Lines Matching refs:cycle

88 	dev_dbg(ds->dev, "longest cycle time %lld ns\n", max_cycle_time);
101 * iterate cyclically through the "schedule". Each "cycle" has an entry point
103 * hardware calls each cycle a "subschedule".
105 * Subschedule (cycle) i starts when
123 * |cycle 0|cycle 1|
151 * - cycle 0: iterates the schedule table from 0 to 2 (and back)
152 * - cycle 1: iterates the schedule table from 3 to 5 (and back)
174 int cycle = 0;
300 schedule_entry_points[cycle].subschindx = cycle;
301 schedule_entry_points[cycle].delta = entry_point_delta;
302 schedule_entry_points[cycle].address = schedule_start_idx;
307 for (i = cycle; i < 8; i++)
319 cycle++;
336 schedule_entry_points[cycle].subschindx = cycle;
337 schedule_entry_points[cycle].delta = entry_point_delta;
338 schedule_entry_points[cycle].address = schedule_start_idx;
340 for (i = cycle; i < 8; i++)
371 * phase given by its cycle's @base_time plus its offset within the cycle
374 * - Collisions within one cycle's (actually the longest cycle's) time frame.
376 * occurrence of each event within one cycle time.
377 * - Collisions in the future. Events may not collide within one cycle time,
378 * but if two port schedules don't have the same periodicity (aka the cycle
400 /* Check if the two cycle times are multiples of one another.
411 * the cycle time.
435 * first cycle time.
442 * within the first cycle time.
536 /* The cycle time extension is the amount of time the last cycle from
541 * So don't add insult over injury and just say we don't support cycle
686 * up a delta, which is 200ns), and wrapping around at the end of each cycle.
707 * cycle times have to be multiples of one another anyway, this means the
708 * correction period can simply be the largest cycle time, hence the current
710 * cycle, and therefore predictable.
715 * your best-effort traffic at the beginning of a cycle, and your