/*
|
/*
|
* linux/net/sunrpc/sched.c
|
* linux/net/sunrpc/sched.c
|
*
|
*
|
* Scheduling for synchronous and asynchronous RPC requests.
|
* Scheduling for synchronous and asynchronous RPC requests.
|
*
|
*
|
* Copyright (C) 1996 Olaf Kirch, <okir@monad.swb.de>
|
* Copyright (C) 1996 Olaf Kirch, <okir@monad.swb.de>
|
*
|
*
|
* TCP NFS related read + write fixes
|
* TCP NFS related read + write fixes
|
* (C) 1999 Dave Airlie, University of Limerick, Ireland <airlied@linux.ie>
|
* (C) 1999 Dave Airlie, University of Limerick, Ireland <airlied@linux.ie>
|
*/
|
*/
|
|
|
#include <linux/module.h>
|
#include <linux/module.h>
|
|
|
#include <linux/sched.h>
|
#include <linux/sched.h>
|
#include <linux/interrupt.h>
|
#include <linux/interrupt.h>
|
#include <linux/slab.h>
|
#include <linux/slab.h>
|
#include <linux/mempool.h>
|
#include <linux/mempool.h>
|
#include <linux/smp.h>
|
#include <linux/smp.h>
|
#include <linux/smp_lock.h>
|
#include <linux/smp_lock.h>
|
#include <linux/spinlock.h>
|
#include <linux/spinlock.h>
|
#include <linux/mutex.h>
|
#include <linux/mutex.h>
|
|
|
#include <linux/sunrpc/clnt.h>
|
#include <linux/sunrpc/clnt.h>
|
|
|
#ifdef RPC_DEBUG
|
#ifdef RPC_DEBUG
|
#define RPCDBG_FACILITY RPCDBG_SCHED
|
#define RPCDBG_FACILITY RPCDBG_SCHED
|
#define RPC_TASK_MAGIC_ID 0xf00baa
|
#define RPC_TASK_MAGIC_ID 0xf00baa
|
#endif
|
#endif
|
|
|
/*
|
/*
|
* RPC slabs and memory pools
|
* RPC slabs and memory pools
|
*/
|
*/
|
#define RPC_BUFFER_MAXSIZE (2048)
|
#define RPC_BUFFER_MAXSIZE (2048)
|
#define RPC_BUFFER_POOLSIZE (8)
|
#define RPC_BUFFER_POOLSIZE (8)
|
#define RPC_TASK_POOLSIZE (8)
|
#define RPC_TASK_POOLSIZE (8)
|
static struct kmem_cache *rpc_task_slabp __read_mostly;
|
static struct kmem_cache *rpc_task_slabp __read_mostly;
|
static struct kmem_cache *rpc_buffer_slabp __read_mostly;
|
static struct kmem_cache *rpc_buffer_slabp __read_mostly;
|
static mempool_t *rpc_task_mempool __read_mostly;
|
static mempool_t *rpc_task_mempool __read_mostly;
|
static mempool_t *rpc_buffer_mempool __read_mostly;
|
static mempool_t *rpc_buffer_mempool __read_mostly;
|
|
|
static void __rpc_default_timer(struct rpc_task *task);
|
static void __rpc_default_timer(struct rpc_task *task);
|
static void rpc_async_schedule(struct work_struct *);
|
static void rpc_async_schedule(struct work_struct *);
|
static void rpc_release_task(struct rpc_task *task);
|
static void rpc_release_task(struct rpc_task *task);
|
|
|
/*
|
/*
|
* RPC tasks sit here while waiting for conditions to improve.
|
* RPC tasks sit here while waiting for conditions to improve.
|
*/
|
*/
|
static RPC_WAITQ(delay_queue, "delayq");
|
static RPC_WAITQ(delay_queue, "delayq");
|
|
|
/*
|
/*
|
* rpciod-related stuff
|
* rpciod-related stuff
|
*/
|
*/
|
struct workqueue_struct *rpciod_workqueue;
|
struct workqueue_struct *rpciod_workqueue;
|
|
|
/*
|
/*
|
* Disable the timer for a given RPC task. Should be called with
|
* Disable the timer for a given RPC task. Should be called with
|
* queue->lock and bh_disabled in order to avoid races within
|
* queue->lock and bh_disabled in order to avoid races within
|
* rpc_run_timer().
|
* rpc_run_timer().
|
*/
|
*/
|
static inline void
|
static inline void
|
__rpc_disable_timer(struct rpc_task *task)
|
__rpc_disable_timer(struct rpc_task *task)
|
{
|
{
|
dprintk("RPC: %5u disabling timer\n", task->tk_pid);
|
dprintk("RPC: %5u disabling timer\n", task->tk_pid);
|
task->tk_timeout_fn = NULL;
|
task->tk_timeout_fn = NULL;
|
task->tk_timeout = 0;
|
task->tk_timeout = 0;
|
}
|
}
|
|
|
/*
|
/*
|
* Run a timeout function.
|
* Run a timeout function.
|
* We use the callback in order to allow __rpc_wake_up_task()
|
* We use the callback in order to allow __rpc_wake_up_task()
|
* and friends to disable the timer synchronously on SMP systems
|
* and friends to disable the timer synchronously on SMP systems
|
* without calling del_timer_sync(). The latter could cause a
|
* without calling del_timer_sync(). The latter could cause a
|
* deadlock if called while we're holding spinlocks...
|
* deadlock if called while we're holding spinlocks...
|
*/
|
*/
|
static void rpc_run_timer(struct rpc_task *task)
|
static void rpc_run_timer(struct rpc_task *task)
|
{
|
{
|
void (*callback)(struct rpc_task *);
|
void (*callback)(struct rpc_task *);
|
|
|
callback = task->tk_timeout_fn;
|
callback = task->tk_timeout_fn;
|
task->tk_timeout_fn = NULL;
|
task->tk_timeout_fn = NULL;
|
if (callback && RPC_IS_QUEUED(task)) {
|
if (callback && RPC_IS_QUEUED(task)) {
|
dprintk("RPC: %5u running timer\n", task->tk_pid);
|
dprintk("RPC: %5u running timer\n", task->tk_pid);
|
callback(task);
|
callback(task);
|
}
|
}
|
smp_mb__before_clear_bit();
|
smp_mb__before_clear_bit();
|
clear_bit(RPC_TASK_HAS_TIMER, &task->tk_runstate);
|
clear_bit(RPC_TASK_HAS_TIMER, &task->tk_runstate);
|
smp_mb__after_clear_bit();
|
smp_mb__after_clear_bit();
|
}
|
}
|
|
|
/*
|
/*
|
* Set up a timer for the current task.
|
* Set up a timer for the current task.
|
*/
|
*/
|
static inline void
|
static inline void
|
__rpc_add_timer(struct rpc_task *task, rpc_action timer)
|
__rpc_add_timer(struct rpc_task *task, rpc_action timer)
|
{
|
{
|
if (!task->tk_timeout)
|
if (!task->tk_timeout)
|
return;
|
return;
|
|
|
dprintk("RPC: %5u setting alarm for %lu ms\n",
|
dprintk("RPC: %5u setting alarm for %lu ms\n",
|
task->tk_pid, task->tk_timeout * 1000 / HZ);
|
task->tk_pid, task->tk_timeout * 1000 / HZ);
|
|
|
if (timer)
|
if (timer)
|
task->tk_timeout_fn = timer;
|
task->tk_timeout_fn = timer;
|
else
|
else
|
task->tk_timeout_fn = __rpc_default_timer;
|
task->tk_timeout_fn = __rpc_default_timer;
|
set_bit(RPC_TASK_HAS_TIMER, &task->tk_runstate);
|
set_bit(RPC_TASK_HAS_TIMER, &task->tk_runstate);
|
mod_timer(&task->tk_timer, jiffies + task->tk_timeout);
|
mod_timer(&task->tk_timer, jiffies + task->tk_timeout);
|
}
|
}
|
|
|
/*
|
/*
|
* Delete any timer for the current task. Because we use del_timer_sync(),
|
* Delete any timer for the current task. Because we use del_timer_sync(),
|
* this function should never be called while holding queue->lock.
|
* this function should never be called while holding queue->lock.
|
*/
|
*/
|
static void
|
static void
|
rpc_delete_timer(struct rpc_task *task)
|
rpc_delete_timer(struct rpc_task *task)
|
{
|
{
|
if (RPC_IS_QUEUED(task))
|
if (RPC_IS_QUEUED(task))
|
return;
|
return;
|
if (test_and_clear_bit(RPC_TASK_HAS_TIMER, &task->tk_runstate)) {
|
if (test_and_clear_bit(RPC_TASK_HAS_TIMER, &task->tk_runstate)) {
|
del_singleshot_timer_sync(&task->tk_timer);
|
del_singleshot_timer_sync(&task->tk_timer);
|
dprintk("RPC: %5u deleting timer\n", task->tk_pid);
|
dprintk("RPC: %5u deleting timer\n", task->tk_pid);
|
}
|
}
|
}
|
}
|
|
|
/*
|
/*
|
* Add new request to a priority queue.
|
* Add new request to a priority queue.
|
*/
|
*/
|
static void __rpc_add_wait_queue_priority(struct rpc_wait_queue *queue, struct rpc_task *task)
|
static void __rpc_add_wait_queue_priority(struct rpc_wait_queue *queue, struct rpc_task *task)
|
{
|
{
|
struct list_head *q;
|
struct list_head *q;
|
struct rpc_task *t;
|
struct rpc_task *t;
|
|
|
INIT_LIST_HEAD(&task->u.tk_wait.links);
|
INIT_LIST_HEAD(&task->u.tk_wait.links);
|
q = &queue->tasks[task->tk_priority];
|
q = &queue->tasks[task->tk_priority];
|
if (unlikely(task->tk_priority > queue->maxpriority))
|
if (unlikely(task->tk_priority > queue->maxpriority))
|
q = &queue->tasks[queue->maxpriority];
|
q = &queue->tasks[queue->maxpriority];
|
list_for_each_entry(t, q, u.tk_wait.list) {
|
list_for_each_entry(t, q, u.tk_wait.list) {
|
if (t->tk_cookie == task->tk_cookie) {
|
if (t->tk_cookie == task->tk_cookie) {
|
list_add_tail(&task->u.tk_wait.list, &t->u.tk_wait.links);
|
list_add_tail(&task->u.tk_wait.list, &t->u.tk_wait.links);
|
return;
|
return;
|
}
|
}
|
}
|
}
|
list_add_tail(&task->u.tk_wait.list, q);
|
list_add_tail(&task->u.tk_wait.list, q);
|
}
|
}
|
|
|
/*
|
/*
|
* Add new request to wait queue.
|
* Add new request to wait queue.
|
*
|
*
|
* Swapper tasks always get inserted at the head of the queue.
|
* Swapper tasks always get inserted at the head of the queue.
|
* This should avoid many nasty memory deadlocks and hopefully
|
* This should avoid many nasty memory deadlocks and hopefully
|
* improve overall performance.
|
* improve overall performance.
|
* Everyone else gets appended to the queue to ensure proper FIFO behavior.
|
* Everyone else gets appended to the queue to ensure proper FIFO behavior.
|
*/
|
*/
|
static void __rpc_add_wait_queue(struct rpc_wait_queue *queue, struct rpc_task *task)
|
static void __rpc_add_wait_queue(struct rpc_wait_queue *queue, struct rpc_task *task)
|
{
|
{
|
BUG_ON (RPC_IS_QUEUED(task));
|
BUG_ON (RPC_IS_QUEUED(task));
|
|
|
if (RPC_IS_PRIORITY(queue))
|
if (RPC_IS_PRIORITY(queue))
|
__rpc_add_wait_queue_priority(queue, task);
|
__rpc_add_wait_queue_priority(queue, task);
|
else if (RPC_IS_SWAPPER(task))
|
else if (RPC_IS_SWAPPER(task))
|
list_add(&task->u.tk_wait.list, &queue->tasks[0]);
|
list_add(&task->u.tk_wait.list, &queue->tasks[0]);
|
else
|
else
|
list_add_tail(&task->u.tk_wait.list, &queue->tasks[0]);
|
list_add_tail(&task->u.tk_wait.list, &queue->tasks[0]);
|
task->u.tk_wait.rpc_waitq = queue;
|
task->u.tk_wait.rpc_waitq = queue;
|
queue->qlen++;
|
queue->qlen++;
|
rpc_set_queued(task);
|
rpc_set_queued(task);
|
|
|
dprintk("RPC: %5u added to queue %p \"%s\"\n",
|
dprintk("RPC: %5u added to queue %p \"%s\"\n",
|
task->tk_pid, queue, rpc_qname(queue));
|
task->tk_pid, queue, rpc_qname(queue));
|
}
|
}
|
|
|
/*
|
/*
|
* Remove request from a priority queue.
|
* Remove request from a priority queue.
|
*/
|
*/
|
static void __rpc_remove_wait_queue_priority(struct rpc_task *task)
|
static void __rpc_remove_wait_queue_priority(struct rpc_task *task)
|
{
|
{
|
struct rpc_task *t;
|
struct rpc_task *t;
|
|
|
if (!list_empty(&task->u.tk_wait.links)) {
|
if (!list_empty(&task->u.tk_wait.links)) {
|
t = list_entry(task->u.tk_wait.links.next, struct rpc_task, u.tk_wait.list);
|
t = list_entry(task->u.tk_wait.links.next, struct rpc_task, u.tk_wait.list);
|
list_move(&t->u.tk_wait.list, &task->u.tk_wait.list);
|
list_move(&t->u.tk_wait.list, &task->u.tk_wait.list);
|
list_splice_init(&task->u.tk_wait.links, &t->u.tk_wait.links);
|
list_splice_init(&task->u.tk_wait.links, &t->u.tk_wait.links);
|
}
|
}
|
list_del(&task->u.tk_wait.list);
|
list_del(&task->u.tk_wait.list);
|
}
|
}
|
|
|
/*
|
/*
|
* Remove request from queue.
|
* Remove request from queue.
|
* Note: must be called with spin lock held.
|
* Note: must be called with spin lock held.
|
*/
|
*/
|
static void __rpc_remove_wait_queue(struct rpc_task *task)
|
static void __rpc_remove_wait_queue(struct rpc_task *task)
|
{
|
{
|
struct rpc_wait_queue *queue;
|
struct rpc_wait_queue *queue;
|
queue = task->u.tk_wait.rpc_waitq;
|
queue = task->u.tk_wait.rpc_waitq;
|
|
|
if (RPC_IS_PRIORITY(queue))
|
if (RPC_IS_PRIORITY(queue))
|
__rpc_remove_wait_queue_priority(task);
|
__rpc_remove_wait_queue_priority(task);
|
else
|
else
|
list_del(&task->u.tk_wait.list);
|
list_del(&task->u.tk_wait.list);
|
queue->qlen--;
|
queue->qlen--;
|
dprintk("RPC: %5u removed from queue %p \"%s\"\n",
|
dprintk("RPC: %5u removed from queue %p \"%s\"\n",
|
task->tk_pid, queue, rpc_qname(queue));
|
task->tk_pid, queue, rpc_qname(queue));
|
}
|
}
|
|
|
static inline void rpc_set_waitqueue_priority(struct rpc_wait_queue *queue, int priority)
|
static inline void rpc_set_waitqueue_priority(struct rpc_wait_queue *queue, int priority)
|
{
|
{
|
queue->priority = priority;
|
queue->priority = priority;
|
queue->count = 1 << (priority * 2);
|
queue->count = 1 << (priority * 2);
|
}
|
}
|
|
|
static inline void rpc_set_waitqueue_cookie(struct rpc_wait_queue *queue, unsigned long cookie)
|
static inline void rpc_set_waitqueue_cookie(struct rpc_wait_queue *queue, unsigned long cookie)
|
{
|
{
|
queue->cookie = cookie;
|
queue->cookie = cookie;
|
queue->nr = RPC_BATCH_COUNT;
|
queue->nr = RPC_BATCH_COUNT;
|
}
|
}
|
|
|
static inline void rpc_reset_waitqueue_priority(struct rpc_wait_queue *queue)
|
static inline void rpc_reset_waitqueue_priority(struct rpc_wait_queue *queue)
|
{
|
{
|
rpc_set_waitqueue_priority(queue, queue->maxpriority);
|
rpc_set_waitqueue_priority(queue, queue->maxpriority);
|
rpc_set_waitqueue_cookie(queue, 0);
|
rpc_set_waitqueue_cookie(queue, 0);
|
}
|
}
|
|
|
static void __rpc_init_priority_wait_queue(struct rpc_wait_queue *queue, const char *qname, int maxprio)
|
static void __rpc_init_priority_wait_queue(struct rpc_wait_queue *queue, const char *qname, int maxprio)
|
{
|
{
|
int i;
|
int i;
|
|
|
spin_lock_init(&queue->lock);
|
spin_lock_init(&queue->lock);
|
for (i = 0; i < ARRAY_SIZE(queue->tasks); i++)
|
for (i = 0; i < ARRAY_SIZE(queue->tasks); i++)
|
INIT_LIST_HEAD(&queue->tasks[i]);
|
INIT_LIST_HEAD(&queue->tasks[i]);
|
queue->maxpriority = maxprio;
|
queue->maxpriority = maxprio;
|
rpc_reset_waitqueue_priority(queue);
|
rpc_reset_waitqueue_priority(queue);
|
#ifdef RPC_DEBUG
|
#ifdef RPC_DEBUG
|
queue->name = qname;
|
queue->name = qname;
|
#endif
|
#endif
|
}
|
}
|
|
|
void rpc_init_priority_wait_queue(struct rpc_wait_queue *queue, const char *qname)
|
void rpc_init_priority_wait_queue(struct rpc_wait_queue *queue, const char *qname)
|
{
|
{
|
__rpc_init_priority_wait_queue(queue, qname, RPC_PRIORITY_HIGH);
|
__rpc_init_priority_wait_queue(queue, qname, RPC_PRIORITY_HIGH);
|
}
|
}
|
|
|
void rpc_init_wait_queue(struct rpc_wait_queue *queue, const char *qname)
|
void rpc_init_wait_queue(struct rpc_wait_queue *queue, const char *qname)
|
{
|
{
|
__rpc_init_priority_wait_queue(queue, qname, 0);
|
__rpc_init_priority_wait_queue(queue, qname, 0);
|
}
|
}
|
EXPORT_SYMBOL(rpc_init_wait_queue);
|
EXPORT_SYMBOL(rpc_init_wait_queue);
|
|
|
static int rpc_wait_bit_interruptible(void *word)
|
static int rpc_wait_bit_interruptible(void *word)
|
{
|
{
|
if (signal_pending(current))
|
if (signal_pending(current))
|
return -ERESTARTSYS;
|
return -ERESTARTSYS;
|
schedule();
|
schedule();
|
return 0;
|
return 0;
|
}
|
}
|
|
|
#ifdef RPC_DEBUG
|
#ifdef RPC_DEBUG
|
static void rpc_task_set_debuginfo(struct rpc_task *task)
|
static void rpc_task_set_debuginfo(struct rpc_task *task)
|
{
|
{
|
static atomic_t rpc_pid;
|
static atomic_t rpc_pid;
|
|
|
task->tk_magic = RPC_TASK_MAGIC_ID;
|
task->tk_magic = RPC_TASK_MAGIC_ID;
|
task->tk_pid = atomic_inc_return(&rpc_pid);
|
task->tk_pid = atomic_inc_return(&rpc_pid);
|
}
|
}
|
#else
|
#else
|
static inline void rpc_task_set_debuginfo(struct rpc_task *task)
|
static inline void rpc_task_set_debuginfo(struct rpc_task *task)
|
{
|
{
|
}
|
}
|
#endif
|
#endif
|
|
|
static void rpc_set_active(struct rpc_task *task)
|
static void rpc_set_active(struct rpc_task *task)
|
{
|
{
|
struct rpc_clnt *clnt;
|
struct rpc_clnt *clnt;
|
if (test_and_set_bit(RPC_TASK_ACTIVE, &task->tk_runstate) != 0)
|
if (test_and_set_bit(RPC_TASK_ACTIVE, &task->tk_runstate) != 0)
|
return;
|
return;
|
rpc_task_set_debuginfo(task);
|
rpc_task_set_debuginfo(task);
|
/* Add to global list of all tasks */
|
/* Add to global list of all tasks */
|
clnt = task->tk_client;
|
clnt = task->tk_client;
|
if (clnt != NULL) {
|
if (clnt != NULL) {
|
spin_lock(&clnt->cl_lock);
|
spin_lock(&clnt->cl_lock);
|
list_add_tail(&task->tk_task, &clnt->cl_tasks);
|
list_add_tail(&task->tk_task, &clnt->cl_tasks);
|
spin_unlock(&clnt->cl_lock);
|
spin_unlock(&clnt->cl_lock);
|
}
|
}
|
}
|
}
|
|
|
/*
|
/*
|
* Mark an RPC call as having completed by clearing the 'active' bit
|
* Mark an RPC call as having completed by clearing the 'active' bit
|
*/
|
*/
|
static void rpc_mark_complete_task(struct rpc_task *task)
|
static void rpc_mark_complete_task(struct rpc_task *task)
|
{
|
{
|
smp_mb__before_clear_bit();
|
smp_mb__before_clear_bit();
|
clear_bit(RPC_TASK_ACTIVE, &task->tk_runstate);
|
clear_bit(RPC_TASK_ACTIVE, &task->tk_runstate);
|
smp_mb__after_clear_bit();
|
smp_mb__after_clear_bit();
|
wake_up_bit(&task->tk_runstate, RPC_TASK_ACTIVE);
|
wake_up_bit(&task->tk_runstate, RPC_TASK_ACTIVE);
|
}
|
}
|
|
|
/*
|
/*
|
* Allow callers to wait for completion of an RPC call
|
* Allow callers to wait for completion of an RPC call
|
*/
|
*/
|
int __rpc_wait_for_completion_task(struct rpc_task *task, int (*action)(void *))
|
int __rpc_wait_for_completion_task(struct rpc_task *task, int (*action)(void *))
|
{
|
{
|
if (action == NULL)
|
if (action == NULL)
|
action = rpc_wait_bit_interruptible;
|
action = rpc_wait_bit_interruptible;
|
return wait_on_bit(&task->tk_runstate, RPC_TASK_ACTIVE,
|
return wait_on_bit(&task->tk_runstate, RPC_TASK_ACTIVE,
|
action, TASK_INTERRUPTIBLE);
|
action, TASK_INTERRUPTIBLE);
|
}
|
}
|
EXPORT_SYMBOL(__rpc_wait_for_completion_task);
|
EXPORT_SYMBOL(__rpc_wait_for_completion_task);
|
|
|
/*
|
/*
|
* Make an RPC task runnable.
|
* Make an RPC task runnable.
|
*
|
*
|
* Note: If the task is ASYNC, this must be called with
|
* Note: If the task is ASYNC, this must be called with
|
* the spinlock held to protect the wait queue operation.
|
* the spinlock held to protect the wait queue operation.
|
*/
|
*/
|
static void rpc_make_runnable(struct rpc_task *task)
|
static void rpc_make_runnable(struct rpc_task *task)
|
{
|
{
|
BUG_ON(task->tk_timeout_fn);
|
BUG_ON(task->tk_timeout_fn);
|
rpc_clear_queued(task);
|
rpc_clear_queued(task);
|
if (rpc_test_and_set_running(task))
|
if (rpc_test_and_set_running(task))
|
return;
|
return;
|
/* We might have raced */
|
/* We might have raced */
|
if (RPC_IS_QUEUED(task)) {
|
if (RPC_IS_QUEUED(task)) {
|
rpc_clear_running(task);
|
rpc_clear_running(task);
|
return;
|
return;
|
}
|
}
|
if (RPC_IS_ASYNC(task)) {
|
if (RPC_IS_ASYNC(task)) {
|
int status;
|
int status;
|
|
|
INIT_WORK(&task->u.tk_work, rpc_async_schedule);
|
INIT_WORK(&task->u.tk_work, rpc_async_schedule);
|
status = queue_work(task->tk_workqueue, &task->u.tk_work);
|
status = queue_work(task->tk_workqueue, &task->u.tk_work);
|
if (status < 0) {
|
if (status < 0) {
|
printk(KERN_WARNING "RPC: failed to add task to queue: error: %d!\n", status);
|
printk(KERN_WARNING "RPC: failed to add task to queue: error: %d!\n", status);
|
task->tk_status = status;
|
task->tk_status = status;
|
return;
|
return;
|
}
|
}
|
} else
|
} else
|
wake_up_bit(&task->tk_runstate, RPC_TASK_QUEUED);
|
wake_up_bit(&task->tk_runstate, RPC_TASK_QUEUED);
|
}
|
}
|
|
|
/*
|
/*
|
* Prepare for sleeping on a wait queue.
|
* Prepare for sleeping on a wait queue.
|
* By always appending tasks to the list we ensure FIFO behavior.
|
* By always appending tasks to the list we ensure FIFO behavior.
|
* NB: An RPC task will only receive interrupt-driven events as long
|
* NB: An RPC task will only receive interrupt-driven events as long
|
* as it's on a wait queue.
|
* as it's on a wait queue.
|
*/
|
*/
|
static void __rpc_sleep_on(struct rpc_wait_queue *q, struct rpc_task *task,
|
static void __rpc_sleep_on(struct rpc_wait_queue *q, struct rpc_task *task,
|
rpc_action action, rpc_action timer)
|
rpc_action action, rpc_action timer)
|
{
|
{
|
dprintk("RPC: %5u sleep_on(queue \"%s\" time %lu)\n",
|
dprintk("RPC: %5u sleep_on(queue \"%s\" time %lu)\n",
|
task->tk_pid, rpc_qname(q), jiffies);
|
task->tk_pid, rpc_qname(q), jiffies);
|
|
|
if (!RPC_IS_ASYNC(task) && !RPC_IS_ACTIVATED(task)) {
|
if (!RPC_IS_ASYNC(task) && !RPC_IS_ACTIVATED(task)) {
|
printk(KERN_ERR "RPC: Inactive synchronous task put to sleep!\n");
|
printk(KERN_ERR "RPC: Inactive synchronous task put to sleep!\n");
|
return;
|
return;
|
}
|
}
|
|
|
__rpc_add_wait_queue(q, task);
|
__rpc_add_wait_queue(q, task);
|
|
|
BUG_ON(task->tk_callback != NULL);
|
BUG_ON(task->tk_callback != NULL);
|
task->tk_callback = action;
|
task->tk_callback = action;
|
__rpc_add_timer(task, timer);
|
__rpc_add_timer(task, timer);
|
}
|
}
|
|
|
void rpc_sleep_on(struct rpc_wait_queue *q, struct rpc_task *task,
|
void rpc_sleep_on(struct rpc_wait_queue *q, struct rpc_task *task,
|
rpc_action action, rpc_action timer)
|
rpc_action action, rpc_action timer)
|
{
|
{
|
/* Mark the task as being activated if so needed */
|
/* Mark the task as being activated if so needed */
|
rpc_set_active(task);
|
rpc_set_active(task);
|
|
|
/*
|
/*
|
* Protect the queue operations.
|
* Protect the queue operations.
|
*/
|
*/
|
spin_lock_bh(&q->lock);
|
spin_lock_bh(&q->lock);
|
__rpc_sleep_on(q, task, action, timer);
|
__rpc_sleep_on(q, task, action, timer);
|
spin_unlock_bh(&q->lock);
|
spin_unlock_bh(&q->lock);
|
}
|
}
|
|
|
/**
|
/**
|
* __rpc_do_wake_up_task - wake up a single rpc_task
|
* __rpc_do_wake_up_task - wake up a single rpc_task
|
* @task: task to be woken up
|
* @task: task to be woken up
|
*
|
*
|
* Caller must hold queue->lock, and have cleared the task queued flag.
|
* Caller must hold queue->lock, and have cleared the task queued flag.
|
*/
|
*/
|
static void __rpc_do_wake_up_task(struct rpc_task *task)
|
static void __rpc_do_wake_up_task(struct rpc_task *task)
|
{
|
{
|
dprintk("RPC: %5u __rpc_wake_up_task (now %lu)\n",
|
dprintk("RPC: %5u __rpc_wake_up_task (now %lu)\n",
|
task->tk_pid, jiffies);
|
task->tk_pid, jiffies);
|
|
|
#ifdef RPC_DEBUG
|
#ifdef RPC_DEBUG
|
BUG_ON(task->tk_magic != RPC_TASK_MAGIC_ID);
|
BUG_ON(task->tk_magic != RPC_TASK_MAGIC_ID);
|
#endif
|
#endif
|
/* Has the task been executed yet? If not, we cannot wake it up! */
|
/* Has the task been executed yet? If not, we cannot wake it up! */
|
if (!RPC_IS_ACTIVATED(task)) {
|
if (!RPC_IS_ACTIVATED(task)) {
|
printk(KERN_ERR "RPC: Inactive task (%p) being woken up!\n", task);
|
printk(KERN_ERR "RPC: Inactive task (%p) being woken up!\n", task);
|
return;
|
return;
|
}
|
}
|
|
|
__rpc_disable_timer(task);
|
__rpc_disable_timer(task);
|
__rpc_remove_wait_queue(task);
|
__rpc_remove_wait_queue(task);
|
|
|
rpc_make_runnable(task);
|
rpc_make_runnable(task);
|
|
|
dprintk("RPC: __rpc_wake_up_task done\n");
|
dprintk("RPC: __rpc_wake_up_task done\n");
|
}
|
}
|
|
|
/*
|
/*
|
* Wake up the specified task
|
* Wake up the specified task
|
*/
|
*/
|
static void __rpc_wake_up_task(struct rpc_task *task)
|
static void __rpc_wake_up_task(struct rpc_task *task)
|
{
|
{
|
if (rpc_start_wakeup(task)) {
|
if (rpc_start_wakeup(task)) {
|
if (RPC_IS_QUEUED(task))
|
if (RPC_IS_QUEUED(task))
|
__rpc_do_wake_up_task(task);
|
__rpc_do_wake_up_task(task);
|
rpc_finish_wakeup(task);
|
rpc_finish_wakeup(task);
|
}
|
}
|
}
|
}
|
|
|
/*
|
/*
|
* Default timeout handler if none specified by user
|
* Default timeout handler if none specified by user
|
*/
|
*/
|
static void
|
static void
|
__rpc_default_timer(struct rpc_task *task)
|
__rpc_default_timer(struct rpc_task *task)
|
{
|
{
|
dprintk("RPC: %5u timeout (default timer)\n", task->tk_pid);
|
dprintk("RPC: %5u timeout (default timer)\n", task->tk_pid);
|
task->tk_status = -ETIMEDOUT;
|
task->tk_status = -ETIMEDOUT;
|
rpc_wake_up_task(task);
|
rpc_wake_up_task(task);
|
}
|
}
|
|
|
/*
|
/*
|
* Wake up the specified task
|
* Wake up the specified task
|
*/
|
*/
|
void rpc_wake_up_task(struct rpc_task *task)
|
void rpc_wake_up_task(struct rpc_task *task)
|
{
|
{
|
rcu_read_lock_bh();
|
rcu_read_lock_bh();
|
if (rpc_start_wakeup(task)) {
|
if (rpc_start_wakeup(task)) {
|
if (RPC_IS_QUEUED(task)) {
|
if (RPC_IS_QUEUED(task)) {
|
struct rpc_wait_queue *queue = task->u.tk_wait.rpc_waitq;
|
struct rpc_wait_queue *queue = task->u.tk_wait.rpc_waitq;
|
|
|
/* Note: we're already in a bh-safe context */
|
/* Note: we're already in a bh-safe context */
|
spin_lock(&queue->lock);
|
spin_lock(&queue->lock);
|
__rpc_do_wake_up_task(task);
|
__rpc_do_wake_up_task(task);
|
spin_unlock(&queue->lock);
|
spin_unlock(&queue->lock);
|
}
|
}
|
rpc_finish_wakeup(task);
|
rpc_finish_wakeup(task);
|
}
|
}
|
rcu_read_unlock_bh();
|
rcu_read_unlock_bh();
|
}
|
}
|
|
|
/*
|
/*
|
* Wake up the next task on a priority queue.
|
* Wake up the next task on a priority queue.
|
*/
|
*/
|
static struct rpc_task * __rpc_wake_up_next_priority(struct rpc_wait_queue *queue)
|
static struct rpc_task * __rpc_wake_up_next_priority(struct rpc_wait_queue *queue)
|
{
|
{
|
struct list_head *q;
|
struct list_head *q;
|
struct rpc_task *task;
|
struct rpc_task *task;
|
|
|
/*
|
/*
|
* Service a batch of tasks from a single cookie.
|
* Service a batch of tasks from a single cookie.
|
*/
|
*/
|
q = &queue->tasks[queue->priority];
|
q = &queue->tasks[queue->priority];
|
if (!list_empty(q)) {
|
if (!list_empty(q)) {
|
task = list_entry(q->next, struct rpc_task, u.tk_wait.list);
|
task = list_entry(q->next, struct rpc_task, u.tk_wait.list);
|
if (queue->cookie == task->tk_cookie) {
|
if (queue->cookie == task->tk_cookie) {
|
if (--queue->nr)
|
if (--queue->nr)
|
goto out;
|
goto out;
|
list_move_tail(&task->u.tk_wait.list, q);
|
list_move_tail(&task->u.tk_wait.list, q);
|
}
|
}
|
/*
|
/*
|
* Check if we need to switch queues.
|
* Check if we need to switch queues.
|
*/
|
*/
|
if (--queue->count)
|
if (--queue->count)
|
goto new_cookie;
|
goto new_cookie;
|
}
|
}
|
|
|
/*
|
/*
|
* Service the next queue.
|
* Service the next queue.
|
*/
|
*/
|
do {
|
do {
|
if (q == &queue->tasks[0])
|
if (q == &queue->tasks[0])
|
q = &queue->tasks[queue->maxpriority];
|
q = &queue->tasks[queue->maxpriority];
|
else
|
else
|
q = q - 1;
|
q = q - 1;
|
if (!list_empty(q)) {
|
if (!list_empty(q)) {
|
task = list_entry(q->next, struct rpc_task, u.tk_wait.list);
|
task = list_entry(q->next, struct rpc_task, u.tk_wait.list);
|
goto new_queue;
|
goto new_queue;
|
}
|
}
|
} while (q != &queue->tasks[queue->priority]);
|
} while (q != &queue->tasks[queue->priority]);
|
|
|
rpc_reset_waitqueue_priority(queue);
|
rpc_reset_waitqueue_priority(queue);
|
return NULL;
|
return NULL;
|
|
|
new_queue:
|
new_queue:
|
rpc_set_waitqueue_priority(queue, (unsigned int)(q - &queue->tasks[0]));
|
rpc_set_waitqueue_priority(queue, (unsigned int)(q - &queue->tasks[0]));
|
new_cookie:
|
new_cookie:
|
rpc_set_waitqueue_cookie(queue, task->tk_cookie);
|
rpc_set_waitqueue_cookie(queue, task->tk_cookie);
|
out:
|
out:
|
__rpc_wake_up_task(task);
|
__rpc_wake_up_task(task);
|
return task;
|
return task;
|
}
|
}
|
|
|
/*
|
/*
|
* Wake up the next task on the wait queue.
|
* Wake up the next task on the wait queue.
|
*/
|
*/
|
struct rpc_task * rpc_wake_up_next(struct rpc_wait_queue *queue)
|
struct rpc_task * rpc_wake_up_next(struct rpc_wait_queue *queue)
|
{
|
{
|
struct rpc_task *task = NULL;
|
struct rpc_task *task = NULL;
|
|
|
dprintk("RPC: wake_up_next(%p \"%s\")\n",
|
dprintk("RPC: wake_up_next(%p \"%s\")\n",
|
queue, rpc_qname(queue));
|
queue, rpc_qname(queue));
|
rcu_read_lock_bh();
|
rcu_read_lock_bh();
|
spin_lock(&queue->lock);
|
spin_lock(&queue->lock);
|
if (RPC_IS_PRIORITY(queue))
|
if (RPC_IS_PRIORITY(queue))
|
task = __rpc_wake_up_next_priority(queue);
|
task = __rpc_wake_up_next_priority(queue);
|
else {
|
else {
|
task_for_first(task, &queue->tasks[0])
|
task_for_first(task, &queue->tasks[0])
|
__rpc_wake_up_task(task);
|
__rpc_wake_up_task(task);
|
}
|
}
|
spin_unlock(&queue->lock);
|
spin_unlock(&queue->lock);
|
rcu_read_unlock_bh();
|
rcu_read_unlock_bh();
|
|
|
return task;
|
return task;
|
}
|
}
|
|
|
/**
|
/**
|
* rpc_wake_up - wake up all rpc_tasks
|
* rpc_wake_up - wake up all rpc_tasks
|
* @queue: rpc_wait_queue on which the tasks are sleeping
|
* @queue: rpc_wait_queue on which the tasks are sleeping
|
*
|
*
|
* Grabs queue->lock
|
* Grabs queue->lock
|
*/
|
*/
|
void rpc_wake_up(struct rpc_wait_queue *queue)
|
void rpc_wake_up(struct rpc_wait_queue *queue)
|
{
|
{
|
struct rpc_task *task, *next;
|
struct rpc_task *task, *next;
|
struct list_head *head;
|
struct list_head *head;
|
|
|
rcu_read_lock_bh();
|
rcu_read_lock_bh();
|
spin_lock(&queue->lock);
|
spin_lock(&queue->lock);
|
head = &queue->tasks[queue->maxpriority];
|
head = &queue->tasks[queue->maxpriority];
|
for (;;) {
|
for (;;) {
|
list_for_each_entry_safe(task, next, head, u.tk_wait.list)
|
list_for_each_entry_safe(task, next, head, u.tk_wait.list)
|
__rpc_wake_up_task(task);
|
__rpc_wake_up_task(task);
|
if (head == &queue->tasks[0])
|
if (head == &queue->tasks[0])
|
break;
|
break;
|
head--;
|
head--;
|
}
|
}
|
spin_unlock(&queue->lock);
|
spin_unlock(&queue->lock);
|
rcu_read_unlock_bh();
|
rcu_read_unlock_bh();
|
}
|
}
|
|
|
/**
|
/**
|
* rpc_wake_up_status - wake up all rpc_tasks and set their status value.
|
* rpc_wake_up_status - wake up all rpc_tasks and set their status value.
|
* @queue: rpc_wait_queue on which the tasks are sleeping
|
* @queue: rpc_wait_queue on which the tasks are sleeping
|
* @status: status value to set
|
* @status: status value to set
|
*
|
*
|
* Grabs queue->lock
|
* Grabs queue->lock
|
*/
|
*/
|
void rpc_wake_up_status(struct rpc_wait_queue *queue, int status)
|
void rpc_wake_up_status(struct rpc_wait_queue *queue, int status)
|
{
|
{
|
struct rpc_task *task, *next;
|
struct rpc_task *task, *next;
|
struct list_head *head;
|
struct list_head *head;
|
|
|
rcu_read_lock_bh();
|
rcu_read_lock_bh();
|
spin_lock(&queue->lock);
|
spin_lock(&queue->lock);
|
head = &queue->tasks[queue->maxpriority];
|
head = &queue->tasks[queue->maxpriority];
|
for (;;) {
|
for (;;) {
|
list_for_each_entry_safe(task, next, head, u.tk_wait.list) {
|
list_for_each_entry_safe(task, next, head, u.tk_wait.list) {
|
task->tk_status = status;
|
task->tk_status = status;
|
__rpc_wake_up_task(task);
|
__rpc_wake_up_task(task);
|
}
|
}
|
if (head == &queue->tasks[0])
|
if (head == &queue->tasks[0])
|
break;
|
break;
|
head--;
|
head--;
|
}
|
}
|
spin_unlock(&queue->lock);
|
spin_unlock(&queue->lock);
|
rcu_read_unlock_bh();
|
rcu_read_unlock_bh();
|
}
|
}
|
|
|
static void __rpc_atrun(struct rpc_task *task)
|
static void __rpc_atrun(struct rpc_task *task)
|
{
|
{
|
rpc_wake_up_task(task);
|
rpc_wake_up_task(task);
|
}
|
}
|
|
|
/*
|
/*
|
* Run a task at a later time
|
* Run a task at a later time
|
*/
|
*/
|
void rpc_delay(struct rpc_task *task, unsigned long delay)
|
void rpc_delay(struct rpc_task *task, unsigned long delay)
|
{
|
{
|
task->tk_timeout = delay;
|
task->tk_timeout = delay;
|
rpc_sleep_on(&delay_queue, task, NULL, __rpc_atrun);
|
rpc_sleep_on(&delay_queue, task, NULL, __rpc_atrun);
|
}
|
}
|
|
|
/*
|
/*
|
* Helper to call task->tk_ops->rpc_call_prepare
|
* Helper to call task->tk_ops->rpc_call_prepare
|
*/
|
*/
|
static void rpc_prepare_task(struct rpc_task *task)
|
static void rpc_prepare_task(struct rpc_task *task)
|
{
|
{
|
lock_kernel();
|
lock_kernel();
|
task->tk_ops->rpc_call_prepare(task, task->tk_calldata);
|
task->tk_ops->rpc_call_prepare(task, task->tk_calldata);
|
unlock_kernel();
|
unlock_kernel();
|
}
|
}
|
|
|
/*
|
/*
|
* Helper that calls task->tk_ops->rpc_call_done if it exists
|
* Helper that calls task->tk_ops->rpc_call_done if it exists
|
*/
|
*/
|
void rpc_exit_task(struct rpc_task *task)
|
void rpc_exit_task(struct rpc_task *task)
|
{
|
{
|
task->tk_action = NULL;
|
task->tk_action = NULL;
|
if (task->tk_ops->rpc_call_done != NULL) {
|
if (task->tk_ops->rpc_call_done != NULL) {
|
lock_kernel();
|
lock_kernel();
|
task->tk_ops->rpc_call_done(task, task->tk_calldata);
|
task->tk_ops->rpc_call_done(task, task->tk_calldata);
|
unlock_kernel();
|
unlock_kernel();
|
if (task->tk_action != NULL) {
|
if (task->tk_action != NULL) {
|
WARN_ON(RPC_ASSASSINATED(task));
|
WARN_ON(RPC_ASSASSINATED(task));
|
/* Always release the RPC slot and buffer memory */
|
/* Always release the RPC slot and buffer memory */
|
xprt_release(task);
|
xprt_release(task);
|
}
|
}
|
}
|
}
|
}
|
}
|
EXPORT_SYMBOL(rpc_exit_task);
|
EXPORT_SYMBOL(rpc_exit_task);
|
|
|
void rpc_release_calldata(const struct rpc_call_ops *ops, void *calldata)
|
void rpc_release_calldata(const struct rpc_call_ops *ops, void *calldata)
|
{
|
{
|
if (ops->rpc_release != NULL) {
|
if (ops->rpc_release != NULL) {
|
lock_kernel();
|
lock_kernel();
|
ops->rpc_release(calldata);
|
ops->rpc_release(calldata);
|
unlock_kernel();
|
unlock_kernel();
|
}
|
}
|
}
|
}
|
|
|
/*
|
/*
|
* This is the RPC `scheduler' (or rather, the finite state machine).
|
* This is the RPC `scheduler' (or rather, the finite state machine).
|
*/
|
*/
|
static void __rpc_execute(struct rpc_task *task)
|
static void __rpc_execute(struct rpc_task *task)
|
{
|
{
|
int status = 0;
|
int status = 0;
|
|
|
dprintk("RPC: %5u __rpc_execute flags=0x%x\n",
|
dprintk("RPC: %5u __rpc_execute flags=0x%x\n",
|
task->tk_pid, task->tk_flags);
|
task->tk_pid, task->tk_flags);
|
|
|
BUG_ON(RPC_IS_QUEUED(task));
|
BUG_ON(RPC_IS_QUEUED(task));
|
|
|
for (;;) {
|
for (;;) {
|
/*
|
/*
|
* Garbage collection of pending timers...
|
* Garbage collection of pending timers...
|
*/
|
*/
|
rpc_delete_timer(task);
|
rpc_delete_timer(task);
|
|
|
/*
|
/*
|
* Execute any pending callback.
|
* Execute any pending callback.
|
*/
|
*/
|
if (RPC_DO_CALLBACK(task)) {
|
if (RPC_DO_CALLBACK(task)) {
|
/* Define a callback save pointer */
|
/* Define a callback save pointer */
|
void (*save_callback)(struct rpc_task *);
|
void (*save_callback)(struct rpc_task *);
|
|
|
/*
|
/*
|
* If a callback exists, save it, reset it,
|
* If a callback exists, save it, reset it,
|
* call it.
|
* call it.
|
* The save is needed to stop from resetting
|
* The save is needed to stop from resetting
|
* another callback set within the callback handler
|
* another callback set within the callback handler
|
* - Dave
|
* - Dave
|
*/
|
*/
|
save_callback=task->tk_callback;
|
save_callback=task->tk_callback;
|
task->tk_callback=NULL;
|
task->tk_callback=NULL;
|
save_callback(task);
|
save_callback(task);
|
}
|
}
|
|
|
/*
|
/*
|
* Perform the next FSM step.
|
* Perform the next FSM step.
|
* tk_action may be NULL when the task has been killed
|
* tk_action may be NULL when the task has been killed
|
* by someone else.
|
* by someone else.
|
*/
|
*/
|
if (!RPC_IS_QUEUED(task)) {
|
if (!RPC_IS_QUEUED(task)) {
|
if (task->tk_action == NULL)
|
if (task->tk_action == NULL)
|
break;
|
break;
|
task->tk_action(task);
|
task->tk_action(task);
|
}
|
}
|
|
|
/*
|
/*
|
* Lockless check for whether task is sleeping or not.
|
* Lockless check for whether task is sleeping or not.
|
*/
|
*/
|
if (!RPC_IS_QUEUED(task))
|
if (!RPC_IS_QUEUED(task))
|
continue;
|
continue;
|
rpc_clear_running(task);
|
rpc_clear_running(task);
|
if (RPC_IS_ASYNC(task)) {
|
if (RPC_IS_ASYNC(task)) {
|
/* Careful! we may have raced... */
|
/* Careful! we may have raced... */
|
if (RPC_IS_QUEUED(task))
|
if (RPC_IS_QUEUED(task))
|
return;
|
return;
|
if (rpc_test_and_set_running(task))
|
if (rpc_test_and_set_running(task))
|
return;
|
return;
|
continue;
|
continue;
|
}
|
}
|
|
|
/* sync task: sleep here */
|
/* sync task: sleep here */
|
dprintk("RPC: %5u sync task going to sleep\n", task->tk_pid);
|
dprintk("RPC: %5u sync task going to sleep\n", task->tk_pid);
|
/* Note: Caller should be using rpc_clnt_sigmask() */
|
/* Note: Caller should be using rpc_clnt_sigmask() */
|
status = out_of_line_wait_on_bit(&task->tk_runstate,
|
status = out_of_line_wait_on_bit(&task->tk_runstate,
|
RPC_TASK_QUEUED, rpc_wait_bit_interruptible,
|
RPC_TASK_QUEUED, rpc_wait_bit_interruptible,
|
TASK_INTERRUPTIBLE);
|
TASK_INTERRUPTIBLE);
|
if (status == -ERESTARTSYS) {
|
if (status == -ERESTARTSYS) {
|
/*
|
/*
|
* When a sync task receives a signal, it exits with
|
* When a sync task receives a signal, it exits with
|
* -ERESTARTSYS. In order to catch any callbacks that
|
* -ERESTARTSYS. In order to catch any callbacks that
|
* clean up after sleeping on some queue, we don't
|
* clean up after sleeping on some queue, we don't
|
* break the loop here, but go around once more.
|
* break the loop here, but go around once more.
|
*/
|
*/
|
dprintk("RPC: %5u got signal\n", task->tk_pid);
|
dprintk("RPC: %5u got signal\n", task->tk_pid);
|
task->tk_flags |= RPC_TASK_KILLED;
|
task->tk_flags |= RPC_TASK_KILLED;
|
rpc_exit(task, -ERESTARTSYS);
|
rpc_exit(task, -ERESTARTSYS);
|
rpc_wake_up_task(task);
|
rpc_wake_up_task(task);
|
}
|
}
|
rpc_set_running(task);
|
rpc_set_running(task);
|
dprintk("RPC: %5u sync task resuming\n", task->tk_pid);
|
dprintk("RPC: %5u sync task resuming\n", task->tk_pid);
|
}
|
}
|
|
|
dprintk("RPC: %5u return %d, status %d\n", task->tk_pid, status,
|
dprintk("RPC: %5u return %d, status %d\n", task->tk_pid, status,
|
task->tk_status);
|
task->tk_status);
|
/* Release all resources associated with the task */
|
/* Release all resources associated with the task */
|
rpc_release_task(task);
|
rpc_release_task(task);
|
}
|
}
|
|
|
/*
|
/*
|
* User-visible entry point to the scheduler.
|
* User-visible entry point to the scheduler.
|
*
|
*
|
* This may be called recursively if e.g. an async NFS task updates
|
* This may be called recursively if e.g. an async NFS task updates
|
* the attributes and finds that dirty pages must be flushed.
|
* the attributes and finds that dirty pages must be flushed.
|
* NOTE: Upon exit of this function the task is guaranteed to be
|
* NOTE: Upon exit of this function the task is guaranteed to be
|
* released. In particular note that tk_release() will have
|
* released. In particular note that tk_release() will have
|
* been called, so your task memory may have been freed.
|
* been called, so your task memory may have been freed.
|
*/
|
*/
|
void rpc_execute(struct rpc_task *task)
|
void rpc_execute(struct rpc_task *task)
|
{
|
{
|
rpc_set_active(task);
|
rpc_set_active(task);
|
rpc_set_running(task);
|
rpc_set_running(task);
|
__rpc_execute(task);
|
__rpc_execute(task);
|
}
|
}
|
|
|
static void rpc_async_schedule(struct work_struct *work)
|
static void rpc_async_schedule(struct work_struct *work)
|
{
|
{
|
__rpc_execute(container_of(work, struct rpc_task, u.tk_work));
|
__rpc_execute(container_of(work, struct rpc_task, u.tk_work));
|
}
|
}
|
|
|
struct rpc_buffer {
|
struct rpc_buffer {
|
size_t len;
|
size_t len;
|
char data[];
|
char data[];
|
};
|
};
|
|
|
/**
|
/**
|
* rpc_malloc - allocate an RPC buffer
|
* rpc_malloc - allocate an RPC buffer
|
* @task: RPC task that will use this buffer
|
* @task: RPC task that will use this buffer
|
* @size: requested byte size
|
* @size: requested byte size
|
*
|
*
|
* To prevent rpciod from hanging, this allocator never sleeps,
|
* To prevent rpciod from hanging, this allocator never sleeps,
|
* returning NULL if the request cannot be serviced immediately.
|
* returning NULL if the request cannot be serviced immediately.
|
* The caller can arrange to sleep in a way that is safe for rpciod.
|
* The caller can arrange to sleep in a way that is safe for rpciod.
|
*
|
*
|
* Most requests are 'small' (under 2KiB) and can be serviced from a
|
* Most requests are 'small' (under 2KiB) and can be serviced from a
|
* mempool, ensuring that NFS reads and writes can always proceed,
|
* mempool, ensuring that NFS reads and writes can always proceed,
|
* and that there is good locality of reference for these buffers.
|
* and that there is good locality of reference for these buffers.
|
*
|
*
|
* In order to avoid memory starvation triggering more writebacks of
|
* In order to avoid memory starvation triggering more writebacks of
|
* NFS requests, we avoid using GFP_KERNEL.
|
* NFS requests, we avoid using GFP_KERNEL.
|
*/
|
*/
|
void *rpc_malloc(struct rpc_task *task, size_t size)
|
void *rpc_malloc(struct rpc_task *task, size_t size)
|
{
|
{
|
struct rpc_buffer *buf;
|
struct rpc_buffer *buf;
|
gfp_t gfp = RPC_IS_SWAPPER(task) ? GFP_ATOMIC : GFP_NOWAIT;
|
gfp_t gfp = RPC_IS_SWAPPER(task) ? GFP_ATOMIC : GFP_NOWAIT;
|
|
|
size += sizeof(struct rpc_buffer);
|
size += sizeof(struct rpc_buffer);
|
if (size <= RPC_BUFFER_MAXSIZE)
|
if (size <= RPC_BUFFER_MAXSIZE)
|
buf = mempool_alloc(rpc_buffer_mempool, gfp);
|
buf = mempool_alloc(rpc_buffer_mempool, gfp);
|
else
|
else
|
buf = kmalloc(size, gfp);
|
buf = kmalloc(size, gfp);
|
|
|
if (!buf)
|
if (!buf)
|
return NULL;
|
return NULL;
|
|
|
buf->len = size;
|
buf->len = size;
|
dprintk("RPC: %5u allocated buffer of size %zu at %p\n",
|
dprintk("RPC: %5u allocated buffer of size %zu at %p\n",
|
task->tk_pid, size, buf);
|
task->tk_pid, size, buf);
|
return &buf->data;
|
return &buf->data;
|
}
|
}
|
EXPORT_SYMBOL_GPL(rpc_malloc);
|
EXPORT_SYMBOL_GPL(rpc_malloc);
|
|
|
/**
|
/**
|
* rpc_free - free buffer allocated via rpc_malloc
|
* rpc_free - free buffer allocated via rpc_malloc
|
* @buffer: buffer to free
|
* @buffer: buffer to free
|
*
|
*
|
*/
|
*/
|
void rpc_free(void *buffer)
|
void rpc_free(void *buffer)
|
{
|
{
|
size_t size;
|
size_t size;
|
struct rpc_buffer *buf;
|
struct rpc_buffer *buf;
|
|
|
if (!buffer)
|
if (!buffer)
|
return;
|
return;
|
|
|
buf = container_of(buffer, struct rpc_buffer, data);
|
buf = container_of(buffer, struct rpc_buffer, data);
|
size = buf->len;
|
size = buf->len;
|
|
|
dprintk("RPC: freeing buffer of size %zu at %p\n",
|
dprintk("RPC: freeing buffer of size %zu at %p\n",
|
size, buf);
|
size, buf);
|
|
|
if (size <= RPC_BUFFER_MAXSIZE)
|
if (size <= RPC_BUFFER_MAXSIZE)
|
mempool_free(buf, rpc_buffer_mempool);
|
mempool_free(buf, rpc_buffer_mempool);
|
else
|
else
|
kfree(buf);
|
kfree(buf);
|
}
|
}
|
EXPORT_SYMBOL_GPL(rpc_free);
|
EXPORT_SYMBOL_GPL(rpc_free);
|
|
|
/*
|
/*
|
* Creation and deletion of RPC task structures
|
* Creation and deletion of RPC task structures
|
*/
|
*/
|
void rpc_init_task(struct rpc_task *task, struct rpc_clnt *clnt, int flags, const struct rpc_call_ops *tk_ops, void *calldata)
|
void rpc_init_task(struct rpc_task *task, struct rpc_clnt *clnt, int flags, const struct rpc_call_ops *tk_ops, void *calldata)
|
{
|
{
|
memset(task, 0, sizeof(*task));
|
memset(task, 0, sizeof(*task));
|
init_timer(&task->tk_timer);
|
init_timer(&task->tk_timer);
|
task->tk_timer.data = (unsigned long) task;
|
task->tk_timer.data = (unsigned long) task;
|
task->tk_timer.function = (void (*)(unsigned long)) rpc_run_timer;
|
task->tk_timer.function = (void (*)(unsigned long)) rpc_run_timer;
|
atomic_set(&task->tk_count, 1);
|
atomic_set(&task->tk_count, 1);
|
task->tk_client = clnt;
|
task->tk_client = clnt;
|
task->tk_flags = flags;
|
task->tk_flags = flags;
|
task->tk_ops = tk_ops;
|
task->tk_ops = tk_ops;
|
if (tk_ops->rpc_call_prepare != NULL)
|
if (tk_ops->rpc_call_prepare != NULL)
|
task->tk_action = rpc_prepare_task;
|
task->tk_action = rpc_prepare_task;
|
task->tk_calldata = calldata;
|
task->tk_calldata = calldata;
|
INIT_LIST_HEAD(&task->tk_task);
|
INIT_LIST_HEAD(&task->tk_task);
|
|
|
/* Initialize retry counters */
|
/* Initialize retry counters */
|
task->tk_garb_retry = 2;
|
task->tk_garb_retry = 2;
|
task->tk_cred_retry = 2;
|
task->tk_cred_retry = 2;
|
|
|
task->tk_priority = RPC_PRIORITY_NORMAL;
|
task->tk_priority = RPC_PRIORITY_NORMAL;
|
task->tk_cookie = (unsigned long)current;
|
task->tk_cookie = (unsigned long)current;
|
|
|
/* Initialize workqueue for async tasks */
|
/* Initialize workqueue for async tasks */
|
task->tk_workqueue = rpciod_workqueue;
|
task->tk_workqueue = rpciod_workqueue;
|
|
|
if (clnt) {
|
if (clnt) {
|
kref_get(&clnt->cl_kref);
|
kref_get(&clnt->cl_kref);
|
if (clnt->cl_softrtry)
|
if (clnt->cl_softrtry)
|
task->tk_flags |= RPC_TASK_SOFT;
|
task->tk_flags |= RPC_TASK_SOFT;
|
if (!clnt->cl_intr)
|
if (!clnt->cl_intr)
|
task->tk_flags |= RPC_TASK_NOINTR;
|
task->tk_flags |= RPC_TASK_NOINTR;
|
}
|
}
|
|
|
BUG_ON(task->tk_ops == NULL);
|
BUG_ON(task->tk_ops == NULL);
|
|
|
/* starting timestamp */
|
/* starting timestamp */
|
task->tk_start = jiffies;
|
task->tk_start = jiffies;
|
|
|
dprintk("RPC: new task initialized, procpid %u\n",
|
dprintk("RPC: new task initialized, procpid %u\n",
|
task_pid_nr(current));
|
task_pid_nr(current));
|
}
|
}
|
|
|
static struct rpc_task *
|
static struct rpc_task *
|
rpc_alloc_task(void)
|
rpc_alloc_task(void)
|
{
|
{
|
return (struct rpc_task *)mempool_alloc(rpc_task_mempool, GFP_NOFS);
|
return (struct rpc_task *)mempool_alloc(rpc_task_mempool, GFP_NOFS);
|
}
|
}
|
|
|
static void rpc_free_task(struct rcu_head *rcu)
|
static void rpc_free_task(struct rcu_head *rcu)
|
{
|
{
|
struct rpc_task *task = container_of(rcu, struct rpc_task, u.tk_rcu);
|
struct rpc_task *task = container_of(rcu, struct rpc_task, u.tk_rcu);
|
dprintk("RPC: %5u freeing task\n", task->tk_pid);
|
dprintk("RPC: %5u freeing task\n", task->tk_pid);
|
mempool_free(task, rpc_task_mempool);
|
mempool_free(task, rpc_task_mempool);
|
}
|
}
|
|
|
/*
|
/*
|
* Create a new task for the specified client.
|
* Create a new task for the specified client.
|
*/
|
*/
|
struct rpc_task *rpc_new_task(struct rpc_clnt *clnt, int flags, const struct rpc_call_ops *tk_ops, void *calldata)
|
struct rpc_task *rpc_new_task(struct rpc_clnt *clnt, int flags, const struct rpc_call_ops *tk_ops, void *calldata)
|
{
|
{
|
struct rpc_task *task;
|
struct rpc_task *task;
|
|
|
task = rpc_alloc_task();
|
task = rpc_alloc_task();
|
if (!task)
|
if (!task)
|
goto out;
|
goto out;
|
|
|
rpc_init_task(task, clnt, flags, tk_ops, calldata);
|
rpc_init_task(task, clnt, flags, tk_ops, calldata);
|
|
|
dprintk("RPC: allocated task %p\n", task);
|
dprintk("RPC: allocated task %p\n", task);
|
task->tk_flags |= RPC_TASK_DYNAMIC;
|
task->tk_flags |= RPC_TASK_DYNAMIC;
|
out:
|
out:
|
return task;
|
return task;
|
}
|
}
|
|
|
|
|
void rpc_put_task(struct rpc_task *task)
|
void rpc_put_task(struct rpc_task *task)
|
{
|
{
|
const struct rpc_call_ops *tk_ops = task->tk_ops;
|
const struct rpc_call_ops *tk_ops = task->tk_ops;
|
void *calldata = task->tk_calldata;
|
void *calldata = task->tk_calldata;
|
|
|
if (!atomic_dec_and_test(&task->tk_count))
|
if (!atomic_dec_and_test(&task->tk_count))
|
return;
|
return;
|
/* Release resources */
|
/* Release resources */
|
if (task->tk_rqstp)
|
if (task->tk_rqstp)
|
xprt_release(task);
|
xprt_release(task);
|
if (task->tk_msg.rpc_cred)
|
if (task->tk_msg.rpc_cred)
|
rpcauth_unbindcred(task);
|
rpcauth_unbindcred(task);
|
if (task->tk_client) {
|
if (task->tk_client) {
|
rpc_release_client(task->tk_client);
|
rpc_release_client(task->tk_client);
|
task->tk_client = NULL;
|
task->tk_client = NULL;
|
}
|
}
|
if (task->tk_flags & RPC_TASK_DYNAMIC)
|
if (task->tk_flags & RPC_TASK_DYNAMIC)
|
call_rcu_bh(&task->u.tk_rcu, rpc_free_task);
|
call_rcu_bh(&task->u.tk_rcu, rpc_free_task);
|
rpc_release_calldata(tk_ops, calldata);
|
rpc_release_calldata(tk_ops, calldata);
|
}
|
}
|
EXPORT_SYMBOL(rpc_put_task);
|
EXPORT_SYMBOL(rpc_put_task);
|
|
|
static void rpc_release_task(struct rpc_task *task)
|
static void rpc_release_task(struct rpc_task *task)
|
{
|
{
|
#ifdef RPC_DEBUG
|
#ifdef RPC_DEBUG
|
BUG_ON(task->tk_magic != RPC_TASK_MAGIC_ID);
|
BUG_ON(task->tk_magic != RPC_TASK_MAGIC_ID);
|
#endif
|
#endif
|
dprintk("RPC: %5u release task\n", task->tk_pid);
|
dprintk("RPC: %5u release task\n", task->tk_pid);
|
|
|
if (!list_empty(&task->tk_task)) {
|
if (!list_empty(&task->tk_task)) {
|
struct rpc_clnt *clnt = task->tk_client;
|
struct rpc_clnt *clnt = task->tk_client;
|
/* Remove from client task list */
|
/* Remove from client task list */
|
spin_lock(&clnt->cl_lock);
|
spin_lock(&clnt->cl_lock);
|
list_del(&task->tk_task);
|
list_del(&task->tk_task);
|
spin_unlock(&clnt->cl_lock);
|
spin_unlock(&clnt->cl_lock);
|
}
|
}
|
BUG_ON (RPC_IS_QUEUED(task));
|
BUG_ON (RPC_IS_QUEUED(task));
|
|
|
/* Synchronously delete any running timer */
|
/* Synchronously delete any running timer */
|
rpc_delete_timer(task);
|
rpc_delete_timer(task);
|
|
|
#ifdef RPC_DEBUG
|
#ifdef RPC_DEBUG
|
task->tk_magic = 0;
|
task->tk_magic = 0;
|
#endif
|
#endif
|
/* Wake up anyone who is waiting for task completion */
|
/* Wake up anyone who is waiting for task completion */
|
rpc_mark_complete_task(task);
|
rpc_mark_complete_task(task);
|
|
|
rpc_put_task(task);
|
rpc_put_task(task);
|
}
|
}
|
|
|
/*
|
/*
|
* Kill all tasks for the given client.
|
* Kill all tasks for the given client.
|
* XXX: kill their descendants as well?
|
* XXX: kill their descendants as well?
|
*/
|
*/
|
void rpc_killall_tasks(struct rpc_clnt *clnt)
|
void rpc_killall_tasks(struct rpc_clnt *clnt)
|
{
|
{
|
struct rpc_task *rovr;
|
struct rpc_task *rovr;
|
|
|
|
|
if (list_empty(&clnt->cl_tasks))
|
if (list_empty(&clnt->cl_tasks))
|
return;
|
return;
|
dprintk("RPC: killing all tasks for client %p\n", clnt);
|
dprintk("RPC: killing all tasks for client %p\n", clnt);
|
/*
|
/*
|
* Spin lock all_tasks to prevent changes...
|
* Spin lock all_tasks to prevent changes...
|
*/
|
*/
|
spin_lock(&clnt->cl_lock);
|
spin_lock(&clnt->cl_lock);
|
list_for_each_entry(rovr, &clnt->cl_tasks, tk_task) {
|
list_for_each_entry(rovr, &clnt->cl_tasks, tk_task) {
|
if (! RPC_IS_ACTIVATED(rovr))
|
if (! RPC_IS_ACTIVATED(rovr))
|
continue;
|
continue;
|
if (!(rovr->tk_flags & RPC_TASK_KILLED)) {
|
if (!(rovr->tk_flags & RPC_TASK_KILLED)) {
|
rovr->tk_flags |= RPC_TASK_KILLED;
|
rovr->tk_flags |= RPC_TASK_KILLED;
|
rpc_exit(rovr, -EIO);
|
rpc_exit(rovr, -EIO);
|
rpc_wake_up_task(rovr);
|
rpc_wake_up_task(rovr);
|
}
|
}
|
}
|
}
|
spin_unlock(&clnt->cl_lock);
|
spin_unlock(&clnt->cl_lock);
|
}
|
}
|
|
|
int rpciod_up(void)
|
int rpciod_up(void)
|
{
|
{
|
return try_module_get(THIS_MODULE) ? 0 : -EINVAL;
|
return try_module_get(THIS_MODULE) ? 0 : -EINVAL;
|
}
|
}
|
|
|
void rpciod_down(void)
|
void rpciod_down(void)
|
{
|
{
|
module_put(THIS_MODULE);
|
module_put(THIS_MODULE);
|
}
|
}
|
|
|
/*
|
/*
|
* Start up the rpciod workqueue.
|
* Start up the rpciod workqueue.
|
*/
|
*/
|
static int rpciod_start(void)
|
static int rpciod_start(void)
|
{
|
{
|
struct workqueue_struct *wq;
|
struct workqueue_struct *wq;
|
|
|
/*
|
/*
|
* Create the rpciod thread and wait for it to start.
|
* Create the rpciod thread and wait for it to start.
|
*/
|
*/
|
dprintk("RPC: creating workqueue rpciod\n");
|
dprintk("RPC: creating workqueue rpciod\n");
|
wq = create_workqueue("rpciod");
|
wq = create_workqueue("rpciod");
|
rpciod_workqueue = wq;
|
rpciod_workqueue = wq;
|
return rpciod_workqueue != NULL;
|
return rpciod_workqueue != NULL;
|
}
|
}
|
|
|
static void rpciod_stop(void)
|
static void rpciod_stop(void)
|
{
|
{
|
struct workqueue_struct *wq = NULL;
|
struct workqueue_struct *wq = NULL;
|
|
|
if (rpciod_workqueue == NULL)
|
if (rpciod_workqueue == NULL)
|
return;
|
return;
|
dprintk("RPC: destroying workqueue rpciod\n");
|
dprintk("RPC: destroying workqueue rpciod\n");
|
|
|
wq = rpciod_workqueue;
|
wq = rpciod_workqueue;
|
rpciod_workqueue = NULL;
|
rpciod_workqueue = NULL;
|
destroy_workqueue(wq);
|
destroy_workqueue(wq);
|
}
|
}
|
|
|
void
|
void
|
rpc_destroy_mempool(void)
|
rpc_destroy_mempool(void)
|
{
|
{
|
rpciod_stop();
|
rpciod_stop();
|
if (rpc_buffer_mempool)
|
if (rpc_buffer_mempool)
|
mempool_destroy(rpc_buffer_mempool);
|
mempool_destroy(rpc_buffer_mempool);
|
if (rpc_task_mempool)
|
if (rpc_task_mempool)
|
mempool_destroy(rpc_task_mempool);
|
mempool_destroy(rpc_task_mempool);
|
if (rpc_task_slabp)
|
if (rpc_task_slabp)
|
kmem_cache_destroy(rpc_task_slabp);
|
kmem_cache_destroy(rpc_task_slabp);
|
if (rpc_buffer_slabp)
|
if (rpc_buffer_slabp)
|
kmem_cache_destroy(rpc_buffer_slabp);
|
kmem_cache_destroy(rpc_buffer_slabp);
|
}
|
}
|
|
|
int
|
int
|
rpc_init_mempool(void)
|
rpc_init_mempool(void)
|
{
|
{
|
rpc_task_slabp = kmem_cache_create("rpc_tasks",
|
rpc_task_slabp = kmem_cache_create("rpc_tasks",
|
sizeof(struct rpc_task),
|
sizeof(struct rpc_task),
|
0, SLAB_HWCACHE_ALIGN,
|
0, SLAB_HWCACHE_ALIGN,
|
NULL);
|
NULL);
|
if (!rpc_task_slabp)
|
if (!rpc_task_slabp)
|
goto err_nomem;
|
goto err_nomem;
|
rpc_buffer_slabp = kmem_cache_create("rpc_buffers",
|
rpc_buffer_slabp = kmem_cache_create("rpc_buffers",
|
RPC_BUFFER_MAXSIZE,
|
RPC_BUFFER_MAXSIZE,
|
0, SLAB_HWCACHE_ALIGN,
|
0, SLAB_HWCACHE_ALIGN,
|
NULL);
|
NULL);
|
if (!rpc_buffer_slabp)
|
if (!rpc_buffer_slabp)
|
goto err_nomem;
|
goto err_nomem;
|
rpc_task_mempool = mempool_create_slab_pool(RPC_TASK_POOLSIZE,
|
rpc_task_mempool = mempool_create_slab_pool(RPC_TASK_POOLSIZE,
|
rpc_task_slabp);
|
rpc_task_slabp);
|
if (!rpc_task_mempool)
|
if (!rpc_task_mempool)
|
goto err_nomem;
|
goto err_nomem;
|
rpc_buffer_mempool = mempool_create_slab_pool(RPC_BUFFER_POOLSIZE,
|
rpc_buffer_mempool = mempool_create_slab_pool(RPC_BUFFER_POOLSIZE,
|
rpc_buffer_slabp);
|
rpc_buffer_slabp);
|
if (!rpc_buffer_mempool)
|
if (!rpc_buffer_mempool)
|
goto err_nomem;
|
goto err_nomem;
|
if (!rpciod_start())
|
if (!rpciod_start())
|
goto err_nomem;
|
goto err_nomem;
|
return 0;
|
return 0;
|
err_nomem:
|
err_nomem:
|
rpc_destroy_mempool();
|
rpc_destroy_mempool();
|
return -ENOMEM;
|
return -ENOMEM;
|
}
|
}
|
|
|
