Changeset ede4f16 in rtems
- Timestamp:
- 03/07/06 22:09:49 (18 years ago)
- Children:
- ca6df52
- Parents:
- 5b07faf
- Location:
- cpukit/score
- Files:
-
- 2 edited
Legend:
- Unmodified
- Added
- Removed
-
cpukit/score/ChangeLog
r5b07faf rede4f16 1 2006-03-07 Steven Johnson <sjohnson@sakuraindustries.com> 2 3 PR 850/rtems 4 * src/watchdogtickle.c: A Watchdog (used to timeout an event) with 5 a delay of 1 sometimes does not seem to timeout. The problem 6 occurs, because for whatever reason when the watchdog tickle function 7 executes, the watchdog->delta_interval is 0. it is then decremented 8 before being tested, becomes huge and so doesnt time out. It is 9 thought there is a race condition where the watchdog->delta_interval 10 is calculated by reference to a head (also with a delay of 1). But 11 before it can be added after the head, the head is removed, so the 12 new head now has a delay of 0. 13 1 14 2006-03-07 Joel Sherrill <joel@OARcorp.com> 2 15 -
cpukit/score/src/watchdogtickle.c
r5b07faf rede4f16 50 50 51 51 the_watchdog = _Watchdog_First( header ); 52 the_watchdog->delta_interval--; 53 if ( the_watchdog->delta_interval != 0 ) 54 goto leave; 52 53 /* 54 * For some reason, on rare occasions the_watchdog->delta_interval 55 * of the head of the watchdog chain is 0. Before this test was 56 * added, on these occasions an event (which usually was supposed 57 * to have a timeout of 1 tick would have a delta_interval of 0, which 58 * would be decremented to 0xFFFFFFFF by the unprotected 59 * "the_watchdog->delta_interval--;" operation. 60 * This would mean the event would not timeout, and also the chain would 61 * be blocked, because a timeout with a very high number would be at the 62 * head, rather than at the end. 63 * The test "if (the_watchdog->delta_interval != 0)" 64 * here prevents this from occuring. 65 * 66 * We were not able to categorically identify the situation that causes 67 * this, but proved it to be true empirically. So this check causes 68 * correct behaviour in this circumstance. 69 * 70 * The belief is that a race condition exists whereby an event at the head 71 * of the chain is removed (by a pending ISR or higher priority task) 72 * during the _ISR_Flash( level ); in _Watchdog_Insert, but the watchdog 73 * to be inserted has already had its delta_interval adjusted to 0, and 74 * so is added to the head of the chain with a delta_interval of 0. 75 * 76 * Steven Johnson - 12/2005 (gcc-3.2.3 -O3 on powerpc) 77 */ 78 if (the_watchdog->delta_interval != 0) { 79 the_watchdog->delta_interval--; 80 if ( the_watchdog->delta_interval != 0 ) 81 goto leave; 82 } 55 83 56 84 do {
Note: See TracChangeset
for help on using the changeset viewer.