vfio-ccw: Reset FSM state to IDLE inside FSM
authorEric Farman <farman@linux.ibm.com>
Tue, 11 May 2021 19:56:30 +0000 (21:56 +0200)
committerCornelia Huck <cohuck@redhat.com>
Wed, 12 May 2021 10:59:50 +0000 (12:59 +0200)
When an I/O request is made, the fsm_io_request() routine
moves the FSM state from IDLE to CP_PROCESSING, and then
fsm_io_helper() moves it to CP_PENDING if the START SUBCHANNEL
received a cc0. Yet, the error case to go from CP_PROCESSING
back to IDLE is done after the FSM call returns.

Let's move this up into the FSM proper, to provide some
better symmetry when unwinding in this case.

Signed-off-by: Eric Farman <farman@linux.ibm.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Acked-by: Matthew Rosato <mjrosato@linux.ibm.com>
Message-Id: <20210511195631.3995081-3-farman@linux.ibm.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
drivers/s390/cio/vfio_ccw_fsm.c
drivers/s390/cio/vfio_ccw_ops.c

index 23e61aa..e435a9c 100644 (file)
@@ -318,6 +318,7 @@ static void fsm_io_request(struct vfio_ccw_private *private,
        }
 
 err_out:
+       private->state = VFIO_CCW_STATE_IDLE;
        trace_vfio_ccw_fsm_io_request(scsw->cmd.fctl, schid,
                                      io_region->ret_code, errstr);
 }
index 491a64c..c57d2a7 100644 (file)
@@ -279,8 +279,6 @@ static ssize_t vfio_ccw_mdev_write_io_region(struct vfio_ccw_private *private,
        }
 
        vfio_ccw_fsm_event(private, VFIO_CCW_EVENT_IO_REQ);
-       if (region->ret_code != 0)
-               private->state = VFIO_CCW_STATE_IDLE;
        ret = (region->ret_code != 0) ? region->ret_code : count;
 
 out_unlock: