Merge remote-tracking branch 'bonzini/threadpool' into staging

* bonzini/threadpool: (39 commits) raw-win32: implement native asynchronous I/O raw-posix: move linux-aio.c to block/ raw-win32: add emulated AIO support raw-posix: rename raw-posix-aio.h, hide unavailable prototypes raw: merge posix-aio-compat.c into block/raw-posix.c block: switch posix-aio-compat to threadpool threadpool: do not take lock in event_notifier_ready aio: add generic thread-pool facility qemu-thread: add QemuSemaphore linux-aio: use event notifiers aio: clean up now-unused functions main-loop: use aio_notify for qemu_notify_event main-loop: use GSource to poll AIO file descriptors aio: call aio_notify after setting I/O handlers aio: add aio_notify aio: make AioContexts GSources aio: add Win32 implementation aio: prepare for introducing GSource-based dispatch aio: add non-blocking variant of aio_wait aio: test node->deleted before calling io_flush ... Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2026-07-06 00:20:22 -07:00 · 2012-11-01 11:13:39 -05:00
parent 43552994c4 f563a5d7a8
commit 5a34dbb559
49 changed files with 2439 additions and 1499 deletions
@@ -171,8 +171,7 @@ endif
 qemu-img.o: qemu-img-cmds.h

 tools-obj-y = $(oslib-obj-y) $(trace-obj-y) qemu-tool.o qemu-timer.o \
-	qemu-timer-common.o main-loop.o notify.o \
-	iohandler.o cutils.o iov.o async.o error.o
+	main-loop.o iohandler.o error.o
 tools-obj-$(CONFIG_POSIX) += compatfd.o

 qemu-img$(EXESUF): qemu-img.o $(tools-obj-y) $(block-obj-y)
@@ -181,7 +180,7 @@ qemu-io$(EXESUF): qemu-io.o cmd.o $(tools-obj-y) $(block-obj-y)

 qemu-bridge-helper$(EXESUF): qemu-bridge-helper.o

-vscclient$(EXESUF): $(libcacard-y) $(oslib-obj-y) $(trace-obj-y) $(tools-obj-y) qemu-timer-common.o libcacard/vscclient.o
+vscclient$(EXESUF): $(libcacard-y) $(oslib-obj-y) $(trace-obj-y) libcacard/vscclient.o
 	$(call quiet-command,$(CC) $(LDFLAGS) -o $@ $^ $(libcacard_libs) $(LIBS),"  LINK  $@")

 fsdev/virtfs-proxy-helper$(EXESUF): fsdev/virtfs-proxy-helper.o fsdev/virtio-9p-marshal.o oslib-posix.o $(trace-obj-y)
@@ -224,7 +223,7 @@ $(SRC_PATH)/qapi-schema.json $(SRC_PATH)/scripts/qapi-commands.py $(qapi-py)
 QGALIB_GEN=$(addprefix qga/qapi-generated/, qga-qapi-types.h qga-qapi-visit.h qga-qmp-commands.h)
 $(qga-obj-y) qemu-ga.o: $(QGALIB_GEN)

-qemu-ga$(EXESUF): qemu-ga.o $(qga-obj-y) $(tools-obj-y) $(qapi-obj-y) $(qobject-obj-y) $(version-obj-y)
+qemu-ga$(EXESUF): qemu-ga.o $(qga-obj-y) $(oslib-obj-y) $(trace-obj-y) $(qapi-obj-y) $(qobject-obj-y) $(version-obj-y)

 QEMULIBS=libuser libdis libdis-user

@@ -19,7 +19,7 @@ universal-obj-y += $(qom-obj-y)

 #######################################################################
 # oslib-obj-y is code depending on the OS (win32 vs posix)
-oslib-obj-y = osdep.o
+oslib-obj-y = osdep.o cutils.o qemu-timer-common.o
 oslib-obj-$(CONFIG_WIN32) += oslib-win32.o qemu-thread-win32.o
 oslib-obj-$(CONFIG_POSIX) += oslib-posix.o qemu-thread-posix.o

@@ -41,12 +41,12 @@ coroutine-obj-$(CONFIG_WIN32) += coroutine-win32.o
 #######################################################################
 # block-obj-y is code used by both qemu system emulation and qemu-img

-block-obj-y = cutils.o iov.o cache-utils.o qemu-option.o module.o async.o
-block-obj-y += nbd.o block.o blockjob.o aio.o aes.o qemu-config.o
-block-obj-y += qemu-progress.o qemu-sockets.o uri.o notify.o
+block-obj-y = iov.o cache-utils.o qemu-option.o module.o async.o
+block-obj-y += nbd.o block.o blockjob.o aes.o qemu-config.o
+block-obj-y += thread-pool.o qemu-progress.o qemu-sockets.o uri.o notify.o
 block-obj-y += $(coroutine-obj-y) $(qobject-obj-y) $(version-obj-y)
-block-obj-$(CONFIG_POSIX) += posix-aio-compat.o
-block-obj-$(CONFIG_LINUX_AIO) += linux-aio.o
+block-obj-$(CONFIG_POSIX) += event_notifier-posix.o aio-posix.o
+block-obj-$(CONFIG_WIN32) += event_notifier-win32.o aio-win32.o
 block-obj-y += block/
 block-obj-y += $(qapi-obj-y) qapi-types.o qapi-visit.o

@@ -92,9 +92,8 @@ common-obj-y += ui/
 common-obj-y += bt-host.o bt-vhci.o

 common-obj-y += dma-helpers.o
-common-obj-y += iov.o acl.o
+common-obj-y += acl.o
 common-obj-$(CONFIG_POSIX) += compatfd.o
-common-obj-y += event_notifier.o
 common-obj-y += qemu-timer.o qemu-timer-common.o
 common-obj-y += qtest.o
 common-obj-y += vl.o
@@ -113,7 +112,7 @@ endif
 user-obj-y =
 user-obj-y += envlist.o path.o
 user-obj-y += tcg-runtime.o host-utils.o
-user-obj-y += cutils.o iov.o cache-utils.o
+user-obj-y += cache-utils.o
 user-obj-y += module.o
 user-obj-y += qemu-user.o
 user-obj-y += $(trace-obj-y)
@@ -228,9 +227,8 @@ universal-obj-y += $(qapi-obj-y)
 ######################################################################
 # guest agent

-qga-obj-y = qga/ qemu-ga.o module.o
-qga-obj-$(CONFIG_WIN32) += oslib-win32.o
-qga-obj-$(CONFIG_POSIX) += oslib-posix.o qemu-sockets.o qemu-option.o
+qga-obj-y = qga/ qemu-ga.o module.o qemu-tool.o
+qga-obj-$(CONFIG_POSIX) += qemu-sockets.o qemu-option.o

 vl.o: QEMU_CFLAGS+=$(GPROF_CFLAGS)

@@ -0,0 +1,268 @@
+/*
+ * QEMU aio implementation
+ *
+ * Copyright IBM, Corp. 2008
+ *
+ * Authors:
+ *  Anthony Liguori   <aliguori@us.ibm.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2.  See
+ * the COPYING file in the top-level directory.
+ *
+ * Contributions after 2012-01-13 are licensed under the terms of the
+ * GNU GPL, version 2 or (at your option) any later version.
+ */
+
+#include "qemu-common.h"
+#include "block.h"
+#include "qemu-queue.h"
+#include "qemu_socket.h"
+
+struct AioHandler
+{
+    GPollFD pfd;
+    IOHandler *io_read;
+    IOHandler *io_write;
+    AioFlushHandler *io_flush;
+    int deleted;
+    void *opaque;
+    QLIST_ENTRY(AioHandler) node;
+};
+
+static AioHandler *find_aio_handler(AioContext *ctx, int fd)
+{
+    AioHandler *node;
+
+    QLIST_FOREACH(node, &ctx->aio_handlers, node) {
+        if (node->pfd.fd == fd)
+            if (!node->deleted)
+                return node;
+    }
+
+    return NULL;
+}
+
+void aio_set_fd_handler(AioContext *ctx,
+                        int fd,
+                        IOHandler *io_read,
+                        IOHandler *io_write,
+                        AioFlushHandler *io_flush,
+                        void *opaque)
+{
+    AioHandler *node;
+
+    node = find_aio_handler(ctx, fd);
+
+    /* Are we deleting the fd handler? */
+    if (!io_read && !io_write) {
+        if (node) {
+            g_source_remove_poll(&ctx->source, &node->pfd);
+
+            /* If the lock is held, just mark the node as deleted */
+            if (ctx->walking_handlers) {
+                node->deleted = 1;
+                node->pfd.revents = 0;
+            } else {
+                /* Otherwise, delete it for real.  We can't just mark it as
+                 * deleted because deleted nodes are only cleaned up after
+                 * releasing the walking_handlers lock.
+                 */
+                QLIST_REMOVE(node, node);
+                g_free(node);
+            }
+        }
+    } else {
+        if (node == NULL) {
+            /* Alloc and insert if it's not already there */
+            node = g_malloc0(sizeof(AioHandler));
+            node->pfd.fd = fd;
+            QLIST_INSERT_HEAD(&ctx->aio_handlers, node, node);
+
+            g_source_add_poll(&ctx->source, &node->pfd);
+        }
+        /* Update handler with latest information */
+        node->io_read = io_read;
+        node->io_write = io_write;
+        node->io_flush = io_flush;
+        node->opaque = opaque;
+
+        node->pfd.events = (io_read ? G_IO_IN | G_IO_HUP : 0);
+        node->pfd.events |= (io_write ? G_IO_OUT : 0);
+    }
+
+    aio_notify(ctx);
+}
+
+void aio_set_event_notifier(AioContext *ctx,
+                            EventNotifier *notifier,
+                            EventNotifierHandler *io_read,
+                            AioFlushEventNotifierHandler *io_flush)
+{
+    aio_set_fd_handler(ctx, event_notifier_get_fd(notifier),
+                       (IOHandler *)io_read, NULL,
+                       (AioFlushHandler *)io_flush, notifier);
+}
+
+bool aio_pending(AioContext *ctx)
+{
+    AioHandler *node;
+
+    QLIST_FOREACH(node, &ctx->aio_handlers, node) {
+        int revents;
+
+        /*
+         * FIXME: right now we cannot get G_IO_HUP and G_IO_ERR because
+         * main-loop.c is still select based (due to the slirp legacy).
+         * If main-loop.c ever switches to poll, G_IO_ERR should be
+         * tested too.  Dispatching G_IO_ERR to both handlers should be
+         * okay, since handlers need to be ready for spurious wakeups.
+         */
+        revents = node->pfd.revents & node->pfd.events;
+        if (revents & (G_IO_IN | G_IO_HUP | G_IO_ERR) && node->io_read) {
+            return true;
+        }
+        if (revents & (G_IO_OUT | G_IO_ERR) && node->io_write) {
+            return true;
+        }
+    }
+
+    return false;
+}
+
+bool aio_poll(AioContext *ctx, bool blocking)
+{
+    static struct timeval tv0;
+    AioHandler *node;
+    fd_set rdfds, wrfds;
+    int max_fd = -1;
+    int ret;
+    bool busy, progress;
+
+    progress = false;
+
+    /*
+     * If there are callbacks left that have been queued, we need to call then.
+     * Do not call select in this case, because it is possible that the caller
+     * does not need a complete flush (as is the case for qemu_aio_wait loops).
+     */
+    if (aio_bh_poll(ctx)) {
+        blocking = false;
+        progress = true;
+    }
+
+    /*
+     * Then dispatch any pending callbacks from the GSource.
+     *
+     * We have to walk very carefully in case qemu_aio_set_fd_handler is
+     * called while we're walking.
+     */
+    node = QLIST_FIRST(&ctx->aio_handlers);
+    while (node) {
+        AioHandler *tmp;
+        int revents;
+
+        ctx->walking_handlers++;
+
+        revents = node->pfd.revents & node->pfd.events;
+        node->pfd.revents = 0;
+
+        /* See comment in aio_pending.  */
+        if (revents & (G_IO_IN | G_IO_HUP | G_IO_ERR) && node->io_read) {
+            node->io_read(node->opaque);
+            progress = true;
+        }
+        if (revents & (G_IO_OUT | G_IO_ERR) && node->io_write) {
+            node->io_write(node->opaque);
+            progress = true;
+        }
+
+        tmp = node;
+        node = QLIST_NEXT(node, node);
+
+        ctx->walking_handlers--;
+
+        if (!ctx->walking_handlers && tmp->deleted) {
+            QLIST_REMOVE(tmp, node);
+            g_free(tmp);
+        }
+    }
+
+    if (progress && !blocking) {
+        return true;
+    }
+
+    ctx->walking_handlers++;
+
+    FD_ZERO(&rdfds);
+    FD_ZERO(&wrfds);
+
+    /* fill fd sets */
+    busy = false;
+    QLIST_FOREACH(node, &ctx->aio_handlers, node) {
+        /* If there aren't pending AIO operations, don't invoke callbacks.
+         * Otherwise, if there are no AIO requests, qemu_aio_wait() would
+         * wait indefinitely.
+         */
+        if (!node->deleted && node->io_flush) {
+            if (node->io_flush(node->opaque) == 0) {
+                continue;
+            }
+            busy = true;
+        }
+        if (!node->deleted && node->io_read) {
+            FD_SET(node->pfd.fd, &rdfds);
+            max_fd = MAX(max_fd, node->pfd.fd + 1);
+        }
+        if (!node->deleted && node->io_write) {
+            FD_SET(node->pfd.fd, &wrfds);
+            max_fd = MAX(max_fd, node->pfd.fd + 1);
+        }
+    }
+
+    ctx->walking_handlers--;
+
+    /* No AIO operations?  Get us out of here */
+    if (!busy) {
+        return progress;
+    }
+
+    /* wait until next event */
+    ret = select(max_fd, &rdfds, &wrfds, NULL, blocking ? NULL : &tv0);
+
+    /* if we have any readable fds, dispatch event */
+    if (ret > 0) {
+        /* we have to walk very carefully in case
+         * qemu_aio_set_fd_handler is called while we're walking */
+        node = QLIST_FIRST(&ctx->aio_handlers);
+        while (node) {
+            AioHandler *tmp;
+
+            ctx->walking_handlers++;
+
+            if (!node->deleted &&
+                FD_ISSET(node->pfd.fd, &rdfds) &&
+                node->io_read) {
+                node->io_read(node->opaque);
+                progress = true;
+            }
+            if (!node->deleted &&
+                FD_ISSET(node->pfd.fd, &wrfds) &&
+                node->io_write) {
+                node->io_write(node->opaque);
+                progress = true;
+            }
+
+            tmp = node;
+            node = QLIST_NEXT(node, node);
+
+            ctx->walking_handlers--;
+
+            if (!ctx->walking_handlers && tmp->deleted) {
+                QLIST_REMOVE(tmp, node);
+                g_free(tmp);
+            }
+        }
+    }
+
+    return progress;
+}
@@ -0,0 +1,215 @@
+/*
+ * QEMU aio implementation
+ *
+ * Copyright IBM Corp., 2008
+ * Copyright Red Hat Inc., 2012
+ *
+ * Authors:
+ *  Anthony Liguori   <aliguori@us.ibm.com>
+ *  Paolo Bonzini     <pbonzini@redhat.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2.  See
+ * the COPYING file in the top-level directory.
+ *
+ * Contributions after 2012-01-13 are licensed under the terms of the
+ * GNU GPL, version 2 or (at your option) any later version.
+ */
+
+#include "qemu-common.h"
+#include "block.h"
+#include "qemu-queue.h"
+#include "qemu_socket.h"
+
+struct AioHandler {
+    EventNotifier *e;
+    EventNotifierHandler *io_notify;
+    AioFlushEventNotifierHandler *io_flush;
+    GPollFD pfd;
+    int deleted;
+    QLIST_ENTRY(AioHandler) node;
+};
+
+void aio_set_event_notifier(AioContext *ctx,
+                            EventNotifier *e,
+                            EventNotifierHandler *io_notify,
+                            AioFlushEventNotifierHandler *io_flush)
+{
+    AioHandler *node;
+
+    QLIST_FOREACH(node, &ctx->aio_handlers, node) {
+        if (node->e == e && !node->deleted) {
+            break;
+        }
+    }
+
+    /* Are we deleting the fd handler? */
+    if (!io_notify) {
+        if (node) {
+            g_source_remove_poll(&ctx->source, &node->pfd);
+
+            /* If the lock is held, just mark the node as deleted */
+            if (ctx->walking_handlers) {
+                node->deleted = 1;
+                node->pfd.revents = 0;
+            } else {
+                /* Otherwise, delete it for real.  We can't just mark it as
+                 * deleted because deleted nodes are only cleaned up after
+                 * releasing the walking_handlers lock.
+                 */
+                QLIST_REMOVE(node, node);
+                g_free(node);
+            }
+        }
+    } else {
+        if (node == NULL) {
+            /* Alloc and insert if it's not already there */
+            node = g_malloc0(sizeof(AioHandler));
+            node->e = e;
+            node->pfd.fd = (uintptr_t)event_notifier_get_handle(e);
+            node->pfd.events = G_IO_IN;
+            QLIST_INSERT_HEAD(&ctx->aio_handlers, node, node);
+
+            g_source_add_poll(&ctx->source, &node->pfd);
+        }
+        /* Update handler with latest information */
+        node->io_notify = io_notify;
+        node->io_flush = io_flush;
+    }
+
+    aio_notify(ctx);
+}
+
+bool aio_pending(AioContext *ctx)
+{
+    AioHandler *node;
+
+    QLIST_FOREACH(node, &ctx->aio_handlers, node) {
+        if (node->pfd.revents && node->io_notify) {
+            return true;
+        }
+    }
+
+    return false;
+}
+
+bool aio_poll(AioContext *ctx, bool blocking)
+{
+    AioHandler *node;
+    HANDLE events[MAXIMUM_WAIT_OBJECTS + 1];
+    bool busy, progress;
+    int count;
+
+    progress = false;
+
+    /*
+     * If there are callbacks left that have been queued, we need to call then.
+     * Do not call select in this case, because it is possible that the caller
+     * does not need a complete flush (as is the case for qemu_aio_wait loops).
+     */
+    if (aio_bh_poll(ctx)) {
+        blocking = false;
+        progress = true;
+    }
+
+    /*
+     * Then dispatch any pending callbacks from the GSource.
+     *
+     * We have to walk very carefully in case qemu_aio_set_fd_handler is
+     * called while we're walking.
+     */
+    node = QLIST_FIRST(&ctx->aio_handlers);
+    while (node) {
+        AioHandler *tmp;
+
+        ctx->walking_handlers++;
+
+        if (node->pfd.revents && node->io_notify) {
+            node->pfd.revents = 0;
+            node->io_notify(node->e);
+            progress = true;
+        }
+
+        tmp = node;
+        node = QLIST_NEXT(node, node);
+
+        ctx->walking_handlers--;
+
+        if (!ctx->walking_handlers && tmp->deleted) {
+            QLIST_REMOVE(tmp, node);
+            g_free(tmp);
+        }
+    }
+
+    if (progress && !blocking) {
+        return true;
+    }
+
+    ctx->walking_handlers++;
+
+    /* fill fd sets */
+    busy = false;
+    count = 0;
+    QLIST_FOREACH(node, &ctx->aio_handlers, node) {
+        /* If there aren't pending AIO operations, don't invoke callbacks.
+         * Otherwise, if there are no AIO requests, qemu_aio_wait() would
+         * wait indefinitely.
+         */
+        if (!node->deleted && node->io_flush) {
+            if (node->io_flush(node->e) == 0) {
+                continue;
+            }
+            busy = true;
+        }
+        if (!node->deleted && node->io_notify) {
+            events[count++] = event_notifier_get_handle(node->e);
+        }
+    }
+
+    ctx->walking_handlers--;
+
+    /* No AIO operations?  Get us out of here */
+    if (!busy) {
+        return progress;
+    }
+
+    /* wait until next event */
+    for (;;) {
+        int timeout = blocking ? INFINITE : 0;
+        int ret = WaitForMultipleObjects(count, events, FALSE, timeout);
+
+        /* if we have any signaled events, dispatch event */
+        if ((DWORD) (ret - WAIT_OBJECT_0) >= count) {
+            break;
+        }
+
+        blocking = false;
+
+        /* we have to walk very carefully in case
+         * qemu_aio_set_fd_handler is called while we're walking */
+        node = QLIST_FIRST(&ctx->aio_handlers);
+        while (node) {
+            AioHandler *tmp;
+
+            ctx->walking_handlers++;
+
+            if (!node->deleted &&
+                event_notifier_get_handle(node->e) == events[ret - WAIT_OBJECT_0] &&
+                node->io_notify) {
+                node->io_notify(node->e);
+                progress = true;
+            }
+
+            tmp = node;
+            node = QLIST_NEXT(node, node);
+
+            ctx->walking_handlers--;
+
+            if (!ctx->walking_handlers && tmp->deleted) {
+                QLIST_REMOVE(tmp, node);
+                g_free(tmp);
+            }
+        }
+    }
+
+    return progress;
+}
@@ -1,194 +0,0 @@
-/*
- * QEMU aio implementation
- *
- * Copyright IBM, Corp. 2008
- *
- * Authors:
- *  Anthony Liguori   <aliguori@us.ibm.com>
- *
- * This work is licensed under the terms of the GNU GPL, version 2.  See
- * the COPYING file in the top-level directory.
- *
- * Contributions after 2012-01-13 are licensed under the terms of the
- * GNU GPL, version 2 or (at your option) any later version.
- */
-
-#include "qemu-common.h"
-#include "block.h"
-#include "qemu-queue.h"
-#include "qemu_socket.h"
-
-typedef struct AioHandler AioHandler;
-
-/* The list of registered AIO handlers */
-static QLIST_HEAD(, AioHandler) aio_handlers;
-
-/* This is a simple lock used to protect the aio_handlers list.  Specifically,
- * it's used to ensure that no callbacks are removed while we're walking and
- * dispatching callbacks.
- */
-static int walking_handlers;
-
-struct AioHandler
-{
-    int fd;
-    IOHandler *io_read;
-    IOHandler *io_write;
-    AioFlushHandler *io_flush;
-    int deleted;
-    void *opaque;
-    QLIST_ENTRY(AioHandler) node;
-};
-
-static AioHandler *find_aio_handler(int fd)
-{
-    AioHandler *node;
-
-    QLIST_FOREACH(node, &aio_handlers, node) {
-        if (node->fd == fd)
-            if (!node->deleted)
-                return node;
-    }
-
-    return NULL;
-}
-
-int qemu_aio_set_fd_handler(int fd,
-                            IOHandler *io_read,
-                            IOHandler *io_write,
-                            AioFlushHandler *io_flush,
-                            void *opaque)
-{
-    AioHandler *node;
-
-    node = find_aio_handler(fd);
-
-    /* Are we deleting the fd handler? */
-    if (!io_read && !io_write) {
-        if (node) {
-            /* If the lock is held, just mark the node as deleted */
-            if (walking_handlers)
-                node->deleted = 1;
-            else {
-                /* Otherwise, delete it for real.  We can't just mark it as
-                 * deleted because deleted nodes are only cleaned up after
-                 * releasing the walking_handlers lock.
-                 */
-                QLIST_REMOVE(node, node);
-                g_free(node);
-            }
-        }
-    } else {
-        if (node == NULL) {
-            /* Alloc and insert if it's not already there */
-            node = g_malloc0(sizeof(AioHandler));
-            node->fd = fd;
-            QLIST_INSERT_HEAD(&aio_handlers, node, node);
-        }
-        /* Update handler with latest information */
-        node->io_read = io_read;
-        node->io_write = io_write;
-        node->io_flush = io_flush;
-        node->opaque = opaque;
-    }
-
-    qemu_set_fd_handler2(fd, NULL, io_read, io_write, opaque);
-
-    return 0;
-}
-
-void qemu_aio_flush(void)
-{
-    while (qemu_aio_wait());
-}
-
-bool qemu_aio_wait(void)
-{
-    AioHandler *node;
-    fd_set rdfds, wrfds;
-    int max_fd = -1;
-    int ret;
-    bool busy;
-
-    /*
-     * If there are callbacks left that have been queued, we need to call then.
-     * Do not call select in this case, because it is possible that the caller
-     * does not need a complete flush (as is the case for qemu_aio_wait loops).
-     */
-    if (qemu_bh_poll()) {
-        return true;
-    }
-
-    walking_handlers++;
-
-    FD_ZERO(&rdfds);
-    FD_ZERO(&wrfds);
-
-    /* fill fd sets */
-    busy = false;
-    QLIST_FOREACH(node, &aio_handlers, node) {
-        /* If there aren't pending AIO operations, don't invoke callbacks.
-         * Otherwise, if there are no AIO requests, qemu_aio_wait() would
-         * wait indefinitely.
-         */
-        if (node->io_flush) {
-            if (node->io_flush(node->opaque) == 0) {
-                continue;
-            }
-            busy = true;
-        }
-        if (!node->deleted && node->io_read) {
-            FD_SET(node->fd, &rdfds);
-            max_fd = MAX(max_fd, node->fd + 1);
-        }
-        if (!node->deleted && node->io_write) {
-            FD_SET(node->fd, &wrfds);
-            max_fd = MAX(max_fd, node->fd + 1);
-        }
-    }
-
-    walking_handlers--;
-
-    /* No AIO operations?  Get us out of here */
-    if (!busy) {
-        return false;
-    }
-
-    /* wait until next event */
-    ret = select(max_fd, &rdfds, &wrfds, NULL, NULL);
-
-    /* if we have any readable fds, dispatch event */
-    if (ret > 0) {
-        /* we have to walk very carefully in case
-         * qemu_aio_set_fd_handler is called while we're walking */
-        node = QLIST_FIRST(&aio_handlers);
-        while (node) {
-            AioHandler *tmp;
-
-            walking_handlers++;
-
-            if (!node->deleted &&
-                FD_ISSET(node->fd, &rdfds) &&
-                node->io_read) {
-                node->io_read(node->opaque);
-            }
-            if (!node->deleted &&
-                FD_ISSET(node->fd, &wrfds) &&
-                node->io_write) {
-                node->io_write(node->opaque);
-            }
-
-            tmp = node;
-            node = QLIST_NEXT(node, node);
-
-            walking_handlers--;
-
-            if (!walking_handlers && tmp->deleted) {
-                QLIST_REMOVE(tmp, node);
-                g_free(tmp);
-            }
-        }
-    }
-
-    return true;
-}
@@ -34,6 +34,6 @@ int tcg_available(void);
 int kvm_available(void);
 int xen_available(void);

-CpuDefinitionInfoList GCC_WEAK_DECL *arch_query_cpu_definitions(Error **errp);
+CpuDefinitionInfoList *arch_query_cpu_definitions(Error **errp);

 #endif
@@ -26,13 +26,11 @@
 #include "qemu-aio.h"
 #include "main-loop.h"

-/* Anchor of the list of Bottom Halves belonging to the context */
-static struct QEMUBH *first_bh;
-
 /***********************************************************/
 /* bottom halves (can be seen as timers which expire ASAP) */

 struct QEMUBH {
+    AioContext *ctx;
    QEMUBHFunc *cb;
    void *opaque;
    QEMUBH *next;
@@ -41,27 +39,27 @@ struct QEMUBH {
    bool deleted;
 };

-QEMUBH *qemu_bh_new(QEMUBHFunc *cb, void *opaque)
+QEMUBH *aio_bh_new(AioContext *ctx, QEMUBHFunc *cb, void *opaque)
 {
    QEMUBH *bh;
    bh = g_malloc0(sizeof(QEMUBH));
+    bh->ctx = ctx;
    bh->cb = cb;
    bh->opaque = opaque;
-    bh->next = first_bh;
-    first_bh = bh;
+    bh->next = ctx->first_bh;
+    ctx->first_bh = bh;
    return bh;
 }

-int qemu_bh_poll(void)
+int aio_bh_poll(AioContext *ctx)
 {
    QEMUBH *bh, **bhp, *next;
    int ret;
-    static int nesting = 0;

-    nesting++;
+    ctx->walking_bh++;

    ret = 0;
-    for (bh = first_bh; bh; bh = next) {
+    for (bh = ctx->first_bh; bh; bh = next) {
        next = bh->next;
        if (!bh->deleted && bh->scheduled) {
            bh->scheduled = 0;
@@ -72,11 +70,11 @@ int qemu_bh_poll(void)
        }
    }

-    nesting--;
+    ctx->walking_bh--;

    /* remove deleted bhs */
-    if (!nesting) {
-        bhp = &first_bh;
+    if (!ctx->walking_bh) {
+        bhp = &ctx->first_bh;
        while (*bhp) {
            bh = *bhp;
            if (bh->deleted) {
@@ -105,8 +103,7 @@ void qemu_bh_schedule(QEMUBH *bh)
        return;
    bh->scheduled = 1;
    bh->idle = 0;
-    /* stop the currently executing CPU to execute the BH ASAP */
-    qemu_notify_event();
+    aio_notify(bh->ctx);
 }

 void qemu_bh_cancel(QEMUBH *bh)
@@ -120,16 +117,20 @@ void qemu_bh_delete(QEMUBH *bh)
    bh->deleted = 1;
 }

-void qemu_bh_update_timeout(uint32_t *timeout)
+static gboolean
+aio_ctx_prepare(GSource *source, gint    *timeout)
 {
+    AioContext *ctx = (AioContext *) source;
    QEMUBH *bh;
+    bool scheduled = false;

-    for (bh = first_bh; bh; bh = bh->next) {
+    for (bh = ctx->first_bh; bh; bh = bh->next) {
        if (!bh->deleted && bh->scheduled) {
+            scheduled = true;
            if (bh->idle) {
                /* idle bottom halves will be polled at least
                 * every 10ms */
-                *timeout = MIN(10, *timeout);
+                *timeout = 10;
            } else {
                /* non-idle bottom halves will be executed
                 * immediately */
@@ -138,5 +139,86 @@ void qemu_bh_update_timeout(uint32_t *timeout)
            }
        }
    }
+
+    return scheduled;
 }

+static gboolean
+aio_ctx_check(GSource *source)
+{
+    AioContext *ctx = (AioContext *) source;
+    QEMUBH *bh;
+
+    for (bh = ctx->first_bh; bh; bh = bh->next) {
+        if (!bh->deleted && bh->scheduled) {
+            return true;
+	}
+    }
+    return aio_pending(ctx);
+}
+
+static gboolean
+aio_ctx_dispatch(GSource     *source,
+                 GSourceFunc  callback,
+                 gpointer     user_data)
+{
+    AioContext *ctx = (AioContext *) source;
+
+    assert(callback == NULL);
+    aio_poll(ctx, false);
+    return true;
+}
+
+static void
+aio_ctx_finalize(GSource     *source)
+{
+    AioContext *ctx = (AioContext *) source;
+
+    aio_set_event_notifier(ctx, &ctx->notifier, NULL, NULL);
+    event_notifier_cleanup(&ctx->notifier);
+}
+
+static GSourceFuncs aio_source_funcs = {
+    aio_ctx_prepare,
+    aio_ctx_check,
+    aio_ctx_dispatch,
+    aio_ctx_finalize
+};
+
+GSource *aio_get_g_source(AioContext *ctx)
+{
+    g_source_ref(&ctx->source);
+    return &ctx->source;
+}
+
+void aio_notify(AioContext *ctx)
+{
+    event_notifier_set(&ctx->notifier);
+}
+
+AioContext *aio_context_new(void)
+{
+    AioContext *ctx;
+    ctx = (AioContext *) g_source_new(&aio_source_funcs, sizeof(AioContext));
+    event_notifier_init(&ctx->notifier, false);
+    aio_set_event_notifier(ctx, &ctx->notifier, 
+                           (EventNotifierHandler *)
+                           event_notifier_test_and_clear, NULL);
+
+    return ctx;
+}
+
+void aio_context_ref(AioContext *ctx)
+{
+    g_source_ref(&ctx->source);
+}
+
+void aio_context_unref(AioContext *ctx)
+{
+    g_source_unref(&ctx->source);
+}
+
+void aio_flush(AioContext *ctx)
+{
+    while (aio_poll(ctx, true));
+}
@@ -2,13 +2,18 @@ block-obj-y += raw.o cow.o qcow.o vdi.o vmdk.o cloop.o dmg.o bochs.o vpc.o vvfat
 block-obj-y += qcow2.o qcow2-refcount.o qcow2-cluster.o qcow2-snapshot.o qcow2-cache.o
 block-obj-y += qed.o qed-gencb.o qed-l2-cache.o qed-table.o qed-cluster.o
 block-obj-y += qed-check.o
-block-obj-y += parallels.o nbd.o blkdebug.o sheepdog.o blkverify.o
-block-obj-$(CONFIG_WIN32) += raw-win32.o
+block-obj-y += parallels.o blkdebug.o blkverify.o
+block-obj-$(CONFIG_WIN32) += raw-win32.o win32-aio.o
 block-obj-$(CONFIG_POSIX) += raw-posix.o
+block-obj-$(CONFIG_LINUX_AIO) += linux-aio.o
+
+ifeq ($(CONFIG_POSIX),y)
+block-obj-y += nbd.o sheepdog.o
 block-obj-$(CONFIG_LIBISCSI) += iscsi.o
 block-obj-$(CONFIG_CURL) += curl.o
 block-obj-$(CONFIG_RBD) += rbd.o
 block-obj-$(CONFIG_GLUSTERFS) += gluster.o
+endif

 common-obj-y += stream.o
 common-obj-y += commit.o
@@ -9,9 +9,10 @@
 */
 #include "qemu-common.h"
 #include "qemu-aio.h"
-#include "block/raw-posix-aio.h"
+#include "qemu-queue.h"
+#include "block/raw-aio.h"
+#include "event_notifier.h"

-#include <sys/eventfd.h>
 #include <libaio.h>

 /*
@@ -37,7 +38,7 @@ struct qemu_laiocb {

 struct qemu_laio_state {
    io_context_t ctx;
-    int efd;
+    EventNotifier e;
    int count;
 };

@@ -76,29 +77,17 @@ static void qemu_laio_process_completion(struct qemu_laio_state *s,
    qemu_aio_release(laiocb);
 }

-static void qemu_laio_completion_cb(void *opaque)
+static void qemu_laio_completion_cb(EventNotifier *e)
 {
-    struct qemu_laio_state *s = opaque;
+    struct qemu_laio_state *s = container_of(e, struct qemu_laio_state, e);

-    while (1) {
+    while (event_notifier_test_and_clear(&s->e)) {
        struct io_event events[MAX_EVENTS];
-        uint64_t val;
-        ssize_t ret;
        struct timespec ts = { 0 };
        int nevents, i;

        do {
-            ret = read(s->efd, &val, sizeof(val));
-        } while (ret == -1 && errno == EINTR);
-
-        if (ret == -1 && errno == EAGAIN)
-            break;
-
-        if (ret != 8)
-            break;
-
-        do {
-            nevents = io_getevents(s->ctx, val, MAX_EVENTS, events, &ts);
+            nevents = io_getevents(s->ctx, MAX_EVENTS, MAX_EVENTS, events, &ts);
        } while (nevents == -EINTR);

        for (i = 0; i < nevents; i++) {
@@ -112,9 +101,9 @@ static void qemu_laio_completion_cb(void *opaque)
    }
 }

-static int qemu_laio_flush_cb(void *opaque)
+static int qemu_laio_flush_cb(EventNotifier *e)
 {
-    struct qemu_laio_state *s = opaque;
+    struct qemu_laio_state *s = container_of(e, struct qemu_laio_state, e);

    return (s->count > 0) ? 1 : 0;
 }
@@ -146,8 +135,9 @@ static void laio_cancel(BlockDriverAIOCB *blockacb)
     * We might be able to do this slightly more optimal by removing the
     * O_NONBLOCK flag.
     */
-    while (laiocb->ret == -EINPROGRESS)
-        qemu_laio_completion_cb(laiocb->ctx);
+    while (laiocb->ret == -EINPROGRESS) {
+        qemu_laio_completion_cb(&laiocb->ctx->e);
+    }
 }

 static AIOPool laio_pool = {
@@ -186,7 +176,7 @@ BlockDriverAIOCB *laio_submit(BlockDriverState *bs, void *aio_ctx, int fd,
                        __func__, type);
        goto out_free_aiocb;
    }
-    io_set_eventfd(&laiocb->iocb, s->efd);
+    io_set_eventfd(&laiocb->iocb, event_notifier_get_fd(&s->e));
    s->count++;

    if (io_submit(s->ctx, 1, &iocbs) < 0)
@@ -205,21 +195,21 @@ void *laio_init(void)
    struct qemu_laio_state *s;

    s = g_malloc0(sizeof(*s));
-    s->efd = eventfd(0, 0);
-    if (s->efd == -1)
+    if (event_notifier_init(&s->e, false) < 0) {
        goto out_free_state;
-    fcntl(s->efd, F_SETFL, O_NONBLOCK);
+    }

-    if (io_setup(MAX_EVENTS, &s->ctx) != 0)
+    if (io_setup(MAX_EVENTS, &s->ctx) != 0) {
        goto out_close_efd;
+    }

-    qemu_aio_set_fd_handler(s->efd, qemu_laio_completion_cb, NULL,
-        qemu_laio_flush_cb, s);
+    qemu_aio_set_event_notifier(&s->e, qemu_laio_completion_cb,
+                                qemu_laio_flush_cb);

    return s;

 out_close_efd:
-    close(s->efd);
+    event_notifier_cleanup(&s->e);
 out_free_state:
    g_free(s);
    return NULL;
@@ -1,5 +1,5 @@
 /*
- * QEMU Posix block I/O backend AIO support
+ * Declarations for AIO in the raw protocol
 *
 * Copyright IBM, Corp. 2008
 *
@@ -12,8 +12,8 @@
 * Contributions after 2012-01-13 are licensed under the terms of the
 * GNU GPL, version 2 or (at your option) any later version.
 */
-#ifndef QEMU_RAW_POSIX_AIO_H
-#define QEMU_RAW_POSIX_AIO_H
+#ifndef QEMU_RAW_AIO_H
+#define QEMU_RAW_AIO_H

 /* AIO request types */
 #define QEMU_AIO_READ         0x0001
@@ -27,19 +27,22 @@
 #define QEMU_AIO_MISALIGNED   0x1000


-/* posix-aio-compat.c - thread pool based implementation */
-int paio_init(void);
-BlockDriverAIOCB *paio_submit(BlockDriverState *bs, int fd,
-        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
-        BlockDriverCompletionFunc *cb, void *opaque, int type);
-BlockDriverAIOCB *paio_ioctl(BlockDriverState *bs, int fd,
-        unsigned long int req, void *buf,
-        BlockDriverCompletionFunc *cb, void *opaque);
-
 /* linux-aio.c - Linux native implementation */
+#ifdef CONFIG_LINUX_AIO
 void *laio_init(void);
 BlockDriverAIOCB *laio_submit(BlockDriverState *bs, void *aio_ctx, int fd,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
        BlockDriverCompletionFunc *cb, void *opaque, int type);
+#endif

-#endif /* QEMU_RAW_POSIX_AIO_H */
+#ifdef _WIN32
+typedef struct QEMUWin32AIOState QEMUWin32AIOState;
+QEMUWin32AIOState *win32_aio_init(void);
+int win32_aio_attach(QEMUWin32AIOState *aio, HANDLE hfile);
+BlockDriverAIOCB *win32_aio_submit(BlockDriverState *bs,
+        QEMUWin32AIOState *aio, HANDLE hfile,
+        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+        BlockDriverCompletionFunc *cb, void *opaque, int type);
+#endif
+
+#endif /* QEMU_RAW_AIO_H */
@@ -27,7 +27,10 @@
 #include "qemu-log.h"
 #include "block_int.h"
 #include "module.h"
-#include "block/raw-posix-aio.h"
+#include "trace.h"
+#include "thread-pool.h"
+#include "iov.h"
+#include "raw-aio.h"

 #if defined(__APPLE__) && (__MACH__)
 #include <paths.h>
@@ -149,6 +152,20 @@ typedef struct BDRVRawReopenState {
 static int fd_open(BlockDriverState *bs);
 static int64_t raw_getlength(BlockDriverState *bs);

+typedef struct RawPosixAIOData {
+    BlockDriverState *bs;
+    int aio_fildes;
+    union {
+        struct iovec *aio_iov;
+        void *aio_ioctl_buf;
+    };
+    int aio_niov;
+    size_t aio_nbytes;
+#define aio_ioctl_cmd   aio_nbytes /* for QEMU_AIO_IOCTL */
+    off_t aio_offset;
+    int aio_type;
+} RawPosixAIOData;
+
 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
 static int cdrom_reopen(BlockDriverState *bs);
 #endif
@@ -266,14 +283,10 @@ static int raw_open_common(BlockDriverState *bs, const char *filename,
    }
    s->fd = fd;

-    /* We're falling back to POSIX AIO in some cases so init always */
-    if (paio_init() < 0) {
-        goto out_close;
-    }
-
 #ifdef CONFIG_LINUX_AIO
    if (raw_set_aio(&s->aio_ctx, &s->use_aio, bdrv_flags)) {
-        goto out_close;
+        qemu_close(fd);
+        return -errno;
    }
 #endif

@@ -284,10 +297,6 @@ static int raw_open_common(BlockDriverState *bs, const char *filename,
 #endif

    return 0;
-
-out_close:
-    qemu_close(fd);
-    return -errno;
 }

 static int raw_open(BlockDriverState *bs, const char *filename, int flags)
@@ -434,6 +443,283 @@ static int qiov_is_aligned(BlockDriverState *bs, QEMUIOVector *qiov)
    return 1;
 }

+static ssize_t handle_aiocb_ioctl(RawPosixAIOData *aiocb)
+{
+    int ret;
+
+    ret = ioctl(aiocb->aio_fildes, aiocb->aio_ioctl_cmd, aiocb->aio_ioctl_buf);
+    if (ret == -1) {
+        return -errno;
+    }
+
+    /*
+     * This looks weird, but the aio code only considers a request
+     * successful if it has written the full number of bytes.
+     *
+     * Now we overload aio_nbytes as aio_ioctl_cmd for the ioctl command,
+     * so in fact we return the ioctl command here to make posix_aio_read()
+     * happy..
+     */
+    return aiocb->aio_nbytes;
+}
+
+static ssize_t handle_aiocb_flush(RawPosixAIOData *aiocb)
+{
+    int ret;
+
+    ret = qemu_fdatasync(aiocb->aio_fildes);
+    if (ret == -1) {
+        return -errno;
+    }
+    return 0;
+}
+
+#ifdef CONFIG_PREADV
+
+static bool preadv_present = true;
+
+static ssize_t
+qemu_preadv(int fd, const struct iovec *iov, int nr_iov, off_t offset)
+{
+    return preadv(fd, iov, nr_iov, offset);
+}
+
+static ssize_t
+qemu_pwritev(int fd, const struct iovec *iov, int nr_iov, off_t offset)
+{
+    return pwritev(fd, iov, nr_iov, offset);
+}
+
+#else
+
+static bool preadv_present = false;
+
+static ssize_t
+qemu_preadv(int fd, const struct iovec *iov, int nr_iov, off_t offset)
+{
+    return -ENOSYS;
+}
+
+static ssize_t
+qemu_pwritev(int fd, const struct iovec *iov, int nr_iov, off_t offset)
+{
+    return -ENOSYS;
+}
+
+#endif
+
+static ssize_t handle_aiocb_rw_vector(RawPosixAIOData *aiocb)
+{
+    ssize_t len;
+
+    do {
+        if (aiocb->aio_type & QEMU_AIO_WRITE)
+            len = qemu_pwritev(aiocb->aio_fildes,
+                               aiocb->aio_iov,
+                               aiocb->aio_niov,
+                               aiocb->aio_offset);
+         else
+            len = qemu_preadv(aiocb->aio_fildes,
+                              aiocb->aio_iov,
+                              aiocb->aio_niov,
+                              aiocb->aio_offset);
+    } while (len == -1 && errno == EINTR);
+
+    if (len == -1) {
+        return -errno;
+    }
+    return len;
+}
+
+/*
+ * Read/writes the data to/from a given linear buffer.
+ *
+ * Returns the number of bytes handles or -errno in case of an error. Short
+ * reads are only returned if the end of the file is reached.
+ */
+static ssize_t handle_aiocb_rw_linear(RawPosixAIOData *aiocb, char *buf)
+{
+    ssize_t offset = 0;
+    ssize_t len;
+
+    while (offset < aiocb->aio_nbytes) {
+        if (aiocb->aio_type & QEMU_AIO_WRITE) {
+            len = pwrite(aiocb->aio_fildes,
+                         (const char *)buf + offset,
+                         aiocb->aio_nbytes - offset,
+                         aiocb->aio_offset + offset);
+        } else {
+            len = pread(aiocb->aio_fildes,
+                        buf + offset,
+                        aiocb->aio_nbytes - offset,
+                        aiocb->aio_offset + offset);
+        }
+        if (len == -1 && errno == EINTR) {
+            continue;
+        } else if (len == -1) {
+            offset = -errno;
+            break;
+        } else if (len == 0) {
+            break;
+        }
+        offset += len;
+    }
+
+    return offset;
+}
+
+static ssize_t handle_aiocb_rw(RawPosixAIOData *aiocb)
+{
+    ssize_t nbytes;
+    char *buf;
+
+    if (!(aiocb->aio_type & QEMU_AIO_MISALIGNED)) {
+        /*
+         * If there is just a single buffer, and it is properly aligned
+         * we can just use plain pread/pwrite without any problems.
+         */
+        if (aiocb->aio_niov == 1) {
+             return handle_aiocb_rw_linear(aiocb, aiocb->aio_iov->iov_base);
+        }
+        /*
+         * We have more than one iovec, and all are properly aligned.
+         *
+         * Try preadv/pwritev first and fall back to linearizing the
+         * buffer if it's not supported.
+         */
+        if (preadv_present) {
+            nbytes = handle_aiocb_rw_vector(aiocb);
+            if (nbytes == aiocb->aio_nbytes ||
+                (nbytes < 0 && nbytes != -ENOSYS)) {
+                return nbytes;
+            }
+            preadv_present = false;
+        }
+
+        /*
+         * XXX(hch): short read/write.  no easy way to handle the reminder
+         * using these interfaces.  For now retry using plain
+         * pread/pwrite?
+         */
+    }
+
+    /*
+     * Ok, we have to do it the hard way, copy all segments into
+     * a single aligned buffer.
+     */
+    buf = qemu_blockalign(aiocb->bs, aiocb->aio_nbytes);
+    if (aiocb->aio_type & QEMU_AIO_WRITE) {
+        char *p = buf;
+        int i;
+
+        for (i = 0; i < aiocb->aio_niov; ++i) {
+            memcpy(p, aiocb->aio_iov[i].iov_base, aiocb->aio_iov[i].iov_len);
+            p += aiocb->aio_iov[i].iov_len;
+        }
+    }
+
+    nbytes = handle_aiocb_rw_linear(aiocb, buf);
+    if (!(aiocb->aio_type & QEMU_AIO_WRITE)) {
+        char *p = buf;
+        size_t count = aiocb->aio_nbytes, copy;
+        int i;
+
+        for (i = 0; i < aiocb->aio_niov && count; ++i) {
+            copy = count;
+            if (copy > aiocb->aio_iov[i].iov_len) {
+                copy = aiocb->aio_iov[i].iov_len;
+            }
+            memcpy(aiocb->aio_iov[i].iov_base, p, copy);
+            p     += copy;
+            count -= copy;
+        }
+    }
+    qemu_vfree(buf);
+
+    return nbytes;
+}
+
+static int aio_worker(void *arg)
+{
+    RawPosixAIOData *aiocb = arg;
+    ssize_t ret = 0;
+
+    switch (aiocb->aio_type & QEMU_AIO_TYPE_MASK) {
+    case QEMU_AIO_READ:
+        ret = handle_aiocb_rw(aiocb);
+        if (ret >= 0 && ret < aiocb->aio_nbytes && aiocb->bs->growable) {
+            iov_memset(aiocb->aio_iov, aiocb->aio_niov, ret,
+                      0, aiocb->aio_nbytes - ret);
+
+            ret = aiocb->aio_nbytes;
+        }
+        if (ret == aiocb->aio_nbytes) {
+            ret = 0;
+        } else if (ret >= 0 && ret < aiocb->aio_nbytes) {
+            ret = -EINVAL;
+        }
+        break;
+    case QEMU_AIO_WRITE:
+        ret = handle_aiocb_rw(aiocb);
+        if (ret == aiocb->aio_nbytes) {
+            ret = 0;
+        } else if (ret >= 0 && ret < aiocb->aio_nbytes) {
+            ret = -EINVAL;
+        }
+        break;
+    case QEMU_AIO_FLUSH:
+        ret = handle_aiocb_flush(aiocb);
+        break;
+    case QEMU_AIO_IOCTL:
+        ret = handle_aiocb_ioctl(aiocb);
+        break;
+    default:
+        fprintf(stderr, "invalid aio request (0x%x)\n", aiocb->aio_type);
+        ret = -EINVAL;
+        break;
+    }
+
+    g_slice_free(RawPosixAIOData, aiocb);
+    return ret;
+}
+
+static BlockDriverAIOCB *paio_submit(BlockDriverState *bs, int fd,
+        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+        BlockDriverCompletionFunc *cb, void *opaque, int type)
+{
+    RawPosixAIOData *acb = g_slice_new(RawPosixAIOData);
+
+    acb->bs = bs;
+    acb->aio_type = type;
+    acb->aio_fildes = fd;
+
+    if (qiov) {
+        acb->aio_iov = qiov->iov;
+        acb->aio_niov = qiov->niov;
+    }
+    acb->aio_nbytes = nb_sectors * 512;
+    acb->aio_offset = sector_num * 512;
+
+    trace_paio_submit(acb, opaque, sector_num, nb_sectors, type);
+    return thread_pool_submit_aio(aio_worker, acb, cb, opaque);
+}
+
+static BlockDriverAIOCB *paio_ioctl(BlockDriverState *bs, int fd,
+        unsigned long int req, void *buf,
+        BlockDriverCompletionFunc *cb, void *opaque)
+{
+    RawPosixAIOData *acb = g_slice_new(RawPosixAIOData);
+
+    acb->bs = bs;
+    acb->aio_type = QEMU_AIO_IOCTL;
+    acb->aio_fildes = fd;
+    acb->aio_offset = 0;
+    acb->aio_ioctl_buf = buf;
+    acb->aio_ioctl_cmd = req;
+
+    return thread_pool_submit_aio(aio_worker, acb, cb, opaque);
+}
+
 static BlockDriverAIOCB *raw_aio_submit(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
        BlockDriverCompletionFunc *cb, void *opaque, int type)
@@ -25,6 +25,10 @@
 #include "qemu-timer.h"
 #include "block_int.h"
 #include "module.h"
+#include "raw-aio.h"
+#include "trace.h"
+#include "thread-pool.h"
+#include "iov.h"
 #include <windows.h>
 #include <winioctl.h>

@@ -32,12 +36,130 @@
 #define FTYPE_CD     1
 #define FTYPE_HARDDISK 2

+static QEMUWin32AIOState *aio;
+
+typedef struct RawWin32AIOData {
+    BlockDriverState *bs;
+    HANDLE hfile;
+    struct iovec *aio_iov;
+    int aio_niov;
+    size_t aio_nbytes;
+    off64_t aio_offset;
+    int aio_type;
+} RawWin32AIOData;
+
 typedef struct BDRVRawState {
    HANDLE hfile;
    int type;
    char drive_path[16]; /* format: "d:\" */
+    QEMUWin32AIOState *aio;
 } BDRVRawState;

+/*
+ * Read/writes the data to/from a given linear buffer.
+ *
+ * Returns the number of bytes handles or -errno in case of an error. Short
+ * reads are only returned if the end of the file is reached.
+ */
+static size_t handle_aiocb_rw(RawWin32AIOData *aiocb)
+{
+    size_t offset = 0;
+    int i;
+
+    for (i = 0; i < aiocb->aio_niov; i++) {
+        OVERLAPPED ov;
+        DWORD ret, ret_count, len;
+
+        memset(&ov, 0, sizeof(ov));
+        ov.Offset = (aiocb->aio_offset + offset);
+        ov.OffsetHigh = (aiocb->aio_offset + offset) >> 32;
+        len = aiocb->aio_iov[i].iov_len;
+        if (aiocb->aio_type & QEMU_AIO_WRITE) {
+            ret = WriteFile(aiocb->hfile, aiocb->aio_iov[i].iov_base,
+                            len, &ret_count, &ov);
+        } else {
+            ret = ReadFile(aiocb->hfile, aiocb->aio_iov[i].iov_base,
+                           len, &ret_count, &ov);
+        }
+        if (!ret) {
+            ret_count = 0;
+        }
+        if (ret_count != len) {
+            break;
+        }
+        offset += len;
+    }
+
+    return offset;
+}
+
+static int aio_worker(void *arg)
+{
+    RawWin32AIOData *aiocb = arg;
+    ssize_t ret = 0;
+    size_t count;
+
+    switch (aiocb->aio_type & QEMU_AIO_TYPE_MASK) {
+    case QEMU_AIO_READ:
+        count = handle_aiocb_rw(aiocb);
+        if (count < aiocb->aio_nbytes && aiocb->bs->growable) {
+            /* A short read means that we have reached EOF. Pad the buffer
+             * with zeros for bytes after EOF. */
+            iov_memset(aiocb->aio_iov, aiocb->aio_niov, count,
+                      0, aiocb->aio_nbytes - count);
+
+            count = aiocb->aio_nbytes;
+        }
+        if (count == aiocb->aio_nbytes) {
+            ret = 0;
+        } else {
+            ret = -EINVAL;
+        }
+        break;
+    case QEMU_AIO_WRITE:
+        count = handle_aiocb_rw(aiocb);
+        if (count == aiocb->aio_nbytes) {
+            count = 0;
+        } else {
+            count = -EINVAL;
+        }
+        break;
+    case QEMU_AIO_FLUSH:
+        if (!FlushFileBuffers(aiocb->hfile)) {
+            return -EIO;
+        }
+        break;
+    default:
+        fprintf(stderr, "invalid aio request (0x%x)\n", aiocb->aio_type);
+        ret = -EINVAL;
+        break;
+    }
+
+    g_slice_free(RawWin32AIOData, aiocb);
+    return ret;
+}
+
+static BlockDriverAIOCB *paio_submit(BlockDriverState *bs, HANDLE hfile,
+        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+        BlockDriverCompletionFunc *cb, void *opaque, int type)
+{
+    RawWin32AIOData *acb = g_slice_new(RawWin32AIOData);
+
+    acb->bs = bs;
+    acb->hfile = hfile;
+    acb->aio_type = type;
+
+    if (qiov) {
+        acb->aio_iov = qiov->iov;
+        acb->aio_niov = qiov->niov;
+    }
+    acb->aio_nbytes = nb_sectors * 512;
+    acb->aio_offset = sector_num * 512;
+
+    trace_paio_submit(acb, opaque, sector_num, nb_sectors, type);
+    return thread_pool_submit_aio(aio_worker, acb, cb, opaque);
+}
+
 int qemu_ftruncate64(int fd, int64_t length)
 {
    LARGE_INTEGER li;
@@ -89,6 +211,9 @@ static void raw_parse_flags(int flags, int *access_flags, DWORD *overlapped)
    }

    *overlapped = FILE_ATTRIBUTE_NORMAL;
+    if (flags & BDRV_O_NATIVE_AIO) {
+        *overlapped |= FILE_FLAG_OVERLAPPED;
+    }
    if (flags & BDRV_O_NOCACHE) {
        *overlapped |= FILE_FLAG_NO_BUFFERING;
    }
@@ -103,6 +228,13 @@ static int raw_open(BlockDriverState *bs, const char *filename, int flags)
    s->type = FTYPE_FILE;

    raw_parse_flags(flags, &access_flags, &overlapped);
+    
+    if ((flags & BDRV_O_NATIVE_AIO) && aio == NULL) {
+        aio = win32_aio_init();
+        if (aio == NULL) {
+            return -EINVAL;
+        }
+    }

    s->hfile = CreateFile(filename, access_flags,
                          FILE_SHARE_READ, NULL,
@@ -112,64 +244,53 @@ static int raw_open(BlockDriverState *bs, const char *filename, int flags)

        if (err == ERROR_ACCESS_DENIED)
            return -EACCES;
-        return -1;
+        return -EINVAL;
+    }
+
+    if (flags & BDRV_O_NATIVE_AIO) {
+        int ret = win32_aio_attach(aio, s->hfile);
+        if (ret < 0) {
+            CloseHandle(s->hfile);
+            return ret;
+        }
+        s->aio = aio;
    }
    return 0;
 }

-static int raw_read(BlockDriverState *bs, int64_t sector_num,
-                    uint8_t *buf, int nb_sectors)
+static BlockDriverAIOCB *raw_aio_readv(BlockDriverState *bs,
+                         int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+                         BlockDriverCompletionFunc *cb, void *opaque)
 {
    BDRVRawState *s = bs->opaque;
-    OVERLAPPED ov;
-    DWORD ret_count;
-    int ret;
-    int64_t offset = sector_num * 512;
-    int count = nb_sectors * 512;
-
-    memset(&ov, 0, sizeof(ov));
-    ov.Offset = offset;
-    ov.OffsetHigh = offset >> 32;
-    ret = ReadFile(s->hfile, buf, count, &ret_count, &ov);
-    if (!ret)
-        return ret_count;
-    if (ret_count == count)
-        ret_count = 0;
-    return ret_count;
-}
-
-static int raw_write(BlockDriverState *bs, int64_t sector_num,
-                     const uint8_t *buf, int nb_sectors)
-{
-    BDRVRawState *s = bs->opaque;
-    OVERLAPPED ov;
-    DWORD ret_count;
-    int ret;
-    int64_t offset = sector_num * 512;
-    int count = nb_sectors * 512;
-
-    memset(&ov, 0, sizeof(ov));
-    ov.Offset = offset;
-    ov.OffsetHigh = offset >> 32;
-    ret = WriteFile(s->hfile, buf, count, &ret_count, &ov);
-    if (!ret)
-        return ret_count;
-    if (ret_count == count)
-        ret_count = 0;
-    return ret_count;
-}
-
-static int raw_flush(BlockDriverState *bs)
-{
-    BDRVRawState *s = bs->opaque;
-    int ret;
-
-    ret = FlushFileBuffers(s->hfile);
-    if (ret == 0) {
-        return -EIO;
+    if (s->aio) {
+        return win32_aio_submit(bs, s->aio, s->hfile, sector_num, qiov,
+                                nb_sectors, cb, opaque, QEMU_AIO_READ); 
+    } else {
+        return paio_submit(bs, s->hfile, sector_num, qiov, nb_sectors,
+                           cb, opaque, QEMU_AIO_READ);
    }
+}

-    return 0;
+static BlockDriverAIOCB *raw_aio_writev(BlockDriverState *bs,
+                          int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+                          BlockDriverCompletionFunc *cb, void *opaque)
+{
+    BDRVRawState *s = bs->opaque;
+    if (s->aio) {
+        return win32_aio_submit(bs, s->aio, s->hfile, sector_num, qiov,
+                                nb_sectors, cb, opaque, QEMU_AIO_WRITE); 
+    } else {
+        return paio_submit(bs, s->hfile, sector_num, qiov, nb_sectors,
+                           cb, opaque, QEMU_AIO_WRITE);
+    }
+}
+
+static BlockDriverAIOCB *raw_aio_flush(BlockDriverState *bs,
+                         BlockDriverCompletionFunc *cb, void *opaque)
+{
+    BDRVRawState *s = bs->opaque;
+    return paio_submit(bs, s->hfile, 0, NULL, 0, cb, opaque, QEMU_AIO_FLUSH);
 }

 static void raw_close(BlockDriverState *bs)
@@ -290,9 +411,9 @@ static BlockDriver bdrv_file = {
    .bdrv_close		= raw_close,
    .bdrv_create	= raw_create,

-    .bdrv_read              = raw_read,
-    .bdrv_write             = raw_write,
-    .bdrv_co_flush_to_disk  = raw_flush,
+    .bdrv_aio_readv     = raw_aio_readv,
+    .bdrv_aio_writev    = raw_aio_writev,
+    .bdrv_aio_flush     = raw_aio_flush,

    .bdrv_truncate	= raw_truncate,
    .bdrv_getlength	= raw_getlength,
@@ -413,9 +534,9 @@ static BlockDriver bdrv_host_device = {
    .bdrv_close		= raw_close,
    .bdrv_has_zero_init = hdev_has_zero_init,

-    .bdrv_read              = raw_read,
-    .bdrv_write             = raw_write,
-    .bdrv_co_flush_to_disk  = raw_flush,
+    .bdrv_aio_readv     = raw_aio_readv,
+    .bdrv_aio_writev    = raw_aio_writev,
+    .bdrv_aio_flush     = raw_aio_flush,

    .bdrv_getlength	= raw_getlength,
    .bdrv_get_allocated_file_size
@@ -0,0 +1,226 @@
+/*
+ * Block driver for RAW files (win32)
+ *
+ * Copyright (c) 2006 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+#include "qemu-common.h"
+#include "qemu-timer.h"
+#include "block_int.h"
+#include "module.h"
+#include "qemu-common.h"
+#include "qemu-aio.h"
+#include "raw-aio.h"
+#include "event_notifier.h"
+#include <windows.h>
+#include <winioctl.h>
+
+#define FTYPE_FILE 0
+#define FTYPE_CD     1
+#define FTYPE_HARDDISK 2
+
+struct QEMUWin32AIOState {
+    HANDLE hIOCP;
+    EventNotifier e;
+    int count;
+};
+
+typedef struct QEMUWin32AIOCB {
+    BlockDriverAIOCB common;
+    struct QEMUWin32AIOState *ctx;
+    int nbytes;
+    OVERLAPPED ov;
+    QEMUIOVector *qiov;
+    void *buf;
+    bool is_read;
+    bool is_linear;
+} QEMUWin32AIOCB;
+
+/*
+ * Completes an AIO request (calls the callback and frees the ACB).
+ */
+static void win32_aio_process_completion(QEMUWin32AIOState *s,
+    QEMUWin32AIOCB *waiocb, DWORD count)
+{
+    int ret;
+    s->count--;
+
+    if (waiocb->ov.Internal != 0) {
+        ret = -EIO;
+    } else {
+        ret = 0;
+        if (count < waiocb->nbytes) {
+            /* Short reads mean EOF, pad with zeros. */
+            if (waiocb->is_read) {
+                qemu_iovec_memset(waiocb->qiov, count, 0,
+                    waiocb->qiov->size - count);
+            } else {
+                ret = -EINVAL;
+            }
+       }
+    }
+
+    if (!waiocb->is_linear) {
+        if (ret == 0 && waiocb->is_read) {
+            QEMUIOVector *qiov = waiocb->qiov;
+            char *p = waiocb->buf;
+            int i;
+
+            for (i = 0; i < qiov->niov; ++i) {
+                memcpy(p, qiov->iov[i].iov_base, qiov->iov[i].iov_len);
+                p += qiov->iov[i].iov_len;
+            }
+            g_free(waiocb->buf);
+        }
+    }
+
+
+    waiocb->common.cb(waiocb->common.opaque, ret);
+    qemu_aio_release(waiocb);
+}
+
+static void win32_aio_completion_cb(EventNotifier *e)
+{
+    QEMUWin32AIOState *s = container_of(e, QEMUWin32AIOState, e);
+    DWORD count;
+    ULONG_PTR key;
+    OVERLAPPED *ov;
+
+    event_notifier_test_and_clear(&s->e);
+    while (GetQueuedCompletionStatus(s->hIOCP, &count, &key, &ov, 0)) {
+        QEMUWin32AIOCB *waiocb = container_of(ov, QEMUWin32AIOCB, ov);
+
+        win32_aio_process_completion(s, waiocb, count);
+    }
+}
+
+static int win32_aio_flush_cb(EventNotifier *e)
+{
+    QEMUWin32AIOState *s = container_of(e, QEMUWin32AIOState, e);
+
+    return (s->count > 0) ? 1 : 0;
+}
+
+static void win32_aio_cancel(BlockDriverAIOCB *blockacb)
+{
+    QEMUWin32AIOCB *waiocb = (QEMUWin32AIOCB *)blockacb;
+
+    /*
+     * CancelIoEx is only supported in Vista and newer.  For now, just
+     * wait for completion.
+     */
+    while (!HasOverlappedIoCompleted(&waiocb->ov)) {
+        qemu_aio_wait();
+    }
+}
+
+static AIOPool win32_aio_pool = {
+    .aiocb_size         = sizeof(QEMUWin32AIOCB),
+    .cancel             = win32_aio_cancel,
+};
+
+BlockDriverAIOCB *win32_aio_submit(BlockDriverState *bs,
+        QEMUWin32AIOState *aio, HANDLE hfile,
+        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+        BlockDriverCompletionFunc *cb, void *opaque, int type)
+{
+    struct QEMUWin32AIOCB *waiocb;
+    uint64_t offset = sector_num * 512;
+    DWORD rc;
+
+    waiocb = qemu_aio_get(&win32_aio_pool, bs, cb, opaque);
+    waiocb->nbytes = nb_sectors * 512;
+    waiocb->qiov = qiov;
+    waiocb->is_read = (type == QEMU_AIO_READ);
+
+    if (qiov->niov > 1) {
+        waiocb->buf = qemu_blockalign(bs, qiov->size);
+        if (type & QEMU_AIO_WRITE) {
+            char *p = waiocb->buf;
+            int i;
+
+            for (i = 0; i < qiov->niov; ++i) {
+                memcpy(p, qiov->iov[i].iov_base, qiov->iov[i].iov_len);
+                p += qiov->iov[i].iov_len;
+            }
+        }
+        waiocb->is_linear = false;
+    } else {
+        waiocb->buf = qiov->iov[0].iov_base;
+        waiocb->is_linear = true;
+    }
+
+    waiocb->ov = (OVERLAPPED) {
+        .Offset = (DWORD) offset,
+        .OffsetHigh = (DWORD) (offset >> 32),
+        .hEvent = event_notifier_get_handle(&aio->e)
+    };
+    aio->count++;
+
+    if (type & QEMU_AIO_READ) {
+        rc = ReadFile(hfile, waiocb->buf, waiocb->nbytes, NULL, &waiocb->ov);
+    } else {
+        rc = WriteFile(hfile, waiocb->buf, waiocb->nbytes, NULL, &waiocb->ov);
+    }
+    if(rc == 0 && GetLastError() != ERROR_IO_PENDING) {
+        goto out_dec_count;
+    }
+    return &waiocb->common;
+
+out_dec_count:
+    aio->count--;
+    qemu_aio_release(waiocb);
+    return NULL;
+}
+
+int win32_aio_attach(QEMUWin32AIOState *aio, HANDLE hfile)
+{
+    if (CreateIoCompletionPort(hfile, aio->hIOCP, (ULONG_PTR) 0, 0) == NULL) {
+        return -EINVAL;
+    } else {
+        return 0;
+    }
+}
+
+QEMUWin32AIOState *win32_aio_init(void)
+{
+    QEMUWin32AIOState *s;
+
+    s = g_malloc0(sizeof(*s));
+    if (event_notifier_init(&s->e, false) < 0) {
+        goto out_free_state;
+    }
+
+    s->hIOCP = CreateIoCompletionPort(INVALID_HANDLE_VALUE, NULL, 0, 0);
+    if (s->hIOCP == NULL) {
+        goto out_close_efd;
+    }
+
+    qemu_aio_set_event_notifier(&s->e, win32_aio_completion_cb,
+                                win32_aio_flush_cb);
+
+    return s;
+
+out_close_efd:
+    event_notifier_cleanup(&s->e);
+out_free_state:
+    g_free(s);
+    return NULL;
+}
@@ -50,16 +50,13 @@
 #   define __printf__ __gnu_printf__
 #  endif
 # endif
-#if defined(_WIN32)
-#define GCC_WEAK __attribute__((weak))
-#define GCC_WEAK_DECL GCC_WEAK
-#else
-#define GCC_WEAK __attribute__((weak))
-#define GCC_WEAK_DECL
-#endif
+# define QEMU_WEAK_ALIAS(newname, oldname) \
+        typeof(oldname) newname __attribute__((weak, alias (#oldname)))
 #else
 #define GCC_ATTR /**/
 #define GCC_FMT_ATTR(n, m)
+#define QEMU_WEAK_ALIAS(newname, oldname) \
+        _Pragma("weak " #newname "=" #oldname)
 #endif

 #endif /* COMPILER_H */
@@ -142,109 +142,6 @@ int qemu_fdatasync(int fd)
 #endif
 }

-/* io vectors */
-
-void qemu_iovec_init(QEMUIOVector *qiov, int alloc_hint)
-{
-    qiov->iov = g_malloc(alloc_hint * sizeof(struct iovec));
-    qiov->niov = 0;
-    qiov->nalloc = alloc_hint;
-    qiov->size = 0;
-}
-
-void qemu_iovec_init_external(QEMUIOVector *qiov, struct iovec *iov, int niov)
-{
-    int i;
-
-    qiov->iov = iov;
-    qiov->niov = niov;
-    qiov->nalloc = -1;
-    qiov->size = 0;
-    for (i = 0; i < niov; i++)
-        qiov->size += iov[i].iov_len;
-}
-
-void qemu_iovec_add(QEMUIOVector *qiov, void *base, size_t len)
-{
-    assert(qiov->nalloc != -1);
-
-    if (qiov->niov == qiov->nalloc) {
-        qiov->nalloc = 2 * qiov->nalloc + 1;
-        qiov->iov = g_realloc(qiov->iov, qiov->nalloc * sizeof(struct iovec));
-    }
-    qiov->iov[qiov->niov].iov_base = base;
-    qiov->iov[qiov->niov].iov_len = len;
-    qiov->size += len;
-    ++qiov->niov;
-}
-
-/*
- * Concatenates (partial) iovecs from src to the end of dst.
- * It starts copying after skipping `soffset' bytes at the
- * beginning of src and adds individual vectors from src to
- * dst copies up to `sbytes' bytes total, or up to the end
- * of src if it comes first.  This way, it is okay to specify
- * very large value for `sbytes' to indicate "up to the end
- * of src".
- * Only vector pointers are processed, not the actual data buffers.
- */
-void qemu_iovec_concat(QEMUIOVector *dst,
-                       QEMUIOVector *src, size_t soffset, size_t sbytes)
-{
-    int i;
-    size_t done;
-    struct iovec *siov = src->iov;
-    assert(dst->nalloc != -1);
-    assert(src->size >= soffset);
-    for (i = 0, done = 0; done < sbytes && i < src->niov; i++) {
-        if (soffset < siov[i].iov_len) {
-            size_t len = MIN(siov[i].iov_len - soffset, sbytes - done);
-            qemu_iovec_add(dst, siov[i].iov_base + soffset, len);
-            done += len;
-            soffset = 0;
-        } else {
-            soffset -= siov[i].iov_len;
-        }
-    }
-    /* return done; */
-}
-
-void qemu_iovec_destroy(QEMUIOVector *qiov)
-{
-    assert(qiov->nalloc != -1);
-
-    qemu_iovec_reset(qiov);
-    g_free(qiov->iov);
-    qiov->nalloc = 0;
-    qiov->iov = NULL;
-}
-
-void qemu_iovec_reset(QEMUIOVector *qiov)
-{
-    assert(qiov->nalloc != -1);
-
-    qiov->niov = 0;
-    qiov->size = 0;
-}
-
-size_t qemu_iovec_to_buf(QEMUIOVector *qiov, size_t offset,
-                         void *buf, size_t bytes)
-{
-    return iov_to_buf(qiov->iov, qiov->niov, offset, buf, bytes);
-}
-
-size_t qemu_iovec_from_buf(QEMUIOVector *qiov, size_t offset,
-                           const void *buf, size_t bytes)
-{
-    return iov_from_buf(qiov->iov, qiov->niov, offset, buf, bytes);
-}
-
-size_t qemu_iovec_memset(QEMUIOVector *qiov, size_t offset,
-                         int fillc, size_t bytes)
-{
-    return iov_memset(qiov->iov, qiov->niov, offset, fillc, bytes);
-}
-
 /*
 * Checks if a buffer is all zeroes
 *
@@ -383,11 +280,6 @@ int qemu_parse_fd(const char *param)
    return fd;
 }

-int qemu_parse_fdset(const char *param)
-{
-    return qemu_parse_fd(param);
-}
-
 /* round down to the nearest power of 2*/
 int64_t pow2floor(int64_t value)
 {
@@ -0,0 +1,120 @@
+/*
+ * event notifier support
+ *
+ * Copyright Red Hat, Inc. 2010
+ *
+ * Authors:
+ *  Michael S. Tsirkin <mst@redhat.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#include "qemu-common.h"
+#include "event_notifier.h"
+#include "qemu-char.h"
+
+#ifdef CONFIG_EVENTFD
+#include <sys/eventfd.h>
+#endif
+
+void event_notifier_init_fd(EventNotifier *e, int fd)
+{
+    e->rfd = fd;
+    e->wfd = fd;
+}
+
+int event_notifier_init(EventNotifier *e, int active)
+{
+    int fds[2];
+    int ret;
+
+#ifdef CONFIG_EVENTFD
+    ret = eventfd(0, EFD_NONBLOCK | EFD_CLOEXEC);
+#else
+    ret = -1;
+    errno = ENOSYS;
+#endif
+    if (ret >= 0) {
+        e->rfd = e->wfd = ret;
+    } else {
+        if (errno != ENOSYS) {
+            return -errno;
+        }
+        if (qemu_pipe(fds) < 0) {
+            return -errno;
+        }
+        ret = fcntl_setfl(fds[0], O_NONBLOCK);
+        if (ret < 0) {
+            ret = -errno;
+            goto fail;
+        }
+        ret = fcntl_setfl(fds[1], O_NONBLOCK);
+        if (ret < 0) {
+            ret = -errno;
+            goto fail;
+        }
+        e->rfd = fds[0];
+        e->wfd = fds[1];
+    }
+    if (active) {
+        event_notifier_set(e);
+    }
+    return 0;
+
+fail:
+    close(fds[0]);
+    close(fds[1]);
+    return ret;
+}
+
+void event_notifier_cleanup(EventNotifier *e)
+{
+    if (e->rfd != e->wfd) {
+        close(e->rfd);
+    }
+    close(e->wfd);
+}
+
+int event_notifier_get_fd(EventNotifier *e)
+{
+    return e->rfd;
+}
+
+int event_notifier_set_handler(EventNotifier *e,
+                               EventNotifierHandler *handler)
+{
+    return qemu_set_fd_handler(e->rfd, (IOHandler *)handler, NULL, e);
+}
+
+int event_notifier_set(EventNotifier *e)
+{
+    static const uint64_t value = 1;
+    ssize_t ret;
+
+    do {
+        ret = write(e->wfd, &value, sizeof(value));
+    } while (ret < 0 && errno == EINTR);
+
+    /* EAGAIN is fine, a read must be pending.  */
+    if (ret < 0 && errno != EAGAIN) {
+        return -errno;
+    }
+    return 0;
+}
+
+int event_notifier_test_and_clear(EventNotifier *e)
+{
+    int value;
+    ssize_t len;
+    char buffer[512];
+
+    /* Drain the notify pipe.  For eventfd, only 8 bytes will be read.  */
+    value = 0;
+    do {
+        len = read(e->rfd, buffer, sizeof(buffer));
+        value |= (len > 0);
+    } while ((len == -1 && errno == EINTR) || len == sizeof(buffer));
+
+    return value;
+}
@@ -0,0 +1,59 @@
+/*
+ * event notifier support
+ *
+ * Copyright Red Hat, Inc. 2010
+ *
+ * Authors:
+ *  Michael S. Tsirkin <mst@redhat.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#include "qemu-common.h"
+#include "event_notifier.h"
+#include "main-loop.h"
+
+int event_notifier_init(EventNotifier *e, int active)
+{
+    e->event = CreateEvent(NULL, FALSE, FALSE, NULL);
+    assert(e->event);
+    return 0;
+}
+
+void event_notifier_cleanup(EventNotifier *e)
+{
+    CloseHandle(e->event);
+}
+
+HANDLE event_notifier_get_handle(EventNotifier *e)
+{
+    return e->event;
+}
+
+int event_notifier_set_handler(EventNotifier *e,
+                               EventNotifierHandler *handler)
+{
+    if (handler) {
+        return qemu_add_wait_object(e->event, (IOHandler *)handler, e);
+    } else {
+        qemu_del_wait_object(e->event, (IOHandler *)handler, e);
+        return 0;
+    }
+}
+
+int event_notifier_set(EventNotifier *e)
+{
+    SetEvent(e->event);
+    return 0;
+}
+
+int event_notifier_test_and_clear(EventNotifier *e)
+{
+    int ret = WaitForSingleObject(e->event, 0);
+    if (ret == WAIT_OBJECT_0) {
+        ResetEvent(e->event);
+        return true;
+    }
+    return false;
+}
@@ -1,67 +0,0 @@
-/*
- * event notifier support
- *
- * Copyright Red Hat, Inc. 2010
- *
- * Authors:
- *  Michael S. Tsirkin <mst@redhat.com>
- *
- * This work is licensed under the terms of the GNU GPL, version 2 or later.
- * See the COPYING file in the top-level directory.
- */
-
-#include "qemu-common.h"
-#include "event_notifier.h"
-#include "qemu-char.h"
-
-#ifdef CONFIG_EVENTFD
-#include <sys/eventfd.h>
-#endif
-
-void event_notifier_init_fd(EventNotifier *e, int fd)
-{
-    e->fd = fd;
-}
-
-int event_notifier_init(EventNotifier *e, int active)
-{
-#ifdef CONFIG_EVENTFD
-    int fd = eventfd(!!active, EFD_NONBLOCK | EFD_CLOEXEC);
-    if (fd < 0)
-        return -errno;
-    e->fd = fd;
-    return 0;
-#else
-    return -ENOSYS;
-#endif
-}
-
-void event_notifier_cleanup(EventNotifier *e)
-{
-    close(e->fd);
-}
-
-int event_notifier_get_fd(EventNotifier *e)
-{
-    return e->fd;
-}
-
-int event_notifier_set_handler(EventNotifier *e,
-                               EventNotifierHandler *handler)
-{
-    return qemu_set_fd_handler(e->fd, (IOHandler *)handler, NULL, e);
-}
-
-int event_notifier_set(EventNotifier *e)
-{
-    uint64_t value = 1;
-    int r = write(e->fd, &value, sizeof(value));
-    return r == sizeof(value);
-}
-
-int event_notifier_test_and_clear(EventNotifier *e)
-{
-    uint64_t value;
-    int r = read(e->fd, &value, sizeof(value));
-    return r == sizeof(value);
-}
@@ -15,18 +15,32 @@

 #include "qemu-common.h"

+#ifdef _WIN32
+#include <windows.h>
+#endif
+
 struct EventNotifier {
-    int fd;
+#ifdef _WIN32
+    HANDLE event;
+#else
+    int rfd;
+    int wfd;
+#endif
 };

 typedef void EventNotifierHandler(EventNotifier *);

-void event_notifier_init_fd(EventNotifier *, int fd);
 int event_notifier_init(EventNotifier *, int active);
 void event_notifier_cleanup(EventNotifier *);
-int event_notifier_get_fd(EventNotifier *);
 int event_notifier_set(EventNotifier *);
 int event_notifier_test_and_clear(EventNotifier *);
 int event_notifier_set_handler(EventNotifier *, EventNotifierHandler *);

+#ifdef CONFIG_POSIX
+void event_notifier_init_fd(EventNotifier *, int fd);
+int event_notifier_get_fd(EventNotifier *);
+#else
+HANDLE event_notifier_get_handle(EventNotifier *);
+#endif
+
 #endif
@@ -10,6 +10,7 @@

 #include "ioport.h"
 #include "irq.h"
+#include "qemu-aio.h"
 #include "qemu-file.h"
 #include "vmstate.h"
 #include "qemu-log.h"
--- a/Show More
+++ b/Show More