diff options
Diffstat (limited to 'include/linux/topology.h')
| -rw-r--r-- | include/linux/topology.h | 287 |
1 files changed, 168 insertions, 119 deletions
diff --git a/include/linux/topology.h b/include/linux/topology.h index d3cf0d6e7712..6575af39fd10 100644 --- a/include/linux/topology.h +++ b/include/linux/topology.h @@ -27,30 +27,25 @@ #ifndef _LINUX_TOPOLOGY_H #define _LINUX_TOPOLOGY_H +#include <linux/arch_topology.h> #include <linux/cpumask.h> +#include <linux/nodemask.h> #include <linux/bitops.h> #include <linux/mmzone.h> #include <linux/smp.h> #include <linux/percpu.h> #include <asm/topology.h> -#ifndef node_has_online_mem -#define node_has_online_mem(nid) (1) -#endif - #ifndef nr_cpus_node #define nr_cpus_node(node) cpumask_weight(cpumask_of_node(node)) #endif -#define for_each_node_with_cpus(node) \ - for_each_online_node(node) \ - if (nr_cpus_node(node)) - int arch_update_cpu_topology(void); /* Conform to ACPI 2.0 SLIT distance definitions */ #define LOCAL_DISTANCE 10 #define REMOTE_DISTANCE 20 +#define DISTANCE_BITS 8 #ifndef node_distance #define node_distance(from,to) ((from) == (to) ? LOCAL_DISTANCE : REMOTE_DISTANCE) #endif @@ -58,122 +53,28 @@ int arch_update_cpu_topology(void); /* * If the distance between nodes in a system is larger than RECLAIM_DISTANCE * (in whatever arch specific measurement units returned by node_distance()) - * then switch on zone reclaim on boot. + * and node_reclaim_mode is enabled then the VM will only call node_reclaim() + * on nodes within this distance. */ #define RECLAIM_DISTANCE 30 #endif -#ifndef PENALTY_FOR_NODE_WITH_CPUS -#define PENALTY_FOR_NODE_WITH_CPUS (1) -#endif /* - * Below are the 3 major initializers used in building sched_domains: - * SD_SIBLING_INIT, for SMT domains - * SD_CPU_INIT, for SMP domains + * The following tunable allows platforms to override the default node + * reclaim distance (RECLAIM_DISTANCE) if remote memory accesses are + * sufficiently fast that the default value actually hurts + * performance. * - * Any architecture that cares to do any tuning to these values should do so - * by defining their own arch-specific initializer in include/asm/topology.h. - * A definition there will automagically override these default initializers - * and allow arch-specific performance tuning of sched_domains. - * (Only non-zero and non-null fields need be specified.) + * AMD EPYC machines use this because even though the 2-hop distance + * is 32 (3.2x slower than a local memory access) performance actually + * *improves* if allowed to reclaim memory and load balance tasks + * between NUMA nodes 2-hops apart. */ +extern int __read_mostly node_reclaim_distance; -#ifdef CONFIG_SCHED_SMT -/* MCD - Do we really need this? It is always on if CONFIG_SCHED_SMT is, - * so can't we drop this in favor of CONFIG_SCHED_SMT? - */ -#define ARCH_HAS_SCHED_WAKE_IDLE -/* Common values for SMT siblings */ -#ifndef SD_SIBLING_INIT -#define SD_SIBLING_INIT (struct sched_domain) { \ - .min_interval = 1, \ - .max_interval = 2, \ - .busy_factor = 64, \ - .imbalance_pct = 110, \ - \ - .flags = 1*SD_LOAD_BALANCE \ - | 1*SD_BALANCE_NEWIDLE \ - | 1*SD_BALANCE_EXEC \ - | 1*SD_BALANCE_FORK \ - | 0*SD_BALANCE_WAKE \ - | 1*SD_WAKE_AFFINE \ - | 1*SD_SHARE_CPUPOWER \ - | 1*SD_SHARE_PKG_RESOURCES \ - | 0*SD_SERIALIZE \ - | 0*SD_PREFER_SIBLING \ - | arch_sd_sibling_asym_packing() \ - , \ - .last_balance = jiffies, \ - .balance_interval = 1, \ - .smt_gain = 1178, /* 15% */ \ -} -#endif -#endif /* CONFIG_SCHED_SMT */ - -#ifdef CONFIG_SCHED_MC -/* Common values for MC siblings. for now mostly derived from SD_CPU_INIT */ -#ifndef SD_MC_INIT -#define SD_MC_INIT (struct sched_domain) { \ - .min_interval = 1, \ - .max_interval = 4, \ - .busy_factor = 64, \ - .imbalance_pct = 125, \ - .cache_nice_tries = 1, \ - .busy_idx = 2, \ - .wake_idx = 0, \ - .forkexec_idx = 0, \ - \ - .flags = 1*SD_LOAD_BALANCE \ - | 1*SD_BALANCE_NEWIDLE \ - | 1*SD_BALANCE_EXEC \ - | 1*SD_BALANCE_FORK \ - | 0*SD_BALANCE_WAKE \ - | 1*SD_WAKE_AFFINE \ - | 0*SD_SHARE_CPUPOWER \ - | 1*SD_SHARE_PKG_RESOURCES \ - | 0*SD_SERIALIZE \ - , \ - .last_balance = jiffies, \ - .balance_interval = 1, \ -} -#endif -#endif /* CONFIG_SCHED_MC */ - -/* Common values for CPUs */ -#ifndef SD_CPU_INIT -#define SD_CPU_INIT (struct sched_domain) { \ - .min_interval = 1, \ - .max_interval = 4, \ - .busy_factor = 64, \ - .imbalance_pct = 125, \ - .cache_nice_tries = 1, \ - .busy_idx = 2, \ - .idle_idx = 1, \ - .newidle_idx = 0, \ - .wake_idx = 0, \ - .forkexec_idx = 0, \ - \ - .flags = 1*SD_LOAD_BALANCE \ - | 1*SD_BALANCE_NEWIDLE \ - | 1*SD_BALANCE_EXEC \ - | 1*SD_BALANCE_FORK \ - | 0*SD_BALANCE_WAKE \ - | 1*SD_WAKE_AFFINE \ - | 0*SD_SHARE_CPUPOWER \ - | 0*SD_SHARE_PKG_RESOURCES \ - | 0*SD_SERIALIZE \ - | 1*SD_PREFER_SIBLING \ - , \ - .last_balance = jiffies, \ - .balance_interval = 1, \ -} -#endif - -#ifdef CONFIG_SCHED_BOOK -#ifndef SD_BOOK_INIT -#error Please define an appropriate SD_BOOK_INIT in include/asm/topology.h!!! +#ifndef PENALTY_FOR_NODE_WITH_CPUS +#define PENALTY_FOR_NODE_WITH_CPUS (1) #endif -#endif /* CONFIG_SCHED_BOOK */ #ifdef CONFIG_USE_PERCPU_NUMA_NODE_ID DECLARE_PER_CPU(int, numa_node); @@ -182,7 +83,7 @@ DECLARE_PER_CPU(int, numa_node); /* Returns the number of the current Node. */ static inline int numa_node_id(void) { - return __this_cpu_read(numa_node); + return raw_cpu_read(numa_node); } #endif @@ -239,7 +140,7 @@ static inline void set_numa_mem(int node) /* Returns the number of the nearest Node with memory */ static inline int numa_mem_id(void) { - return __this_cpu_read(_numa_mem_); + return raw_cpu_read(_numa_mem_); } #endif @@ -276,17 +177,165 @@ static inline int cpu_to_mem(int cpu) #endif /* [!]CONFIG_HAVE_MEMORYLESS_NODES */ +#if defined(topology_die_id) && defined(topology_die_cpumask) +#define TOPOLOGY_DIE_SYSFS +#endif +#if defined(topology_cluster_id) && defined(topology_cluster_cpumask) +#define TOPOLOGY_CLUSTER_SYSFS +#endif +#if defined(topology_book_id) && defined(topology_book_cpumask) +#define TOPOLOGY_BOOK_SYSFS +#endif +#if defined(topology_drawer_id) && defined(topology_drawer_cpumask) +#define TOPOLOGY_DRAWER_SYSFS +#endif + #ifndef topology_physical_package_id #define topology_physical_package_id(cpu) ((void)(cpu), -1) #endif +#ifndef topology_die_id +#define topology_die_id(cpu) ((void)(cpu), -1) +#endif +#ifndef topology_cluster_id +#define topology_cluster_id(cpu) ((void)(cpu), -1) +#endif #ifndef topology_core_id #define topology_core_id(cpu) ((void)(cpu), 0) #endif -#ifndef topology_thread_cpumask -#define topology_thread_cpumask(cpu) cpumask_of(cpu) +#ifndef topology_book_id +#define topology_book_id(cpu) ((void)(cpu), -1) +#endif +#ifndef topology_drawer_id +#define topology_drawer_id(cpu) ((void)(cpu), -1) +#endif +#ifndef topology_ppin +#define topology_ppin(cpu) ((void)(cpu), 0ull) +#endif +#ifndef topology_sibling_cpumask +#define topology_sibling_cpumask(cpu) cpumask_of(cpu) #endif #ifndef topology_core_cpumask #define topology_core_cpumask(cpu) cpumask_of(cpu) #endif +#ifndef topology_cluster_cpumask +#define topology_cluster_cpumask(cpu) cpumask_of(cpu) +#endif +#ifndef topology_die_cpumask +#define topology_die_cpumask(cpu) cpumask_of(cpu) +#endif +#ifndef topology_book_cpumask +#define topology_book_cpumask(cpu) cpumask_of(cpu) +#endif +#ifndef topology_drawer_cpumask +#define topology_drawer_cpumask(cpu) cpumask_of(cpu) +#endif + +#if defined(CONFIG_SCHED_SMT) && !defined(cpu_smt_mask) +static inline const struct cpumask *cpu_smt_mask(int cpu) +{ + return topology_sibling_cpumask(cpu); +} +#endif + +#ifndef topology_is_primary_thread + +static inline bool topology_is_primary_thread(unsigned int cpu) +{ + /* + * When disabling SMT, the primary thread of the SMT will remain + * enabled/active. Architectures that have a special primary thread + * (e.g. x86) need to override this function. Otherwise the first + * thread in the SMT can be made the primary thread. + * + * The sibling cpumask of an offline CPU always contains the CPU + * itself on architectures using the implementation of + * CONFIG_GENERIC_ARCH_TOPOLOGY for building their topology. + * Other architectures not using CONFIG_GENERIC_ARCH_TOPOLOGY for + * building their topology have to check whether to use this default + * implementation or to override it. + */ + return cpu == cpumask_first(topology_sibling_cpumask(cpu)); +} +#define topology_is_primary_thread topology_is_primary_thread + +#endif + +static inline const struct cpumask *cpu_node_mask(int cpu) +{ + return cpumask_of_node(cpu_to_node(cpu)); +} + +#ifdef CONFIG_NUMA +int sched_numa_find_nth_cpu(const struct cpumask *cpus, int cpu, int node); +extern const struct cpumask *sched_numa_hop_mask(unsigned int node, unsigned int hops); +#else +static __always_inline int sched_numa_find_nth_cpu(const struct cpumask *cpus, int cpu, int node) +{ + return cpumask_nth_and(cpu, cpus, cpu_online_mask); +} + +static inline const struct cpumask * +sched_numa_hop_mask(unsigned int node, unsigned int hops) +{ + return ERR_PTR(-EOPNOTSUPP); +} +#endif /* CONFIG_NUMA */ + +/** + * for_each_node_numadist() - iterate over nodes in increasing distance + * order, starting from a given node + * @node: the iteration variable and the starting node. + * @unvisited: a nodemask to keep track of the unvisited nodes. + * + * This macro iterates over NUMA node IDs in increasing distance from the + * starting @node and yields MAX_NUMNODES when all the nodes have been + * visited. + * + * Note that by the time the loop completes, the @unvisited nodemask will + * be fully cleared, unless the loop exits early. + * + * The difference between for_each_node() and for_each_node_numadist() is + * that the former allows to iterate over nodes in numerical order, whereas + * the latter iterates over nodes in increasing order of distance. + * + * This complexity of this iterator is O(N^2), where N represents the + * number of nodes, as each iteration involves scanning all nodes to + * find the one with the shortest distance. + * + * Requires rcu_lock to be held. + */ +#define for_each_node_numadist(node, unvisited) \ + for (int __start = (node), \ + (node) = nearest_node_nodemask((__start), &(unvisited)); \ + (node) < MAX_NUMNODES; \ + node_clear((node), (unvisited)), \ + (node) = nearest_node_nodemask((__start), &(unvisited))) + +/** + * for_each_numa_hop_mask - iterate over cpumasks of increasing NUMA distance + * from a given node. + * @mask: the iteration variable. + * @node: the NUMA node to start the search from. + * + * Requires rcu_lock to be held. + * + * Yields cpu_online_mask for @node == NUMA_NO_NODE. + */ +#define for_each_numa_hop_mask(mask, node) \ + for (unsigned int __hops = 0; \ + mask = (node != NUMA_NO_NODE || __hops) ? \ + sched_numa_hop_mask(node, __hops) : \ + cpu_online_mask, \ + !IS_ERR_OR_NULL(mask); \ + __hops++) + +DECLARE_PER_CPU(unsigned long, cpu_scale); + +static inline unsigned long topology_get_cpu_scale(int cpu) +{ + return per_cpu(cpu_scale, cpu); +} + +void topology_set_cpu_scale(unsigned int cpu, unsigned long capacity); #endif /* _LINUX_TOPOLOGY_H */ |