[Qemu-devel] [PATCH 012/156] mirror: fix throttling delay calculation

[Michael Roth <mdroth@linux.vnet.ibm.com>]

From: Paolo Bonzini <pbonzini@redhat.com>

The throttling delay calculation was using an inaccurate sector count to
calculate the time to sleep.  This broke rate-limiting for the block
mirror job.

Move the delay calculation into mirror_iteration() where we know how
many sectors were transferred.  This lets us calculate an accurate delay
time.

Reported-by: Joaquim Barrera <jbarrera@ac.upc.edu>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
(cherry picked from commit cc8c9d6c6f28e4e376a6561a2a31524fd069bc2d)
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
---
 block/mirror.c | 28 +++++++++++++++-------------
 trace-events   |  2 +-
 2 files changed, 16 insertions(+), 14 deletions(-)

diff --git a/block/mirror.c b/block/mirror.c
index 7b95acf..9bfc22f 100644
--- a/block/mirror.c
+++ b/block/mirror.c
@@ -136,11 +136,12 @@ static void mirror_read_complete(void *opaque, int ret)
                     mirror_write_complete, op);
 }
 
-static void coroutine_fn mirror_iteration(MirrorBlockJob *s)
+static uint64_t coroutine_fn mirror_iteration(MirrorBlockJob *s)
 {
     BlockDriverState *source = s->common.bs;
     int nb_sectors, sectors_per_chunk, nb_chunks;
     int64_t end, sector_num, next_chunk, next_sector, hbitmap_next_sector;
+    uint64_t delay_ns;
     MirrorOp *op;
 
     s->sector_num = hbitmap_iter_next(&s->hbi);
@@ -227,7 +228,12 @@ static void coroutine_fn mirror_iteration(MirrorBlockJob *s)
         nb_chunks += added_chunks;
         next_sector += added_sectors;
         next_chunk += added_chunks;
-    } while (next_sector < end);
+        if (!s->synced && s->common.speed) {
+            delay_ns = ratelimit_calculate_delay(&s->limit, added_sectors);
+        } else {
+            delay_ns = 0;
+        }
+    } while (delay_ns == 0 && next_sector < end);
 
     /* Allocate a MirrorOp that is used as an AIO callback.  */
     op = g_slice_new(MirrorOp);
@@ -263,6 +269,7 @@ static void coroutine_fn mirror_iteration(MirrorBlockJob *s)
     trace_mirror_one_iteration(s, sector_num, nb_sectors);
     bdrv_aio_readv(source, sector_num, &op->qiov, nb_sectors,
                    mirror_read_complete, op);
+    return delay_ns;
 }
 
 static void mirror_free_init(MirrorBlockJob *s)
@@ -358,7 +365,7 @@ static void coroutine_fn mirror_run(void *opaque)
     bdrv_dirty_iter_init(bs, &s->hbi);
     last_pause_ns = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
     for (;;) {
-        uint64_t delay_ns;
+        uint64_t delay_ns = 0;
         int64_t cnt;
         bool should_complete;
 
@@ -382,8 +389,10 @@ static void coroutine_fn mirror_run(void *opaque)
                 qemu_coroutine_yield();
                 continue;
             } else if (cnt != 0) {
-                mirror_iteration(s);
-                continue;
+                delay_ns = mirror_iteration(s);
+                if (delay_ns == 0) {
+                    continue;
+                }
             }
         }
 
@@ -428,17 +437,10 @@ static void coroutine_fn mirror_run(void *opaque)
         }
 
         ret = 0;
-        trace_mirror_before_sleep(s, cnt, s->synced);
+        trace_mirror_before_sleep(s, cnt, s->synced, delay_ns);
         if (!s->synced) {
             /* Publish progress */
             s->common.offset = (end - cnt) * BDRV_SECTOR_SIZE;
-
-            if (s->common.speed) {
-                delay_ns = ratelimit_calculate_delay(&s->limit, sectors_per_chunk);
-            } else {
-                delay_ns = 0;
-            }
-
             block_job_sleep_ns(&s->common, QEMU_CLOCK_REALTIME, delay_ns);
             if (block_job_is_cancelled(&s->common)) {
                 break;
diff --git a/trace-events b/trace-events
index 8695e9e..b8887c1 100644
--- a/trace-events
+++ b/trace-events
@@ -81,7 +81,7 @@ mirror_start(void *bs, void *s, void *co, void *opaque) "bs %p s %p co %p opaque
 mirror_restart_iter(void *s, int64_t cnt) "s %p dirty count %"PRId64
 mirror_before_flush(void *s) "s %p"
 mirror_before_drain(void *s, int64_t cnt) "s %p dirty count %"PRId64
-mirror_before_sleep(void *s, int64_t cnt, int synced) "s %p dirty count %"PRId64" synced %d"
+mirror_before_sleep(void *s, int64_t cnt, int synced, uint64_t delay_ns) "s %p dirty count %"PRId64" synced %d delay %"PRIu64"ns"
 mirror_one_iteration(void *s, int64_t sector_num, int nb_sectors) "s %p sector_num %"PRId64" nb_sectors %d"
 mirror_iteration_done(void *s, int64_t sector_num, int nb_sectors, int ret) "s %p sector_num %"PRId64" nb_sectors %d ret %d"
 mirror_yield(void *s, int64_t cnt, int buf_free_count, int in_flight) "s %p dirty count %"PRId64" free buffers %d in_flight %d"
-- 
1.9.1