Stop dhcpcd before starting if it was running

[people/pmueller/ipfire-2.x.git] / src / patches / xen_aio.diff

--- linux-2.6.27.21/fs/aio.c	2009-03-23 22:04:09.000000000 +0000
+++ linux-2.6.27.8/fs/aio.c	2009-03-29 15:53:57.000000000 +0000
@@ -36,6 +36,11 @@
 #include <asm/uaccess.h>
 #include <asm/mmu_context.h>
 
+#ifdef CONFIG_EPOLL
+#include <linux/poll.h>
+#include <linux/eventpoll.h>
+#endif
+
 #if DEBUG > 1
 #define dprintk		printk
 #else
@@ -428,7 +433,7 @@
 	req->private = NULL;
 	req->ki_iovec = NULL;
 	INIT_LIST_HEAD(&req->ki_run_list);
-	req->ki_eventfd = NULL;
+	req->ki_eventfd = ERR_PTR(-EINVAL);
 
 	/* Check if the completion queue has enough free space to
 	 * accept an event from this io.
@@ -470,6 +475,8 @@
 {
 	assert_spin_locked(&ctx->ctx_lock);
 
+	if (!IS_ERR(req->ki_eventfd))
+		fput(req->ki_eventfd);
 	if (req->ki_dtor)
 		req->ki_dtor(req);
 	if (req->ki_iovec != &req->ki_inline_vec)
@@ -491,11 +498,8 @@
 		list_del(&req->ki_list);
 		spin_unlock_irq(&fput_lock);
 
-		/* Complete the fput(s) */
-		if (req->ki_filp != NULL)
-			__fput(req->ki_filp);
-		if (req->ki_eventfd != NULL)
-			__fput(req->ki_eventfd);
+		/* Complete the fput */
+		__fput(req->ki_filp);
 
 		/* Link the iocb into the context's free list */
 		spin_lock_irq(&ctx->ctx_lock);
@@ -513,14 +517,12 @@
  */
 static int __aio_put_req(struct kioctx *ctx, struct kiocb *req)
 {
-	int schedule_putreq = 0;
-
 	dprintk(KERN_DEBUG "aio_put(%p): f_count=%ld\n",
 		req, atomic_long_read(&req->ki_filp->f_count));
 
 	assert_spin_locked(&ctx->ctx_lock);
 
-	req->ki_users--;
+	req->ki_users --;
 	BUG_ON(req->ki_users < 0);
 	if (likely(req->ki_users))
 		return 0;
@@ -528,23 +530,10 @@
 	req->ki_cancel = NULL;
 	req->ki_retry = NULL;
 
-	/*
-	 * Try to optimize the aio and eventfd file* puts, by avoiding to
-	 * schedule work in case it is not __fput() time. In normal cases,
-	 * we would not be holding the last reference to the file*, so
-	 * this function will be executed w/out any aio kthread wakeup.
+	/* Must be done under the lock to serialise against cancellation.
+	 * Call this aio_fput as it duplicates fput via the fput_work.
 	 */
-	if (unlikely(atomic_long_dec_and_test(&req->ki_filp->f_count)))
-		schedule_putreq++;
-	else
-		req->ki_filp = NULL;
-	if (req->ki_eventfd != NULL) {
-		if (unlikely(atomic_long_dec_and_test(&req->ki_eventfd->f_count)))
-			schedule_putreq++;
-		else
-			req->ki_eventfd = NULL;
-	}
-	if (unlikely(schedule_putreq)) {
+	if (unlikely(atomic_long_dec_and_test(&req->ki_filp->f_count))) {
 		get_ioctx(ctx);
 		spin_lock(&fput_lock);
 		list_add(&req->ki_list, &fput_head);
@@ -1008,7 +997,7 @@
 	 * eventfd. The eventfd_signal() function is safe to be called
 	 * from IRQ context.
 	 */
-	if (iocb->ki_eventfd != NULL)
+	if (!IS_ERR(iocb->ki_eventfd))
 		eventfd_signal(iocb->ki_eventfd, 1);
 
 put_rq:
@@ -1026,6 +1015,11 @@
 	if (waitqueue_active(&ctx->wait))
 		wake_up(&ctx->wait);
 
+#ifdef CONFIG_EPOLL
+	if (ctx->file && waitqueue_active(&ctx->poll_wait))
+		wake_up(&ctx->poll_wait);
+#endif
+
 	spin_unlock_irqrestore(&ctx->ctx_lock, flags);
 	return ret;
 }
@@ -1033,6 +1027,8 @@
 /* aio_read_evt
  *	Pull an event off of the ioctx's event ring.  Returns the number of 
  *	events fetched (0 or 1 ;-)
+ *	If ent parameter is 0, just returns the number of events that would
+ *	be fetched.
  *	FIXME: make this use cmpxchg.
  *	TODO: make the ringbuffer user mmap()able (requires FIXME).
  */
@@ -1055,13 +1051,18 @@
 
 	head = ring->head % info->nr;
 	if (head != ring->tail) {
-		struct io_event *evp = aio_ring_event(info, head, KM_USER1);
-		*ent = *evp;
-		head = (head + 1) % info->nr;
-		smp_mb(); /* finish reading the event before updatng the head */
-		ring->head = head;
-		ret = 1;
-		put_aio_ring_event(evp, KM_USER1);
+		if (ent) { /* event requested */
+			struct io_event *evp =
+				aio_ring_event(info, head, KM_USER1);
+			*ent = *evp;
+			head = (head + 1) % info->nr;
+			/* finish reading the event before updatng the head */
+			smp_mb();
+			ring->head = head;
+			ret = 1;
+			put_aio_ring_event(evp, KM_USER1);
+		} else /* only need to know availability */
+			ret = 1;
 	}
 	spin_unlock(&info->ring_lock);
 
@@ -1251,6 +1252,13 @@
 
 	aio_cancel_all(ioctx);
 	wait_for_all_aios(ioctx);
+#ifdef CONFIG_EPOLL
+	/* forget the poll file, but it's up to the user to close it */
+	if (ioctx->file) {
+		ioctx->file->private_data = 0;
+		ioctx->file = 0;
+	}
+#endif
 
 	/*
 	 * Wake up any waiters.  The setting of ctx->dead must be seen
@@ -1261,6 +1269,67 @@
 	put_ioctx(ioctx);	/* once for the lookup */
 }
 
+#ifdef CONFIG_EPOLL
+
+static int aio_queue_fd_close(struct inode *inode, struct file *file)
+{
+	struct kioctx *ioctx = file->private_data;
+	if (ioctx) {
+		file->private_data = 0;
+		spin_lock_irq(&ioctx->ctx_lock);
+		ioctx->file = 0;
+		spin_unlock_irq(&ioctx->ctx_lock);
+	}
+	return 0;
+}
+
+static unsigned int aio_queue_fd_poll(struct file *file, poll_table *wait)
+{	unsigned int pollflags = 0;
+	struct kioctx *ioctx = file->private_data;
+
+	if (ioctx) {
+
+		spin_lock_irq(&ioctx->ctx_lock);
+		/* Insert inside our poll wait queue */
+		poll_wait(file, &ioctx->poll_wait, wait);
+
+		/* Check our condition */
+		if (aio_read_evt(ioctx, 0))
+			pollflags = POLLIN | POLLRDNORM;
+		spin_unlock_irq(&ioctx->ctx_lock);
+	}
+
+	return pollflags;
+}
+
+static const struct file_operations aioq_fops = {
+	.release	= aio_queue_fd_close,
+	.poll		= aio_queue_fd_poll
+};
+
+/* make_aio_fd:
+ *  Create a file descriptor that can be used to poll the event queue.
+ *  Based and piggybacked on the excellent epoll code.
+ */
+
+static int make_aio_fd(struct kioctx *ioctx)
+{
+	int error, fd;
+	struct inode *inode;
+	struct file *file;
+
+	error = ep_getfd(&fd, &inode, &file, NULL, &aioq_fops);
+	if (error)
+		return error;
+
+	/* associate the file with the IO context */
+	file->private_data = ioctx;
+	ioctx->file = file;
+	init_waitqueue_head(&ioctx->poll_wait);
+	return fd;
+}
+#endif
+
 /* sys_io_setup:
  *	Create an aio_context capable of receiving at least nr_events.
  *	ctxp must not point to an aio_context that already exists, and
@@ -1273,18 +1342,30 @@
  *	resources are available.  May fail with -EFAULT if an invalid
  *	pointer is passed for ctxp.  Will fail with -ENOSYS if not
  *	implemented.
+ *
+ *	To request a selectable fd, the user context has to be initialized
+ *	to 1, instead of 0, and the return value is the fd.
+ *	This keeps the system call compatible, since a non-zero value
+ *	was not allowed so far.
  */
-SYSCALL_DEFINE2(io_setup, unsigned, nr_events, aio_context_t __user *, ctxp)
+asmlinkage long sys_io_setup(unsigned nr_events, aio_context_t __user *ctxp)
 {
 	struct kioctx *ioctx = NULL;
 	unsigned long ctx;
 	long ret;
+	int make_fd = 0;
 
 	ret = get_user(ctx, ctxp);
 	if (unlikely(ret))
 		goto out;
 
 	ret = -EINVAL;
+#ifdef CONFIG_EPOLL
+	if (ctx == 1) {
+		make_fd = 1;
+		ctx = 0;
+	}
+#endif
 	if (unlikely(ctx || nr_events == 0)) {
 		pr_debug("EINVAL: io_setup: ctx %lu nr_events %u\n",
 		         ctx, nr_events);
@@ -1295,8 +1376,12 @@
 	ret = PTR_ERR(ioctx);
 	if (!IS_ERR(ioctx)) {
 		ret = put_user(ioctx->user_id, ctxp);
-		if (!ret)
-			return 0;
+#ifdef CONFIG_EPOLL
+		if (make_fd && ret >= 0)
+			ret = make_aio_fd(ioctx);
+#endif
+		if (ret >= 0)
+			return ret;
 
 		get_ioctx(ioctx); /* io_destroy() expects us to hold a ref */
 		io_destroy(ioctx);
@@ -1312,7 +1397,7 @@
  *	implemented.  May fail with -EFAULT if the context pointed to
  *	is invalid.
  */
-SYSCALL_DEFINE1(io_destroy, aio_context_t, ctx)
+asmlinkage long sys_io_destroy(aio_context_t ctx)
 {
 	struct kioctx *ioctx = lookup_ioctx(ctx);
 	if (likely(NULL != ioctx)) {
@@ -1612,7 +1697,6 @@
 		req->ki_eventfd = eventfd_fget((int) iocb->aio_resfd);
 		if (IS_ERR(req->ki_eventfd)) {
 			ret = PTR_ERR(req->ki_eventfd);
-			req->ki_eventfd = NULL;
 			goto out_put_req;
 		}
 	}
@@ -1667,8 +1751,8 @@
  *	are available to queue any iocbs.  Will return 0 if nr is 0.  Will
  *	fail with -ENOSYS if not implemented.
  */
-SYSCALL_DEFINE3(io_submit, aio_context_t, ctx_id, long, nr,
-		struct iocb __user * __user *, iocbpp)
+asmlinkage long sys_io_submit(aio_context_t ctx_id, long nr,
+			      struct iocb __user * __user *iocbpp)
 {
 	struct kioctx *ctx;
 	long ret = 0;
@@ -1742,8 +1826,8 @@
  *	invalid.  May fail with -EAGAIN if the iocb specified was not
  *	cancelled.  Will fail with -ENOSYS if not implemented.
  */
-SYSCALL_DEFINE3(io_cancel, aio_context_t, ctx_id, struct iocb __user *, iocb,
-		struct io_event __user *, result)
+asmlinkage long sys_io_cancel(aio_context_t ctx_id, struct iocb __user *iocb,
+			      struct io_event __user *result)
 {
 	int (*cancel)(struct kiocb *iocb, struct io_event *res);
 	struct kioctx *ctx;
@@ -1804,11 +1888,11 @@
  *	will be updated if not NULL and the operation blocks.  Will fail
  *	with -ENOSYS if not implemented.
  */
-SYSCALL_DEFINE5(io_getevents, aio_context_t, ctx_id,
-		long, min_nr,
-		long, nr,
-		struct io_event __user *, events,
-		struct timespec __user *, timeout)
+asmlinkage long sys_io_getevents(aio_context_t ctx_id,
+				 long min_nr,
+				 long nr,
+				 struct io_event __user *events,
+				 struct timespec __user *timeout)
 {
 	struct kioctx *ioctx = lookup_ioctx(ctx_id);
 	long ret = -EINVAL;