X-Git-Url: https://repo.jachan.dev/qemu.git/blobdiff_plain/a97ff30e934c4d673122cf709e8e87f0effbe2f7..2d710006a0da4a9b7ddf5c02d072e178906d0ef6:/aio-posix.c

diff --git a/aio-posix.c b/aio-posix.c
index 88d09e1cfb..d4770336c5 100644
--- a/aio-posix.c
+++ b/aio-posix.c
@@ -23,7 +23,6 @@ struct AioHandler
     GPollFD pfd;
     IOHandler *io_read;
     IOHandler *io_write;
-    AioFlushHandler *io_flush;
     int deleted;
     void *opaque;
     QLIST_ENTRY(AioHandler) node;
@@ -46,7 +45,6 @@ void aio_set_fd_handler(AioContext *ctx,
                         int fd,
                         IOHandler *io_read,
                         IOHandler *io_write,
-                        AioFlushHandler *io_flush,
                         void *opaque)
 {
     AioHandler *node;
@@ -74,7 +72,7 @@ void aio_set_fd_handler(AioContext *ctx,
     } else {
         if (node == NULL) {
             /* Alloc and insert if it's not already there */
-            node = g_malloc0(sizeof(AioHandler));
+            node = g_new0(AioHandler, 1);
             node->pfd.fd = fd;
             QLIST_INSERT_HEAD(&ctx->aio_handlers, node, node);
 
@@ -83,11 +81,10 @@ void aio_set_fd_handler(AioContext *ctx,
         /* 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);
+        node->pfd.events = (io_read ? G_IO_IN | G_IO_HUP | G_IO_ERR : 0);
+        node->pfd.events |= (io_write ? G_IO_OUT | G_IO_ERR : 0);
     }
 
     aio_notify(ctx);
@@ -95,12 +92,15 @@ void aio_set_fd_handler(AioContext *ctx,
 
 void aio_set_event_notifier(AioContext *ctx,
                             EventNotifier *notifier,
-                            EventNotifierHandler *io_read,
-                            AioFlushEventNotifierHandler *io_flush)
+                            EventNotifierHandler *io_read)
 {
     aio_set_fd_handler(ctx, event_notifier_get_fd(notifier),
-                       (IOHandler *)io_read, NULL,
-                       (AioFlushHandler *)io_flush, notifier);
+                       (IOHandler *)io_read, NULL, notifier);
+}
+
+bool aio_prepare(AioContext *ctx)
+{
+    return false;
 }
 
 bool aio_pending(AioContext *ctx)
@@ -110,13 +110,6 @@ bool aio_pending(AioContext *ctx)
     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;
@@ -129,31 +122,22 @@ bool aio_pending(AioContext *ctx)
     return false;
 }
 
-bool aio_poll(AioContext *ctx, bool blocking)
+bool aio_dispatch(AioContext *ctx)
 {
-    static struct timeval tv0;
     AioHandler *node;
-    fd_set rdfds, wrfds;
-    int max_fd = -1;
-    int ret;
-    bool busy, progress;
-
-    progress = false;
+    bool progress = false;
 
     /*
-     * If there are callbacks left that have been queued, we need to call then.
+     * If there are callbacks left that have been queued, we need to call them.
      * 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).
+     * does not need a complete flush (as is the case for aio_poll 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
+     * We have to walk very carefully in case aio_set_fd_handler is
      * called while we're walking.
      */
     node = QLIST_FIRST(&ctx->aio_handlers);
@@ -166,12 +150,19 @@ bool aio_poll(AioContext *ctx, bool blocking)
         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) {
+        if (!node->deleted &&
+            (revents & (G_IO_IN | G_IO_HUP | G_IO_ERR)) &&
+            node->io_read) {
             node->io_read(node->opaque);
-            progress = true;
+
+            /* aio_notify() does not count as progress */
+            if (node->opaque != &ctx->notifier) {
+                progress = true;
+            }
         }
-        if (revents & (G_IO_OUT | G_IO_ERR) && node->io_write) {
+        if (!node->deleted &&
+            (revents & (G_IO_OUT | G_IO_ERR)) &&
+            node->io_write) {
             node->io_write(node->opaque);
             progress = true;
         }
@@ -187,82 +178,122 @@ bool aio_poll(AioContext *ctx, bool blocking)
         }
     }
 
-    if (progress && !blocking) {
-        return true;
+    /* Run our timers */
+    progress |= timerlistgroup_run_timers(&ctx->tlg);
+
+    return progress;
+}
+
+/* These thread-local variables are used only in a small part of aio_poll
+ * around the call to the poll() system call.  In particular they are not
+ * used while aio_poll is performing callbacks, which makes it much easier
+ * to think about reentrancy!
+ *
+ * Stack-allocated arrays would be perfect but they have size limitations;
+ * heap allocation is expensive enough that we want to reuse arrays across
+ * calls to aio_poll().  And because poll() has to be called without holding
+ * any lock, the arrays cannot be stored in AioContext.  Thread-local data
+ * has none of the disadvantages of these three options.
+ */
+static __thread GPollFD *pollfds;
+static __thread AioHandler **nodes;
+static __thread unsigned npfd, nalloc;
+static __thread Notifier pollfds_cleanup_notifier;
+
+static void pollfds_cleanup(Notifier *n, void *unused)
+{
+    g_assert(npfd == 0);
+    g_free(pollfds);
+    g_free(nodes);
+    nalloc = 0;
+}
+
+static void add_pollfd(AioHandler *node)
+{
+    if (npfd == nalloc) {
+        if (nalloc == 0) {
+            pollfds_cleanup_notifier.notify = pollfds_cleanup;
+            qemu_thread_atexit_add(&pollfds_cleanup_notifier);
+            nalloc = 8;
+        } else {
+            g_assert(nalloc <= INT_MAX);
+            nalloc *= 2;
+        }
+        pollfds = g_renew(GPollFD, pollfds, nalloc);
+        nodes = g_renew(AioHandler *, nodes, nalloc);
+    }
+    nodes[npfd] = node;
+    pollfds[npfd] = (GPollFD) {
+        .fd = node->pfd.fd,
+        .events = node->pfd.events,
+    };
+    npfd++;
+}
+
+bool aio_poll(AioContext *ctx, bool blocking)
+{
+    AioHandler *node;
+    int i, ret;
+    bool progress;
+    int64_t timeout;
+
+    aio_context_acquire(ctx);
+    progress = false;
+
+    /* aio_notify can avoid the expensive event_notifier_set if
+     * everything (file descriptors, bottom halves, timers) will
+     * be re-evaluated before the next blocking poll().  This is
+     * already true when aio_poll is called with blocking == false;
+     * if blocking == true, it is only true after poll() returns,
+     * so disable the optimization now.
+     */
+    if (blocking) {
+        atomic_add(&ctx->notify_me, 2);
     }
 
     ctx->walking_handlers++;
 
-    FD_ZERO(&rdfds);
-    FD_ZERO(&wrfds);
+    assert(npfd == 0);
 
-    /* fill fd sets */
-    busy = false;
+    /* fill pollfds */
     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);
+        if (!node->deleted && node->pfd.events) {
+            add_pollfd(node);
         }
     }
 
-    ctx->walking_handlers--;
+    timeout = blocking ? aio_compute_timeout(ctx) : 0;
 
-    /* No AIO operations?  Get us out of here */
-    if (!busy) {
-        return progress;
+    /* wait until next event */
+    if (timeout) {
+        aio_context_release(ctx);
+    }
+    ret = qemu_poll_ns((GPollFD *)pollfds, npfd, timeout);
+    if (blocking) {
+        atomic_sub(&ctx->notify_me, 2);
+    }
+    if (timeout) {
+        aio_context_acquire(ctx);
     }
 
-    /* wait until next event */
-    ret = select(max_fd, &rdfds, &wrfds, NULL, blocking ? NULL : &tv0);
+    aio_notify_accept(ctx);
 
     /* 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);
+        for (i = 0; i < npfd; i++) {
+            nodes[i]->pfd.revents = pollfds[i].revents;
+        }
+    }
 
-            ctx->walking_handlers--;
+    npfd = 0;
+    ctx->walking_handlers--;
 
-            if (!ctx->walking_handlers && tmp->deleted) {
-                QLIST_REMOVE(tmp, node);
-                g_free(tmp);
-            }
-        }
+    /* Run dispatch even if there were no readable fds to run timers */
+    if (aio_dispatch(ctx)) {
+        progress = true;
     }
 
+    aio_context_release(ctx);
+
     return progress;
 }