linux/drivers/scsi/bnx2i/bnx2i_init.c
Eddie Wai b5cf6b63f7 [SCSI] bnx2i: Added the use of kthreads to handle SCSI cmd completion
This patch breaks the SCSI cmd completion into two parts:
1. The bh will allocate and queued work to the cmd specific CPU IO
completion kthread.  The CPU for the cmd is from the sc->request->cpu.

2. The CPU specific IO completion kthread will call the scsi_cmd_resp
routine to do the actual cmd completion.

In the normal case, these IO completion kthreads should complete before
the blk IO times out at 60s.  However, in the case when these kthreads
are blocked for whatever reason and exceeded the timeout, the call
to conn_destroy will have to iterate and exhaust all related work in the
percpu work list for all online CPUs.  This will guarantee the protection
of the work->session and conn pointers before they get freed.

Also modified the event coalescing formula to have at least the
event_coal_min outstanding cmds in the pipeline so the SCSI producer
would not get underrun.

Also changed the following SCSI parameters:
- can_queue from 1024 to 2048
- cmds_per_lun from 24 to 128

Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Acked-by: Benjamin Li <benli@broadcom.com>
Acked-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
2011-06-29 16:48:23 -05:00

555 lines
15 KiB
C

/* bnx2i.c: Broadcom NetXtreme II iSCSI driver.
*
* Copyright (c) 2006 - 2010 Broadcom Corporation
* Copyright (c) 2007, 2008 Red Hat, Inc. All rights reserved.
* Copyright (c) 2007, 2008 Mike Christie
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation.
*
* Written by: Anil Veerabhadrappa (anilgv@broadcom.com)
* Maintained by: Eddie Wai (eddie.wai@broadcom.com)
*/
#include "bnx2i.h"
static struct list_head adapter_list = LIST_HEAD_INIT(adapter_list);
static u32 adapter_count;
#define DRV_MODULE_NAME "bnx2i"
#define DRV_MODULE_VERSION "2.6.2.3"
#define DRV_MODULE_RELDATE "Dec 31, 2010"
static char version[] __devinitdata =
"Broadcom NetXtreme II iSCSI Driver " DRV_MODULE_NAME \
" v" DRV_MODULE_VERSION " (" DRV_MODULE_RELDATE ")\n";
MODULE_AUTHOR("Anil Veerabhadrappa <anilgv@broadcom.com> and "
"Eddie Wai <eddie.wai@broadcom.com>");
MODULE_DESCRIPTION("Broadcom NetXtreme II BCM5706/5708/5709/57710/57711/57712"
" iSCSI Driver");
MODULE_LICENSE("GPL");
MODULE_VERSION(DRV_MODULE_VERSION);
static DEFINE_MUTEX(bnx2i_dev_lock);
unsigned int event_coal_min = 24;
module_param(event_coal_min, int, 0664);
MODULE_PARM_DESC(event_coal_min, "Event Coalescing Minimum Commands");
unsigned int event_coal_div = 2;
module_param(event_coal_div, int, 0664);
MODULE_PARM_DESC(event_coal_div, "Event Coalescing Divide Factor");
unsigned int en_tcp_dack = 1;
module_param(en_tcp_dack, int, 0664);
MODULE_PARM_DESC(en_tcp_dack, "Enable TCP Delayed ACK");
unsigned int error_mask1 = 0x00;
module_param(error_mask1, int, 0664);
MODULE_PARM_DESC(error_mask1, "Config FW iSCSI Error Mask #1");
unsigned int error_mask2 = 0x00;
module_param(error_mask2, int, 0664);
MODULE_PARM_DESC(error_mask2, "Config FW iSCSI Error Mask #2");
unsigned int sq_size;
module_param(sq_size, int, 0664);
MODULE_PARM_DESC(sq_size, "Configure SQ size");
unsigned int rq_size = BNX2I_RQ_WQES_DEFAULT;
module_param(rq_size, int, 0664);
MODULE_PARM_DESC(rq_size, "Configure RQ size");
u64 iscsi_error_mask = 0x00;
DEFINE_PER_CPU(struct bnx2i_percpu_s, bnx2i_percpu);
static int bnx2i_cpu_callback(struct notifier_block *nfb,
unsigned long action, void *hcpu);
/* notification function for CPU hotplug events */
static struct notifier_block bnx2i_cpu_notifier = {
.notifier_call = bnx2i_cpu_callback,
};
/**
* bnx2i_identify_device - identifies NetXtreme II device type
* @hba: Adapter structure pointer
*
* This function identifies the NX2 device type and sets appropriate
* queue mailbox register access method, 5709 requires driver to
* access MBOX regs using *bin* mode
*/
void bnx2i_identify_device(struct bnx2i_hba *hba)
{
hba->cnic_dev_type = 0;
if ((hba->pci_did == PCI_DEVICE_ID_NX2_5706) ||
(hba->pci_did == PCI_DEVICE_ID_NX2_5706S))
set_bit(BNX2I_NX2_DEV_5706, &hba->cnic_dev_type);
else if ((hba->pci_did == PCI_DEVICE_ID_NX2_5708) ||
(hba->pci_did == PCI_DEVICE_ID_NX2_5708S))
set_bit(BNX2I_NX2_DEV_5708, &hba->cnic_dev_type);
else if ((hba->pci_did == PCI_DEVICE_ID_NX2_5709) ||
(hba->pci_did == PCI_DEVICE_ID_NX2_5709S)) {
set_bit(BNX2I_NX2_DEV_5709, &hba->cnic_dev_type);
hba->mail_queue_access = BNX2I_MQ_BIN_MODE;
} else if (hba->pci_did == PCI_DEVICE_ID_NX2_57710 ||
hba->pci_did == PCI_DEVICE_ID_NX2_57711 ||
hba->pci_did == PCI_DEVICE_ID_NX2_57711E ||
hba->pci_did == PCI_DEVICE_ID_NX2_57712 ||
hba->pci_did == PCI_DEVICE_ID_NX2_57712E)
set_bit(BNX2I_NX2_DEV_57710, &hba->cnic_dev_type);
else
printk(KERN_ALERT "bnx2i: unknown device, 0x%x\n",
hba->pci_did);
}
/**
* get_adapter_list_head - returns head of adapter list
*/
struct bnx2i_hba *get_adapter_list_head(void)
{
struct bnx2i_hba *hba = NULL;
struct bnx2i_hba *tmp_hba;
if (!adapter_count)
goto hba_not_found;
mutex_lock(&bnx2i_dev_lock);
list_for_each_entry(tmp_hba, &adapter_list, link) {
if (tmp_hba->cnic && tmp_hba->cnic->cm_select_dev) {
hba = tmp_hba;
break;
}
}
mutex_unlock(&bnx2i_dev_lock);
hba_not_found:
return hba;
}
/**
* bnx2i_find_hba_for_cnic - maps cnic device instance to bnx2i adapter instance
* @cnic: pointer to cnic device instance
*
*/
struct bnx2i_hba *bnx2i_find_hba_for_cnic(struct cnic_dev *cnic)
{
struct bnx2i_hba *hba, *temp;
mutex_lock(&bnx2i_dev_lock);
list_for_each_entry_safe(hba, temp, &adapter_list, link) {
if (hba->cnic == cnic) {
mutex_unlock(&bnx2i_dev_lock);
return hba;
}
}
mutex_unlock(&bnx2i_dev_lock);
return NULL;
}
/**
* bnx2i_start - cnic callback to initialize & start adapter instance
* @handle: transparent handle pointing to adapter structure
*
* This function maps adapter structure to pcidev structure and initiates
* firmware handshake to enable/initialize on chip iscsi components
* This bnx2i - cnic interface api callback is issued after following
* 2 conditions are met -
* a) underlying network interface is up (marked by event 'NETDEV_UP'
* from netdev
* b) bnx2i adapter instance is registered
*/
void bnx2i_start(void *handle)
{
#define BNX2I_INIT_POLL_TIME (1000 / HZ)
struct bnx2i_hba *hba = handle;
int i = HZ;
if (!hba->cnic->max_iscsi_conn) {
printk(KERN_ALERT "bnx2i: dev %s does not support "
"iSCSI\n", hba->netdev->name);
if (test_bit(BNX2I_CNIC_REGISTERED, &hba->reg_with_cnic)) {
mutex_lock(&bnx2i_dev_lock);
list_del_init(&hba->link);
adapter_count--;
hba->cnic->unregister_device(hba->cnic, CNIC_ULP_ISCSI);
clear_bit(BNX2I_CNIC_REGISTERED, &hba->reg_with_cnic);
mutex_unlock(&bnx2i_dev_lock);
bnx2i_free_hba(hba);
}
return;
}
bnx2i_send_fw_iscsi_init_msg(hba);
while (!test_bit(ADAPTER_STATE_UP, &hba->adapter_state) && i--)
msleep(BNX2I_INIT_POLL_TIME);
}
/**
* bnx2i_chip_cleanup - local routine to handle chip cleanup
* @hba: Adapter instance to register
*
* Driver checks if adapter still has any active connections before
* executing the cleanup process
*/
static void bnx2i_chip_cleanup(struct bnx2i_hba *hba)
{
struct bnx2i_endpoint *bnx2i_ep;
struct list_head *pos, *tmp;
if (hba->ofld_conns_active) {
/* Stage to force the disconnection
* This is the case where the daemon is either slow or
* not present
*/
printk(KERN_ALERT "bnx2i: (%s) chip cleanup for %d active "
"connections\n", hba->netdev->name,
hba->ofld_conns_active);
mutex_lock(&hba->net_dev_lock);
list_for_each_safe(pos, tmp, &hba->ep_active_list) {
bnx2i_ep = list_entry(pos, struct bnx2i_endpoint, link);
/* Clean up the chip only */
bnx2i_hw_ep_disconnect(bnx2i_ep);
bnx2i_ep->cm_sk = NULL;
}
mutex_unlock(&hba->net_dev_lock);
}
}
/**
* bnx2i_stop - cnic callback to shutdown adapter instance
* @handle: transparent handle pointing to adapter structure
*
* driver checks if adapter is already in shutdown mode, if not start
* the shutdown process
*/
void bnx2i_stop(void *handle)
{
struct bnx2i_hba *hba = handle;
int conns_active;
int wait_delay = 1 * HZ;
/* check if cleanup happened in GOING_DOWN context */
if (!test_and_set_bit(ADAPTER_STATE_GOING_DOWN,
&hba->adapter_state)) {
iscsi_host_for_each_session(hba->shost,
bnx2i_drop_session);
wait_delay = hba->hba_shutdown_tmo;
}
/* Wait for inflight offload connection tasks to complete before
* proceeding. Forcefully terminate all connection recovery in
* progress at the earliest, either in bind(), send_pdu(LOGIN),
* or conn_start()
*/
wait_event_interruptible_timeout(hba->eh_wait,
(list_empty(&hba->ep_ofld_list) &&
list_empty(&hba->ep_destroy_list)),
2 * HZ);
/* Wait for all endpoints to be torn down, Chip will be reset once
* control returns to network driver. So it is required to cleanup and
* release all connection resources before returning from this routine.
*/
while (hba->ofld_conns_active) {
conns_active = hba->ofld_conns_active;
wait_event_interruptible_timeout(hba->eh_wait,
(hba->ofld_conns_active != conns_active),
wait_delay);
if (hba->ofld_conns_active == conns_active)
break;
}
bnx2i_chip_cleanup(hba);
/* This flag should be cleared last so that ep_disconnect() gracefully
* cleans up connection context
*/
clear_bit(ADAPTER_STATE_GOING_DOWN, &hba->adapter_state);
clear_bit(ADAPTER_STATE_UP, &hba->adapter_state);
}
/**
* bnx2i_init_one - initialize an adapter instance and allocate memory resources
* @hba: bnx2i adapter instance
* @cnic: cnic device handle
*
* Global resource lock is held during critical sections below. This routine is
* called from either cnic_register_driver() or device hot plug context and
* and does majority of device specific initialization
*/
static int bnx2i_init_one(struct bnx2i_hba *hba, struct cnic_dev *cnic)
{
int rc;
mutex_lock(&bnx2i_dev_lock);
hba->cnic = cnic;
rc = cnic->register_device(cnic, CNIC_ULP_ISCSI, hba);
if (!rc) {
hba->age++;
set_bit(BNX2I_CNIC_REGISTERED, &hba->reg_with_cnic);
list_add_tail(&hba->link, &adapter_list);
adapter_count++;
} else if (rc == -EBUSY) /* duplicate registration */
printk(KERN_ALERT "bnx2i, duplicate registration"
"hba=%p, cnic=%p\n", hba, cnic);
else if (rc == -EAGAIN)
printk(KERN_ERR "bnx2i, driver not registered\n");
else if (rc == -EINVAL)
printk(KERN_ERR "bnx2i, invalid type %d\n", CNIC_ULP_ISCSI);
else
printk(KERN_ERR "bnx2i dev reg, unknown error, %d\n", rc);
mutex_unlock(&bnx2i_dev_lock);
return rc;
}
/**
* bnx2i_ulp_init - initialize an adapter instance
* @dev: cnic device handle
*
* Called from cnic_register_driver() context to initialize all enumerated
* cnic devices. This routine allocate adapter structure and other
* device specific resources.
*/
void bnx2i_ulp_init(struct cnic_dev *dev)
{
struct bnx2i_hba *hba;
/* Allocate a HBA structure for this device */
hba = bnx2i_alloc_hba(dev);
if (!hba) {
printk(KERN_ERR "bnx2i init: hba initialization failed\n");
return;
}
/* Get PCI related information and update hba struct members */
clear_bit(BNX2I_CNIC_REGISTERED, &hba->reg_with_cnic);
if (bnx2i_init_one(hba, dev)) {
printk(KERN_ERR "bnx2i - hba %p init failed\n", hba);
bnx2i_free_hba(hba);
}
}
/**
* bnx2i_ulp_exit - shuts down adapter instance and frees all resources
* @dev: cnic device handle
*
*/
void bnx2i_ulp_exit(struct cnic_dev *dev)
{
struct bnx2i_hba *hba;
hba = bnx2i_find_hba_for_cnic(dev);
if (!hba) {
printk(KERN_INFO "bnx2i_ulp_exit: hba not "
"found, dev 0x%p\n", dev);
return;
}
mutex_lock(&bnx2i_dev_lock);
list_del_init(&hba->link);
adapter_count--;
if (test_bit(BNX2I_CNIC_REGISTERED, &hba->reg_with_cnic)) {
hba->cnic->unregister_device(hba->cnic, CNIC_ULP_ISCSI);
clear_bit(BNX2I_CNIC_REGISTERED, &hba->reg_with_cnic);
}
mutex_unlock(&bnx2i_dev_lock);
bnx2i_free_hba(hba);
}
/**
* bnx2i_percpu_thread_create - Create a receive thread for an
* online CPU
*
* @cpu: cpu index for the online cpu
*/
static void bnx2i_percpu_thread_create(unsigned int cpu)
{
struct bnx2i_percpu_s *p;
struct task_struct *thread;
p = &per_cpu(bnx2i_percpu, cpu);
thread = kthread_create(bnx2i_percpu_io_thread, (void *)p,
"bnx2i_thread/%d", cpu);
/* bind thread to the cpu */
if (likely(!IS_ERR(thread))) {
kthread_bind(thread, cpu);
p->iothread = thread;
wake_up_process(thread);
}
}
static void bnx2i_percpu_thread_destroy(unsigned int cpu)
{
struct bnx2i_percpu_s *p;
struct task_struct *thread;
struct bnx2i_work *work, *tmp;
/* Prevent any new work from being queued for this CPU */
p = &per_cpu(bnx2i_percpu, cpu);
spin_lock_bh(&p->p_work_lock);
thread = p->iothread;
p->iothread = NULL;
/* Free all work in the list */
list_for_each_entry_safe(work, tmp, &p->work_list, list) {
list_del_init(&work->list);
bnx2i_process_scsi_cmd_resp(work->session,
work->bnx2i_conn, &work->cqe);
kfree(work);
}
spin_unlock_bh(&p->p_work_lock);
if (thread)
kthread_stop(thread);
}
/**
* bnx2i_cpu_callback - Handler for CPU hotplug events
*
* @nfb: The callback data block
* @action: The event triggering the callback
* @hcpu: The index of the CPU that the event is for
*
* This creates or destroys per-CPU data for iSCSI
*
* Returns NOTIFY_OK always.
*/
static int bnx2i_cpu_callback(struct notifier_block *nfb,
unsigned long action, void *hcpu)
{
unsigned cpu = (unsigned long)hcpu;
switch (action) {
case CPU_ONLINE:
case CPU_ONLINE_FROZEN:
printk(KERN_INFO "bnx2i: CPU %x online: Create Rx thread\n",
cpu);
bnx2i_percpu_thread_create(cpu);
break;
case CPU_DEAD:
case CPU_DEAD_FROZEN:
printk(KERN_INFO "CPU %x offline: Remove Rx thread\n", cpu);
bnx2i_percpu_thread_destroy(cpu);
break;
default:
break;
}
return NOTIFY_OK;
}
/**
* bnx2i_mod_init - module init entry point
*
* initialize any driver wide global data structures such as endpoint pool,
* tcp port manager/queue, sysfs. finally driver will register itself
* with the cnic module
*/
static int __init bnx2i_mod_init(void)
{
int err;
unsigned cpu = 0;
struct bnx2i_percpu_s *p;
printk(KERN_INFO "%s", version);
if (sq_size && !is_power_of_2(sq_size))
sq_size = roundup_pow_of_two(sq_size);
mutex_init(&bnx2i_dev_lock);
bnx2i_scsi_xport_template =
iscsi_register_transport(&bnx2i_iscsi_transport);
if (!bnx2i_scsi_xport_template) {
printk(KERN_ERR "Could not register bnx2i transport.\n");
err = -ENOMEM;
goto out;
}
err = cnic_register_driver(CNIC_ULP_ISCSI, &bnx2i_cnic_cb);
if (err) {
printk(KERN_ERR "Could not register bnx2i cnic driver.\n");
goto unreg_xport;
}
/* Create percpu kernel threads to handle iSCSI I/O completions */
for_each_possible_cpu(cpu) {
p = &per_cpu(bnx2i_percpu, cpu);
INIT_LIST_HEAD(&p->work_list);
spin_lock_init(&p->p_work_lock);
p->iothread = NULL;
}
for_each_online_cpu(cpu)
bnx2i_percpu_thread_create(cpu);
/* Initialize per CPU interrupt thread */
register_hotcpu_notifier(&bnx2i_cpu_notifier);
return 0;
unreg_xport:
iscsi_unregister_transport(&bnx2i_iscsi_transport);
out:
return err;
}
/**
* bnx2i_mod_exit - module cleanup/exit entry point
*
* Global resource lock and host adapter lock is held during critical sections
* in this function. Driver will browse through the adapter list, cleans-up
* each instance, unregisters iscsi transport name and finally driver will
* unregister itself with the cnic module
*/
static void __exit bnx2i_mod_exit(void)
{
struct bnx2i_hba *hba;
unsigned cpu = 0;
mutex_lock(&bnx2i_dev_lock);
while (!list_empty(&adapter_list)) {
hba = list_entry(adapter_list.next, struct bnx2i_hba, link);
list_del(&hba->link);
adapter_count--;
if (test_bit(BNX2I_CNIC_REGISTERED, &hba->reg_with_cnic)) {
bnx2i_chip_cleanup(hba);
hba->cnic->unregister_device(hba->cnic, CNIC_ULP_ISCSI);
clear_bit(BNX2I_CNIC_REGISTERED, &hba->reg_with_cnic);
}
bnx2i_free_hba(hba);
}
mutex_unlock(&bnx2i_dev_lock);
unregister_hotcpu_notifier(&bnx2i_cpu_notifier);
for_each_online_cpu(cpu)
bnx2i_percpu_thread_destroy(cpu);
iscsi_unregister_transport(&bnx2i_iscsi_transport);
cnic_unregister_driver(CNIC_ULP_ISCSI);
}
module_init(bnx2i_mod_init);
module_exit(bnx2i_mod_exit);