This commit adds to the poll_state_synchronize_srcu() header comment describing the issues surrounding SRCU cookie overflow/wrap for the different kernel configurations.
Unified · Split
| @@ -1058,6 +1058,21 @@ EXPORT_SYMBOL_GPL(start_poll_synchronize_srcu); · 1057 unmodified lines | ||
| 1058 | 1058 | * get_state_synchronize_srcu() or start_poll_synchronize_srcu(), and |
| 1059 | 1059 | * returns @true if an SRCU grace period elapsed since the time that the |
| 1060 | 1060 | * cookie was created. |
| 1061 | + * | |
| 1062 | + * Because cookies are finite in size, wrapping/overflow is possible. | |
| 1063 | + * This is more pronounced on 32-bit systems where cookies are 32 bits, | |
| 1064 | + * where in theory wrapping could happen in about 14 hours assuming | |
| 1065 | + * 25-microsecond expedited SRCU grace periods. However, a more likely | |
| 1066 | + * overflow lower bound is on the order of 24 days in the case of | |
| 1067 | + * one-millisecond SRCU grace periods. Of course, wrapping in a 64-bit | |
| 1068 | + * system requires geologic timespans, as in more than seven million years | |
| 1069 | + * even for expedited SRCU grace periods. | |
| 1070 | + * | |
| 1071 | + * Wrapping/overflow is much more of an issue for CONFIG_SMP=n systems | |
| 1072 | + * that also have CONFIG_PREEMPTION=n, which selects Tiny SRCU. This uses | |
| 1073 | + * a 16-bit cookie, which rcutorture routinely wraps in a matter of a | |
| 1074 | + * few minutes. If this proves to be a problem, this counter will be | |
| 1075 | + * expanded to the same size as for Tree SRCU. | |
| 1061 | 1076 | */ |
| 1062 | 1077 | bool poll_state_synchronize_srcu(struct srcu_struct *ssp, unsigned long cookie) |
| 1063 | 1078 | { |