*/
/*
* Copyright (c) 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2013, Intel Corporation.
+ *
+ * Copyright (c) 2012, 2016, Intel Corporation.
*/
/*
* This file is part of Lustre, http://www.lustre.org/
* 1 : disable multiple partitions
* >1 : specify number of partitions
*/
-static int cpu_npartitions;
-CFS_MODULE_PARM(cpu_npartitions, "i", int, 0444, "# of CPU partitions");
+static int cpu_npartitions;
+module_param(cpu_npartitions, int, 0444);
+MODULE_PARM_DESC(cpu_npartitions, "# of CPU partitions");
/**
* modparam for setting CPU partitions patterns:
* i.e: "N 0[0,1] 1[2,3]" the first character 'N' means numbers in bracket
* are NUMA node ID, number before bracket is CPU partition ID.
*
+ * i.e: "N", shortcut expression to create CPT from NUMA & CPU topology
+ *
* NB: If user specified cpu_pattern, cpu_npartitions will be ignored
*/
-static char *cpu_pattern = "";
-CFS_MODULE_PARM(cpu_pattern, "s", charp, 0444, "CPU partitions pattern");
-
-struct cfs_cpt_data {
- /* serialize hotplug etc */
- spinlock_t cpt_lock;
- /* reserved for hotplug */
- unsigned long cpt_version;
- /* mutex to protect cpt_cpumask */
- struct semaphore cpt_mutex;
- /* scratch buffer for set/unset_node */
- cpumask_t *cpt_cpumask;
-};
-
-static struct cfs_cpt_data cpt_data;
-
-void
-cfs_cpu_core_siblings(int cpu, cpumask_t *mask)
-{
- /* return cpumask of cores in the same socket */
- cpumask_copy(mask, topology_core_cpumask(cpu));
-}
-EXPORT_SYMBOL(cfs_cpu_core_siblings);
+static char *cpu_pattern = "N";
+module_param(cpu_pattern, charp, 0444);
+MODULE_PARM_DESC(cpu_pattern, "CPU partitions pattern");
-/* return number of cores in the same socket of \a cpu */
-int
-cfs_cpu_core_nsiblings(int cpu)
+void cfs_cpt_table_free(struct cfs_cpt_table *cptab)
{
- int num;
-
- down(&cpt_data.cpt_mutex);
-
- cfs_cpu_core_siblings(cpu, cpt_data.cpt_cpumask);
- num = cpus_weight(*cpt_data.cpt_cpumask);
-
- up(&cpt_data.cpt_mutex);
-
- return num;
-}
-EXPORT_SYMBOL(cfs_cpu_core_nsiblings);
-
-/* return cpumask of HTs in the same core */
-void
-cfs_cpu_ht_siblings(int cpu, cpumask_t *mask)
-{
- cpumask_copy(mask, topology_thread_cpumask(cpu));
-}
-EXPORT_SYMBOL(cfs_cpu_ht_siblings);
-
-/* return number of HTs in the same core of \a cpu */
-int
-cfs_cpu_ht_nsiblings(int cpu)
-{
- int num;
-
- down(&cpt_data.cpt_mutex);
-
- cfs_cpu_ht_siblings(cpu, cpt_data.cpt_cpumask);
- num = cpus_weight(*cpt_data.cpt_cpumask);
-
- up(&cpt_data.cpt_mutex);
-
- return num;
-}
-EXPORT_SYMBOL(cfs_cpu_ht_nsiblings);
-
-void
-cfs_node_to_cpumask(int node, cpumask_t *mask)
-{
- cpumask_copy(mask, cpumask_of_node(node));
-}
-EXPORT_SYMBOL(cfs_node_to_cpumask);
-
-void
-cfs_cpt_table_free(struct cfs_cpt_table *cptab)
-{
- int i;
+ int i;
if (cptab->ctb_cpu2cpt != NULL) {
LIBCFS_FREE(cptab->ctb_cpu2cpt,
- num_possible_cpus() *
- sizeof(cptab->ctb_cpu2cpt[0]));
+ nr_cpu_ids * sizeof(cptab->ctb_cpu2cpt[0]));
+ }
+
+ if (cptab->ctb_node2cpt != NULL) {
+ LIBCFS_FREE(cptab->ctb_node2cpt,
+ nr_node_ids * sizeof(cptab->ctb_node2cpt[0]));
}
for (i = 0; cptab->ctb_parts != NULL && i < cptab->ctb_nparts; i++) {
if (part->cpt_cpumask != NULL)
LIBCFS_FREE(part->cpt_cpumask, cpumask_size());
+
+ if (part->cpt_distance) {
+ LIBCFS_FREE(part->cpt_distance,
+ cptab->ctb_nparts *
+ sizeof(part->cpt_distance[0]));
+ }
}
if (cptab->ctb_parts != NULL) {
}
EXPORT_SYMBOL(cfs_cpt_table_free);
-struct cfs_cpt_table *
-cfs_cpt_table_alloc(unsigned int ncpt)
+struct cfs_cpt_table *cfs_cpt_table_alloc(int ncpt)
{
struct cfs_cpt_table *cptab;
- int i;
+ int i;
LIBCFS_ALLOC(cptab, sizeof(*cptab));
if (cptab == NULL)
goto failed;
LIBCFS_ALLOC(cptab->ctb_cpu2cpt,
- num_possible_cpus() * sizeof(cptab->ctb_cpu2cpt[0]));
+ nr_cpu_ids * sizeof(cptab->ctb_cpu2cpt[0]));
if (cptab->ctb_cpu2cpt == NULL)
goto failed;
memset(cptab->ctb_cpu2cpt, -1,
- num_possible_cpus() * sizeof(cptab->ctb_cpu2cpt[0]));
+ nr_cpu_ids * sizeof(cptab->ctb_cpu2cpt[0]));
+
+ LIBCFS_ALLOC(cptab->ctb_node2cpt,
+ nr_node_ids * sizeof(cptab->ctb_node2cpt[0]));
+ if (cptab->ctb_node2cpt == NULL)
+ goto failed;
+
+ memset(cptab->ctb_node2cpt, -1,
+ nr_node_ids * sizeof(cptab->ctb_node2cpt[0]));
LIBCFS_ALLOC(cptab->ctb_parts, ncpt * sizeof(cptab->ctb_parts[0]));
if (cptab->ctb_parts == NULL)
struct cfs_cpu_partition *part = &cptab->ctb_parts[i];
LIBCFS_ALLOC(part->cpt_cpumask, cpumask_size());
+ if (!part->cpt_cpumask)
+ goto failed;
+
LIBCFS_ALLOC(part->cpt_nodemask, sizeof(*part->cpt_nodemask));
- if (part->cpt_cpumask == NULL || part->cpt_nodemask == NULL)
+ if (!part->cpt_nodemask)
goto failed;
- }
- spin_lock(&cpt_data.cpt_lock);
- /* Reserved for hotplug */
- cptab->ctb_version = cpt_data.cpt_version;
- spin_unlock(&cpt_data.cpt_lock);
+ LIBCFS_ALLOC(part->cpt_distance,
+ cptab->ctb_nparts * sizeof(part->cpt_distance[0]));
+ if (!part->cpt_distance)
+ goto failed;
+ }
return cptab;
- failed:
+failed:
cfs_cpt_table_free(cptab);
return NULL;
}
EXPORT_SYMBOL(cfs_cpt_table_alloc);
-int
-cfs_cpt_table_print(struct cfs_cpt_table *cptab, char *buf, int len)
+int cfs_cpt_table_print(struct cfs_cpt_table *cptab, char *buf, int len)
{
char *tmp = buf;
- int rc = 0;
+ int rc = -EFBIG;
int i;
int j;
for (i = 0; i < cptab->ctb_nparts; i++) {
- if (len > 0) {
- rc = snprintf(tmp, len, "%d\t: ", i);
+ if (len <= 0)
+ goto out;
+
+ rc = snprintf(tmp, len, "%d\t:", i);
+ len -= rc;
+
+ if (len <= 0)
+ goto out;
+
+ tmp += rc;
+ for_each_cpu(j, cptab->ctb_parts[i].cpt_cpumask) {
+ rc = snprintf(tmp, len, " %d", j);
len -= rc;
+ if (len <= 0)
+ goto out;
+ tmp += rc;
}
- if (len <= 0) {
- rc = -EFBIG;
+ *tmp = '\n';
+ tmp++;
+ len--;
+ }
+ rc = 0;
+ out:
+ if (rc < 0)
+ return rc;
+
+ return tmp - buf;
+}
+EXPORT_SYMBOL(cfs_cpt_table_print);
+
+int cfs_cpt_distance_print(struct cfs_cpt_table *cptab, char *buf, int len)
+{
+ char *tmp = buf;
+ int rc = -EFBIG;
+ int i;
+ int j;
+
+ for (i = 0; i < cptab->ctb_nparts; i++) {
+ if (len <= 0)
+ goto out;
+
+ rc = snprintf(tmp, len, "%d\t:", i);
+ len -= rc;
+
+ if (len <= 0)
goto out;
- }
tmp += rc;
- for_each_cpu_mask(j, *cptab->ctb_parts[i].cpt_cpumask) {
- rc = snprintf(tmp, len, "%d ", j);
+ for (j = 0; j < cptab->ctb_nparts; j++) {
+ rc = snprintf(tmp, len, " %d:%d",
+ j, cptab->ctb_parts[i].cpt_distance[j]);
len -= rc;
- if (len <= 0) {
- rc = -EFBIG;
+ if (len <= 0)
goto out;
- }
tmp += rc;
}
tmp++;
len--;
}
-
+ rc = 0;
out:
if (rc < 0)
return rc;
return tmp - buf;
}
-EXPORT_SYMBOL(cfs_cpt_table_print);
+EXPORT_SYMBOL(cfs_cpt_distance_print);
-int
-cfs_cpt_number(struct cfs_cpt_table *cptab)
+int cfs_cpt_number(struct cfs_cpt_table *cptab)
{
return cptab->ctb_nparts;
}
EXPORT_SYMBOL(cfs_cpt_number);
-int
-cfs_cpt_weight(struct cfs_cpt_table *cptab, int cpt)
+int cfs_cpt_weight(struct cfs_cpt_table *cptab, int cpt)
{
LASSERT(cpt == CFS_CPT_ANY || (cpt >= 0 && cpt < cptab->ctb_nparts));
return cpt == CFS_CPT_ANY ?
- cpus_weight(*cptab->ctb_cpumask) :
- cpus_weight(*cptab->ctb_parts[cpt].cpt_cpumask);
+ cpumask_weight(cptab->ctb_cpumask) :
+ cpumask_weight(cptab->ctb_parts[cpt].cpt_cpumask);
}
EXPORT_SYMBOL(cfs_cpt_weight);
-int
-cfs_cpt_online(struct cfs_cpt_table *cptab, int cpt)
+int cfs_cpt_online(struct cfs_cpt_table *cptab, int cpt)
{
LASSERT(cpt == CFS_CPT_ANY || (cpt >= 0 && cpt < cptab->ctb_nparts));
return cpt == CFS_CPT_ANY ?
- any_online_cpu(*cptab->ctb_cpumask) != NR_CPUS :
- any_online_cpu(*cptab->ctb_parts[cpt].cpt_cpumask) != NR_CPUS;
+ cpumask_any_and(cptab->ctb_cpumask,
+ cpu_online_mask) < nr_cpu_ids :
+ cpumask_any_and(cptab->ctb_parts[cpt].cpt_cpumask,
+ cpu_online_mask) < nr_cpu_ids;
}
EXPORT_SYMBOL(cfs_cpt_online);
-cpumask_t *
-cfs_cpt_cpumask(struct cfs_cpt_table *cptab, int cpt)
+cpumask_t *cfs_cpt_cpumask(struct cfs_cpt_table *cptab, int cpt)
{
LASSERT(cpt == CFS_CPT_ANY || (cpt >= 0 && cpt < cptab->ctb_nparts));
}
EXPORT_SYMBOL(cfs_cpt_cpumask);
-nodemask_t *
-cfs_cpt_nodemask(struct cfs_cpt_table *cptab, int cpt)
+nodemask_t *cfs_cpt_nodemask(struct cfs_cpt_table *cptab, int cpt)
{
LASSERT(cpt == CFS_CPT_ANY || (cpt >= 0 && cpt < cptab->ctb_nparts));
}
EXPORT_SYMBOL(cfs_cpt_nodemask);
-int
-cfs_cpt_set_cpu(struct cfs_cpt_table *cptab, int cpt, int cpu)
+unsigned cfs_cpt_distance(struct cfs_cpt_table *cptab, int cpt1, int cpt2)
+{
+ LASSERT(cpt1 == CFS_CPT_ANY || (cpt1 >= 0 && cpt1 < cptab->ctb_nparts));
+ LASSERT(cpt2 == CFS_CPT_ANY || (cpt2 >= 0 && cpt2 < cptab->ctb_nparts));
+
+ if (cpt1 == CFS_CPT_ANY || cpt2 == CFS_CPT_ANY)
+ return cptab->ctb_distance;
+
+ return cptab->ctb_parts[cpt1].cpt_distance[cpt2];
+}
+EXPORT_SYMBOL(cfs_cpt_distance);
+
+/*
+ * Calculate the maximum NUMA distance between all nodes in the
+ * from_mask and all nodes in the to_mask.
+ */
+static unsigned cfs_cpt_distance_calculate(nodemask_t *from_mask,
+ nodemask_t *to_mask)
+{
+ unsigned maximum;
+ unsigned distance;
+ int to;
+ int from;
+
+ maximum = 0;
+ for_each_node_mask(from, *from_mask) {
+ for_each_node_mask(to, *to_mask) {
+ distance = node_distance(from, to);
+ if (maximum < distance)
+ maximum = distance;
+ }
+ }
+ return maximum;
+}
+
+static void cfs_cpt_add_cpu(struct cfs_cpt_table *cptab, int cpt, int cpu)
+{
+ cptab->ctb_cpu2cpt[cpu] = cpt;
+
+ cpumask_set_cpu(cpu, cptab->ctb_cpumask);
+ cpumask_set_cpu(cpu, cptab->ctb_parts[cpt].cpt_cpumask);
+}
+
+static void cfs_cpt_del_cpu(struct cfs_cpt_table *cptab, int cpt, int cpu)
+{
+ cpumask_clear_cpu(cpu, cptab->ctb_parts[cpt].cpt_cpumask);
+ cpumask_clear_cpu(cpu, cptab->ctb_cpumask);
+
+ cptab->ctb_cpu2cpt[cpu] = -1;
+}
+
+static void cfs_cpt_add_node(struct cfs_cpt_table *cptab, int cpt, int node)
+{
+ int cpt2;
+ struct cfs_cpu_partition *part;
+ struct cfs_cpu_partition *part2;
+
+ if (!node_isset(node, *cptab->ctb_nodemask)) {
+ /* first time node is added to the CPT table */
+ node_set(node, *cptab->ctb_nodemask);
+ cptab->ctb_node2cpt[node] = cpt;
+ cptab->ctb_distance = cfs_cpt_distance_calculate(
+ cptab->ctb_nodemask,
+ cptab->ctb_nodemask);
+ }
+
+ part = &cptab->ctb_parts[cpt];
+ if (!node_isset(node, *part->cpt_nodemask)) {
+ /* first time node is added to this CPT */
+ node_set(node, *part->cpt_nodemask);
+ for (cpt2 = 0; cpt2 < cptab->ctb_nparts; cpt2++) {
+ part2 = &cptab->ctb_parts[cpt2];
+ part->cpt_distance[cpt2] = cfs_cpt_distance_calculate(
+ part->cpt_nodemask,
+ part2->cpt_nodemask);
+ part2->cpt_distance[cpt] = cfs_cpt_distance_calculate(
+ part2->cpt_nodemask,
+ part->cpt_nodemask);
+ }
+ }
+}
+
+static void cfs_cpt_del_node(struct cfs_cpt_table *cptab, int cpt, int node)
{
- int node;
+ int cpu;
+ int cpt2;
+ struct cfs_cpu_partition *part;
+ struct cfs_cpu_partition *part2;
+ part = &cptab->ctb_parts[cpt];
+
+ for_each_cpu(cpu, part->cpt_cpumask) {
+ /* this CPT has other CPU belonging to this node? */
+ if (cpu_to_node(cpu) == node)
+ break;
+ }
+
+ if (cpu >= nr_cpu_ids && node_isset(node, *part->cpt_nodemask)) {
+ /* No more CPUs in the node for this CPT. */
+ node_clear(node, *part->cpt_nodemask);
+ for (cpt2 = 0; cpt2 < cptab->ctb_nparts; cpt2++) {
+ part2 = &cptab->ctb_parts[cpt2];
+ if (node_isset(node, *part2->cpt_nodemask))
+ cptab->ctb_node2cpt[node] = cpt2;
+ part->cpt_distance[cpt2] = cfs_cpt_distance_calculate(
+ part->cpt_nodemask,
+ part2->cpt_nodemask);
+ part2->cpt_distance[cpt] = cfs_cpt_distance_calculate(
+ part2->cpt_nodemask,
+ part->cpt_nodemask);
+ }
+ }
+
+ for_each_cpu(cpu, cptab->ctb_cpumask) {
+ /* this CPT-table has other CPUs belonging to this node? */
+ if (cpu_to_node(cpu) == node)
+ break;
+ }
+
+ if (cpu >= nr_cpu_ids && node_isset(node, *cptab->ctb_nodemask)) {
+ /* No more CPUs in the table for this node. */
+ node_clear(node, *cptab->ctb_nodemask);
+ cptab->ctb_node2cpt[node] = -1;
+ cptab->ctb_distance =
+ cfs_cpt_distance_calculate(cptab->ctb_nodemask,
+ cptab->ctb_nodemask);
+ }
+}
+
+int cfs_cpt_set_cpu(struct cfs_cpt_table *cptab, int cpt, int cpu)
+{
LASSERT(cpt >= 0 && cpt < cptab->ctb_nparts);
- if (cpu < 0 || cpu >= NR_CPUS || !cpu_online(cpu)) {
+ if (cpu < 0 || cpu >= nr_cpu_ids || !cpu_online(cpu)) {
CDEBUG(D_INFO, "CPU %d is invalid or it's offline\n", cpu);
return 0;
}
return 0;
}
- cptab->ctb_cpu2cpt[cpu] = cpt;
-
- LASSERT(!cpu_isset(cpu, *cptab->ctb_cpumask));
- LASSERT(!cpu_isset(cpu, *cptab->ctb_parts[cpt].cpt_cpumask));
+ LASSERT(!cpumask_test_cpu(cpu, cptab->ctb_cpumask));
+ LASSERT(!cpumask_test_cpu(cpu, cptab->ctb_parts[cpt].cpt_cpumask));
- cpu_set(cpu, *cptab->ctb_cpumask);
- cpu_set(cpu, *cptab->ctb_parts[cpt].cpt_cpumask);
-
- node = cpu_to_node(cpu);
-
- /* first CPU of @node in this CPT table */
- if (!node_isset(node, *cptab->ctb_nodemask))
- node_set(node, *cptab->ctb_nodemask);
-
- /* first CPU of @node in this partition */
- if (!node_isset(node, *cptab->ctb_parts[cpt].cpt_nodemask))
- node_set(node, *cptab->ctb_parts[cpt].cpt_nodemask);
+ cfs_cpt_add_cpu(cptab, cpt, cpu);
+ cfs_cpt_add_node(cptab, cpt, cpu_to_node(cpu));
return 1;
}
EXPORT_SYMBOL(cfs_cpt_set_cpu);
-void
-cfs_cpt_unset_cpu(struct cfs_cpt_table *cptab, int cpt, int cpu)
+void cfs_cpt_unset_cpu(struct cfs_cpt_table *cptab, int cpt, int cpu)
{
- int node;
- int i;
-
LASSERT(cpt == CFS_CPT_ANY || (cpt >= 0 && cpt < cptab->ctb_nparts));
- if (cpu < 0 || cpu >= NR_CPUS) {
+ if (cpu < 0 || cpu >= nr_cpu_ids) {
CDEBUG(D_INFO, "Invalid CPU id %d\n", cpu);
return;
}
return;
}
- LASSERT(cpu_isset(cpu, *cptab->ctb_parts[cpt].cpt_cpumask));
- LASSERT(cpu_isset(cpu, *cptab->ctb_cpumask));
-
- cpu_clear(cpu, *cptab->ctb_parts[cpt].cpt_cpumask);
- cpu_clear(cpu, *cptab->ctb_cpumask);
- cptab->ctb_cpu2cpt[cpu] = -1;
-
- node = cpu_to_node(cpu);
+ LASSERT(cpumask_test_cpu(cpu, cptab->ctb_parts[cpt].cpt_cpumask));
+ LASSERT(cpumask_test_cpu(cpu, cptab->ctb_cpumask));
- LASSERT(node_isset(node, *cptab->ctb_parts[cpt].cpt_nodemask));
- LASSERT(node_isset(node, *cptab->ctb_nodemask));
-
- for_each_cpu_mask(i, *cptab->ctb_parts[cpt].cpt_cpumask) {
- /* this CPT has other CPU belonging to this node? */
- if (cpu_to_node(i) == node)
- break;
- }
-
- if (i == NR_CPUS)
- node_clear(node, *cptab->ctb_parts[cpt].cpt_nodemask);
-
- for_each_cpu_mask(i, *cptab->ctb_cpumask) {
- /* this CPT-table has other CPU belonging to this node? */
- if (cpu_to_node(i) == node)
- break;
- }
-
- if (i == NR_CPUS)
- node_clear(node, *cptab->ctb_nodemask);
-
- return;
+ cfs_cpt_del_cpu(cptab, cpt, cpu);
+ cfs_cpt_del_node(cptab, cpt, cpu_to_node(cpu));
}
EXPORT_SYMBOL(cfs_cpt_unset_cpu);
-int
-cfs_cpt_set_cpumask(struct cfs_cpt_table *cptab, int cpt, cpumask_t *mask)
+int cfs_cpt_set_cpumask(struct cfs_cpt_table *cptab, int cpt,
+ const cpumask_t *mask)
{
- int i;
+ int cpu;
- if (cpus_weight(*mask) == 0 || any_online_cpu(*mask) == NR_CPUS) {
+ if (cpumask_weight(mask) == 0 ||
+ cpumask_any_and(mask, cpu_online_mask) >= nr_cpu_ids) {
CDEBUG(D_INFO, "No online CPU is found in the CPU mask "
"for CPU partition %d\n", cpt);
return 0;
}
- for_each_cpu_mask(i, *mask) {
- if (!cfs_cpt_set_cpu(cptab, cpt, i))
- return 0;
+ for_each_cpu(cpu, mask) {
+ cfs_cpt_add_cpu(cptab, cpt, cpu);
+ cfs_cpt_add_node(cptab, cpt, cpu_to_node(cpu));
}
return 1;
}
EXPORT_SYMBOL(cfs_cpt_set_cpumask);
-void
-cfs_cpt_unset_cpumask(struct cfs_cpt_table *cptab, int cpt, cpumask_t *mask)
+void cfs_cpt_unset_cpumask(struct cfs_cpt_table *cptab, int cpt,
+ const cpumask_t *mask)
{
- int i;
+ int cpu;
- for_each_cpu_mask(i, *mask)
- cfs_cpt_unset_cpu(cptab, cpt, i);
+ for_each_cpu(cpu, mask)
+ cfs_cpt_unset_cpu(cptab, cpt, cpu);
}
EXPORT_SYMBOL(cfs_cpt_unset_cpumask);
-int
-cfs_cpt_set_node(struct cfs_cpt_table *cptab, int cpt, int node)
+int cfs_cpt_set_node(struct cfs_cpt_table *cptab, int cpt, int node)
{
- cpumask_t *mask;
- int rc;
+ const cpumask_t *mask;
+ int cpu;
- if (node < 0 || node >= MAX_NUMNODES) {
+ if (node < 0 || node >= nr_node_ids) {
CDEBUG(D_INFO,
"Invalid NUMA id %d for CPU partition %d\n", node, cpt);
return 0;
}
- down(&cpt_data.cpt_mutex);
-
- mask = cpt_data.cpt_cpumask;
- cfs_node_to_cpumask(node, mask);
+ mask = cpumask_of_node(node);
- rc = cfs_cpt_set_cpumask(cptab, cpt, mask);
+ for_each_cpu(cpu, mask)
+ cfs_cpt_add_cpu(cptab, cpt, cpu);
- up(&cpt_data.cpt_mutex);
+ cfs_cpt_add_node(cptab, cpt, node);
- return rc;
+ return 1;
}
EXPORT_SYMBOL(cfs_cpt_set_node);
-void
-cfs_cpt_unset_node(struct cfs_cpt_table *cptab, int cpt, int node)
+void cfs_cpt_unset_node(struct cfs_cpt_table *cptab, int cpt, int node)
{
- cpumask_t *mask;
+ const cpumask_t *mask;
+ int cpu;
- if (node < 0 || node >= MAX_NUMNODES) {
+ if (node < 0 || node >= nr_node_ids) {
CDEBUG(D_INFO,
"Invalid NUMA id %d for CPU partition %d\n", node, cpt);
return;
}
- down(&cpt_data.cpt_mutex);
-
- mask = cpt_data.cpt_cpumask;
- cfs_node_to_cpumask(node, mask);
+ mask = cpumask_of_node(node);
- cfs_cpt_unset_cpumask(cptab, cpt, mask);
+ for_each_cpu(cpu, mask)
+ cfs_cpt_del_cpu(cptab, cpt, cpu);
- up(&cpt_data.cpt_mutex);
+ cfs_cpt_del_node(cptab, cpt, node);
}
EXPORT_SYMBOL(cfs_cpt_unset_node);
-int
-cfs_cpt_set_nodemask(struct cfs_cpt_table *cptab, int cpt, nodemask_t *mask)
+int cfs_cpt_set_nodemask(struct cfs_cpt_table *cptab, int cpt,
+ const nodemask_t *mask)
{
int i;
}
EXPORT_SYMBOL(cfs_cpt_set_nodemask);
-void
-cfs_cpt_unset_nodemask(struct cfs_cpt_table *cptab, int cpt, nodemask_t *mask)
+void cfs_cpt_unset_nodemask(struct cfs_cpt_table *cptab, int cpt,
+ const nodemask_t *mask)
{
int i;
}
EXPORT_SYMBOL(cfs_cpt_unset_nodemask);
-void
-cfs_cpt_clear(struct cfs_cpt_table *cptab, int cpt)
-{
- int last;
- int i;
-
- if (cpt == CFS_CPT_ANY) {
- last = cptab->ctb_nparts - 1;
- cpt = 0;
- } else {
- last = cpt;
- }
-
- for (; cpt <= last; cpt++) {
- for_each_cpu_mask(i, *cptab->ctb_parts[cpt].cpt_cpumask)
- cfs_cpt_unset_cpu(cptab, cpt, i);
- }
-}
-EXPORT_SYMBOL(cfs_cpt_clear);
-
-int
-cfs_cpt_spread_node(struct cfs_cpt_table *cptab, int cpt)
+int cfs_cpt_spread_node(struct cfs_cpt_table *cptab, int cpt)
{
nodemask_t *mask;
int weight;
}
EXPORT_SYMBOL(cfs_cpt_spread_node);
-int
-cfs_cpt_current(struct cfs_cpt_table *cptab, int remap)
+int cfs_cpt_current(struct cfs_cpt_table *cptab, int remap)
{
- int cpu = smp_processor_id();
- int cpt = cptab->ctb_cpu2cpt[cpu];
+ int cpu = smp_processor_id();
+ int cpt = cptab->ctb_cpu2cpt[cpu];
if (cpt < 0) {
if (!remap)
}
EXPORT_SYMBOL(cfs_cpt_current);
-int
-cfs_cpt_of_cpu(struct cfs_cpt_table *cptab, int cpu)
+int cfs_cpt_of_cpu(struct cfs_cpt_table *cptab, int cpu)
{
- LASSERT(cpu >= 0 && cpu < NR_CPUS);
+ LASSERT(cpu >= 0 && cpu < nr_cpu_ids);
return cptab->ctb_cpu2cpt[cpu];
}
EXPORT_SYMBOL(cfs_cpt_of_cpu);
-int
-cfs_cpt_bind(struct cfs_cpt_table *cptab, int cpt)
+int cfs_cpt_of_node(struct cfs_cpt_table *cptab, int node)
+{
+ if (node < 0 || node > nr_node_ids)
+ return CFS_CPT_ANY;
+
+ return cptab->ctb_node2cpt[node];
+}
+EXPORT_SYMBOL(cfs_cpt_of_node);
+
+int cfs_cpt_bind(struct cfs_cpt_table *cptab, int cpt)
{
cpumask_t *cpumask;
nodemask_t *nodemask;
nodemask = cptab->ctb_parts[cpt].cpt_nodemask;
}
- if (any_online_cpu(*cpumask) == NR_CPUS) {
+ if (cpumask_any_and(cpumask, cpu_online_mask) >= nr_cpu_ids) {
CERROR("No online CPU found in CPU partition %d, did someone "
"do CPU hotplug on system? You might need to reload "
"Lustre modules to keep system working well.\n", cpt);
}
for_each_online_cpu(i) {
- if (cpu_isset(i, *cpumask))
+ if (cpumask_test_cpu(i, cpumask))
continue;
- rc = set_cpus_allowed_ptr(cfs_current(), cpumask);
+ rc = set_cpus_allowed_ptr(current, cpumask);
set_mems_allowed(*nodemask);
if (rc == 0)
schedule(); /* switch to allowed CPU */
* Choose max to \a number CPUs from \a node and set them in \a cpt.
* We always prefer to choose CPU in the same core/socket.
*/
-static int
-cfs_cpt_choose_ncpus(struct cfs_cpt_table *cptab, int cpt,
- cpumask_t *node, int number)
+static int cfs_cpt_choose_ncpus(struct cfs_cpt_table *cptab, int cpt,
+ cpumask_t *node, int number)
{
- cpumask_t *socket = NULL;
- cpumask_t *core = NULL;
- int rc = 0;
- int cpu;
+ cpumask_t *socket = NULL;
+ cpumask_t *core = NULL;
+ int rc = 0;
+ int cpu;
+ int i;
LASSERT(number > 0);
- if (number >= cpus_weight(*node)) {
- while (!cpus_empty(*node)) {
- cpu = first_cpu(*node);
+ if (number >= cpumask_weight(node)) {
+ while (!cpumask_empty(node)) {
+ cpu = cpumask_first(node);
rc = cfs_cpt_set_cpu(cptab, cpt, cpu);
if (!rc)
return -EINVAL;
- cpu_clear(cpu, *node);
+ cpumask_clear_cpu(cpu, node);
}
return 0;
}
goto out;
}
- while (!cpus_empty(*node)) {
- cpu = first_cpu(*node);
+ while (!cpumask_empty(node)) {
+ cpu = cpumask_first(node);
/* get cpumask for cores in the same socket */
- cfs_cpu_core_siblings(cpu, socket);
- cpus_and(*socket, *socket, *node);
+ cpumask_copy(socket, topology_core_cpumask(cpu));
+ cpumask_and(socket, socket, node);
- LASSERT(!cpus_empty(*socket));
-
- while (!cpus_empty(*socket)) {
- int i;
+ LASSERT(!cpumask_empty(socket));
+ while (!cpumask_empty(socket)) {
/* get cpumask for hts in the same core */
- cfs_cpu_ht_siblings(cpu, core);
- cpus_and(*core, *core, *node);
+ cpumask_copy(core, topology_sibling_cpumask(cpu));
+ cpumask_and(core, core, node);
- LASSERT(!cpus_empty(*core));
+ LASSERT(!cpumask_empty(core));
- for_each_cpu_mask(i, *core) {
- cpu_clear(i, *socket);
- cpu_clear(i, *node);
+ for_each_cpu(i, core) {
+ cpumask_clear_cpu(i, socket);
+ cpumask_clear_cpu(i, node);
rc = cfs_cpt_set_cpu(cptab, cpt, i);
if (!rc) {
if (--number == 0)
goto out;
}
- cpu = first_cpu(*socket);
+ cpu = cpumask_first(socket);
}
}
- out:
+out:
if (socket != NULL)
LIBCFS_FREE(socket, cpumask_size());
if (core != NULL)
return rc;
}
-#define CPT_WEIGHT_MIN 4u
+#define CPT_WEIGHT_MIN 4
-static unsigned int
-cfs_cpt_num_estimate(void)
+static int cfs_cpt_num_estimate(void)
{
- unsigned nnode = num_online_nodes();
- unsigned ncpu = num_online_cpus();
- unsigned ncpt;
+ int nnode = num_online_nodes();
+ int ncpu = num_online_cpus();
+ int ncpt;
if (ncpu <= CPT_WEIGHT_MIN) {
ncpt = 1;
ncpt = nnode;
- out:
+out:
#if (BITS_PER_LONG == 32)
/* config many CPU partitions on 32-bit system could consume
* too much memory */
- ncpt = min(2U, ncpt);
+ ncpt = min(2, ncpt);
#endif
while (ncpu % ncpt != 0)
ncpt--; /* worst case is 1 */
return ncpt;
}
-static struct cfs_cpt_table *
-cfs_cpt_table_create(int ncpt)
+static struct cfs_cpt_table *cfs_cpt_table_create(int ncpt)
{
struct cfs_cpt_table *cptab = NULL;
- cpumask_t *mask = NULL;
- int cpt = 0;
- int num;
- int rc;
- int i;
+ cpumask_t *mask = NULL;
+ int cpt = 0;
+ int num;
+ int rc;
+ int i;
rc = cfs_cpt_num_estimate();
if (ncpt <= 0)
}
for_each_online_node(i) {
- cfs_node_to_cpumask(i, mask);
+ cpumask_copy(mask, cpumask_of_node(i));
- while (!cpus_empty(*mask)) {
+ while (!cpumask_empty(mask)) {
struct cfs_cpu_partition *part;
- int n;
+ int n;
- if (cpt >= ncpt)
- goto failed;
+ /* Each emulated NUMA node has all allowed CPUs in
+ * the mask.
+ * End loop when all partitions have assigned CPUs.
+ */
+ if (cpt == ncpt)
+ break;
part = &cptab->ctb_parts[cpt];
- n = num - cpus_weight(*part->cpt_cpumask);
+ n = num - cpumask_weight(part->cpt_cpumask);
LASSERT(n > 0);
rc = cfs_cpt_choose_ncpus(cptab, cpt, mask, n);
if (rc < 0)
goto failed;
- LASSERT(num >= cpus_weight(*part->cpt_cpumask));
- if (num == cpus_weight(*part->cpt_cpumask))
+ LASSERT(num >= cpumask_weight(part->cpt_cpumask));
+ if (num == cpumask_weight(part->cpt_cpumask))
cpt++;
}
}
if (cpt != ncpt ||
- num != cpus_weight(*cptab->ctb_parts[ncpt - 1].cpt_cpumask)) {
+ num != cpumask_weight(cptab->ctb_parts[ncpt - 1].cpt_cpumask)) {
CERROR("Expect %d(%d) CPU partitions but got %d(%d), "
"CPU hotplug/unplug while setting?\n",
cptab->ctb_nparts, num, cpt,
- cpus_weight(*cptab->ctb_parts[ncpt - 1].cpt_cpumask));
+ cpumask_weight(cptab->ctb_parts[ncpt - 1].cpt_cpumask));
goto failed;
}
cfs_cpt_table_create_pattern(char *pattern)
{
struct cfs_cpt_table *cptab;
- char *str = pattern;
+ char *str;
int node = 0;
+ int ncpt = 0;
int high;
- int ncpt;
+ int cpt;
+ int rc;
int c;
-
- for (ncpt = 0;; ncpt++) { /* quick scan bracket */
- str = strchr(str, '[');
- if (str == NULL)
- break;
- str++;
- }
+ int i;
str = cfs_trimwhite(pattern);
if (*str == 'n' || *str == 'N') {
pattern = str + 1;
- node = 1;
+ if (*pattern != '\0') {
+ node = 1; /* numa pattern */
+
+ } else { /* shortcut to create CPT from NUMA & CPU topology */
+ node = -1;
+ ncpt = num_online_nodes();
+ }
+ }
+
+ if (ncpt == 0) { /* scanning bracket which is mark of partition */
+ for (str = pattern;; str++, ncpt++) {
+ str = strchr(str, '[');
+ if (str == NULL)
+ break;
+ }
}
if (ncpt == 0 ||
return NULL;
}
- high = node ? MAX_NUMNODES - 1 : NR_CPUS - 1;
-
cptab = cfs_cpt_table_alloc(ncpt);
if (cptab == NULL) {
CERROR("Failed to allocate cpu partition table\n");
return NULL;
}
+ if (node < 0) { /* shortcut to create CPT from NUMA & CPU topology */
+ cpt = 0;
+ for_each_online_node(i) {
+ if (cpt >= ncpt) {
+ CERROR("CPU changed while setting CPU "
+ "partition table, %d/%d\n", cpt, ncpt);
+ goto failed;
+ }
+
+ rc = cfs_cpt_set_node(cptab, cpt++, i);
+ if (!rc)
+ goto failed;
+ }
+ return cptab;
+ }
+
+ high = node ? nr_node_ids - 1 : nr_cpu_ids - 1;
+
for (str = cfs_trimwhite(pattern), c = 0;; c++) {
struct cfs_range_expr *range;
struct cfs_expr_list *el;
char *bracket = strchr(str, '[');
- int cpt;
- int rc;
- int i;
int n;
if (bracket == NULL) {
break;
}
- if (sscanf(str, "%u%n", &cpt, &n) < 1) {
+ if (sscanf(str, "%d%n", &cpt, &n) < 1) {
CERROR("Invalid cpu pattern %s\n", str);
goto failed;
}
goto failed;
}
- cfs_list_for_each_entry(range, &el->el_exprs, re_link) {
+ list_for_each_entry(range, &el->el_exprs, re_link) {
for (i = range->re_lo; i <= range->re_hi; i++) {
if ((i - range->re_lo) % range->re_stride != 0)
continue;
}
#ifdef CONFIG_HOTPLUG_CPU
-static int
-cfs_cpu_notify(struct notifier_block *self, unsigned long action, void *hcpu)
+static int cfs_cpu_notify(struct notifier_block *self, unsigned long action,
+ void *hcpu)
{
- unsigned int cpu = (unsigned long)hcpu;
+ int cpu = (unsigned long)hcpu;
+ bool warn;
switch (action) {
case CPU_DEAD:
case CPU_DEAD_FROZEN:
case CPU_ONLINE:
case CPU_ONLINE_FROZEN:
- spin_lock(&cpt_data.cpt_lock);
- cpt_data.cpt_version++;
- spin_unlock(&cpt_data.cpt_lock);
default:
- CWARN("Lustre: can't support CPU hotplug well now, "
- "performance and stability could be impacted"
- "[CPU %u notify: %lx]\n", cpu, action);
+ if (action != CPU_DEAD && action != CPU_DEAD_FROZEN) {
+ CDEBUG(D_INFO, "CPU changed [cpu %u action %lx]\n",
+ cpu, action);
+ break;
+ }
+
+ /* if all HTs in a core are offline, it may break affinity */
+ warn = cpumask_any_and(topology_sibling_cpumask(cpu),
+ cpu_online_mask) >= nr_cpu_ids;
+ CDEBUG(warn ? D_WARNING : D_INFO,
+ "Lustre: can't support CPU plug-out well now, "
+ "performance and stability could be impacted"
+ "[CPU %u action: %lx]\n", cpu, action);
}
return NOTIFY_OK;
#endif
-void
-cfs_cpu_fini(void)
+void cfs_cpu_fini(void)
{
if (cfs_cpt_table != NULL)
cfs_cpt_table_free(cfs_cpt_table);
#ifdef CONFIG_HOTPLUG_CPU
unregister_hotcpu_notifier(&cfs_cpu_notifier);
#endif
- if (cpt_data.cpt_cpumask != NULL)
- LIBCFS_FREE(cpt_data.cpt_cpumask, cpumask_size());
}
-int
-cfs_cpu_init(void)
+int cfs_cpu_init(void)
{
LASSERT(cfs_cpt_table == NULL);
- memset(&cpt_data, 0, sizeof(cpt_data));
-
- LIBCFS_ALLOC(cpt_data.cpt_cpumask, cpumask_size());
- if (cpt_data.cpt_cpumask == NULL) {
- CERROR("Failed to allocate scratch buffer\n");
- return -1;
- }
-
- spin_lock_init(&cpt_data.cpt_lock);
- sema_init(&cpt_data.cpt_mutex, 1);
-
#ifdef CONFIG_HOTPLUG_CPU
register_hotcpu_notifier(&cfs_cpu_notifier);
#endif
-
+ get_online_cpus();
if (*cpu_pattern != 0) {
- cfs_cpt_table = cfs_cpt_table_create_pattern(cpu_pattern);
+ char *cpu_pattern_dup = kstrdup(cpu_pattern, GFP_KERNEL);
+
+ if (cpu_pattern_dup == NULL) {
+ CERROR("Failed to duplicate cpu_pattern\n");
+ goto failed;
+ }
+
+ cfs_cpt_table = cfs_cpt_table_create_pattern(cpu_pattern_dup);
+ kfree(cpu_pattern_dup);
if (cfs_cpt_table == NULL) {
CERROR("Failed to create cptab from pattern %s\n",
cpu_pattern);
goto failed;
}
}
+ put_online_cpus();
- spin_lock(&cpt_data.cpt_lock);
- if (cfs_cpt_table->ctb_version != cpt_data.cpt_version) {
- spin_unlock(&cpt_data.cpt_lock);
- CERROR("CPU hotplug/unplug during setup\n");
- goto failed;
- }
- spin_unlock(&cpt_data.cpt_lock);
-
- LCONSOLE(0, "HW CPU cores: %d, npartitions: %d\n",
- num_online_cpus(), cfs_cpt_number(cfs_cpt_table));
+ LCONSOLE(0, "HW nodes: %d, HW CPU cores: %d, npartitions: %d\n",
+ num_online_nodes(), num_online_cpus(),
+ cfs_cpt_number(cfs_cpt_table));
return 0;
- failed:
+failed:
+ put_online_cpus();
cfs_cpu_fini();
return -1;
}