[RTEMS Project] #2230: PowerPC with mpc6xx clock: Nanoseconds extension doesn't compensate for pending clock tick, time may go backwards

[RTEMS trac]

#2230: PowerPC with mpc6xx clock: Nanoseconds extension doesn't compensate for
pending clock tick, time may go backwards
-------------------------+-------------------------------------------------
 Reporter:               |      Owner:
  nick.withers           |
     Type:  defect       |     Status:  new
 Priority:  normal       |  Milestone:  4.11
Component:  bsps         |    Version:  4.11
 Severity:  normal       |   Keywords:  clock, time, backwards, race,
                         |  nanoseconds, monotonic
-------------------------+-------------------------------------------------
 The PowerPC mpc6xx clock driver nanoseconds extension
 (''Clock_driver_nanoseconds_since_last_tick()'') is called with interrupts
 disabled and does not compensate for a pending clock tick interrupt when
 called.

 As a result, the times returned by (e.g.) ''clock_gettime(CLOCK_MONOTONIC,
 ...)'' may not be monotonic - "time may go backwards".

 This problem has been discussed on the users and devel mailing lists:
 http://lists.rtems.org/pipermail/users/2014-November/028354.html
 http://lists.rtems.org/pipermail/devel/2014-December/009446.html

 I thought I'd move the discussion into a ticket as I'm so lacking in
 confidence with this change...

--
Ticket URL: <http://devel.rtems.org/ticket/2230>
RTEMS Project <http://www.rtems.org/>
RTEMS Project