OpenCloudOS-Kernel/drivers/scsi/qla2xxx/qla_init.c

3898 lines
100 KiB
C
Raw Normal View History

/*
* QLogic Fibre Channel HBA Driver
* Copyright (c) 2003-2005 QLogic Corporation
*
* See LICENSE.qla2xxx for copyright and licensing details.
*/
#include "qla_def.h"
#include <linux/delay.h>
#include <linux/vmalloc.h>
#include "qla_devtbl.h"
/* XXX(hch): this is ugly, but we don't want to pull in exioctl.h */
#ifndef EXT_IS_LUN_BIT_SET
#define EXT_IS_LUN_BIT_SET(P,L) \
(((P)->mask[L/8] & (0x80 >> (L%8)))?1:0)
#define EXT_SET_LUN_BIT(P,L) \
((P)->mask[L/8] |= (0x80 >> (L%8)))
#endif
/*
* QLogic ISP2x00 Hardware Support Function Prototypes.
*/
static int qla2x00_isp_firmware(scsi_qla_host_t *);
static void qla2x00_resize_request_q(scsi_qla_host_t *);
static int qla2x00_setup_chip(scsi_qla_host_t *);
static void qla2x00_init_response_q_entries(scsi_qla_host_t *);
static int qla2x00_init_rings(scsi_qla_host_t *);
static int qla2x00_fw_ready(scsi_qla_host_t *);
static int qla2x00_configure_hba(scsi_qla_host_t *);
static int qla2x00_configure_loop(scsi_qla_host_t *);
static int qla2x00_configure_local_loop(scsi_qla_host_t *);
static int qla2x00_configure_fabric(scsi_qla_host_t *);
static int qla2x00_find_all_fabric_devs(scsi_qla_host_t *, struct list_head *);
static int qla2x00_device_resync(scsi_qla_host_t *);
static int qla2x00_fabric_dev_login(scsi_qla_host_t *, fc_port_t *,
uint16_t *);
static int qla2x00_restart_isp(scsi_qla_host_t *);
static int qla2x00_find_new_loop_id(scsi_qla_host_t *ha, fc_port_t *dev);
/****************************************************************************/
/* QLogic ISP2x00 Hardware Support Functions. */
/****************************************************************************/
/*
* qla2x00_initialize_adapter
* Initialize board.
*
* Input:
* ha = adapter block pointer.
*
* Returns:
* 0 = success
*/
int
qla2x00_initialize_adapter(scsi_qla_host_t *ha)
{
int rval;
/* Clear adapter flags. */
ha->flags.online = 0;
ha->flags.reset_active = 0;
atomic_set(&ha->loop_down_timer, LOOP_DOWN_TIME);
atomic_set(&ha->loop_state, LOOP_DOWN);
ha->device_flags = 0;
ha->dpc_flags = 0;
ha->flags.management_server_logged_in = 0;
ha->marker_needed = 0;
ha->mbx_flags = 0;
ha->isp_abort_cnt = 0;
ha->beacon_blink_led = 0;
set_bit(REGISTER_FDMI_NEEDED, &ha->dpc_flags);
qla_printk(KERN_INFO, ha, "Configuring PCI space...\n");
rval = ha->isp_ops.pci_config(ha);
if (rval) {
DEBUG2(printk("scsi(%ld): Unable to configure PCI space=n",
ha->host_no));
return (rval);
}
ha->isp_ops.reset_chip(ha);
qla_printk(KERN_INFO, ha, "Configure NVRAM parameters...\n");
ha->isp_ops.nvram_config(ha);
if (ha->flags.disable_serdes) {
/* Mask HBA via NVRAM settings? */
qla_printk(KERN_INFO, ha, "Masking HBA WWPN "
"%02x%02x%02x%02x%02x%02x%02x%02x (via NVRAM).\n",
ha->port_name[0], ha->port_name[1],
ha->port_name[2], ha->port_name[3],
ha->port_name[4], ha->port_name[5],
ha->port_name[6], ha->port_name[7]);
return QLA_FUNCTION_FAILED;
}
qla_printk(KERN_INFO, ha, "Verifying loaded RISC code...\n");
if (qla2x00_isp_firmware(ha) != QLA_SUCCESS) {
rval = ha->isp_ops.chip_diag(ha);
if (rval)
return (rval);
rval = qla2x00_setup_chip(ha);
if (rval)
return (rval);
}
rval = qla2x00_init_rings(ha);
return (rval);
}
/**
* qla2100_pci_config() - Setup ISP21xx PCI configuration registers.
* @ha: HA context
*
* Returns 0 on success.
*/
int
qla2100_pci_config(scsi_qla_host_t *ha)
{
uint16_t w, mwi;
uint32_t d;
unsigned long flags;
struct device_reg_2xxx __iomem *reg = &ha->iobase->isp;
pci_set_master(ha->pdev);
mwi = 0;
if (pci_set_mwi(ha->pdev))
mwi = PCI_COMMAND_INVALIDATE;
pci_read_config_word(ha->pdev, PCI_COMMAND, &w);
w |= mwi | (PCI_COMMAND_PARITY | PCI_COMMAND_SERR);
pci_write_config_word(ha->pdev, PCI_COMMAND, w);
/* Reset expansion ROM address decode enable */
pci_read_config_dword(ha->pdev, PCI_ROM_ADDRESS, &d);
d &= ~PCI_ROM_ADDRESS_ENABLE;
pci_write_config_dword(ha->pdev, PCI_ROM_ADDRESS, d);
/* Get PCI bus information. */
spin_lock_irqsave(&ha->hardware_lock, flags);
ha->pci_attr = RD_REG_WORD(&reg->ctrl_status);
spin_unlock_irqrestore(&ha->hardware_lock, flags);
return QLA_SUCCESS;
}
/**
* qla2300_pci_config() - Setup ISP23xx PCI configuration registers.
* @ha: HA context
*
* Returns 0 on success.
*/
int
qla2300_pci_config(scsi_qla_host_t *ha)
{
uint16_t w, mwi;
uint32_t d;
unsigned long flags = 0;
uint32_t cnt;
struct device_reg_2xxx __iomem *reg = &ha->iobase->isp;
pci_set_master(ha->pdev);
mwi = 0;
if (pci_set_mwi(ha->pdev))
mwi = PCI_COMMAND_INVALIDATE;
pci_read_config_word(ha->pdev, PCI_COMMAND, &w);
w |= mwi | (PCI_COMMAND_PARITY | PCI_COMMAND_SERR);
if (IS_QLA2322(ha) || IS_QLA6322(ha))
w &= ~PCI_COMMAND_INTX_DISABLE;
/*
* If this is a 2300 card and not 2312, reset the
* COMMAND_INVALIDATE due to a bug in the 2300. Unfortunately,
* the 2310 also reports itself as a 2300 so we need to get the
* fb revision level -- a 6 indicates it really is a 2300 and
* not a 2310.
*/
if (IS_QLA2300(ha)) {
spin_lock_irqsave(&ha->hardware_lock, flags);
/* Pause RISC. */
WRT_REG_WORD(&reg->hccr, HCCR_PAUSE_RISC);
for (cnt = 0; cnt < 30000; cnt++) {
if ((RD_REG_WORD(&reg->hccr) & HCCR_RISC_PAUSE) != 0)
break;
udelay(10);
}
/* Select FPM registers. */
WRT_REG_WORD(&reg->ctrl_status, 0x20);
RD_REG_WORD(&reg->ctrl_status);
/* Get the fb rev level */
ha->fb_rev = RD_FB_CMD_REG(ha, reg);
if (ha->fb_rev == FPM_2300)
w &= ~PCI_COMMAND_INVALIDATE;
/* Deselect FPM registers. */
WRT_REG_WORD(&reg->ctrl_status, 0x0);
RD_REG_WORD(&reg->ctrl_status);
/* Release RISC module. */
WRT_REG_WORD(&reg->hccr, HCCR_RELEASE_RISC);
for (cnt = 0; cnt < 30000; cnt++) {
if ((RD_REG_WORD(&reg->hccr) & HCCR_RISC_PAUSE) == 0)
break;
udelay(10);
}
spin_unlock_irqrestore(&ha->hardware_lock, flags);
}
pci_write_config_word(ha->pdev, PCI_COMMAND, w);
pci_write_config_byte(ha->pdev, PCI_LATENCY_TIMER, 0x80);
/* Reset expansion ROM address decode enable */
pci_read_config_dword(ha->pdev, PCI_ROM_ADDRESS, &d);
d &= ~PCI_ROM_ADDRESS_ENABLE;
pci_write_config_dword(ha->pdev, PCI_ROM_ADDRESS, d);
/* Get PCI bus information. */
spin_lock_irqsave(&ha->hardware_lock, flags);
ha->pci_attr = RD_REG_WORD(&reg->ctrl_status);
spin_unlock_irqrestore(&ha->hardware_lock, flags);
return QLA_SUCCESS;
}
/**
* qla24xx_pci_config() - Setup ISP24xx PCI configuration registers.
* @ha: HA context
*
* Returns 0 on success.
*/
int
qla24xx_pci_config(scsi_qla_host_t *ha)
{
uint16_t w, mwi;
uint32_t d;
unsigned long flags = 0;
struct device_reg_24xx __iomem *reg = &ha->iobase->isp24;
int pcix_cmd_reg, pcie_dctl_reg;
pci_set_master(ha->pdev);
mwi = 0;
if (pci_set_mwi(ha->pdev))
mwi = PCI_COMMAND_INVALIDATE;
pci_read_config_word(ha->pdev, PCI_COMMAND, &w);
w |= mwi | (PCI_COMMAND_PARITY | PCI_COMMAND_SERR);
w &= ~PCI_COMMAND_INTX_DISABLE;
pci_write_config_word(ha->pdev, PCI_COMMAND, w);
pci_write_config_byte(ha->pdev, PCI_LATENCY_TIMER, 0x80);
/* PCI-X -- adjust Maximum Memory Read Byte Count (2048). */
pcix_cmd_reg = pci_find_capability(ha->pdev, PCI_CAP_ID_PCIX);
if (pcix_cmd_reg) {
uint16_t pcix_cmd;
pcix_cmd_reg += PCI_X_CMD;
pci_read_config_word(ha->pdev, pcix_cmd_reg, &pcix_cmd);
pcix_cmd &= ~PCI_X_CMD_MAX_READ;
pcix_cmd |= 0x0008;
pci_write_config_word(ha->pdev, pcix_cmd_reg, pcix_cmd);
}
/* PCIe -- adjust Maximum Read Request Size (2048). */
pcie_dctl_reg = pci_find_capability(ha->pdev, PCI_CAP_ID_EXP);
if (pcie_dctl_reg) {
uint16_t pcie_dctl;
pcie_dctl_reg += PCI_EXP_DEVCTL;
pci_read_config_word(ha->pdev, pcie_dctl_reg, &pcie_dctl);
pcie_dctl &= ~PCI_EXP_DEVCTL_READRQ;
pcie_dctl |= 0x4000;
pci_write_config_word(ha->pdev, pcie_dctl_reg, pcie_dctl);
}
/* Reset expansion ROM address decode enable */
pci_read_config_dword(ha->pdev, PCI_ROM_ADDRESS, &d);
d &= ~PCI_ROM_ADDRESS_ENABLE;
pci_write_config_dword(ha->pdev, PCI_ROM_ADDRESS, d);
pci_read_config_word(ha->pdev, PCI_REVISION_ID, &ha->chip_revision);
/* Get PCI bus information. */
spin_lock_irqsave(&ha->hardware_lock, flags);
ha->pci_attr = RD_REG_DWORD(&reg->ctrl_status);
spin_unlock_irqrestore(&ha->hardware_lock, flags);
return QLA_SUCCESS;
}
/**
* qla2x00_isp_firmware() - Choose firmware image.
* @ha: HA context
*
* Returns 0 on success.
*/
static int
qla2x00_isp_firmware(scsi_qla_host_t *ha)
{
int rval;
/* Assume loading risc code */
rval = QLA_FUNCTION_FAILED;
if (ha->flags.disable_risc_code_load) {
DEBUG2(printk("scsi(%ld): RISC CODE NOT loaded\n",
ha->host_no));
qla_printk(KERN_INFO, ha, "RISC CODE NOT loaded\n");
/* Verify checksum of loaded RISC code. */
rval = qla2x00_verify_checksum(ha, ha->fw_srisc_address);
}
if (rval) {
DEBUG2_3(printk("scsi(%ld): **** Load RISC code ****\n",
ha->host_no));
}
return (rval);
}
/**
* qla2x00_reset_chip() - Reset ISP chip.
* @ha: HA context
*
* Returns 0 on success.
*/
void
qla2x00_reset_chip(scsi_qla_host_t *ha)
{
unsigned long flags = 0;
struct device_reg_2xxx __iomem *reg = &ha->iobase->isp;
uint32_t cnt;
uint16_t cmd;
ha->isp_ops.disable_intrs(ha);
spin_lock_irqsave(&ha->hardware_lock, flags);
/* Turn off master enable */
cmd = 0;
pci_read_config_word(ha->pdev, PCI_COMMAND, &cmd);
cmd &= ~PCI_COMMAND_MASTER;
pci_write_config_word(ha->pdev, PCI_COMMAND, cmd);
if (!IS_QLA2100(ha)) {
/* Pause RISC. */
WRT_REG_WORD(&reg->hccr, HCCR_PAUSE_RISC);
if (IS_QLA2200(ha) || IS_QLA2300(ha)) {
for (cnt = 0; cnt < 30000; cnt++) {
if ((RD_REG_WORD(&reg->hccr) &
HCCR_RISC_PAUSE) != 0)
break;
udelay(100);
}
} else {
RD_REG_WORD(&reg->hccr); /* PCI Posting. */
udelay(10);
}
/* Select FPM registers. */
WRT_REG_WORD(&reg->ctrl_status, 0x20);
RD_REG_WORD(&reg->ctrl_status); /* PCI Posting. */
/* FPM Soft Reset. */
WRT_REG_WORD(&reg->fpm_diag_config, 0x100);
RD_REG_WORD(&reg->fpm_diag_config); /* PCI Posting. */
/* Toggle Fpm Reset. */
if (!IS_QLA2200(ha)) {
WRT_REG_WORD(&reg->fpm_diag_config, 0x0);
RD_REG_WORD(&reg->fpm_diag_config); /* PCI Posting. */
}
/* Select frame buffer registers. */
WRT_REG_WORD(&reg->ctrl_status, 0x10);
RD_REG_WORD(&reg->ctrl_status); /* PCI Posting. */
/* Reset frame buffer FIFOs. */
if (IS_QLA2200(ha)) {
WRT_FB_CMD_REG(ha, reg, 0xa000);
RD_FB_CMD_REG(ha, reg); /* PCI Posting. */
} else {
WRT_FB_CMD_REG(ha, reg, 0x00fc);
/* Read back fb_cmd until zero or 3 seconds max */
for (cnt = 0; cnt < 3000; cnt++) {
if ((RD_FB_CMD_REG(ha, reg) & 0xff) == 0)
break;
udelay(100);
}
}
/* Select RISC module registers. */
WRT_REG_WORD(&reg->ctrl_status, 0);
RD_REG_WORD(&reg->ctrl_status); /* PCI Posting. */
/* Reset RISC processor. */
WRT_REG_WORD(&reg->hccr, HCCR_RESET_RISC);
RD_REG_WORD(&reg->hccr); /* PCI Posting. */
/* Release RISC processor. */
WRT_REG_WORD(&reg->hccr, HCCR_RELEASE_RISC);
RD_REG_WORD(&reg->hccr); /* PCI Posting. */
}
WRT_REG_WORD(&reg->hccr, HCCR_CLR_RISC_INT);
WRT_REG_WORD(&reg->hccr, HCCR_CLR_HOST_INT);
/* Reset ISP chip. */
WRT_REG_WORD(&reg->ctrl_status, CSR_ISP_SOFT_RESET);
/* Wait for RISC to recover from reset. */
if (IS_QLA2100(ha) || IS_QLA2200(ha) || IS_QLA2300(ha)) {
/*
* It is necessary to for a delay here since the card doesn't
* respond to PCI reads during a reset. On some architectures
* this will result in an MCA.
*/
udelay(20);
for (cnt = 30000; cnt; cnt--) {
if ((RD_REG_WORD(&reg->ctrl_status) &
CSR_ISP_SOFT_RESET) == 0)
break;
udelay(100);
}
} else
udelay(10);
/* Reset RISC processor. */
WRT_REG_WORD(&reg->hccr, HCCR_RESET_RISC);
WRT_REG_WORD(&reg->semaphore, 0);
/* Release RISC processor. */
WRT_REG_WORD(&reg->hccr, HCCR_RELEASE_RISC);
RD_REG_WORD(&reg->hccr); /* PCI Posting. */
if (IS_QLA2100(ha) || IS_QLA2200(ha) || IS_QLA2300(ha)) {
for (cnt = 0; cnt < 30000; cnt++) {
if (RD_MAILBOX_REG(ha, reg, 0) != MBS_BUSY)
break;
udelay(100);
}
} else
udelay(100);
/* Turn on master enable */
cmd |= PCI_COMMAND_MASTER;
pci_write_config_word(ha->pdev, PCI_COMMAND, cmd);
/* Disable RISC pause on FPM parity error. */
if (!IS_QLA2100(ha)) {
WRT_REG_WORD(&reg->hccr, HCCR_DISABLE_PARITY_PAUSE);
RD_REG_WORD(&reg->hccr); /* PCI Posting. */
}
spin_unlock_irqrestore(&ha->hardware_lock, flags);
}
/**
* qla24xx_reset_risc() - Perform full reset of ISP24xx RISC.
* @ha: HA context
*
* Returns 0 on success.
*/
static inline void
qla24xx_reset_risc(scsi_qla_host_t *ha)
{
unsigned long flags = 0;
struct device_reg_24xx __iomem *reg = &ha->iobase->isp24;
uint32_t cnt, d2;
uint16_t wd;
spin_lock_irqsave(&ha->hardware_lock, flags);
/* Reset RISC. */
WRT_REG_DWORD(&reg->ctrl_status, CSRX_DMA_SHUTDOWN|MWB_4096_BYTES);
for (cnt = 0; cnt < 30000; cnt++) {
if ((RD_REG_DWORD(&reg->ctrl_status) & CSRX_DMA_ACTIVE) == 0)
break;
udelay(10);
}
WRT_REG_DWORD(&reg->ctrl_status,
CSRX_ISP_SOFT_RESET|CSRX_DMA_SHUTDOWN|MWB_4096_BYTES);
pci_read_config_word(ha->pdev, PCI_COMMAND, &wd);
udelay(100);
/* Wait for firmware to complete NVRAM accesses. */
d2 = (uint32_t) RD_REG_WORD(&reg->mailbox0);
for (cnt = 10000 ; cnt && d2; cnt--) {
udelay(5);
d2 = (uint32_t) RD_REG_WORD(&reg->mailbox0);
barrier();
}
/* Wait for soft-reset to complete. */
d2 = RD_REG_DWORD(&reg->ctrl_status);
for (cnt = 6000000 ; cnt && (d2 & CSRX_ISP_SOFT_RESET); cnt--) {
udelay(5);
d2 = RD_REG_DWORD(&reg->ctrl_status);
barrier();
}
WRT_REG_DWORD(&reg->hccr, HCCRX_SET_RISC_RESET);
RD_REG_DWORD(&reg->hccr);
WRT_REG_DWORD(&reg->hccr, HCCRX_REL_RISC_PAUSE);
RD_REG_DWORD(&reg->hccr);
WRT_REG_DWORD(&reg->hccr, HCCRX_CLR_RISC_RESET);
RD_REG_DWORD(&reg->hccr);
d2 = (uint32_t) RD_REG_WORD(&reg->mailbox0);
for (cnt = 6000000 ; cnt && d2; cnt--) {
udelay(5);
d2 = (uint32_t) RD_REG_WORD(&reg->mailbox0);
barrier();
}
spin_unlock_irqrestore(&ha->hardware_lock, flags);
}
/**
* qla24xx_reset_chip() - Reset ISP24xx chip.
* @ha: HA context
*
* Returns 0 on success.
*/
void
qla24xx_reset_chip(scsi_qla_host_t *ha)
{
ha->isp_ops.disable_intrs(ha);
/* Perform RISC reset. */
qla24xx_reset_risc(ha);
}
/**
* qla2x00_chip_diag() - Test chip for proper operation.
* @ha: HA context
*
* Returns 0 on success.
*/
int
qla2x00_chip_diag(scsi_qla_host_t *ha)
{
int rval;
struct device_reg_2xxx __iomem *reg = &ha->iobase->isp;
unsigned long flags = 0;
uint16_t data;
uint32_t cnt;
uint16_t mb[5];
/* Assume a failed state */
rval = QLA_FUNCTION_FAILED;
DEBUG3(printk("scsi(%ld): Testing device at %lx.\n",
ha->host_no, (u_long)&reg->flash_address));
spin_lock_irqsave(&ha->hardware_lock, flags);
/* Reset ISP chip. */
WRT_REG_WORD(&reg->ctrl_status, CSR_ISP_SOFT_RESET);
/*
* We need to have a delay here since the card will not respond while
* in reset causing an MCA on some architectures.
*/
udelay(20);
data = qla2x00_debounce_register(&reg->ctrl_status);
for (cnt = 6000000 ; cnt && (data & CSR_ISP_SOFT_RESET); cnt--) {
udelay(5);
data = RD_REG_WORD(&reg->ctrl_status);
barrier();
}
if (!cnt)
goto chip_diag_failed;
DEBUG3(printk("scsi(%ld): Reset register cleared by chip reset\n",
ha->host_no));
/* Reset RISC processor. */
WRT_REG_WORD(&reg->hccr, HCCR_RESET_RISC);
WRT_REG_WORD(&reg->hccr, HCCR_RELEASE_RISC);
/* Workaround for QLA2312 PCI parity error */
if (IS_QLA2100(ha) || IS_QLA2200(ha) || IS_QLA2300(ha)) {
data = qla2x00_debounce_register(MAILBOX_REG(ha, reg, 0));
for (cnt = 6000000; cnt && (data == MBS_BUSY); cnt--) {
udelay(5);
data = RD_MAILBOX_REG(ha, reg, 0);
barrier();
}
} else
udelay(10);
if (!cnt)
goto chip_diag_failed;
/* Check product ID of chip */
DEBUG3(printk("scsi(%ld): Checking product ID of chip\n", ha->host_no));
mb[1] = RD_MAILBOX_REG(ha, reg, 1);
mb[2] = RD_MAILBOX_REG(ha, reg, 2);
mb[3] = RD_MAILBOX_REG(ha, reg, 3);
mb[4] = qla2x00_debounce_register(MAILBOX_REG(ha, reg, 4));
if (mb[1] != PROD_ID_1 || (mb[2] != PROD_ID_2 && mb[2] != PROD_ID_2a) ||
mb[3] != PROD_ID_3) {
qla_printk(KERN_WARNING, ha,
"Wrong product ID = 0x%x,0x%x,0x%x\n", mb[1], mb[2], mb[3]);
goto chip_diag_failed;
}
ha->product_id[0] = mb[1];
ha->product_id[1] = mb[2];
ha->product_id[2] = mb[3];
ha->product_id[3] = mb[4];
/* Adjust fw RISC transfer size */
if (ha->request_q_length > 1024)
ha->fw_transfer_size = REQUEST_ENTRY_SIZE * 1024;
else
ha->fw_transfer_size = REQUEST_ENTRY_SIZE *
ha->request_q_length;
if (IS_QLA2200(ha) &&
RD_MAILBOX_REG(ha, reg, 7) == QLA2200A_RISC_ROM_VER) {
/* Limit firmware transfer size with a 2200A */
DEBUG3(printk("scsi(%ld): Found QLA2200A chip.\n",
ha->host_no));
ha->device_type |= DT_ISP2200A;
ha->fw_transfer_size = 128;
}
/* Wrap Incoming Mailboxes Test. */
spin_unlock_irqrestore(&ha->hardware_lock, flags);
DEBUG3(printk("scsi(%ld): Checking mailboxes.\n", ha->host_no));
rval = qla2x00_mbx_reg_test(ha);
if (rval) {
DEBUG(printk("scsi(%ld): Failed mailbox send register test\n",
ha->host_no));
qla_printk(KERN_WARNING, ha,
"Failed mailbox send register test\n");
}
else {
/* Flag a successful rval */
rval = QLA_SUCCESS;
}
spin_lock_irqsave(&ha->hardware_lock, flags);
chip_diag_failed:
if (rval)
DEBUG2_3(printk("scsi(%ld): Chip diagnostics **** FAILED "
"****\n", ha->host_no));
spin_unlock_irqrestore(&ha->hardware_lock, flags);
return (rval);
}
/**
* qla24xx_chip_diag() - Test ISP24xx for proper operation.
* @ha: HA context
*
* Returns 0 on success.
*/
int
qla24xx_chip_diag(scsi_qla_host_t *ha)
{
int rval;
/* Perform RISC reset. */
qla24xx_reset_risc(ha);
ha->fw_transfer_size = REQUEST_ENTRY_SIZE * 1024;
rval = qla2x00_mbx_reg_test(ha);
if (rval) {
DEBUG(printk("scsi(%ld): Failed mailbox send register test\n",
ha->host_no));
qla_printk(KERN_WARNING, ha,
"Failed mailbox send register test\n");
} else {
/* Flag a successful rval */
rval = QLA_SUCCESS;
}
return rval;
}
void
qla2x00_alloc_fw_dump(scsi_qla_host_t *ha)
{
int rval;
uint32_t dump_size, fixed_size, mem_size, req_q_size, rsp_q_size,
eft_size;
dma_addr_t eft_dma;
void *eft;
if (ha->fw_dump) {
qla_printk(KERN_WARNING, ha,
"Firmware dump previously allocated.\n");
return;
}
ha->fw_dumped = 0;
fixed_size = mem_size = eft_size = 0;
if (IS_QLA2100(ha) || IS_QLA2200(ha)) {
fixed_size = sizeof(struct qla2100_fw_dump);
} else if (IS_QLA23XX(ha)) {
fixed_size = offsetof(struct qla2300_fw_dump, data_ram);
mem_size = (ha->fw_memory_size - 0x11000 + 1) *
sizeof(uint16_t);
} else if (IS_QLA24XX(ha) || IS_QLA54XX(ha)) {
fixed_size = offsetof(struct qla24xx_fw_dump, ext_mem);
mem_size = (ha->fw_memory_size - 0x100000 + 1) *
sizeof(uint32_t);
/* Allocate memory for Extended Trace Buffer. */
eft = dma_alloc_coherent(&ha->pdev->dev, EFT_SIZE, &eft_dma,
GFP_KERNEL);
if (!eft) {
qla_printk(KERN_WARNING, ha, "Unable to allocate "
"(%d KB) for EFT.\n", EFT_SIZE / 1024);
goto cont_alloc;
}
rval = qla2x00_trace_control(ha, TC_ENABLE, eft_dma,
EFT_NUM_BUFFERS);
if (rval) {
qla_printk(KERN_WARNING, ha, "Unable to initialize "
"EFT (%d).\n", rval);
dma_free_coherent(&ha->pdev->dev, EFT_SIZE, eft,
eft_dma);
goto cont_alloc;
}
qla_printk(KERN_INFO, ha, "Allocated (%d KB) for EFT...\n",
EFT_SIZE / 1024);
eft_size = EFT_SIZE;
memset(eft, 0, eft_size);
ha->eft_dma = eft_dma;
ha->eft = eft;
}
cont_alloc:
req_q_size = ha->request_q_length * sizeof(request_t);
rsp_q_size = ha->response_q_length * sizeof(response_t);
dump_size = offsetof(struct qla2xxx_fw_dump, isp);
dump_size += fixed_size + mem_size + req_q_size + rsp_q_size +
eft_size;
ha->fw_dump = vmalloc(dump_size);
if (!ha->fw_dump) {
qla_printk(KERN_WARNING, ha, "Unable to allocate (%d KB) for "
"firmware dump!!!\n", dump_size / 1024);
if (ha->eft) {
dma_free_coherent(&ha->pdev->dev, eft_size, ha->eft,
ha->eft_dma);
ha->eft = NULL;
ha->eft_dma = 0;
}
return;
}
qla_printk(KERN_INFO, ha, "Allocated (%d KB) for firmware dump...\n",
dump_size / 1024);
ha->fw_dump_len = dump_size;
ha->fw_dump->signature[0] = 'Q';
ha->fw_dump->signature[1] = 'L';
ha->fw_dump->signature[2] = 'G';
ha->fw_dump->signature[3] = 'C';
ha->fw_dump->version = __constant_htonl(1);
ha->fw_dump->fixed_size = htonl(fixed_size);
ha->fw_dump->mem_size = htonl(mem_size);
ha->fw_dump->req_q_size = htonl(req_q_size);
ha->fw_dump->rsp_q_size = htonl(rsp_q_size);
ha->fw_dump->eft_size = htonl(eft_size);
ha->fw_dump->eft_addr_l = htonl(LSD(ha->eft_dma));
ha->fw_dump->eft_addr_h = htonl(MSD(ha->eft_dma));
ha->fw_dump->header_size =
htonl(offsetof(struct qla2xxx_fw_dump, isp));
}
/**
* qla2x00_resize_request_q() - Resize request queue given available ISP memory.
* @ha: HA context
*
* Returns 0 on success.
*/
static void
qla2x00_resize_request_q(scsi_qla_host_t *ha)
{
int rval;
uint16_t fw_iocb_cnt = 0;
uint16_t request_q_length = REQUEST_ENTRY_CNT_2XXX_EXT_MEM;
dma_addr_t request_dma;
request_t *request_ring;
/* Valid only on recent ISPs. */
if (IS_QLA2100(ha) || IS_QLA2200(ha))
return;
/* Retrieve IOCB counts available to the firmware. */
rval = qla2x00_get_resource_cnts(ha, NULL, NULL, NULL, &fw_iocb_cnt);
if (rval)
return;
/* No point in continuing if current settings are sufficient. */
if (fw_iocb_cnt < 1024)
return;
if (ha->request_q_length >= request_q_length)
return;
/* Attempt to claim larger area for request queue. */
request_ring = dma_alloc_coherent(&ha->pdev->dev,
(request_q_length + 1) * sizeof(request_t), &request_dma,
GFP_KERNEL);
if (request_ring == NULL)
return;
/* Resize successful, report extensions. */
qla_printk(KERN_INFO, ha, "Extended memory detected (%d KB)...\n",
(ha->fw_memory_size + 1) / 1024);
qla_printk(KERN_INFO, ha, "Resizing request queue depth "
"(%d -> %d)...\n", ha->request_q_length, request_q_length);
/* Clear old allocations. */
dma_free_coherent(&ha->pdev->dev,
(ha->request_q_length + 1) * sizeof(request_t), ha->request_ring,
ha->request_dma);
/* Begin using larger queue. */
ha->request_q_length = request_q_length;
ha->request_ring = request_ring;
ha->request_dma = request_dma;
}
/**
* qla2x00_setup_chip() - Load and start RISC firmware.
* @ha: HA context
*
* Returns 0 on success.
*/
static int
qla2x00_setup_chip(scsi_qla_host_t *ha)
{
int rval;
uint32_t srisc_address = 0;
/* Load firmware sequences */
rval = ha->isp_ops.load_risc(ha, &srisc_address);
if (rval == QLA_SUCCESS) {
DEBUG(printk("scsi(%ld): Verifying Checksum of loaded RISC "
"code.\n", ha->host_no));
rval = qla2x00_verify_checksum(ha, srisc_address);
if (rval == QLA_SUCCESS) {
/* Start firmware execution. */
DEBUG(printk("scsi(%ld): Checksum OK, start "
"firmware.\n", ha->host_no));
rval = qla2x00_execute_fw(ha, srisc_address);
/* Retrieve firmware information. */
if (rval == QLA_SUCCESS && ha->fw_major_version == 0) {
qla2x00_get_fw_version(ha,
&ha->fw_major_version,
&ha->fw_minor_version,
&ha->fw_subminor_version,
&ha->fw_attributes, &ha->fw_memory_size);
qla2x00_resize_request_q(ha);
if (ql2xallocfwdump)
qla2x00_alloc_fw_dump(ha);
}
} else {
DEBUG2(printk(KERN_INFO
"scsi(%ld): ISP Firmware failed checksum.\n",
ha->host_no));
}
}
if (rval) {
DEBUG2_3(printk("scsi(%ld): Setup chip **** FAILED ****.\n",
ha->host_no));
}
return (rval);
}
/**
* qla2x00_init_response_q_entries() - Initializes response queue entries.
* @ha: HA context
*
* Beginning of request ring has initialization control block already built
* by nvram config routine.
*
* Returns 0 on success.
*/
static void
qla2x00_init_response_q_entries(scsi_qla_host_t *ha)
{
uint16_t cnt;
response_t *pkt;
pkt = ha->response_ring_ptr;
for (cnt = 0; cnt < ha->response_q_length; cnt++) {
pkt->signature = RESPONSE_PROCESSED;
pkt++;
}
}
/**
* qla2x00_update_fw_options() - Read and process firmware options.
* @ha: HA context
*
* Returns 0 on success.
*/
void
qla2x00_update_fw_options(scsi_qla_host_t *ha)
{
uint16_t swing, emphasis, tx_sens, rx_sens;
memset(ha->fw_options, 0, sizeof(ha->fw_options));
qla2x00_get_fw_options(ha, ha->fw_options);
if (IS_QLA2100(ha) || IS_QLA2200(ha))
return;
/* Serial Link options. */
DEBUG3(printk("scsi(%ld): Serial link options:\n",
ha->host_no));
DEBUG3(qla2x00_dump_buffer((uint8_t *)&ha->fw_seriallink_options,
sizeof(ha->fw_seriallink_options)));
ha->fw_options[1] &= ~FO1_SET_EMPHASIS_SWING;
if (ha->fw_seriallink_options[3] & BIT_2) {
ha->fw_options[1] |= FO1_SET_EMPHASIS_SWING;
/* 1G settings */
swing = ha->fw_seriallink_options[2] & (BIT_2 | BIT_1 | BIT_0);
emphasis = (ha->fw_seriallink_options[2] &
(BIT_4 | BIT_3)) >> 3;
tx_sens = ha->fw_seriallink_options[0] &
(BIT_3 | BIT_2 | BIT_1 | BIT_0);
rx_sens = (ha->fw_seriallink_options[0] &
(BIT_7 | BIT_6 | BIT_5 | BIT_4)) >> 4;
ha->fw_options[10] = (emphasis << 14) | (swing << 8);
if (IS_QLA2300(ha) || IS_QLA2312(ha) || IS_QLA6312(ha)) {
if (rx_sens == 0x0)
rx_sens = 0x3;
ha->fw_options[10] |= (tx_sens << 4) | rx_sens;
} else if (IS_QLA2322(ha) || IS_QLA6322(ha))
ha->fw_options[10] |= BIT_5 |
((rx_sens & (BIT_1 | BIT_0)) << 2) |
(tx_sens & (BIT_1 | BIT_0));
/* 2G settings */
swing = (ha->fw_seriallink_options[2] &
(BIT_7 | BIT_6 | BIT_5)) >> 5;
emphasis = ha->fw_seriallink_options[3] & (BIT_1 | BIT_0);
tx_sens = ha->fw_seriallink_options[1] &
(BIT_3 | BIT_2 | BIT_1 | BIT_0);
rx_sens = (ha->fw_seriallink_options[1] &
(BIT_7 | BIT_6 | BIT_5 | BIT_4)) >> 4;
ha->fw_options[11] = (emphasis << 14) | (swing << 8);
if (IS_QLA2300(ha) || IS_QLA2312(ha) || IS_QLA6312(ha)) {
if (rx_sens == 0x0)
rx_sens = 0x3;
ha->fw_options[11] |= (tx_sens << 4) | rx_sens;
} else if (IS_QLA2322(ha) || IS_QLA6322(ha))
ha->fw_options[11] |= BIT_5 |
((rx_sens & (BIT_1 | BIT_0)) << 2) |
(tx_sens & (BIT_1 | BIT_0));
}
/* FCP2 options. */
/* Return command IOCBs without waiting for an ABTS to complete. */
ha->fw_options[3] |= BIT_13;
/* LED scheme. */
if (ha->flags.enable_led_scheme)
ha->fw_options[2] |= BIT_12;
/* Detect ISP6312. */
if (IS_QLA6312(ha))
ha->fw_options[2] |= BIT_13;
/* Update firmware options. */
qla2x00_set_fw_options(ha, ha->fw_options);
}
void
qla24xx_update_fw_options(scsi_qla_host_t *ha)
{
int rval;
/* Update Serial Link options. */
if ((le16_to_cpu(ha->fw_seriallink_options24[0]) & BIT_0) == 0)
return;
rval = qla2x00_set_serdes_params(ha,
le16_to_cpu(ha->fw_seriallink_options24[1]),
le16_to_cpu(ha->fw_seriallink_options24[2]),
le16_to_cpu(ha->fw_seriallink_options24[3]));
if (rval != QLA_SUCCESS) {
qla_printk(KERN_WARNING, ha,
"Unable to update Serial Link options (%x).\n", rval);
}
}
void
qla2x00_config_rings(struct scsi_qla_host *ha)
{
struct device_reg_2xxx __iomem *reg = &ha->iobase->isp;
/* Setup ring parameters in initialization control block. */
ha->init_cb->request_q_outpointer = __constant_cpu_to_le16(0);
ha->init_cb->response_q_inpointer = __constant_cpu_to_le16(0);
ha->init_cb->request_q_length = cpu_to_le16(ha->request_q_length);
ha->init_cb->response_q_length = cpu_to_le16(ha->response_q_length);
ha->init_cb->request_q_address[0] = cpu_to_le32(LSD(ha->request_dma));
ha->init_cb->request_q_address[1] = cpu_to_le32(MSD(ha->request_dma));
ha->init_cb->response_q_address[0] = cpu_to_le32(LSD(ha->response_dma));
ha->init_cb->response_q_address[1] = cpu_to_le32(MSD(ha->response_dma));
WRT_REG_WORD(ISP_REQ_Q_IN(ha, reg), 0);
WRT_REG_WORD(ISP_REQ_Q_OUT(ha, reg), 0);
WRT_REG_WORD(ISP_RSP_Q_IN(ha, reg), 0);
WRT_REG_WORD(ISP_RSP_Q_OUT(ha, reg), 0);
RD_REG_WORD(ISP_RSP_Q_OUT(ha, reg)); /* PCI Posting. */
}
void
qla24xx_config_rings(struct scsi_qla_host *ha)
{
struct device_reg_24xx __iomem *reg = &ha->iobase->isp24;
struct init_cb_24xx *icb;
/* Setup ring parameters in initialization control block. */
icb = (struct init_cb_24xx *)ha->init_cb;
icb->request_q_outpointer = __constant_cpu_to_le16(0);
icb->response_q_inpointer = __constant_cpu_to_le16(0);
icb->request_q_length = cpu_to_le16(ha->request_q_length);
icb->response_q_length = cpu_to_le16(ha->response_q_length);
icb->request_q_address[0] = cpu_to_le32(LSD(ha->request_dma));
icb->request_q_address[1] = cpu_to_le32(MSD(ha->request_dma));
icb->response_q_address[0] = cpu_to_le32(LSD(ha->response_dma));
icb->response_q_address[1] = cpu_to_le32(MSD(ha->response_dma));
WRT_REG_DWORD(&reg->req_q_in, 0);
WRT_REG_DWORD(&reg->req_q_out, 0);
WRT_REG_DWORD(&reg->rsp_q_in, 0);
WRT_REG_DWORD(&reg->rsp_q_out, 0);
RD_REG_DWORD(&reg->rsp_q_out);
}
/**
* qla2x00_init_rings() - Initializes firmware.
* @ha: HA context
*
* Beginning of request ring has initialization control block already built
* by nvram config routine.
*
* Returns 0 on success.
*/
static int
qla2x00_init_rings(scsi_qla_host_t *ha)
{
int rval;
unsigned long flags = 0;
int cnt;
spin_lock_irqsave(&ha->hardware_lock, flags);
/* Clear outstanding commands array. */
for (cnt = 0; cnt < MAX_OUTSTANDING_COMMANDS; cnt++)
ha->outstanding_cmds[cnt] = NULL;
ha->current_outstanding_cmd = 0;
/* Clear RSCN queue. */
ha->rscn_in_ptr = 0;
ha->rscn_out_ptr = 0;
/* Initialize firmware. */
ha->request_ring_ptr = ha->request_ring;
ha->req_ring_index = 0;
ha->req_q_cnt = ha->request_q_length;
ha->response_ring_ptr = ha->response_ring;
ha->rsp_ring_index = 0;
/* Initialize response queue entries */
qla2x00_init_response_q_entries(ha);
ha->isp_ops.config_rings(ha);
spin_unlock_irqrestore(&ha->hardware_lock, flags);
/* Update any ISP specific firmware options before initialization. */
ha->isp_ops.update_fw_options(ha);
DEBUG(printk("scsi(%ld): Issue init firmware.\n", ha->host_no));
rval = qla2x00_init_firmware(ha, ha->init_cb_size);
if (rval) {
DEBUG2_3(printk("scsi(%ld): Init firmware **** FAILED ****.\n",
ha->host_no));
} else {
DEBUG3(printk("scsi(%ld): Init firmware -- success.\n",
ha->host_no));
}
return (rval);
}
/**
* qla2x00_fw_ready() - Waits for firmware ready.
* @ha: HA context
*
* Returns 0 on success.
*/
static int
qla2x00_fw_ready(scsi_qla_host_t *ha)
{
int rval;
unsigned long wtime, mtime;
uint16_t min_wait; /* Minimum wait time if loop is down */
uint16_t wait_time; /* Wait time if loop is coming ready */
uint16_t fw_state;
rval = QLA_SUCCESS;
/* 20 seconds for loop down. */
min_wait = 20;
/*
* Firmware should take at most one RATOV to login, plus 5 seconds for
* our own processing.
*/
if ((wait_time = (ha->retry_count*ha->login_timeout) + 5) < min_wait) {
wait_time = min_wait;
}
/* Min wait time if loop down */
mtime = jiffies + (min_wait * HZ);
/* wait time before firmware ready */
wtime = jiffies + (wait_time * HZ);
/* Wait for ISP to finish LIP */
if (!ha->flags.init_done)
qla_printk(KERN_INFO, ha, "Waiting for LIP to complete...\n");
DEBUG3(printk("scsi(%ld): Waiting for LIP to complete...\n",
ha->host_no));
do {
rval = qla2x00_get_firmware_state(ha, &fw_state);
if (rval == QLA_SUCCESS) {
if (fw_state < FSTATE_LOSS_OF_SYNC) {
ha->device_flags &= ~DFLG_NO_CABLE;
}
if (fw_state == FSTATE_READY) {
DEBUG(printk("scsi(%ld): F/W Ready - OK \n",
ha->host_no));
qla2x00_get_retry_cnt(ha, &ha->retry_count,
&ha->login_timeout, &ha->r_a_tov);
rval = QLA_SUCCESS;
break;
}
rval = QLA_FUNCTION_FAILED;
if (atomic_read(&ha->loop_down_timer) &&
fw_state != FSTATE_READY) {
/* Loop down. Timeout on min_wait for states
* other than Wait for Login.
*/
if (time_after_eq(jiffies, mtime)) {
qla_printk(KERN_INFO, ha,
"Cable is unplugged...\n");
ha->device_flags |= DFLG_NO_CABLE;
break;
}
}
} else {
/* Mailbox cmd failed. Timeout on min_wait. */
if (time_after_eq(jiffies, mtime))
break;
}
if (time_after_eq(jiffies, wtime))
break;
/* Delay for a while */
msleep(500);
DEBUG3(printk("scsi(%ld): fw_state=%x curr time=%lx.\n",
ha->host_no, fw_state, jiffies));
} while (1);
DEBUG(printk("scsi(%ld): fw_state=%x curr time=%lx.\n",
ha->host_no, fw_state, jiffies));
if (rval) {
DEBUG2_3(printk("scsi(%ld): Firmware ready **** FAILED ****.\n",
ha->host_no));
}
return (rval);
}
/*
* qla2x00_configure_hba
* Setup adapter context.
*
* Input:
* ha = adapter state pointer.
*
* Returns:
* 0 = success
*
* Context:
* Kernel context.
*/
static int
qla2x00_configure_hba(scsi_qla_host_t *ha)
{
int rval;
uint16_t loop_id;
uint16_t topo;
uint8_t al_pa;
uint8_t area;
uint8_t domain;
char connect_type[22];
/* Get host addresses. */
rval = qla2x00_get_adapter_id(ha,
&loop_id, &al_pa, &area, &domain, &topo);
if (rval != QLA_SUCCESS) {
if (LOOP_TRANSITION(ha) || atomic_read(&ha->loop_down_timer) ||
(rval == QLA_COMMAND_ERROR && loop_id == 0x7)) {
DEBUG2(printk("%s(%ld) Loop is in a transition state\n",
__func__, ha->host_no));
} else {
qla_printk(KERN_WARNING, ha,
"ERROR -- Unable to get host loop ID.\n");
set_bit(ISP_ABORT_NEEDED, &ha->dpc_flags);
}
return (rval);
}
if (topo == 4) {
qla_printk(KERN_INFO, ha,
"Cannot get topology - retrying.\n");
return (QLA_FUNCTION_FAILED);
}
ha->loop_id = loop_id;
/* initialize */
ha->min_external_loopid = SNS_FIRST_LOOP_ID;
ha->operating_mode = LOOP;
switch (topo) {
case 0:
DEBUG3(printk("scsi(%ld): HBA in NL topology.\n",
ha->host_no));
ha->current_topology = ISP_CFG_NL;
strcpy(connect_type, "(Loop)");
break;
case 1:
DEBUG3(printk("scsi(%ld): HBA in FL topology.\n",
ha->host_no));
ha->current_topology = ISP_CFG_FL;
strcpy(connect_type, "(FL_Port)");
break;
case 2:
DEBUG3(printk("scsi(%ld): HBA in N P2P topology.\n",
ha->host_no));
ha->operating_mode = P2P;
ha->current_topology = ISP_CFG_N;
strcpy(connect_type, "(N_Port-to-N_Port)");
break;
case 3:
DEBUG3(printk("scsi(%ld): HBA in F P2P topology.\n",
ha->host_no));
ha->operating_mode = P2P;
ha->current_topology = ISP_CFG_F;
strcpy(connect_type, "(F_Port)");
break;
default:
DEBUG3(printk("scsi(%ld): HBA in unknown topology %x. "
"Using NL.\n",
ha->host_no, topo));
ha->current_topology = ISP_CFG_NL;
strcpy(connect_type, "(Loop)");
break;
}
/* Save Host port and loop ID. */
/* byte order - Big Endian */
ha->d_id.b.domain = domain;
ha->d_id.b.area = area;
ha->d_id.b.al_pa = al_pa;
if (!ha->flags.init_done)
qla_printk(KERN_INFO, ha,
"Topology - %s, Host Loop address 0x%x\n",
connect_type, ha->loop_id);
if (rval) {
DEBUG2_3(printk("scsi(%ld): FAILED.\n", ha->host_no));
} else {
DEBUG3(printk("scsi(%ld): exiting normally.\n", ha->host_no));
}
return(rval);
}
/*
* NVRAM configuration for ISP 2xxx
*
* Input:
* ha = adapter block pointer.
*
* Output:
* initialization control block in response_ring
* host adapters parameters in host adapter block
*
* Returns:
* 0 = success.
*/
int
qla2x00_nvram_config(scsi_qla_host_t *ha)
{
int rval;
uint8_t chksum = 0;
uint16_t cnt;
uint8_t *dptr1, *dptr2;
init_cb_t *icb = ha->init_cb;
nvram_t *nv = (nvram_t *)ha->request_ring;
uint8_t *ptr = (uint8_t *)ha->request_ring;
struct device_reg_2xxx __iomem *reg = &ha->iobase->isp;
rval = QLA_SUCCESS;
/* Determine NVRAM starting address. */
ha->nvram_size = sizeof(nvram_t);
ha->nvram_base = 0;
if (!IS_QLA2100(ha) && !IS_QLA2200(ha) && !IS_QLA2300(ha))
if ((RD_REG_WORD(&reg->ctrl_status) >> 14) == 1)
ha->nvram_base = 0x80;
/* Get NVRAM data and calculate checksum. */
ha->isp_ops.read_nvram(ha, ptr, ha->nvram_base, ha->nvram_size);
for (cnt = 0, chksum = 0; cnt < ha->nvram_size; cnt++)
chksum += *ptr++;
DEBUG5(printk("scsi(%ld): Contents of NVRAM\n", ha->host_no));
DEBUG5(qla2x00_dump_buffer((uint8_t *)ha->request_ring,
ha->nvram_size));
/* Bad NVRAM data, set defaults parameters. */
if (chksum || nv->id[0] != 'I' || nv->id[1] != 'S' ||
nv->id[2] != 'P' || nv->id[3] != ' ' || nv->nvram_version < 1) {
/* Reset NVRAM data. */
qla_printk(KERN_WARNING, ha, "Inconsistent NVRAM detected: "
"checksum=0x%x id=%c version=0x%x.\n", chksum, nv->id[0],
nv->nvram_version);
qla_printk(KERN_WARNING, ha, "Falling back to functioning (yet "
"invalid -- WWPN) defaults.\n");
/*
* Set default initialization control block.
*/
memset(nv, 0, ha->nvram_size);
nv->parameter_block_version = ICB_VERSION;
if (IS_QLA23XX(ha)) {
nv->firmware_options[0] = BIT_2 | BIT_1;
nv->firmware_options[1] = BIT_7 | BIT_5;
nv->add_firmware_options[0] = BIT_5;
nv->add_firmware_options[1] = BIT_5 | BIT_4;
nv->frame_payload_size = __constant_cpu_to_le16(2048);
nv->special_options[1] = BIT_7;
} else if (IS_QLA2200(ha)) {
nv->firmware_options[0] = BIT_2 | BIT_1;
nv->firmware_options[1] = BIT_7 | BIT_5;
nv->add_firmware_options[0] = BIT_5;
nv->add_firmware_options[1] = BIT_5 | BIT_4;
nv->frame_payload_size = __constant_cpu_to_le16(1024);
} else if (IS_QLA2100(ha)) {
nv->firmware_options[0] = BIT_3 | BIT_1;
nv->firmware_options[1] = BIT_5;
nv->frame_payload_size = __constant_cpu_to_le16(1024);
}
nv->max_iocb_allocation = __constant_cpu_to_le16(256);
nv->execution_throttle = __constant_cpu_to_le16(16);
nv->retry_count = 8;
nv->retry_delay = 1;
nv->port_name[0] = 33;
nv->port_name[3] = 224;
nv->port_name[4] = 139;
nv->login_timeout = 4;
/*
* Set default host adapter parameters
*/
nv->host_p[1] = BIT_2;
nv->reset_delay = 5;
nv->port_down_retry_count = 8;
nv->max_luns_per_target = __constant_cpu_to_le16(8);
nv->link_down_timeout = 60;
rval = 1;
}
#if defined(CONFIG_IA64_GENERIC) || defined(CONFIG_IA64_SGI_SN2)
/*
* The SN2 does not provide BIOS emulation which means you can't change
* potentially bogus BIOS settings. Force the use of default settings
* for link rate and frame size. Hope that the rest of the settings
* are valid.
*/
if (ia64_platform_is("sn2")) {
nv->frame_payload_size = __constant_cpu_to_le16(2048);
if (IS_QLA23XX(ha))
nv->special_options[1] = BIT_7;
}
#endif
/* Reset Initialization control block */
memset(icb, 0, ha->init_cb_size);
/*
* Setup driver NVRAM options.
*/
nv->firmware_options[0] |= (BIT_6 | BIT_1);
nv->firmware_options[0] &= ~(BIT_5 | BIT_4);
nv->firmware_options[1] |= (BIT_5 | BIT_0);
nv->firmware_options[1] &= ~BIT_4;
if (IS_QLA23XX(ha)) {
nv->firmware_options[0] |= BIT_2;
nv->firmware_options[0] &= ~BIT_3;
nv->add_firmware_options[1] |= BIT_5 | BIT_4;
if (IS_QLA2300(ha)) {
if (ha->fb_rev == FPM_2310) {
strcpy(ha->model_number, "QLA2310");
} else {
strcpy(ha->model_number, "QLA2300");
}
} else {
if (rval == 0 &&
memcmp(nv->model_number, BINZERO,
sizeof(nv->model_number)) != 0) {
char *st, *en;
strncpy(ha->model_number, nv->model_number,
sizeof(nv->model_number));
st = en = ha->model_number;
en += sizeof(nv->model_number) - 1;
while (en > st) {
if (*en != 0x20 && *en != 0x00)
break;
*en-- = '\0';
}
} else {
uint16_t index;
index = (ha->pdev->subsystem_device & 0xff);
if (index < QLA_MODEL_NAMES) {
strcpy(ha->model_number,
qla2x00_model_name[index * 2]);
ha->model_desc =
qla2x00_model_name[index * 2 + 1];
} else {
strcpy(ha->model_number, "QLA23xx");
}
}
}
} else if (IS_QLA2200(ha)) {
nv->firmware_options[0] |= BIT_2;
/*
* 'Point-to-point preferred, else loop' is not a safe
* connection mode setting.
*/
if ((nv->add_firmware_options[0] & (BIT_6 | BIT_5 | BIT_4)) ==
(BIT_5 | BIT_4)) {
/* Force 'loop preferred, else point-to-point'. */
nv->add_firmware_options[0] &= ~(BIT_6 | BIT_5 | BIT_4);
nv->add_firmware_options[0] |= BIT_5;
}
strcpy(ha->model_number, "QLA22xx");
} else /*if (IS_QLA2100(ha))*/ {
strcpy(ha->model_number, "QLA2100");
}
/*
* Copy over NVRAM RISC parameter block to initialization control block.
*/
dptr1 = (uint8_t *)icb;
dptr2 = (uint8_t *)&nv->parameter_block_version;
cnt = (uint8_t *)&icb->request_q_outpointer - (uint8_t *)&icb->version;
while (cnt--)
*dptr1++ = *dptr2++;
/* Copy 2nd half. */
dptr1 = (uint8_t *)icb->add_firmware_options;
cnt = (uint8_t *)icb->reserved_3 - (uint8_t *)icb->add_firmware_options;
while (cnt--)
*dptr1++ = *dptr2++;
/* Use alternate WWN? */
if (nv->host_p[1] & BIT_7) {
memcpy(icb->node_name, nv->alternate_node_name, WWN_SIZE);
memcpy(icb->port_name, nv->alternate_port_name, WWN_SIZE);
}
/* Prepare nodename */
if ((icb->firmware_options[1] & BIT_6) == 0) {
/*
* Firmware will apply the following mask if the nodename was
* not provided.
*/
memcpy(icb->node_name, icb->port_name, WWN_SIZE);
icb->node_name[0] &= 0xF0;
}
/*
* Set host adapter parameters.
*/
if (nv->host_p[0] & BIT_7)
ql2xextended_error_logging = 1;
ha->flags.disable_risc_code_load = ((nv->host_p[0] & BIT_4) ? 1 : 0);
/* Always load RISC code on non ISP2[12]00 chips. */
if (!IS_QLA2100(ha) && !IS_QLA2200(ha))
ha->flags.disable_risc_code_load = 0;
ha->flags.enable_lip_reset = ((nv->host_p[1] & BIT_1) ? 1 : 0);
ha->flags.enable_lip_full_login = ((nv->host_p[1] & BIT_2) ? 1 : 0);
ha->flags.enable_target_reset = ((nv->host_p[1] & BIT_3) ? 1 : 0);
ha->flags.enable_led_scheme = (nv->special_options[1] & BIT_4) ? 1 : 0;
ha->flags.disable_serdes = 0;
ha->operating_mode =
(icb->add_firmware_options[0] & (BIT_6 | BIT_5 | BIT_4)) >> 4;
memcpy(ha->fw_seriallink_options, nv->seriallink_options,
sizeof(ha->fw_seriallink_options));
/* save HBA serial number */
ha->serial0 = icb->port_name[5];
ha->serial1 = icb->port_name[6];
ha->serial2 = icb->port_name[7];
ha->node_name = icb->node_name;
ha->port_name = icb->port_name;
icb->execution_throttle = __constant_cpu_to_le16(0xFFFF);
ha->retry_count = nv->retry_count;
/* Set minimum login_timeout to 4 seconds. */
if (nv->login_timeout < ql2xlogintimeout)
nv->login_timeout = ql2xlogintimeout;
if (nv->login_timeout < 4)
nv->login_timeout = 4;
ha->login_timeout = nv->login_timeout;
icb->login_timeout = nv->login_timeout;
/* Set minimum RATOV to 200 tenths of a second. */
ha->r_a_tov = 200;
ha->loop_reset_delay = nv->reset_delay;
/* Link Down Timeout = 0:
*
* When Port Down timer expires we will start returning
* I/O's to OS with "DID_NO_CONNECT".
*
* Link Down Timeout != 0:
*
* The driver waits for the link to come up after link down
* before returning I/Os to OS with "DID_NO_CONNECT".
*/
if (nv->link_down_timeout == 0) {
ha->loop_down_abort_time =
(LOOP_DOWN_TIME - LOOP_DOWN_TIMEOUT);
} else {
ha->link_down_timeout = nv->link_down_timeout;
ha->loop_down_abort_time =
(LOOP_DOWN_TIME - ha->link_down_timeout);
}
/*
* Need enough time to try and get the port back.
*/
ha->port_down_retry_count = nv->port_down_retry_count;
if (qlport_down_retry)
ha->port_down_retry_count = qlport_down_retry;
/* Set login_retry_count */
ha->login_retry_count = nv->retry_count;
if (ha->port_down_retry_count == nv->port_down_retry_count &&
ha->port_down_retry_count > 3)
ha->login_retry_count = ha->port_down_retry_count;
else if (ha->port_down_retry_count > (int)ha->login_retry_count)
ha->login_retry_count = ha->port_down_retry_count;
if (ql2xloginretrycount)
ha->login_retry_count = ql2xloginretrycount;
icb->lun_enables = __constant_cpu_to_le16(0);
icb->command_resource_count = 0;
icb->immediate_notify_resource_count = 0;
icb->timeout = __constant_cpu_to_le16(0);
if (IS_QLA2100(ha) || IS_QLA2200(ha)) {
/* Enable RIO */
icb->firmware_options[0] &= ~BIT_3;
icb->add_firmware_options[0] &=
~(BIT_3 | BIT_2 | BIT_1 | BIT_0);
icb->add_firmware_options[0] |= BIT_2;
icb->response_accumulation_timer = 3;
icb->interrupt_delay_timer = 5;
ha->flags.process_response_queue = 1;
} else {
/* Enable ZIO. */
if (!ha->flags.init_done) {
ha->zio_mode = icb->add_firmware_options[0] &
(BIT_3 | BIT_2 | BIT_1 | BIT_0);
ha->zio_timer = icb->interrupt_delay_timer ?
icb->interrupt_delay_timer: 2;
}
icb->add_firmware_options[0] &=
~(BIT_3 | BIT_2 | BIT_1 | BIT_0);
ha->flags.process_response_queue = 0;
if (ha->zio_mode != QLA_ZIO_DISABLED) {
ha->zio_mode = QLA_ZIO_MODE_6;
DEBUG2(printk("scsi(%ld): ZIO mode %d enabled; timer "
"delay (%d us).\n", ha->host_no, ha->zio_mode,
ha->zio_timer * 100));
qla_printk(KERN_INFO, ha,
"ZIO mode %d enabled; timer delay (%d us).\n",
ha->zio_mode, ha->zio_timer * 100);
icb->add_firmware_options[0] |= (uint8_t)ha->zio_mode;
icb->interrupt_delay_timer = (uint8_t)ha->zio_timer;
ha->flags.process_response_queue = 1;
}
}
if (rval) {
DEBUG2_3(printk(KERN_WARNING
"scsi(%ld): NVRAM configuration failed!\n", ha->host_no));
}
return (rval);
}
[SCSI] update fc_transport for removal of block/unblock functions We recently went back to implement a board reset. When we perform the reset, we wanted to tear down the internal data structures and rebuild them. Unfortunately, when it came to the rport structure, things were odd. If we deleted them, the scsi targets and sdevs would be torn down. Not a good thing for a temporary reset. We could block the rports, but we either maintain the internal structures to keep the rport reference (perhaps even replicating what's in the transport), or we have to fatten the fc transport with new search routines to find the rport (and deal with a case of a dangling rport that the driver forgets). It dawned on me that we had actually reached this state incorrectly. When the fc transport first started, we did the block/unblock first, then added the rport interface. The purpose of block/unblock is to hide the temporary disappearance of the rport (e.g. being deleted, then readded). Why are we making the driver do the block/unblock ? We should be making the transport have only an rport add/delete, and the let the transport handle the block/unblock. So... This patch removes the existing fc_remote_port_block/unblock functions. It moves the block/unblock functionality into the fc_remote_port_add/delete functions. Updates for the lpfc driver are included. Qlogic driver updates are also enclosed, thanks to the contributions of Andrew Vasquez. [Note: the qla2xxx changes are relative to the scsi-misc-2.6 tree as of this morning - which does not include the recent patches sent by Andrew]. The zfcp driver does not use the block/unblock functions. One last comment: The resulting behavior feels very clean. The LLDD is concerned only with add/delete, which corresponds to the physical disappearance. However, the fact that the scsi target and sdevs are not immediately torn down after the LLDD calls delete causes an interesting scenario... the midlayer can call the xxx_slave_alloc and xxx_queuecommand functions with a sdev that is at the location the rport used to be. The driver must validate the device exists when it first enters these functions. In thinking about it, this has always been the case for the LLDD and these routines. The existing drivers already check for existence. However, this highlights that simple validation via data structure dereferencing needs to be watched. To deal with this, a new transport function, fc_remote_port_chkready() was created that LLDDs should call when they first enter these two routines. It validates the rport state, and returns a scsi result which could be returned. In addition to solving the above, it also creates consistent behavior from the LLDD's when the block and deletes are occuring. Rejections fixed up and Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-19 00:03:35 +08:00
static void
qla2x00_rport_del(void *data)
{
fc_port_t *fcport = data;
struct fc_rport *rport;
unsigned long flags;
spin_lock_irqsave(&fcport->rport_lock, flags);
rport = fcport->drport;
fcport->drport = NULL;
spin_unlock_irqrestore(&fcport->rport_lock, flags);
if (rport)
fc_remote_port_delete(rport);
[SCSI] update fc_transport for removal of block/unblock functions We recently went back to implement a board reset. When we perform the reset, we wanted to tear down the internal data structures and rebuild them. Unfortunately, when it came to the rport structure, things were odd. If we deleted them, the scsi targets and sdevs would be torn down. Not a good thing for a temporary reset. We could block the rports, but we either maintain the internal structures to keep the rport reference (perhaps even replicating what's in the transport), or we have to fatten the fc transport with new search routines to find the rport (and deal with a case of a dangling rport that the driver forgets). It dawned on me that we had actually reached this state incorrectly. When the fc transport first started, we did the block/unblock first, then added the rport interface. The purpose of block/unblock is to hide the temporary disappearance of the rport (e.g. being deleted, then readded). Why are we making the driver do the block/unblock ? We should be making the transport have only an rport add/delete, and the let the transport handle the block/unblock. So... This patch removes the existing fc_remote_port_block/unblock functions. It moves the block/unblock functionality into the fc_remote_port_add/delete functions. Updates for the lpfc driver are included. Qlogic driver updates are also enclosed, thanks to the contributions of Andrew Vasquez. [Note: the qla2xxx changes are relative to the scsi-misc-2.6 tree as of this morning - which does not include the recent patches sent by Andrew]. The zfcp driver does not use the block/unblock functions. One last comment: The resulting behavior feels very clean. The LLDD is concerned only with add/delete, which corresponds to the physical disappearance. However, the fact that the scsi target and sdevs are not immediately torn down after the LLDD calls delete causes an interesting scenario... the midlayer can call the xxx_slave_alloc and xxx_queuecommand functions with a sdev that is at the location the rport used to be. The driver must validate the device exists when it first enters these functions. In thinking about it, this has always been the case for the LLDD and these routines. The existing drivers already check for existence. However, this highlights that simple validation via data structure dereferencing needs to be watched. To deal with this, a new transport function, fc_remote_port_chkready() was created that LLDDs should call when they first enter these two routines. It validates the rport state, and returns a scsi result which could be returned. In addition to solving the above, it also creates consistent behavior from the LLDD's when the block and deletes are occuring. Rejections fixed up and Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-19 00:03:35 +08:00
}
/**
* qla2x00_alloc_fcport() - Allocate a generic fcport.
* @ha: HA context
* @flags: allocation flags
*
* Returns a pointer to the allocated fcport, or NULL, if none available.
*/
static fc_port_t *
qla2x00_alloc_fcport(scsi_qla_host_t *ha, gfp_t flags)
{
fc_port_t *fcport;
fcport = kmalloc(sizeof(fc_port_t), flags);
if (fcport == NULL)
return (fcport);
/* Setup fcport template structure. */
memset(fcport, 0, sizeof (fc_port_t));
fcport->ha = ha;
fcport->port_type = FCT_UNKNOWN;
fcport->loop_id = FC_NO_LOOP_ID;
atomic_set(&fcport->state, FCS_UNCONFIGURED);
fcport->flags = FCF_RLC_SUPPORT;
fcport->supported_classes = FC_COS_UNSPECIFIED;
spin_lock_init(&fcport->rport_lock);
return (fcport);
}
/*
* qla2x00_configure_loop
* Updates Fibre Channel Device Database with what is actually on loop.
*
* Input:
* ha = adapter block pointer.
*
* Returns:
* 0 = success.
* 1 = error.
* 2 = database was full and device was not configured.
*/
static int
qla2x00_configure_loop(scsi_qla_host_t *ha)
{
int rval;
unsigned long flags, save_flags;
rval = QLA_SUCCESS;
/* Get Initiator ID */
if (test_bit(LOCAL_LOOP_UPDATE, &ha->dpc_flags)) {
rval = qla2x00_configure_hba(ha);
if (rval != QLA_SUCCESS) {
DEBUG(printk("scsi(%ld): Unable to configure HBA.\n",
ha->host_no));
return (rval);
}
}
save_flags = flags = ha->dpc_flags;
DEBUG(printk("scsi(%ld): Configure loop -- dpc flags =0x%lx\n",
ha->host_no, flags));
/*
* If we have both an RSCN and PORT UPDATE pending then handle them
* both at the same time.
*/
clear_bit(LOCAL_LOOP_UPDATE, &ha->dpc_flags);
clear_bit(RSCN_UPDATE, &ha->dpc_flags);
/* Determine what we need to do */
if (ha->current_topology == ISP_CFG_FL &&
(test_bit(LOCAL_LOOP_UPDATE, &flags))) {
ha->flags.rscn_queue_overflow = 1;
set_bit(RSCN_UPDATE, &flags);
} else if (ha->current_topology == ISP_CFG_F &&
(test_bit(LOCAL_LOOP_UPDATE, &flags))) {
ha->flags.rscn_queue_overflow = 1;
set_bit(RSCN_UPDATE, &flags);
clear_bit(LOCAL_LOOP_UPDATE, &flags);
} else if (ha->current_topology == ISP_CFG_N) {
clear_bit(RSCN_UPDATE, &flags);
} else if (!ha->flags.online ||
(test_bit(ABORT_ISP_ACTIVE, &flags))) {
ha->flags.rscn_queue_overflow = 1;
set_bit(RSCN_UPDATE, &flags);
set_bit(LOCAL_LOOP_UPDATE, &flags);
}
if (test_bit(LOCAL_LOOP_UPDATE, &flags)) {
if (test_bit(LOOP_RESYNC_NEEDED, &ha->dpc_flags)) {
rval = QLA_FUNCTION_FAILED;
} else {
rval = qla2x00_configure_local_loop(ha);
}
}
if (rval == QLA_SUCCESS && test_bit(RSCN_UPDATE, &flags)) {
if (LOOP_TRANSITION(ha)) {
rval = QLA_FUNCTION_FAILED;
} else {
rval = qla2x00_configure_fabric(ha);
}
}
if (rval == QLA_SUCCESS) {
if (atomic_read(&ha->loop_down_timer) ||
test_bit(LOOP_RESYNC_NEEDED, &ha->dpc_flags)) {
rval = QLA_FUNCTION_FAILED;
} else {
atomic_set(&ha->loop_state, LOOP_READY);
DEBUG(printk("scsi(%ld): LOOP READY\n", ha->host_no));
}
}
if (rval) {
DEBUG2_3(printk("%s(%ld): *** FAILED ***\n",
__func__, ha->host_no));
} else {
DEBUG3(printk("%s: exiting normally\n", __func__));
}
/* Restore state if a resync event occured during processing */
if (test_bit(LOOP_RESYNC_NEEDED, &ha->dpc_flags)) {
if (test_bit(LOCAL_LOOP_UPDATE, &save_flags))
set_bit(LOCAL_LOOP_UPDATE, &ha->dpc_flags);
if (test_bit(RSCN_UPDATE, &save_flags))
set_bit(RSCN_UPDATE, &ha->dpc_flags);
}
return (rval);
}
/*
* qla2x00_configure_local_loop
* Updates Fibre Channel Device Database with local loop devices.
*
* Input:
* ha = adapter block pointer.
*
* Returns:
* 0 = success.
*/
static int
qla2x00_configure_local_loop(scsi_qla_host_t *ha)
{
int rval, rval2;
int found_devs;
int found;
fc_port_t *fcport, *new_fcport;
uint16_t index;
uint16_t entries;
char *id_iter;
uint16_t loop_id;
uint8_t domain, area, al_pa;
found_devs = 0;
new_fcport = NULL;
entries = MAX_FIBRE_DEVICES;
DEBUG3(printk("scsi(%ld): Getting FCAL position map\n", ha->host_no));
DEBUG3(qla2x00_get_fcal_position_map(ha, NULL));
/* Get list of logged in devices. */
memset(ha->gid_list, 0, GID_LIST_SIZE);
rval = qla2x00_get_id_list(ha, ha->gid_list, ha->gid_list_dma,
&entries);
if (rval != QLA_SUCCESS)
goto cleanup_allocation;
DEBUG3(printk("scsi(%ld): Entries in ID list (%d)\n",
ha->host_no, entries));
DEBUG3(qla2x00_dump_buffer((uint8_t *)ha->gid_list,
entries * sizeof(struct gid_list_info)));
/* Allocate temporary fcport for any new fcports discovered. */
new_fcport = qla2x00_alloc_fcport(ha, GFP_KERNEL);
if (new_fcport == NULL) {
rval = QLA_MEMORY_ALLOC_FAILED;
goto cleanup_allocation;
}
new_fcport->flags &= ~FCF_FABRIC_DEVICE;
/*
* Mark local devices that were present with FCF_DEVICE_LOST for now.
*/
list_for_each_entry(fcport, &ha->fcports, list) {
if (atomic_read(&fcport->state) == FCS_ONLINE &&
fcport->port_type != FCT_BROADCAST &&
(fcport->flags & FCF_FABRIC_DEVICE) == 0) {
DEBUG(printk("scsi(%ld): Marking port lost, "
"loop_id=0x%04x\n",
ha->host_no, fcport->loop_id));
atomic_set(&fcport->state, FCS_DEVICE_LOST);
fcport->flags &= ~FCF_FARP_DONE;
}
}
/* Add devices to port list. */
id_iter = (char *)ha->gid_list;
for (index = 0; index < entries; index++) {
domain = ((struct gid_list_info *)id_iter)->domain;
area = ((struct gid_list_info *)id_iter)->area;
al_pa = ((struct gid_list_info *)id_iter)->al_pa;
if (IS_QLA2100(ha) || IS_QLA2200(ha))
loop_id = (uint16_t)
((struct gid_list_info *)id_iter)->loop_id_2100;
else
loop_id = le16_to_cpu(
((struct gid_list_info *)id_iter)->loop_id);
id_iter += ha->gid_list_info_size;
/* Bypass reserved domain fields. */
if ((domain & 0xf0) == 0xf0)
continue;
/* Bypass if not same domain and area of adapter. */
if (area && domain &&
(area != ha->d_id.b.area || domain != ha->d_id.b.domain))
continue;
/* Bypass invalid local loop ID. */
if (loop_id > LAST_LOCAL_LOOP_ID)
continue;
/* Fill in member data. */
new_fcport->d_id.b.domain = domain;
new_fcport->d_id.b.area = area;
new_fcport->d_id.b.al_pa = al_pa;
new_fcport->loop_id = loop_id;
rval2 = qla2x00_get_port_database(ha, new_fcport, 0);
if (rval2 != QLA_SUCCESS) {
DEBUG2(printk("scsi(%ld): Failed to retrieve fcport "
"information -- get_port_database=%x, "
"loop_id=0x%04x\n",
ha->host_no, rval2, new_fcport->loop_id));
DEBUG2(printk("scsi(%ld): Scheduling resync...\n",
ha->host_no));
set_bit(LOOP_RESYNC_NEEDED, &ha->dpc_flags);
continue;
}
/* Check for matching device in port list. */
found = 0;
fcport = NULL;
list_for_each_entry(fcport, &ha->fcports, list) {
if (memcmp(new_fcport->port_name, fcport->port_name,
WWN_SIZE))
continue;
fcport->flags &= ~(FCF_FABRIC_DEVICE |
FCF_PERSISTENT_BOUND);
fcport->loop_id = new_fcport->loop_id;
fcport->port_type = new_fcport->port_type;
fcport->d_id.b24 = new_fcport->d_id.b24;
memcpy(fcport->node_name, new_fcport->node_name,
WWN_SIZE);
found++;
break;
}
if (!found) {
/* New device, add to fcports list. */
new_fcport->flags &= ~FCF_PERSISTENT_BOUND;
list_add_tail(&new_fcport->list, &ha->fcports);
/* Allocate a new replacement fcport. */
fcport = new_fcport;
new_fcport = qla2x00_alloc_fcport(ha, GFP_KERNEL);
if (new_fcport == NULL) {
rval = QLA_MEMORY_ALLOC_FAILED;
goto cleanup_allocation;
}
new_fcport->flags &= ~FCF_FABRIC_DEVICE;
}
/* Base iIDMA settings on HBA port speed. */
switch (ha->link_data_rate) {
case PORT_SPEED_1GB:
fcport->fp_speed = cpu_to_be16(BIT_15);
break;
case PORT_SPEED_2GB:
fcport->fp_speed = cpu_to_be16(BIT_14);
break;
case PORT_SPEED_4GB:
fcport->fp_speed = cpu_to_be16(BIT_13);
break;
}
qla2x00_update_fcport(ha, fcport);
found_devs++;
}
cleanup_allocation:
kfree(new_fcport);
if (rval != QLA_SUCCESS) {
DEBUG2(printk("scsi(%ld): Configure local loop error exit: "
"rval=%x\n", ha->host_no, rval));
}
if (found_devs) {
ha->device_flags |= DFLG_LOCAL_DEVICES;
ha->device_flags &= ~DFLG_RETRY_LOCAL_DEVICES;
}
return (rval);
}
static void
qla2x00_probe_for_all_luns(scsi_qla_host_t *ha)
{
fc_port_t *fcport;
qla2x00_mark_all_devices_lost(ha, 0);
list_for_each_entry(fcport, &ha->fcports, list) {
if (fcport->port_type != FCT_TARGET)
continue;
qla2x00_update_fcport(ha, fcport);
}
}
static void
qla2x00_iidma_fcport(scsi_qla_host_t *ha, fc_port_t *fcport)
{
#define LS_UNKNOWN 2
static char *link_speeds[5] = { "1", "2", "?", "4" };
int rval;
uint16_t port_speed, mb[6];
if (!IS_QLA24XX(ha))
return;
switch (be16_to_cpu(fcport->fp_speed)) {
case BIT_15:
port_speed = PORT_SPEED_1GB;
break;
case BIT_14:
port_speed = PORT_SPEED_2GB;
break;
case BIT_13:
port_speed = PORT_SPEED_4GB;
break;
default:
DEBUG2(printk("scsi(%ld): %02x%02x%02x%02x%02x%02x%02x%02x -- "
"unsupported FM port operating speed (%04x).\n",
ha->host_no, fcport->port_name[0], fcport->port_name[1],
fcport->port_name[2], fcport->port_name[3],
fcport->port_name[4], fcport->port_name[5],
fcport->port_name[6], fcport->port_name[7],
be16_to_cpu(fcport->fp_speed)));
port_speed = PORT_SPEED_UNKNOWN;
break;
}
if (port_speed == PORT_SPEED_UNKNOWN)
return;
rval = qla2x00_set_idma_speed(ha, fcport->loop_id, port_speed, mb);
if (rval != QLA_SUCCESS) {
DEBUG2(printk("scsi(%ld): Unable to adjust iIDMA "
"%02x%02x%02x%02x%02x%02x%02x%02x -- %04x %x %04x %04x.\n",
ha->host_no, fcport->port_name[0], fcport->port_name[1],
fcport->port_name[2], fcport->port_name[3],
fcport->port_name[4], fcport->port_name[5],
fcport->port_name[6], fcport->port_name[7], rval,
port_speed, mb[0], mb[1]));
} else {
DEBUG2(qla_printk(KERN_INFO, ha,
"iIDMA adjusted to %s GB/s on "
"%02x%02x%02x%02x%02x%02x%02x%02x.\n",
link_speeds[port_speed], fcport->port_name[0],
fcport->port_name[1], fcport->port_name[2],
fcport->port_name[3], fcport->port_name[4],
fcport->port_name[5], fcport->port_name[6],
fcport->port_name[7]));
}
}
/*
* qla2x00_update_fcport
* Updates device on list.
*
* Input:
* ha = adapter block pointer.
* fcport = port structure pointer.
*
* Return:
* 0 - Success
* BIT_0 - error
*
* Context:
* Kernel context.
*/
void
qla2x00_update_fcport(scsi_qla_host_t *ha, fc_port_t *fcport)
{
fcport->ha = ha;
fcport->login_retry = 0;
fcport->port_login_retry_count = ha->port_down_retry_count *
PORT_RETRY_TIME;
atomic_set(&fcport->port_down_timer, ha->port_down_retry_count *
PORT_RETRY_TIME);
fcport->flags &= ~FCF_LOGIN_NEEDED;
qla2x00_iidma_fcport(ha, fcport);
atomic_set(&fcport->state, FCS_ONLINE);
qla2x00_reg_remote_port(ha, fcport);
}
void
qla2x00_reg_remote_port(scsi_qla_host_t *ha, fc_port_t *fcport)
{
struct fc_rport_identifiers rport_ids;
struct fc_rport *rport;
unsigned long flags;
if (fcport->drport)
qla2x00_rport_del(fcport);
if (fcport->rport)
return;
rport_ids.node_name = wwn_to_u64(fcport->node_name);
rport_ids.port_name = wwn_to_u64(fcport->port_name);
rport_ids.port_id = fcport->d_id.b.domain << 16 |
fcport->d_id.b.area << 8 | fcport->d_id.b.al_pa;
rport_ids.roles = FC_RPORT_ROLE_UNKNOWN;
rport = fc_remote_port_add(ha->host, 0, &rport_ids);
if (!rport) {
qla_printk(KERN_WARNING, ha,
"Unable to allocate fc remote port!\n");
return;
}
spin_lock_irqsave(&fcport->rport_lock, flags);
fcport->rport = rport;
[SCSI] update fc_transport for removal of block/unblock functions We recently went back to implement a board reset. When we perform the reset, we wanted to tear down the internal data structures and rebuild them. Unfortunately, when it came to the rport structure, things were odd. If we deleted them, the scsi targets and sdevs would be torn down. Not a good thing for a temporary reset. We could block the rports, but we either maintain the internal structures to keep the rport reference (perhaps even replicating what's in the transport), or we have to fatten the fc transport with new search routines to find the rport (and deal with a case of a dangling rport that the driver forgets). It dawned on me that we had actually reached this state incorrectly. When the fc transport first started, we did the block/unblock first, then added the rport interface. The purpose of block/unblock is to hide the temporary disappearance of the rport (e.g. being deleted, then readded). Why are we making the driver do the block/unblock ? We should be making the transport have only an rport add/delete, and the let the transport handle the block/unblock. So... This patch removes the existing fc_remote_port_block/unblock functions. It moves the block/unblock functionality into the fc_remote_port_add/delete functions. Updates for the lpfc driver are included. Qlogic driver updates are also enclosed, thanks to the contributions of Andrew Vasquez. [Note: the qla2xxx changes are relative to the scsi-misc-2.6 tree as of this morning - which does not include the recent patches sent by Andrew]. The zfcp driver does not use the block/unblock functions. One last comment: The resulting behavior feels very clean. The LLDD is concerned only with add/delete, which corresponds to the physical disappearance. However, the fact that the scsi target and sdevs are not immediately torn down after the LLDD calls delete causes an interesting scenario... the midlayer can call the xxx_slave_alloc and xxx_queuecommand functions with a sdev that is at the location the rport used to be. The driver must validate the device exists when it first enters these functions. In thinking about it, this has always been the case for the LLDD and these routines. The existing drivers already check for existence. However, this highlights that simple validation via data structure dereferencing needs to be watched. To deal with this, a new transport function, fc_remote_port_chkready() was created that LLDDs should call when they first enter these two routines. It validates the rport state, and returns a scsi result which could be returned. In addition to solving the above, it also creates consistent behavior from the LLDD's when the block and deletes are occuring. Rejections fixed up and Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
2005-10-19 00:03:35 +08:00
*((fc_port_t **)rport->dd_data) = fcport;
spin_unlock_irqrestore(&fcport->rport_lock, flags);
rport->supported_classes = fcport->supported_classes;
rport_ids.roles = FC_RPORT_ROLE_UNKNOWN;
if (fcport->port_type == FCT_INITIATOR)
rport_ids.roles |= FC_RPORT_ROLE_FCP_INITIATOR;
if (fcport->port_type == FCT_TARGET)
rport_ids.roles |= FC_RPORT_ROLE_FCP_TARGET;
fc_remote_port_rolechg(rport, rport_ids.roles);
if (rport->scsi_target_id != -1 &&
rport->scsi_target_id < ha->host->max_id)
fcport->os_target_id = rport->scsi_target_id;
}
/*
* qla2x00_configure_fabric
* Setup SNS devices with loop ID's.
*
* Input:
* ha = adapter block pointer.
*
* Returns:
* 0 = success.
* BIT_0 = error
*/
static int
qla2x00_configure_fabric(scsi_qla_host_t *ha)
{
int rval, rval2;
fc_port_t *fcport, *fcptemp;
uint16_t next_loopid;
uint16_t mb[MAILBOX_REGISTER_COUNT];
uint16_t loop_id;
LIST_HEAD(new_fcports);
/* If FL port exists, then SNS is present */
if (IS_QLA24XX(ha) || IS_QLA54XX(ha))
loop_id = NPH_F_PORT;
else
loop_id = SNS_FL_PORT;
rval = qla2x00_get_port_name(ha, loop_id, ha->fabric_node_name, 1);
if (rval != QLA_SUCCESS) {
DEBUG2(printk("scsi(%ld): MBC_GET_PORT_NAME Failed, No FL "
"Port\n", ha->host_no));
ha->device_flags &= ~SWITCH_FOUND;
return (QLA_SUCCESS);
}
ha->device_flags |= SWITCH_FOUND;
/* Mark devices that need re-synchronization. */
rval2 = qla2x00_device_resync(ha);
if (rval2 == QLA_RSCNS_HANDLED) {
/* No point doing the scan, just continue. */
return (QLA_SUCCESS);
}
do {
/* FDMI support. */
if (ql2xfdmienable &&
test_and_clear_bit(REGISTER_FDMI_NEEDED, &ha->dpc_flags))
qla2x00_fdmi_register(ha);
/* Ensure we are logged into the SNS. */
if (IS_QLA24XX(ha) || IS_QLA54XX(ha))
loop_id = NPH_SNS;
else
loop_id = SIMPLE_NAME_SERVER;
ha->isp_ops.fabric_login(ha, loop_id, 0xff, 0xff,
0xfc, mb, BIT_1 | BIT_0);
if (mb[0] != MBS_COMMAND_COMPLETE) {
DEBUG2(qla_printk(KERN_INFO, ha,
"Failed SNS login: loop_id=%x mb[0]=%x mb[1]=%x "
"mb[2]=%x mb[6]=%x mb[7]=%x\n", loop_id,
mb[0], mb[1], mb[2], mb[6], mb[7]));
return (QLA_SUCCESS);
}
if (test_and_clear_bit(REGISTER_FC4_NEEDED, &ha->dpc_flags)) {
if (qla2x00_rft_id(ha)) {
/* EMPTY */
DEBUG2(printk("scsi(%ld): Register FC-4 "
"TYPE failed.\n", ha->host_no));
}
if (qla2x00_rff_id(ha)) {
/* EMPTY */
DEBUG2(printk("scsi(%ld): Register FC-4 "
"Features failed.\n", ha->host_no));
}
if (qla2x00_rnn_id(ha)) {
/* EMPTY */
DEBUG2(printk("scsi(%ld): Register Node Name "
"failed.\n", ha->host_no));
} else if (qla2x00_rsnn_nn(ha)) {
/* EMPTY */
DEBUG2(printk("scsi(%ld): Register Symbolic "
"Node Name failed.\n", ha->host_no));
}
}
rval = qla2x00_find_all_fabric_devs(ha, &new_fcports);
if (rval != QLA_SUCCESS)
break;
/*
* Logout all previous fabric devices marked lost, except
* tape devices.
*/
list_for_each_entry(fcport, &ha->fcports, list) {
if (test_bit(LOOP_RESYNC_NEEDED, &ha->dpc_flags))
break;
if ((fcport->flags & FCF_FABRIC_DEVICE) == 0)
continue;
if (atomic_read(&fcport->state) == FCS_DEVICE_LOST) {
qla2x00_mark_device_lost(ha, fcport,
ql2xplogiabsentdevice, 0);
if (fcport->loop_id != FC_NO_LOOP_ID &&
(fcport->flags & FCF_TAPE_PRESENT) == 0 &&
fcport->port_type != FCT_INITIATOR &&
fcport->port_type != FCT_BROADCAST) {
ha->isp_ops.fabric_logout(ha,
fcport->loop_id,
fcport->d_id.b.domain,
fcport->d_id.b.area,
fcport->d_id.b.al_pa);
fcport->loop_id = FC_NO_LOOP_ID;
}
}
}
/* Starting free loop ID. */
next_loopid = ha->min_external_loopid;
/*
* Scan through our port list and login entries that need to be
* logged in.
*/
list_for_each_entry(fcport, &ha->fcports, list) {
if (atomic_read(&ha->loop_down_timer) ||
test_bit(LOOP_RESYNC_NEEDED, &ha->dpc_flags))
break;
if ((fcport->flags & FCF_FABRIC_DEVICE) == 0 ||
(fcport->flags & FCF_LOGIN_NEEDED) == 0)
continue;
if (fcport->loop_id == FC_NO_LOOP_ID) {
fcport->loop_id = next_loopid;
rval = qla2x00_find_new_loop_id(ha, fcport);
if (rval != QLA_SUCCESS) {
/* Ran out of IDs to use */
break;
}
}
/* Login and update database */
qla2x00_fabric_dev_login(ha, fcport, &next_loopid);
}
/* Exit if out of loop IDs. */
if (rval != QLA_SUCCESS) {
break;
}
/*
* Login and add the new devices to our port list.
*/
list_for_each_entry_safe(fcport, fcptemp, &new_fcports, list) {
if (atomic_read(&ha->loop_down_timer) ||
test_bit(LOOP_RESYNC_NEEDED, &ha->dpc_flags))
break;
/* Find a new loop ID to use. */
fcport->loop_id = next_loopid;
rval = qla2x00_find_new_loop_id(ha, fcport);
if (rval != QLA_SUCCESS) {
/* Ran out of IDs to use */
break;
}
/* Remove device from the new list and add it to DB */
list_move_tail(&fcport->list, &ha->fcports);
/* Login and update database */
qla2x00_fabric_dev_login(ha, fcport, &next_loopid);
}
} while (0);
/* Free all new device structures not processed. */
list_for_each_entry_safe(fcport, fcptemp, &new_fcports, list) {
list_del(&fcport->list);
kfree(fcport);
}
if (rval) {
DEBUG2(printk("scsi(%ld): Configure fabric error exit: "
"rval=%d\n", ha->host_no, rval));
}
return (rval);
}
/*
* qla2x00_find_all_fabric_devs
*
* Input:
* ha = adapter block pointer.
* dev = database device entry pointer.
*
* Returns:
* 0 = success.
*
* Context:
* Kernel context.
*/
static int
qla2x00_find_all_fabric_devs(scsi_qla_host_t *ha, struct list_head *new_fcports)
{
int rval;
uint16_t loop_id;
fc_port_t *fcport, *new_fcport, *fcptemp;
int found;
sw_info_t *swl;
int swl_idx;
int first_dev, last_dev;
port_id_t wrap, nxt_d_id;
rval = QLA_SUCCESS;
/* Try GID_PT to get device list, else GAN. */
swl = kmalloc(sizeof(sw_info_t) * MAX_FIBRE_DEVICES, GFP_ATOMIC);
if (swl == NULL) {
/*EMPTY*/
DEBUG2(printk("scsi(%ld): GID_PT allocations failed, fallback "
"on GA_NXT\n", ha->host_no));
} else {
memset(swl, 0, sizeof(sw_info_t) * MAX_FIBRE_DEVICES);
if (qla2x00_gid_pt(ha, swl) != QLA_SUCCESS) {
kfree(swl);
swl = NULL;
} else if (qla2x00_gpn_id(ha, swl) != QLA_SUCCESS) {
kfree(swl);
swl = NULL;
} else if (qla2x00_gnn_id(ha, swl) != QLA_SUCCESS) {
kfree(swl);
swl = NULL;
} else if (qla2x00_gfpn_id(ha, swl) == QLA_SUCCESS) {
qla2x00_gpsc(ha, swl);
}
}
swl_idx = 0;
/* Allocate temporary fcport for any new fcports discovered. */
new_fcport = qla2x00_alloc_fcport(ha, GFP_KERNEL);
if (new_fcport == NULL) {
kfree(swl);
return (QLA_MEMORY_ALLOC_FAILED);
}
new_fcport->flags |= (FCF_FABRIC_DEVICE | FCF_LOGIN_NEEDED);
/* Set start port ID scan at adapter ID. */
first_dev = 1;
last_dev = 0;
/* Starting free loop ID. */
loop_id = ha->min_external_loopid;
for (; loop_id <= ha->last_loop_id; loop_id++) {
if (qla2x00_is_reserved_id(ha, loop_id))
continue;
if (atomic_read(&ha->loop_down_timer) || LOOP_TRANSITION(ha))
break;
if (swl != NULL) {
if (last_dev) {
wrap.b24 = new_fcport->d_id.b24;
} else {
new_fcport->d_id.b24 = swl[swl_idx].d_id.b24;
memcpy(new_fcport->node_name,
swl[swl_idx].node_name, WWN_SIZE);
memcpy(new_fcport->port_name,
swl[swl_idx].port_name, WWN_SIZE);
memcpy(new_fcport->fabric_port_name,
swl[swl_idx].fabric_port_name, WWN_SIZE);
new_fcport->fp_speed = swl[swl_idx].fp_speed;
if (swl[swl_idx].d_id.b.rsvd_1 != 0) {
last_dev = 1;
}
swl_idx++;
}
} else {
/* Send GA_NXT to the switch */
rval = qla2x00_ga_nxt(ha, new_fcport);
if (rval != QLA_SUCCESS) {
qla_printk(KERN_WARNING, ha,
"SNS scan failed -- assuming zero-entry "
"result...\n");
list_for_each_entry_safe(fcport, fcptemp,
new_fcports, list) {
list_del(&fcport->list);
kfree(fcport);
}
rval = QLA_SUCCESS;
break;
}
}
/* If wrap on switch device list, exit. */
if (first_dev) {
wrap.b24 = new_fcport->d_id.b24;
first_dev = 0;
} else if (new_fcport->d_id.b24 == wrap.b24) {
DEBUG2(printk("scsi(%ld): device wrap (%02x%02x%02x)\n",
ha->host_no, new_fcport->d_id.b.domain,
new_fcport->d_id.b.area, new_fcport->d_id.b.al_pa));
break;
}
/* Bypass if host adapter. */
if (new_fcport->d_id.b24 == ha->d_id.b24)
continue;
/* Bypass if same domain and area of adapter. */
if (((new_fcport->d_id.b24 & 0xffff00) ==
(ha->d_id.b24 & 0xffff00)) && ha->current_topology ==
ISP_CFG_FL)
continue;
/* Bypass reserved domain fields. */
if ((new_fcport->d_id.b.domain & 0xf0) == 0xf0)
continue;
/* Locate matching device in database. */
found = 0;
list_for_each_entry(fcport, &ha->fcports, list) {
if (memcmp(new_fcport->port_name, fcport->port_name,
WWN_SIZE))
continue;
found++;
/* Update port state. */
memcpy(fcport->fabric_port_name,
new_fcport->fabric_port_name, WWN_SIZE);
fcport->fp_speed = new_fcport->fp_speed;
/*
* If address the same and state FCS_ONLINE, nothing
* changed.
*/
if (fcport->d_id.b24 == new_fcport->d_id.b24 &&
atomic_read(&fcport->state) == FCS_ONLINE) {
break;
}
/*
* If device was not a fabric device before.
*/
if ((fcport->flags & FCF_FABRIC_DEVICE) == 0) {
fcport->d_id.b24 = new_fcport->d_id.b24;
fcport->loop_id = FC_NO_LOOP_ID;
fcport->flags |= (FCF_FABRIC_DEVICE |
FCF_LOGIN_NEEDED);
fcport->flags &= ~FCF_PERSISTENT_BOUND;
break;
}
/*
* Port ID changed or device was marked to be updated;
* Log it out if still logged in and mark it for
* relogin later.
*/
fcport->d_id.b24 = new_fcport->d_id.b24;
fcport->flags |= FCF_LOGIN_NEEDED;
if (fcport->loop_id != FC_NO_LOOP_ID &&
(fcport->flags & FCF_TAPE_PRESENT) == 0 &&
fcport->port_type != FCT_INITIATOR &&
fcport->port_type != FCT_BROADCAST) {
ha->isp_ops.fabric_logout(ha, fcport->loop_id,
fcport->d_id.b.domain, fcport->d_id.b.area,
fcport->d_id.b.al_pa);
fcport->loop_id = FC_NO_LOOP_ID;
}
break;
}
if (found)
continue;
/* If device was not in our fcports list, then add it. */
list_add_tail(&new_fcport->list, new_fcports);
/* Allocate a new replacement fcport. */
nxt_d_id.b24 = new_fcport->d_id.b24;
new_fcport = qla2x00_alloc_fcport(ha, GFP_KERNEL);
if (new_fcport == NULL) {
kfree(swl);
return (QLA_MEMORY_ALLOC_FAILED);
}
new_fcport->flags |= (FCF_FABRIC_DEVICE | FCF_LOGIN_NEEDED);
new_fcport->d_id.b24 = nxt_d_id.b24;
}
kfree(swl);
kfree(new_fcport);
if (!list_empty(new_fcports))
ha->device_flags |= DFLG_FABRIC_DEVICES;
return (rval);
}
/*
* qla2x00_find_new_loop_id
* Scan through our port list and find a new usable loop ID.
*
* Input:
* ha: adapter state pointer.
* dev: port structure pointer.
*
* Returns:
* qla2x00 local function return status code.
*
* Context:
* Kernel context.
*/
static int
qla2x00_find_new_loop_id(scsi_qla_host_t *ha, fc_port_t *dev)
{
int rval;
int found;
fc_port_t *fcport;
uint16_t first_loop_id;
rval = QLA_SUCCESS;
/* Save starting loop ID. */
first_loop_id = dev->loop_id;
for (;;) {
/* Skip loop ID if already used by adapter. */
if (dev->loop_id == ha->loop_id) {
dev->loop_id++;
}
/* Skip reserved loop IDs. */
while (qla2x00_is_reserved_id(ha, dev->loop_id)) {
dev->loop_id++;
}
/* Reset loop ID if passed the end. */
if (dev->loop_id > ha->last_loop_id) {
/* first loop ID. */
dev->loop_id = ha->min_external_loopid;
}
/* Check for loop ID being already in use. */
found = 0;
fcport = NULL;
list_for_each_entry(fcport, &ha->fcports, list) {
if (fcport->loop_id == dev->loop_id && fcport != dev) {
/* ID possibly in use */
found++;
break;
}
}
/* If not in use then it is free to use. */
if (!found) {
break;
}
/* ID in use. Try next value. */
dev->loop_id++;
/* If wrap around. No free ID to use. */
if (dev->loop_id == first_loop_id) {
dev->loop_id = FC_NO_LOOP_ID;
rval = QLA_FUNCTION_FAILED;
break;
}
}
return (rval);
}
/*
* qla2x00_device_resync
* Marks devices in the database that needs resynchronization.
*
* Input:
* ha = adapter block pointer.
*
* Context:
* Kernel context.
*/
static int
qla2x00_device_resync(scsi_qla_host_t *ha)
{
int rval;
uint32_t mask;
fc_port_t *fcport;
uint32_t rscn_entry;
uint8_t rscn_out_iter;
uint8_t format;
port_id_t d_id;
rval = QLA_RSCNS_HANDLED;
while (ha->rscn_out_ptr != ha->rscn_in_ptr ||
ha->flags.rscn_queue_overflow) {
rscn_entry = ha->rscn_queue[ha->rscn_out_ptr];
format = MSB(MSW(rscn_entry));
d_id.b.domain = LSB(MSW(rscn_entry));
d_id.b.area = MSB(LSW(rscn_entry));
d_id.b.al_pa = LSB(LSW(rscn_entry));
DEBUG(printk("scsi(%ld): RSCN queue entry[%d] = "
"[%02x/%02x%02x%02x].\n",
ha->host_no, ha->rscn_out_ptr, format, d_id.b.domain,
d_id.b.area, d_id.b.al_pa));
ha->rscn_out_ptr++;
if (ha->rscn_out_ptr == MAX_RSCN_COUNT)
ha->rscn_out_ptr = 0;
/* Skip duplicate entries. */
for (rscn_out_iter = ha->rscn_out_ptr;
!ha->flags.rscn_queue_overflow &&
rscn_out_iter != ha->rscn_in_ptr;
rscn_out_iter = (rscn_out_iter ==
(MAX_RSCN_COUNT - 1)) ? 0: rscn_out_iter + 1) {
if (rscn_entry != ha->rscn_queue[rscn_out_iter])
break;
DEBUG(printk("scsi(%ld): Skipping duplicate RSCN queue "
"entry found at [%d].\n", ha->host_no,
rscn_out_iter));
ha->rscn_out_ptr = rscn_out_iter;
}
/* Queue overflow, set switch default case. */
if (ha->flags.rscn_queue_overflow) {
DEBUG(printk("scsi(%ld): device_resync: rscn "
"overflow.\n", ha->host_no));
format = 3;
ha->flags.rscn_queue_overflow = 0;
}
switch (format) {
case 0:
mask = 0xffffff;
break;
case 1:
mask = 0xffff00;
break;
case 2:
mask = 0xff0000;
break;
default:
mask = 0x0;
d_id.b24 = 0;
ha->rscn_out_ptr = ha->rscn_in_ptr;
break;
}
rval = QLA_SUCCESS;
list_for_each_entry(fcport, &ha->fcports, list) {
if ((fcport->flags & FCF_FABRIC_DEVICE) == 0 ||
(fcport->d_id.b24 & mask) != d_id.b24 ||
fcport->port_type == FCT_BROADCAST)
continue;
if (atomic_read(&fcport->state) == FCS_ONLINE) {
if (format != 3 ||
fcport->port_type != FCT_INITIATOR) {
qla2x00_mark_device_lost(ha, fcport,
0, 0);
}
}
fcport->flags &= ~FCF_FARP_DONE;
}
}
return (rval);
}
/*
* qla2x00_fabric_dev_login
* Login fabric target device and update FC port database.
*
* Input:
* ha: adapter state pointer.
* fcport: port structure list pointer.
* next_loopid: contains value of a new loop ID that can be used
* by the next login attempt.
*
* Returns:
* qla2x00 local function return status code.
*
* Context:
* Kernel context.
*/
static int
qla2x00_fabric_dev_login(scsi_qla_host_t *ha, fc_port_t *fcport,
uint16_t *next_loopid)
{
int rval;
int retry;
uint8_t opts;
rval = QLA_SUCCESS;
retry = 0;
rval = qla2x00_fabric_login(ha, fcport, next_loopid);
if (rval == QLA_SUCCESS) {
/* Send an ADISC to tape devices.*/
opts = 0;
if (fcport->flags & FCF_TAPE_PRESENT)
opts |= BIT_1;
rval = qla2x00_get_port_database(ha, fcport, opts);
if (rval != QLA_SUCCESS) {
ha->isp_ops.fabric_logout(ha, fcport->loop_id,
fcport->d_id.b.domain, fcport->d_id.b.area,
fcport->d_id.b.al_pa);
qla2x00_mark_device_lost(ha, fcport, 1, 0);
} else {
qla2x00_update_fcport(ha, fcport);
}
}
return (rval);
}
/*
* qla2x00_fabric_login
* Issue fabric login command.
*
* Input:
* ha = adapter block pointer.
* device = pointer to FC device type structure.
*
* Returns:
* 0 - Login successfully
* 1 - Login failed
* 2 - Initiator device
* 3 - Fatal error
*/
int
qla2x00_fabric_login(scsi_qla_host_t *ha, fc_port_t *fcport,
uint16_t *next_loopid)
{
int rval;
int retry;
uint16_t tmp_loopid;
uint16_t mb[MAILBOX_REGISTER_COUNT];
retry = 0;
tmp_loopid = 0;
for (;;) {
DEBUG(printk("scsi(%ld): Trying Fabric Login w/loop id 0x%04x "
"for port %02x%02x%02x.\n",
ha->host_no, fcport->loop_id, fcport->d_id.b.domain,
fcport->d_id.b.area, fcport->d_id.b.al_pa));
/* Login fcport on switch. */
ha->isp_ops.fabric_login(ha, fcport->loop_id,
fcport->d_id.b.domain, fcport->d_id.b.area,
fcport->d_id.b.al_pa, mb, BIT_0);
if (mb[0] == MBS_PORT_ID_USED) {
/*
* Device has another loop ID. The firmware team
* recommends the driver perform an implicit login with
* the specified ID again. The ID we just used is save
* here so we return with an ID that can be tried by
* the next login.
*/
retry++;
tmp_loopid = fcport->loop_id;
fcport->loop_id = mb[1];
DEBUG(printk("Fabric Login: port in use - next "
"loop id=0x%04x, port Id=%02x%02x%02x.\n",
fcport->loop_id, fcport->d_id.b.domain,
fcport->d_id.b.area, fcport->d_id.b.al_pa));
} else if (mb[0] == MBS_COMMAND_COMPLETE) {
/*
* Login succeeded.
*/
if (retry) {
/* A retry occurred before. */
*next_loopid = tmp_loopid;
} else {
/*
* No retry occurred before. Just increment the
* ID value for next login.
*/
*next_loopid = (fcport->loop_id + 1);
}
if (mb[1] & BIT_0) {
fcport->port_type = FCT_INITIATOR;
} else {
fcport->port_type = FCT_TARGET;
if (mb[1] & BIT_1) {
fcport->flags |= FCF_TAPE_PRESENT;
}
}
if (mb[10] & BIT_0)
fcport->supported_classes |= FC_COS_CLASS2;
if (mb[10] & BIT_1)
fcport->supported_classes |= FC_COS_CLASS3;
rval = QLA_SUCCESS;
break;
} else if (mb[0] == MBS_LOOP_ID_USED) {
/*
* Loop ID already used, try next loop ID.
*/
fcport->loop_id++;
rval = qla2x00_find_new_loop_id(ha, fcport);
if (rval != QLA_SUCCESS) {
/* Ran out of loop IDs to use */
break;
}
} else if (mb[0] == MBS_COMMAND_ERROR) {
/*
* Firmware possibly timed out during login. If NO
* retries are left to do then the device is declared
* dead.
*/
*next_loopid = fcport->loop_id;
ha->isp_ops.fabric_logout(ha, fcport->loop_id,
fcport->d_id.b.domain, fcport->d_id.b.area,
fcport->d_id.b.al_pa);
qla2x00_mark_device_lost(ha, fcport, 1, 0);
rval = 1;
break;
} else {
/*
* unrecoverable / not handled error
*/
DEBUG2(printk("%s(%ld): failed=%x port_id=%02x%02x%02x "
"loop_id=%x jiffies=%lx.\n",
__func__, ha->host_no, mb[0],
fcport->d_id.b.domain, fcport->d_id.b.area,
fcport->d_id.b.al_pa, fcport->loop_id, jiffies));
*next_loopid = fcport->loop_id;
ha->isp_ops.fabric_logout(ha, fcport->loop_id,
fcport->d_id.b.domain, fcport->d_id.b.area,
fcport->d_id.b.al_pa);
fcport->loop_id = FC_NO_LOOP_ID;
fcport->login_retry = 0;
rval = 3;
break;
}
}
return (rval);
}
/*
* qla2x00_local_device_login
* Issue local device login command.
*
* Input:
* ha = adapter block pointer.
* loop_id = loop id of device to login to.
*
* Returns (Where's the #define!!!!):
* 0 - Login successfully
* 1 - Login failed
* 3 - Fatal error
*/
int
qla2x00_local_device_login(scsi_qla_host_t *ha, fc_port_t *fcport)
{
int rval;
uint16_t mb[MAILBOX_REGISTER_COUNT];
memset(mb, 0, sizeof(mb));
rval = qla2x00_login_local_device(ha, fcport, mb, BIT_0);
if (rval == QLA_SUCCESS) {
/* Interrogate mailbox registers for any errors */
if (mb[0] == MBS_COMMAND_ERROR)
rval = 1;
else if (mb[0] == MBS_COMMAND_PARAMETER_ERROR)
/* device not in PCB table */
rval = 3;
}
return (rval);
}
/*
* qla2x00_loop_resync
* Resync with fibre channel devices.
*
* Input:
* ha = adapter block pointer.
*
* Returns:
* 0 = success
*/
int
qla2x00_loop_resync(scsi_qla_host_t *ha)
{
int rval;
uint32_t wait_time;
rval = QLA_SUCCESS;
atomic_set(&ha->loop_state, LOOP_UPDATE);
clear_bit(ISP_ABORT_RETRY, &ha->dpc_flags);
if (ha->flags.online) {
if (!(rval = qla2x00_fw_ready(ha))) {
/* Wait at most MAX_TARGET RSCNs for a stable link. */
wait_time = 256;
do {
atomic_set(&ha->loop_state, LOOP_UPDATE);
/* Issue a marker after FW becomes ready. */
qla2x00_marker(ha, 0, 0, MK_SYNC_ALL);
ha->marker_needed = 0;
/* Remap devices on Loop. */
clear_bit(LOOP_RESYNC_NEEDED, &ha->dpc_flags);
qla2x00_configure_loop(ha);
wait_time--;
} while (!atomic_read(&ha->loop_down_timer) &&
!(test_bit(ISP_ABORT_NEEDED, &ha->dpc_flags)) &&
wait_time &&
(test_bit(LOOP_RESYNC_NEEDED, &ha->dpc_flags)));
}
}
if (test_bit(ISP_ABORT_NEEDED, &ha->dpc_flags)) {
return (QLA_FUNCTION_FAILED);
}
if (rval) {
DEBUG2_3(printk("%s(): **** FAILED ****\n", __func__));
}
return (rval);
}
void
qla2x00_rescan_fcports(scsi_qla_host_t *ha)
{
int rescan_done;
fc_port_t *fcport;
rescan_done = 0;
list_for_each_entry(fcport, &ha->fcports, list) {
if ((fcport->flags & FCF_RESCAN_NEEDED) == 0)
continue;
qla2x00_update_fcport(ha, fcport);
fcport->flags &= ~FCF_RESCAN_NEEDED;
rescan_done = 1;
}
qla2x00_probe_for_all_luns(ha);
}
void
qla2x00_update_fcports(scsi_qla_host_t *ha)
{
fc_port_t *fcport;
/* Go with deferred removal of rport references. */
list_for_each_entry(fcport, &ha->fcports, list)
if (fcport->drport)
qla2x00_rport_del(fcport);
}
/*
* qla2x00_abort_isp
* Resets ISP and aborts all outstanding commands.
*
* Input:
* ha = adapter block pointer.
*
* Returns:
* 0 = success
*/
int
qla2x00_abort_isp(scsi_qla_host_t *ha)
{
int rval;
unsigned long flags = 0;
uint16_t cnt;
srb_t *sp;
uint8_t status = 0;
if (ha->flags.online) {
ha->flags.online = 0;
clear_bit(ISP_ABORT_NEEDED, &ha->dpc_flags);
qla_printk(KERN_INFO, ha,
"Performing ISP error recovery - ha= %p.\n", ha);
ha->isp_ops.reset_chip(ha);
atomic_set(&ha->loop_down_timer, LOOP_DOWN_TIME);
if (atomic_read(&ha->loop_state) != LOOP_DOWN) {
atomic_set(&ha->loop_state, LOOP_DOWN);
qla2x00_mark_all_devices_lost(ha, 0);
} else {
if (!atomic_read(&ha->loop_down_timer))
atomic_set(&ha->loop_down_timer,
LOOP_DOWN_TIME);
}
spin_lock_irqsave(&ha->hardware_lock, flags);
/* Requeue all commands in outstanding command list. */
for (cnt = 1; cnt < MAX_OUTSTANDING_COMMANDS; cnt++) {
sp = ha->outstanding_cmds[cnt];
if (sp) {
ha->outstanding_cmds[cnt] = NULL;
sp->flags = 0;
sp->cmd->result = DID_RESET << 16;
sp->cmd->host_scribble = (unsigned char *)NULL;
qla2x00_sp_compl(ha, sp);
}
}
spin_unlock_irqrestore(&ha->hardware_lock, flags);
ha->isp_ops.nvram_config(ha);
if (!qla2x00_restart_isp(ha)) {
clear_bit(RESET_MARKER_NEEDED, &ha->dpc_flags);
if (!atomic_read(&ha->loop_down_timer)) {
/*
* Issue marker command only when we are going
* to start the I/O .
*/
ha->marker_needed = 1;
}
ha->flags.online = 1;
ha->isp_ops.enable_intrs(ha);
ha->isp_abort_cnt = 0;
clear_bit(ISP_ABORT_RETRY, &ha->dpc_flags);
if (ha->eft) {
rval = qla2x00_trace_control(ha, TC_ENABLE,
ha->eft_dma, EFT_NUM_BUFFERS);
if (rval) {
qla_printk(KERN_WARNING, ha,
"Unable to reinitialize EFT "
"(%d).\n", rval);
}
}
} else { /* failed the ISP abort */
ha->flags.online = 1;
if (test_bit(ISP_ABORT_RETRY, &ha->dpc_flags)) {
if (ha->isp_abort_cnt == 0) {
qla_printk(KERN_WARNING, ha,
"ISP error recovery failed - "
"board disabled\n");
/*
* The next call disables the board
* completely.
*/
ha->isp_ops.reset_adapter(ha);
ha->flags.online = 0;
clear_bit(ISP_ABORT_RETRY,
&ha->dpc_flags);
status = 0;
} else { /* schedule another ISP abort */
ha->isp_abort_cnt--;
DEBUG(printk("qla%ld: ISP abort - "
"retry remaining %d\n",
ha->host_no, ha->isp_abort_cnt));
status = 1;
}
} else {
ha->isp_abort_cnt = MAX_RETRIES_OF_ISP_ABORT;
DEBUG(printk("qla2x00(%ld): ISP error recovery "
"- retrying (%d) more times\n",
ha->host_no, ha->isp_abort_cnt));
set_bit(ISP_ABORT_RETRY, &ha->dpc_flags);
status = 1;
}
}
}
if (status) {
qla_printk(KERN_INFO, ha,
"qla2x00_abort_isp: **** FAILED ****\n");
} else {
DEBUG(printk(KERN_INFO
"qla2x00_abort_isp(%ld): exiting.\n",
ha->host_no));
}
return(status);
}
/*
* qla2x00_restart_isp
* restarts the ISP after a reset
*
* Input:
* ha = adapter block pointer.
*
* Returns:
* 0 = success
*/
static int
qla2x00_restart_isp(scsi_qla_host_t *ha)
{
uint8_t status = 0;
struct device_reg_2xxx __iomem *reg = &ha->iobase->isp;
unsigned long flags = 0;
uint32_t wait_time;
/* If firmware needs to be loaded */
if (qla2x00_isp_firmware(ha)) {
ha->flags.online = 0;
if (!(status = ha->isp_ops.chip_diag(ha))) {
if (IS_QLA2100(ha) || IS_QLA2200(ha)) {
status = qla2x00_setup_chip(ha);
goto done;
}
spin_lock_irqsave(&ha->hardware_lock, flags);
if (!IS_QLA24XX(ha) && !IS_QLA54XX(ha)) {
/*
* Disable SRAM, Instruction RAM and GP RAM
* parity.
*/
WRT_REG_WORD(&reg->hccr,
(HCCR_ENABLE_PARITY + 0x0));
RD_REG_WORD(&reg->hccr);
}
spin_unlock_irqrestore(&ha->hardware_lock, flags);
status = qla2x00_setup_chip(ha);
spin_lock_irqsave(&ha->hardware_lock, flags);
if (!IS_QLA24XX(ha) && !IS_QLA54XX(ha)) {
/* Enable proper parity */
if (IS_QLA2300(ha))
/* SRAM parity */
WRT_REG_WORD(&reg->hccr,
(HCCR_ENABLE_PARITY + 0x1));
else
/*
* SRAM, Instruction RAM and GP RAM
* parity.
*/
WRT_REG_WORD(&reg->hccr,
(HCCR_ENABLE_PARITY + 0x7));
RD_REG_WORD(&reg->hccr);
}
spin_unlock_irqrestore(&ha->hardware_lock, flags);
}
}
done:
if (!status && !(status = qla2x00_init_rings(ha))) {
clear_bit(RESET_MARKER_NEEDED, &ha->dpc_flags);
if (!(status = qla2x00_fw_ready(ha))) {
DEBUG(printk("%s(): Start configure loop, "
"status = %d\n", __func__, status));
/* Issue a marker after FW becomes ready. */
qla2x00_marker(ha, 0, 0, MK_SYNC_ALL);
ha->flags.online = 1;
/* Wait at most MAX_TARGET RSCNs for a stable link. */
wait_time = 256;
do {
clear_bit(LOOP_RESYNC_NEEDED, &ha->dpc_flags);
qla2x00_configure_loop(ha);
wait_time--;
} while (!atomic_read(&ha->loop_down_timer) &&
!(test_bit(ISP_ABORT_NEEDED, &ha->dpc_flags)) &&
wait_time &&
(test_bit(LOOP_RESYNC_NEEDED, &ha->dpc_flags)));
}
/* if no cable then assume it's good */
if ((ha->device_flags & DFLG_NO_CABLE))
status = 0;
DEBUG(printk("%s(): Configure loop done, status = 0x%x\n",
__func__,
status));
}
return (status);
}
/*
* qla2x00_reset_adapter
* Reset adapter.
*
* Input:
* ha = adapter block pointer.
*/
void
qla2x00_reset_adapter(scsi_qla_host_t *ha)
{
unsigned long flags = 0;
struct device_reg_2xxx __iomem *reg = &ha->iobase->isp;
ha->flags.online = 0;
ha->isp_ops.disable_intrs(ha);
spin_lock_irqsave(&ha->hardware_lock, flags);
WRT_REG_WORD(&reg->hccr, HCCR_RESET_RISC);
RD_REG_WORD(&reg->hccr); /* PCI Posting. */
WRT_REG_WORD(&reg->hccr, HCCR_RELEASE_RISC);
RD_REG_WORD(&reg->hccr); /* PCI Posting. */
spin_unlock_irqrestore(&ha->hardware_lock, flags);
}
void
qla24xx_reset_adapter(scsi_qla_host_t *ha)
{
unsigned long flags = 0;
struct device_reg_24xx __iomem *reg = &ha->iobase->isp24;
ha->flags.online = 0;
ha->isp_ops.disable_intrs(ha);
spin_lock_irqsave(&ha->hardware_lock, flags);
WRT_REG_DWORD(&reg->hccr, HCCRX_SET_RISC_RESET);
RD_REG_DWORD(&reg->hccr);
WRT_REG_DWORD(&reg->hccr, HCCRX_REL_RISC_PAUSE);
RD_REG_DWORD(&reg->hccr);
spin_unlock_irqrestore(&ha->hardware_lock, flags);
}
int
qla24xx_nvram_config(scsi_qla_host_t *ha)
{
int rval;
struct init_cb_24xx *icb;
struct nvram_24xx *nv;
uint32_t *dptr;
uint8_t *dptr1, *dptr2;
uint32_t chksum;
uint16_t cnt;
rval = QLA_SUCCESS;
icb = (struct init_cb_24xx *)ha->init_cb;
nv = (struct nvram_24xx *)ha->request_ring;
/* Determine NVRAM starting address. */
ha->nvram_size = sizeof(struct nvram_24xx);
ha->nvram_base = FA_NVRAM_FUNC0_ADDR;
ha->vpd_size = FA_NVRAM_VPD_SIZE;
ha->vpd_base = FA_NVRAM_VPD0_ADDR;
if (PCI_FUNC(ha->pdev->devfn)) {
ha->nvram_base = FA_NVRAM_FUNC1_ADDR;
ha->vpd_base = FA_NVRAM_VPD1_ADDR;
}
/* Get NVRAM data and calculate checksum. */
dptr = (uint32_t *)nv;
ha->isp_ops.read_nvram(ha, (uint8_t *)dptr, ha->nvram_base,
ha->nvram_size);
for (cnt = 0, chksum = 0; cnt < ha->nvram_size >> 2; cnt++)
chksum += le32_to_cpu(*dptr++);
DEBUG5(printk("scsi(%ld): Contents of NVRAM\n", ha->host_no));
DEBUG5(qla2x00_dump_buffer((uint8_t *)ha->request_ring,
ha->nvram_size));
/* Bad NVRAM data, set defaults parameters. */
if (chksum || nv->id[0] != 'I' || nv->id[1] != 'S' || nv->id[2] != 'P'
|| nv->id[3] != ' ' ||
nv->nvram_version < __constant_cpu_to_le16(ICB_VERSION)) {
/* Reset NVRAM data. */
qla_printk(KERN_WARNING, ha, "Inconsistent NVRAM detected: "
"checksum=0x%x id=%c version=0x%x.\n", chksum, nv->id[0],
le16_to_cpu(nv->nvram_version));
qla_printk(KERN_WARNING, ha, "Falling back to functioning (yet "
"invalid -- WWPN) defaults.\n");
/*
* Set default initialization control block.
*/
memset(nv, 0, ha->nvram_size);
nv->nvram_version = __constant_cpu_to_le16(ICB_VERSION);
nv->version = __constant_cpu_to_le16(ICB_VERSION);
nv->frame_payload_size = __constant_cpu_to_le16(2048);
nv->execution_throttle = __constant_cpu_to_le16(0xFFFF);
nv->exchange_count = __constant_cpu_to_le16(0);
nv->hard_address = __constant_cpu_to_le16(124);
nv->port_name[0] = 0x21;
nv->port_name[1] = 0x00 + PCI_FUNC(ha->pdev->devfn);
nv->port_name[2] = 0x00;
nv->port_name[3] = 0xe0;
nv->port_name[4] = 0x8b;
nv->port_name[5] = 0x1c;
nv->port_name[6] = 0x55;
nv->port_name[7] = 0x86;
nv->node_name[0] = 0x20;
nv->node_name[1] = 0x00;
nv->node_name[2] = 0x00;
nv->node_name[3] = 0xe0;
nv->node_name[4] = 0x8b;
nv->node_name[5] = 0x1c;
nv->node_name[6] = 0x55;
nv->node_name[7] = 0x86;
nv->login_retry_count = __constant_cpu_to_le16(8);
nv->interrupt_delay_timer = __constant_cpu_to_le16(0);
nv->login_timeout = __constant_cpu_to_le16(0);
nv->firmware_options_1 =
__constant_cpu_to_le32(BIT_14|BIT_13|BIT_2|BIT_1);
nv->firmware_options_2 = __constant_cpu_to_le32(2 << 4);
nv->firmware_options_2 |= __constant_cpu_to_le32(BIT_12);
nv->firmware_options_3 = __constant_cpu_to_le32(2 << 13);
nv->host_p = __constant_cpu_to_le32(BIT_11|BIT_10);
nv->efi_parameters = __constant_cpu_to_le32(0);
nv->reset_delay = 5;
nv->max_luns_per_target = __constant_cpu_to_le16(128);
nv->port_down_retry_count = __constant_cpu_to_le16(30);
nv->link_down_timeout = __constant_cpu_to_le16(30);
rval = 1;
}
/* Reset Initialization control block */
memset(icb, 0, sizeof(struct init_cb_24xx));
/* Copy 1st segment. */
dptr1 = (uint8_t *)icb;
dptr2 = (uint8_t *)&nv->version;
cnt = (uint8_t *)&icb->response_q_inpointer - (uint8_t *)&icb->version;
while (cnt--)
*dptr1++ = *dptr2++;
icb->login_retry_count = nv->login_retry_count;
icb->link_down_on_nos = nv->link_down_on_nos;
/* Copy 2nd segment. */
dptr1 = (uint8_t *)&icb->interrupt_delay_timer;
dptr2 = (uint8_t *)&nv->interrupt_delay_timer;
cnt = (uint8_t *)&icb->reserved_3 -
(uint8_t *)&icb->interrupt_delay_timer;
while (cnt--)
*dptr1++ = *dptr2++;
/*
* Setup driver NVRAM options.
*/
if (memcmp(nv->model_name, BINZERO, sizeof(nv->model_name)) != 0) {
char *st, *en;
uint16_t index;
strncpy(ha->model_number, nv->model_name,
sizeof(nv->model_name));
st = en = ha->model_number;
en += sizeof(nv->model_name) - 1;
while (en > st) {
if (*en != 0x20 && *en != 0x00)
break;
*en-- = '\0';
}
index = (ha->pdev->subsystem_device & 0xff);
if (index < QLA_MODEL_NAMES)
ha->model_desc = qla2x00_model_name[index * 2 + 1];
} else
strcpy(ha->model_number, "QLA2462");
/* Use alternate WWN? */
if (nv->host_p & __constant_cpu_to_le32(BIT_15)) {
memcpy(icb->node_name, nv->alternate_node_name, WWN_SIZE);
memcpy(icb->port_name, nv->alternate_port_name, WWN_SIZE);
}
/* Prepare nodename */
if ((icb->firmware_options_1 & __constant_cpu_to_le32(BIT_14)) == 0) {
/*
* Firmware will apply the following mask if the nodename was
* not provided.
*/
memcpy(icb->node_name, icb->port_name, WWN_SIZE);
icb->node_name[0] &= 0xF0;
}
/* Set host adapter parameters. */
ha->flags.disable_risc_code_load = 0;
ha->flags.enable_lip_reset = 1;
ha->flags.enable_lip_full_login = 1;
ha->flags.enable_target_reset = 1;
ha->flags.enable_led_scheme = 0;
ha->flags.disable_serdes = le32_to_cpu(nv->host_p) & BIT_5 ? 1: 0;
ha->operating_mode = (le32_to_cpu(icb->firmware_options_2) &
(BIT_6 | BIT_5 | BIT_4)) >> 4;
memcpy(ha->fw_seriallink_options24, nv->seriallink_options,
sizeof(ha->fw_seriallink_options24));
/* save HBA serial number */
ha->serial0 = icb->port_name[5];
ha->serial1 = icb->port_name[6];
ha->serial2 = icb->port_name[7];
ha->node_name = icb->node_name;
ha->port_name = icb->port_name;
icb->execution_throttle = __constant_cpu_to_le16(0xFFFF);
ha->retry_count = le16_to_cpu(nv->login_retry_count);
/* Set minimum login_timeout to 4 seconds. */
if (le16_to_cpu(nv->login_timeout) < ql2xlogintimeout)
nv->login_timeout = cpu_to_le16(ql2xlogintimeout);
if (le16_to_cpu(nv->login_timeout) < 4)
nv->login_timeout = __constant_cpu_to_le16(4);
ha->login_timeout = le16_to_cpu(nv->login_timeout);
icb->login_timeout = cpu_to_le16(nv->login_timeout);
/* Set minimum RATOV to 200 tenths of a second. */
ha->r_a_tov = 200;
ha->loop_reset_delay = nv->reset_delay;
/* Link Down Timeout = 0:
*
* When Port Down timer expires we will start returning
* I/O's to OS with "DID_NO_CONNECT".
*
* Link Down Timeout != 0:
*
* The driver waits for the link to come up after link down
* before returning I/Os to OS with "DID_NO_CONNECT".
*/
if (le16_to_cpu(nv->link_down_timeout) == 0) {
ha->loop_down_abort_time =
(LOOP_DOWN_TIME - LOOP_DOWN_TIMEOUT);
} else {
ha->link_down_timeout = le16_to_cpu(nv->link_down_timeout);
ha->loop_down_abort_time =
(LOOP_DOWN_TIME - ha->link_down_timeout);
}
/* Need enough time to try and get the port back. */
ha->port_down_retry_count = le16_to_cpu(nv->port_down_retry_count);
if (qlport_down_retry)
ha->port_down_retry_count = qlport_down_retry;
/* Set login_retry_count */
ha->login_retry_count = le16_to_cpu(nv->login_retry_count);
if (ha->port_down_retry_count ==
le16_to_cpu(nv->port_down_retry_count) &&
ha->port_down_retry_count > 3)
ha->login_retry_count = ha->port_down_retry_count;
else if (ha->port_down_retry_count > (int)ha->login_retry_count)
ha->login_retry_count = ha->port_down_retry_count;
if (ql2xloginretrycount)
ha->login_retry_count = ql2xloginretrycount;
/* Enable ZIO. */
if (!ha->flags.init_done) {
ha->zio_mode = le32_to_cpu(icb->firmware_options_2) &
(BIT_3 | BIT_2 | BIT_1 | BIT_0);
ha->zio_timer = le16_to_cpu(icb->interrupt_delay_timer) ?
le16_to_cpu(icb->interrupt_delay_timer): 2;
}
icb->firmware_options_2 &= __constant_cpu_to_le32(
~(BIT_3 | BIT_2 | BIT_1 | BIT_0));
ha->flags.process_response_queue = 0;
if (ha->zio_mode != QLA_ZIO_DISABLED) {
ha->zio_mode = QLA_ZIO_MODE_6;
DEBUG2(printk("scsi(%ld): ZIO mode %d enabled; timer delay "
"(%d us).\n", ha->host_no, ha->zio_mode,
ha->zio_timer * 100));
qla_printk(KERN_INFO, ha,
"ZIO mode %d enabled; timer delay (%d us).\n",
ha->zio_mode, ha->zio_timer * 100);
icb->firmware_options_2 |= cpu_to_le32(
(uint32_t)ha->zio_mode);
icb->interrupt_delay_timer = cpu_to_le16(ha->zio_timer);
ha->flags.process_response_queue = 1;
}
if (rval) {
DEBUG2_3(printk(KERN_WARNING
"scsi(%ld): NVRAM configuration failed!\n", ha->host_no));
}
return (rval);
}
static int
qla24xx_load_risc_flash(scsi_qla_host_t *ha, uint32_t *srisc_addr)
{
int rval;
int segments, fragment;
uint32_t faddr;
uint32_t *dcode, dlen;
uint32_t risc_addr;
uint32_t risc_size;
uint32_t i;
rval = QLA_SUCCESS;
segments = FA_RISC_CODE_SEGMENTS;
faddr = FA_RISC_CODE_ADDR;
dcode = (uint32_t *)ha->request_ring;
*srisc_addr = 0;
/* Validate firmware image by checking version. */
qla24xx_read_flash_data(ha, dcode, faddr + 4, 4);
for (i = 0; i < 4; i++)
dcode[i] = be32_to_cpu(dcode[i]);
if ((dcode[0] == 0xffffffff && dcode[1] == 0xffffffff &&
dcode[2] == 0xffffffff && dcode[3] == 0xffffffff) ||
(dcode[0] == 0 && dcode[1] == 0 && dcode[2] == 0 &&
dcode[3] == 0)) {
qla_printk(KERN_WARNING, ha,
"Unable to verify integrity of flash firmware image!\n");
qla_printk(KERN_WARNING, ha,
"Firmware data: %08x %08x %08x %08x!\n", dcode[0],
dcode[1], dcode[2], dcode[3]);
return QLA_FUNCTION_FAILED;
}
while (segments && rval == QLA_SUCCESS) {
/* Read segment's load information. */
qla24xx_read_flash_data(ha, dcode, faddr, 4);
risc_addr = be32_to_cpu(dcode[2]);
*srisc_addr = *srisc_addr == 0 ? risc_addr : *srisc_addr;
risc_size = be32_to_cpu(dcode[3]);
fragment = 0;
while (risc_size > 0 && rval == QLA_SUCCESS) {
dlen = (uint32_t)(ha->fw_transfer_size >> 2);
if (dlen > risc_size)
dlen = risc_size;
DEBUG7(printk("scsi(%ld): Loading risc segment@ risc "
"addr %x, number of dwords 0x%x, offset 0x%x.\n",
ha->host_no, risc_addr, dlen, faddr));
qla24xx_read_flash_data(ha, dcode, faddr, dlen);
for (i = 0; i < dlen; i++)
dcode[i] = swab32(dcode[i]);
rval = qla2x00_load_ram(ha, ha->request_dma, risc_addr,
dlen);
if (rval) {
DEBUG(printk("scsi(%ld):[ERROR] Failed to load "
"segment %d of firmware\n", ha->host_no,
fragment));
qla_printk(KERN_WARNING, ha,
"[ERROR] Failed to load segment %d of "
"firmware\n", fragment);
break;
}
faddr += dlen;
risc_addr += dlen;
risc_size -= dlen;
fragment++;
}
/* Next segment. */
segments--;
}
return rval;
}
#define QLA_FW_URL "ftp://ftp.qlogic.com/outgoing/linux/firmware/"
int
qla2x00_load_risc(scsi_qla_host_t *ha, uint32_t *srisc_addr)
{
int rval;
int i, fragment;
uint16_t *wcode, *fwcode;
uint32_t risc_addr, risc_size, fwclen, wlen, *seg;
struct fw_blob *blob;
/* Load firmware blob. */
blob = qla2x00_request_firmware(ha);
if (!blob) {
qla_printk(KERN_ERR, ha, "Firmware image unavailable.\n");
qla_printk(KERN_ERR, ha, "Firmware images can be retrieved "
"from: " QLA_FW_URL ".\n");
return QLA_FUNCTION_FAILED;
}
rval = QLA_SUCCESS;
wcode = (uint16_t *)ha->request_ring;
*srisc_addr = 0;
fwcode = (uint16_t *)blob->fw->data;
fwclen = 0;
/* Validate firmware image by checking version. */
if (blob->fw->size < 8 * sizeof(uint16_t)) {
qla_printk(KERN_WARNING, ha,
"Unable to verify integrity of firmware image (%Zd)!\n",
blob->fw->size);
goto fail_fw_integrity;
}
for (i = 0; i < 4; i++)
wcode[i] = be16_to_cpu(fwcode[i + 4]);
if ((wcode[0] == 0xffff && wcode[1] == 0xffff && wcode[2] == 0xffff &&
wcode[3] == 0xffff) || (wcode[0] == 0 && wcode[1] == 0 &&
wcode[2] == 0 && wcode[3] == 0)) {
qla_printk(KERN_WARNING, ha,
"Unable to verify integrity of firmware image!\n");
qla_printk(KERN_WARNING, ha,
"Firmware data: %04x %04x %04x %04x!\n", wcode[0],
wcode[1], wcode[2], wcode[3]);
goto fail_fw_integrity;
}
seg = blob->segs;
while (*seg && rval == QLA_SUCCESS) {
risc_addr = *seg;
*srisc_addr = *srisc_addr == 0 ? *seg : *srisc_addr;
risc_size = be16_to_cpu(fwcode[3]);
/* Validate firmware image size. */
fwclen += risc_size * sizeof(uint16_t);
if (blob->fw->size < fwclen) {
qla_printk(KERN_WARNING, ha,
"Unable to verify integrity of firmware image "
"(%Zd)!\n", blob->fw->size);
goto fail_fw_integrity;
}
fragment = 0;
while (risc_size > 0 && rval == QLA_SUCCESS) {
wlen = (uint16_t)(ha->fw_transfer_size >> 1);
if (wlen > risc_size)
wlen = risc_size;
DEBUG7(printk("scsi(%ld): Loading risc segment@ risc "
"addr %x, number of words 0x%x.\n", ha->host_no,
risc_addr, wlen));
for (i = 0; i < wlen; i++)
wcode[i] = swab16(fwcode[i]);
rval = qla2x00_load_ram(ha, ha->request_dma, risc_addr,
wlen);
if (rval) {
DEBUG(printk("scsi(%ld):[ERROR] Failed to load "
"segment %d of firmware\n", ha->host_no,
fragment));
qla_printk(KERN_WARNING, ha,
"[ERROR] Failed to load segment %d of "
"firmware\n", fragment);
break;
}
fwcode += wlen;
risc_addr += wlen;
risc_size -= wlen;
fragment++;
}
/* Next segment. */
seg++;
}
return rval;
fail_fw_integrity:
return QLA_FUNCTION_FAILED;
}
int
qla24xx_load_risc(scsi_qla_host_t *ha, uint32_t *srisc_addr)
{
int rval;
int segments, fragment;
uint32_t *dcode, dlen;
uint32_t risc_addr;
uint32_t risc_size;
uint32_t i;
struct fw_blob *blob;
uint32_t *fwcode, fwclen;
/* Load firmware blob. */
blob = qla2x00_request_firmware(ha);
if (!blob) {
qla_printk(KERN_ERR, ha, "Firmware image unavailable.\n");
qla_printk(KERN_ERR, ha, "Firmware images can be retrieved "
"from: " QLA_FW_URL ".\n");
/* Try to load RISC code from flash. */
qla_printk(KERN_ERR, ha, "Attempting to load (potentially "
"outdated) firmware from flash.\n");
return qla24xx_load_risc_flash(ha, srisc_addr);
}
rval = QLA_SUCCESS;
segments = FA_RISC_CODE_SEGMENTS;
dcode = (uint32_t *)ha->request_ring;
*srisc_addr = 0;
fwcode = (uint32_t *)blob->fw->data;
fwclen = 0;
/* Validate firmware image by checking version. */
if (blob->fw->size < 8 * sizeof(uint32_t)) {
qla_printk(KERN_WARNING, ha,
"Unable to verify integrity of firmware image (%Zd)!\n",
blob->fw->size);
goto fail_fw_integrity;
}
for (i = 0; i < 4; i++)
dcode[i] = be32_to_cpu(fwcode[i + 4]);
if ((dcode[0] == 0xffffffff && dcode[1] == 0xffffffff &&
dcode[2] == 0xffffffff && dcode[3] == 0xffffffff) ||
(dcode[0] == 0 && dcode[1] == 0 && dcode[2] == 0 &&
dcode[3] == 0)) {
qla_printk(KERN_WARNING, ha,
"Unable to verify integrity of firmware image!\n");
qla_printk(KERN_WARNING, ha,
"Firmware data: %08x %08x %08x %08x!\n", dcode[0],
dcode[1], dcode[2], dcode[3]);
goto fail_fw_integrity;
}
while (segments && rval == QLA_SUCCESS) {
risc_addr = be32_to_cpu(fwcode[2]);
*srisc_addr = *srisc_addr == 0 ? risc_addr : *srisc_addr;
risc_size = be32_to_cpu(fwcode[3]);
/* Validate firmware image size. */
fwclen += risc_size * sizeof(uint32_t);
if (blob->fw->size < fwclen) {
qla_printk(KERN_WARNING, ha,
"Unable to verify integrity of firmware image "
"(%Zd)!\n", blob->fw->size);
goto fail_fw_integrity;
}
fragment = 0;
while (risc_size > 0 && rval == QLA_SUCCESS) {
dlen = (uint32_t)(ha->fw_transfer_size >> 2);
if (dlen > risc_size)
dlen = risc_size;
DEBUG7(printk("scsi(%ld): Loading risc segment@ risc "
"addr %x, number of dwords 0x%x.\n", ha->host_no,
risc_addr, dlen));
for (i = 0; i < dlen; i++)
dcode[i] = swab32(fwcode[i]);
rval = qla2x00_load_ram(ha, ha->request_dma, risc_addr,
dlen);
if (rval) {
DEBUG(printk("scsi(%ld):[ERROR] Failed to load "
"segment %d of firmware\n", ha->host_no,
fragment));
qla_printk(KERN_WARNING, ha,
"[ERROR] Failed to load segment %d of "
"firmware\n", fragment);
break;
}
fwcode += dlen;
risc_addr += dlen;
risc_size -= dlen;
fragment++;
}
/* Next segment. */
segments--;
}
return rval;
fail_fw_integrity:
return QLA_FUNCTION_FAILED;
}
void
qla2x00_try_to_stop_firmware(scsi_qla_host_t *ha)
{
int ret, retries;
if (!IS_QLA24XX(ha) && !IS_QLA54XX(ha))
return;
ret = qla2x00_stop_firmware(ha);
for (retries = 5; ret != QLA_SUCCESS && retries ; retries--) {
qla2x00_reset_chip(ha);
if (qla2x00_chip_diag(ha) != QLA_SUCCESS)
continue;
if (qla2x00_setup_chip(ha) != QLA_SUCCESS)
continue;
qla_printk(KERN_INFO, ha,
"Attempting retry of stop-firmware command...\n");
ret = qla2x00_stop_firmware(ha);
}
}