vmalloc.c 69 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>
Arun Sharma's avatar
Arun Sharma committed
29
#include <linux/atomic.h>
30
#include <linux/compiler.h>
31
#include <linux/llist.h>
32
#include <linux/bitops.h>
33

Linus Torvalds's avatar
Linus Torvalds committed
34 35
#include <asm/uaccess.h>
#include <asm/tlbflush.h>
36
#include <asm/shmparam.h>
Linus Torvalds's avatar
Linus Torvalds committed
37

38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56
struct vfree_deferred {
	struct llist_head list;
	struct work_struct wq;
};
static DEFINE_PER_CPU(struct vfree_deferred, vfree_deferred);

static void __vunmap(const void *, int);

static void free_work(struct work_struct *w)
{
	struct vfree_deferred *p = container_of(w, struct vfree_deferred, wq);
	struct llist_node *llnode = llist_del_all(&p->list);
	while (llnode) {
		void *p = llnode;
		llnode = llist_next(llnode);
		__vunmap(p, 1);
	}
}

Nick Piggin's avatar
Nick Piggin committed
57
/*** Page table manipulation functions ***/
58

Linus Torvalds's avatar
Linus Torvalds committed
59 60 61 62 63 64 65 66 67 68 69
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
70
static void vunmap_pmd_range(pud_t *pud, unsigned long addr, unsigned long end)
Linus Torvalds's avatar
Linus Torvalds committed
71 72 73 74 75 76 77
{
	pmd_t *pmd;
	unsigned long next;

	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
78 79
		if (pmd_clear_huge(pmd))
			continue;
Linus Torvalds's avatar
Linus Torvalds committed
80 81 82 83 84 85
		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
86
static void vunmap_pud_range(pgd_t *pgd, unsigned long addr, unsigned long end)
Linus Torvalds's avatar
Linus Torvalds committed
87 88 89 90 91 92 93
{
	pud_t *pud;
	unsigned long next;

	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
94 95
		if (pud_clear_huge(pud))
			continue;
Linus Torvalds's avatar
Linus Torvalds committed
96 97 98 99 100 101
		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
102
static void vunmap_page_range(unsigned long addr, unsigned long end)
Linus Torvalds's avatar
Linus Torvalds committed
103 104 105 106 107 108 109 110 111 112 113 114 115 116 117
{
	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
118
		unsigned long end, pgprot_t prot, struct page **pages, int *nr)
Linus Torvalds's avatar
Linus Torvalds committed
119 120 121
{
	pte_t *pte;

Nick Piggin's avatar
Nick Piggin committed
122 123 124 125 126
	/*
	 * nr is a running index into the array which helps higher level
	 * callers keep track of where we're up to.
	 */

127
	pte = pte_alloc_kernel(pmd, addr);
Linus Torvalds's avatar
Linus Torvalds committed
128 129 130
	if (!pte)
		return -ENOMEM;
	do {
Nick Piggin's avatar
Nick Piggin committed
131 132 133 134 135
		struct page *page = pages[*nr];

		if (WARN_ON(!pte_none(*pte)))
			return -EBUSY;
		if (WARN_ON(!page))
Linus Torvalds's avatar
Linus Torvalds committed
136 137
			return -ENOMEM;
		set_pte_at(&init_mm, addr, pte, mk_pte(page, prot));
Nick Piggin's avatar
Nick Piggin committed
138
		(*nr)++;
Linus Torvalds's avatar
Linus Torvalds committed
139 140 141 142
	} while (pte++, addr += PAGE_SIZE, addr != end);
	return 0;
}

Nick Piggin's avatar
Nick Piggin committed
143 144
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
145 146 147 148 149 150 151 152 153
{
	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
154
		if (vmap_pte_range(pmd, addr, next, prot, pages, nr))
Linus Torvalds's avatar
Linus Torvalds committed
155 156 157 158 159
			return -ENOMEM;
	} while (pmd++, addr = next, addr != end);
	return 0;
}

Nick Piggin's avatar
Nick Piggin committed
160 161
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
162 163 164 165 166 167 168 169 170
{
	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
171
		if (vmap_pmd_range(pud, addr, next, prot, pages, nr))
Linus Torvalds's avatar
Linus Torvalds committed
172 173 174 175 176
			return -ENOMEM;
	} while (pud++, addr = next, addr != end);
	return 0;
}

Nick Piggin's avatar
Nick Piggin committed
177 178 179 180 181 182
/*
 * 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]
 */
183 184
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
185 186 187
{
	pgd_t *pgd;
	unsigned long next;
188
	unsigned long addr = start;
Nick Piggin's avatar
Nick Piggin committed
189 190
	int err = 0;
	int nr = 0;
Linus Torvalds's avatar
Linus Torvalds committed
191 192 193 194 195

	BUG_ON(addr >= end);
	pgd = pgd_offset_k(addr);
	do {
		next = pgd_addr_end(addr, end);
Nick Piggin's avatar
Nick Piggin committed
196
		err = vmap_pud_range(pgd, addr, next, prot, pages, &nr);
Linus Torvalds's avatar
Linus Torvalds committed
197
		if (err)
198
			return err;
Linus Torvalds's avatar
Linus Torvalds committed
199
	} while (pgd++, addr = next, addr != end);
Nick Piggin's avatar
Nick Piggin committed
200 201

	return nr;
Linus Torvalds's avatar
Linus Torvalds committed
202 203
}

204 205 206 207 208 209 210 211 212 213
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;
}

214
int is_vmalloc_or_module_addr(const void *x)
215 216
{
	/*
217
	 * ARM, x86-64 and sparc64 put modules in a special place,
218 219 220 221 222 223 224 225 226 227 228
	 * 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);
}

229
/*
230
 * Walk a vmap address to the struct page it maps.
231
 */
232
struct page *vmalloc_to_page(const void *vmalloc_addr)
233 234
{
	unsigned long addr = (unsigned long) vmalloc_addr;
235
	struct page *page = NULL;
236 237
	pgd_t *pgd = pgd_offset_k(addr);

238 239 240 241
	/*
	 * XXX we might need to change this if we add VIRTUAL_BUG_ON for
	 * architectures that do not vmalloc module space
	 */
242
	VIRTUAL_BUG_ON(!is_vmalloc_or_module_addr(vmalloc_addr));
Jiri Slaby's avatar
Jiri Slaby committed
243

244
	if (!pgd_none(*pgd)) {
Nick Piggin's avatar
Nick Piggin committed
245
		pud_t *pud = pud_offset(pgd, addr);
246
		if (!pud_none(*pud)) {
Nick Piggin's avatar
Nick Piggin committed
247
			pmd_t *pmd = pmd_offset(pud, addr);
248
			if (!pmd_none(*pmd)) {
Nick Piggin's avatar
Nick Piggin committed
249 250
				pte_t *ptep, pte;

251 252 253
				ptep = pte_offset_map(pmd, addr);
				pte = *ptep;
				if (pte_present(pte))
254
					page = pte_page(pte);
255 256 257 258
				pte_unmap(ptep);
			}
		}
	}
259
	return page;
260
}
261
EXPORT_SYMBOL(vmalloc_to_page);
262 263

/*
264
 * Map a vmalloc()-space virtual address to the physical page frame number.
265
 */
266
unsigned long vmalloc_to_pfn(const void *vmalloc_addr)
267
{
268
	return page_to_pfn(vmalloc_to_page(vmalloc_addr));
269
}
270
EXPORT_SYMBOL(vmalloc_to_pfn);
271

Nick Piggin's avatar
Nick Piggin committed
272 273 274 275 276 277 278 279

/*** Global kva allocator ***/

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

static DEFINE_SPINLOCK(vmap_area_lock);
280 281
/* Export for kexec only */
LIST_HEAD(vmap_area_list);
Nick Piggin's avatar
Nick Piggin committed
282 283 284 285 286 287 288 289
static struct rb_root vmap_area_root = RB_ROOT;

/* The vmap cache globals are protected by vmap_area_lock */
static struct rb_node *free_vmap_cache;
static unsigned long cached_hole_size;
static unsigned long cached_vstart;
static unsigned long cached_align;

290
static unsigned long vmap_area_pcpu_hole;
Nick Piggin's avatar
Nick Piggin committed
291 292

static struct vmap_area *__find_vmap_area(unsigned long addr)
Linus Torvalds's avatar
Linus Torvalds committed
293
{
Nick Piggin's avatar
Nick Piggin committed
294 295 296 297 298 299 300 301
	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;
302
		else if (addr >= va->va_end)
Nick Piggin's avatar
Nick Piggin committed
303 304 305 306 307 308 309 310 311 312 313 314 315 316 317
			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) {
318
		struct vmap_area *tmp_va;
Nick Piggin's avatar
Nick Piggin committed
319 320

		parent = *p;
321 322
		tmp_va = rb_entry(parent, struct vmap_area, rb_node);
		if (va->va_start < tmp_va->va_end)
Nick Piggin's avatar
Nick Piggin committed
323
			p = &(*p)->rb_left;
324
		else if (va->va_end > tmp_va->va_start)
Nick Piggin's avatar
Nick Piggin committed
325 326 327 328 329 330 331 332
			p = &(*p)->rb_right;
		else
			BUG();
	}

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

333
	/* address-sort this list */
Nick Piggin's avatar
Nick Piggin committed
334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355
	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
356
	unsigned long addr;
Nick Piggin's avatar
Nick Piggin committed
357
	int purged = 0;
Nick Piggin's avatar
Nick Piggin committed
358
	struct vmap_area *first;
Nick Piggin's avatar
Nick Piggin committed
359

Nick Piggin's avatar
Nick Piggin committed
360
	BUG_ON(!size);
Nick Piggin's avatar
Nick Piggin committed
361
	BUG_ON(size & ~PAGE_MASK);
Nick Piggin's avatar
Nick Piggin committed
362
	BUG_ON(!is_power_of_2(align));
Nick Piggin's avatar
Nick Piggin committed
363 364 365 366 367 368

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

369 370 371 372 373 374
	/*
	 * Only scan the relevant parts containing pointers to other objects
	 * to avoid false negatives.
	 */
	kmemleak_scan_area(&va->rb_node, SIZE_MAX, gfp_mask & GFP_RECLAIM_MASK);

Nick Piggin's avatar
Nick Piggin committed
375 376
retry:
	spin_lock(&vmap_area_lock);
Nick Piggin's avatar
Nick Piggin committed
377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400
	/*
	 * Invalidate cache if we have more permissive parameters.
	 * cached_hole_size notes the largest hole noticed _below_
	 * the vmap_area cached in free_vmap_cache: if size fits
	 * into that hole, we want to scan from vstart to reuse
	 * the hole instead of allocating above free_vmap_cache.
	 * Note that __free_vmap_area may update free_vmap_cache
	 * without updating cached_hole_size or cached_align.
	 */
	if (!free_vmap_cache ||
			size < cached_hole_size ||
			vstart < cached_vstart ||
			align < cached_align) {
nocache:
		cached_hole_size = 0;
		free_vmap_cache = NULL;
	}
	/* record if we encounter less permissive parameters */
	cached_vstart = vstart;
	cached_align = align;

	/* find starting point for our search */
	if (free_vmap_cache) {
		first = rb_entry(free_vmap_cache, struct vmap_area, rb_node);
401
		addr = ALIGN(first->va_end, align);
Nick Piggin's avatar
Nick Piggin committed
402 403
		if (addr < vstart)
			goto nocache;
404
		if (addr + size < addr)
Nick Piggin's avatar
Nick Piggin committed
405 406 407 408
			goto overflow;

	} else {
		addr = ALIGN(vstart, align);
409
		if (addr + size < addr)
Nick Piggin's avatar
Nick Piggin committed
410 411 412 413 414 415
			goto overflow;

		n = vmap_area_root.rb_node;
		first = NULL;

		while (n) {
Nick Piggin's avatar
Nick Piggin committed
416 417 418 419
			struct vmap_area *tmp;
			tmp = rb_entry(n, struct vmap_area, rb_node);
			if (tmp->va_end >= addr) {
				first = tmp;
Nick Piggin's avatar
Nick Piggin committed
420 421 422 423
				if (tmp->va_start <= addr)
					break;
				n = n->rb_left;
			} else
Nick Piggin's avatar
Nick Piggin committed
424
				n = n->rb_right;
Nick Piggin's avatar
Nick Piggin committed
425
		}
Nick Piggin's avatar
Nick Piggin committed
426 427 428 429

		if (!first)
			goto found;
	}
Nick Piggin's avatar
Nick Piggin committed
430 431

	/* from the starting point, walk areas until a suitable hole is found */
432
	while (addr + size > first->va_start && addr + size <= vend) {
Nick Piggin's avatar
Nick Piggin committed
433 434
		if (addr + cached_hole_size < first->va_start)
			cached_hole_size = first->va_start - addr;
435
		addr = ALIGN(first->va_end, align);
436
		if (addr + size < addr)
Nick Piggin's avatar
Nick Piggin committed
437 438
			goto overflow;

439
		if (list_is_last(&first->list, &vmap_area_list))
Nick Piggin's avatar
Nick Piggin committed
440
			goto found;
441 442 443

		first = list_entry(first->list.next,
				struct vmap_area, list);
Nick Piggin's avatar
Nick Piggin committed
444 445
	}

Nick Piggin's avatar
Nick Piggin committed
446 447 448
found:
	if (addr + size > vend)
		goto overflow;
Nick Piggin's avatar
Nick Piggin committed
449 450 451 452 453

	va->va_start = addr;
	va->va_end = addr + size;
	va->flags = 0;
	__insert_vmap_area(va);
Nick Piggin's avatar
Nick Piggin committed
454
	free_vmap_cache = &va->rb_node;
Nick Piggin's avatar
Nick Piggin committed
455 456
	spin_unlock(&vmap_area_lock);

Nick Piggin's avatar
Nick Piggin committed
457 458 459 460
	BUG_ON(va->va_start & (align-1));
	BUG_ON(va->va_start < vstart);
	BUG_ON(va->va_end > vend);

Nick Piggin's avatar
Nick Piggin committed
461
	return va;
Nick Piggin's avatar
Nick Piggin committed
462 463 464 465 466 467 468 469 470

overflow:
	spin_unlock(&vmap_area_lock);
	if (!purged) {
		purge_vmap_area_lazy();
		purged = 1;
		goto retry;
	}
	if (printk_ratelimit())
471
		pr_warn("vmap allocation for size %lu failed: "
Nick Piggin's avatar
Nick Piggin committed
472 473 474
			"use vmalloc=<size> to increase size.\n", size);
	kfree(va);
	return ERR_PTR(-EBUSY);
Nick Piggin's avatar
Nick Piggin committed
475 476 477 478 479
}

static void __free_vmap_area(struct vmap_area *va)
{
	BUG_ON(RB_EMPTY_NODE(&va->rb_node));
Nick Piggin's avatar
Nick Piggin committed
480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495

	if (free_vmap_cache) {
		if (va->va_end < cached_vstart) {
			free_vmap_cache = NULL;
		} else {
			struct vmap_area *cache;
			cache = rb_entry(free_vmap_cache, struct vmap_area, rb_node);
			if (va->va_start <= cache->va_start) {
				free_vmap_cache = rb_prev(&va->rb_node);
				/*
				 * We don't try to update cached_hole_size or
				 * cached_align, but it won't go very wrong.
				 */
			}
		}
	}
Nick Piggin's avatar
Nick Piggin committed
496 497 498 499
	rb_erase(&va->rb_node, &vmap_area_root);
	RB_CLEAR_NODE(&va->rb_node);
	list_del_rcu(&va->list);

500 501 502 503 504 505 506 507 508
	/*
	 * 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);

509
	kfree_rcu(va, rcu_head);
Nick Piggin's avatar
Nick Piggin committed
510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529
}

/*
 * 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);
}

530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550
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
551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577
/*
 * 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);

578 579 580
/* for per-CPU blocks */
static void purge_fragmented_blocks_allcpus(void);

581 582 583 584 585 586 587 588 589
/*
 * called before a call to iounmap() if the caller wants vm_area_struct's
 * immediately freed.
 */
void set_iounmap_nonlazy(void)
{
	atomic_set(&vmap_lazy_nr, lazy_max_pages()+1);
}

Nick Piggin's avatar
Nick Piggin committed
590 591 592 593 594 595 596 597 598 599 600 601 602
/*
 * 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)
{
603
	static DEFINE_SPINLOCK(purge_lock);
Nick Piggin's avatar
Nick Piggin committed
604 605
	LIST_HEAD(valist);
	struct vmap_area *va;
606
	struct vmap_area *n_va;
Nick Piggin's avatar
Nick Piggin committed
607 608 609 610 611 612 613 614
	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) {
615
		if (!spin_trylock(&purge_lock))
Nick Piggin's avatar
Nick Piggin committed
616 617
			return;
	} else
618
		spin_lock(&purge_lock);
Nick Piggin's avatar
Nick Piggin committed
619

620 621 622
	if (sync)
		purge_fragmented_blocks_allcpus();

Nick Piggin's avatar
Nick Piggin committed
623 624 625 626 627 628 629 630 631 632 633 634 635 636 637
	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;
			list_add_tail(&va->purge_list, &valist);
			va->flags |= VM_LAZY_FREEING;
			va->flags &= ~VM_LAZY_FREE;
		}
	}
	rcu_read_unlock();

638
	if (nr)
Nick Piggin's avatar
Nick Piggin committed
639 640 641 642 643 644 645
		atomic_sub(nr, &vmap_lazy_nr);

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

	if (nr) {
		spin_lock(&vmap_area_lock);
646
		list_for_each_entry_safe(va, n_va, &valist, purge_list)
Nick Piggin's avatar
Nick Piggin committed
647 648 649
			__free_vmap_area(va);
		spin_unlock(&vmap_area_lock);
	}
650
	spin_unlock(&purge_lock);
Nick Piggin's avatar
Nick Piggin committed
651 652
}

Nick Piggin's avatar
Nick Piggin committed
653 654 655 656 657 658 659 660 661 662 663
/*
 * 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
664 665 666 667 668 669 670
/*
 * 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
671
	__purge_vmap_area_lazy(&start, &end, 1, 0);
Nick Piggin's avatar
Nick Piggin committed
672 673 674
}

/*
675 676 677
 * Free a vmap area, caller ensuring that the area has been unmapped
 * and flush_cache_vunmap had been called for the correct range
 * previously.
Nick Piggin's avatar
Nick Piggin committed
678
 */
679
static void free_vmap_area_noflush(struct vmap_area *va)
Nick Piggin's avatar
Nick Piggin committed
680 681 682 683
{
	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
684
		try_purge_vmap_area_lazy();
Nick Piggin's avatar
Nick Piggin committed
685 686
}

687 688 689 690 691 692 693 694 695 696
/*
 * Free and unmap a vmap area, caller ensuring flush_cache_vunmap had been
 * called for the correct range previously.
 */
static void free_unmap_vmap_area_noflush(struct vmap_area *va)
{
	unmap_vmap_area(va);
	free_vmap_area_noflush(va);
}

697 698 699 700 701 702 703 704 705
/*
 * 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
706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749
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() */
750 751 752 753
#define VMAP_BBMAP_BITS		\
		VMAP_MIN(VMAP_BBMAP_BITS_MAX,	\
		VMAP_MAX(VMAP_BBMAP_BITS_MIN,	\
			VMALLOC_PAGES / roundup_pow_of_two(NR_CPUS) / 16))
Nick Piggin's avatar
Nick Piggin committed
754 755 756

#define VMAP_BLOCK_SIZE		(VMAP_BBMAP_BITS * PAGE_SIZE)

757 758
static bool vmap_initialized __read_mostly = false;

Nick Piggin's avatar
Nick Piggin committed
759 760 761 762 763 764 765 766 767
struct vmap_block_queue {
	spinlock_t lock;
	struct list_head free;
};

struct vmap_block {
	spinlock_t lock;
	struct vmap_area *va;
	unsigned long free, dirty;
768
	unsigned long dirty_min, dirty_max; /*< dirty range */
769 770
	struct list_head free_list;
	struct rcu_head rcu_head;
771
	struct list_head purge;
Nick Piggin's avatar
Nick Piggin committed
772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798
};

/* 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;
}

799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816
static void *vmap_block_vaddr(unsigned long va_start, unsigned long pages_off)
{
	unsigned long addr;

	addr = va_start + (pages_off << PAGE_SHIFT);
	BUG_ON(addr_to_vb_idx(addr) != addr_to_vb_idx(va_start));
	return (void *)addr;
}

/**
 * new_vmap_block - allocates new vmap_block and occupies 2^order pages in this
 *                  block. Of course pages number can't exceed VMAP_BBMAP_BITS
 * @order:    how many 2^order pages should be occupied in newly allocated block
 * @gfp_mask: flags for the page level allocator
 *
 * Returns: virtual address in a newly allocated block or ERR_PTR(-errno)
 */
static void *new_vmap_block(unsigned int order, gfp_t gfp_mask)
Nick Piggin's avatar
Nick Piggin committed
817 818 819 820 821 822
{
	struct vmap_block_queue *vbq;
	struct vmap_block *vb;
	struct vmap_area *va;
	unsigned long vb_idx;
	int node, err;
823
	void *vaddr;
Nick Piggin's avatar
Nick Piggin committed
824 825 826 827 828 829 830 831 832 833 834

	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);
835
	if (IS_ERR(va)) {
Nick Piggin's avatar
Nick Piggin committed
836
		kfree(vb);
Julia Lawall's avatar
Julia Lawall committed
837
		return ERR_CAST(va);
Nick Piggin's avatar
Nick Piggin committed
838 839 840 841 842 843 844 845 846
	}

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

847
	vaddr = vmap_block_vaddr(va->va_start, 0);
Nick Piggin's avatar
Nick Piggin committed
848 849
	spin_lock_init(&vb->lock);
	vb->va = va;
850 851 852
	/* At least something should be left free */
	BUG_ON(VMAP_BBMAP_BITS <= (1UL << order));
	vb->free = VMAP_BBMAP_BITS - (1UL << order);
Nick Piggin's avatar
Nick Piggin committed
853
	vb->dirty = 0;
854 855
	vb->dirty_min = VMAP_BBMAP_BITS;
	vb->dirty_max = 0;
Nick Piggin's avatar
Nick Piggin committed
856 857 858 859 860 861 862 863 864 865 866
	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);
	spin_lock(&vbq->lock);
867
	list_add_tail_rcu(&vb->free_list, &vbq->free);
Nick Piggin's avatar
Nick Piggin committed
868
	spin_unlock(&vbq->lock);
869
	put_cpu_var(vmap_block_queue);
Nick Piggin's avatar
Nick Piggin committed
870

871
	return vaddr;
Nick Piggin's avatar
Nick Piggin committed
872 873 874 875 876 877 878 879 880 881 882 883 884
}

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);

885
	free_vmap_area_noflush(vb->va);
886
	kfree_rcu(vb, rcu_head);
Nick Piggin's avatar
Nick Piggin committed
887 888
}

889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905
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 */
906 907
			vb->dirty_min = 0;
			vb->dirty_max = VMAP_BBMAP_BITS;
908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931
			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_allcpus(void)
{
	int cpu;

	for_each_possible_cpu(cpu)
		purge_fragmented_blocks(cpu);
}

Nick Piggin's avatar
Nick Piggin committed
932 933 934 935
static void *vb_alloc(unsigned long size, gfp_t gfp_mask)
{
	struct vmap_block_queue *vbq;
	struct vmap_block *vb;
936
	void *vaddr = NULL;
Nick Piggin's avatar
Nick Piggin committed
937 938 939 940
	unsigned int order;

	BUG_ON(size & ~PAGE_MASK);
	BUG_ON(size > PAGE_SIZE*VMAP_MAX_ALLOC);
Jan Kara's avatar
Jan Kara committed
941 942 943 944 945 946 947 948
	if (WARN_ON(size == 0)) {
		/*
		 * Allocating 0 bytes isn't what caller wants since
		 * get_order(0) returns funny result. Just warn and terminate
		 * early.
		 */
		return NULL;
	}
Nick Piggin's avatar
Nick Piggin committed
949 950 951 952 953
	order = get_order(size);

	rcu_read_lock();
	vbq = &get_cpu_var(vmap_block_queue);
	list_for_each_entry_rcu(vb, &vbq->free, free_list) {
954
		unsigned long pages_off;
Nick Piggin's avatar
Nick Piggin committed
955 956

		spin_lock(&vb->lock);
957 958 959 960
		if (vb->free < (1UL << order)) {
			spin_unlock(&vb->lock);
			continue;
		}
961

962 963
		pages_off = VMAP_BBMAP_BITS - vb->free;
		vaddr = vmap_block_vaddr(vb->va->va_start, pages_off);
964 965 966 967 968 969
		vb->free -= 1UL << order;
		if (vb->free == 0) {
			spin_lock(&vbq->lock);
			list_del_rcu(&vb->free_list);
			spin_unlock(&vbq->lock);
		}
970

971 972
		spin_unlock(&vb->lock);
		break;
Nick Piggin's avatar
Nick Piggin committed
973
	}
974

975
	put_cpu_var(vmap_block_queue);
Nick Piggin's avatar
Nick Piggin committed
976 977
	rcu_read_unlock();

978 979 980
	/* Allocate new block if nothing was found */
	if (!vaddr)
		vaddr = new_vmap_block(order, gfp_mask);
Nick Piggin's avatar
Nick Piggin committed
981

982
	return vaddr;
Nick Piggin's avatar
Nick Piggin committed
983 984 985 986 987 988 989 990 991 992 993
}

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);
994 995 996

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

Nick Piggin's avatar
Nick Piggin committed
997 998 999
	order = get_order(size);

	offset = (unsigned long)addr & (VMAP_BLOCK_SIZE - 1);
1000
	offset >>= PAGE_SHIFT;
Nick Piggin's avatar
Nick Piggin committed
1001 1002 1003 1004 1005 1006 1007

	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);

1008 1009
	vunmap_page_range((unsigned long)addr, (unsigned long)addr + size);

Nick Piggin's avatar
Nick Piggin committed
1010
	spin_lock(&vb->lock);
1011 1012 1013 1014

	/* Expand dirty range */
	vb->dirty_min = min(vb->dirty_min, offset);
	vb->dirty_max = max(vb->dirty_max, offset + (1UL << order));
1015

Nick Piggin's avatar
Nick Piggin committed
1016 1017
	vb->dirty += 1UL << order;
	if (vb->dirty == VMAP_BBMAP_BITS) {
1018
		BUG_ON(vb->free);
Nick Piggin's avatar
Nick Piggin committed
1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043
		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;

1044 1045 1046
	if (unlikely(!vmap_initialized))
		return;

Nick Piggin's avatar
Nick Piggin committed
1047 1048 1049 1050 1051 1052 1053
	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) {
			spin_lock(&vb->lock);
1054 1055
			if (vb->dirty) {
				unsigned long va_start = vb->va->va_start;
Nick Piggin's avatar
Nick Piggin committed
1056
				unsigned long s, e;
1057

1058 1059
				s = va_start + (vb->dirty_min << PAGE_SHIFT);
				e = va_start + (vb->dirty_max << PAGE_SHIFT);
Nick Piggin's avatar
Nick Piggin committed
1060

1061 1062
				start = min(s, start);
				end   = max(e, end);
Nick Piggin's avatar
Nick Piggin committed
1063

1064
				flush = 1;
Nick Piggin's avatar
Nick Piggin committed
1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090
			}
			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);
1091
	vmap_debug_free_range(addr, addr+size);
Nick Piggin's avatar
Nick Piggin committed
1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105

	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
1106
 *
1107 1108 1109 1110 1111 1112
 * If you use this function for less than VMAP_MAX_ALLOC pages, it could be
 * faster than vmap so it's good.  But if you mix long-life and short-life
 * objects with vm_map_ram(), it could consume lots of address space through
 * fragmentation (especially on a 32bit machine).  You could see failures in
 * the end.  Please use this function for short-lived objects.
 *
1113
 * Returns: a pointer to the address that has been mapped, or %NULL on failure
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
 */
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);

1144
static struct vm_struct *vmlist __initdata;
Nicolas Pitre's avatar
Nicolas Pitre committed
1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170
/**
 * vm_area_add_early - add vmap area early during boot
 * @vm: vm_struct to add
 *
 * This function is used to add fixed kernel vm area to vmlist before
 * vmalloc_init() is called.  @vm->addr, @vm->size, and @vm->flags
 * should contain proper values and the other fields should be zero.
 *
 * DO NOT USE THIS FUNCTION UNLESS YOU KNOW WHAT YOU'RE DOING.
 */
void __init vm_area_add_early(struct vm_struct *vm)
{
	struct vm_struct *tmp, **p;

	BUG_ON(vmap_initialized);
	for (p = &vmlist; (tmp = *p) != NULL; p = &tmp->next) {
		if (tmp->addr >= vm->addr) {
			BUG_ON(tmp->addr < vm->addr + vm->size);
			break;
		} else
			BUG_ON(tmp->addr + tmp->size > vm->addr);
	}
	vm->next = *p;
	*p = vm;
}

1171 1172 1173
/**
 * vm_area_register_early - register vmap area early during boot
 * @vm: vm_struct to register
1174
 * @align: requested alignment
1175 1176 1177 1178 1179 1180 1181 1182
 *
 * 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.
 */
1183
void __init vm_area_register_early(struct vm_struct *vm, size_t align)
1184 1185
{
	static size_t vm_init_off __initdata;
1186 1187 1188 1189
	unsigned long addr;

	addr = ALIGN(VMALLOC_START + vm_init_off, align);
	vm_init_off = PFN_ALIGN(addr + vm->size) - VMALLOC_START;
1190

1191
	vm->addr = (void *)addr;
1192

Nicolas Pitre's avatar
Nicolas Pitre committed
1193
	vm_area_add_early(vm);
1194 1195
}

Nick Piggin's avatar
Nick Piggin committed
1196 1197
void __init vmalloc_init(void)
{
Ivan Kokshaysky's avatar
Ivan Kokshaysky committed
1198 1199
	struct vmap_area *va;
	struct vm_struct *tmp;
Nick Piggin's avatar
Nick Piggin committed
1200 1201 1202 1203
	int i;

	for_each_possible_cpu(i) {
		struct vmap_block_queue *vbq;
1204
		struct vfree_deferred *p;
Nick Piggin's avatar
Nick Piggin committed
1205 1206 1207 1208

		vbq = &per_cpu(vmap_block_queue, i);
		spin_lock_init(&vbq->lock);
		INIT_LIST_HEAD(&vbq->free);
1209 1210 1211
		p = &per_cpu(vfree_deferred, i);
		init_llist_head(&p->list);
		INIT_WORK(&p->wq, free_work);
Nick Piggin's avatar
Nick Piggin committed
1212
	}
1213

Ivan Kokshaysky's avatar
Ivan Kokshaysky committed
1214 1215
	/* Import existing vmlist entries. */
	for (tmp = vmlist; tmp; tmp = tmp->next) {
1216
		va = kzalloc(sizeof(struct vmap_area), GFP_NOWAIT);
1217
		va->flags = VM_VM_AREA;
Ivan Kokshaysky's avatar
Ivan Kokshaysky committed
1218 1219
		va->va_start = (unsigned long)tmp->addr;
		va->va_end = va->va_start + tmp->size;
1220
		va->vm = tmp;
Ivan Kokshaysky's avatar
Ivan Kokshaysky committed
1221 1222
		__insert_vmap_area(va);
	}
1223 1224 1225

	vmap_area_pcpu_hole = VMALLOC_END;

1226
	vmap_initialized = true;
Nick Piggin's avatar
Nick Piggin committed
1227 1228
}

1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271
/**
 * 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);
}
1272
EXPORT_SYMBOL_GPL(unmap_kernel_range_noflush);
1273 1274 1275 1276 1277 1278 1279 1280 1281

/**
 * 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
1282 1283 1284
void unmap_kernel_range(unsigned long addr, unsigned long size)
{
	unsigned long end = addr + size;
1285 1286

	flush_cache_vunmap(addr, end);
Nick Piggin's avatar
Nick Piggin committed
1287 1288 1289
	vunmap_page_range(addr, end);
	flush_tlb_kernel_range(addr, end);
}
1290
EXPORT_SYMBOL_GPL(unmap_kernel_range);
Nick Piggin's avatar
Nick Piggin committed
1291

1292
int map_vm_area(struct vm_struct *area, pgprot_t prot, struct page **pages)
Nick Piggin's avatar
Nick Piggin committed
1293 1294
{
	unsigned long addr = (unsigned long)area->addr;
1295
	unsigned long end = addr + get_vm_area_size(area);
Nick Piggin's avatar
Nick Piggin committed
1296 1297
	int err;

1298
	err = vmap_page_range(addr, end, prot, pages);
Nick Piggin's avatar
Nick Piggin committed
1299

1300
	return err > 0 ? 0 : err;
Nick Piggin's avatar
Nick Piggin committed
1301 1302 1303
}
EXPORT_SYMBOL_GPL(map_vm_area);

1304
static void setup_vmalloc_vm(struct vm_struct *vm, struct vmap_area *va,
1305
			      unsigned long flags, const void *caller)
1306
{
1307
	spin_lock(&vmap_area_lock);
1308 1309 1310 1311
	vm->flags = flags;
	vm->addr = (void *)va->va_start;
	vm->size = va->va_end - va->va_start;
	vm->caller = caller;
1312
	va->vm = vm;