pageattr.c 30.8 KB
Newer Older
1 2
/*
 * Copyright 2002 Andi Kleen, SuSE Labs.
L
Linus Torvalds 已提交
3
 * Thanks to Ben LaHaise for precious feedback.
4
 */
L
Linus Torvalds 已提交
5
#include <linux/highmem.h>
I
Ingo Molnar 已提交
6
#include <linux/bootmem.h>
L
Linus Torvalds 已提交
7
#include <linux/module.h>
8
#include <linux/sched.h>
L
Linus Torvalds 已提交
9
#include <linux/slab.h>
10
#include <linux/mm.h>
11
#include <linux/interrupt.h>
12 13
#include <linux/seq_file.h>
#include <linux/debugfs.h>
14
#include <linux/pfn.h>
15

16
#include <asm/e820.h>
L
Linus Torvalds 已提交
17 18
#include <asm/processor.h>
#include <asm/tlbflush.h>
D
Dave Jones 已提交
19
#include <asm/sections.h>
20
#include <asm/setup.h>
21 22
#include <asm/uaccess.h>
#include <asm/pgalloc.h>
T
Thomas Gleixner 已提交
23
#include <asm/proto.h>
24
#include <asm/pat.h>
L
Linus Torvalds 已提交
25

I
Ingo Molnar 已提交
26 27 28
/*
 * The current flushing context - we pass it instead of 5 arguments:
 */
T
Thomas Gleixner 已提交
29
struct cpa_data {
30
	unsigned long	*vaddr;
T
Thomas Gleixner 已提交
31 32
	pgprot_t	mask_set;
	pgprot_t	mask_clr;
33
	int		numpages;
34
	int		flags;
T
Thomas Gleixner 已提交
35
	unsigned long	pfn;
36
	unsigned	force_split : 1;
37
	int		curpage;
38
	struct page	**pages;
T
Thomas Gleixner 已提交
39 40
};

41 42 43 44 45 46 47 48
/*
 * Serialize cpa() (for !DEBUG_PAGEALLOC which uses large identity mappings)
 * using cpa_lock. So that we don't allow any other cpu, with stale large tlb
 * entries change the page attribute in parallel to some other cpu
 * splitting a large page entry along with changing the attribute.
 */
static DEFINE_SPINLOCK(cpa_lock);

49 50
#define CPA_FLUSHTLB 1
#define CPA_ARRAY 2
51
#define CPA_PAGES_ARRAY 4
52

53
#ifdef CONFIG_PROC_FS
54 55
static unsigned long direct_pages_count[PG_LEVEL_NUM];

56
void update_page_count(int level, unsigned long pages)
57 58
{
	unsigned long flags;
59

60 61 62 63
	/* Protect against CPA */
	spin_lock_irqsave(&pgd_lock, flags);
	direct_pages_count[level] += pages;
	spin_unlock_irqrestore(&pgd_lock, flags);
64 65 66 67 68 69 70 71
}

static void split_page_count(int level)
{
	direct_pages_count[level]--;
	direct_pages_count[level - 1] += PTRS_PER_PTE;
}

72
void arch_report_meminfo(struct seq_file *m)
73
{
74
	seq_printf(m, "DirectMap4k:    %8lu kB\n",
H
Hugh Dickins 已提交
75 76
			direct_pages_count[PG_LEVEL_4K] << 2);
#if defined(CONFIG_X86_64) || defined(CONFIG_X86_PAE)
77
	seq_printf(m, "DirectMap2M:    %8lu kB\n",
H
Hugh Dickins 已提交
78 79
			direct_pages_count[PG_LEVEL_2M] << 11);
#else
80
	seq_printf(m, "DirectMap4M:    %8lu kB\n",
H
Hugh Dickins 已提交
81 82
			direct_pages_count[PG_LEVEL_2M] << 12);
#endif
83
#ifdef CONFIG_X86_64
H
Hugh Dickins 已提交
84
	if (direct_gbpages)
85
		seq_printf(m, "DirectMap1G:    %8lu kB\n",
H
Hugh Dickins 已提交
86
			direct_pages_count[PG_LEVEL_1G] << 20);
87 88
#endif
}
89 90 91
#else
static inline void split_page_count(int level) { }
#endif
92

T
Thomas Gleixner 已提交
93 94 95 96 97 98 99 100 101
#ifdef CONFIG_X86_64

static inline unsigned long highmap_start_pfn(void)
{
	return __pa(_text) >> PAGE_SHIFT;
}

static inline unsigned long highmap_end_pfn(void)
{
102
	return __pa(roundup(_brk_end, PMD_SIZE)) >> PAGE_SHIFT;
T
Thomas Gleixner 已提交
103 104 105 106
}

#endif

107 108 109 110 111 112
#ifdef CONFIG_DEBUG_PAGEALLOC
# define debug_pagealloc 1
#else
# define debug_pagealloc 0
#endif

113 114
static inline int
within(unsigned long addr, unsigned long start, unsigned long end)
I
Ingo Molnar 已提交
115
{
116 117 118
	return addr >= start && addr < end;
}

T
Thomas Gleixner 已提交
119 120 121
/*
 * Flushing functions
 */
122 123 124 125 126 127 128 129 130

/**
 * clflush_cache_range - flush a cache range with clflush
 * @addr:	virtual start address
 * @size:	number of bytes to flush
 *
 * clflush is an unordered instruction which needs fencing with mfence
 * to avoid ordering issues.
 */
I
Ingo Molnar 已提交
131
void clflush_cache_range(void *vaddr, unsigned int size)
T
Thomas Gleixner 已提交
132
{
I
Ingo Molnar 已提交
133
	void *vend = vaddr + size - 1;
T
Thomas Gleixner 已提交
134

135
	mb();
I
Ingo Molnar 已提交
136 137 138 139 140 141 142 143

	for (; vaddr < vend; vaddr += boot_cpu_data.x86_clflush_size)
		clflush(vaddr);
	/*
	 * Flush any possible final partial cacheline:
	 */
	clflush(vend);

144
	mb();
T
Thomas Gleixner 已提交
145 146
}

147
static void __cpa_flush_all(void *arg)
T
Thomas Gleixner 已提交
148
{
149 150
	unsigned long cache = (unsigned long)arg;

T
Thomas Gleixner 已提交
151 152 153 154 155 156
	/*
	 * Flush all to work around Errata in early athlons regarding
	 * large page flushing.
	 */
	__flush_tlb_all();

157
	if (cache && boot_cpu_data.x86 >= 4)
T
Thomas Gleixner 已提交
158 159 160
		wbinvd();
}

161
static void cpa_flush_all(unsigned long cache)
T
Thomas Gleixner 已提交
162 163 164
{
	BUG_ON(irqs_disabled());

165
	on_each_cpu(__cpa_flush_all, (void *) cache, 1);
T
Thomas Gleixner 已提交
166 167
}

168 169 170 171 172 173 174 175 176 177
static void __cpa_flush_range(void *arg)
{
	/*
	 * We could optimize that further and do individual per page
	 * tlb invalidates for a low number of pages. Caveat: we must
	 * flush the high aliases on 64bit as well.
	 */
	__flush_tlb_all();
}

178
static void cpa_flush_range(unsigned long start, int numpages, int cache)
179
{
I
Ingo Molnar 已提交
180 181 182
	unsigned int i, level;
	unsigned long addr;

183
	BUG_ON(irqs_disabled());
I
Ingo Molnar 已提交
184
	WARN_ON(PAGE_ALIGN(start) != start);
185

186
	on_each_cpu(__cpa_flush_range, NULL, 1);
187

188 189 190
	if (!cache)
		return;

T
Thomas Gleixner 已提交
191 192 193 194 195 196
	/*
	 * We only need to flush on one CPU,
	 * clflush is a MESI-coherent instruction that
	 * will cause all other CPUs to flush the same
	 * cachelines:
	 */
I
Ingo Molnar 已提交
197 198 199 200 201 202
	for (i = 0, addr = start; i < numpages; i++, addr += PAGE_SIZE) {
		pte_t *pte = lookup_address(addr, &level);

		/*
		 * Only flush present addresses:
		 */
203
		if (pte && (pte_val(*pte) & _PAGE_PRESENT))
I
Ingo Molnar 已提交
204 205
			clflush_cache_range((void *) addr, PAGE_SIZE);
	}
206 207
}

208 209
static void cpa_flush_array(unsigned long *start, int numpages, int cache,
			    int in_flags, struct page **pages)
210 211
{
	unsigned int i, level;
212
	unsigned long do_wbinvd = cache && numpages >= 1024; /* 4M threshold */
213 214 215

	BUG_ON(irqs_disabled());

216
	on_each_cpu(__cpa_flush_all, (void *) do_wbinvd, 1);
217

218
	if (!cache || do_wbinvd)
219 220 221 222 223 224 225 226
		return;

	/*
	 * We only need to flush on one CPU,
	 * clflush is a MESI-coherent instruction that
	 * will cause all other CPUs to flush the same
	 * cachelines:
	 */
227 228 229 230 231 232 233 234 235 236
	for (i = 0; i < numpages; i++) {
		unsigned long addr;
		pte_t *pte;

		if (in_flags & CPA_PAGES_ARRAY)
			addr = (unsigned long)page_address(pages[i]);
		else
			addr = start[i];

		pte = lookup_address(addr, &level);
237 238 239 240 241

		/*
		 * Only flush present addresses:
		 */
		if (pte && (pte_val(*pte) & _PAGE_PRESENT))
242
			clflush_cache_range((void *)addr, PAGE_SIZE);
243 244 245
	}
}

246 247 248 249 250 251
/*
 * Certain areas of memory on x86 require very specific protection flags,
 * for example the BIOS area or kernel text. Callers don't always get this
 * right (again, ioremap() on BIOS memory is not uncommon) so this function
 * checks and fixes these known static required protection bits.
 */
T
Thomas Gleixner 已提交
252 253
static inline pgprot_t static_protections(pgprot_t prot, unsigned long address,
				   unsigned long pfn)
254 255 256
{
	pgprot_t forbidden = __pgprot(0);

I
Ingo Molnar 已提交
257
	/*
258 259
	 * The BIOS area between 640k and 1Mb needs to be executable for
	 * PCI BIOS based config access (CONFIG_PCI_GOBIOS) support.
I
Ingo Molnar 已提交
260
	 */
T
Thomas Gleixner 已提交
261
	if (within(pfn, BIOS_BEGIN >> PAGE_SHIFT, BIOS_END >> PAGE_SHIFT))
262 263 264 265
		pgprot_val(forbidden) |= _PAGE_NX;

	/*
	 * The kernel text needs to be executable for obvious reasons
T
Thomas Gleixner 已提交
266 267
	 * Does not cover __inittext since that is gone later on. On
	 * 64bit we do not enforce !NX on the low mapping
268 269 270
	 */
	if (within(address, (unsigned long)_text, (unsigned long)_etext))
		pgprot_val(forbidden) |= _PAGE_NX;
271 272

	/*
T
Thomas Gleixner 已提交
273 274
	 * The .rodata section needs to be read-only. Using the pfn
	 * catches all aliases.
275
	 */
T
Thomas Gleixner 已提交
276 277
	if (within(pfn, __pa((unsigned long)__start_rodata) >> PAGE_SHIFT,
		   __pa((unsigned long)__end_rodata) >> PAGE_SHIFT))
278
		pgprot_val(forbidden) |= _PAGE_RW;
279 280

	prot = __pgprot(pgprot_val(prot) & ~pgprot_val(forbidden));
I
Ingo Molnar 已提交
281 282 283 284

	return prot;
}

T
Thomas Gleixner 已提交
285 286 287 288 289 290 291 292
/*
 * Lookup the page table entry for a virtual address. Return a pointer
 * to the entry and the level of the mapping.
 *
 * Note: We return pud and pmd either when the entry is marked large
 * or when the present bit is not set. Otherwise we would return a
 * pointer to a nonexisting mapping.
 */
293
pte_t *lookup_address(unsigned long address, unsigned int *level)
294
{
L
Linus Torvalds 已提交
295 296 297
	pgd_t *pgd = pgd_offset_k(address);
	pud_t *pud;
	pmd_t *pmd;
298

T
Thomas Gleixner 已提交
299 300
	*level = PG_LEVEL_NONE;

L
Linus Torvalds 已提交
301 302
	if (pgd_none(*pgd))
		return NULL;
I
Ingo Molnar 已提交
303

L
Linus Torvalds 已提交
304 305 306
	pud = pud_offset(pgd, address);
	if (pud_none(*pud))
		return NULL;
307 308 309 310 311

	*level = PG_LEVEL_1G;
	if (pud_large(*pud) || !pud_present(*pud))
		return (pte_t *)pud;

L
Linus Torvalds 已提交
312 313 314
	pmd = pmd_offset(pud, address);
	if (pmd_none(*pmd))
		return NULL;
T
Thomas Gleixner 已提交
315 316

	*level = PG_LEVEL_2M;
T
Thomas Gleixner 已提交
317
	if (pmd_large(*pmd) || !pmd_present(*pmd))
L
Linus Torvalds 已提交
318 319
		return (pte_t *)pmd;

T
Thomas Gleixner 已提交
320
	*level = PG_LEVEL_4K;
I
Ingo Molnar 已提交
321

322 323
	return pte_offset_kernel(pmd, address);
}
324
EXPORT_SYMBOL_GPL(lookup_address);
325

I
Ingo Molnar 已提交
326 327 328
/*
 * Set the new pmd in all the pgds we know about:
 */
I
Ingo Molnar 已提交
329
static void __set_pmd_pte(pte_t *kpte, unsigned long address, pte_t pte)
330 331 332
{
	/* change init_mm */
	set_pte_atomic(kpte, pte);
333
#ifdef CONFIG_X86_32
334
	if (!SHARED_KERNEL_PMD) {
335 336
		struct page *page;

337
		list_for_each_entry(page, &pgd_list, lru) {
338 339 340 341 342 343 344 345 346
			pgd_t *pgd;
			pud_t *pud;
			pmd_t *pmd;

			pgd = (pgd_t *)page_address(page) + pgd_index(address);
			pud = pud_offset(pgd, address);
			pmd = pmd_offset(pud, address);
			set_pte_atomic((pte_t *)pmd, pte);
		}
L
Linus Torvalds 已提交
347
	}
348
#endif
L
Linus Torvalds 已提交
349 350
}

I
Ingo Molnar 已提交
351 352 353
static int
try_preserve_large_page(pte_t *kpte, unsigned long address,
			struct cpa_data *cpa)
354
{
T
Thomas Gleixner 已提交
355
	unsigned long nextpage_addr, numpages, pmask, psize, flags, addr, pfn;
356 357
	pte_t new_pte, old_pte, *tmp;
	pgprot_t old_prot, new_prot;
358
	int i, do_split = 1;
359
	unsigned int level;
360

361 362 363
	if (cpa->force_split)
		return 1;

364 365 366 367 368 369 370 371 372 373 374
	spin_lock_irqsave(&pgd_lock, flags);
	/*
	 * Check for races, another CPU might have split this page
	 * up already:
	 */
	tmp = lookup_address(address, &level);
	if (tmp != kpte)
		goto out_unlock;

	switch (level) {
	case PG_LEVEL_2M:
375 376
		psize = PMD_PAGE_SIZE;
		pmask = PMD_PAGE_MASK;
377
		break;
378
#ifdef CONFIG_X86_64
379
	case PG_LEVEL_1G:
380 381
		psize = PUD_PAGE_SIZE;
		pmask = PUD_PAGE_MASK;
382 383
		break;
#endif
384
	default:
I
Ingo Molnar 已提交
385
		do_split = -EINVAL;
386 387 388 389 390 391 392 393 394
		goto out_unlock;
	}

	/*
	 * Calculate the number of pages, which fit into this large
	 * page starting at address:
	 */
	nextpage_addr = (address + psize) & pmask;
	numpages = (nextpage_addr - address) >> PAGE_SHIFT;
395 396
	if (numpages < cpa->numpages)
		cpa->numpages = numpages;
397 398 399 400 401 402 403 404 405

	/*
	 * We are safe now. Check whether the new pgprot is the same:
	 */
	old_pte = *kpte;
	old_prot = new_prot = pte_pgprot(old_pte);

	pgprot_val(new_prot) &= ~pgprot_val(cpa->mask_clr);
	pgprot_val(new_prot) |= pgprot_val(cpa->mask_set);
T
Thomas Gleixner 已提交
406 407 408 409 410 411 412 413 414

	/*
	 * old_pte points to the large page base address. So we need
	 * to add the offset of the virtual address:
	 */
	pfn = pte_pfn(old_pte) + ((address & (psize - 1)) >> PAGE_SHIFT);
	cpa->pfn = pfn;

	new_prot = static_protections(new_prot, address, pfn);
415

416 417 418 419 420 421
	/*
	 * We need to check the full range, whether
	 * static_protection() requires a different pgprot for one of
	 * the pages in the range we try to preserve:
	 */
	addr = address + PAGE_SIZE;
T
Thomas Gleixner 已提交
422
	pfn++;
423
	for (i = 1; i < cpa->numpages; i++, addr += PAGE_SIZE, pfn++) {
T
Thomas Gleixner 已提交
424
		pgprot_t chk_prot = static_protections(new_prot, addr, pfn);
425 426 427 428 429

		if (pgprot_val(chk_prot) != pgprot_val(new_prot))
			goto out_unlock;
	}

430 431 432 433 434
	/*
	 * If there are no changes, return. maxpages has been updated
	 * above:
	 */
	if (pgprot_val(new_prot) == pgprot_val(old_prot)) {
I
Ingo Molnar 已提交
435
		do_split = 0;
436 437 438 439 440 441 442 443 444 445 446
		goto out_unlock;
	}

	/*
	 * We need to change the attributes. Check, whether we can
	 * change the large page in one go. We request a split, when
	 * the address is not aligned and the number of pages is
	 * smaller than the number of pages in the large page. Note
	 * that we limited the number of possible pages already to
	 * the number of pages in the large page.
	 */
447
	if (address == (nextpage_addr - psize) && cpa->numpages == numpages) {
448 449 450 451 452 453
		/*
		 * The address is aligned and the number of pages
		 * covers the full page.
		 */
		new_pte = pfn_pte(pte_pfn(old_pte), canon_pgprot(new_prot));
		__set_pmd_pte(kpte, address, new_pte);
454
		cpa->flags |= CPA_FLUSHTLB;
I
Ingo Molnar 已提交
455
		do_split = 0;
456 457 458 459
	}

out_unlock:
	spin_unlock_irqrestore(&pgd_lock, flags);
I
Ingo Molnar 已提交
460

I
Ingo Molnar 已提交
461
	return do_split;
462 463
}

464
static int split_large_page(pte_t *kpte, unsigned long address)
465
{
T
Thomas Gleixner 已提交
466
	unsigned long flags, pfn, pfninc = 1;
I
Ingo Molnar 已提交
467
	unsigned int i, level;
468
	pte_t *pbase, *tmp;
I
Ingo Molnar 已提交
469
	pgprot_t ref_prot;
470 471 472 473
	struct page *base;

	if (!debug_pagealloc)
		spin_unlock(&cpa_lock);
474
	base = alloc_pages(GFP_KERNEL | __GFP_NOTRACK, 0);
475 476
	if (!debug_pagealloc)
		spin_lock(&cpa_lock);
S
Suresh Siddha 已提交
477 478
	if (!base)
		return -ENOMEM;
479

T
Thomas Gleixner 已提交
480
	spin_lock_irqsave(&pgd_lock, flags);
481 482 483 484 485
	/*
	 * Check for races, another CPU might have split this page
	 * up for us already:
	 */
	tmp = lookup_address(address, &level);
I
Ingo Molnar 已提交
486
	if (tmp != kpte)
487 488 489
		goto out_unlock;

	pbase = (pte_t *)page_address(base);
490
	paravirt_alloc_pte(&init_mm, page_to_pfn(base));
T
Thomas Gleixner 已提交
491
	ref_prot = pte_pgprot(pte_clrhuge(*kpte));
492 493 494 495 496 497 498
	/*
	 * If we ever want to utilize the PAT bit, we need to
	 * update this function to make sure it's converted from
	 * bit 12 to bit 7 when we cross from the 2MB level to
	 * the 4K level:
	 */
	WARN_ON_ONCE(pgprot_val(ref_prot) & _PAGE_PAT_LARGE);
499

500 501 502 503 504 505 506
#ifdef CONFIG_X86_64
	if (level == PG_LEVEL_1G) {
		pfninc = PMD_PAGE_SIZE >> PAGE_SHIFT;
		pgprot_val(ref_prot) |= _PAGE_PSE;
	}
#endif

507 508 509 510
	/*
	 * Get the target pfn from the original entry:
	 */
	pfn = pte_pfn(*kpte);
511
	for (i = 0; i < PTRS_PER_PTE; i++, pfn += pfninc)
512
		set_pte(&pbase[i], pfn_pte(pfn, ref_prot));
513

514
	if (address >= (unsigned long)__va(0) &&
515 516 517 518 519
		address < (unsigned long)__va(max_low_pfn_mapped << PAGE_SHIFT))
		split_page_count(level);

#ifdef CONFIG_X86_64
	if (address >= (unsigned long)__va(1UL<<32) &&
520 521
		address < (unsigned long)__va(max_pfn_mapped << PAGE_SHIFT))
		split_page_count(level);
522
#endif
523

524
	/*
525
	 * Install the new, split up pagetable.
526
	 *
527 528 529
	 * We use the standard kernel pagetable protections for the new
	 * pagetable protections, the actual ptes set above control the
	 * primary protection behavior:
530
	 */
531
	__set_pmd_pte(kpte, address, mk_pte(base, __pgprot(_KERNPG_TABLE)));
532 533 534 535 536 537 538 539 540 541 542

	/*
	 * Intel Atom errata AAH41 workaround.
	 *
	 * The real fix should be in hw or in a microcode update, but
	 * we also probabilistically try to reduce the window of having
	 * a large TLB mixed with 4K TLBs while instruction fetches are
	 * going on.
	 */
	__flush_tlb_all();

543 544 545
	base = NULL;

out_unlock:
T
Thomas Gleixner 已提交
546 547 548 549
	/*
	 * If we dropped out via the lookup_address check under
	 * pgd_lock then stick the page back into the pool:
	 */
S
Suresh Siddha 已提交
550 551
	if (base)
		__free_page(base);
I
Ingo Molnar 已提交
552
	spin_unlock_irqrestore(&pgd_lock, flags);
553 554 555 556

	return 0;
}

557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586
static int __cpa_process_fault(struct cpa_data *cpa, unsigned long vaddr,
			       int primary)
{
	/*
	 * Ignore all non primary paths.
	 */
	if (!primary)
		return 0;

	/*
	 * Ignore the NULL PTE for kernel identity mapping, as it is expected
	 * to have holes.
	 * Also set numpages to '1' indicating that we processed cpa req for
	 * one virtual address page and its pfn. TBD: numpages can be set based
	 * on the initial value and the level returned by lookup_address().
	 */
	if (within(vaddr, PAGE_OFFSET,
		   PAGE_OFFSET + (max_pfn_mapped << PAGE_SHIFT))) {
		cpa->numpages = 1;
		cpa->pfn = __pa(vaddr) >> PAGE_SHIFT;
		return 0;
	} else {
		WARN(1, KERN_WARNING "CPA: called for zero pte. "
			"vaddr = %lx cpa->vaddr = %lx\n", vaddr,
			*cpa->vaddr);

		return -EFAULT;
	}
}

T
Thomas Gleixner 已提交
587
static int __change_page_attr(struct cpa_data *cpa, int primary)
588
{
589
	unsigned long address;
590 591
	int do_split, err;
	unsigned int level;
T
Thomas Gleixner 已提交
592
	pte_t *kpte, old_pte;
L
Linus Torvalds 已提交
593

594 595 596
	if (cpa->flags & CPA_PAGES_ARRAY)
		address = (unsigned long)page_address(cpa->pages[cpa->curpage]);
	else if (cpa->flags & CPA_ARRAY)
597 598 599
		address = cpa->vaddr[cpa->curpage];
	else
		address = *cpa->vaddr;
600
repeat:
601
	kpte = lookup_address(address, &level);
L
Linus Torvalds 已提交
602
	if (!kpte)
603
		return __cpa_process_fault(cpa, address, primary);
T
Thomas Gleixner 已提交
604 605

	old_pte = *kpte;
606 607
	if (!pte_val(old_pte))
		return __cpa_process_fault(cpa, address, primary);
608

T
Thomas Gleixner 已提交
609
	if (level == PG_LEVEL_4K) {
T
Thomas Gleixner 已提交
610
		pte_t new_pte;
611
		pgprot_t new_prot = pte_pgprot(old_pte);
T
Thomas Gleixner 已提交
612
		unsigned long pfn = pte_pfn(old_pte);
I
Ingo Molnar 已提交
613

T
Thomas Gleixner 已提交
614 615
		pgprot_val(new_prot) &= ~pgprot_val(cpa->mask_clr);
		pgprot_val(new_prot) |= pgprot_val(cpa->mask_set);
I
Ingo Molnar 已提交
616

T
Thomas Gleixner 已提交
617
		new_prot = static_protections(new_prot, address, pfn);
I
Ingo Molnar 已提交
618

619 620 621 622 623
		/*
		 * We need to keep the pfn from the existing PTE,
		 * after all we're only going to change it's attributes
		 * not the memory it points to
		 */
T
Thomas Gleixner 已提交
624 625
		new_pte = pfn_pte(pfn, canon_pgprot(new_prot));
		cpa->pfn = pfn;
626 627 628 629 630
		/*
		 * Do we really change anything ?
		 */
		if (pte_val(old_pte) != pte_val(new_pte)) {
			set_pte_atomic(kpte, new_pte);
631
			cpa->flags |= CPA_FLUSHTLB;
632
		}
633
		cpa->numpages = 1;
634
		return 0;
L
Linus Torvalds 已提交
635
	}
636 637 638 639 640

	/*
	 * Check, whether we can keep the large page intact
	 * and just change the pte:
	 */
I
Ingo Molnar 已提交
641
	do_split = try_preserve_large_page(kpte, address, cpa);
642 643
	/*
	 * When the range fits into the existing large page,
644
	 * return. cp->numpages and cpa->tlbflush have been updated in
645 646
	 * try_large_page:
	 */
I
Ingo Molnar 已提交
647 648
	if (do_split <= 0)
		return do_split;
649 650 651 652

	/*
	 * We have to split the large page:
	 */
I
Ingo Molnar 已提交
653 654
	err = split_large_page(kpte, address);
	if (!err) {
655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673
		/*
	 	 * Do a global flush tlb after splitting the large page
	 	 * and before we do the actual change page attribute in the PTE.
	 	 *
	 	 * With out this, we violate the TLB application note, that says
	 	 * "The TLBs may contain both ordinary and large-page
		 *  translations for a 4-KByte range of linear addresses. This
		 *  may occur if software modifies the paging structures so that
		 *  the page size used for the address range changes. If the two
		 *  translations differ with respect to page frame or attributes
		 *  (e.g., permissions), processor behavior is undefined and may
		 *  be implementation-specific."
	 	 *
	 	 * We do this global tlb flush inside the cpa_lock, so that we
		 * don't allow any other cpu, with stale tlb entries change the
		 * page attribute in parallel, that also falls into the
		 * just split large page entry.
	 	 */
		flush_tlb_all();
I
Ingo Molnar 已提交
674 675
		goto repeat;
	}
I
Ingo Molnar 已提交
676

I
Ingo Molnar 已提交
677
	return err;
678
}
L
Linus Torvalds 已提交
679

T
Thomas Gleixner 已提交
680 681 682
static int __change_page_attr_set_clr(struct cpa_data *cpa, int checkalias);

static int cpa_process_alias(struct cpa_data *cpa)
L
Linus Torvalds 已提交
683
{
T
Thomas Gleixner 已提交
684
	struct cpa_data alias_cpa;
T
Tejun Heo 已提交
685
	unsigned long laddr = (unsigned long)__va(cpa->pfn << PAGE_SHIFT);
686
	unsigned long vaddr, remapped;
T
Tejun Heo 已提交
687
	int ret;
688

Y
Yinghai Lu 已提交
689
	if (cpa->pfn >= max_pfn_mapped)
T
Thomas Gleixner 已提交
690
		return 0;
691

692
#ifdef CONFIG_X86_64
Y
Yinghai Lu 已提交
693
	if (cpa->pfn >= max_low_pfn_mapped && cpa->pfn < (1UL<<(32-PAGE_SHIFT)))
694 695
		return 0;
#endif
696 697 698 699
	/*
	 * No need to redo, when the primary call touched the direct
	 * mapping already:
	 */
700 701 702
	if (cpa->flags & CPA_PAGES_ARRAY)
		vaddr = (unsigned long)page_address(cpa->pages[cpa->curpage]);
	else if (cpa->flags & CPA_ARRAY)
703 704 705 706 707
		vaddr = cpa->vaddr[cpa->curpage];
	else
		vaddr = *cpa->vaddr;

	if (!(within(vaddr, PAGE_OFFSET,
708
		    PAGE_OFFSET + (max_pfn_mapped << PAGE_SHIFT)))) {
709

710
		alias_cpa = *cpa;
T
Tejun Heo 已提交
711
		alias_cpa.vaddr = &laddr;
712
		alias_cpa.flags &= ~(CPA_PAGES_ARRAY | CPA_ARRAY);
713

714
		ret = __change_page_attr_set_clr(&alias_cpa, 0);
T
Tejun Heo 已提交
715 716
		if (ret)
			return ret;
717
	}
718 719

#ifdef CONFIG_X86_64
A
Arjan van de Ven 已提交
720
	/*
T
Tejun Heo 已提交
721 722
	 * If the primary call didn't touch the high mapping already
	 * and the physical address is inside the kernel map, we need
723
	 * to touch the high mapped kernel as well:
A
Arjan van de Ven 已提交
724
	 */
T
Tejun Heo 已提交
725 726 727 728 729 730 731
	if (!within(vaddr, (unsigned long)_text, _brk_end) &&
	    within(cpa->pfn, highmap_start_pfn(), highmap_end_pfn())) {
		unsigned long temp_cpa_vaddr = (cpa->pfn << PAGE_SHIFT) +
					       __START_KERNEL_map - phys_base;
		alias_cpa = *cpa;
		alias_cpa.vaddr = &temp_cpa_vaddr;
		alias_cpa.flags &= ~(CPA_PAGES_ARRAY | CPA_ARRAY);
T
Thomas Gleixner 已提交
732

T
Tejun Heo 已提交
733 734 735 736 737 738
		/*
		 * The high mapping range is imprecise, so ignore the
		 * return value.
		 */
		__change_page_attr_set_clr(&alias_cpa, 0);
	}
A
Arjan van de Ven 已提交
739
#endif
T
Tejun Heo 已提交
740

741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758
	/*
	 * If the PMD page was partially used for per-cpu remapping,
	 * the recycled area needs to be split and modified.  Because
	 * the area is always proper subset of a PMD page
	 * cpa->numpages is guaranteed to be 1 for these areas, so
	 * there's no need to loop over and check for further remaps.
	 */
	remapped = (unsigned long)pcpu_lpage_remapped((void *)laddr);
	if (remapped) {
		WARN_ON(cpa->numpages > 1);
		alias_cpa = *cpa;
		alias_cpa.vaddr = &remapped;
		alias_cpa.flags &= ~(CPA_PAGES_ARRAY | CPA_ARRAY);
		ret = __change_page_attr_set_clr(&alias_cpa, 0);
		if (ret)
			return ret;
	}

T
Tejun Heo 已提交
759
	return 0;
L
Linus Torvalds 已提交
760 761
}

T
Thomas Gleixner 已提交
762
static int __change_page_attr_set_clr(struct cpa_data *cpa, int checkalias)
763
{
764
	int ret, numpages = cpa->numpages;
765

766 767 768 769 770
	while (numpages) {
		/*
		 * Store the remaining nr of pages for the large page
		 * preservation check.
		 */
771
		cpa->numpages = numpages;
772
		/* for array changes, we can't use large page */
773
		if (cpa->flags & (CPA_ARRAY | CPA_PAGES_ARRAY))
774
			cpa->numpages = 1;
T
Thomas Gleixner 已提交
775

776 777
		if (!debug_pagealloc)
			spin_lock(&cpa_lock);
T
Thomas Gleixner 已提交
778
		ret = __change_page_attr(cpa, checkalias);
779 780
		if (!debug_pagealloc)
			spin_unlock(&cpa_lock);
781 782 783
		if (ret)
			return ret;

T
Thomas Gleixner 已提交
784 785 786 787 788 789
		if (checkalias) {
			ret = cpa_process_alias(cpa);
			if (ret)
				return ret;
		}

790 791 792 793 794
		/*
		 * Adjust the number of pages with the result of the
		 * CPA operation. Either a large page has been
		 * preserved or a single page update happened.
		 */
795 796
		BUG_ON(cpa->numpages > numpages);
		numpages -= cpa->numpages;
797
		if (cpa->flags & (CPA_PAGES_ARRAY | CPA_ARRAY))
798 799 800 801
			cpa->curpage++;
		else
			*cpa->vaddr += cpa->numpages * PAGE_SIZE;

802
	}
803 804 805
	return 0;
}

806 807 808 809 810 811
static inline int cache_attr(pgprot_t attr)
{
	return pgprot_val(attr) &
		(_PAGE_PAT | _PAGE_PAT_LARGE | _PAGE_PWT | _PAGE_PCD);
}

812
static int change_page_attr_set_clr(unsigned long *addr, int numpages,
813
				    pgprot_t mask_set, pgprot_t mask_clr,
814 815
				    int force_split, int in_flag,
				    struct page **pages)
816
{
T
Thomas Gleixner 已提交
817
	struct cpa_data cpa;
818
	int ret, cache, checkalias;
819 820 821 822 823 824 825

	/*
	 * Check, if we are requested to change a not supported
	 * feature:
	 */
	mask_set = canon_pgprot(mask_set);
	mask_clr = canon_pgprot(mask_clr);
826
	if (!pgprot_val(mask_set) && !pgprot_val(mask_clr) && !force_split)
827 828
		return 0;

829
	/* Ensure we are PAGE_SIZE aligned */
830
	if (in_flag & CPA_ARRAY) {
831 832 833 834 835 836 837
		int i;
		for (i = 0; i < numpages; i++) {
			if (addr[i] & ~PAGE_MASK) {
				addr[i] &= PAGE_MASK;
				WARN_ON_ONCE(1);
			}
		}
838 839 840 841 842 843 844 845 846 847 848 849
	} else if (!(in_flag & CPA_PAGES_ARRAY)) {
		/*
		 * in_flag of CPA_PAGES_ARRAY implies it is aligned.
		 * No need to cehck in that case
		 */
		if (*addr & ~PAGE_MASK) {
			*addr &= PAGE_MASK;
			/*
			 * People should not be passing in unaligned addresses:
			 */
			WARN_ON_ONCE(1);
		}
850 851
	}

852 853 854
	/* Must avoid aliasing mappings in the highmem code */
	kmap_flush_unused();

N
Nick Piggin 已提交
855 856
	vm_unmap_aliases();

T
Thomas Gleixner 已提交
857
	cpa.vaddr = addr;
858
	cpa.pages = pages;
T
Thomas Gleixner 已提交
859 860 861
	cpa.numpages = numpages;
	cpa.mask_set = mask_set;
	cpa.mask_clr = mask_clr;
862 863
	cpa.flags = 0;
	cpa.curpage = 0;
864
	cpa.force_split = force_split;
T
Thomas Gleixner 已提交
865

866 867
	if (in_flag & (CPA_ARRAY | CPA_PAGES_ARRAY))
		cpa.flags |= in_flag;
868

869 870 871 872
	/* No alias checking for _NX bit modifications */
	checkalias = (pgprot_val(mask_set) | pgprot_val(mask_clr)) != _PAGE_NX;

	ret = __change_page_attr_set_clr(&cpa, checkalias);
873

874 875 876
	/*
	 * Check whether we really changed something:
	 */
877
	if (!(cpa.flags & CPA_FLUSHTLB))
878
		goto out;
879

880 881 882 883 884 885
	/*
	 * No need to flush, when we did not set any of the caching
	 * attributes:
	 */
	cache = cache_attr(mask_set);

886 887 888
	/*
	 * On success we use clflush, when the CPU supports it to
	 * avoid the wbindv. If the CPU does not support it and in the
889
	 * error case we fall back to cpa_flush_all (which uses
890 891
	 * wbindv):
	 */
892
	if (!ret && cpu_has_clflush) {
893 894 895 896
		if (cpa.flags & (CPA_PAGES_ARRAY | CPA_ARRAY)) {
			cpa_flush_array(addr, numpages, cache,
					cpa.flags, pages);
		} else
897 898
			cpa_flush_range(*addr, numpages, cache);
	} else
899
		cpa_flush_all(cache);
900

901
out:
902 903 904
	return ret;
}

905 906
static inline int change_page_attr_set(unsigned long *addr, int numpages,
				       pgprot_t mask, int array)
907
{
908
	return change_page_attr_set_clr(addr, numpages, mask, __pgprot(0), 0,
909
		(array ? CPA_ARRAY : 0), NULL);
910 911
}

912 913
static inline int change_page_attr_clear(unsigned long *addr, int numpages,
					 pgprot_t mask, int array)
914
{
915
	return change_page_attr_set_clr(addr, numpages, __pgprot(0), mask, 0,
916
		(array ? CPA_ARRAY : 0), NULL);
917 918
}

919 920 921 922 923 924 925 926 927 928 929 930 931 932
static inline int cpa_set_pages_array(struct page **pages, int numpages,
				       pgprot_t mask)
{
	return change_page_attr_set_clr(NULL, numpages, mask, __pgprot(0), 0,
		CPA_PAGES_ARRAY, pages);
}

static inline int cpa_clear_pages_array(struct page **pages, int numpages,
					 pgprot_t mask)
{
	return change_page_attr_set_clr(NULL, numpages, __pgprot(0), mask, 0,
		CPA_PAGES_ARRAY, pages);
}

933
int _set_memory_uc(unsigned long addr, int numpages)
934
{
935 936 937
	/*
	 * for now UC MINUS. see comments in ioremap_nocache()
	 */
938 939
	return change_page_attr_set(&addr, numpages,
				    __pgprot(_PAGE_CACHE_UC_MINUS), 0);
940
}
941 942 943

int set_memory_uc(unsigned long addr, int numpages)
{
944 945
	int ret;

946 947 948
	/*
	 * for now UC MINUS. see comments in ioremap_nocache()
	 */
949 950 951 952 953 954 955 956 957 958
	ret = reserve_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE,
			    _PAGE_CACHE_UC_MINUS, NULL);
	if (ret)
		goto out_err;

	ret = _set_memory_uc(addr, numpages);
	if (ret)
		goto out_free;

	return 0;
959

960 961 962 963
out_free:
	free_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE);
out_err:
	return ret;
964
}
965 966
EXPORT_SYMBOL(set_memory_uc);

967 968
int set_memory_array_uc(unsigned long *addr, int addrinarray)
{
969 970 971
	int i, j;
	int ret;

972 973 974 975
	/*
	 * for now UC MINUS. see comments in ioremap_nocache()
	 */
	for (i = 0; i < addrinarray; i++) {
976 977 978 979
		ret = reserve_memtype(__pa(addr[i]), __pa(addr[i]) + PAGE_SIZE,
					_PAGE_CACHE_UC_MINUS, NULL);
		if (ret)
			goto out_free;
980 981
	}

982
	ret = change_page_attr_set(addr, addrinarray,
983
				    __pgprot(_PAGE_CACHE_UC_MINUS), 1);
984 985 986 987 988 989 990 991 992 993
	if (ret)
		goto out_free;

	return 0;

out_free:
	for (j = 0; j < i; j++)
		free_memtype(__pa(addr[j]), __pa(addr[j]) + PAGE_SIZE);

	return ret;
994 995 996
}
EXPORT_SYMBOL(set_memory_array_uc);

997 998
int _set_memory_wc(unsigned long addr, int numpages)
{
999 1000 1001 1002 1003 1004
	int ret;
	ret = change_page_attr_set(&addr, numpages,
				    __pgprot(_PAGE_CACHE_UC_MINUS), 0);

	if (!ret) {
		ret = change_page_attr_set(&addr, numpages,
1005
				    __pgprot(_PAGE_CACHE_WC), 0);
1006 1007
	}
	return ret;
1008 1009 1010 1011
}

int set_memory_wc(unsigned long addr, int numpages)
{
1012 1013
	int ret;

1014
	if (!pat_enabled)
1015 1016
		return set_memory_uc(addr, numpages);

1017 1018 1019 1020
	ret = reserve_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE,
		_PAGE_CACHE_WC, NULL);
	if (ret)
		goto out_err;
1021

1022 1023 1024 1025 1026 1027 1028 1029 1030 1031
	ret = _set_memory_wc(addr, numpages);
	if (ret)
		goto out_free;

	return 0;

out_free:
	free_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE);
out_err:
	return ret;
1032 1033 1034
}
EXPORT_SYMBOL(set_memory_wc);

1035
int _set_memory_wb(unsigned long addr, int numpages)
1036
{
1037 1038
	return change_page_attr_clear(&addr, numpages,
				      __pgprot(_PAGE_CACHE_MASK), 0);
1039
}
1040 1041 1042

int set_memory_wb(unsigned long addr, int numpages)
{
1043 1044 1045 1046 1047 1048
	int ret;

	ret = _set_memory_wb(addr, numpages);
	if (ret)
		return ret;

1049
	free_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE);
1050
	return 0;
1051
}
1052 1053
EXPORT_SYMBOL(set_memory_wb);

1054 1055 1056
int set_memory_array_wb(unsigned long *addr, int addrinarray)
{
	int i;
1057 1058 1059 1060
	int ret;

	ret = change_page_attr_clear(addr, addrinarray,
				      __pgprot(_PAGE_CACHE_MASK), 1);
1061 1062
	if (ret)
		return ret;
1063

1064 1065
	for (i = 0; i < addrinarray; i++)
		free_memtype(__pa(addr[i]), __pa(addr[i]) + PAGE_SIZE);
1066

1067
	return 0;
1068 1069 1070
}
EXPORT_SYMBOL(set_memory_array_wb);

1071 1072
int set_memory_x(unsigned long addr, int numpages)
{
1073
	return change_page_attr_clear(&addr, numpages, __pgprot(_PAGE_NX), 0);
1074 1075 1076 1077 1078
}
EXPORT_SYMBOL(set_memory_x);

int set_memory_nx(unsigned long addr, int numpages)
{
1079
	return change_page_attr_set(&addr, numpages, __pgprot(_PAGE_NX), 0);
1080 1081 1082 1083 1084
}
EXPORT_SYMBOL(set_memory_nx);

int set_memory_ro(unsigned long addr, int numpages)
{
1085
	return change_page_attr_clear(&addr, numpages, __pgprot(_PAGE_RW), 0);
1086
}
1087
EXPORT_SYMBOL_GPL(set_memory_ro);
1088 1089 1090

int set_memory_rw(unsigned long addr, int numpages)
{
1091
	return change_page_attr_set(&addr, numpages, __pgprot(_PAGE_RW), 0);
1092
}
1093
EXPORT_SYMBOL_GPL(set_memory_rw);
I
Ingo Molnar 已提交
1094 1095 1096

int set_memory_np(unsigned long addr, int numpages)
{
1097
	return change_page_attr_clear(&addr, numpages, __pgprot(_PAGE_PRESENT), 0);
I
Ingo Molnar 已提交
1098
}
1099

1100 1101
int set_memory_4k(unsigned long addr, int numpages)
{
1102
	return change_page_attr_set_clr(&addr, numpages, __pgprot(0),
1103
					__pgprot(0), 1, 0, NULL);
1104 1105
}

1106 1107 1108 1109
int set_pages_uc(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);

T
Thomas Gleixner 已提交
1110
	return set_memory_uc(addr, numpages);
1111 1112 1113
}
EXPORT_SYMBOL(set_pages_uc);

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
int set_pages_array_uc(struct page **pages, int addrinarray)
{
	unsigned long start;
	unsigned long end;
	int i;
	int free_idx;

	for (i = 0; i < addrinarray; i++) {
		start = (unsigned long)page_address(pages[i]);
		end = start + PAGE_SIZE;
		if (reserve_memtype(start, end, _PAGE_CACHE_UC_MINUS, NULL))
			goto err_out;
	}

	if (cpa_set_pages_array(pages, addrinarray,
			__pgprot(_PAGE_CACHE_UC_MINUS)) == 0) {
		return 0; /* Success */
	}
err_out:
	free_idx = i;
	for (i = 0; i < free_idx; i++) {
		start = (unsigned long)page_address(pages[i]);
		end = start + PAGE_SIZE;
		free_memtype(start, end);
	}
	return -EINVAL;
}
EXPORT_SYMBOL(set_pages_array_uc);

1143 1144 1145 1146
int set_pages_wb(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);

T
Thomas Gleixner 已提交
1147
	return set_memory_wb(addr, numpages);
1148 1149 1150
}
EXPORT_SYMBOL(set_pages_wb);

1151 1152 1153 1154 1155 1156 1157 1158 1159
int set_pages_array_wb(struct page **pages, int addrinarray)
{
	int retval;
	unsigned long start;
	unsigned long end;
	int i;

	retval = cpa_clear_pages_array(pages, addrinarray,
			__pgprot(_PAGE_CACHE_MASK));
1160 1161
	if (retval)
		return retval;
1162 1163 1164 1165 1166 1167 1168

	for (i = 0; i < addrinarray; i++) {
		start = (unsigned long)page_address(pages[i]);
		end = start + PAGE_SIZE;
		free_memtype(start, end);
	}

1169
	return 0;
1170 1171 1172
}
EXPORT_SYMBOL(set_pages_array_wb);

1173 1174 1175 1176
int set_pages_x(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);

T
Thomas Gleixner 已提交
1177
	return set_memory_x(addr, numpages);
1178 1179 1180 1181 1182 1183 1184
}
EXPORT_SYMBOL(set_pages_x);

int set_pages_nx(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);

T
Thomas Gleixner 已提交
1185
	return set_memory_nx(addr, numpages);
1186 1187 1188 1189 1190 1191 1192
}
EXPORT_SYMBOL(set_pages_nx);

int set_pages_ro(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);

T
Thomas Gleixner 已提交
1193
	return set_memory_ro(addr, numpages);
1194 1195 1196 1197 1198
}

int set_pages_rw(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);
1199

T
Thomas Gleixner 已提交
1200
	return set_memory_rw(addr, numpages);
I
Ingo Molnar 已提交
1201 1202
}

L
Linus Torvalds 已提交
1203
#ifdef CONFIG_DEBUG_PAGEALLOC
I
Ingo Molnar 已提交
1204 1205 1206

static int __set_pages_p(struct page *page, int numpages)
{
1207 1208
	unsigned long tempaddr = (unsigned long) page_address(page);
	struct cpa_data cpa = { .vaddr = &tempaddr,
T
Thomas Gleixner 已提交
1209 1210
				.numpages = numpages,
				.mask_set = __pgprot(_PAGE_PRESENT | _PAGE_RW),
1211 1212
				.mask_clr = __pgprot(0),
				.flags = 0};
1213

1214 1215 1216 1217 1218 1219 1220
	/*
	 * No alias checking needed for setting present flag. otherwise,
	 * we may need to break large pages for 64-bit kernel text
	 * mappings (this adds to complexity if we want to do this from
	 * atomic context especially). Let's keep it simple!
	 */
	return __change_page_attr_set_clr(&cpa, 0);
I
Ingo Molnar 已提交
1221 1222 1223 1224
}

static int __set_pages_np(struct page *page, int numpages)
{
1225 1226
	unsigned long tempaddr = (unsigned long) page_address(page);
	struct cpa_data cpa = { .vaddr = &tempaddr,
T
Thomas Gleixner 已提交
1227 1228
				.numpages = numpages,
				.mask_set = __pgprot(0),
1229 1230
				.mask_clr = __pgprot(_PAGE_PRESENT | _PAGE_RW),
				.flags = 0};
1231

1232 1233 1234 1235 1236 1237 1238
	/*
	 * No alias checking needed for setting not present flag. otherwise,
	 * we may need to break large pages for 64-bit kernel text
	 * mappings (this adds to complexity if we want to do this from
	 * atomic context especially). Let's keep it simple!
	 */
	return __change_page_attr_set_clr(&cpa, 0);
I
Ingo Molnar 已提交
1239 1240
}

L
Linus Torvalds 已提交
1241 1242 1243 1244
void kernel_map_pages(struct page *page, int numpages, int enable)
{
	if (PageHighMem(page))
		return;
1245
	if (!enable) {
1246 1247
		debug_check_no_locks_freed(page_address(page),
					   numpages * PAGE_SIZE);
1248
	}
1249

1250 1251 1252 1253 1254 1255
	/*
	 * If page allocator is not up yet then do not call c_p_a():
	 */
	if (!debug_pagealloc_enabled)
		return;

1256
	/*
I
Ingo Molnar 已提交
1257
	 * The return value is ignored as the calls cannot fail.
1258 1259
	 * Large pages for identity mappings are not used at boot time
	 * and hence no memory allocations during large page split.
L
Linus Torvalds 已提交
1260
	 */
I
Ingo Molnar 已提交
1261 1262 1263 1264
	if (enable)
		__set_pages_p(page, numpages);
	else
		__set_pages_np(page, numpages);
1265 1266

	/*
1267 1268
	 * We should perform an IPI and flush all tlbs,
	 * but that can deadlock->flush only current cpu:
L
Linus Torvalds 已提交
1269 1270
	 */
	__flush_tlb_all();
1271 1272
}

1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289
#ifdef CONFIG_HIBERNATION

bool kernel_page_present(struct page *page)
{
	unsigned int level;
	pte_t *pte;

	if (PageHighMem(page))
		return false;

	pte = lookup_address((unsigned long)page_address(page), &level);
	return (pte_val(*pte) & _PAGE_PRESENT);
}

#endif /* CONFIG_HIBERNATION */

#endif /* CONFIG_DEBUG_PAGEALLOC */
1290 1291 1292 1293 1294 1295 1296 1297

/*
 * The testcases use internal knowledge of the implementation that shouldn't
 * be exposed to the rest of the kernel. Include these directly here.
 */
#ifdef CONFIG_CPA_DEBUG
#include "pageattr-test.c"
#endif