Message ID | 20180410131550.20064.30595.stgit@pasha-VirtualBox (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
diff --git a/util/qemu-timer.c b/util/qemu-timer.c index 2ed1bf2..e1c703c 100644 --- a/util/qemu-timer.c +++ b/util/qemu-timer.c @@ -587,7 +587,10 @@ int64_t timerlistgroup_deadline_ns(QEMUTimerListGroup *tlg) } else { /* Read clock from the replay file and do not calculate the deadline, based on virtual clock. */ - qemu_clock_get_ns(type); + if (atomic_read(&tlg->tl[type]->active_timers) + && tlg->tl[type]->clock->enabled) { + qemu_clock_get_ns(type); + } } } }
Ciro Santilli reported that commit a5ed352596a8b7eb2f9acce34371b944ac3056c4 breaks the execution replay. It happens due to the probing the clock for the new instances of iothread. Probing of the timerlists' clock instead of using them for the deadlines calculation is critical for the replay. However, this probing was made for the timer lists that are empty. In record mode such lists are not used for the deadlines and corresponding clock does not queried. Therefore this patch disables quering the clock for the empty timerlists and disabled clocks. Signed-off-by: Pavel Dovgalyuk <Pavel.Dovgaluk@ispras.ru> --- util/qemu-timer.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)