vmalloc.c 60.8 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1
2
3
4
5
6
7
/*
 *  linux/mm/vmalloc.c
 *
 *  Copyright (C) 1993  Linus Torvalds
 *  Support of BIGMEM added by Gerhard Wichert, Siemens AG, July 1999
 *  SMP-safe vmalloc/vfree/ioremap, Tigran Aivazian <tigran@veritas.com>, May 2000
 *  Major rework to support vmap/vunmap, Christoph Hellwig, SGI, August 2002
Christoph Lameter's avatar
Christoph Lameter committed
8
 *  Numa awareness, Christoph Lameter, SGI, June 2005
Linus Torvalds's avatar
Linus Torvalds committed
9
10
 */

Nick Piggin's avatar
Nick Piggin committed
11
#include <linux/vmalloc.h>
Linus Torvalds's avatar
Linus Torvalds committed
12
13
14
#include <linux/mm.h>
#include <linux/module.h>
#include <linux/highmem.h>
15
#include <linux/sched.h>
Linus Torvalds's avatar
Linus Torvalds committed
16
17
18
#include <linux/slab.h>
#include <linux/spinlock.h>
#include <linux/interrupt.h>
19
#include <linux/proc_fs.h>
20
#include <linux/seq_file.h>
21
#include <linux/debugobjects.h>
22
#include <linux/kallsyms.h>
Nick Piggin's avatar
Nick Piggin committed
23
24
25
26
#include <linux/list.h>
#include <linux/rbtree.h>
#include <linux/radix-tree.h>
#include <linux/rcupdate.h>
27
#include <linux/pfn.h>
28
#include <linux/kmemleak.h>
Nick Piggin's avatar
Nick Piggin committed
29
#include <asm/atomic.h>
Linus Torvalds's avatar
Linus Torvalds committed
30
31
#include <asm/uaccess.h>
#include <asm/tlbflush.h>
32
#include <asm/shmparam.h>
Linus Torvalds's avatar
Linus Torvalds committed
33
34


Nick Piggin's avatar
Nick Piggin committed
35
/*** Page table manipulation functions ***/
36

Linus Torvalds's avatar
Linus Torvalds committed
37
38
39
40
41
42
43
44
45
46
47
static void vunmap_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end)
{
	pte_t *pte;

	pte = pte_offset_kernel(pmd, addr);
	do {
		pte_t ptent = ptep_get_and_clear(&init_mm, addr, pte);
		WARN_ON(!pte_none(ptent) && !pte_present(ptent));
	} while (pte++, addr += PAGE_SIZE, addr != end);
}

Nick Piggin's avatar
Nick Piggin committed
48
static void vunmap_pmd_range(pud_t *pud, unsigned long addr, unsigned long end)
Linus Torvalds's avatar
Linus Torvalds committed
49
50
51
52
53
54
55
56
57
58
59
60
61
{
	pmd_t *pmd;
	unsigned long next;

	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
		if (pmd_none_or_clear_bad(pmd))
			continue;
		vunmap_pte_range(pmd, addr, next);
	} while (pmd++, addr = next, addr != end);
}

Nick Piggin's avatar
Nick Piggin committed
62
static void vunmap_pud_range(pgd_t *pgd, unsigned long addr, unsigned long end)
Linus Torvalds's avatar
Linus Torvalds committed
63
64
65
66
67
68
69
70
71
72
73
74
75
{
	pud_t *pud;
	unsigned long next;

	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
		if (pud_none_or_clear_bad(pud))
			continue;
		vunmap_pmd_range(pud, addr, next);
	} while (pud++, addr = next, addr != end);
}

Nick Piggin's avatar
Nick Piggin committed
76
static void vunmap_page_range(unsigned long addr, unsigned long end)
Linus Torvalds's avatar
Linus Torvalds committed
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
{
	pgd_t *pgd;
	unsigned long next;

	BUG_ON(addr >= end);
	pgd = pgd_offset_k(addr);
	do {
		next = pgd_addr_end(addr, end);
		if (pgd_none_or_clear_bad(pgd))
			continue;
		vunmap_pud_range(pgd, addr, next);
	} while (pgd++, addr = next, addr != end);
}

static int vmap_pte_range(pmd_t *pmd, unsigned long addr,
Nick Piggin's avatar
Nick Piggin committed
92
		unsigned long end, pgprot_t prot, struct page **pages, int *nr)
Linus Torvalds's avatar
Linus Torvalds committed
93
94
95
{
	pte_t *pte;

Nick Piggin's avatar
Nick Piggin committed
96
97
98
99
100
	/*
	 * nr is a running index into the array which helps higher level
	 * callers keep track of where we're up to.
	 */

101
	pte = pte_alloc_kernel(pmd, addr);
Linus Torvalds's avatar
Linus Torvalds committed
102
103
104
	if (!pte)
		return -ENOMEM;
	do {
Nick Piggin's avatar
Nick Piggin committed
105
106
107
108
109
		struct page *page = pages[*nr];

		if (WARN_ON(!pte_none(*pte)))
			return -EBUSY;
		if (WARN_ON(!page))
Linus Torvalds's avatar
Linus Torvalds committed
110
111
			return -ENOMEM;
		set_pte_at(&init_mm, addr, pte, mk_pte(page, prot));
Nick Piggin's avatar
Nick Piggin committed
112
		(*nr)++;
Linus Torvalds's avatar
Linus Torvalds committed
113
114
115
116
	} while (pte++, addr += PAGE_SIZE, addr != end);
	return 0;
}

Nick Piggin's avatar
Nick Piggin committed
117
118
static int vmap_pmd_range(pud_t *pud, unsigned long addr,
		unsigned long end, pgprot_t prot, struct page **pages, int *nr)
Linus Torvalds's avatar
Linus Torvalds committed
119
120
121
122
123
124
125
126
127
{
	pmd_t *pmd;
	unsigned long next;

	pmd = pmd_alloc(&init_mm, pud, addr);
	if (!pmd)
		return -ENOMEM;
	do {
		next = pmd_addr_end(addr, end);
Nick Piggin's avatar
Nick Piggin committed
128
		if (vmap_pte_range(pmd, addr, next, prot, pages, nr))
Linus Torvalds's avatar
Linus Torvalds committed
129
130
131
132
133
			return -ENOMEM;
	} while (pmd++, addr = next, addr != end);
	return 0;
}

Nick Piggin's avatar
Nick Piggin committed
134
135
static int vmap_pud_range(pgd_t *pgd, unsigned long addr,
		unsigned long end, pgprot_t prot, struct page **pages, int *nr)
Linus Torvalds's avatar
Linus Torvalds committed
136
137
138
139
140
141
142
143
144
{
	pud_t *pud;
	unsigned long next;

	pud = pud_alloc(&init_mm, pgd, addr);
	if (!pud)
		return -ENOMEM;
	do {
		next = pud_addr_end(addr, end);
Nick Piggin's avatar
Nick Piggin committed
145
		if (vmap_pmd_range(pud, addr, next, prot, pages, nr))
Linus Torvalds's avatar
Linus Torvalds committed
146
147
148
149
150
			return -ENOMEM;
	} while (pud++, addr = next, addr != end);
	return 0;
}

Nick Piggin's avatar
Nick Piggin committed
151
152
153
154
155
156
/*
 * Set up page tables in kva (addr, end). The ptes shall have prot "prot", and
 * will have pfns corresponding to the "pages" array.
 *
 * Ie. pte at addr+N*PAGE_SIZE shall point to pfn corresponding to pages[N]
 */
157
158
static int vmap_page_range_noflush(unsigned long start, unsigned long end,
				   pgprot_t prot, struct page **pages)
Linus Torvalds's avatar
Linus Torvalds committed
159
160
161
{
	pgd_t *pgd;
	unsigned long next;
162
	unsigned long addr = start;
Nick Piggin's avatar
Nick Piggin committed
163
164
	int err = 0;
	int nr = 0;
Linus Torvalds's avatar
Linus Torvalds committed
165
166
167
168
169

	BUG_ON(addr >= end);
	pgd = pgd_offset_k(addr);
	do {
		next = pgd_addr_end(addr, end);
Nick Piggin's avatar
Nick Piggin committed
170
		err = vmap_pud_range(pgd, addr, next, prot, pages, &nr);
Linus Torvalds's avatar
Linus Torvalds committed
171
		if (err)
172
			return err;
Linus Torvalds's avatar
Linus Torvalds committed
173
	} while (pgd++, addr = next, addr != end);
Nick Piggin's avatar
Nick Piggin committed
174
175

	return nr;
Linus Torvalds's avatar
Linus Torvalds committed
176
177
}

178
179
180
181
182
183
184
185
186
187
static int vmap_page_range(unsigned long start, unsigned long end,
			   pgprot_t prot, struct page **pages)
{
	int ret;

	ret = vmap_page_range_noflush(start, end, prot, pages);
	flush_cache_vmap(start, end);
	return ret;
}

188
int is_vmalloc_or_module_addr(const void *x)
189
190
{
	/*
191
	 * ARM, x86-64 and sparc64 put modules in a special place,
192
193
194
195
196
197
198
199
200
201
202
	 * and fall back on vmalloc() if that fails. Others
	 * just put it in the vmalloc space.
	 */
#if defined(CONFIG_MODULES) && defined(MODULES_VADDR)
	unsigned long addr = (unsigned long)x;
	if (addr >= MODULES_VADDR && addr < MODULES_END)
		return 1;
#endif
	return is_vmalloc_addr(x);
}

203
/*
Nick Piggin's avatar
Nick Piggin committed
204
 * Walk a vmap address to the struct page it maps.
205
 */
206
struct page *vmalloc_to_page(const void *vmalloc_addr)
207
208
209
210
211
{
	unsigned long addr = (unsigned long) vmalloc_addr;
	struct page *page = NULL;
	pgd_t *pgd = pgd_offset_k(addr);

212
213
214
215
	/*
	 * XXX we might need to change this if we add VIRTUAL_BUG_ON for
	 * architectures that do not vmalloc module space
	 */
216
	VIRTUAL_BUG_ON(!is_vmalloc_or_module_addr(vmalloc_addr));
Jiri Slaby's avatar
Jiri Slaby committed
217

218
	if (!pgd_none(*pgd)) {
Nick Piggin's avatar
Nick Piggin committed
219
		pud_t *pud = pud_offset(pgd, addr);
220
		if (!pud_none(*pud)) {
Nick Piggin's avatar
Nick Piggin committed
221
			pmd_t *pmd = pmd_offset(pud, addr);
222
			if (!pmd_none(*pmd)) {
Nick Piggin's avatar
Nick Piggin committed
223
224
				pte_t *ptep, pte;

225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
				ptep = pte_offset_map(pmd, addr);
				pte = *ptep;
				if (pte_present(pte))
					page = pte_page(pte);
				pte_unmap(ptep);
			}
		}
	}
	return page;
}
EXPORT_SYMBOL(vmalloc_to_page);

/*
 * Map a vmalloc()-space virtual address to the physical page frame number.
 */
240
unsigned long vmalloc_to_pfn(const void *vmalloc_addr)
241
242
243
244
245
{
	return page_to_pfn(vmalloc_to_page(vmalloc_addr));
}
EXPORT_SYMBOL(vmalloc_to_pfn);

Nick Piggin's avatar
Nick Piggin committed
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266

/*** Global kva allocator ***/

#define VM_LAZY_FREE	0x01
#define VM_LAZY_FREEING	0x02
#define VM_VM_AREA	0x04

struct vmap_area {
	unsigned long va_start;
	unsigned long va_end;
	unsigned long flags;
	struct rb_node rb_node;		/* address sorted rbtree */
	struct list_head list;		/* address sorted list */
	struct list_head purge_list;	/* "lazy purge" list */
	void *private;
	struct rcu_head rcu_head;
};

static DEFINE_SPINLOCK(vmap_area_lock);
static struct rb_root vmap_area_root = RB_ROOT;
static LIST_HEAD(vmap_area_list);
267
static unsigned long vmap_area_pcpu_hole;
Nick Piggin's avatar
Nick Piggin committed
268
269

static struct vmap_area *__find_vmap_area(unsigned long addr)
Linus Torvalds's avatar
Linus Torvalds committed
270
{
Nick Piggin's avatar
Nick Piggin committed
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
	struct rb_node *n = vmap_area_root.rb_node;

	while (n) {
		struct vmap_area *va;

		va = rb_entry(n, struct vmap_area, rb_node);
		if (addr < va->va_start)
			n = n->rb_left;
		else if (addr > va->va_start)
			n = n->rb_right;
		else
			return va;
	}

	return NULL;
}

static void __insert_vmap_area(struct vmap_area *va)
{
	struct rb_node **p = &vmap_area_root.rb_node;
	struct rb_node *parent = NULL;
	struct rb_node *tmp;

	while (*p) {
		struct vmap_area *tmp;

		parent = *p;
		tmp = rb_entry(parent, struct vmap_area, rb_node);
		if (va->va_start < tmp->va_end)
			p = &(*p)->rb_left;
		else if (va->va_end > tmp->va_start)
			p = &(*p)->rb_right;
		else
			BUG();
	}

	rb_link_node(&va->rb_node, parent, p);
	rb_insert_color(&va->rb_node, &vmap_area_root);

	/* address-sort this list so it is usable like the vmlist */
	tmp = rb_prev(&va->rb_node);
	if (tmp) {
		struct vmap_area *prev;
		prev = rb_entry(tmp, struct vmap_area, rb_node);
		list_add_rcu(&va->list, &prev->list);
	} else
		list_add_rcu(&va->list, &vmap_area_list);
}

static void purge_vmap_area_lazy(void);

/*
 * Allocate a region of KVA of the specified size and alignment, within the
 * vstart and vend.
 */
static struct vmap_area *alloc_vmap_area(unsigned long size,
				unsigned long align,
				unsigned long vstart, unsigned long vend,
				int node, gfp_t gfp_mask)
{
	struct vmap_area *va;
	struct rb_node *n;
Linus Torvalds's avatar
Linus Torvalds committed
333
	unsigned long addr;
Nick Piggin's avatar
Nick Piggin committed
334
335
	int purged = 0;

Nick Piggin's avatar
Nick Piggin committed
336
	BUG_ON(!size);
Nick Piggin's avatar
Nick Piggin committed
337
338
339
340
341
342
343
344
	BUG_ON(size & ~PAGE_MASK);

	va = kmalloc_node(sizeof(struct vmap_area),
			gfp_mask & GFP_RECLAIM_MASK, node);
	if (unlikely(!va))
		return ERR_PTR(-ENOMEM);

retry:
Glauber Costa's avatar
Glauber Costa committed
345
346
	addr = ALIGN(vstart, align);

Nick Piggin's avatar
Nick Piggin committed
347
	spin_lock(&vmap_area_lock);
Nick Piggin's avatar
Nick Piggin committed
348
349
350
	if (addr + size - 1 < addr)
		goto overflow;

Nick Piggin's avatar
Nick Piggin committed
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
	/* XXX: could have a last_hole cache */
	n = vmap_area_root.rb_node;
	if (n) {
		struct vmap_area *first = NULL;

		do {
			struct vmap_area *tmp;
			tmp = rb_entry(n, struct vmap_area, rb_node);
			if (tmp->va_end >= addr) {
				if (!first && tmp->va_start < addr + size)
					first = tmp;
				n = n->rb_left;
			} else {
				first = tmp;
				n = n->rb_right;
			}
		} while (n);

		if (!first)
			goto found;

		if (first->va_end < addr) {
			n = rb_next(&first->rb_node);
			if (n)
				first = rb_entry(n, struct vmap_area, rb_node);
			else
				goto found;
		}

Nick Piggin's avatar
Nick Piggin committed
380
		while (addr + size > first->va_start && addr + size <= vend) {
Nick Piggin's avatar
Nick Piggin committed
381
			addr = ALIGN(first->va_end + PAGE_SIZE, align);
Nick Piggin's avatar
Nick Piggin committed
382
383
			if (addr + size - 1 < addr)
				goto overflow;
Nick Piggin's avatar
Nick Piggin committed
384
385
386
387
388
389
390
391
392
393

			n = rb_next(&first->rb_node);
			if (n)
				first = rb_entry(n, struct vmap_area, rb_node);
			else
				goto found;
		}
	}
found:
	if (addr + size > vend) {
Nick Piggin's avatar
Nick Piggin committed
394
overflow:
Nick Piggin's avatar
Nick Piggin committed
395
396
397
398
399
400
401
		spin_unlock(&vmap_area_lock);
		if (!purged) {
			purge_vmap_area_lazy();
			purged = 1;
			goto retry;
		}
		if (printk_ratelimit())
402
403
404
			printk(KERN_WARNING
				"vmap allocation for size %lu failed: "
				"use vmalloc=<size> to increase size.\n", size);
405
		kfree(va);
Nick Piggin's avatar
Nick Piggin committed
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
		return ERR_PTR(-EBUSY);
	}

	BUG_ON(addr & (align-1));

	va->va_start = addr;
	va->va_end = addr + size;
	va->flags = 0;
	__insert_vmap_area(va);
	spin_unlock(&vmap_area_lock);

	return va;
}

static void rcu_free_va(struct rcu_head *head)
{
	struct vmap_area *va = container_of(head, struct vmap_area, rcu_head);

	kfree(va);
}

static void __free_vmap_area(struct vmap_area *va)
{
	BUG_ON(RB_EMPTY_NODE(&va->rb_node));
	rb_erase(&va->rb_node, &vmap_area_root);
	RB_CLEAR_NODE(&va->rb_node);
	list_del_rcu(&va->list);

434
435
436
437
438
439
440
441
442
	/*
	 * Track the highest possible candidate for pcpu area
	 * allocation.  Areas outside of vmalloc area can be returned
	 * here too, consider only end addresses which fall inside
	 * vmalloc area proper.
	 */
	if (va->va_end > VMALLOC_START && va->va_end <= VMALLOC_END)
		vmap_area_pcpu_hole = max(vmap_area_pcpu_hole, va->va_end);

Nick Piggin's avatar
Nick Piggin committed
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
	call_rcu(&va->rcu_head, rcu_free_va);
}

/*
 * Free a region of KVA allocated by alloc_vmap_area
 */
static void free_vmap_area(struct vmap_area *va)
{
	spin_lock(&vmap_area_lock);
	__free_vmap_area(va);
	spin_unlock(&vmap_area_lock);
}

/*
 * Clear the pagetable entries of a given vmap_area
 */
static void unmap_vmap_area(struct vmap_area *va)
{
	vunmap_page_range(va->va_start, va->va_end);
}

464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
static void vmap_debug_free_range(unsigned long start, unsigned long end)
{
	/*
	 * Unmap page tables and force a TLB flush immediately if
	 * CONFIG_DEBUG_PAGEALLOC is set. This catches use after free
	 * bugs similarly to those in linear kernel virtual address
	 * space after a page has been freed.
	 *
	 * All the lazy freeing logic is still retained, in order to
	 * minimise intrusiveness of this debugging feature.
	 *
	 * This is going to be *slow* (linear kernel virtual address
	 * debugging doesn't do a broadcast TLB flush so it is a lot
	 * faster).
	 */
#ifdef CONFIG_DEBUG_PAGEALLOC
	vunmap_page_range(start, end);
	flush_tlb_kernel_range(start, end);
#endif
}

Nick Piggin's avatar
Nick Piggin committed
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
/*
 * lazy_max_pages is the maximum amount of virtual address space we gather up
 * before attempting to purge with a TLB flush.
 *
 * There is a tradeoff here: a larger number will cover more kernel page tables
 * and take slightly longer to purge, but it will linearly reduce the number of
 * global TLB flushes that must be performed. It would seem natural to scale
 * this number up linearly with the number of CPUs (because vmapping activity
 * could also scale linearly with the number of CPUs), however it is likely
 * that in practice, workloads might be constrained in other ways that mean
 * vmap activity will not scale linearly with CPUs. Also, I want to be
 * conservative and not introduce a big latency on huge systems, so go with
 * a less aggressive log scale. It will still be an improvement over the old
 * code, and it will be simple to change the scale factor if we find that it
 * becomes a problem on bigger systems.
 */
static unsigned long lazy_max_pages(void)
{
	unsigned int log;

	log = fls(num_online_cpus());

	return log * (32UL * 1024 * 1024 / PAGE_SIZE);
}

static atomic_t vmap_lazy_nr = ATOMIC_INIT(0);

512
513
514
/* for per-CPU blocks */
static void purge_fragmented_blocks_allcpus(void);

Nick Piggin's avatar
Nick Piggin committed
515
516
517
518
519
520
521
522
523
524
525
526
527
/*
 * Purges all lazily-freed vmap areas.
 *
 * If sync is 0 then don't purge if there is already a purge in progress.
 * If force_flush is 1, then flush kernel TLBs between *start and *end even
 * if we found no lazy vmap areas to unmap (callers can use this to optimise
 * their own TLB flushing).
 * Returns with *start = min(*start, lowest purged address)
 *              *end = max(*end, highest purged address)
 */
static void __purge_vmap_area_lazy(unsigned long *start, unsigned long *end,
					int sync, int force_flush)
{
528
	static DEFINE_SPINLOCK(purge_lock);
Nick Piggin's avatar
Nick Piggin committed
529
530
	LIST_HEAD(valist);
	struct vmap_area *va;
531
	struct vmap_area *n_va;
Nick Piggin's avatar
Nick Piggin committed
532
533
534
535
536
537
538
539
	int nr = 0;

	/*
	 * If sync is 0 but force_flush is 1, we'll go sync anyway but callers
	 * should not expect such behaviour. This just simplifies locking for
	 * the case that isn't actually used at the moment anyway.
	 */
	if (!sync && !force_flush) {
540
		if (!spin_trylock(&purge_lock))
Nick Piggin's avatar
Nick Piggin committed
541
542
			return;
	} else
543
		spin_lock(&purge_lock);
Nick Piggin's avatar
Nick Piggin committed
544

545
546
547
	if (sync)
		purge_fragmented_blocks_allcpus();

Nick Piggin's avatar
Nick Piggin committed
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
	rcu_read_lock();
	list_for_each_entry_rcu(va, &vmap_area_list, list) {
		if (va->flags & VM_LAZY_FREE) {
			if (va->va_start < *start)
				*start = va->va_start;
			if (va->va_end > *end)
				*end = va->va_end;
			nr += (va->va_end - va->va_start) >> PAGE_SHIFT;
			unmap_vmap_area(va);
			list_add_tail(&va->purge_list, &valist);
			va->flags |= VM_LAZY_FREEING;
			va->flags &= ~VM_LAZY_FREE;
		}
	}
	rcu_read_unlock();

564
	if (nr)
Nick Piggin's avatar
Nick Piggin committed
565
566
567
568
569
570
571
		atomic_sub(nr, &vmap_lazy_nr);

	if (nr || force_flush)
		flush_tlb_kernel_range(*start, *end);

	if (nr) {
		spin_lock(&vmap_area_lock);
572
		list_for_each_entry_safe(va, n_va, &valist, purge_list)
Nick Piggin's avatar
Nick Piggin committed
573
574
575
			__free_vmap_area(va);
		spin_unlock(&vmap_area_lock);
	}
576
	spin_unlock(&purge_lock);
Nick Piggin's avatar
Nick Piggin committed
577
578
}

Nick Piggin's avatar
Nick Piggin committed
579
580
581
582
583
584
585
586
587
588
589
/*
 * Kick off a purge of the outstanding lazy areas. Don't bother if somebody
 * is already purging.
 */
static void try_purge_vmap_area_lazy(void)
{
	unsigned long start = ULONG_MAX, end = 0;

	__purge_vmap_area_lazy(&start, &end, 0, 0);
}

Nick Piggin's avatar
Nick Piggin committed
590
591
592
593
594
595
596
/*
 * Kick off a purge of the outstanding lazy areas.
 */
static void purge_vmap_area_lazy(void)
{
	unsigned long start = ULONG_MAX, end = 0;

Nick Piggin's avatar
Nick Piggin committed
597
	__purge_vmap_area_lazy(&start, &end, 1, 0);
Nick Piggin's avatar
Nick Piggin committed
598
599
600
}

/*
601
602
 * Free and unmap a vmap area, caller ensuring flush_cache_vunmap had been
 * called for the correct range previously.
Nick Piggin's avatar
Nick Piggin committed
603
 */
604
static void free_unmap_vmap_area_noflush(struct vmap_area *va)
Nick Piggin's avatar
Nick Piggin committed
605
606
607
608
{
	va->flags |= VM_LAZY_FREE;
	atomic_add((va->va_end - va->va_start) >> PAGE_SHIFT, &vmap_lazy_nr);
	if (unlikely(atomic_read(&vmap_lazy_nr) > lazy_max_pages()))
Nick Piggin's avatar
Nick Piggin committed
609
		try_purge_vmap_area_lazy();
Nick Piggin's avatar
Nick Piggin committed
610
611
}

612
613
614
615
616
617
618
619
620
/*
 * Free and unmap a vmap area
 */
static void free_unmap_vmap_area(struct vmap_area *va)
{
	flush_cache_vunmap(va->va_start, va->va_end);
	free_unmap_vmap_area_noflush(va);
}

Nick Piggin's avatar
Nick Piggin committed
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
static struct vmap_area *find_vmap_area(unsigned long addr)
{
	struct vmap_area *va;

	spin_lock(&vmap_area_lock);
	va = __find_vmap_area(addr);
	spin_unlock(&vmap_area_lock);

	return va;
}

static void free_unmap_vmap_area_addr(unsigned long addr)
{
	struct vmap_area *va;

	va = find_vmap_area(addr);
	BUG_ON(!va);
	free_unmap_vmap_area(va);
}


/*** Per cpu kva allocator ***/

/*
 * vmap space is limited especially on 32 bit architectures. Ensure there is
 * room for at least 16 percpu vmap blocks per CPU.
 */
/*
 * If we had a constant VMALLOC_START and VMALLOC_END, we'd like to be able
 * to #define VMALLOC_SPACE		(VMALLOC_END-VMALLOC_START). Guess
 * instead (we just need a rough idea)
 */
#if BITS_PER_LONG == 32
#define VMALLOC_SPACE		(128UL*1024*1024)
#else
#define VMALLOC_SPACE		(128UL*1024*1024*1024)
#endif

#define VMALLOC_PAGES		(VMALLOC_SPACE / PAGE_SIZE)
#define VMAP_MAX_ALLOC		BITS_PER_LONG	/* 256K with 4K pages */
#define VMAP_BBMAP_BITS_MAX	1024	/* 4MB with 4K pages */
#define VMAP_BBMAP_BITS_MIN	(VMAP_MAX_ALLOC*2)
#define VMAP_MIN(x, y)		((x) < (y) ? (x) : (y)) /* can't use min() */
#define VMAP_MAX(x, y)		((x) > (y) ? (x) : (y)) /* can't use max() */
#define VMAP_BBMAP_BITS		VMAP_MIN(VMAP_BBMAP_BITS_MAX,		\
					VMAP_MAX(VMAP_BBMAP_BITS_MIN,	\
						VMALLOC_PAGES / NR_CPUS / 16))

#define VMAP_BLOCK_SIZE		(VMAP_BBMAP_BITS * PAGE_SIZE)

671
672
static bool vmap_initialized __read_mostly = false;

Nick Piggin's avatar
Nick Piggin committed
673
674
675
676
677
678
679
680
681
682
683
684
struct vmap_block_queue {
	spinlock_t lock;
	struct list_head free;
};

struct vmap_block {
	spinlock_t lock;
	struct vmap_area *va;
	struct vmap_block_queue *vbq;
	unsigned long free, dirty;
	DECLARE_BITMAP(alloc_map, VMAP_BBMAP_BITS);
	DECLARE_BITMAP(dirty_map, VMAP_BBMAP_BITS);
685
686
	struct list_head free_list;
	struct rcu_head rcu_head;
687
	struct list_head purge;
Nick Piggin's avatar
Nick Piggin committed
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
};

/* Queue of free and dirty vmap blocks, for allocation and flushing purposes */
static DEFINE_PER_CPU(struct vmap_block_queue, vmap_block_queue);

/*
 * Radix tree of vmap blocks, indexed by address, to quickly find a vmap block
 * in the free path. Could get rid of this if we change the API to return a
 * "cookie" from alloc, to be passed to free. But no big deal yet.
 */
static DEFINE_SPINLOCK(vmap_block_tree_lock);
static RADIX_TREE(vmap_block_tree, GFP_ATOMIC);

/*
 * We should probably have a fallback mechanism to allocate virtual memory
 * out of partially filled vmap blocks. However vmap block sizing should be
 * fairly reasonable according to the vmalloc size, so it shouldn't be a
 * big problem.
 */

static unsigned long addr_to_vb_idx(unsigned long addr)
{
	addr -= VMALLOC_START & ~(VMAP_BLOCK_SIZE-1);
	addr /= VMAP_BLOCK_SIZE;
	return addr;
}

static struct vmap_block *new_vmap_block(gfp_t gfp_mask)
{
	struct vmap_block_queue *vbq;
	struct vmap_block *vb;
	struct vmap_area *va;
	unsigned long vb_idx;
	int node, err;

	node = numa_node_id();

	vb = kmalloc_node(sizeof(struct vmap_block),
			gfp_mask & GFP_RECLAIM_MASK, node);
	if (unlikely(!vb))
		return ERR_PTR(-ENOMEM);

	va = alloc_vmap_area(VMAP_BLOCK_SIZE, VMAP_BLOCK_SIZE,
					VMALLOC_START, VMALLOC_END,
					node, gfp_mask);
	if (unlikely(IS_ERR(va))) {
		kfree(vb);
Julia Lawall's avatar
Julia Lawall committed
735
		return ERR_CAST(va);
Nick Piggin's avatar
Nick Piggin committed
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
	}

	err = radix_tree_preload(gfp_mask);
	if (unlikely(err)) {
		kfree(vb);
		free_vmap_area(va);
		return ERR_PTR(err);
	}

	spin_lock_init(&vb->lock);
	vb->va = va;
	vb->free = VMAP_BBMAP_BITS;
	vb->dirty = 0;
	bitmap_zero(vb->alloc_map, VMAP_BBMAP_BITS);
	bitmap_zero(vb->dirty_map, VMAP_BBMAP_BITS);
	INIT_LIST_HEAD(&vb->free_list);

	vb_idx = addr_to_vb_idx(va->va_start);
	spin_lock(&vmap_block_tree_lock);
	err = radix_tree_insert(&vmap_block_tree, vb_idx, vb);
	spin_unlock(&vmap_block_tree_lock);
	BUG_ON(err);
	radix_tree_preload_end();

	vbq = &get_cpu_var(vmap_block_queue);
	vb->vbq = vbq;
	spin_lock(&vbq->lock);
763
	list_add_rcu(&vb->free_list, &vbq->free);
Nick Piggin's avatar
Nick Piggin committed
764
	spin_unlock(&vbq->lock);
765
	put_cpu_var(vmap_block_queue);
Nick Piggin's avatar
Nick Piggin committed
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787

	return vb;
}

static void rcu_free_vb(struct rcu_head *head)
{
	struct vmap_block *vb = container_of(head, struct vmap_block, rcu_head);

	kfree(vb);
}

static void free_vmap_block(struct vmap_block *vb)
{
	struct vmap_block *tmp;
	unsigned long vb_idx;

	vb_idx = addr_to_vb_idx(vb->va->va_start);
	spin_lock(&vmap_block_tree_lock);
	tmp = radix_tree_delete(&vmap_block_tree, vb_idx);
	spin_unlock(&vmap_block_tree_lock);
	BUG_ON(tmp != vb);

788
	free_unmap_vmap_area_noflush(vb->va);
Nick Piggin's avatar
Nick Piggin committed
789
790
791
	call_rcu(&vb->rcu_head, rcu_free_vb);
}

792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
static void purge_fragmented_blocks(int cpu)
{
	LIST_HEAD(purge);
	struct vmap_block *vb;
	struct vmap_block *n_vb;
	struct vmap_block_queue *vbq = &per_cpu(vmap_block_queue, cpu);

	rcu_read_lock();
	list_for_each_entry_rcu(vb, &vbq->free, free_list) {

		if (!(vb->free + vb->dirty == VMAP_BBMAP_BITS && vb->dirty != VMAP_BBMAP_BITS))
			continue;

		spin_lock(&vb->lock);
		if (vb->free + vb->dirty == VMAP_BBMAP_BITS && vb->dirty != VMAP_BBMAP_BITS) {
			vb->free = 0; /* prevent further allocs after releasing lock */
			vb->dirty = VMAP_BBMAP_BITS; /* prevent purging it again */
			bitmap_fill(vb->alloc_map, VMAP_BBMAP_BITS);
			bitmap_fill(vb->dirty_map, VMAP_BBMAP_BITS);
			spin_lock(&vbq->lock);
			list_del_rcu(&vb->free_list);
			spin_unlock(&vbq->lock);
			spin_unlock(&vb->lock);
			list_add_tail(&vb->purge, &purge);
		} else
			spin_unlock(&vb->lock);
	}
	rcu_read_unlock();

	list_for_each_entry_safe(vb, n_vb, &purge, purge) {
		list_del(&vb->purge);
		free_vmap_block(vb);
	}
}

static void purge_fragmented_blocks_thiscpu(void)
{
	purge_fragmented_blocks(smp_processor_id());
}

static void purge_fragmented_blocks_allcpus(void)
{
	int cpu;

	for_each_possible_cpu(cpu)
		purge_fragmented_blocks(cpu);
}

Nick Piggin's avatar
Nick Piggin committed
840
841
842
843
844
845
static void *vb_alloc(unsigned long size, gfp_t gfp_mask)
{
	struct vmap_block_queue *vbq;
	struct vmap_block *vb;
	unsigned long addr = 0;
	unsigned int order;
846
	int purge = 0;
Nick Piggin's avatar
Nick Piggin committed
847
848
849
850
851
852
853
854
855
856
857
858

	BUG_ON(size & ~PAGE_MASK);
	BUG_ON(size > PAGE_SIZE*VMAP_MAX_ALLOC);
	order = get_order(size);

again:
	rcu_read_lock();
	vbq = &get_cpu_var(vmap_block_queue);
	list_for_each_entry_rcu(vb, &vbq->free, free_list) {
		int i;

		spin_lock(&vb->lock);
859
860
861
		if (vb->free < 1UL << order)
			goto next;

Nick Piggin's avatar
Nick Piggin committed
862
863
864
		i = bitmap_find_free_region(vb->alloc_map,
						VMAP_BBMAP_BITS, order);

865
866
867
868
869
		if (i < 0) {
			if (vb->free + vb->dirty == VMAP_BBMAP_BITS) {
				/* fragmented and no outstanding allocations */
				BUG_ON(vb->dirty != VMAP_BBMAP_BITS);
				purge = 1;
Nick Piggin's avatar
Nick Piggin committed
870
			}
871
			goto next;
Nick Piggin's avatar
Nick Piggin committed
872
		}
873
874
875
876
877
878
879
880
881
882
883
884
		addr = vb->va->va_start + (i << PAGE_SHIFT);
		BUG_ON(addr_to_vb_idx(addr) !=
				addr_to_vb_idx(vb->va->va_start));
		vb->free -= 1UL << order;
		if (vb->free == 0) {
			spin_lock(&vbq->lock);
			list_del_rcu(&vb->free_list);
			spin_unlock(&vbq->lock);
		}
		spin_unlock(&vb->lock);
		break;
next:
Nick Piggin's avatar
Nick Piggin committed
885
886
		spin_unlock(&vb->lock);
	}
887
888
889
890

	if (purge)
		purge_fragmented_blocks_thiscpu();

891
	put_cpu_var(vmap_block_queue);
Nick Piggin's avatar
Nick Piggin committed
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
	rcu_read_unlock();

	if (!addr) {
		vb = new_vmap_block(gfp_mask);
		if (IS_ERR(vb))
			return vb;
		goto again;
	}

	return (void *)addr;
}

static void vb_free(const void *addr, unsigned long size)
{
	unsigned long offset;
	unsigned long vb_idx;
	unsigned int order;
	struct vmap_block *vb;

	BUG_ON(size & ~PAGE_MASK);
	BUG_ON(size > PAGE_SIZE*VMAP_MAX_ALLOC);
913
914
915

	flush_cache_vunmap((unsigned long)addr, (unsigned long)addr + size);

Nick Piggin's avatar
Nick Piggin committed
916
917
918
919
920
921
922
923
924
925
926
	order = get_order(size);

	offset = (unsigned long)addr & (VMAP_BLOCK_SIZE - 1);

	vb_idx = addr_to_vb_idx((unsigned long)addr);
	rcu_read_lock();
	vb = radix_tree_lookup(&vmap_block_tree, vb_idx);
	rcu_read_unlock();
	BUG_ON(!vb);

	spin_lock(&vb->lock);
927
	BUG_ON(bitmap_allocate_region(vb->dirty_map, offset >> PAGE_SHIFT, order));
928

Nick Piggin's avatar
Nick Piggin committed
929
930
	vb->dirty += 1UL << order;
	if (vb->dirty == VMAP_BBMAP_BITS) {
931
		BUG_ON(vb->free);
Nick Piggin's avatar
Nick Piggin committed
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
		spin_unlock(&vb->lock);
		free_vmap_block(vb);
	} else
		spin_unlock(&vb->lock);
}

/**
 * vm_unmap_aliases - unmap outstanding lazy aliases in the vmap layer
 *
 * The vmap/vmalloc layer lazily flushes kernel virtual mappings primarily
 * to amortize TLB flushing overheads. What this means is that any page you
 * have now, may, in a former life, have been mapped into kernel virtual
 * address by the vmap layer and so there might be some CPUs with TLB entries
 * still referencing that page (additional to the regular 1:1 kernel mapping).
 *
 * vm_unmap_aliases flushes all such lazy mappings. After it returns, we can
 * be sure that none of the pages we have control over will have any aliases
 * from the vmap layer.
 */
void vm_unmap_aliases(void)
{
	unsigned long start = ULONG_MAX, end = 0;
	int cpu;
	int flush = 0;

957
958
959
	if (unlikely(!vmap_initialized))
		return;

Nick Piggin's avatar
Nick Piggin committed
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
	for_each_possible_cpu(cpu) {
		struct vmap_block_queue *vbq = &per_cpu(vmap_block_queue, cpu);
		struct vmap_block *vb;

		rcu_read_lock();
		list_for_each_entry_rcu(vb, &vbq->free, free_list) {
			int i;

			spin_lock(&vb->lock);
			i = find_first_bit(vb->dirty_map, VMAP_BBMAP_BITS);
			while (i < VMAP_BBMAP_BITS) {
				unsigned long s, e;
				int j;
				j = find_next_zero_bit(vb->dirty_map,
					VMAP_BBMAP_BITS, i);

				s = vb->va->va_start + (i << PAGE_SHIFT);
				e = vb->va->va_start + (j << PAGE_SHIFT);
				vunmap_page_range(s, e);
				flush = 1;

				if (s < start)
					start = s;
				if (e > end)
					end = e;

				i = j;
				i = find_next_bit(vb->dirty_map,
							VMAP_BBMAP_BITS, i);
			}
			spin_unlock(&vb->lock);
		}
		rcu_read_unlock();
	}

	__purge_vmap_area_lazy(&start, &end, 1, flush);
}
EXPORT_SYMBOL_GPL(vm_unmap_aliases);

/**
 * vm_unmap_ram - unmap linear kernel address space set up by vm_map_ram
 * @mem: the pointer returned by vm_map_ram
 * @count: the count passed to that vm_map_ram call (cannot unmap partial)
 */
void vm_unmap_ram(const void *mem, unsigned int count)
{
	unsigned long size = count << PAGE_SHIFT;
	unsigned long addr = (unsigned long)mem;

	BUG_ON(!addr);
	BUG_ON(addr < VMALLOC_START);
	BUG_ON(addr > VMALLOC_END);
	BUG_ON(addr & (PAGE_SIZE-1));

	debug_check_no_locks_freed(mem, size);
1015
	vmap_debug_free_range(addr, addr+size);
Nick Piggin's avatar
Nick Piggin committed
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029

	if (likely(count <= VMAP_MAX_ALLOC))
		vb_free(mem, size);
	else
		free_unmap_vmap_area_addr(addr);
}
EXPORT_SYMBOL(vm_unmap_ram);

/**
 * vm_map_ram - map pages linearly into kernel virtual address (vmalloc space)
 * @pages: an array of pointers to the pages to be mapped
 * @count: number of pages
 * @node: prefer to allocate data structures on this node
 * @prot: memory protection to use. PAGE_KERNEL for regular RAM
1030
1031
 *
 * Returns: a pointer to the address that has been mapped, or %NULL on failure
Nick Piggin's avatar
Nick Piggin committed
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
 */
void *vm_map_ram(struct page **pages, unsigned int count, int node, pgprot_t prot)
{
	unsigned long size = count << PAGE_SHIFT;
	unsigned long addr;
	void *mem;

	if (likely(count <= VMAP_MAX_ALLOC)) {
		mem = vb_alloc(size, GFP_KERNEL);
		if (IS_ERR(mem))
			return NULL;
		addr = (unsigned long)mem;
	} else {
		struct vmap_area *va;
		va = alloc_vmap_area(size, PAGE_SIZE,
				VMALLOC_START, VMALLOC_END, node, GFP_KERNEL);
		if (IS_ERR(va))
			return NULL;

		addr = va->va_start;
		mem = (void *)addr;
	}
	if (vmap_page_range(addr, addr + size, prot, pages) < 0) {
		vm_unmap_ram(mem, count);
		return NULL;
	}
	return mem;
}
EXPORT_SYMBOL(vm_map_ram);

1062
1063
1064
/**
 * vm_area_register_early - register vmap area early during boot
 * @vm: vm_struct to register
1065
 * @align: requested alignment
1066
1067
1068
1069
1070
1071
1072
1073
 *
 * This function is used to register kernel vm area before
 * vmalloc_init() is called.  @vm->size and @vm->flags should contain
 * proper values on entry and other fields should be zero.  On return,
 * vm->addr contains the allocated address.
 *
 * DO NOT USE THIS FUNCTION UNLESS YOU KNOW WHAT YOU'RE DOING.
 */
1074
void __init vm_area_register_early(struct vm_struct *vm, size_t align)
1075
1076
{
	static size_t vm_init_off __initdata;
1077
1078
1079
1080
	unsigned long addr;

	addr = ALIGN(VMALLOC_START + vm_init_off, align);
	vm_init_off = PFN_ALIGN(addr + vm->size) - VMALLOC_START;
1081

1082
	vm->addr = (void *)addr;
1083
1084
1085
1086
1087

	vm->next = vmlist;
	vmlist = vm;
}

Nick Piggin's avatar
Nick Piggin committed
1088
1089
void __init vmalloc_init(void)
{
Ivan Kokshaysky's avatar
Ivan Kokshaysky committed
1090
1091
	struct vmap_area *va;
	struct vm_struct *tmp;
Nick Piggin's avatar
Nick Piggin committed
1092
1093
1094
1095
1096
1097
1098
1099
1100
	int i;

	for_each_possible_cpu(i) {
		struct vmap_block_queue *vbq;

		vbq = &per_cpu(vmap_block_queue, i);
		spin_lock_init(&vbq->lock);
		INIT_LIST_HEAD(&vbq->free);
	}
1101

Ivan Kokshaysky's avatar
Ivan Kokshaysky committed
1102
1103
	/* Import existing vmlist entries. */
	for (tmp = vmlist; tmp; tmp = tmp->next) {
1104
		va = kzalloc(sizeof(struct vmap_area), GFP_NOWAIT);
Ivan Kokshaysky's avatar
Ivan Kokshaysky committed
1105
1106
1107
1108
1109
		va->flags = tmp->flags | VM_VM_AREA;
		va->va_start = (unsigned long)tmp->addr;
		va->va_end = va->va_start + tmp->size;
		__insert_vmap_area(va);
	}
1110
1111
1112

	vmap_area_pcpu_hole = VMALLOC_END;

1113
	vmap_initialized = true;
Nick Piggin's avatar
Nick Piggin committed
1114
1115
}

1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
/**
 * map_kernel_range_noflush - map kernel VM area with the specified pages
 * @addr: start of the VM area to map
 * @size: size of the VM area to map
 * @prot: page protection flags to use
 * @pages: pages to map
 *
 * Map PFN_UP(@size) pages at @addr.  The VM area @addr and @size
 * specify should have been allocated using get_vm_area() and its
 * friends.
 *
 * NOTE:
 * This function does NOT do any cache flushing.  The caller is
 * responsible for calling flush_cache_vmap() on to-be-mapped areas
 * before calling this function.
 *
 * RETURNS:
 * The number of pages mapped on success, -errno on failure.
 */
int map_kernel_range_noflush(unsigned long addr, unsigned long size,
			     pgprot_t prot, struct page **pages)
{
	return vmap_page_range_noflush(addr, addr + size, prot, pages);
}

/**
 * unmap_kernel_range_noflush - unmap kernel VM area
 * @addr: start of the VM area to unmap
 * @size: size of the VM area to unmap
 *
 * Unmap PFN_UP(@size) pages at @addr.  The VM area @addr and @size
 * specify should have been allocated using get_vm_area() and its
 * friends.
 *
 * NOTE:
 * This function does NOT do any cache flushing.  The caller is
 * responsible for calling flush_cache_vunmap() on to-be-mapped areas
 * before calling this function and flush_tlb_kernel_range() after.
 */
void unmap_kernel_range_noflush(unsigned long addr, unsigned long size)
{
	vunmap_page_range(addr, addr + size);
}

/**
 * unmap_kernel_range - unmap kernel VM area and flush cache and TLB
 * @addr: start of the VM area to unmap
 * @size: size of the VM area to unmap
 *
 * Similar to unmap_kernel_range_noflush() but flushes vcache before
 * the unmapping and tlb after.
 */
Nick Piggin's avatar
Nick Piggin committed
1168
1169
1170
void unmap_kernel_range(unsigned long addr, unsigned long size)
{
	unsigned long end = addr + size;
1171
1172

	flush_cache_vunmap(addr, end);
Nick Piggin's avatar
Nick Piggin committed
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
	vunmap_page_range(addr, end);
	flush_tlb_kernel_range(addr, end);
}

int map_vm_area(struct vm_struct *area, pgprot_t prot, struct page ***pages)
{
	unsigned long addr = (unsigned long)area->addr;
	unsigned long end = addr + area->size - PAGE_SIZE;
	int err;

	err = vmap_page_range(addr, end, prot, *pages);
	if (err > 0) {
		*pages += err;
		err = 0;
	}

	return err;
}
EXPORT_SYMBOL_GPL(map_vm_area);

/*** Old vmalloc interfaces ***/
DEFINE_RWLOCK(vmlist_lock);
struct vm_struct *vmlist;

1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
static void insert_vmalloc_vm(struct vm_struct *vm, struct vmap_area *va,
			      unsigned long flags, void *caller)
{
	struct vm_struct *tmp, **p;

	vm->flags = flags;
	vm->addr = (void *)va->va_start;
	vm->size = va->va_end - va->va_start;
	vm->caller = caller;
	va->private = vm;
	va->flags |= VM_VM_AREA;

	write_lock(&vmlist_lock);
	for (p = &vmlist; (tmp = *p) != NULL; p = &tmp->next) {
		if (tmp->addr >= vm->addr)
			break;
	}
	vm->next = *p;
	*p = vm;
	write_unlock(&vmlist_lock);
}

Nick Piggin's avatar
Nick Piggin committed
1219
static struct vm_struct *__get_vm_area_node(unsigned long size,
1220
1221
		unsigned long align, unsigned long flags, unsigned long start,
		unsigned long end, int node, gfp_t gfp_mask, void *caller)
Nick Piggin's avatar
Nick Piggin committed
1222
1223
1224
{
	static struct vmap_area *va;
	struct vm_struct *area;
Linus Torvalds's avatar
Linus Torvalds committed
1225

1226
	BUG_ON(in_interrupt());
Linus Torvalds's avatar
Linus Torvalds committed
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
	if (flags & VM_IOREMAP) {
		int bit = fls(size);

		if (bit > IOREMAP_MAX_ORDER)
			bit = IOREMAP_MAX_ORDER;
		else if (bit < PAGE_SHIFT)
			bit = PAGE_SHIFT;

		align = 1ul << bit;
	}
Nick Piggin's avatar
Nick Piggin committed
1237

Linus Torvalds's avatar
Linus Torvalds committed
1238
	size = PAGE_ALIGN(size);
1239
1240
	if (unlikely(!size))
		return NULL;
Linus Torvalds's avatar
Linus Torvalds committed
1241

1242
	area = kzalloc_node(sizeof(*area), gfp_mask & GFP_RECLAIM_MASK, node);
Linus Torvalds's avatar
Linus Torvalds committed
1243
1244
1245
1246
1247
1248
1249
1250
	if (unlikely(!area))
		return NULL;

	/*
	 * We always allocate a guard page.
	 */
	size += PAGE_SIZE;

Nick Piggin's avatar
Nick Piggin committed
1251
1252
1253
1254
	va = alloc_vmap_area(size, align, start, end, node, gfp_mask);
	if (IS_ERR(va)) {
		kfree(area);
		return NULL;
Linus Torvalds's avatar
Linus Torvalds committed
1255
1256
	}

1257
	insert_vmalloc_vm(area, va, flags, caller);
Linus Torvalds's avatar
Linus Torvalds committed
1258
1259
1260
	return area;
}

Christoph Lameter's avatar
Christoph Lameter committed
1261
1262
1263
struct vm_struct *__get_vm_area(unsigned long size, unsigned long flags,
				unsigned long start, unsigned long end)
{
1264
	return __get_vm_area_node(size, 1, flags, start, end, -1, GFP_KERNEL,
1265
						__builtin_return_address(0));
Christoph Lameter's avatar
Christoph Lameter committed
1266
}
1267
EXPORT_SYMBOL_GPL(__get_vm_area);
Christoph Lameter's avatar
Christoph Lameter committed
1268

1269
1270
1271
1272
struct vm_struct *__get_vm_area_caller(unsigned long size, unsigned long flags,
				       unsigned long start, unsigned long end,
				       void *caller)
{
1273
	return __get_vm_area_node(size, 1, flags, start, end, -1, GFP_KERNEL,
1274
1275
1276
				  caller);
}

Linus Torvalds's avatar
Linus Torvalds committed
1277
/**
Simon Arlott's avatar
Simon Arlott committed
1278
 *	get_vm_area  -  reserve a contiguous kernel virtual area
Linus Torvalds's avatar
Linus Torvalds committed
1279
1280
1281
1282
1283
1284
1285
1286
1287
 *	@size:		size of the area
 *	@flags:		%VM_IOREMAP for I/O mappings or VM_ALLOC
 *
 *	Search an area of @size in the kernel virtual mapping area,
 *	and reserved it for out purposes.  Returns the area descriptor
 *	on success or %NULL on failure.
 */
struct vm_struct *get_vm_area(unsigned long size, unsigned long flags)
{
1288
	return __get_vm_area_node(size, 1, flags, VMALLOC_START, VMALLOC_END,
1289
1290
1291
1292
1293
1294
				-1, GFP_KERNEL, __builtin_return_address(0));
}

struct vm_struct *get_vm_area_caller(unsigned long size, unsigned long flags,
				void *caller)
{
1295
	return __get_vm_area_node(size, 1, flags, VMALLOC_START, VMALLOC_END,
1296
						-1, GFP_KERNEL, caller);
Linus Torvalds's avatar
Linus Torvalds committed
1297
1298
}

1299
1300
struct vm_struct *get_vm_area_node(unsigned long size, unsigned long flags,
				   int node, gfp_t gfp_mask)
Christoph Lameter's avatar
Christoph Lameter committed
1301
{
1302
1303
	return __get_vm_area_node(size, 1, flags, VMALLOC_START, VMALLOC_END,
				  node, gfp_mask, __builtin_return_address(0));
Christoph Lameter's avatar
Christoph Lameter committed
1304
1305
}

Nick Piggin's avatar
Nick Piggin committed
1306
static struct vm_struct *find_vm_area(const void *addr)
1307
{
Nick Piggin's avatar
Nick Piggin committed
1308
	struct vmap_area *va;
1309

Nick Piggin's avatar
Nick Piggin committed
1310
1311
1312
	va = find_vmap_area((unsigned long)addr);
	if (va && va->flags & VM_VM_AREA)
		return va->private;
Linus Torvalds's avatar
Linus Torvalds committed
1313
1314
1315
1316

	return NULL;
}

1317
/**
Simon Arlott's avatar
Simon Arlott committed
1318
 *	remove_vm_area  -  find and remove a continuous kernel virtual area
1319
1320
1321
1322
1323
1324
 *	@addr:		base address
 *
 *	Search for the kernel VM area starting at @addr, and remove it.
 *	This function returns the found VM area, but using it is NOT safe
 *	on SMP machines, except for its size or flags.
 */
1325
struct vm_struct *remove_vm_area(const void *addr)
1326
{
Nick Piggin's avatar
Nick Piggin committed
1327
1328
1329
1330
1331
1332
	struct vmap_area *va;

	va = find_vmap_area((unsigned long)addr);
	if (va && va->flags & VM_VM_AREA) {
		struct vm_struct *vm = va->private;
		struct vm_struct *tmp, **p;
1333
1334
1335
1336
1337
		/*
		 * remove from list and disallow access to this vm_struct
		 * before unmap. (address range confliction is maintained by
		 * vmap.)
		 */
Nick Piggin's avatar
Nick Piggin committed
1338
1339
1340
1341
1342
1343
		write_lock(&vmlist_lock);
		for (p = &vmlist; (tmp = *p) != vm; p = &tmp->next)
			;
		*p = tmp->next;
		write_unlock(&vmlist_lock);

1344
1345
1346
1347
		vmap_debug_free_range(va->va_start, va->va_end);
		free_unmap_vmap_area(va);
		vm->size -= PAGE_SIZE;

Nick Piggin's avatar
Nick Piggin committed
1348
1349
1350
		return vm;
	}
	return NULL;
1351
1352
}

1353
static void __vunmap(const void *addr, int deallocate_pages)
Linus Torvalds's avatar
Linus Torvalds committed
1354
1355
1356
1357
1358
1359
1360
{
	struct vm_struct *area;

	if (!addr)
		return;

	if ((PAGE_SIZE-1) & (unsigned long)addr) {
Arjan van de Ven's avatar
Arjan van de Ven committed
1361
		WARN(1, KERN_ERR "Trying to vfree() bad address (%p)\n", addr);
Linus Torvalds's avatar
Linus Torvalds committed
1362
1363
1364
1365
1366
		return;
	}

	area = remove_vm_area(addr);
	if (unlikely(!area)) {
Arjan van de Ven's avatar
Arjan van de Ven committed
1367
		WARN(1, KERN_ERR "Trying to vfree() nonexistent vm area (%p)\n",
Linus Torvalds's avatar
Linus Torvalds committed
1368
1369
1370
1371
				addr);
		return;
	}

1372
	debug_check_no_locks_freed(addr, area->size);
1373
	debug_check_no_obj_freed(addr, area->size);
1374

Linus Torvalds's avatar
Linus Torvalds committed
1375
1376
1377
1378
	if (deallocate_pages) {
		int i;

		for (i = 0; i < area->nr_pages; i++) {
1379
1380
1381
1382
			struct page *page = area->pages[i];

			BUG_ON(!page);
			__free_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
1383
1384
		}

1385
		if (area->flags & VM_VPAGES)
Linus Torvalds's avatar
Linus Torvalds committed
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
			vfree(area->pages);
		else
			kfree(area->pages);
	}

	kfree(area);
	return;
}

/**
 *	vfree  -  release memory allocated by vmalloc()
 *	@addr:		memory base address
 *
Simon Arlott's avatar
Simon Arlott committed
1399
 *	Free the virtually continuous memory area starting at @addr, as
1400
1401
 *	obtained from vmalloc(), vmalloc_32() or __vmalloc(). If @addr is
 *	NULL, no operation is performed.
Linus Torvalds's avatar
Linus Torvalds committed
1402
 *
1403
 *	Must not be called in interrupt context.
Linus Torvalds's avatar
Linus Torvalds committed
1404
 */
1405
void vfree(const void *addr)
Linus Torvalds's avatar
Linus Torvalds committed
1406
1407
{
	BUG_ON(in_interrupt());
1408
1409
1410

	kmemleak_free(addr);

Linus Torvalds's avatar
Linus Torvalds committed
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
	__vunmap(addr, 1);
}
EXPORT_SYMBOL(vfree);

/**
 *	vunmap  -  release virtual mapping obtained by vmap()
 *	@addr:		memory base address
 *
 *	Free the virtually contiguous memory area starting at @addr,
 *	which was created from the page array passed to vmap().
 *
1422
 *	Must not be called in interrupt context.
Linus Torvalds's avatar
Linus Torvalds committed
1423
 */
1424
void vunmap(const void *addr)
Linus Torvalds's avatar
Linus Torvalds committed
1425
1426
{
	BUG_ON(in_interrupt());
1427
	might_sleep();
Linus Torvalds's avatar
Linus Torvalds committed
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
	__vunmap(addr, 0);
}
EXPORT_SYMBOL(vunmap);

/**
 *	vmap  -  map an array of pages into virtually contiguous space
 *	@pages:		array of page pointers
 *	@count:		number of pages to map
 *	@flags:		vm_area->flags
 *	@prot:		page protection for the mapping
 *
 *	Maps @count pages from @pages into contiguous kernel virtual
 *	space.
 */
void *vmap(struct page **pages, unsigned int count,
		unsigned long flags, pgprot_t prot)
{
	struct vm_struct *area;

1447
1448
	might_sleep();

1449
	if (count > totalram_pages)
Linus Torvalds's avatar
Linus Torvalds committed
1450
1451
		return NULL;

1452
1453
	area = get_vm_area_caller((count << PAGE_SHIFT), flags,
					__builtin_return_address(0));
Linus Torvalds's avatar
Linus Torvalds committed
1454
1455
	if (!area)
		return NULL;
1456

Linus Torvalds's avatar
Linus Torvalds committed
1457
1458
1459
1460
1461
1462
1463
1464
1465
	if (map_vm_area(area, prot, &pages)) {
		vunmap(area->addr);
		return NULL;
	}

	return area->addr;
}
EXPORT_SYMBOL(vmap);

1466
1467
static void *__vmalloc_node(unsigned long size, unsigned long align,
			    gfp_t gfp_mask, pgprot_t prot,
Nick Piggin's avatar
Nick Piggin committed
1468
			    int node, void *caller);
1469
static void *__vmalloc_area_node(struct vm_struct *area, gfp_t gfp_mask,
1470
				 pgprot_t prot, int node, void *caller)
Linus Torvalds's avatar
Linus Torvalds committed
1471
1472
1473
{
	struct page **pages;
	unsigned int nr_pages, array_size, i;
1474
	gfp_t nested_gfp = (gfp_mask & GFP_RECLAIM_MASK) | __GFP_ZERO;
Linus Torvalds's avatar
Linus Torvalds committed
1475
1476
1477
1478
1479
1480

	nr_pages = (area->size - PAGE_SIZE) >> PAGE_SHIFT;
	array_size = (nr_pages * sizeof(struct page *));

	area->nr_pages = nr_pages;
	/* Please note that the recursion is strictly bounded. */
1481
	if (array_size > PAGE_SIZE) {
1482
		pages = __vmalloc_node(array_size, 1, nested_gfp|__GFP_HIGHMEM,
1483
				PAGE_KERNEL, node, caller);
1484
		area->flags |= VM_VPAGES;
1485
	} else {
1486
		pages = kmalloc_node(array_size, nested_gfp, node);
1487
	}
Linus Torvalds's avatar
Linus Torvalds committed
1488
	area->pages = pages;
1489
	area->caller = caller;
Linus Torvalds's avatar
Linus Torvalds committed
1490
1491
1492
1493
1494
1495
1496
	if (!area->pages) {
		remove_vm_area(area->addr);
		kfree(area);
		return NULL;
	}

	for (i = 0; i < area->nr_pages; i++) {
1497
1498
		struct page *page;

Christoph Lameter's avatar
Christoph Lameter committed
1499
		if (node < 0)
1500
			page = alloc_page(gfp_mask);
Christoph Lameter's avatar
Christoph Lameter committed
1501
		else
1502
1503
1504
			page = alloc_pages_node(node, gfp_mask, 0);

		if (unlikely(!page)) {
Linus Torvalds's avatar
Linus Torvalds committed
1505
1506
1507
1508
			/* Successfully allocated i pages, free them in __vunmap() */
			area->nr_pages = i;
			goto fail;
		}
1509
		area->pages[i] = page;
Linus Torvalds's avatar
Linus Torvalds committed
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
	}

	if (map_vm_area(area, prot, &pages))
		goto fail;
	return area->addr;

fail:
	vfree(area->addr);
	return NULL;
}

Christoph Lameter's avatar
Christoph Lameter committed
1521
1522
void *__vmalloc_area(struct vm_struct *area, gfp_t gfp_mask, pgprot_t prot)
{
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
	void *addr = __vmalloc_area_node(area, gfp_mask, prot, -1,
					 __builtin_return_address(0));

	/*
	 * A ref_count = 3 is needed because the vm_struct and vmap_area
	 * structures allocated in the __get_vm_area_node() function contain
	 * references to the virtual address of the vmalloc'ed block.
	 */
	kmemleak_alloc(addr, area->size - PAGE_SIZE, 3, gfp_mask);

	return addr;
Christoph Lameter's avatar
Christoph Lameter committed
1534
1535
}

Linus Torvalds's avatar
Linus Torvalds committed
1536
/**
Christoph Lameter's avatar
Christoph Lameter committed
1537
 *	__vmalloc_node  -  allocate virtually contiguous memory
Linus Torvalds's avatar
Linus Torvalds committed
1538
 *	@size:		allocation size
1539
 *	@align:		desired alignment
Linus Torvalds's avatar
Linus Torvalds committed
1540
1541
 *	@gfp_mask:	flags for the page level allocator
 *	@prot:		protection mask for the allocated pages
1542
 *	@node:		node to use for allocation or -1
1543
 *	@caller:	caller's return address
Linus Torvalds's avatar
Linus Torvalds committed
1544
1545
1546
1547
1548
 *
 *	Allocate enough pages to cover @size from the page level
 *	allocator with @gfp_mask flags.  Map them into contiguous
 *	kernel virtual space, using a pagetable protection of @prot.
 */
1549
1550
1551
static void *__vmalloc_node(unsigned long size, unsigned long align,
			    gfp_t gfp_mask, pgprot_t prot,
			    int node, void *caller)
Linus Torvalds's avatar
Linus Torvalds committed
1552
1553
{
	struct vm_struct *area;
1554
1555
	void *addr;
	unsigned long real_size = size;
Linus Torvalds's avatar
Linus Torvalds committed
1556
1557

	size = PAGE_ALIGN(size);
1558
	if (!size || (size >> PAGE_SHIFT) > totalram_pages)
Linus Torvalds's avatar
Linus Torvalds committed
1559
1560
		return NULL;

1561
1562
	area = __get_vm_area_node(size, align, VM_ALLOC, VMALLOC_START,
				  VMALLOC_END, node, gfp_mask, caller);
1563

Linus Torvalds's avatar
Linus Torvalds committed
1564
1565
1566
	if (!area)
		return NULL;

1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
	addr = __vmalloc_area_node(area, gfp_mask, prot, node, caller);

	/*
	 * A ref_count = 3 is needed because the vm_struct and vmap_area
	 * structures allocated in the __get_vm_area_node() function contain
	 * references to the virtual address of the vmalloc'ed block.
	 */
	kmemleak_alloc(addr, real_size, 3, gfp_mask);

	return addr;
Linus Torvalds's avatar
Linus Torvalds committed
1577
1578
}

Christoph Lameter's avatar
Christoph Lameter committed
1579
1580
void *__vmalloc(unsigned long size, gfp_t gfp_mask, pgprot_t prot)
{
1581
	return __vmalloc_node(size, 1, gfp_mask, prot, -1,
1582
				__builtin_return_address(0));
Christoph Lameter's avatar
Christoph Lameter committed
1583
}
Linus Torvalds's avatar
Linus Torvalds committed
1584
1585
1586
1587
1588
1589
1590
1591
EXPORT_SYMBOL(__vmalloc);

/**
 *	vmalloc  -  allocate virtually contiguous memory
 *	@size:		allocation size
 *	Allocate enough pages to cover @size from the page level
 *	allocator and map them into contiguous kernel virtual space.
 *
1592
 *	For tight control over page level allocator and protection flags