sched.c

Linux内核源代码 为压缩文件 是<<Linux内核>>一书中的源代码

			 */
			if (task->tk_client->cl_intr && signalled()) {
				dprintk("RPC: %4d got signal\n", task->tk_pid);
				task->tk_flags |= RPC_TASK_KILLED;
				rpc_exit(task, -ERESTARTSYS);
				rpc_wake_up_task(task);
			}
		}
	}

	if (task->tk_exit) {
		task->tk_exit(task);
		/* If tk_action is non-null, the user wants us to restart */
		if (task->tk_action) {
			if (!RPC_ASSASSINATED(task)) {
				/* Release RPC slot and buffer memory */
				if (task->tk_rqstp)
					xprt_release(task);
				if (task->tk_buffer) {
					rpc_free(task->tk_buffer);
					task->tk_buffer = NULL;
				}
				goto restarted;
			}
			printk(KERN_ERR "RPC: dead task tries to walk away.\n");
		}
	}

	dprintk("RPC: %4d exit() = %d\n", task->tk_pid, task->tk_status);
	status = task->tk_status;

	/* Release all resources associated with the task */
	rpc_release_task(task);

	return status;
}

/*
 * User-visible entry point to the scheduler.
 *
 * This may be called recursively if e.g. an async NFS task updates
 * the attributes and finds that dirty pages must be flushed.
 * NOTE: Upon exit of this function the task is guaranteed to be
 *	 released. In particular note that tk_release() will have
 *	 been called, so your task memory may have been freed.
 */
int
rpc_execute(struct rpc_task *task)
{
	int status = -EIO;
	if (rpc_inhibit) {
		printk(KERN_INFO "RPC: execution inhibited!\n");
		goto out_release;
	}

	status = -EWOULDBLOCK;
	if (task->tk_active) {
		printk(KERN_ERR "RPC: active task was run twice!\n");
		goto out_err;
	}

	task->tk_active = 1;
	rpc_set_running(task);
	return __rpc_execute(task);
 out_release:
	rpc_release_task(task);
 out_err:
	return status;
}

/*
 * This is our own little scheduler for async RPC tasks.
 */
static void
__rpc_schedule(void)
{
	struct rpc_task	*task;
	int		count = 0;

	dprintk("RPC:      rpc_schedule enter\n");
	while (1) {
		/* Ensure equal rights for tcp tasks... */
		rpciod_tcp_dispatcher();

		spin_lock_bh(&rpc_queue_lock);
		if (!(task = schedq.task)) {
			spin_unlock_bh(&rpc_queue_lock);
			break;
		}
		if (task->tk_lock) {
			spin_unlock_bh(&rpc_queue_lock);
			printk(KERN_ERR "RPC: Locked task was scheduled !!!!\n");
#ifdef RPC_DEBUG			
			rpc_debug = ~0;
			rpc_show_tasks();
#endif			
			break;
		}
		__rpc_remove_wait_queue(task);
		spin_unlock_bh(&rpc_queue_lock);

		__rpc_execute(task);

		if (++count >= 200 || current->need_resched) {
			count = 0;
			schedule();
		}
	}
	dprintk("RPC:      rpc_schedule leave\n");
}

/*
 * Allocate memory for RPC purpose.
 *
 * This is yet another tricky issue: For sync requests issued by
 * a user process, we want to make kmalloc sleep if there isn't
 * enough memory. Async requests should not sleep too excessively
 * because that will block rpciod (but that's not dramatic when
 * it's starved of memory anyway). Finally, swapout requests should
 * never sleep at all, and should not trigger another swap_out
 * request through kmalloc which would just increase memory contention.
 *
 * I hope the following gets it right, which gives async requests
 * a slight advantage over sync requests (good for writeback, debatable
 * for readahead):
 *
 *   sync user requests:	GFP_KERNEL
 *   async requests:		GFP_RPC		(== GFP_NFS)
 *   swap requests:		GFP_ATOMIC	(or new GFP_SWAPPER)
 */
void *
rpc_allocate(unsigned int flags, unsigned int size)
{
	u32	*buffer;
	int	gfp;

	if (flags & RPC_TASK_SWAPPER)
		gfp = GFP_ATOMIC;
	else if (flags & RPC_TASK_ASYNC)
		gfp = GFP_RPC;
	else
		gfp = GFP_KERNEL;

	do {
		if ((buffer = (u32 *) kmalloc(size, gfp)) != NULL) {
			dprintk("RPC:      allocated buffer %p\n", buffer);
			return buffer;
		}
		if ((flags & RPC_TASK_SWAPPER) && size <= sizeof(swap_buffer)
		    && rpc_lock_swapbuf()) {
			dprintk("RPC:      used last-ditch swap buffer\n");
			return swap_buffer;
		}
		if (flags & RPC_TASK_ASYNC)
			return NULL;
		set_current_state(TASK_INTERRUPTIBLE);
		schedule_timeout(HZ>>4);
	} while (!signalled());

	return NULL;
}

void
rpc_free(void *buffer)
{
	if (buffer != swap_buffer) {
		kfree(buffer);
		return;
	}
	rpc_unlock_swapbuf();
}

/*
 * Creation and deletion of RPC task structures
 */
inline void
rpc_init_task(struct rpc_task *task, struct rpc_clnt *clnt,
				rpc_action callback, int flags)
{
	memset(task, 0, sizeof(*task));
	init_timer(&task->tk_timer);
	task->tk_timer.data     = (unsigned long) task;
	task->tk_timer.function = (void (*)(unsigned long)) rpc_run_timer;
	task->tk_client = clnt;
	task->tk_flags  = flags;
	task->tk_exit   = callback;
	init_waitqueue_head(&task->tk_wait);
	if (current->uid != current->fsuid || current->gid != current->fsgid)
		task->tk_flags |= RPC_TASK_SETUID;

	/* Initialize retry counters */
	task->tk_garb_retry = 2;
	task->tk_cred_retry = 2;
	task->tk_suid_retry = 1;

	/* Add to global list of all tasks */
	spin_lock(&rpc_sched_lock);
	task->tk_next_task = all_tasks;
	task->tk_prev_task = NULL;
	if (all_tasks)
		all_tasks->tk_prev_task = task;
	all_tasks = task;
	spin_unlock(&rpc_sched_lock);

	if (clnt)
		atomic_inc(&clnt->cl_users);

#ifdef RPC_DEBUG
	task->tk_magic = 0xf00baa;
	task->tk_pid = rpc_task_id++;
#endif
	dprintk("RPC: %4d new task procpid %d\n", task->tk_pid,
				current->pid);
}

static void
rpc_default_free_task(struct rpc_task *task)
{
	dprintk("RPC: %4d freeing task\n", task->tk_pid);
	rpc_free(task);
}

/*
 * Create a new task for the specified client.  We have to
 * clean up after an allocation failure, as the client may
 * have specified "oneshot".
 */
struct rpc_task *
rpc_new_task(struct rpc_clnt *clnt, rpc_action callback, int flags)
{
	struct rpc_task	*task;

	task = (struct rpc_task *) rpc_allocate(flags, sizeof(*task));
	if (!task)
		goto cleanup;

	rpc_init_task(task, clnt, callback, flags);

	/* Replace tk_release */
	task->tk_release = rpc_default_free_task;

	dprintk("RPC: %4d allocated task\n", task->tk_pid);
	task->tk_flags |= RPC_TASK_DYNAMIC;
out:
	return task;

cleanup:
	/* Check whether to release the client */
	if (clnt) {
		printk("rpc_new_task: failed, users=%d, oneshot=%d\n",
			atomic_read(&clnt->cl_users), clnt->cl_oneshot);
		atomic_inc(&clnt->cl_users); /* pretend we were used ... */
		rpc_release_client(clnt);
	}
	goto out;
}

void
rpc_release_task(struct rpc_task *task)
{
	struct rpc_task	*next, *prev;

	dprintk("RPC: %4d release task\n", task->tk_pid);

#ifdef RPC_DEBUG
	if (task->tk_magic != 0xf00baa) {
		printk(KERN_ERR "RPC: attempt to release a non-existing task!\n");
		rpc_debug = ~0;
		rpc_show_tasks();
		return;
	}
#endif

	/* Remove from global task list */
	spin_lock(&rpc_sched_lock);
	prev = task->tk_prev_task;
	next = task->tk_next_task;
	if (next)
		next->tk_prev_task = prev;
	if (prev)
		prev->tk_next_task = next;
	else
		all_tasks = next;
	task->tk_next_task = task->tk_prev_task = NULL;
	spin_unlock(&rpc_sched_lock);

	/* Protect the execution below. */
	spin_lock_bh(&rpc_queue_lock);

	/* Disable timer to prevent zombie wakeup */
	__rpc_disable_timer(task);

	/* Remove from any wait queue we're still on */
	__rpc_remove_wait_queue(task);

	task->tk_active = 0;

	spin_unlock_bh(&rpc_queue_lock);

	/* Synchronously delete any running timer */
	rpc_delete_timer(task);

	/* Release resources */
	if (task->tk_rqstp)
		xprt_release(task);
	if (task->tk_msg.rpc_cred)
		rpcauth_unbindcred(task);
	if (task->tk_buffer) {
		rpc_free(task->tk_buffer);
		task->tk_buffer = NULL;
	}
	if (task->tk_client) {
		rpc_release_client(task->tk_client);
		task->tk_client = NULL;
	}

#ifdef RPC_DEBUG
	task->tk_magic = 0;
#endif
	if (task->tk_release)
		task->tk_release(task);
}

/**
 * rpc_find_parent - find the parent of a child task.
 * @child: child task
 *
 * Checks that the parent task is still sleeping on the
 * queue 'childq'. If so returns a pointer to the parent.
 * Upon failure returns NULL.
 *
 * Caller must hold rpc_queue_lock
 */
static inline struct rpc_task *
rpc_find_parent(struct rpc_task *child)
{
	struct rpc_task	*task, *parent;

	parent = (struct rpc_task *) child->tk_calldata;
	if ((task = childq.task) != NULL) {
		do {
			if (task == parent)
				return parent;
		} while ((task = task->tk_next) != childq.task);
	}
	return NULL;
}

static void
rpc_child_exit(struct rpc_task *child)
{
	struct rpc_task	*parent;

	spin_lock_bh(&rpc_queue_lock);
	if ((parent = rpc_find_parent(child)) != NULL) {
		parent->tk_status = child->tk_status;
		__rpc_wake_up_task(parent);
	}
	spin_unlock_bh(&rpc_queue_lock);
}

/*
 * Note: rpc_new_task releases the client after a failure.
 */
struct rpc_task *
rpc_new_child(struct rpc_clnt *clnt, struct rpc_task *parent)
{
	struct rpc_task	*task;

	task = rpc_new_task(clnt, NULL, RPC_TASK_ASYNC | RPC_TASK_CHILD);
	if (!task)
		goto fail;
	task->tk_exit = rpc_child_exit;
	task->tk_calldata = parent;
	return task;

fail:
	parent->tk_status = -ENOMEM;
	return NULL;
}

void
rpc_run_child(struct rpc_task *task, struct rpc_task *child, rpc_action func)
{
	spin_lock_bh(&rpc_queue_lock);
	/* N.B. Is it possible for the child to have already finished? */
	__rpc_sleep_on(&childq, task, func, NULL);
	rpc_schedule_run(child);
	spin_unlock_bh(&rpc_queue_lock);
}

/*
 * Kill all tasks for the given client.
 * XXX: kill their descendants as well?
 */
void
rpc_killall_tasks(struct rpc_clnt *clnt)
{
	struct rpc_task	**q, *rovr;

	dprintk("RPC:      killing all tasks for client %p\n", clnt);

	/*
	 * Spin lock all_tasks to prevent changes...
	 */
	spin_lock(&rpc_sched_lock);
	for (q = &all_tasks; (rovr = *q); q = &rovr->tk_next_task) {
		if (!clnt || rovr->tk_client == clnt) {
			rovr->tk_flags |= RPC_TASK_KILLED;
			rpc_exit(rovr, -EIO);
			rpc_wake_up_task(rovr);
		}
	}
	spin_unlock(&rpc_sched_lock);
}

static DECLARE_MUTEX_LOCKED(rpciod_running);

static inline int
rpciod_task_pending(void)
{
	return schedq.task != NULL || xprt_tcp_pending();
}


/*
 * This is the rpciod kernel thread
 */
static int
rpciod(void *ptr)
{
	wait_queue_head_t *assassin = (wait_queue_head_t*) ptr;
	int		rounds = 0;

	MOD_INC_USE_COUNT;
	lock_kernel();
	/*
	 * Let our maker know we're running ...
	 */
	rpciod_pid = current->pid;
	up(&rpciod_running);

	exit_fs(current);
	exit_files(current);
	exit_mm(current);

	spin_lock_irq(&current->sigmask_lock);
	siginitsetinv(&current->blocked, sigmask(SIGKILL));
	recalc_sigpending(current);
	spin_unlock_irq(&current->sigmask_lock);

	current->session = 1;
	current->pgrp = 1;
	strcpy(current->comm, "rpciod");

	dprintk("RPC: rpciod starting (pid %d)\n", rpciod_pid);
	while (rpciod_users) {
		if (signalled()) {
			rpciod_killall();
			flush_signals(current);
		}
		__rpc_schedule();

		if (++rounds >= 64) {	/* safeguard */
			schedule();
			rounds = 0;
		}

		if (!rpciod_task_pending()) {
			dprintk("RPC: rpciod back to sleep\n");
			wait_event_interruptible(rpciod_idle, rpciod_task_pending());
			dprintk("RPC: switch to rpciod\n");
			rounds = 0;
		}
	}

	dprintk("RPC: rpciod shutdown commences\n");
	if (all_tasks) {
		printk(KERN_ERR "rpciod: active tasks at shutdown?!\n");
		rpciod_killall();
	}

	rpciod_pid = 0;
	wake_up(assassin);

	dprintk("RPC: rpciod exiting\n");
	MOD_DEC_USE_COUNT;
	return 0;
}

static void
rpciod_killall(void)
{
	unsigned long flags;

	while (all_tasks) {
		current->sigpending = 0;
		rpc_killall_tasks(NULL);
		__rpc_schedule();
		if (all_tasks) {
			dprintk("rpciod_killall: waiting for tasks to exit\n");
			set_current_state(TASK_INTERRUPTIBLE);
			schedule_timeout(1);
		}
	}

	spin_lock_irqsave(&current->sigmask_lock, flags);
	recalc_sigpending(current);
	spin_unlock_irqrestore(&current->sigmask_lock, flags);
}

/*
 * Start up the rpciod process if it's not already running.
 */
int
rpciod_up(void)
{
	int error = 0;

	MOD_INC_USE_COUNT;
	down(&rpciod_sema);
	dprintk("rpciod_up: pid %d, users %d\n", rpciod_pid, rpciod_users);
	rpciod_users++;
	if (rpciod_pid)
		goto out;
	/*
	 * If there's no pid, we should be the first user.
	 */
	if (rpciod_users > 1)
		printk(KERN_WARNING "rpciod_up: no pid, %d users??\n", rpciod_users);
	/*
	 * Create the rpciod thread and wait for it to start.
	 */
	error = kernel_thread(rpciod, &rpciod_killer, 0);
	if (error < 0) {
		printk(KERN_WARNING "rpciod_up: create thread failed, error=%d\n", error);
		rpciod_users--;
		goto out;
	}
	down(&rpciod_running);
	error = 0;
out:
	up(&rpciod_sema);
	MOD_DEC_USE_COUNT;
	return error;
}

void
rpciod_down(void)
{
	unsigned long flags;

	MOD_INC_USE_COUNT;
	down(&rpciod_sema);
	dprintk("rpciod_down pid %d sema %d\n", rpciod_pid, rpciod_users);
	if (rpciod_users) {
		if (--rpciod_users)
			goto out;
	} else
		printk(KERN_WARNING "rpciod_down: pid=%d, no users??\n", rpciod_pid);

	if (!rpciod_pid) {
		dprintk("rpciod_down: Nothing to do!\n");
		goto out;
	}

	kill_proc(rpciod_pid, SIGKILL, 1);
	/*
	 * Usually rpciod will exit very quickly, so we
	 * wait briefly before checking the process id.
	 */
	current->sigpending = 0;
	set_current_state(TASK_INTERRUPTIBLE);
	schedule_timeout(1);
	/*
	 * Display a message if we're going to wait longer.
	 */
	while (rpciod_pid) {
		dprintk("rpciod_down: waiting for pid %d to exit\n", rpciod_pid);
		if (signalled()) {
			dprintk("rpciod_down: caught signal\n");
			break;
		}
		interruptible_sleep_on(&rpciod_killer);
	}
	spin_lock_irqsave(&current->sigmask_lock, flags);
	recalc_sigpending(current);
	spin_unlock_irqrestore(&current->sigmask_lock, flags);
out:
	up(&rpciod_sema);
	MOD_DEC_USE_COUNT;
}

#ifdef RPC_DEBUG
void rpc_show_tasks(void)
{
	struct rpc_task *t = all_tasks, *next;

	spin_lock(&rpc_sched_lock);
	t = all_tasks;
	if (!t) {
		spin_unlock(&rpc_sched_lock);
		return;
	}
	printk("-pid- proc flgs status -client- -prog- --rqstp- -timeout "
		"-rpcwait -action- --exit--\n");
	for (; t; t = next) {
		next = t->tk_next_task;
		printk("%05d %04d %04x %06d %8p %6d %8p %08ld %8s %8p %8p\n",
			t->tk_pid, t->tk_msg.rpc_proc, t->tk_flags, t->tk_status,
			t->tk_client, t->tk_client->cl_prog,
			t->tk_rqstp, t->tk_timeout,
			t->tk_rpcwait ? rpc_qname(t->tk_rpcwait) : " <NULL> ",
			t->tk_action, t->tk_exit);
	}
	spin_unlock(&rpc_sched_lock);
}
#endif