update notes

1 files changed, 11 insertions, 11 deletions

diff --git a/xhfc/xhfc_st_state.c b/xhfc/xhfc_st_state.c
index 537b8d9..538fe30 100644
--- a/xhfc/xhfc_st_state.c
+++ b/xhfc/xhfc_st_state.c
@@ -70,30 +70,30 @@ void deactivate_request(struct xhfc_span* s) /* XXX not called */
 }
 
 /* This function is meant to be ran each time an interface changes state;
- * however, if the state change is polled instead of generating an interrupt
- * (what is done currently during the timer interrupt; see enable_interrupts
- * and xhfc_interrupt). In this case, it is possible that two or more state
- * changes happen on the same interface between two interruptions.
- * Additionally, even when using the ST state change interruptions, an
- * interruption can still be overlooked and lead to one of these conditions.
+ * however, the state change is polled (during the timer interrupt; see
+ * enable_interrupts and xhfc_interrupt). Therefore case, it is possible that
+ * two or more state changes happen on the same interface between two
+ * interruptions and that this functions gets only called once. 
+ * Additionally, even when using the ST state change interrupt, an interruption
+ * can still be overlooked and lead to one of these conditions.
  *
  * However, there is no notable consequence because the actions are either:
  *
  * - Starting timers:
- *   there is a problem iff a timer that should be started is not started and
- *   cannot expire when it should
  *   - T3 is only started after an activation request, therefore we always now
  *     when T3 has to start
  *   - T1 is only started when going to state G2, which can only yield G3
- *     automatically. This can be disallowed with V_G2_G3_EN so that the timer
+ *     automatically. This can be disallowed in V_G2_G3_EN so that the timer
  *     can be started before allowing the transition with V_SU_SET_G2_G3.
  * - Stopping timers:
  *   - T1 is only stopped on G2->G3.
  *     From G3:
  *      - G3->G2: if the timer is already running, stop it
  *      - on deactivate requests stop the timer
- *   - T3 is stopped on going to state F7. I don't see how it is possible to
- *     guarantee this.
+ *   - T3 is stopped on going to state F7. I don't think it's possible to
+ *     guarantee this; however except F2->F7 which can not happen when T3 is
+ *     running, every transition to F7 stops the timer, so there is no incoherent
+ *     behaviour;
  * - Changing alarms in DAHDI, which will be done anyway. No point notifying
  *   something that lasts less that 1ms.
  */