summaryrefslogtreecommitdiffstats
path: root/arch/x86/kernel/setup_percpu.c
blob: b5c35af2011d0325a2f3b3d331362d20b8dda41f (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
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
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
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
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
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/bootmem.h>
#include <linux/percpu.h>
#include <linux/kexec.h>
#include <linux/crash_dump.h>
#include <linux/smp.h>
#include <linux/topology.h>
#include <asm/sections.h>
#include <asm/processor.h>
#include <asm/setup.h>
#include <asm/mpspec.h>
#include <asm/apicdef.h>
#include <asm/highmem.h>
#include <asm/proto.h>
#include <asm/cpumask.h>

#ifdef CONFIG_DEBUG_PER_CPU_MAPS
# define DBG(x...) printk(KERN_DEBUG x)
#else
# define DBG(x...)
#endif

#ifdef CONFIG_X86_LOCAL_APIC
unsigned int num_processors;
unsigned disabled_cpus __cpuinitdata;
/* Processor that is doing the boot up */
unsigned int boot_cpu_physical_apicid = -1U;
EXPORT_SYMBOL(boot_cpu_physical_apicid);
unsigned int max_physical_apicid;

/* Bitmask of physically existing CPUs */
physid_mask_t phys_cpu_present_map;
#endif

/*
 * Map cpu index to physical APIC ID
 */
DEFINE_EARLY_PER_CPU(u16, x86_cpu_to_apicid, BAD_APICID);
DEFINE_EARLY_PER_CPU(u16, x86_bios_cpu_apicid, BAD_APICID);
EXPORT_EARLY_PER_CPU_SYMBOL(x86_cpu_to_apicid);
EXPORT_EARLY_PER_CPU_SYMBOL(x86_bios_cpu_apicid);

#if defined(CONFIG_NUMA) && defined(CONFIG_X86_64)
#define	X86_64_NUMA	1	/* (used later) */

/*
 * Map cpu index to node index
 */
DEFINE_EARLY_PER_CPU(int, x86_cpu_to_node_map, NUMA_NO_NODE);
EXPORT_EARLY_PER_CPU_SYMBOL(x86_cpu_to_node_map);

/*
 * Which logical CPUs are on which nodes
 */
cpumask_t *node_to_cpumask_map;
EXPORT_SYMBOL(node_to_cpumask_map);

/*
 * Setup node_to_cpumask_map
 */
static void __init setup_node_to_cpumask_map(void);

#else
static inline void setup_node_to_cpumask_map(void) { }
#endif

/*
 * Define load_pda_offset() and per-cpu __pda for x86_64.
 * load_pda_offset() is responsible for loading the offset of pda into
 * %gs.
 *
 * On SMP, pda offset also duals as percpu base address and thus it
 * should be at the start of per-cpu area.  To achieve this, it's
 * preallocated in vmlinux_64.lds.S directly instead of using
 * DEFINE_PER_CPU().
 */
#ifdef CONFIG_X86_64
void __cpuinit load_pda_offset(int cpu)
{
	/* Memory clobbers used to order pda/percpu accesses */
	mb();
	wrmsrl(MSR_GS_BASE, cpu_pda(cpu));
	mb();
}
#ifndef CONFIG_SMP
DEFINE_PER_CPU(struct x8664_pda, __pda);
#endif
EXPORT_PER_CPU_SYMBOL(__pda);
#endif /* CONFIG_SMP && CONFIG_X86_64 */

#ifdef CONFIG_X86_64

/* correctly size the local cpu masks */
static void setup_cpu_local_masks(void)
{
	alloc_bootmem_cpumask_var(&cpu_initialized_mask);
	alloc_bootmem_cpumask_var(&cpu_callin_mask);
	alloc_bootmem_cpumask_var(&cpu_callout_mask);
	alloc_bootmem_cpumask_var(&cpu_sibling_setup_mask);
}

#else /* CONFIG_X86_32 */

static inline void setup_cpu_local_masks(void)
{
}

#endif /* CONFIG_X86_32 */

#ifdef CONFIG_HAVE_SETUP_PER_CPU_AREA
/*
 * Copy data used in early init routines from the initial arrays to the
 * per cpu data areas.  These arrays then become expendable and the
 * *_early_ptr's are zeroed indicating that the static arrays are gone.
 */
static void __init setup_per_cpu_maps(void)
{
	int cpu;

	for_each_possible_cpu(cpu) {
		per_cpu(x86_cpu_to_apicid, cpu) =
				early_per_cpu_map(x86_cpu_to_apicid, cpu);
		per_cpu(x86_bios_cpu_apicid, cpu) =
				early_per_cpu_map(x86_bios_cpu_apicid, cpu);
#ifdef X86_64_NUMA
		per_cpu(x86_cpu_to_node_map, cpu) =
				early_per_cpu_map(x86_cpu_to_node_map, cpu);
#endif
	}

	/* indicate the early static arrays will soon be gone */
	early_per_cpu_ptr(x86_cpu_to_apicid) = NULL;
	early_per_cpu_ptr(x86_bios_cpu_apicid) = NULL;
#ifdef X86_64_NUMA
	early_per_cpu_ptr(x86_cpu_to_node_map) = NULL;
#endif
}

#ifdef CONFIG_X86_64
unsigned long __per_cpu_offset[NR_CPUS] __read_mostly = {
	[0] = (unsigned long)__per_cpu_load,
};
#else
unsigned long __per_cpu_offset[NR_CPUS] __read_mostly;
#endif
EXPORT_SYMBOL(__per_cpu_offset);

/*
 * Great future plan:
 * Declare PDA itself and support (irqstack,tss,pgd) as per cpu data.
 * Always point %gs to its beginning
 */
void __init setup_per_cpu_areas(void)
{
	ssize_t size, old_size;
	char *ptr;
	int cpu;
	unsigned long align = 1;

	/* Copy section for each CPU (we discard the original) */
	old_size = PERCPU_ENOUGH_ROOM;
	align = max_t(unsigned long, PAGE_SIZE, align);
	size = roundup(old_size, align);

	pr_info("NR_CPUS:%d nr_cpumask_bits:%d nr_cpu_ids:%d nr_node_ids:%d\n",
		NR_CPUS, nr_cpumask_bits, nr_cpu_ids, nr_node_ids);

	pr_info("PERCPU: Allocating %zd bytes of per cpu data\n", size);

	for_each_possible_cpu(cpu) {
#ifndef CONFIG_NEED_MULTIPLE_NODES
		ptr = __alloc_bootmem(size, align,
				 __pa(MAX_DMA_ADDRESS));
#else
		int node = early_cpu_to_node(cpu);
		if (!node_online(node) || !NODE_DATA(node)) {
			ptr = __alloc_bootmem(size, align,
					 __pa(MAX_DMA_ADDRESS));
			pr_info("cpu %d has no node %d or node-local memory\n",
				cpu, node);
			pr_debug("per cpu data for cpu%d at %016lx\n",
				 cpu, __pa(ptr));
		} else {
			ptr = __alloc_bootmem_node(NODE_DATA(node), size, align,
							__pa(MAX_DMA_ADDRESS));
			pr_debug("per cpu data for cpu%d on node%d at %016lx\n",
				cpu, node, __pa(ptr));
		}
#endif

		memcpy(ptr, __per_cpu_load, __per_cpu_end - __per_cpu_start);
		per_cpu_offset(cpu) = ptr - __per_cpu_start;
#ifdef CONFIG_X86_64
		/*
		 * CPU0 modified pda in the init data area, reload pda
		 * offset for CPU0 and clear the area for others.
		 */
		if (cpu == 0)
			load_pda_offset(0);
		else
			memset(cpu_pda(cpu), 0, sizeof(*cpu_pda(cpu)));
#endif
		per_cpu(this_cpu_off, cpu) = per_cpu_offset(cpu);

		DBG("PERCPU: cpu %4d %p\n", cpu, ptr);
	}

	/* Setup percpu data maps */
	setup_per_cpu_maps();

	/* Setup node to cpumask map */
	setup_node_to_cpumask_map();

	/* Setup cpu initialized, callin, callout masks */
	setup_cpu_local_masks();
}

#endif

#ifdef X86_64_NUMA

/*
 * Allocate node_to_cpumask_map based on number of available nodes
 * Requires node_possible_map to be valid.
 *
 * Note: node_to_cpumask() is not valid until after this is done.
 * (Use CONFIG_DEBUG_PER_CPU_MAPS to check this.)
 */
static void __init setup_node_to_cpumask_map(void)
{
	unsigned int node, num = 0;
	cpumask_t *map;

	/* setup nr_node_ids if not done yet */
	if (nr_node_ids == MAX_NUMNODES) {
		for_each_node_mask(node, node_possible_map)
			num = node;
		nr_node_ids = num + 1;
	}

	/* allocate the map */
	map = alloc_bootmem_low(nr_node_ids * sizeof(cpumask_t));
	DBG("node_to_cpumask_map at %p for %d nodes\n", map, nr_node_ids);

	pr_debug("Node to cpumask map at %p for %d nodes\n",
		 map, nr_node_ids);

	/* node_to_cpumask() will now work */
	node_to_cpumask_map = map;
}

void __cpuinit numa_set_node(int cpu, int node)
{
	int *cpu_to_node_map = early_per_cpu_ptr(x86_cpu_to_node_map);

	/* early setting, no percpu area yet */
	if (cpu_to_node_map) {
		cpu_to_node_map[cpu] = node;
		return;
	}

#ifdef CONFIG_DEBUG_PER_CPU_MAPS
	if (cpu >= nr_cpu_ids || !per_cpu_offset(cpu)) {
		printk(KERN_ERR "numa_set_node: invalid cpu# (%d)\n", cpu);
		dump_stack();
		return;
	}
#endif
	per_cpu(x86_cpu_to_node_map, cpu) = node;

	if (node != NUMA_NO_NODE)
		cpu_pda(cpu)->nodenumber = node;
}

void __cpuinit numa_clear_node(int cpu)
{
	numa_set_node(cpu, NUMA_NO_NODE);
}

#ifndef CONFIG_DEBUG_PER_CPU_MAPS

void __cpuinit numa_add_cpu(int cpu)
{
	cpu_set(cpu, node_to_cpumask_map[early_cpu_to_node(cpu)]);
}

void __cpuinit numa_remove_cpu(int cpu)
{
	cpu_clear(cpu, node_to_cpumask_map[early_cpu_to_node(cpu)]);
}

#else /* CONFIG_DEBUG_PER_CPU_MAPS */

/*
 * --------- debug versions of the numa functions ---------
 */
static void __cpuinit numa_set_cpumask(int cpu, int enable)
{
	int node = early_cpu_to_node(cpu);
	cpumask_t *mask;
	char buf[64];

	if (node_to_cpumask_map == NULL) {
		printk(KERN_ERR "node_to_cpumask_map NULL\n");
		dump_stack();
		return;
	}

	mask = &node_to_cpumask_map[node];
	if (enable)
		cpu_set(cpu, *mask);
	else
		cpu_clear(cpu, *mask);

	cpulist_scnprintf(buf, sizeof(buf), mask);
	printk(KERN_DEBUG "%s cpu %d node %d: mask now %s\n",
		enable ? "numa_add_cpu" : "numa_remove_cpu", cpu, node, buf);
}

void __cpuinit numa_add_cpu(int cpu)
{
	numa_set_cpumask(cpu, 1);
}

void __cpuinit numa_remove_cpu(int cpu)
{
	numa_set_cpumask(cpu, 0);
}

int cpu_to_node(int cpu)
{
	if (early_per_cpu_ptr(x86_cpu_to_node_map)) {
		printk(KERN_WARNING
			"cpu_to_node(%d): usage too early!\n", cpu);
		dump_stack();
		return early_per_cpu_ptr(x86_cpu_to_node_map)[cpu];
	}
	return per_cpu(x86_cpu_to_node_map, cpu);
}
EXPORT_SYMBOL(cpu_to_node);

/*
 * Same function as cpu_to_node() but used if called before the
 * per_cpu areas are setup.
 */
int early_cpu_to_node(int cpu)
{
	if (early_per_cpu_ptr(x86_cpu_to_node_map))
		return early_per_cpu_ptr(x86_cpu_to_node_map)[cpu];

	if (!per_cpu_offset(cpu)) {
		printk(KERN_WARNING
			"early_cpu_to_node(%d): no per_cpu area!\n", cpu);
		dump_stack();
		return NUMA_NO_NODE;
	}
	return per_cpu(x86_cpu_to_node_map, cpu);
}


/* empty cpumask */
static const cpumask_t cpu_mask_none;

/*
 * Returns a pointer to the bitmask of CPUs on Node 'node'.
 */
const cpumask_t *cpumask_of_node(int node)
{
	if (node_to_cpumask_map == NULL) {
		printk(KERN_WARNING
			"cpumask_of_node(%d): no node_to_cpumask_map!\n",
			node);
		dump_stack();
		return (const cpumask_t *)&cpu_online_map;
	}
	if (node >= nr_node_ids) {
		printk(KERN_WARNING
			"cpumask_of_node(%d): node > nr_node_ids(%d)\n",
			node, nr_node_ids);
		dump_stack();
		return &cpu_mask_none;
	}
	return &node_to_cpumask_map[node];
}
EXPORT_SYMBOL(cpumask_of_node);

/*
 * Returns a bitmask of CPUs on Node 'node'.
 *
 * Side note: this function creates the returned cpumask on the stack
 * so with a high NR_CPUS count, excessive stack space is used.  The
 * node_to_cpumask_ptr function should be used whenever possible.
 */
cpumask_t node_to_cpumask(int node)
{
	if (node_to_cpumask_map == NULL) {
		printk(KERN_WARNING
			"node_to_cpumask(%d): no node_to_cpumask_map!\n", node);
		dump_stack();
		return cpu_online_map;
	}
	if (node >= nr_node_ids) {
		printk(KERN_WARNING
			"node_to_cpumask(%d): node > nr_node_ids(%d)\n",
			node, nr_node_ids);
		dump_stack();
		return cpu_mask_none;
	}
	return node_to_cpumask_map[node];
}
EXPORT_SYMBOL(node_to_cpumask);

/*
 * --------- end of debug versions of the numa functions ---------
 */

#endif /* CONFIG_DEBUG_PER_CPU_MAPS */

#endif /* X86_64_NUMA */

OpenPOWER on IntegriCloud