tree: https://github.com/ammarfaizi2/linux-block akpm/mm/mm-unstable head: 1b745fa43c32e6a6ffb4d922114d75019830a9e2 commit: f66f3c26148785882c6292f7ef32fd34d13d1a97 [157/170] mm/demotion: update node_is_toptier to work with memory tiers config: x86_64-rhel-8.3-kselftests (https://download.01.org/0day-ci/archive/20220820/202208201813.3qpwXbu9-lkp@xxxxxxxxx/config) compiler: gcc-11 (Debian 11.3.0-5) 11.3.0 reproduce (this is a W=1 build): # https://github.com/ammarfaizi2/linux-block/commit/f66f3c26148785882c6292f7ef32fd34d13d1a97 git remote add ammarfaizi2-block https://github.com/ammarfaizi2/linux-block git fetch --no-tags ammarfaizi2-block akpm/mm/mm-unstable git checkout f66f3c26148785882c6292f7ef32fd34d13d1a97 # save the config file mkdir build_dir && cp config build_dir/.config make W=1 O=build_dir ARCH=x86_64 SHELL=/bin/bash If you fix the issue, kindly add following tag where applicable Reported-by: kernel test robot <lkp@xxxxxxxxx> All errors (new ones prefixed by >>): kernel/sched/fair.c: In function 'should_numa_migrate_memory': >> kernel/sched/fair.c:1571:14: error: implicit declaration of function 'node_is_toptier' [-Werror=implicit-function-declaration] 1571 | !node_is_toptier(src_nid)) { | ^~~~~~~~~~~~~~~ cc1: some warnings being treated as errors vim +/node_is_toptier +1571 kernel/sched/fair.c 118aa9e6092758 Huang Ying 2022-07-13 1558 10f39042711ba2 Rik van Riel 2014-01-27 1559 bool should_numa_migrate_memory(struct task_struct *p, struct page * page, 10f39042711ba2 Rik van Riel 2014-01-27 1560 int src_nid, int dst_cpu) 10f39042711ba2 Rik van Riel 2014-01-27 1561 { cb361d8cdef699 Jann Horn 2019-07-16 1562 struct numa_group *ng = deref_curr_numa_group(p); 10f39042711ba2 Rik van Riel 2014-01-27 1563 int dst_nid = cpu_to_node(dst_cpu); 10f39042711ba2 Rik van Riel 2014-01-27 1564 int last_cpupid, this_cpupid; 10f39042711ba2 Rik van Riel 2014-01-27 1565 301a39f0d73fef Huang Ying 2022-07-13 1566 /* 301a39f0d73fef Huang Ying 2022-07-13 1567 * The pages in slow memory node should be migrated according 301a39f0d73fef Huang Ying 2022-07-13 1568 * to hot/cold instead of private/shared. 301a39f0d73fef Huang Ying 2022-07-13 1569 */ 301a39f0d73fef Huang Ying 2022-07-13 1570 if (sysctl_numa_balancing_mode & NUMA_BALANCING_MEMORY_TIERING && 301a39f0d73fef Huang Ying 2022-07-13 @1571 !node_is_toptier(src_nid)) { 301a39f0d73fef Huang Ying 2022-07-13 1572 struct pglist_data *pgdat; 118aa9e6092758 Huang Ying 2022-07-13 1573 unsigned long rate_limit; 118aa9e6092758 Huang Ying 2022-07-13 1574 unsigned int latency, th, def_th; 301a39f0d73fef Huang Ying 2022-07-13 1575 301a39f0d73fef Huang Ying 2022-07-13 1576 pgdat = NODE_DATA(dst_nid); 118aa9e6092758 Huang Ying 2022-07-13 1577 if (pgdat_free_space_enough(pgdat)) { 118aa9e6092758 Huang Ying 2022-07-13 1578 /* workload changed, reset hot threshold */ 118aa9e6092758 Huang Ying 2022-07-13 1579 pgdat->nbp_threshold = 0; 301a39f0d73fef Huang Ying 2022-07-13 1580 return true; 118aa9e6092758 Huang Ying 2022-07-13 1581 } 301a39f0d73fef Huang Ying 2022-07-13 1582 118aa9e6092758 Huang Ying 2022-07-13 1583 def_th = sysctl_numa_balancing_hot_threshold; 118aa9e6092758 Huang Ying 2022-07-13 1584 rate_limit = sysctl_numa_balancing_promote_rate_limit << \ 118aa9e6092758 Huang Ying 2022-07-13 1585 (20 - PAGE_SHIFT); 118aa9e6092758 Huang Ying 2022-07-13 1586 numa_promotion_adjust_threshold(pgdat, rate_limit, def_th); 118aa9e6092758 Huang Ying 2022-07-13 1587 118aa9e6092758 Huang Ying 2022-07-13 1588 th = pgdat->nbp_threshold ? : def_th; 301a39f0d73fef Huang Ying 2022-07-13 1589 latency = numa_hint_fault_latency(page); 301a39f0d73fef Huang Ying 2022-07-13 1590 if (latency >= th) 301a39f0d73fef Huang Ying 2022-07-13 1591 return false; 301a39f0d73fef Huang Ying 2022-07-13 1592 23441fcb084b9e Huang Ying 2022-07-13 1593 return !numa_promotion_rate_limit(pgdat, rate_limit, 23441fcb084b9e Huang Ying 2022-07-13 1594 thp_nr_pages(page)); 301a39f0d73fef Huang Ying 2022-07-13 1595 } 301a39f0d73fef Huang Ying 2022-07-13 1596 10f39042711ba2 Rik van Riel 2014-01-27 1597 this_cpupid = cpu_pid_to_cpupid(dst_cpu, current->pid); 37355bdc5a1298 Mel Gorman 2018-10-01 1598 last_cpupid = page_cpupid_xchg_last(page, this_cpupid); 37355bdc5a1298 Mel Gorman 2018-10-01 1599 301a39f0d73fef Huang Ying 2022-07-13 1600 if (!(sysctl_numa_balancing_mode & NUMA_BALANCING_MEMORY_TIERING) && 301a39f0d73fef Huang Ying 2022-07-13 1601 !node_is_toptier(src_nid) && !cpupid_valid(last_cpupid)) 301a39f0d73fef Huang Ying 2022-07-13 1602 return false; 301a39f0d73fef Huang Ying 2022-07-13 1603 37355bdc5a1298 Mel Gorman 2018-10-01 1604 /* 37355bdc5a1298 Mel Gorman 2018-10-01 1605 * Allow first faults or private faults to migrate immediately early in 37355bdc5a1298 Mel Gorman 2018-10-01 1606 * the lifetime of a task. The magic number 4 is based on waiting for 37355bdc5a1298 Mel Gorman 2018-10-01 1607 * two full passes of the "multi-stage node selection" test that is 37355bdc5a1298 Mel Gorman 2018-10-01 1608 * executed below. 37355bdc5a1298 Mel Gorman 2018-10-01 1609 */ 98fa15f34cb379 Anshuman Khandual 2019-03-05 1610 if ((p->numa_preferred_nid == NUMA_NO_NODE || p->numa_scan_seq <= 4) && 37355bdc5a1298 Mel Gorman 2018-10-01 1611 (cpupid_pid_unset(last_cpupid) || cpupid_match_pid(p, last_cpupid))) 37355bdc5a1298 Mel Gorman 2018-10-01 1612 return true; 10f39042711ba2 Rik van Riel 2014-01-27 1613 10f39042711ba2 Rik van Riel 2014-01-27 1614 /* 10f39042711ba2 Rik van Riel 2014-01-27 1615 * Multi-stage node selection is used in conjunction with a periodic 10f39042711ba2 Rik van Riel 2014-01-27 1616 * migration fault to build a temporal task<->page relation. By using 10f39042711ba2 Rik van Riel 2014-01-27 1617 * a two-stage filter we remove short/unlikely relations. 10f39042711ba2 Rik van Riel 2014-01-27 1618 * 10f39042711ba2 Rik van Riel 2014-01-27 1619 * Using P(p) ~ n_p / n_t as per frequentist probability, we can equate 10f39042711ba2 Rik van Riel 2014-01-27 1620 * a task's usage of a particular page (n_p) per total usage of this 10f39042711ba2 Rik van Riel 2014-01-27 1621 * page (n_t) (in a given time-span) to a probability. 10f39042711ba2 Rik van Riel 2014-01-27 1622 * 10f39042711ba2 Rik van Riel 2014-01-27 1623 * Our periodic faults will sample this probability and getting the 10f39042711ba2 Rik van Riel 2014-01-27 1624 * same result twice in a row, given these samples are fully 10f39042711ba2 Rik van Riel 2014-01-27 1625 * independent, is then given by P(n)^2, provided our sample period 10f39042711ba2 Rik van Riel 2014-01-27 1626 * is sufficiently short compared to the usage pattern. 10f39042711ba2 Rik van Riel 2014-01-27 1627 * 10f39042711ba2 Rik van Riel 2014-01-27 1628 * This quadric squishes small probabilities, making it less likely we 10f39042711ba2 Rik van Riel 2014-01-27 1629 * act on an unlikely task<->page relation. 10f39042711ba2 Rik van Riel 2014-01-27 1630 */ 10f39042711ba2 Rik van Riel 2014-01-27 1631 if (!cpupid_pid_unset(last_cpupid) && 10f39042711ba2 Rik van Riel 2014-01-27 1632 cpupid_to_nid(last_cpupid) != dst_nid) 10f39042711ba2 Rik van Riel 2014-01-27 1633 return false; 10f39042711ba2 Rik van Riel 2014-01-27 1634 10f39042711ba2 Rik van Riel 2014-01-27 1635 /* Always allow migrate on private faults */ 10f39042711ba2 Rik van Riel 2014-01-27 1636 if (cpupid_match_pid(p, last_cpupid)) 10f39042711ba2 Rik van Riel 2014-01-27 1637 return true; 10f39042711ba2 Rik van Riel 2014-01-27 1638 10f39042711ba2 Rik van Riel 2014-01-27 1639 /* A shared fault, but p->numa_group has not been set up yet. */ 10f39042711ba2 Rik van Riel 2014-01-27 1640 if (!ng) 10f39042711ba2 Rik van Riel 2014-01-27 1641 return true; 10f39042711ba2 Rik van Riel 2014-01-27 1642 10f39042711ba2 Rik van Riel 2014-01-27 1643 /* 4142c3ebb685bb Rik van Riel 2016-01-25 1644 * Destination node is much more heavily used than the source 4142c3ebb685bb Rik van Riel 2016-01-25 1645 * node? Allow migration. 10f39042711ba2 Rik van Riel 2014-01-27 1646 */ 4142c3ebb685bb Rik van Riel 2016-01-25 1647 if (group_faults_cpu(ng, dst_nid) > group_faults_cpu(ng, src_nid) * 4142c3ebb685bb Rik van Riel 2016-01-25 1648 ACTIVE_NODE_FRACTION) 10f39042711ba2 Rik van Riel 2014-01-27 1649 return true; 10f39042711ba2 Rik van Riel 2014-01-27 1650 10f39042711ba2 Rik van Riel 2014-01-27 1651 /* 4142c3ebb685bb Rik van Riel 2016-01-25 1652 * Distribute memory according to CPU & memory use on each node, 4142c3ebb685bb Rik van Riel 2016-01-25 1653 * with 3/4 hysteresis to avoid unnecessary memory migrations: 4142c3ebb685bb Rik van Riel 2016-01-25 1654 * 4142c3ebb685bb Rik van Riel 2016-01-25 1655 * faults_cpu(dst) 3 faults_cpu(src) 4142c3ebb685bb Rik van Riel 2016-01-25 1656 * --------------- * - > --------------- 4142c3ebb685bb Rik van Riel 2016-01-25 1657 * faults_mem(dst) 4 faults_mem(src) 10f39042711ba2 Rik van Riel 2014-01-27 1658 */ 4142c3ebb685bb Rik van Riel 2016-01-25 1659 return group_faults_cpu(ng, dst_nid) * group_faults(p, src_nid) * 3 > 4142c3ebb685bb Rik van Riel 2016-01-25 1660 group_faults_cpu(ng, src_nid) * group_faults(p, dst_nid) * 4; 10f39042711ba2 Rik van Riel 2014-01-27 1661 } 10f39042711ba2 Rik van Riel 2014-01-27 1662 :::::: The code at line 1571 was first introduced by commit :::::: 301a39f0d73fef87b631e25726f507f244273fa9 memory tiering: hot page selection with hint page fault latency :::::: TO: Huang Ying <ying.huang@xxxxxxxxx> :::::: CC: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> -- 0-DAY CI Kernel Test Service https://01.org/lkp
