summary refs log tree commit diff stats
path: root/util
diff options
context:
space:
mode:
Diffstat (limited to 'util')
-rw-r--r--util/aio-posix.c88
-rw-r--r--util/aio-wait.c11
-rw-r--r--util/async.c2
-rw-r--r--util/memfd.c1
-rw-r--r--util/qemu-coroutine.c5
-rw-r--r--util/qemu-error.c40
-rw-r--r--util/trace-events4
7 files changed, 106 insertions, 45 deletions
diff --git a/util/aio-posix.c b/util/aio-posix.c
index 131ba6b4a8..621b3025d8 100644
--- a/util/aio-posix.c
+++ b/util/aio-posix.c
@@ -211,6 +211,7 @@ void aio_set_fd_handler(AioContext *ctx,
     AioHandler *node;
     bool is_new = false;
     bool deleted = false;
+    int poll_disable_change;
 
     qemu_lockcnt_lock(&ctx->list_lock);
 
@@ -244,11 +245,9 @@ void aio_set_fd_handler(AioContext *ctx,
             QLIST_REMOVE(node, node);
             deleted = true;
         }
-
-        if (!node->io_poll) {
-            ctx->poll_disable_cnt--;
-        }
+        poll_disable_change = -!node->io_poll;
     } else {
+        poll_disable_change = !io_poll - (node && !node->io_poll);
         if (node == NULL) {
             /* Alloc and insert if it's not already there */
             node = g_new0(AioHandler, 1);
@@ -257,10 +256,6 @@ void aio_set_fd_handler(AioContext *ctx,
 
             g_source_add_poll(&ctx->source, &node->pfd);
             is_new = true;
-
-            ctx->poll_disable_cnt += !io_poll;
-        } else {
-            ctx->poll_disable_cnt += !io_poll - !node->io_poll;
         }
 
         /* Update handler with latest information */
@@ -274,6 +269,15 @@ void aio_set_fd_handler(AioContext *ctx,
         node->pfd.events |= (io_write ? G_IO_OUT | G_IO_ERR : 0);
     }
 
+    /* No need to order poll_disable_cnt writes against other updates;
+     * the counter is only used to avoid wasting time and latency on
+     * iterated polling when the system call will be ultimately necessary.
+     * Changing handlers is a rare event, and a little wasted polling until
+     * the aio_notify below is not an issue.
+     */
+    atomic_set(&ctx->poll_disable_cnt,
+               atomic_read(&ctx->poll_disable_cnt) + poll_disable_change);
+
     aio_epoll_update(ctx, node, is_new);
     qemu_lockcnt_unlock(&ctx->list_lock);
     aio_notify(ctx);
@@ -486,7 +490,7 @@ static void add_pollfd(AioHandler *node)
     npfd++;
 }
 
-static bool run_poll_handlers_once(AioContext *ctx)
+static bool run_poll_handlers_once(AioContext *ctx, int64_t *timeout)
 {
     bool progress = false;
     AioHandler *node;
@@ -494,9 +498,11 @@ static bool run_poll_handlers_once(AioContext *ctx)
     QLIST_FOREACH_RCU(node, &ctx->aio_handlers, node) {
         if (!node->deleted && node->io_poll &&
             aio_node_check(ctx, node->is_external) &&
-            node->io_poll(node->opaque) &&
-            node->opaque != &ctx->notifier) {
-            progress = true;
+            node->io_poll(node->opaque)) {
+            *timeout = 0;
+            if (node->opaque != &ctx->notifier) {
+                progress = true;
+            }
         }
 
         /* Caller handles freeing deleted nodes.  Don't do it here. */
@@ -518,31 +524,38 @@ static bool run_poll_handlers_once(AioContext *ctx)
  *
  * Returns: true if progress was made, false otherwise
  */
-static bool run_poll_handlers(AioContext *ctx, int64_t max_ns)
+static bool run_poll_handlers(AioContext *ctx, int64_t max_ns, int64_t *timeout)
 {
     bool progress;
-    int64_t end_time;
+    int64_t start_time, elapsed_time;
 
     assert(ctx->notify_me);
     assert(qemu_lockcnt_count(&ctx->list_lock) > 0);
-    assert(ctx->poll_disable_cnt == 0);
-
-    trace_run_poll_handlers_begin(ctx, max_ns);
 
-    end_time = qemu_clock_get_ns(QEMU_CLOCK_REALTIME) + max_ns;
+    trace_run_poll_handlers_begin(ctx, max_ns, *timeout);
 
+    start_time = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
     do {
-        progress = run_poll_handlers_once(ctx);
-    } while (!progress && qemu_clock_get_ns(QEMU_CLOCK_REALTIME) < end_time);
+        progress = run_poll_handlers_once(ctx, timeout);
+        elapsed_time = qemu_clock_get_ns(QEMU_CLOCK_REALTIME) - start_time;
+    } while (!progress && elapsed_time < max_ns
+             && !atomic_read(&ctx->poll_disable_cnt));
 
-    trace_run_poll_handlers_end(ctx, progress);
+    /* If time has passed with no successful polling, adjust *timeout to
+     * keep the same ending time.
+     */
+    if (*timeout != -1) {
+        *timeout -= MIN(*timeout, elapsed_time);
+    }
 
+    trace_run_poll_handlers_end(ctx, progress, *timeout);
     return progress;
 }
 
 /* try_poll_mode:
  * @ctx: the AioContext
- * @blocking: busy polling is only attempted when blocking is true
+ * @timeout: timeout for blocking wait, computed by the caller and updated if
+ *    polling succeeds.
  *
  * ctx->notify_me must be non-zero so this function can detect aio_notify().
  *
@@ -550,19 +563,16 @@ static bool run_poll_handlers(AioContext *ctx, int64_t max_ns)
  *
  * Returns: true if progress was made, false otherwise
  */
-static bool try_poll_mode(AioContext *ctx, bool blocking)
+static bool try_poll_mode(AioContext *ctx, int64_t *timeout)
 {
-    if (blocking && ctx->poll_max_ns && ctx->poll_disable_cnt == 0) {
-        /* See qemu_soonest_timeout() uint64_t hack */
-        int64_t max_ns = MIN((uint64_t)aio_compute_timeout(ctx),
-                             (uint64_t)ctx->poll_ns);
+    /* See qemu_soonest_timeout() uint64_t hack */
+    int64_t max_ns = MIN((uint64_t)*timeout, (uint64_t)ctx->poll_ns);
 
-        if (max_ns) {
-            poll_set_started(ctx, true);
+    if (max_ns && !atomic_read(&ctx->poll_disable_cnt)) {
+        poll_set_started(ctx, true);
 
-            if (run_poll_handlers(ctx, max_ns)) {
-                return true;
-            }
+        if (run_poll_handlers(ctx, max_ns, timeout)) {
+            return true;
         }
     }
 
@@ -571,7 +581,7 @@ static bool try_poll_mode(AioContext *ctx, bool blocking)
     /* Even if we don't run busy polling, try polling once in case it can make
      * progress and the caller will be able to avoid ppoll(2)/epoll_wait(2).
      */
-    return run_poll_handlers_once(ctx);
+    return run_poll_handlers_once(ctx, timeout);
 }
 
 bool aio_poll(AioContext *ctx, bool blocking)
@@ -601,8 +611,14 @@ bool aio_poll(AioContext *ctx, bool blocking)
         start = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
     }
 
-    progress = try_poll_mode(ctx, blocking);
-    if (!progress) {
+    timeout = blocking ? aio_compute_timeout(ctx) : 0;
+    progress = try_poll_mode(ctx, &timeout);
+    assert(!(timeout && progress));
+
+    /* If polling is allowed, non-blocking aio_poll does not need the
+     * system call---a single round of run_poll_handlers_once suffices.
+     */
+    if (timeout || atomic_read(&ctx->poll_disable_cnt)) {
         assert(npfd == 0);
 
         /* fill pollfds */
@@ -616,8 +632,6 @@ bool aio_poll(AioContext *ctx, bool blocking)
             }
         }
 
-        timeout = blocking ? aio_compute_timeout(ctx) : 0;
-
         /* wait until next event */
         if (aio_epoll_check_poll(ctx, pollfds, npfd, timeout)) {
             AioHandler epoll_handler;
diff --git a/util/aio-wait.c b/util/aio-wait.c
index b8a8f86dba..b4877493f8 100644
--- a/util/aio-wait.c
+++ b/util/aio-wait.c
@@ -26,21 +26,22 @@
 #include "qemu/main-loop.h"
 #include "block/aio-wait.h"
 
+AioWait global_aio_wait;
+
 static void dummy_bh_cb(void *opaque)
 {
     /* The point is to make AIO_WAIT_WHILE()'s aio_poll() return */
 }
 
-void aio_wait_kick(AioWait *wait)
+void aio_wait_kick(void)
 {
     /* The barrier (or an atomic op) is in the caller.  */
-    if (atomic_read(&wait->num_waiters)) {
+    if (atomic_read(&global_aio_wait.num_waiters)) {
         aio_bh_schedule_oneshot(qemu_get_aio_context(), dummy_bh_cb, NULL);
     }
 }
 
 typedef struct {
-    AioWait wait;
     bool done;
     QEMUBHFunc *cb;
     void *opaque;
@@ -54,7 +55,7 @@ static void aio_wait_bh(void *opaque)
     data->cb(data->opaque);
 
     data->done = true;
-    aio_wait_kick(&data->wait);
+    aio_wait_kick();
 }
 
 void aio_wait_bh_oneshot(AioContext *ctx, QEMUBHFunc *cb, void *opaque)
@@ -67,5 +68,5 @@ void aio_wait_bh_oneshot(AioContext *ctx, QEMUBHFunc *cb, void *opaque)
     assert(qemu_get_current_aio_context() == qemu_get_aio_context());
 
     aio_bh_schedule_oneshot(ctx, aio_wait_bh, &data);
-    AIO_WAIT_WHILE(&data.wait, ctx, !data.done);
+    AIO_WAIT_WHILE(ctx, !data.done);
 }
diff --git a/util/async.c b/util/async.c
index 05979f8014..c10642a385 100644
--- a/util/async.c
+++ b/util/async.c
@@ -400,7 +400,7 @@ static void co_schedule_bh_cb(void *opaque)
 
         /* Protected by write barrier in qemu_aio_coroutine_enter */
         atomic_set(&co->scheduled, NULL);
-        qemu_coroutine_enter(co);
+        qemu_aio_coroutine_enter(ctx, co);
         aio_context_release(ctx);
     }
 }
diff --git a/util/memfd.c b/util/memfd.c
index d248a53c3c..6287946b61 100644
--- a/util/memfd.c
+++ b/util/memfd.c
@@ -187,6 +187,7 @@ bool qemu_memfd_alloc_check(void)
         int fd;
         void *ptr;
 
+        fd = -1;
         ptr = qemu_memfd_alloc("test", 4096, 0, &fd, NULL);
         memfd_check = ptr ? MEMFD_OK : MEMFD_KO;
         qemu_memfd_free(ptr, 4096, fd);
diff --git a/util/qemu-coroutine.c b/util/qemu-coroutine.c
index 1ba4191b84..2295928d33 100644
--- a/util/qemu-coroutine.c
+++ b/util/qemu-coroutine.c
@@ -198,3 +198,8 @@ bool qemu_coroutine_entered(Coroutine *co)
 {
     return co->caller;
 }
+
+AioContext *coroutine_fn qemu_coroutine_get_aio_context(Coroutine *co)
+{
+    return co->ctx;
+}
diff --git a/util/qemu-error.c b/util/qemu-error.c
index a25d3b94c6..4ab428f7e4 100644
--- a/util/qemu-error.c
+++ b/util/qemu-error.c
@@ -310,3 +310,43 @@ void info_report(const char *fmt, ...)
     vreport(REPORT_TYPE_INFO, fmt, ap);
     va_end(ap);
 }
+
+/*
+ * Like error_report(), except print just once.
+ * If *printed is false, print the message, and flip *printed to true.
+ * Return whether the message was printed.
+ */
+bool error_report_once_cond(bool *printed, const char *fmt, ...)
+{
+    va_list ap;
+
+    assert(printed);
+    if (*printed) {
+        return false;
+    }
+    *printed = true;
+    va_start(ap, fmt);
+    vreport(REPORT_TYPE_ERROR, fmt, ap);
+    va_end(ap);
+    return true;
+}
+
+/*
+ * Like warn_report(), except print just once.
+ * If *printed is false, print the message, and flip *printed to true.
+ * Return whether the message was printed.
+ */
+bool warn_report_once_cond(bool *printed, const char *fmt, ...)
+{
+    va_list ap;
+
+    assert(printed);
+    if (*printed) {
+        return false;
+    }
+    *printed = true;
+    va_start(ap, fmt);
+    vreport(REPORT_TYPE_WARNING, fmt, ap);
+    va_end(ap);
+    return true;
+}
diff --git a/util/trace-events b/util/trace-events
index 4822434c89..79569b7fdf 100644
--- a/util/trace-events
+++ b/util/trace-events
@@ -1,8 +1,8 @@
 # See docs/devel/tracing.txt for syntax documentation.
 
 # util/aio-posix.c
-run_poll_handlers_begin(void *ctx, int64_t max_ns) "ctx %p max_ns %"PRId64
-run_poll_handlers_end(void *ctx, bool progress) "ctx %p progress %d"
+run_poll_handlers_begin(void *ctx, int64_t max_ns, int64_t timeout) "ctx %p max_ns %"PRId64 " timeout %"PRId64
+run_poll_handlers_end(void *ctx, bool progress, int64_t timeout) "ctx %p progress %d new timeout %"PRId64
 poll_shrink(void *ctx, int64_t old, int64_t new) "ctx %p old %"PRId64" new %"PRId64
 poll_grow(void *ctx, int64_t old, int64_t new) "ctx %p old %"PRId64" new %"PRId64