On 10/17/2018 05:18 AM, Pierre Morel wrote:
There is a risk that the mediated device is used before all the
data are initialized if it is register too early.
s/register/registered/
Let's register the mediated device when all the data structures
which could be used are initialized.
I need to look at the later patches more, because I think that helps
this situation too. But fixing up this ordering seems good to me.
Signed-off-by: Pierre Morel <pmorel@xxxxxxxxxxxxx>
Reviewed-by: Eric Farman <farman@xxxxxxxxxxxxx>
---
drivers/s390/cio/vfio_ccw_drv.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/s390/cio/vfio_ccw_drv.c b/drivers/s390/cio/vfio_ccw_drv.c
index f47d16b..33fd53f 100644
--- a/drivers/s390/cio/vfio_ccw_drv.c
+++ b/drivers/s390/cio/vfio_ccw_drv.c
@@ -134,14 +134,14 @@ static int vfio_ccw_sch_probe(struct subchannel *sch)
if (ret)
goto out_free;
- ret = vfio_ccw_mdev_reg(sch);
- if (ret)
- goto out_disable;
-
INIT_WORK(&private->io_work, vfio_ccw_sch_io_todo);
atomic_set(&private->avail, 1);
private->state = VFIO_CCW_STATE_STANDBY;
+ ret = vfio_ccw_mdev_reg(sch);
+ if (ret)
+ goto out_disable;
+
return 0;
out_disable:
