From: Nick Piggin This changes sched domains to contain all possible CPUs, and check for online as needed. It's in order to play nicely with CPU hotplug. --- arch/i386/kernel/smpboot.c | 22 +++++++++--------- kernel/sched.c | 53 ++++++++++++++++++++------------------------- 2 files changed, 35 insertions(+), 40 deletions(-) diff -puN arch/i386/kernel/smpboot.c~sched-domains-use-cpu_possible_map arch/i386/kernel/smpboot.c --- 25/arch/i386/kernel/smpboot.c~sched-domains-use-cpu_possible_map 2004-02-28 16:31:46.000000000 -0800 +++ 25-akpm/arch/i386/kernel/smpboot.c 2004-02-28 16:31:46.000000000 -0800 @@ -1136,7 +1136,7 @@ __init void arch_init_sched_domains(void struct sched_group *first_cpu = NULL, *last_cpu = NULL; /* Set up domains */ - for_each_cpu_mask(i, cpu_online_map) { + for_each_cpu(i) { struct sched_domain *cpu_domain = cpu_sched_domain(i); struct sched_domain *phys_domain = &per_cpu(phys_domains, i); struct sched_domain *node_domain = &per_cpu(node_domains, i); @@ -1150,11 +1150,11 @@ __init void arch_init_sched_domains(void phys_domain->span = nodemask; *node_domain = SD_NODE_INIT; - node_domain->span = cpu_online_map; + node_domain->span = cpu_possible_map; } /* Set up CPU (sibling) groups */ - for_each_cpu_mask(i, cpu_online_map) { + for_each_cpu(i) { struct sched_domain *cpu_domain = cpu_sched_domain(i); int j; first_cpu = last_cpu = NULL; @@ -1182,7 +1182,7 @@ __init void arch_init_sched_domains(void int j; cpumask_t nodemask; struct sched_group *node = &sched_group_nodes[i]; - cpus_and(nodemask, node_to_cpumask(i), cpu_online_map); + cpus_and(nodemask, node_to_cpumask(i), cpu_possible_map); if (cpus_empty(nodemask)) continue; @@ -1218,7 +1218,7 @@ __init void arch_init_sched_domains(void for (i = 0; i < MAX_NUMNODES; i++) { struct sched_group *cpu = &sched_group_nodes[i]; cpumask_t nodemask; - cpus_and(nodemask, node_to_cpumask(i), cpu_online_map); + cpus_and(nodemask, node_to_cpumask(i), cpu_possible_map); if (cpus_empty(nodemask)) continue; @@ -1235,7 +1235,7 @@ __init void arch_init_sched_domains(void last_cpu->next = first_cpu; mb(); - for_each_cpu_mask(i, cpu_online_map) { + for_each_cpu(i) { int node = cpu_to_node(i); struct sched_domain *cpu_domain = cpu_sched_domain(i); struct sched_domain *phys_domain = &per_cpu(phys_domains, i); @@ -1262,7 +1262,7 @@ __init void arch_init_sched_domains(void struct sched_group *first_cpu = NULL, *last_cpu = NULL; /* Set up domains */ - for_each_cpu_mask(i, cpu_online_map) { + for_each_cpu(i) { struct sched_domain *cpu_domain = cpu_sched_domain(i); struct sched_domain *phys_domain = &per_cpu(phys_domains, i); @@ -1270,11 +1270,11 @@ __init void arch_init_sched_domains(void cpu_domain->span = cpu_sibling_map[i]; *phys_domain = SD_CPU_INIT; - phys_domain->span = cpu_online_map; + phys_domain->span = cpu_possible_map; } /* Set up CPU (sibling) groups */ - for_each_cpu_mask(i, cpu_online_map) { + for_each_cpu(i) { struct sched_domain *cpu_domain = cpu_sched_domain(i); int j; first_cpu = last_cpu = NULL; @@ -1300,7 +1300,7 @@ __init void arch_init_sched_domains(void first_cpu = last_cpu = NULL; /* Set up physical groups */ - for_each_cpu_mask(i, cpu_online_map) { + for_each_cpu(i) { struct sched_domain *cpu_domain = cpu_sched_domain(i); struct sched_group *cpu = &sched_group_phys[i]; @@ -1320,7 +1320,7 @@ __init void arch_init_sched_domains(void last_cpu->next = first_cpu; mb(); - for_each_cpu_mask(i, cpu_online_map) { + for_each_cpu(i) { struct sched_domain *cpu_domain = cpu_sched_domain(i); struct sched_domain *phys_domain = &per_cpu(phys_domains, i); struct sched_group *cpu_group = &sched_group_cpus[i]; diff -puN kernel/sched.c~sched-domains-use-cpu_possible_map kernel/sched.c --- 25/kernel/sched.c~sched-domains-use-cpu_possible_map 2004-02-28 16:31:46.000000000 -0800 +++ 25-akpm/kernel/sched.c 2004-02-28 16:31:46.000000000 -0800 @@ -645,6 +645,7 @@ static inline unsigned long get_high_cpu #if defined(CONFIG_SMP) && defined(ARCH_HAS_SCHED_WAKE_BALANCE) static int sched_balance_wake(int cpu, task_t *p) { + cpumask_t tmp; struct sched_domain *domain; int i; @@ -655,10 +656,8 @@ static int sched_balance_wake(int cpu, t if (!(domain->flags & SD_FLAG_WAKE)) return cpu; - for_each_cpu_mask(i, domain->span) { - if (!cpu_online(i)) - continue; - + cpus_and(tmp, domain->span, cpu_online_map); + for_each_cpu_mask(i, tmp) { if (!cpu_isset(i, p->cpus_allowed)) continue; @@ -1147,16 +1146,15 @@ out: */ static int sched_best_cpu(struct task_struct *p, struct sched_domain *domain) { + cpumask_t tmp; int i, min_load, this_cpu, best_cpu; best_cpu = this_cpu = task_cpu(p); min_load = INT_MAX; - for_each_online_cpu(i) { + cpus_and(tmp, domain->span, cpu_online_map); + for_each_cpu_mask(i, tmp) { unsigned long load; - if (!cpu_isset(i, domain->span)) - continue; - if (i == this_cpu) load = get_low_cpu_load(i, 0); else @@ -1371,6 +1369,7 @@ find_busiest_group(struct sched_domain * modify = 1; do { + cpumask_t tmp; unsigned long load; int local_group; int i, nr_cpus = 0; @@ -1379,10 +1378,8 @@ find_busiest_group(struct sched_domain * /* Tally up the load of all CPUs in the group */ avg_load = 0; - for_each_cpu_mask(i, group->cpumask) { - if (!cpu_online(i)) - continue; - + cpus_and(tmp, group->cpumask, cpu_online_map); + for_each_cpu_mask(i, tmp) { /* Bias balancing toward cpus of our domain */ if (local_group) { load = get_high_cpu_load(i, modify); @@ -1494,16 +1491,15 @@ out_balanced: */ static runqueue_t *find_busiest_queue(struct sched_group *group) { + cpumask_t tmp; int i; unsigned long max_load = 0; runqueue_t *busiest = NULL; - for_each_cpu_mask(i, group->cpumask) { + cpus_and(tmp, group->cpumask, cpu_online_map); + for_each_cpu_mask(i, tmp) { unsigned long load; - if (!cpu_online(i)) - continue; - load = get_low_cpu_load(i, 0); if (load >= max_load) { @@ -1689,16 +1685,15 @@ static void active_load_balance(runqueue group = sd->groups; do { + cpumask_t tmp; runqueue_t *rq; int push_cpu = 0, nr = 0; if (group == busy_group) goto next_group; - for_each_cpu_mask(i, group->cpumask) { - if (!cpu_online(i)) - continue; - + cpus_and(tmp, group->cpumask, cpu_online_map); + for_each_cpu_mask(i, tmp) { if (!idle_cpu(i)) goto next_group; push_cpu = i; @@ -3224,17 +3219,17 @@ static void __init arch_init_sched_domai struct sched_group *first_node = NULL, *last_node = NULL; /* Set up domains */ - for_each_cpu_mask(i, cpu_online_map) { + for_each_cpu(i) { int node = cpu_to_node(i); cpumask_t nodemask = node_to_cpumask(node); struct sched_domain *node_domain = &per_cpu(node_domains, i); struct sched_domain *cpu_domain = cpu_sched_domain(i); *node_domain = SD_NODE_INIT; - node_domain->span = cpu_online_map; + node_domain->span = cpu_possible_map; *cpu_domain = SD_CPU_INIT; - cpus_and(cpu_domain->span, nodemask, cpu_online_map); + cpus_and(cpu_domain->span, nodemask, cpu_possible_map); cpu_domain->parent = node_domain; } @@ -3246,7 +3241,7 @@ static void __init arch_init_sched_domai struct sched_group *node = &sched_group_nodes[i]; cpumask_t tmp = node_to_cpumask(i); - cpus_and(nodemask, tmp, cpu_online_map); + cpus_and(nodemask, tmp, cpu_possible_map); if (cpus_empty(nodemask)) continue; @@ -3278,7 +3273,7 @@ static void __init arch_init_sched_domai last_node->next = first_node; mb(); - for_each_cpu_mask(i, cpu_online_map) { + for_each_cpu(i) { struct sched_domain *node_domain = &per_cpu(node_domains, i); struct sched_domain *cpu_domain = cpu_sched_domain(i); node_domain->groups = &sched_group_nodes[cpu_to_node(i)]; @@ -3293,15 +3288,15 @@ static void __init arch_init_sched_domai struct sched_group *first_cpu = NULL, *last_cpu = NULL; /* Set up domains */ - for_each_cpu_mask(i, cpu_online_map) { + for_each_cpu(i) { struct sched_domain *cpu_domain = cpu_sched_domain(i); *cpu_domain = SD_CPU_INIT; - cpu_domain->span = cpu_online_map; + cpu_domain->span = cpu_possible_map; } /* Set up CPU groups */ - for_each_cpu_mask(i, cpu_online_map) { + for_each_cpu_mask(i, cpu_possible_map) { struct sched_group *cpu = &sched_group_cpus[i]; cpus_clear(cpu->cpumask); @@ -3317,7 +3312,7 @@ static void __init arch_init_sched_domai last_cpu->next = first_cpu; mb(); - for_each_cpu_mask(i, cpu_online_map) { + for_each_cpu(i) { struct sched_domain *cpu_domain = cpu_sched_domain(i); cpu_domain->groups = &sched_group_cpus[i]; } _