
| PCI_DMA_FROMDEVICE); |
| prefetch(skb2->data); |
| |
| skb_checksum_none_assert(skb2); |
| if (qdev->device_id == QL3022_DEVICE_ID) { |
| /* |
| * Copy the ethhdr from first buffer to second. This |
| * is necessary for 3022 IP completions. |
| */ |
| skb_copy_from_linear_data_offset(skb1, VLAN_ID_LEN, |
| skb_push(skb2, size), size); |
| } else { |
| u16 checksum = le16_to_cpu(ib_ip_rsp_ptr->checksum); |
| if (checksum & |
| (IB_IP_IOCB_RSP_3032_ICE | |
| IB_IP_IOCB_RSP_3032_CE)) { |
| netdev_err(ndev, |
| "%s: Bad checksum for this %s packet, checksum = %x\n", |
| __func__, |
| ((checksum & IB_IP_IOCB_RSP_3032_TCP) ? |
| "TCP" : "UDP"), checksum); |
| } else if ((checksum & IB_IP_IOCB_RSP_3032_TCP) || |
| (checksum & IB_IP_IOCB_RSP_3032_UDP && |
| !(checksum & IB_IP_IOCB_RSP_3032_NUC))) { |
| skb2->ip_summed = CHECKSUM_UNNECESSARY; |
| } |
| } |
| skb2->protocol = eth_type_trans(skb2, qdev->ndev); |
| |
| netif_receive_skb(skb2); |
| ndev->stats.rx_packets++; |
| ndev->stats.rx_bytes += length; |
| lrg_buf_cb2->skb = NULL; |
| |
| if (qdev->device_id == QL3022_DEVICE_ID) |
| ql_release_to_lrg_buf_free_list(qdev, lrg_buf_cb1); |
| ql_release_to_lrg_buf_free_list(qdev, lrg_buf_cb2); |
| } |
| |
| static int ql_tx_rx_clean(struct ql3_adapter *qdev, |
| int *tx_cleaned, int *rx_cleaned, int work_to_do) |
| { |
| struct net_rsp_iocb *net_rsp; |
| struct net_device *ndev = qdev->ndev; |
| int work_done = 0; |
| |
| /* While there are entries in the completion queue. */ |
| while ((le32_to_cpu(*(qdev->prsp_producer_index)) != |
| qdev->rsp_consumer_index) && (work_done < work_to_do)) { |
| |
| net_rsp = qdev->rsp_current; |
| rmb(); |
| /* |
| * Fix 4032 chip's undocumented "feature" where bit-8 is set |
| * if the inbound completion is for a VLAN. |
| */ |
| if (qdev->device_id == QL3032_DEVICE_ID) |
| net_rsp->opcode &= 0x7f; |
| switch (net_rsp->opcode) { |
| |
| case OPCODE_OB_MAC_IOCB_FN0: |
| case OPCODE_OB_MAC_IOCB_FN2: |
| ql_process_mac_tx_intr(qdev, (struct ob_mac_iocb_rsp *) |
| net_rsp); |
| (*tx_cleaned)++; |
| break; |
| |
| case OPCODE_IB_MAC_IOCB: |
| case OPCODE_IB_3032_MAC_IOCB: |
| ql_process_mac_rx_intr(qdev, (struct ib_mac_iocb_rsp *) |
| net_rsp); |
| (*rx_cleaned)++; |
| break; |
| |
| case OPCODE_IB_IP_IOCB: |
| case OPCODE_IB_3032_IP_IOCB: |
| ql_process_macip_rx_intr(qdev, (struct ib_ip_iocb_rsp *) |
| net_rsp); |
| (*rx_cleaned)++; |
| break; |
| default: { |
| u32 *tmp = (u32 *)net_rsp; |
| netdev_err(ndev, |
| "Hit default case, not handled!\n" |
| " dropping the packet, opcode = %x\n" |
| "0x%08lx 0x%08lx 0x%08lx 0x%08lx\n", |
| net_rsp->opcode, |
| (unsigned long int)tmp[0], |
| (unsigned long int)tmp[1], |
| (unsigned long int)tmp[2], |
| (unsigned long int)tmp[3]); |
| } |
| } |
| |
| qdev->rsp_consumer_index++; |
| |
| if (qdev->rsp_consumer_index == NUM_RSP_Q_ENTRIES) { |
| qdev->rsp_consumer_index = 0; |
| qdev->rsp_current = qdev->rsp_q_virt_addr; |
| } else { |
| qdev->rsp_current++; |
| } |
| |
| work_done = *tx_cleaned + *rx_cleaned; |
| } |
| |
| return work_done; |
| } |
| |
| static int ql_poll(struct napi_struct *napi, int budget) |
| { |
| struct ql3_adapter *qdev = container_of(napi, struct ql3_adapter, napi); |
| int rx_cleaned = 0, tx_cleaned = 0; |
| unsigned long hw_flags; |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| |
| ql_tx_rx_clean(qdev, &tx_cleaned, &rx_cleaned, budget); |
| |
| if (tx_cleaned + rx_cleaned != budget) { |
| spin_lock_irqsave(&qdev->hw_lock, hw_flags); |
| __napi_complete(napi); |
| ql_update_small_bufq_prod_index(qdev); |
| ql_update_lrg_bufq_prod_index(qdev); |
| writel(qdev->rsp_consumer_index, |
| &port_regs->CommonRegs.rspQConsumerIndex); |
| spin_unlock_irqrestore(&qdev->hw_lock, hw_flags); |
| |
| ql_enable_interrupts(qdev); |
| } |
| return tx_cleaned + rx_cleaned; |
| } |
| |
| static irqreturn_t ql3xxx_isr(int irq, void *dev_id) |
| { |
| |
| struct net_device *ndev = dev_id; |
| struct ql3_adapter *qdev = netdev_priv(ndev); |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 value; |
| int handled = 1; |
| u32 var; |
| |
| value = ql_read_common_reg_l(qdev, |
| &port_regs->CommonRegs.ispControlStatus); |
| |
| if (value & (ISP_CONTROL_FE | ISP_CONTROL_RI)) { |
| spin_lock(&qdev->adapter_lock); |
| netif_stop_queue(qdev->ndev); |
| netif_carrier_off(qdev->ndev); |
| ql_disable_interrupts(qdev); |
| qdev->port_link_state = LS_DOWN; |
| set_bit(QL_RESET_ACTIVE, &qdev->flags) ; |
| |
| if (value & ISP_CONTROL_FE) { |
| /* |
| * Chip Fatal Error. |
| */ |
| var = |
| ql_read_page0_reg_l(qdev, |
| &port_regs->PortFatalErrStatus); |
| netdev_warn(ndev, |
| "Resetting chip. PortFatalErrStatus register = 0x%x\n", |
| var); |
| set_bit(QL_RESET_START, &qdev->flags) ; |
| } else { |
| /* |
| * Soft Reset Requested. |
| */ |
| set_bit(QL_RESET_PER_SCSI, &qdev->flags) ; |
| netdev_err(ndev, |
| "Another function issued a reset to the chip. ISR value = %x\n", |
| value); |
| } |
| queue_delayed_work(qdev->workqueue, &qdev->reset_work, 0); |
| spin_unlock(&qdev->adapter_lock); |
| } else if (value & ISP_IMR_DISABLE_CMPL_INT) { |
| ql_disable_interrupts(qdev); |
| if (likely(napi_schedule_prep(&qdev->napi))) |
| __napi_schedule(&qdev->napi); |
| } else |
| return IRQ_NONE; |
| |
| return IRQ_RETVAL(handled); |
| } |
| |
| /* |
| * Get the total number of segments needed for the given number of fragments. |
| * This is necessary because outbound address lists (OAL) will be used when |
| * more than two frags are given. Each address list has 5 addr/len pairs. |
| * The 5th pair in each OAL is used to point to the next OAL if more frags |
| * are coming. That is why the frags:segment count ratio is not linear. |
| */ |
| static int ql_get_seg_count(struct ql3_adapter *qdev, unsigned short frags) |
| { |
| if (qdev->device_id == QL3022_DEVICE_ID) |
| return 1; |
| |
| if (frags <= 2) |
| return frags + 1; |
| else if (frags <= 6) |
| return frags + 2; |
| else if (frags <= 10) |
| return frags + 3; |
| else if (frags <= 14) |
| return frags + 4; |
| else if (frags <= 18) |
| return frags + 5; |
| return -1; |
| } |
| |
| static void ql_hw_csum_setup(const struct sk_buff *skb, |
| struct ob_mac_iocb_req *mac_iocb_ptr) |
| { |
| const struct iphdr *ip = ip_hdr(skb); |
| |
| mac_iocb_ptr->ip_hdr_off = skb_network_offset(skb); |
| mac_iocb_ptr->ip_hdr_len = ip->ihl; |
| |
| if (ip->protocol == IPPROTO_TCP) { |
| mac_iocb_ptr->flags1 |= OB_3032MAC_IOCB_REQ_TC | |
| OB_3032MAC_IOCB_REQ_IC; |
| } else { |
| mac_iocb_ptr->flags1 |= OB_3032MAC_IOCB_REQ_UC | |
| OB_3032MAC_IOCB_REQ_IC; |
| } |
| |
| } |
| |
| /* |
| * Map the buffers for this transmit. |
| * This will return NETDEV_TX_BUSY or NETDEV_TX_OK based on success. |
| */ |
| static int ql_send_map(struct ql3_adapter *qdev, |
| struct ob_mac_iocb_req *mac_iocb_ptr, |
| struct ql_tx_buf_cb *tx_cb, |
| struct sk_buff *skb) |
| { |
| struct oal *oal; |
| struct oal_entry *oal_entry; |
| int len = skb_headlen(skb); |
| dma_addr_t map; |
| int err; |
| int completed_segs, i; |
| int seg_cnt, seg = 0; |
| int frag_cnt = (int)skb_shinfo(skb)->nr_frags; |
| |
| seg_cnt = tx_cb->seg_count; |
| /* |
| * Map the skb buffer first. |
| */ |
| map = pci_map_single(qdev->pdev, skb->data, len, PCI_DMA_TODEVICE); |
| |
| err = pci_dma_mapping_error(qdev->pdev, map); |
| if (err) { |
| netdev_err(qdev->ndev, "PCI mapping failed with error: %d\n", |
| err); |
| |
| return NETDEV_TX_BUSY; |
| } |
| |
| oal_entry = (struct oal_entry *)&mac_iocb_ptr->buf_addr0_low; |
| oal_entry->dma_lo = cpu_to_le32(LS_64BITS(map)); |
| oal_entry->dma_hi = cpu_to_le32(MS_64BITS(map)); |
| oal_entry->len = cpu_to_le32(len); |
| dma_unmap_addr_set(&tx_cb->map[seg], mapaddr, map); |
| dma_unmap_len_set(&tx_cb->map[seg], maplen, len); |
| seg++; |
| |
| if (seg_cnt == 1) { |
| /* Terminate the last segment. */ |
| oal_entry->len |= cpu_to_le32(OAL_LAST_ENTRY); |
| return NETDEV_TX_OK; |
| } |
| oal = tx_cb->oal; |
| for (completed_segs = 0; |
| completed_segs < frag_cnt; |
| completed_segs++, seg++) { |
| skb_frag_t *frag = &skb_shinfo(skb)->frags[completed_segs]; |
| oal_entry++; |
| /* |
| * Check for continuation requirements. |
| * It's strange but necessary. |
| * Continuation entry points to outbound address list. |
| */ |
| if ((seg == 2 && seg_cnt > 3) || |
| (seg == 7 && seg_cnt > 8) || |
| (seg == 12 && seg_cnt > 13) || |
| (seg == 17 && seg_cnt > 18)) { |
| map = pci_map_single(qdev->pdev, oal, |
| sizeof(struct oal), |
| PCI_DMA_TODEVICE); |
| |
| err = pci_dma_mapping_error(qdev->pdev, map); |
| if (err) { |
| netdev_err(qdev->ndev, |
| "PCI mapping outbound address list with error: %d\n", |
| err); |
| goto map_error; |
| } |
| |
| oal_entry->dma_lo = cpu_to_le32(LS_64BITS(map)); |
| oal_entry->dma_hi = cpu_to_le32(MS_64BITS(map)); |
| oal_entry->len = cpu_to_le32(sizeof(struct oal) | |
| OAL_CONT_ENTRY); |
| dma_unmap_addr_set(&tx_cb->map[seg], mapaddr, map); |
| dma_unmap_len_set(&tx_cb->map[seg], maplen, |
| sizeof(struct oal)); |
| oal_entry = (struct oal_entry *)oal; |
| oal++; |
| seg++; |
| } |
| |
| map = pci_map_page(qdev->pdev, frag->page, |
| frag->page_offset, frag->size, |
| PCI_DMA_TODEVICE); |
| |
| err = pci_dma_mapping_error(qdev->pdev, map); |
| if (err) { |
| netdev_err(qdev->ndev, |
| "PCI mapping frags failed with error: %d\n", |
| err); |
| goto map_error; |
| } |
| |
| oal_entry->dma_lo = cpu_to_le32(LS_64BITS(map)); |
| oal_entry->dma_hi = cpu_to_le32(MS_64BITS(map)); |
| oal_entry->len = cpu_to_le32(frag->size); |
| dma_unmap_addr_set(&tx_cb->map[seg], mapaddr, map); |
| dma_unmap_len_set(&tx_cb->map[seg], maplen, frag->size); |
| } |
| /* Terminate the last segment. */ |
| oal_entry->len |= cpu_to_le32(OAL_LAST_ENTRY); |
| return NETDEV_TX_OK; |
| |
| map_error: |
| /* A PCI mapping failed and now we will need to back out |
| * We need to traverse through the oal's and associated pages which |
| * have been mapped and now we must unmap them to clean up properly |
| */ |
| |
| seg = 1; |
| oal_entry = (struct oal_entry *)&mac_iocb_ptr->buf_addr0_low; |
| oal = tx_cb->oal; |
| for (i = 0; i < completed_segs; i++, seg++) { |
| oal_entry++; |
| |
| /* |
| * Check for continuation requirements. |
| * It's strange but necessary. |
| */ |
| |
| if ((seg == 2 && seg_cnt > 3) || |
| (seg == 7 && seg_cnt > 8) || |
| (seg == 12 && seg_cnt > 13) || |
| (seg == 17 && seg_cnt > 18)) { |
| pci_unmap_single(qdev->pdev, |
| dma_unmap_addr(&tx_cb->map[seg], mapaddr), |
| dma_unmap_len(&tx_cb->map[seg], maplen), |
| PCI_DMA_TODEVICE); |
| oal++; |
| seg++; |
| } |
| |
| pci_unmap_page(qdev->pdev, |
| dma_unmap_addr(&tx_cb->map[seg], mapaddr), |
| dma_unmap_len(&tx_cb->map[seg], maplen), |
| PCI_DMA_TODEVICE); |
| } |
| |
| pci_unmap_single(qdev->pdev, |
| dma_unmap_addr(&tx_cb->map[0], mapaddr), |
| dma_unmap_addr(&tx_cb->map[0], maplen), |
| PCI_DMA_TODEVICE); |
| |
| return NETDEV_TX_BUSY; |
| |
| } |
| |
| /* |
| * The difference between 3022 and 3032 sends: |
| * 3022 only supports a simple single segment transmission. |
| * 3032 supports checksumming and scatter/gather lists (fragments). |
| * The 3032 supports sglists by using the 3 addr/len pairs (ALP) |
| * in the IOCB plus a chain of outbound address lists (OAL) that |
| * each contain 5 ALPs. The last ALP of the IOCB (3rd) or OAL (5th) |
| * will used to point to an OAL when more ALP entries are required. |
| * The IOCB is always the top of the chain followed by one or more |
| * OALs (when necessary). |
| */ |
| static netdev_tx_t ql3xxx_send(struct sk_buff *skb, |
| struct net_device *ndev) |
| { |
| struct ql3_adapter *qdev = (struct ql3_adapter *)netdev_priv(ndev); |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| struct ql_tx_buf_cb *tx_cb; |
| u32 tot_len = skb->len; |
| struct ob_mac_iocb_req *mac_iocb_ptr; |
| |
| if (unlikely(atomic_read(&qdev->tx_count) < 2)) |
| return NETDEV_TX_BUSY; |
| |
| tx_cb = &qdev->tx_buf[qdev->req_producer_index]; |
| tx_cb->seg_count = ql_get_seg_count(qdev, |
| skb_shinfo(skb)->nr_frags); |
| if (tx_cb->seg_count == -1) { |
| netdev_err(ndev, "%s: invalid segment count!\n", __func__); |
| return NETDEV_TX_OK; |
| } |
| |
| mac_iocb_ptr = tx_cb->queue_entry; |
| memset((void *)mac_iocb_ptr, 0, sizeof(struct ob_mac_iocb_req)); |
| mac_iocb_ptr->opcode = qdev->mac_ob_opcode; |
| mac_iocb_ptr->flags = OB_MAC_IOCB_REQ_X; |
| mac_iocb_ptr->flags |= qdev->mb_bit_mask; |
| mac_iocb_ptr->transaction_id = qdev->req_producer_index; |
| mac_iocb_ptr->data_len = cpu_to_le16((u16) tot_len); |
| tx_cb->skb = skb; |
| if (qdev->device_id == QL3032_DEVICE_ID && |
| skb->ip_summed == CHECKSUM_PARTIAL) |
| ql_hw_csum_setup(skb, mac_iocb_ptr); |
| |
| if (ql_send_map(qdev, mac_iocb_ptr, tx_cb, skb) != NETDEV_TX_OK) { |
| netdev_err(ndev, "%s: Could not map the segments!\n", __func__); |
| return NETDEV_TX_BUSY; |
| } |
| |
| wmb(); |
| qdev->req_producer_index++; |
| if (qdev->req_producer_index == NUM_REQ_Q_ENTRIES) |
| qdev->req_producer_index = 0; |
| wmb(); |
| ql_write_common_reg_l(qdev, |
| &port_regs->CommonRegs.reqQProducerIndex, |
| qdev->req_producer_index); |
| |
| netif_printk(qdev, tx_queued, KERN_DEBUG, ndev, |
| "tx queued, slot %d, len %d\n", |
| qdev->req_producer_index, skb->len); |
| |
| atomic_dec(&qdev->tx_count); |
| return NETDEV_TX_OK; |
| } |
| |
| static int ql_alloc_net_req_rsp_queues(struct ql3_adapter *qdev) |
| { |
| qdev->req_q_size = |
| (u32) (NUM_REQ_Q_ENTRIES * sizeof(struct ob_mac_iocb_req)); |
| |
| qdev->req_q_virt_addr = |
| pci_alloc_consistent(qdev->pdev, |
| (size_t) qdev->req_q_size, |
| &qdev->req_q_phy_addr); |
| |
| if ((qdev->req_q_virt_addr == NULL) || |
| LS_64BITS(qdev->req_q_phy_addr) & (qdev->req_q_size - 1)) { |
| netdev_err(qdev->ndev, "reqQ failed\n"); |
| return -ENOMEM; |
| } |
| |
| qdev->rsp_q_size = NUM_RSP_Q_ENTRIES * sizeof(struct net_rsp_iocb); |
| |
| qdev->rsp_q_virt_addr = |
| pci_alloc_consistent(qdev->pdev, |
| (size_t) qdev->rsp_q_size, |
| &qdev->rsp_q_phy_addr); |
| |
| if ((qdev->rsp_q_virt_addr == NULL) || |
| LS_64BITS(qdev->rsp_q_phy_addr) & (qdev->rsp_q_size - 1)) { |
| netdev_err(qdev->ndev, "rspQ allocation failed\n"); |
| pci_free_consistent(qdev->pdev, (size_t) qdev->req_q_size, |
| qdev->req_q_virt_addr, |
| qdev->req_q_phy_addr); |
| return -ENOMEM; |
| } |
| |
| set_bit(QL_ALLOC_REQ_RSP_Q_DONE, &qdev->flags); |
| |
| return 0; |
| } |
| |
| static void ql_free_net_req_rsp_queues(struct ql3_adapter *qdev) |
| { |
| if (!test_bit(QL_ALLOC_REQ_RSP_Q_DONE, &qdev->flags)) { |
| netdev_info(qdev->ndev, "Already done\n"); |
| return; |
| } |
| |
| pci_free_consistent(qdev->pdev, |
| qdev->req_q_size, |
| qdev->req_q_virt_addr, qdev->req_q_phy_addr); |
| |
| qdev->req_q_virt_addr = NULL; |
| |
| pci_free_consistent(qdev->pdev, |
| qdev->rsp_q_size, |
| qdev->rsp_q_virt_addr, qdev->rsp_q_phy_addr); |
| |
| qdev->rsp_q_virt_addr = NULL; |
| |
| clear_bit(QL_ALLOC_REQ_RSP_Q_DONE, &qdev->flags); |
| } |
| |
| static int ql_alloc_buffer_queues(struct ql3_adapter *qdev) |
| { |
| /* Create Large Buffer Queue */ |
| qdev->lrg_buf_q_size = |
| qdev->num_lbufq_entries * sizeof(struct lrg_buf_q_entry); |
| if (qdev->lrg_buf_q_size < PAGE_SIZE) |
| qdev->lrg_buf_q_alloc_size = PAGE_SIZE; |
| else |
| qdev->lrg_buf_q_alloc_size = qdev->lrg_buf_q_size * 2; |
| |
| qdev->lrg_buf = |
| kmalloc(qdev->num_large_buffers * sizeof(struct ql_rcv_buf_cb), |
| GFP_KERNEL); |
| if (qdev->lrg_buf == NULL) { |
| netdev_err(qdev->ndev, "qdev->lrg_buf alloc failed\n"); |
| return -ENOMEM; |
| } |
| |
| qdev->lrg_buf_q_alloc_virt_addr = |
| pci_alloc_consistent(qdev->pdev, |
| qdev->lrg_buf_q_alloc_size, |
| &qdev->lrg_buf_q_alloc_phy_addr); |
| |
| if (qdev->lrg_buf_q_alloc_virt_addr == NULL) { |
| netdev_err(qdev->ndev, "lBufQ failed\n"); |
| return -ENOMEM; |
| } |
| qdev->lrg_buf_q_virt_addr = qdev->lrg_buf_q_alloc_virt_addr; |
| qdev->lrg_buf_q_phy_addr = qdev->lrg_buf_q_alloc_phy_addr; |
| |
| /* Create Small Buffer Queue */ |
| qdev->small_buf_q_size = |
| NUM_SBUFQ_ENTRIES * sizeof(struct lrg_buf_q_entry); |
| if (qdev->small_buf_q_size < PAGE_SIZE) |
| qdev->small_buf_q_alloc_size = PAGE_SIZE; |
| else |
| qdev->small_buf_q_alloc_size = qdev->small_buf_q_size * 2; |
| |
| qdev->small_buf_q_alloc_virt_addr = |
| pci_alloc_consistent(qdev->pdev, |
| qdev->small_buf_q_alloc_size, |
| &qdev->small_buf_q_alloc_phy_addr); |
| |
| if (qdev->small_buf_q_alloc_virt_addr == NULL) { |
| netdev_err(qdev->ndev, "Small Buffer Queue allocation failed\n"); |
| pci_free_consistent(qdev->pdev, qdev->lrg_buf_q_alloc_size, |
| qdev->lrg_buf_q_alloc_virt_addr, |
| qdev->lrg_buf_q_alloc_phy_addr); |
| return -ENOMEM; |
| } |
| |
| qdev->small_buf_q_virt_addr = qdev->small_buf_q_alloc_virt_addr; |
| qdev->small_buf_q_phy_addr = qdev->small_buf_q_alloc_phy_addr; |
| set_bit(QL_ALLOC_BUFQS_DONE, &qdev->flags); |
| return 0; |
| } |
| |
| static void ql_free_buffer_queues(struct ql3_adapter *qdev) |
| { |
| if (!test_bit(QL_ALLOC_BUFQS_DONE, &qdev->flags)) { |
| netdev_info(qdev->ndev, "Already done\n"); |
| return; |
| } |
| kfree(qdev->lrg_buf); |
| pci_free_consistent(qdev->pdev, |
| qdev->lrg_buf_q_alloc_size, |
| qdev->lrg_buf_q_alloc_virt_addr, |
| qdev->lrg_buf_q_alloc_phy_addr); |
| |
| qdev->lrg_buf_q_virt_addr = NULL; |
| |
| pci_free_consistent(qdev->pdev, |
| qdev->small_buf_q_alloc_size, |
| qdev->small_buf_q_alloc_virt_addr, |
| qdev->small_buf_q_alloc_phy_addr); |
| |
| qdev->small_buf_q_virt_addr = NULL; |
| |
| clear_bit(QL_ALLOC_BUFQS_DONE, &qdev->flags); |
| } |
| |
| static int ql_alloc_small_buffers(struct ql3_adapter *qdev) |
| { |
| int i; |
| struct bufq_addr_element *small_buf_q_entry; |
| |
| /* Currently we allocate on one of memory and use it for smallbuffers */ |
| qdev->small_buf_total_size = |
| (QL_ADDR_ELE_PER_BUFQ_ENTRY * NUM_SBUFQ_ENTRIES * |
| QL_SMALL_BUFFER_SIZE); |
| |
| qdev->small_buf_virt_addr = |
| pci_alloc_consistent(qdev->pdev, |
| qdev->small_buf_total_size, |
| &qdev->small_buf_phy_addr); |
| |
| if (qdev->small_buf_virt_addr == NULL) { |
| netdev_err(qdev->ndev, "Failed to get small buffer memory\n"); |
| return -ENOMEM; |
| } |
| |
| qdev->small_buf_phy_addr_low = LS_64BITS(qdev->small_buf_phy_addr); |
| qdev->small_buf_phy_addr_high = MS_64BITS(qdev->small_buf_phy_addr); |
| |
| small_buf_q_entry = qdev->small_buf_q_virt_addr; |
| |
| /* Initialize the small buffer queue. */ |
| for (i = 0; i < (QL_ADDR_ELE_PER_BUFQ_ENTRY * NUM_SBUFQ_ENTRIES); i++) { |
| small_buf_q_entry->addr_high = |
| cpu_to_le32(qdev->small_buf_phy_addr_high); |
| small_buf_q_entry->addr_low = |
| cpu_to_le32(qdev->small_buf_phy_addr_low + |
| (i * QL_SMALL_BUFFER_SIZE)); |
| small_buf_q_entry++; |
| } |
| qdev->small_buf_index = 0; |
| set_bit(QL_ALLOC_SMALL_BUF_DONE, &qdev->flags); |
| return 0; |
| } |
| |
| static void ql_free_small_buffers(struct ql3_adapter *qdev) |
| { |
| if (!test_bit(QL_ALLOC_SMALL_BUF_DONE, &qdev->flags)) { |
| netdev_info(qdev->ndev, "Already done\n"); |
| return; |
| } |
| if (qdev->small_buf_virt_addr != NULL) { |
| pci_free_consistent(qdev->pdev, |
| qdev->small_buf_total_size, |
| qdev->small_buf_virt_addr, |
| qdev->small_buf_phy_addr); |
| |
| qdev->small_buf_virt_addr = NULL; |
| } |
| } |
| |
| static void ql_free_large_buffers(struct ql3_adapter *qdev) |
| { |
| int i = 0; |
| struct ql_rcv_buf_cb *lrg_buf_cb; |
| |
| for (i = 0; i < qdev->num_large_buffers; i++) { |
| lrg_buf_cb = &qdev->lrg_buf[i]; |
| if (lrg_buf_cb->skb) { |
| dev_kfree_skb(lrg_buf_cb->skb); |
| pci_unmap_single(qdev->pdev, |
| dma_unmap_addr(lrg_buf_cb, mapaddr), |
| dma_unmap_len(lrg_buf_cb, maplen), |
| PCI_DMA_FROMDEVICE); |
| memset(lrg_buf_cb, 0, sizeof(struct ql_rcv_buf_cb)); |
| } else { |
| break; |
| } |
| } |
| } |
| |
| static void ql_init_large_buffers(struct ql3_adapter *qdev) |
| { |
| int i; |
| struct ql_rcv_buf_cb *lrg_buf_cb; |
| struct bufq_addr_element *buf_addr_ele = qdev->lrg_buf_q_virt_addr; |
| |
| for (i = 0; i < qdev->num_large_buffers; i++) { |
| lrg_buf_cb = &qdev->lrg_buf[i]; |
| buf_addr_ele->addr_high = lrg_buf_cb->buf_phy_addr_high; |
| buf_addr_ele->addr_low = lrg_buf_cb->buf_phy_addr_low; |
| buf_addr_ele++; |
| } |
| qdev->lrg_buf_index = 0; |
| qdev->lrg_buf_skb_check = 0; |
| } |
| |
| static int ql_alloc_large_buffers(struct ql3_adapter *qdev) |
| { |
| int i; |
| struct ql_rcv_buf_cb *lrg_buf_cb; |
| struct sk_buff *skb; |
| dma_addr_t map; |
| int err; |
| |
| for (i = 0; i < qdev->num_large_buffers; i++) { |
| skb = netdev_alloc_skb(qdev->ndev, |
| qdev->lrg_buffer_len); |
| if (unlikely(!skb)) { |
| /* Better luck next round */ |
| netdev_err(qdev->ndev, |
| "large buff alloc failed for %d bytes at index %d\n", |
| qdev->lrg_buffer_len * 2, i); |
| ql_free_large_buffers(qdev); |
| return -ENOMEM; |
| } else { |
| |
| lrg_buf_cb = &qdev->lrg_buf[i]; |
| memset(lrg_buf_cb, 0, sizeof(struct ql_rcv_buf_cb)); |
| lrg_buf_cb->index = i; |
| lrg_buf_cb->skb = skb; |
| /* |
| * We save some space to copy the ethhdr from first |
| * buffer |
| */ |
| skb_reserve(skb, QL_HEADER_SPACE); |
| map = pci_map_single(qdev->pdev, |
| skb->data, |
| qdev->lrg_buffer_len - |
| QL_HEADER_SPACE, |
| PCI_DMA_FROMDEVICE); |
| |
| err = pci_dma_mapping_error(qdev->pdev, map); |
| if (err) { |
| netdev_err(qdev->ndev, |
| "PCI mapping failed with error: %d\n", |
| err); |
| ql_free_large_buffers(qdev); |
| return -ENOMEM; |
| } |
| |
| dma_unmap_addr_set(lrg_buf_cb, mapaddr, map); |
| dma_unmap_len_set(lrg_buf_cb, maplen, |
| qdev->lrg_buffer_len - |
| QL_HEADER_SPACE); |
| lrg_buf_cb->buf_phy_addr_low = |
| cpu_to_le32(LS_64BITS(map)); |
| lrg_buf_cb->buf_phy_addr_high = |
| cpu_to_le32(MS_64BITS(map)); |
| } |
| } |
| return 0; |
| } |
| |
| static void ql_free_send_free_list(struct ql3_adapter *qdev) |
| { |
| struct ql_tx_buf_cb *tx_cb; |
| int i; |
| |
| tx_cb = &qdev->tx_buf[0]; |
| for (i = 0; i < NUM_REQ_Q_ENTRIES; i++) { |
| kfree(tx_cb->oal); |
| tx_cb->oal = NULL; |
| tx_cb++; |
| } |
| } |
| |
| static int ql_create_send_free_list(struct ql3_adapter *qdev) |
| { |
| struct ql_tx_buf_cb *tx_cb; |
| int i; |
| struct ob_mac_iocb_req *req_q_curr = qdev->req_q_virt_addr; |
| |
| /* Create free list of transmit buffers */ |
| for (i = 0; i < NUM_REQ_Q_ENTRIES; i++) { |
| |
| tx_cb = &qdev->tx_buf[i]; |
| tx_cb->skb = NULL; |
| tx_cb->queue_entry = req_q_curr; |
| req_q_curr++; |
| tx_cb->oal = kmalloc(512, GFP_KERNEL); |
| if (tx_cb->oal == NULL) |
| return -1; |
| } |
| return 0; |
| } |
| |
| static int ql_alloc_mem_resources(struct ql3_adapter *qdev) |
| { |
| if (qdev->ndev->mtu == NORMAL_MTU_SIZE) { |
| qdev->num_lbufq_entries = NUM_LBUFQ_ENTRIES; |
| qdev->lrg_buffer_len = NORMAL_MTU_SIZE; |
| } else if (qdev->ndev->mtu == JUMBO_MTU_SIZE) { |
| /* |
| * Bigger buffers, so less of them. |
| */ |
| qdev->num_lbufq_entries = JUMBO_NUM_LBUFQ_ENTRIES; |
| qdev->lrg_buffer_len = JUMBO_MTU_SIZE; |
| } else { |
| netdev_err(qdev->ndev, "Invalid mtu size: %d. Only %d and %d are accepted.\n", |
| qdev->ndev->mtu, NORMAL_MTU_SIZE, JUMBO_MTU_SIZE); |
| return -ENOMEM; |
| } |
| qdev->num_large_buffers = |
| qdev->num_lbufq_entries * QL_ADDR_ELE_PER_BUFQ_ENTRY; |
| qdev->lrg_buffer_len += VLAN_ETH_HLEN + VLAN_ID_LEN + QL_HEADER_SPACE; |
| qdev->max_frame_size = |
| (qdev->lrg_buffer_len - QL_HEADER_SPACE) + ETHERNET_CRC_SIZE; |
| |
| /* |
| * First allocate a page of shared memory and use it for shadow |
| * locations of Network Request Queue Consumer Address Register and |
| * Network Completion Queue Producer Index Register |
| */ |
| qdev->shadow_reg_virt_addr = |
| pci_alloc_consistent(qdev->pdev, |
| PAGE_SIZE, &qdev->shadow_reg_phy_addr); |
| |
| if (qdev->shadow_reg_virt_addr != NULL) { |
| qdev->preq_consumer_index = (u16 *) qdev->shadow_reg_virt_addr; |
| qdev->req_consumer_index_phy_addr_high = |
| MS_64BITS(qdev->shadow_reg_phy_addr); |
| qdev->req_consumer_index_phy_addr_low = |
| LS_64BITS(qdev->shadow_reg_phy_addr); |
| |
| qdev->prsp_producer_index = |
| (__le32 *) (((u8 *) qdev->preq_consumer_index) + 8); |
| qdev->rsp_producer_index_phy_addr_high = |
| qdev->req_consumer_index_phy_addr_high; |
| qdev->rsp_producer_index_phy_addr_low = |
| qdev->req_consumer_index_phy_addr_low + 8; |
| } else { |
| netdev_err(qdev->ndev, "shadowReg Alloc failed\n"); |
| return -ENOMEM; |
| } |
| |
| if (ql_alloc_net_req_rsp_queues(qdev) != 0) { |
| netdev_err(qdev->ndev, "ql_alloc_net_req_rsp_queues failed\n"); |
| goto err_req_rsp; |
| } |
| |
| if (ql_alloc_buffer_queues(qdev) != 0) { |
| netdev_err(qdev->ndev, "ql_alloc_buffer_queues failed\n"); |
| goto err_buffer_queues; |
| } |
| |
| if (ql_alloc_small_buffers(qdev) != 0) { |
| netdev_err(qdev->ndev, "ql_alloc_small_buffers failed\n"); |
| goto err_small_buffers; |
| } |
| |
| if (ql_alloc_large_buffers(qdev) != 0) { |
| netdev_err(qdev->ndev, "ql_alloc_large_buffers failed\n"); |
| goto err_small_buffers; |
| } |
| |
| /* Initialize the large buffer queue. */ |
| ql_init_large_buffers(qdev); |
| if (ql_create_send_free_list(qdev)) |
| goto err_free_list; |
| |
| qdev->rsp_current = qdev->rsp_q_virt_addr; |
| |
| return 0; |
| err_free_list: |
| ql_free_send_free_list(qdev); |
| err_small_buffers: |
| ql_free_buffer_queues(qdev); |
| err_buffer_queues: |
| ql_free_net_req_rsp_queues(qdev); |
| err_req_rsp: |
| pci_free_consistent(qdev->pdev, |
| PAGE_SIZE, |
| qdev->shadow_reg_virt_addr, |
| qdev->shadow_reg_phy_addr); |
| |
| return -ENOMEM; |
| } |
| |
| static void ql_free_mem_resources(struct ql3_adapter *qdev) |
| { |
| ql_free_send_free_list(qdev); |
| ql_free_large_buffers(qdev); |
| ql_free_small_buffers(qdev); |
| ql_free_buffer_queues(qdev); |
| ql_free_net_req_rsp_queues(qdev); |
| if (qdev->shadow_reg_virt_addr != NULL) { |
| pci_free_consistent(qdev->pdev, |
| PAGE_SIZE, |
| qdev->shadow_reg_virt_addr, |
| qdev->shadow_reg_phy_addr); |
| qdev->shadow_reg_virt_addr = NULL; |
| } |
| } |
| |
| static int ql_init_misc_registers(struct ql3_adapter *qdev) |
| { |
| struct ql3xxx_local_ram_registers __iomem *local_ram = |
| (void __iomem *)qdev->mem_map_registers; |
| |
| if (ql_sem_spinlock(qdev, QL_DDR_RAM_SEM_MASK, |
| (QL_RESOURCE_BITS_BASE_CODE | (qdev->mac_index) * |
| 2) << 4)) |
| return -1; |
| |
| ql_write_page2_reg(qdev, |
| &local_ram->bufletSize, qdev->nvram_data.bufletSize); |
| |
| ql_write_page2_reg(qdev, |
| &local_ram->maxBufletCount, |
| qdev->nvram_data.bufletCount); |
| |
| ql_write_page2_reg(qdev, |
| &local_ram->freeBufletThresholdLow, |
| (qdev->nvram_data.tcpWindowThreshold25 << 16) | |
| (qdev->nvram_data.tcpWindowThreshold0)); |
| |
| ql_write_page2_reg(qdev, |
| &local_ram->freeBufletThresholdHigh, |
| qdev->nvram_data.tcpWindowThreshold50); |
| |
| ql_write_page2_reg(qdev, |
| &local_ram->ipHashTableBase, |
| (qdev->nvram_data.ipHashTableBaseHi << 16) | |
| qdev->nvram_data.ipHashTableBaseLo); |
| ql_write_page2_reg(qdev, |
| &local_ram->ipHashTableCount, |
| qdev->nvram_data.ipHashTableSize); |
| ql_write_page2_reg(qdev, |
| &local_ram->tcpHashTableBase, |
| (qdev->nvram_data.tcpHashTableBaseHi << 16) | |
| qdev->nvram_data.tcpHashTableBaseLo); |
| ql_write_page2_reg(qdev, |
| &local_ram->tcpHashTableCount, |
| qdev->nvram_data.tcpHashTableSize); |
| ql_write_page2_reg(qdev, |
| &local_ram->ncbBase, |
| (qdev->nvram_data.ncbTableBaseHi << 16) | |
| qdev->nvram_data.ncbTableBaseLo); |
| ql_write_page2_reg(qdev, |
| &local_ram->maxNcbCount, |
| qdev->nvram_data.ncbTableSize); |
| ql_write_page2_reg(qdev, |
| &local_ram->drbBase, |
| (qdev->nvram_data.drbTableBaseHi << 16) | |
| qdev->nvram_data.drbTableBaseLo); |
| ql_write_page2_reg(qdev, |
| &local_ram->maxDrbCount, |
| qdev->nvram_data.drbTableSize); |
| ql_sem_unlock(qdev, QL_DDR_RAM_SEM_MASK); |
| return 0; |
| } |
| |
| static int ql_adapter_initialize(struct ql3_adapter *qdev) |
| { |
| u32 value; |
| struct ql3xxx_port_registers __iomem *port_regs = |
| qdev->mem_map_registers; |
| u32 *spir = &port_regs->CommonRegs.serialPortInterfaceReg; |
| struct ql3xxx_host_memory_registers __iomem *hmem_regs = |
| (void __iomem *)port_regs; |
| u32 delay = 10; |
| int status = 0; |
| unsigned long hw_flags = 0; |
| |
| if (ql_mii_setup(qdev)) |
| return -1; |
| |
| /* Bring out PHY out of reset */ |
| ql_write_common_reg(qdev, spir, |
| (ISP_SERIAL_PORT_IF_WE | |
| (ISP_SERIAL_PORT_IF_WE << 16))); |
| /* Give the PHY time to come out of reset. */ |
| mdelay(100); |
| qdev->port_link_state = LS_DOWN; |
| netif_carrier_off(qdev->ndev); |
| |
| /* V2 chip fix for ARS-39168. */ |
| ql_write_common_reg(qdev, spir, |
| (ISP_SERIAL_PORT_IF_SDE | |
| (ISP_SERIAL_PORT_IF_SDE << 16))); |
| |
| /* Request Queue Registers */ |
| *((u32 *)(qdev->preq_consumer_index)) = 0; |
| atomic_set(&qdev->tx_count, NUM_REQ_Q_ENTRIES); |
| qdev->req_producer_index = 0; |
| |
| ql_write_page1_reg(qdev, |
| &hmem_regs->reqConsumerIndexAddrHigh, |
| qdev->req_consumer_index_phy_addr_high); |
| ql_write_page1_reg(qdev, |
| &hmem_regs->reqConsumerIndexAddrLow, |
| qdev->req_consumer_index_phy_addr_low); |
| |
| ql_write_page1_reg(qdev, |
| &hmem_regs->reqBaseAddrHigh, |
| MS_64BITS(qdev->req_q_phy_addr)); |
| ql_write_page1_reg(qdev, |
| &hmem_regs->reqBaseAddrLow, |
| LS_64BITS(qdev->req_q_phy_addr)); |
| ql_write_page1_reg(qdev, &hmem_regs->reqLength, NUM_REQ_Q_ENTRIES); |
| |
| /* Response Queue Registers */ |
| *((__le16 *) (qdev->prsp_producer_index)) = 0; |
| qdev->rsp_consumer_index = 0; |
| qdev->rsp_current = qdev->rsp_q_virt_addr; |
| |
| ql_write_page1_reg(qdev, |
| &hmem_regs->rspProducerIndexAddrHigh, |
| qdev->rsp_producer_index_phy_addr_high); |
| |
| ql_write_page1_reg(qdev, |
| &hmem_regs->rspProducerIndexAddrLow, |
| qdev->rsp_producer_index_phy_addr_low); |
| |
| ql_write_page1_reg(qdev, |
| &hmem_regs->rspBaseAddrHigh, |
| MS_64BITS(qdev->rsp_q_phy_addr)); |
| |
| ql_write_page1_reg(qdev, |
| &hmem_regs->rspBaseAddrLow, |
| LS_64BITS(qdev->rsp_q_phy_addr)); |
| |
| ql_write_page1_reg(qdev, &hmem_regs->rspLength, NUM_RSP_Q_ENTRIES); |
| |
| /* Large Buffer Queue */ |
| ql_write_page1_reg(qdev, |
| &hmem_regs->rxLargeQBaseAddrHigh, |
| MS_64BITS(qdev->lrg_buf_q_phy_addr)); |
| |
| ql_write_page1_reg(qdev, |
| &hmem_regs->rxLargeQBaseAddrLow, |
| LS_64BITS(qdev->lrg_buf_q_phy_addr)); |
| |
| ql_write_page1_reg(qdev, |
| &hmem_regs->rxLargeQLength, |
| qdev->num_lbufq_entries); |
| |
| ql_write_page1_reg(qdev, |
| &hmem_regs->rxLargeBufferLength, |
| qdev->lrg_buffer_len); |
| |
| /* Small Buffer Queue */ |
| ql_write_page1_reg(qdev, |
| &hmem_regs->rxSmallQBaseAddrHigh, |
| MS_64BITS(qdev->small_buf_q_phy_addr)); |
| |
| ql_write_page1_reg(qdev, |
| &hmem_regs->rxSmallQBaseAddrLow, |
| LS_64BITS(qdev->small_buf_q_phy_addr)); |
| |
| ql_write_page1_reg(qdev, &hmem_regs->rxSmallQLength, NUM_SBUFQ_ENTRIES); |
| ql_write_page1_reg(qdev, |
| &hmem_regs->rxSmallBufferLength, |
| QL_SMALL_BUFFER_SIZE); |
|