vmalloc.c 67.9 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/llist.h>
Linus Torvalds's avatar
Linus Torvalds committed
31 32
#include <asm/uaccess.h>
#include <asm/tlbflush.h>
33
#include <asm/shmparam.h>
Linus Torvalds's avatar
Linus Torvalds committed
34

35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53
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
54
/*** Page table manipulation functions ***/
55

Linus Torvalds's avatar
Linus Torvalds committed
56 57 58 59 60 61 62 63 64 65 66
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
67
static void vunmap_pmd_range(pud_t *pud, unsigned long addr, unsigned long end)
Linus Torvalds's avatar
Linus Torvalds committed
68 69 70 71 72 73 74 75 76 77 78 79 80
{
	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
81
static void vunmap_pud_range(pgd_t *pgd, unsigned long addr, unsigned long end)
Linus Torvalds's avatar
Linus Torvalds committed
82 83 84 85 86 87 88 89 90 91 92 93 94
{
	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
95
static void vunmap_page_range(unsigned long addr, unsigned long end)
Linus Torvalds's avatar
Linus Torvalds committed
96 97 98 99 100 101 102 103 104 105 106 107 108 109 110
{
	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
111
		unsigned long end, pgprot_t prot, struct page **pages, int *nr)
Linus Torvalds's avatar
Linus Torvalds committed
112 113 114
{
	pte_t *pte;

Nick Piggin's avatar
Nick Piggin committed
115 116 117 118 119
	/*
	 * nr is a running index into the array which helps higher level
	 * callers keep track of where we're up to.
	 */

120
	pte = pte_alloc_kernel(pmd, addr);
Linus Torvalds's avatar
Linus Torvalds committed
121 122 123
	if (!pte)
		return -ENOMEM;
	do {
Nick Piggin's avatar
Nick Piggin committed
124 125 126 127 128
		struct page *page = pages[*nr];

		if (WARN_ON(!pte_none(*pte)))
			return -EBUSY;
		if (WARN_ON(!page))
Linus Torvalds's avatar
Linus Torvalds committed
129 130
			return -ENOMEM;
		set_pte_at(&init_mm, addr, pte, mk_pte(page, prot));
Nick Piggin's avatar
Nick Piggin committed
131
		(*nr)++;
Linus Torvalds's avatar
Linus Torvalds committed
132 133 134 135
	} while (pte++, addr += PAGE_SIZE, addr != end);
	return 0;
}

Nick Piggin's avatar
Nick Piggin committed
136 137
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
138 139 140 141 142 143 144 145 146
{
	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
147
		if (vmap_pte_range(pmd, addr, next, prot, pages, nr))
Linus Torvalds's avatar
Linus Torvalds committed
148 149 150 151 152
			return -ENOMEM;
	} while (pmd++, addr = next, addr != end);
	return 0;
}

Nick Piggin's avatar
Nick Piggin committed
153 154
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
155 156 157 158 159 160 161 162 163
{
	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
164
		if (vmap_pmd_range(pud, addr, next, prot, pages, nr))
Linus Torvalds's avatar
Linus Torvalds committed
165 166 167 168 169
			return -ENOMEM;
	} while (pud++, addr = next, addr != end);
	return 0;
}

Nick Piggin's avatar
Nick Piggin committed
170 171 172 173 174 175
/*
 * 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]
 */
176 177
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
178 179 180
{
	pgd_t *pgd;
	unsigned long next;
181
	unsigned long addr = start;
Nick Piggin's avatar
Nick Piggin committed
182 183
	int err = 0;
	int nr = 0;
Linus Torvalds's avatar
Linus Torvalds committed
184 185 186 187 188

	BUG_ON(addr >= end);
	pgd = pgd_offset_k(addr);
	do {
		next = pgd_addr_end(addr, end);
Nick Piggin's avatar
Nick Piggin committed
189
		err = vmap_pud_range(pgd, addr, next, prot, pages, &nr);
Linus Torvalds's avatar
Linus Torvalds committed
190
		if (err)
191
			return err;
Linus Torvalds's avatar
Linus Torvalds committed
192
	} while (pgd++, addr = next, addr != end);
Nick Piggin's avatar
Nick Piggin committed
193 194

	return nr;
Linus Torvalds's avatar
Linus Torvalds committed
195 196
}

197 198 199 200 201 202 203 204 205 206
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;
}

207
int is_vmalloc_or_module_addr(const void *x)
208 209
{
	/*
210
	 * ARM, x86-64 and sparc64 put modules in a special place,
211 212 213 214 215 216 217 218 219 220 221
	 * 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);
}

222
/*
Nick Piggin's avatar
Nick Piggin committed
223
 * Walk a vmap address to the struct page it maps.
224
 */
225
struct page *vmalloc_to_page(const void *vmalloc_addr)
226 227 228 229 230
{
	unsigned long addr = (unsigned long) vmalloc_addr;
	struct page *page = NULL;
	pgd_t *pgd = pgd_offset_k(addr);

231 232 233 234
	/*
	 * XXX we might need to change this if we add VIRTUAL_BUG_ON for
	 * architectures that do not vmalloc module space
	 */
235
	VIRTUAL_BUG_ON(!is_vmalloc_or_module_addr(vmalloc_addr));
Jiri Slaby's avatar
Jiri Slaby committed
236

237
	if (!pgd_none(*pgd)) {
Nick Piggin's avatar
Nick Piggin committed
238
		pud_t *pud = pud_offset(pgd, addr);
239
		if (!pud_none(*pud)) {
Nick Piggin's avatar
Nick Piggin committed
240
			pmd_t *pmd = pmd_offset(pud, addr);
241
			if (!pmd_none(*pmd)) {
Nick Piggin's avatar
Nick Piggin committed
242 243
				pte_t *ptep, pte;

244 245 246 247 248 249 250 251 252 253 254 255 256 257 258
				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.
 */
259
unsigned long vmalloc_to_pfn(const void *vmalloc_addr)
260 261 262 263 264
{
	return page_to_pfn(vmalloc_to_page(vmalloc_addr));
}
EXPORT_SYMBOL(vmalloc_to_pfn);

Nick Piggin's avatar
Nick Piggin committed
265 266 267 268 269 270 271 272

/*** Global kva allocator ***/

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

static DEFINE_SPINLOCK(vmap_area_lock);
273 274
/* Export for kexec only */
LIST_HEAD(vmap_area_list);
Nick Piggin's avatar
Nick Piggin committed
275 276 277 278 279 280 281 282
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;

283
static unsigned long vmap_area_pcpu_hole;
Nick Piggin's avatar
Nick Piggin committed
284 285

static struct vmap_area *__find_vmap_area(unsigned long addr)
Linus Torvalds's avatar
Linus Torvalds committed
286
{
Nick Piggin's avatar
Nick Piggin committed
287 288 289 290 291 292 293 294
	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;
295
		else if (addr >= va->va_end)
Nick Piggin's avatar
Nick Piggin committed
296 297 298 299 300 301 302 303 304 305 306 307 308 309 310
			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) {
311
		struct vmap_area *tmp_va;
Nick Piggin's avatar
Nick Piggin committed
312 313

		parent = *p;
314 315
		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
316
			p = &(*p)->rb_left;
317
		else if (va->va_end > tmp_va->va_start)
Nick Piggin's avatar
Nick Piggin committed
318 319 320 321 322 323 324 325
			p = &(*p)->rb_right;
		else
			BUG();
	}

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

326
	/* address-sort this list */
Nick Piggin's avatar
Nick Piggin committed
327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348
	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
349
	unsigned long addr;
Nick Piggin's avatar
Nick Piggin committed
350
	int purged = 0;
Nick Piggin's avatar
Nick Piggin committed
351
	struct vmap_area *first;
Nick Piggin's avatar
Nick Piggin committed
352

Nick Piggin's avatar
Nick Piggin committed
353
	BUG_ON(!size);
Nick Piggin's avatar
Nick Piggin committed
354
	BUG_ON(size & ~PAGE_MASK);
Nick Piggin's avatar
Nick Piggin committed
355
	BUG_ON(!is_power_of_2(align));
Nick Piggin's avatar
Nick Piggin committed
356 357 358 359 360 361

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

362 363 364 365 366 367
	/*
	 * 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
368 369
retry:
	spin_lock(&vmap_area_lock);
Nick Piggin's avatar
Nick Piggin committed
370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393
	/*
	 * 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);
394
		addr = ALIGN(first->va_end, align);
Nick Piggin's avatar
Nick Piggin committed
395 396
		if (addr < vstart)
			goto nocache;
397
		if (addr + size < addr)
Nick Piggin's avatar
Nick Piggin committed
398 399 400 401
			goto overflow;

	} else {
		addr = ALIGN(vstart, align);
402
		if (addr + size < addr)
Nick Piggin's avatar
Nick Piggin committed
403 404 405 406 407 408
			goto overflow;

		n = vmap_area_root.rb_node;
		first = NULL;

		while (n) {
Nick Piggin's avatar
Nick Piggin committed
409 410 411 412
			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
413 414 415 416
				if (tmp->va_start <= addr)
					break;
				n = n->rb_left;
			} else
Nick Piggin's avatar
Nick Piggin committed
417
				n = n->rb_right;
Nick Piggin's avatar
Nick Piggin committed
418
		}
Nick Piggin's avatar
Nick Piggin committed
419 420 421 422

		if (!first)
			goto found;
	}
Nick Piggin's avatar
Nick Piggin committed
423 424

	/* from the starting point, walk areas until a suitable hole is found */
425
	while (addr + size > first->va_start && addr + size <= vend) {
Nick Piggin's avatar
Nick Piggin committed
426 427
		if (addr + cached_hole_size < first->va_start)
			cached_hole_size = first->va_start - addr;
428
		addr = ALIGN(first->va_end, align);
429
		if (addr + size < addr)
Nick Piggin's avatar
Nick Piggin committed
430 431
			goto overflow;

432
		if (list_is_last(&first->list, &vmap_area_list))
Nick Piggin's avatar
Nick Piggin committed
433
			goto found;
434 435 436

		first = list_entry(first->list.next,
				struct vmap_area, list);
Nick Piggin's avatar
Nick Piggin committed
437 438
	}

Nick Piggin's avatar
Nick Piggin committed
439 440 441
found:
	if (addr + size > vend)
		goto overflow;
Nick Piggin's avatar
Nick Piggin committed
442 443 444 445 446

	va->va_start = addr;
	va->va_end = addr + size;
	va->flags = 0;
	__insert_vmap_area(va);
Nick Piggin's avatar
Nick Piggin committed
447
	free_vmap_cache = &va->rb_node;
Nick Piggin's avatar
Nick Piggin committed
448 449
	spin_unlock(&vmap_area_lock);

Nick Piggin's avatar
Nick Piggin committed
450 451 452 453
	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
454
	return va;
Nick Piggin's avatar
Nick Piggin committed
455 456 457 458 459 460 461 462 463 464 465 466 467 468

overflow:
	spin_unlock(&vmap_area_lock);
	if (!purged) {
		purge_vmap_area_lazy();
		purged = 1;
		goto retry;
	}
	if (printk_ratelimit())
		printk(KERN_WARNING
			"vmap allocation for size %lu failed: "
			"use vmalloc=<size> to increase size.\n", size);
	kfree(va);
	return ERR_PTR(-EBUSY);
Nick Piggin's avatar
Nick Piggin committed
469 470 471 472 473
}

static void __free_vmap_area(struct vmap_area *va)
{
	BUG_ON(RB_EMPTY_NODE(&va->rb_node));
Nick Piggin's avatar
Nick Piggin committed
474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489

	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
490 491 492 493
	rb_erase(&va->rb_node, &vmap_area_root);
	RB_CLEAR_NODE(&va->rb_node);
	list_del_rcu(&va->list);

494 495 496 497 498 499 500 501 502
	/*
	 * 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);

503
	kfree_rcu(va, rcu_head);
Nick Piggin's avatar
Nick Piggin committed
504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523
}

/*
 * 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);
}

524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544
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
545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571
/*
 * 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);

572 573 574
/* for per-CPU blocks */
static void purge_fragmented_blocks_allcpus(void);

575 576 577 578 579 580 581 582 583
/*
 * 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
584 585 586 587 588 589 590 591 592 593 594 595 596
/*
 * 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)
{
597
	static DEFINE_SPINLOCK(purge_lock);
Nick Piggin's avatar
Nick Piggin committed
598 599
	LIST_HEAD(valist);
	struct vmap_area *va;
600
	struct vmap_area *n_va;
Nick Piggin's avatar
Nick Piggin committed
601 602 603 604 605 606 607 608
	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) {
609
		if (!spin_trylock(&purge_lock))
Nick Piggin's avatar
Nick Piggin committed
610 611
			return;
	} else
612
		spin_lock(&purge_lock);
Nick Piggin's avatar
Nick Piggin committed
613

614 615 616
	if (sync)
		purge_fragmented_blocks_allcpus();

Nick Piggin's avatar
Nick Piggin committed
617 618 619 620 621 622 623 624 625 626 627 628 629 630 631
	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();

632
	if (nr)
Nick Piggin's avatar
Nick Piggin committed
633 634 635 636 637 638 639
		atomic_sub(nr, &vmap_lazy_nr);

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

	if (nr) {
		spin_lock(&vmap_area_lock);
640
		list_for_each_entry_safe(va, n_va, &valist, purge_list)
Nick Piggin's avatar
Nick Piggin committed
641 642 643
			__free_vmap_area(va);
		spin_unlock(&vmap_area_lock);
	}
644
	spin_unlock(&purge_lock);
Nick Piggin's avatar
Nick Piggin committed
645 646
}

Nick Piggin's avatar
Nick Piggin committed
647 648 649 650 651 652 653 654 655 656 657
/*
 * 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
658 659 660 661 662 663 664
/*
 * 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
665
	__purge_vmap_area_lazy(&start, &end, 1, 0);
Nick Piggin's avatar
Nick Piggin committed
666 667 668
}

/*
669 670 671
 * 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
672
 */
673
static void free_vmap_area_noflush(struct vmap_area *va)
Nick Piggin's avatar
Nick Piggin committed
674 675 676 677
{
	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
678
		try_purge_vmap_area_lazy();
Nick Piggin's avatar
Nick Piggin committed
679 680
}

681 682 683 684 685 686 687 688 689 690
/*
 * 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);
}

691 692 693 694 695 696 697 698 699
/*
 * 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
700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743
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() */
744 745 746 747
#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
748 749 750

#define VMAP_BLOCK_SIZE		(VMAP_BBMAP_BITS * PAGE_SIZE)

751 752
static bool vmap_initialized __read_mostly = false;

Nick Piggin's avatar
Nick Piggin committed
753 754 755 756 757 758 759 760 761 762
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;
	DECLARE_BITMAP(dirty_map, VMAP_BBMAP_BITS);
763 764
	struct list_head free_list;
	struct rcu_head rcu_head;
765
	struct list_head purge;
Nick Piggin's avatar
Nick Piggin committed
766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810
};

/* 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);
811
	if (IS_ERR(va)) {
Nick Piggin's avatar
Nick Piggin committed
812
		kfree(vb);
Julia Lawall's avatar
Julia Lawall committed
813
		return ERR_CAST(va);
Nick Piggin's avatar
Nick Piggin committed
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
	}

	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->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);
	spin_lock(&vbq->lock);
839
	list_add_rcu(&vb->free_list, &vbq->free);
Nick Piggin's avatar
Nick Piggin committed
840
	spin_unlock(&vbq->lock);
841
	put_cpu_var(vmap_block_queue);
Nick Piggin's avatar
Nick Piggin committed
842 843 844 845 846 847 848 849 850 851 852 853 854 855 856

	return 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);

857
	free_vmap_area_noflush(vb->va);
858
	kfree_rcu(vb, rcu_head);
Nick Piggin's avatar
Nick Piggin committed
859 860
}

861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902
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->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_allcpus(void)
{
	int cpu;

	for_each_possible_cpu(cpu)
		purge_fragmented_blocks(cpu);
}

Nick Piggin's avatar
Nick Piggin committed
903 904 905 906 907 908 909 910 911
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;

	BUG_ON(size & ~PAGE_MASK);
	BUG_ON(size > PAGE_SIZE*VMAP_MAX_ALLOC);
Jan Kara's avatar
Jan Kara committed
912 913 914 915 916 917 918 919
	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
920 921 922 923 924 925 926 927 928
	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);
929 930 931
		if (vb->free < 1UL << order)
			goto next;

932
		i = VMAP_BBMAP_BITS - vb->free;
933 934 935 936 937 938 939 940 941 942 943 944
		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
945 946
		spin_unlock(&vb->lock);
	}
947

948
	put_cpu_var(vmap_block_queue);
Nick Piggin's avatar
Nick Piggin committed
949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969
	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);
970 971 972

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

Nick Piggin's avatar
Nick Piggin committed
973 974 975 976 977 978 979 980 981 982
	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);

983 984
	vunmap_page_range((unsigned long)addr, (unsigned long)addr + size);

Nick Piggin's avatar
Nick Piggin committed
985
	spin_lock(&vb->lock);
986
	BUG_ON(bitmap_allocate_region(vb->dirty_map, offset >> PAGE_SHIFT, order));
987

Nick Piggin's avatar
Nick Piggin committed
988 989
	vb->dirty += 1UL << order;
	if (vb->dirty == VMAP_BBMAP_BITS) {
990
		BUG_ON(vb->free);
Nick Piggin's avatar
Nick Piggin committed
991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015
		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;

1016 1017 1018
	if (unlikely(!vmap_initialized))
		return;

Nick Piggin's avatar
Nick Piggin committed
1019 1020 1021 1022 1023 1024
	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) {
1025
			int i, j;
Nick Piggin's avatar
Nick Piggin committed
1026 1027 1028

			spin_lock(&vb->lock);
			i = find_first_bit(vb->dirty_map, VMAP_BBMAP_BITS);
1029
			if (i < VMAP_BBMAP_BITS) {
Nick Piggin's avatar
Nick Piggin committed
1030
				unsigned long s, e;
1031 1032 1033 1034

				j = find_last_bit(vb->dirty_map,
							VMAP_BBMAP_BITS);
				j = j + 1; /* need exclusive index */
Nick Piggin's avatar
Nick Piggin committed
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 1062 1063 1064 1065 1066 1067 1068 1069

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

				if (s < start)
					start = s;
				if (e > end)
					end = e;
			}
			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);
1070
	vmap_debug_free_range(addr, addr+size);
Nick Piggin's avatar
Nick Piggin committed
1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084

	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
1085 1086
 *
 * Returns: a pointer to the address that has been mapped, or %NULL on failure
Nick Piggin's avatar
Nick Piggin committed
1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116
 */
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);

1117
static struct vm_struct *vmlist __initdata;
Nicolas Pitre's avatar
Nicolas Pitre committed
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
/**
 * 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;
}

1144 1145 1146
/**
 * vm_area_register_early - register vmap area early during boot
 * @vm: vm_struct to register
1147
 * @align: requested alignment
1148 1149 1150 1151 1152 1153 1154 1155
 *
 * 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.
 */
1156
void __init vm_area_register_early(struct vm_struct *vm, size_t align)
1157 1158
{
	static size_t vm_init_off __initdata;
1159 1160 1161 1162
	unsigned long addr;

	addr = ALIGN(VMALLOC_START + vm_init_off, align);
	vm_init_off = PFN_ALIGN(addr + vm->size) - VMALLOC_START;
1163

1164
	vm->addr = (void *)addr;
1165

Nicolas Pitre's avatar
Nicolas Pitre committed
1166
	vm_area_add_early(vm);
1167 1168
}

Nick Piggin's avatar
Nick Piggin committed
1169 1170
void __init vmalloc_init(void)
{
Ivan Kokshaysky's avatar
Ivan Kokshaysky committed
1171 1172
	struct vmap_area *va;
	struct vm_struct *tmp;
Nick Piggin's avatar
Nick Piggin committed
1173 1174 1175 1176
	int i;

	for_each_possible_cpu(i) {
		struct vmap_block_queue *vbq;
1177
		struct vfree_deferred *p;
Nick Piggin's avatar
Nick Piggin committed
1178 1179 1180 1181

		vbq = &per_cpu(vmap_block_queue, i);
		spin_lock_init(&vbq->lock);
		INIT_LIST_HEAD(&vbq->free);
1182 1183 1184
		p = &per_cpu(vfree_deferred, i);
		init_llist_head(&p->list);
		INIT_WORK(&p->wq, free_work);
Nick Piggin's avatar
Nick Piggin committed
1185
	}
1186

Ivan Kokshaysky's avatar
Ivan Kokshaysky committed
1187 1188
	/* Import existing vmlist entries. */
	for (tmp = vmlist; tmp; tmp = tmp->next) {
1189
		va = kzalloc(sizeof(struct vmap_area), GFP_NOWAIT);
1190
		va->flags = VM_VM_AREA;
Ivan Kokshaysky's avatar
Ivan Kokshaysky committed
1191 1192
		va->va_start = (unsigned long)tmp->addr;
		va->va_end = va->va_start + tmp->size;
1193
		va->vm = tmp;
Ivan Kokshaysky's avatar
Ivan Kokshaysky committed
1194 1195
		__insert_vmap_area(va);
	}
1196 1197 1198

	vmap_area_pcpu_hole = VMALLOC_END;

1199
	vmap_initialized = true;
Nick Piggin's avatar
Nick Piggin committed
1200 1201
}

1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244
/**
 * 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);
}
1245
EXPORT_SYMBOL_GPL(unmap_kernel_range_noflush);
1246 1247 1248 1249 1250 1251 1252 1253 1254

/**
 * 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
1255 1256 1257
void unmap_kernel_range(unsigned long addr, unsigned long size)
{
	unsigned long end = addr + size;
1258 1259

	flush_cache_vunmap(addr, end);
Nick Piggin's avatar
Nick Piggin committed
1260 1261 1262 1263 1264 1265 1266
	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;
1267
	unsigned long end = addr + get_vm_area_size(area);
Nick Piggin's avatar
Nick Piggin committed
1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279
	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);

1280
static void setup_vmalloc_vm(struct vm_struct *vm, struct vmap_area *va,
1281
			      unsigned long flags, const void *caller)
1282
{
1283
	spin_lock(&vmap_area_lock);
1284 1285 1286 1287
	vm->flags = flags;
	vm->addr = (void *)va->va_start;
	vm->size = va->va_end - va->va_start;
	vm->caller = caller;
1288
	va->vm = vm;
1289
	va->flags |= VM_VM_AREA;
1290
	spin_unlock(&vmap_area_lock);
1291
}
1292

1293
static void clear_vm_uninitialized_flag(struct vm_struct *vm)
1294
{
1295
	/*
1296
	 * Before removing VM_UNINITIALIZED,
1297 1298 1299 1300
	 * we should make sure that vm has proper values.
	 * Pair with smp_rmb() in show_numa_info().
	 */
	smp_wmb();
1301
	vm->flags &= ~VM_UNINITIALIZED;
1302 1303
}

Nick Piggin's avatar
Nick Piggin committed
1304
static struct vm_struct *__get_vm_area_node(unsigned long size,
1305
		unsigned long align, unsigned long flags, unsigned long start,
1306
		unsigned long end, int node, gfp_t gfp_mask, const void *caller)
Nick Piggin's avatar
Nick Piggin committed
1307
{
1308
	struct vmap_area *va;
Nick Piggin's avatar
Nick Piggin committed
1309
	struct vm_struct *area;
Linus Torvalds's avatar
Linus Torvalds committed
1310

1311
	BUG_ON(in_interrupt());
1312 1313
	if (flags & VM_IOREMAP)
		align = 1ul << clamp(fls(size), PAGE_SHIFT, IOREMAP_MAX_ORDER);
Nick Piggin's avatar
Nick Piggin committed
1314

Linus Torvalds's avatar
Linus Torvalds committed
1315
	size = PAGE_ALIGN(size);
1316 1317
	if (unlikely(!size))
		return NULL;
Linus Torvalds's avatar
Linus Torvalds committed
1318

1319
	area = kzalloc_node(sizeof(*area), gfp_mask & GFP_RECLAIM_MASK, node);
Linus Torvalds's avatar
Linus Torvalds committed
1320 1321 1322 1323 1324 1325 1326 1327
	if (unlikely(!area))
		return NULL;

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

Nick Piggin's avatar
Nick Piggin committed
1328 1329 1330 1331
	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
1332 1333
	}

1334
	setup_vmalloc_vm(area, va, flags, caller);
1335

Linus Torvalds's avatar
Linus Torvalds committed
1336 1337 1338
	return area;
}

Christoph Lameter's avatar
Christoph Lameter committed
1339 1340 1341
struct vm_struct *__get_vm_area(unsigned long size, unsigned long flags,
				unsigned long start, unsigned long end)
{
David Rientjes's avatar
David Rientjes committed
1342 1343
	return __get_vm_area_node(size, 1, flags, start, end, NUMA_NO_NODE,
				  GFP_KERNEL, __builtin_return_address(0));
Christoph Lameter's avatar
Christoph Lameter committed
1344
}
1345
EXPORT_SYMBOL_GPL(__get_vm_area);
Christoph Lameter's avatar
Christoph Lameter committed
1346

1347 1348
struct vm_struct *__get_vm_area_caller(unsigned long size, unsigned long flags,
				       unsigned long start, unsigned long end,
1349
				       const void *caller)
1350
{
David Rientjes's avatar
David Rientjes committed
1351 1352
	return __get_vm_area_node(size, 1, flags, start, end, NUMA_NO_NODE,
				  GFP_KERNEL, caller);
1353 1354
}

Linus Torvalds's avatar
Linus Torvalds committed
1355
/**
Simon Arlott's avatar
Simon Arlott committed
1356
 *	get_vm_area  -  reserve a contiguous kernel virtual area
Linus Torvalds's avatar
Linus Torvalds committed
1357 1358 1359 1360 1361 1362 1363 1364 1365
 *	@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)
{
1366
	return __get_vm_area_node(size, 1, flags, VMALLOC_START, VMALLOC_END,
David Rientjes's avatar
David Rientjes committed
1367 1368
				  NUMA_NO_NODE, GFP_KERNEL,
				  __builtin_return_address(0));
1369 1370 1371
}

struct vm_struct *get_vm_area_caller(unsigned long size, unsigned long flags,
1372
				const void *caller)
1373
{