[v7,18/22] replay: don't process async events when warping the clock

Commit Message

Pavel Dovgalyuk Feb. 27, 2018, 9:53 a.m. UTC

Virtual clock is warped from iothread and vcpu thread. When the hardware
events associated with warp checkpoint, then interrupt delivering may be
non-deterministic if checkpoint is processed in different threads in record
and replay.
This patch disables event processing for clock warp checkpoint and leaves
all hardware events to other checkpoints (e.g., virtual clock).

Signed-off-by: Pavel Dovgalyuk <[e-mail]>

--

v4: added assert for replay_save_events function
---
 replay/replay-events.c |    1 +
 replay/replay.c        |    7 ++++++-
 2 files changed, 7 insertions(+), 1 deletion(-)

Patch

diff --git a/replay/replay-events.c b/replay/replay-events.c
index 54dd9d2..3d5fc8a 100644
--- a/replay/replay-events.c
+++ b/replay/replay-events.c
@@ -205,6 +205,7 @@  static void replay_save_event(Event *event, int checkpoint)
 void replay_save_events(int checkpoint)
 {
     g_assert(replay_mutex_locked());
+    g_assert(checkpoint != CHECKPOINT_CLOCK_WARP_START);
     while (!QTAILQ_EMPTY(&events_list)) {
         Event *event = QTAILQ_FIRST(&events_list);
         replay_save_event(event, checkpoint);
diff --git a/replay/replay.c b/replay/replay.c
index d5c3a66..19721b0 100644
--- a/replay/replay.c
+++ b/replay/replay.c
@@ -211,7 +211,12 @@  bool replay_checkpoint(ReplayCheckpoint checkpoint)
     } else if (replay_mode == REPLAY_MODE_RECORD) {
         g_assert(replay_mutex_locked());
         replay_put_event(EVENT_CHECKPOINT + checkpoint);
-        replay_save_events(checkpoint);
+        /* This checkpoint belongs to several threads.
+           Processing events from different threads is
+           non-deterministic */
+        if (checkpoint != CHECKPOINT_CLOCK_WARP_START) {
+            replay_save_events(checkpoint);
+        }
         res = true;
     }
 out: