topology.c 8.6 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
/*
 * arch/arm/kernel/topology.c
 *
 * Copyright (C) 2011 Linaro Limited.
 * Written by: Vincent Guittot
 *
 * based on arch/sh/kernel/topology.c
 *
 * This file is subject to the terms and conditions of the GNU General Public
 * License.  See the file "COPYING" in the main directory of this archive
 * for more details.
 */

#include <linux/cpu.h>
#include <linux/cpumask.h>
16
#include <linux/export.h>
17 18 19 20
#include <linux/init.h>
#include <linux/percpu.h>
#include <linux/node.h>
#include <linux/nodemask.h>
21
#include <linux/of.h>
22
#include <linux/sched.h>
23
#include <linux/slab.h>
24 25 26 27

#include <asm/cputype.h>
#include <asm/topology.h>

28
/*
29
 * cpu capacity scale management
30 31 32
 */

/*
33
 * cpu capacity table
34 35
 * This per cpu data structure describes the relative capacity of each core.
 * On a heteregenous system, cores don't have the same computation capacity
36 37 38 39 40 41
 * and we reflect that difference in the cpu_capacity field so the scheduler
 * can take this difference into account during load balance. A per cpu
 * structure is preferred because each CPU updates its own cpu_capacity field
 * during the load balance except for idle cores. One idle core is selected
 * to run the rebalance_domains for all idle cores and the cpu_capacity can be
 * updated during this sequence.
42 43 44
 */
static DEFINE_PER_CPU(unsigned long, cpu_scale);

45
unsigned long arch_scale_freq_capacity(struct sched_domain *sd, int cpu)
46 47 48 49
{
	return per_cpu(cpu_scale, cpu);
}

50
static void set_capacity_scale(unsigned int cpu, unsigned long capacity)
51
{
52
	per_cpu(cpu_scale, cpu) = capacity;
53 54
}

55 56 57 58 59 60 61 62 63 64
#ifdef CONFIG_OF
struct cpu_efficiency {
	const char *compatible;
	unsigned long efficiency;
};

/*
 * Table of relative efficiency of each processors
 * The efficiency value must fit in 20bit and the final
 * cpu_scale value must be in the range
65
 *   0 < cpu_scale < 3*SCHED_CAPACITY_SCALE/2
66 67 68
 * in order to return at most 1 when DIV_ROUND_CLOSEST
 * is used to compute the capacity of a CPU.
 * Processors that are not defined in the table,
69
 * use the default SCHED_CAPACITY_SCALE value for cpu_scale.
70
 */
71
static const struct cpu_efficiency table_efficiency[] = {
72 73 74 75 76
	{"arm,cortex-a15", 3891},
	{"arm,cortex-a7",  2048},
	{NULL, },
};

77
static unsigned long *__cpu_capacity;
78
#define cpu_capacity(cpu)	__cpu_capacity[cpu]
79

80
static unsigned long middle_capacity = 1;
81 82 83 84 85

/*
 * Iterate all CPUs' descriptor in DT and compute the efficiency
 * (as per table_efficiency). Also calculate a middle efficiency
 * as close as possible to  (max{eff_i} - min{eff_i}) / 2
86 87 88
 * This is later used to scale the cpu_capacity field such that an
 * 'average' CPU is of middle capacity. Also see the comments near
 * table_efficiency[] and update_cpu_capacity().
89 90 91
 */
static void __init parse_dt_topology(void)
{
92
	const struct cpu_efficiency *cpu_eff;
93
	struct device_node *cn = NULL;
94
	unsigned long min_capacity = ULONG_MAX;
95 96
	unsigned long max_capacity = 0;
	unsigned long capacity = 0;
97
	int cpu = 0;
98

99 100
	__cpu_capacity = kcalloc(nr_cpu_ids, sizeof(*__cpu_capacity),
				 GFP_NOWAIT);
101

102 103
	for_each_possible_cpu(cpu) {
		const u32 *rate;
104 105
		int len;

106 107 108 109 110 111
		/* too early to use cpu->of_node */
		cn = of_get_cpu_node(cpu, NULL);
		if (!cn) {
			pr_err("missing device node for CPU %d\n", cpu);
			continue;
		}
112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136

		for (cpu_eff = table_efficiency; cpu_eff->compatible; cpu_eff++)
			if (of_device_is_compatible(cn, cpu_eff->compatible))
				break;

		if (cpu_eff->compatible == NULL)
			continue;

		rate = of_get_property(cn, "clock-frequency", &len);
		if (!rate || len != 4) {
			pr_err("%s missing clock-frequency property\n",
				cn->full_name);
			continue;
		}

		capacity = ((be32_to_cpup(rate)) >> 20) * cpu_eff->efficiency;

		/* Save min capacity of the system */
		if (capacity < min_capacity)
			min_capacity = capacity;

		/* Save max capacity of the system */
		if (capacity > max_capacity)
			max_capacity = capacity;

137
		cpu_capacity(cpu) = capacity;
138 139 140 141 142 143
	}

	/* If min and max capacities are equals, we bypass the update of the
	 * cpu_scale because all CPUs have the same capacity. Otherwise, we
	 * compute a middle_capacity factor that will ensure that the capacity
	 * of an 'average' CPU of the system will be as close as possible to
144
	 * SCHED_CAPACITY_SCALE, which is the default value, but with the
145 146
	 * constraint explained near table_efficiency[].
	 */
147
	if (4*max_capacity < (3*(max_capacity + min_capacity)))
148
		middle_capacity = (min_capacity + max_capacity)
149
				>> (SCHED_CAPACITY_SHIFT+1);
150 151
	else
		middle_capacity = ((max_capacity / 3)
152
				>> (SCHED_CAPACITY_SHIFT-1)) + 1;
153 154 155 156 157 158 159 160

}

/*
 * Look for a customed capacity of a CPU in the cpu_capacity table during the
 * boot. The update of all CPUs is in O(n^2) for heteregeneous system but the
 * function returns directly for SMP system.
 */
161
static void update_cpu_capacity(unsigned int cpu)
162
{
163
	if (!cpu_capacity(cpu))
164 165
		return;

166
	set_capacity_scale(cpu, cpu_capacity(cpu) / middle_capacity);
167

168 169
	printk(KERN_INFO "CPU%u: update cpu_capacity %lu\n",
		cpu, arch_scale_freq_capacity(NULL, cpu));
170 171 172 173
}

#else
static inline void parse_dt_topology(void) {}
174
static inline void update_cpu_capacity(unsigned int cpuid) {}
175 176
#endif

177
 /*
178 179
 * cpu topology table
 */
180
struct cputopo_arm cpu_topology[NR_CPUS];
181
EXPORT_SYMBOL_GPL(cpu_topology);
182

183
const struct cpumask *cpu_coregroup_mask(int cpu)
184 185 186 187
{
	return &cpu_topology[cpu].core_sibling;
}

188 189 190 191 192 193 194 195 196
/*
 * The current assumption is that we can power gate each core independently.
 * This will be superseded by DT binding once available.
 */
const struct cpumask *cpu_corepower_mask(int cpu)
{
	return &cpu_topology[cpu].thread_sibling;
}

197
static void update_siblings_masks(unsigned int cpuid)
198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222
{
	struct cputopo_arm *cpu_topo, *cpuid_topo = &cpu_topology[cpuid];
	int cpu;

	/* update core and thread sibling masks */
	for_each_possible_cpu(cpu) {
		cpu_topo = &cpu_topology[cpu];

		if (cpuid_topo->socket_id != cpu_topo->socket_id)
			continue;

		cpumask_set_cpu(cpuid, &cpu_topo->core_sibling);
		if (cpu != cpuid)
			cpumask_set_cpu(cpu, &cpuid_topo->core_sibling);

		if (cpuid_topo->core_id != cpu_topo->core_id)
			continue;

		cpumask_set_cpu(cpuid, &cpu_topo->thread_sibling);
		if (cpu != cpuid)
			cpumask_set_cpu(cpu, &cpuid_topo->thread_sibling);
	}
	smp_wmb();
}

223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247
/*
 * store_cpu_topology is called at boot when only one cpu is running
 * and with the mutex cpu_hotplug.lock locked, when several cpus have booted,
 * which prevents simultaneous write access to cpu_topology array
 */
void store_cpu_topology(unsigned int cpuid)
{
	struct cputopo_arm *cpuid_topo = &cpu_topology[cpuid];
	unsigned int mpidr;

	/* If the cpu topology has been already set, just return */
	if (cpuid_topo->core_id != -1)
		return;

	mpidr = read_cpuid_mpidr();

	/* create cpu topology mapping */
	if ((mpidr & MPIDR_SMP_BITMASK) == MPIDR_SMP_VALUE) {
		/*
		 * This is a multiprocessor system
		 * multiprocessor format & multiprocessor mode field are set
		 */

		if (mpidr & MPIDR_MT_BITMASK) {
			/* core performance interdependency */
248 249 250
			cpuid_topo->thread_id = MPIDR_AFFINITY_LEVEL(mpidr, 0);
			cpuid_topo->core_id = MPIDR_AFFINITY_LEVEL(mpidr, 1);
			cpuid_topo->socket_id = MPIDR_AFFINITY_LEVEL(mpidr, 2);
251 252 253
		} else {
			/* largely independent cores */
			cpuid_topo->thread_id = -1;
254 255
			cpuid_topo->core_id = MPIDR_AFFINITY_LEVEL(mpidr, 0);
			cpuid_topo->socket_id = MPIDR_AFFINITY_LEVEL(mpidr, 1);
256 257 258 259 260 261 262 263 264 265 266 267
		}
	} else {
		/*
		 * This is an uniprocessor system
		 * we are in multiprocessor format but uniprocessor system
		 * or in the old uniprocessor format
		 */
		cpuid_topo->thread_id = -1;
		cpuid_topo->core_id = 0;
		cpuid_topo->socket_id = -1;
	}

268
	update_siblings_masks(cpuid);
269

270
	update_cpu_capacity(cpuid);
271

272 273 274 275 276 277
	printk(KERN_INFO "CPU%u: thread %d, cpu %d, socket %d, mpidr %x\n",
		cpuid, cpu_topology[cpuid].thread_id,
		cpu_topology[cpuid].core_id,
		cpu_topology[cpuid].socket_id, mpidr);
}

G
Guenter Roeck 已提交
278
static inline int cpu_corepower_flags(void)
279 280 281 282 283 284 285 286 287 288 289 290 291
{
	return SD_SHARE_PKG_RESOURCES  | SD_SHARE_POWERDOMAIN;
}

static struct sched_domain_topology_level arm_topology[] = {
#ifdef CONFIG_SCHED_MC
	{ cpu_corepower_mask, cpu_corepower_flags, SD_INIT_NAME(GMC) },
	{ cpu_coregroup_mask, cpu_core_flags, SD_INIT_NAME(MC) },
#endif
	{ cpu_cpu_mask, SD_INIT_NAME(DIE) },
	{ NULL, },
};

292 293 294 295
/*
 * init_cpu_topology is called at boot when only one cpu is running
 * which prevent simultaneous write access to cpu_topology array
 */
296
void __init init_cpu_topology(void)
297 298 299
{
	unsigned int cpu;

300
	/* init core mask and capacity */
301 302 303 304 305 306 307 308
	for_each_possible_cpu(cpu) {
		struct cputopo_arm *cpu_topo = &(cpu_topology[cpu]);

		cpu_topo->thread_id = -1;
		cpu_topo->core_id =  -1;
		cpu_topo->socket_id = -1;
		cpumask_clear(&cpu_topo->core_sibling);
		cpumask_clear(&cpu_topo->thread_sibling);
309

310
		set_capacity_scale(cpu, SCHED_CAPACITY_SCALE);
311 312
	}
	smp_wmb();
313 314

	parse_dt_topology();
315 316 317

	/* Set scheduler topology descriptor */
	set_sched_topology(arm_topology);
318
}