Bug in the driver initialization causes soft-lockup if firmware
initialization timeout is reached. Polling function bfa_ioc_poll_fwinit()
incorrectly calls bfa_nw_iocpf_timeout() when the timeout is reached.
The problem is that bfa_nw_iocpf_timeout() calls again
bfa_ioc_poll_fwinit()... etc. The bfa_ioc_poll_fwinit() should directly
send timeout event for iocpf and the same should be done if firmware
download into HW fails.
Cc: Rasesh Mody <rasesh.mody@qlogic.com>
Signed-off-by: Ivan Vecera <ivecera@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
        if (status == BFA_STATUS_OK)
                bfa_ioc_lpu_start(ioc);
        else
-               bfa_nw_iocpf_timeout(ioc);
+               bfa_fsm_send_event(&ioc->iocpf, IOCPF_E_TIMEOUT);
 
        return status;
 }
        }
 
        if (ioc->iocpf.poll_time >= BFA_IOC_TOV) {
-               bfa_nw_iocpf_timeout(ioc);
+               bfa_fsm_send_event(&ioc->iocpf, IOCPF_E_TIMEOUT);
        } else {
                ioc->iocpf.poll_time += BFA_IOC_POLL_TOV;
                mod_timer(&ioc->iocpf_timer, jiffies +