X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=net%2Fsunrpc%2Fsched.c;h=d1c8b47f0d0946bc9a3c7e1bd3e38cb065e7a13f;hb=64ba3f394c830ec48a1c31b53dcae312c56f1604;hp=be26e4c5b1a720b607b496b671c307d98a850f97;hpb=6a77f38946aaee1cd85eeec6cf4229b204c15071;p=linux-2.6.git

diff --git a/net/sunrpc/sched.c b/net/sunrpc/sched.c
index be26e4c5b..d1c8b47f0 100644
--- a/net/sunrpc/sched.c
+++ b/net/sunrpc/sched.c
@@ -34,15 +34,13 @@ static int			rpc_task_id;
 #define RPC_BUFFER_MAXSIZE	(2048)
 #define RPC_BUFFER_POOLSIZE	(8)
 #define RPC_TASK_POOLSIZE	(8)
-static kmem_cache_t	*rpc_task_slabp;
-static kmem_cache_t	*rpc_buffer_slabp;
-static mempool_t	*rpc_task_mempool;
-static mempool_t	*rpc_buffer_mempool;
+static kmem_cache_t	*rpc_task_slabp __read_mostly;
+static kmem_cache_t	*rpc_buffer_slabp __read_mostly;
+static mempool_t	*rpc_task_mempool __read_mostly;
+static mempool_t	*rpc_buffer_mempool __read_mostly;
 
 static void			__rpc_default_timer(struct rpc_task *task);
 static void			rpciod_killall(void);
-static void			rpc_free(struct rpc_task *task);
-
 static void			rpc_async_schedule(void *);
 
 /*
@@ -132,9 +130,11 @@ __rpc_add_timer(struct rpc_task *task, rpc_action timer)
  * Delete any timer for the current task. Because we use del_timer_sync(),
  * this function should never be called while holding queue->lock.
  */
-static inline void
+static void
 rpc_delete_timer(struct rpc_task *task)
 {
+	if (RPC_IS_QUEUED(task))
+		return;
 	if (test_and_clear_bit(RPC_TASK_HAS_TIMER, &task->tk_runstate)) {
 		del_singleshot_timer_sync(&task->tk_timer);
 		dprintk("RPC: %4d deleting timer\n", task->tk_pid);
@@ -262,6 +262,35 @@ void rpc_init_wait_queue(struct rpc_wait_queue *queue, const char *qname)
 }
 EXPORT_SYMBOL(rpc_init_wait_queue);
 
+static int rpc_wait_bit_interruptible(void *word)
+{
+	if (signal_pending(current))
+		return -ERESTARTSYS;
+	schedule();
+	return 0;
+}
+
+/*
+ * Mark an RPC call as having completed by clearing the 'active' bit
+ */
+static inline void rpc_mark_complete_task(struct rpc_task *task)
+{
+	rpc_clear_active(task);
+	wake_up_bit(&task->tk_runstate, RPC_TASK_ACTIVE);
+}
+
+/*
+ * Allow callers to wait for completion of an RPC call
+ */
+int __rpc_wait_for_completion_task(struct rpc_task *task, int (*action)(void *))
+{
+	if (action == NULL)
+		action = rpc_wait_bit_interruptible;
+	return wait_on_bit(&task->tk_runstate, RPC_TASK_ACTIVE,
+			action, TASK_INTERRUPTIBLE);
+}
+EXPORT_SYMBOL(__rpc_wait_for_completion_task);
+
 /*
  * Make an RPC task runnable.
  *
@@ -270,13 +299,15 @@ EXPORT_SYMBOL(rpc_init_wait_queue);
  */
 static void rpc_make_runnable(struct rpc_task *task)
 {
-	int do_ret;
-
 	BUG_ON(task->tk_timeout_fn);
-	do_ret = rpc_test_and_set_running(task);
 	rpc_clear_queued(task);
-	if (do_ret)
+	if (rpc_test_and_set_running(task))
+		return;
+	/* We might have raced */
+	if (RPC_IS_QUEUED(task)) {
+		rpc_clear_running(task);
 		return;
+	}
 	if (RPC_IS_ASYNC(task)) {
 		int status;
 
@@ -288,7 +319,7 @@ static void rpc_make_runnable(struct rpc_task *task)
 			return;
 		}
 	} else
-		wake_up(&task->u.tk_wait.waitq);
+		wake_up_bit(&task->tk_runstate, RPC_TASK_QUEUED);
 }
 
 /*
@@ -297,10 +328,7 @@ static void rpc_make_runnable(struct rpc_task *task)
 static inline void
 rpc_schedule_run(struct rpc_task *task)
 {
-	/* Don't run a child twice! */
-	if (RPC_IS_ACTIVATED(task))
-		return;
-	task->tk_active = 1;
+	rpc_set_active(task);
 	rpc_make_runnable(task);
 }
 
@@ -322,8 +350,7 @@ static void __rpc_sleep_on(struct rpc_wait_queue *q, struct rpc_task *task,
 	}
 
 	/* Mark the task as being activated if so needed */
-	if (!RPC_IS_ACTIVATED(task))
-		task->tk_active = 1;
+	rpc_set_active(task);
 
 	__rpc_add_wait_queue(q, task);
 
@@ -490,16 +517,14 @@ struct rpc_task * rpc_wake_up_next(struct rpc_wait_queue *queue)
  */
 void rpc_wake_up(struct rpc_wait_queue *queue)
 {
-	struct rpc_task *task;
-
+	struct rpc_task *task, *next;
 	struct list_head *head;
+
 	spin_lock_bh(&queue->lock);
 	head = &queue->tasks[queue->maxpriority];
 	for (;;) {
-		while (!list_empty(head)) {
-			task = list_entry(head->next, struct rpc_task, u.tk_wait.list);
+		list_for_each_entry_safe(task, next, head, u.tk_wait.list)
 			__rpc_wake_up_task(task);
-		}
 		if (head == &queue->tasks[0])
 			break;
 		head--;
@@ -516,14 +541,13 @@ void rpc_wake_up(struct rpc_wait_queue *queue)
  */
 void rpc_wake_up_status(struct rpc_wait_queue *queue, int status)
 {
+	struct rpc_task *task, *next;
 	struct list_head *head;
-	struct rpc_task *task;
 
 	spin_lock_bh(&queue->lock);
 	head = &queue->tasks[queue->maxpriority];
 	for (;;) {
-		while (!list_empty(head)) {
-			task = list_entry(head->next, struct rpc_task, u.tk_wait.list);
+		list_for_each_entry_safe(task, next, head, u.tk_wait.list) {
 			task->tk_status = status;
 			__rpc_wake_up_task(task);
 		}
@@ -552,6 +576,31 @@ __rpc_atrun(struct rpc_task *task)
 	rpc_wake_up_task(task);
 }
 
+/*
+ * Helper to call task->tk_ops->rpc_call_prepare
+ */
+static void rpc_prepare_task(struct rpc_task *task)
+{
+	task->tk_ops->rpc_call_prepare(task, task->tk_calldata);
+}
+
+/*
+ * Helper that calls task->tk_ops->rpc_call_done if it exists
+ */
+void rpc_exit_task(struct rpc_task *task)
+{
+	task->tk_action = NULL;
+	if (task->tk_ops->rpc_call_done != NULL) {
+		task->tk_ops->rpc_call_done(task, task->tk_calldata);
+		if (task->tk_action != NULL) {
+			WARN_ON(RPC_ASSASSINATED(task));
+			/* Always release the RPC slot and buffer memory */
+			xprt_release(task);
+		}
+	}
+}
+EXPORT_SYMBOL(rpc_exit_task);
+
 /*
  * This is the RPC `scheduler' (or rather, the finite state machine).
  */
@@ -564,8 +613,7 @@ static int __rpc_execute(struct rpc_task *task)
 
 	BUG_ON(RPC_IS_QUEUED(task));
 
- restarted:
-	while (1) {
+	for (;;) {
 		/*
 		 * Garbage collection of pending timers...
 		 */
@@ -598,7 +646,7 @@ static int __rpc_execute(struct rpc_task *task)
 		 * by someone else.
 		 */
 		if (!RPC_IS_QUEUED(task)) {
-			if (!task->tk_action)
+			if (task->tk_action == NULL)
 				break;
 			lock_kernel();
 			task->tk_action(task);
@@ -622,47 +670,29 @@ static int __rpc_execute(struct rpc_task *task)
 
 		/* sync task: sleep here */
 		dprintk("RPC: %4d sync task going to sleep\n", task->tk_pid);
-		if (RPC_TASK_UNINTERRUPTIBLE(task)) {
-			__wait_event(task->u.tk_wait.waitq, !RPC_IS_QUEUED(task));
-		} else {
-			__wait_event_interruptible(task->u.tk_wait.waitq, !RPC_IS_QUEUED(task), status);
+		/* Note: Caller should be using rpc_clnt_sigmask() */
+		status = out_of_line_wait_on_bit(&task->tk_runstate,
+				RPC_TASK_QUEUED, rpc_wait_bit_interruptible,
+				TASK_INTERRUPTIBLE);
+		if (status == -ERESTARTSYS) {
 			/*
 			 * When a sync task receives a signal, it exits with
 			 * -ERESTARTSYS. In order to catch any callbacks that
 			 * clean up after sleeping on some queue, we don't
 			 * break the loop here, but go around once more.
 			 */
-			if (status == -ERESTARTSYS) {
-				dprintk("RPC: %4d got signal\n", task->tk_pid);
-				task->tk_flags |= RPC_TASK_KILLED;
-				rpc_exit(task, -ERESTARTSYS);
-				rpc_wake_up_task(task);
-			}
+			dprintk("RPC: %4d got signal\n", task->tk_pid);
+			task->tk_flags |= RPC_TASK_KILLED;
+			rpc_exit(task, -ERESTARTSYS);
+			rpc_wake_up_task(task);
 		}
 		rpc_set_running(task);
 		dprintk("RPC: %4d sync task resuming\n", task->tk_pid);
 	}
 
-	if (task->tk_exit) {
-		lock_kernel();
-		task->tk_exit(task);
-		unlock_kernel();
-		/* 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);
-				rpc_free(task);
-				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;
-
+	dprintk("RPC: %4d, return %d, status %d\n", task->tk_pid, status, task->tk_status);
+	/* Wake up anyone who is waiting for task completion */
+	rpc_mark_complete_task(task);
 	/* Release all resources associated with the task */
 	rpc_release_task(task);
 	return status;
@@ -680,9 +710,7 @@ static int __rpc_execute(struct rpc_task *task)
 int
 rpc_execute(struct rpc_task *task)
 {
-	BUG_ON(task->tk_active);
-
-	task->tk_active = 1;
+	rpc_set_active(task);
 	rpc_set_running(task);
 	return __rpc_execute(task);
 }
@@ -692,18 +720,20 @@ static void rpc_async_schedule(void *arg)
 	__rpc_execute((struct rpc_task *)arg);
 }
 
-/*
- * Allocate memory for RPC purposes.
+/**
+ * rpc_malloc - allocate an RPC buffer
+ * @task: RPC task that will use this buffer
+ * @size: requested byte size
  *
  * We try to ensure that some NFS reads and writes can always proceed
  * by using a mempool when allocating 'small' buffers.
  * In order to avoid memory starvation triggering more writebacks of
  * NFS requests, we use GFP_NOFS rather than GFP_KERNEL.
  */
-void *
-rpc_malloc(struct rpc_task *task, size_t size)
+void * rpc_malloc(struct rpc_task *task, size_t size)
 {
-	int	gfp;
+	struct rpc_rqst *req = task->tk_rqstp;
+	gfp_t	gfp;
 
 	if (task->tk_flags & RPC_TASK_SWAPPER)
 		gfp = GFP_ATOMIC;
@@ -711,57 +741,62 @@ rpc_malloc(struct rpc_task *task, size_t size)
 		gfp = GFP_NOFS;
 
 	if (size > RPC_BUFFER_MAXSIZE) {
-		task->tk_buffer =  kmalloc(size, gfp);
-		if (task->tk_buffer)
-			task->tk_bufsize = size;
+		req->rq_buffer = kmalloc(size, gfp);
+		if (req->rq_buffer)
+			req->rq_bufsize = size;
 	} else {
-		task->tk_buffer =  mempool_alloc(rpc_buffer_mempool, gfp);
-		if (task->tk_buffer)
-			task->tk_bufsize = RPC_BUFFER_MAXSIZE;
+		req->rq_buffer = mempool_alloc(rpc_buffer_mempool, gfp);
+		if (req->rq_buffer)
+			req->rq_bufsize = RPC_BUFFER_MAXSIZE;
 	}
-	return task->tk_buffer;
+	return req->rq_buffer;
 }
 
-static void
-rpc_free(struct rpc_task *task)
+/**
+ * rpc_free - free buffer allocated via rpc_malloc
+ * @task: RPC task with a buffer to be freed
+ *
+ */
+void rpc_free(struct rpc_task *task)
 {
-	if (task->tk_buffer) {
-		if (task->tk_bufsize == RPC_BUFFER_MAXSIZE)
-			mempool_free(task->tk_buffer, rpc_buffer_mempool);
+	struct rpc_rqst *req = task->tk_rqstp;
+
+	if (req->rq_buffer) {
+		if (req->rq_bufsize == RPC_BUFFER_MAXSIZE)
+			mempool_free(req->rq_buffer, rpc_buffer_mempool);
 		else
-			kfree(task->tk_buffer);
-		task->tk_buffer = NULL;
-		task->tk_bufsize = 0;
+			kfree(req->rq_buffer);
+		req->rq_buffer = NULL;
+		req->rq_bufsize = 0;
 	}
 }
 
 /*
  * Creation and deletion of RPC task structures
  */
-void rpc_init_task(struct rpc_task *task, struct rpc_clnt *clnt, rpc_action callback, int flags)
+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));
 	init_timer(&task->tk_timer);
 	task->tk_timer.data     = (unsigned long) task;
 	task->tk_timer.function = (void (*)(unsigned long)) rpc_run_timer;
+	atomic_set(&task->tk_count, 1);
 	task->tk_client = clnt;
 	task->tk_flags  = flags;
-	task->tk_exit   = callback;
-	if (current->uid != current->fsuid || current->gid != current->fsgid)
-		task->tk_flags |= RPC_TASK_SETUID;
+	task->tk_ops = tk_ops;
+	if (tk_ops->rpc_call_prepare != NULL)
+		task->tk_action = rpc_prepare_task;
+	task->tk_calldata = calldata;
 
 	/* Initialize retry counters */
 	task->tk_garb_retry = 2;
 	task->tk_cred_retry = 2;
-	task->tk_suid_retry = 1;
 
 	task->tk_priority = RPC_PRIORITY_NORMAL;
 	task->tk_cookie = (unsigned long)current;
 
 	/* Initialize workqueue for async tasks */
 	task->tk_workqueue = rpciod_workqueue;
-	if (!RPC_IS_ASYNC(task))
-		init_waitqueue_head(&task->u.tk_wait.waitq);
 
 	if (clnt) {
 		atomic_inc(&clnt->cl_users);
@@ -780,6 +815,8 @@ void rpc_init_task(struct rpc_task *task, struct rpc_clnt *clnt, rpc_action call
 	list_add_tail(&task->tk_task, &all_tasks);
 	spin_unlock(&rpc_sched_lock);
 
+	BUG_ON(task->tk_ops == NULL);
+
 	dprintk("RPC: %4d new task procpid %d\n", task->tk_pid,
 				current->pid);
 }
@@ -790,8 +827,7 @@ rpc_alloc_task(void)
 	return (struct rpc_task *)mempool_alloc(rpc_task_mempool, GFP_NOFS);
 }
 
-static void
-rpc_default_free_task(struct rpc_task *task)
+static void rpc_free_task(struct rpc_task *task)
 {
 	dprintk("RPC: %4d freeing task\n", task->tk_pid);
 	mempool_free(task, rpc_task_mempool);
@@ -802,8 +838,7 @@ rpc_default_free_task(struct rpc_task *task)
  * 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 *rpc_new_task(struct rpc_clnt *clnt, int flags, const struct rpc_call_ops *tk_ops, void *calldata)
 {
 	struct rpc_task	*task;
 
@@ -811,10 +846,7 @@ rpc_new_task(struct rpc_clnt *clnt, rpc_action callback, int flags)
 	if (!task)
 		goto cleanup;
 
-	rpc_init_task(task, clnt, callback, flags);
-
-	/* Replace tk_release */
-	task->tk_release = rpc_default_free_task;
+	rpc_init_task(task, clnt, flags, tk_ops, calldata);
 
 	dprintk("RPC: %4d allocated task\n", task->tk_pid);
 	task->tk_flags |= RPC_TASK_DYNAMIC;
@@ -834,11 +866,15 @@ cleanup:
 
 void rpc_release_task(struct rpc_task *task)
 {
-	dprintk("RPC: %4d release task\n", task->tk_pid);
+	const struct rpc_call_ops *tk_ops = task->tk_ops;
+	void *calldata = task->tk_calldata;
 
 #ifdef RPC_DEBUG
 	BUG_ON(task->tk_magic != RPC_TASK_MAGIC_ID);
 #endif
+	if (!atomic_dec_and_test(&task->tk_count))
+		return;
+	dprintk("RPC: %4d release task\n", task->tk_pid);
 
 	/* Remove from global task list */
 	spin_lock(&rpc_sched_lock);
@@ -846,7 +882,6 @@ void rpc_release_task(struct rpc_task *task)
 	spin_unlock(&rpc_sched_lock);
 
 	BUG_ON (RPC_IS_QUEUED(task));
-	task->tk_active = 0;
 
 	/* Synchronously delete any running timer */
 	rpc_delete_timer(task);
@@ -856,7 +891,6 @@ void rpc_release_task(struct rpc_task *task)
 		xprt_release(task);
 	if (task->tk_msg.rpc_cred)
 		rpcauth_unbindcred(task);
-	rpc_free(task);
 	if (task->tk_client) {
 		rpc_release_client(task->tk_client);
 		task->tk_client = NULL;
@@ -865,13 +899,37 @@ void rpc_release_task(struct rpc_task *task)
 #ifdef RPC_DEBUG
 	task->tk_magic = 0;
 #endif
-	if (task->tk_release)
-		task->tk_release(task);
+	if (task->tk_flags & RPC_TASK_DYNAMIC)
+		rpc_free_task(task);
+	if (tk_ops->rpc_release)
+		tk_ops->rpc_release(calldata);
 }
 
+/**
+ * rpc_run_task - Allocate a new RPC task, then run rpc_execute against it
+ * @clnt: pointer to RPC client
+ * @flags: RPC flags
+ * @ops: RPC call ops
+ * @data: user call data
+ */
+struct rpc_task *rpc_run_task(struct rpc_clnt *clnt, int flags,
+					const struct rpc_call_ops *ops,
+					void *data)
+{
+	struct rpc_task *task;
+	task = rpc_new_task(clnt, flags, ops, data);
+	if (task == NULL)
+		return ERR_PTR(-ENOMEM);
+	atomic_inc(&task->tk_count);
+	rpc_execute(task);
+	return task;
+}
+EXPORT_SYMBOL(rpc_run_task);
+
 /**
  * rpc_find_parent - find the parent of a child task.
  * @child: child task
+ * @parent: parent task
  *
  * Checks that the parent task is still sleeping on the
  * queue 'childq'. If so returns a pointer to the parent.
@@ -879,12 +937,11 @@ void rpc_release_task(struct rpc_task *task)
  *
  * Caller must hold childq.lock
  */
-static inline struct rpc_task *rpc_find_parent(struct rpc_task *child)
+static inline struct rpc_task *rpc_find_parent(struct rpc_task *child, struct rpc_task *parent)
 {
-	struct rpc_task	*task, *parent;
+	struct rpc_task	*task;
 	struct list_head *le;
 
-	parent = (struct rpc_task *) child->tk_calldata;
 	task_for_each(task, le, &childq.tasks[0])
 		if (task == parent)
 			return parent;
@@ -892,18 +949,22 @@ static inline struct rpc_task *rpc_find_parent(struct rpc_task *child)
 	return NULL;
 }
 
-static void rpc_child_exit(struct rpc_task *child)
+static void rpc_child_exit(struct rpc_task *child, void *calldata)
 {
 	struct rpc_task	*parent;
 
 	spin_lock_bh(&childq.lock);
-	if ((parent = rpc_find_parent(child)) != NULL) {
+	if ((parent = rpc_find_parent(child, calldata)) != NULL) {
 		parent->tk_status = child->tk_status;
 		__rpc_wake_up_task(parent);
 	}
 	spin_unlock_bh(&childq.lock);
 }
 
+static const struct rpc_call_ops rpc_child_ops = {
+	.rpc_call_done = rpc_child_exit,
+};
+
 /*
  * Note: rpc_new_task releases the client after a failure.
  */
@@ -912,11 +973,9 @@ 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);
+	task = rpc_new_task(clnt, RPC_TASK_ASYNC | RPC_TASK_CHILD, &rpc_child_ops, parent);
 	if (!task)
 		goto fail;
-	task->tk_exit = rpc_child_exit;
-	task->tk_calldata = parent;
 	return task;
 
 fail:
@@ -1052,7 +1111,7 @@ void rpc_show_tasks(void)
 		return;
 	}
 	printk("-pid- proc flgs status -client- -prog- --rqstp- -timeout "
-		"-rpcwait -action- --exit--\n");
+		"-rpcwait -action- ---ops--\n");
 	alltask_for_each(t, le, &all_tasks) {
 		const char *rpc_waitq = "none";
 
@@ -1067,7 +1126,7 @@ void rpc_show_tasks(void)
 			(t->tk_client ? t->tk_client->cl_prog : 0),
 			t->tk_rqstp, t->tk_timeout,
 			rpc_waitq,
-			t->tk_action, t->tk_exit);
+			t->tk_action, t->tk_ops);
 	}
 	spin_unlock(&rpc_sched_lock);
 }