Convert aio_bh_schedule_oneshot() to aio_bh_schedule_oneshot_event(),
which can specify the clock type, making it compatible with
record-replay.
Operations on SCSI reqs do affect target machine state, so it should
use QEMU_CLOCK_VIRTUAL to recorded and replay the bh.
This fixes hangs in record/replay when using SCSI devices.
Signed-off-by: Nicholas Piggin <npig...@gmail.com>
---
hw/scsi/scsi-bus.c | 14 +++++++++++---
1 file changed, 11 insertions(+), 3 deletions(-)
diff --git a/hw/scsi/scsi-bus.c b/hw/scsi/scsi-bus.c
index 2f1678d51e7..16f294ce6b7 100644
--- a/hw/scsi/scsi-bus.c
+++ b/hw/scsi/scsi-bus.c
@@ -166,9 +166,17 @@ static void scsi_device_for_each_req_async(SCSIDevice *s,
/* Paired with blk_dec_in_flight() in scsi_device_for_each_req_async_bh()
*/
blk_inc_in_flight(s->conf.blk);
- aio_bh_schedule_oneshot(blk_get_aio_context(s->conf.blk),
- scsi_device_for_each_req_async_bh,
- data);
+
+ /*
+ * This is called by device reset and does not affect the observable state
+ * of the target (because it is being reset), and by scsi_dma_restart_cb
+ * to restart DMA on vmstate change which also should not affect the state
+ * of the target (XXX is this really true?), so QEMU_CLOCK_REALTIME should
+ * be used to avoid record-replay of the bh event.
+ */
+ aio_bh_schedule_oneshot_event(blk_get_aio_context(s->conf.blk),
+ scsi_device_for_each_req_async_bh,
+ data, QEMU_CLOCK_REALTIME);
}
static void scsi_device_realize(SCSIDevice *s, Error **errp)
--
2.45.2
