Lines Matching refs:scan
1117 * Approximate time to scan a full NUMA task in ms. The task scan period is
1124 /* Portion of address space to scan in MB */
1188 /* For sanitys sake, never scan more PTEs than MAX_SCAN_WINDOW MB/sec. */
1194 unsigned int scan, floor;
1202 scan = sysctl_numa_balancing_scan_period_min / task_nr_scan_windows(p);
1203 return max_t(unsigned int, floor, scan);
1212 /* Scale the maximum scan period with the amount of shared memory. */
1237 /* Scale the maximum scan period with the amount of shared memory. */
2236 * When adapting the scan rate, the period is divided into NUMA_PERIOD_SLOTS
2237 * increments. The more local the fault statistics are, the higher the scan
2238 * period will be for the next scan window. If local/(local+remote) ratio is
2240 * the scan period will decrease. Aim for 70% local accesses.
2246 * Increase the scan period (slow down scanning) if the majority of
2249 * Otherwise, decrease the scan period.
2265 * node is overloaded. In either case, scan slower
2276 * Prepare to scale scan period relative to the current period.
2277 * == NUMA_PERIOD_THRESHOLD scan period stays the same
2278 * < NUMA_PERIOD_THRESHOLD scan period decreases (scan faster)
2279 * >= NUMA_PERIOD_THRESHOLD scan period increases (scan slower)
2495 /* Decay existing window, copy faults since last scan */
2778 * scan rate to slow down when a workload has settled down.
2856 * Enforce maximal scan/migration frequency..
2925 * Try to scan sysctl_numa_balancing_size worth of
2929 * PTEs, scan up to virtpages, to skip through those
2950 * would find the !migratable VMA on the next scan but not reset the
3069 * Allow resets if faults have been trapped before one scan
3075 * Avoid scan adjustments if moving to the preferred
6484 * comparing the average scan cost (tracked in sd->avg_scan_cost) against the