pageattr.c 33.4 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 9
#include <linux/sched.h>
#include <linux/mm.h>
10
#include <linux/interrupt.h>
11 12
#include <linux/seq_file.h>
#include <linux/debugfs.h>
13
#include <linux/pfn.h>
14
#include <linux/percpu.h>
15
#include <linux/gfp.h>
16
#include <linux/pci.h>
17

18
#include <asm/e820.h>
L
Linus Torvalds 已提交
19 20
#include <asm/processor.h>
#include <asm/tlbflush.h>
D
Dave Jones 已提交
21
#include <asm/sections.h>
22
#include <asm/setup.h>
23 24
#include <asm/uaccess.h>
#include <asm/pgalloc.h>
T
Thomas Gleixner 已提交
25
#include <asm/proto.h>
26
#include <asm/pat.h>
L
Linus Torvalds 已提交
27

I
Ingo Molnar 已提交
28 29 30
/*
 * The current flushing context - we pass it instead of 5 arguments:
 */
T
Thomas Gleixner 已提交
31
struct cpa_data {
32
	unsigned long	*vaddr;
T
Thomas Gleixner 已提交
33 34
	pgprot_t	mask_set;
	pgprot_t	mask_clr;
35
	int		numpages;
36
	int		flags;
T
Thomas Gleixner 已提交
37
	unsigned long	pfn;
38
	unsigned	force_split : 1;
39
	int		curpage;
40
	struct page	**pages;
T
Thomas Gleixner 已提交
41 42
};

43 44 45 46 47 48 49 50
/*
 * 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);

51 52
#define CPA_FLUSHTLB 1
#define CPA_ARRAY 2
53
#define CPA_PAGES_ARRAY 4
54

55
#ifdef CONFIG_PROC_FS
56 57
static unsigned long direct_pages_count[PG_LEVEL_NUM];

58
void update_page_count(int level, unsigned long pages)
59 60
{
	/* Protect against CPA */
A
Andrea Arcangeli 已提交
61
	spin_lock(&pgd_lock);
62
	direct_pages_count[level] += pages;
A
Andrea Arcangeli 已提交
63
	spin_unlock(&pgd_lock);
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

/**
 * clflush_cache_range - flush a cache range with clflush
125
 * @vaddr:	virtual start address
126 127 128 129 130
 * @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
EXPORT_SYMBOL_GPL(clflush_cache_range);
T
Thomas Gleixner 已提交
147

148
static void __cpa_flush_all(void *arg)
T
Thomas Gleixner 已提交
149
{
150 151
	unsigned long cache = (unsigned long)arg;

T
Thomas Gleixner 已提交
152 153 154 155 156 157
	/*
	 * Flush all to work around Errata in early athlons regarding
	 * large page flushing.
	 */
	__flush_tlb_all();

158
	if (cache && boot_cpu_data.x86 >= 4)
T
Thomas Gleixner 已提交
159 160 161
		wbinvd();
}

162
static void cpa_flush_all(unsigned long cache)
T
Thomas Gleixner 已提交
163 164 165
{
	BUG_ON(irqs_disabled());

166
	on_each_cpu(__cpa_flush_all, (void *) cache, 1);
T
Thomas Gleixner 已提交
167 168
}

169 170 171 172 173 174 175 176 177 178
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();
}

179
static void cpa_flush_range(unsigned long start, int numpages, int cache)
180
{
I
Ingo Molnar 已提交
181 182 183
	unsigned int i, level;
	unsigned long addr;

184
	BUG_ON(irqs_disabled());
I
Ingo Molnar 已提交
185
	WARN_ON(PAGE_ALIGN(start) != start);
186

187
	on_each_cpu(__cpa_flush_range, NULL, 1);
188

189 190 191
	if (!cache)
		return;

T
Thomas Gleixner 已提交
192 193 194 195 196 197
	/*
	 * 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
198 199 200 201 202 203
	for (i = 0, addr = start; i < numpages; i++, addr += PAGE_SIZE) {
		pte_t *pte = lookup_address(addr, &level);

		/*
		 * Only flush present addresses:
		 */
204
		if (pte && (pte_val(*pte) & _PAGE_PRESENT))
I
Ingo Molnar 已提交
205 206
			clflush_cache_range((void *) addr, PAGE_SIZE);
	}
207 208
}

209 210
static void cpa_flush_array(unsigned long *start, int numpages, int cache,
			    int in_flags, struct page **pages)
211 212
{
	unsigned int i, level;
213
	unsigned long do_wbinvd = cache && numpages >= 1024; /* 4M threshold */
214 215 216

	BUG_ON(irqs_disabled());

217
	on_each_cpu(__cpa_flush_all, (void *) do_wbinvd, 1);
218

219
	if (!cache || do_wbinvd)
220 221 222 223 224 225 226 227
		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:
	 */
228 229 230 231 232 233 234 235 236 237
	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);
238 239 240 241 242

		/*
		 * Only flush present addresses:
		 */
		if (pte && (pte_val(*pte) & _PAGE_PRESENT))
243
			clflush_cache_range((void *)addr, PAGE_SIZE);
244 245 246
	}
}

247 248 249 250 251 252
/*
 * 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
253 254
static inline pgprot_t static_protections(pgprot_t prot, unsigned long address,
				   unsigned long pfn)
255 256 257
{
	pgprot_t forbidden = __pgprot(0);

I
Ingo Molnar 已提交
258
	/*
259 260
	 * 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
261
	 */
262 263
#ifdef CONFIG_PCI_BIOS
	if (pcibios_enabled && within(pfn, BIOS_BEGIN >> PAGE_SHIFT, BIOS_END >> PAGE_SHIFT))
264
		pgprot_val(forbidden) |= _PAGE_NX;
265
#endif
266 267 268

	/*
	 * The kernel text needs to be executable for obvious reasons
T
Thomas Gleixner 已提交
269 270
	 * Does not cover __inittext since that is gone later on. On
	 * 64bit we do not enforce !NX on the low mapping
271 272 273
	 */
	if (within(address, (unsigned long)_text, (unsigned long)_etext))
		pgprot_val(forbidden) |= _PAGE_NX;
274 275

	/*
T
Thomas Gleixner 已提交
276 277
	 * The .rodata section needs to be read-only. Using the pfn
	 * catches all aliases.
278
	 */
T
Thomas Gleixner 已提交
279 280
	if (within(pfn, __pa((unsigned long)__start_rodata) >> PAGE_SHIFT,
		   __pa((unsigned long)__end_rodata) >> PAGE_SHIFT))
281
		pgprot_val(forbidden) |= _PAGE_RW;
282

283
#if defined(CONFIG_X86_64) && defined(CONFIG_DEBUG_RODATA)
284
	/*
285 286 287 288
	 * Once the kernel maps the text as RO (kernel_set_to_readonly is set),
	 * kernel text mappings for the large page aligned text, rodata sections
	 * will be always read-only. For the kernel identity mappings covering
	 * the holes caused by this alignment can be anything that user asks.
289 290 291 292
	 *
	 * This will preserve the large page mappings for kernel text/data
	 * at no extra cost.
	 */
293 294
	if (kernel_set_to_readonly &&
	    within(address, (unsigned long)_text,
295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312
		   (unsigned long)__end_rodata_hpage_align)) {
		unsigned int level;

		/*
		 * Don't enforce the !RW mapping for the kernel text mapping,
		 * if the current mapping is already using small page mapping.
		 * No need to work hard to preserve large page mappings in this
		 * case.
		 *
		 * This also fixes the Linux Xen paravirt guest boot failure
		 * (because of unexpected read-only mappings for kernel identity
		 * mappings). In this paravirt guest case, the kernel text
		 * mapping and the kernel identity mapping share the same
		 * page-table pages. Thus we can't really use different
		 * protections for the kernel text and identity mappings. Also,
		 * these shared mappings are made of small page mappings.
		 * Thus this don't enforce !RW mapping for small page kernel
		 * text mapping logic will help Linux Xen parvirt guest boot
L
Lucas De Marchi 已提交
313
		 * as well.
314 315 316 317
		 */
		if (lookup_address(address, &level) && (level != PG_LEVEL_4K))
			pgprot_val(forbidden) |= _PAGE_RW;
	}
318 319
#endif

320
	prot = __pgprot(pgprot_val(prot) & ~pgprot_val(forbidden));
I
Ingo Molnar 已提交
321 322 323 324

	return prot;
}

T
Thomas Gleixner 已提交
325 326 327 328 329 330 331 332
/*
 * 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.
 */
333
pte_t *lookup_address(unsigned long address, unsigned int *level)
334
{
L
Linus Torvalds 已提交
335 336 337
	pgd_t *pgd = pgd_offset_k(address);
	pud_t *pud;
	pmd_t *pmd;
338

T
Thomas Gleixner 已提交
339 340
	*level = PG_LEVEL_NONE;

L
Linus Torvalds 已提交
341 342
	if (pgd_none(*pgd))
		return NULL;
I
Ingo Molnar 已提交
343

L
Linus Torvalds 已提交
344 345 346
	pud = pud_offset(pgd, address);
	if (pud_none(*pud))
		return NULL;
347 348 349 350 351

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

L
Linus Torvalds 已提交
352 353 354
	pmd = pmd_offset(pud, address);
	if (pmd_none(*pmd))
		return NULL;
T
Thomas Gleixner 已提交
355 356

	*level = PG_LEVEL_2M;
T
Thomas Gleixner 已提交
357
	if (pmd_large(*pmd) || !pmd_present(*pmd))
L
Linus Torvalds 已提交
358 359
		return (pte_t *)pmd;

T
Thomas Gleixner 已提交
360
	*level = PG_LEVEL_4K;
I
Ingo Molnar 已提交
361

362 363
	return pte_offset_kernel(pmd, address);
}
364
EXPORT_SYMBOL_GPL(lookup_address);
365

I
Ingo Molnar 已提交
366 367 368
/*
 * Set the new pmd in all the pgds we know about:
 */
I
Ingo Molnar 已提交
369
static void __set_pmd_pte(pte_t *kpte, unsigned long address, pte_t pte)
370 371 372
{
	/* change init_mm */
	set_pte_atomic(kpte, pte);
373
#ifdef CONFIG_X86_32
374
	if (!SHARED_KERNEL_PMD) {
375 376
		struct page *page;

377
		list_for_each_entry(page, &pgd_list, lru) {
378 379 380 381 382 383 384 385 386
			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
387
	}
388
#endif
L
Linus Torvalds 已提交
389 390
}

I
Ingo Molnar 已提交
391 392 393
static int
try_preserve_large_page(pte_t *kpte, unsigned long address,
			struct cpa_data *cpa)
394
{
A
Andrea Arcangeli 已提交
395
	unsigned long nextpage_addr, numpages, pmask, psize, addr, pfn;
396
	pte_t new_pte, old_pte, *tmp;
397
	pgprot_t old_prot, new_prot, req_prot;
398
	int i, do_split = 1;
399
	unsigned int level;
400

401 402 403
	if (cpa->force_split)
		return 1;

A
Andrea Arcangeli 已提交
404
	spin_lock(&pgd_lock);
405 406 407 408 409 410 411 412 413 414
	/*
	 * 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:
415 416
		psize = PMD_PAGE_SIZE;
		pmask = PMD_PAGE_MASK;
417
		break;
418
#ifdef CONFIG_X86_64
419
	case PG_LEVEL_1G:
420 421
		psize = PUD_PAGE_SIZE;
		pmask = PUD_PAGE_MASK;
422 423
		break;
#endif
424
	default:
I
Ingo Molnar 已提交
425
		do_split = -EINVAL;
426 427 428 429 430 431 432 433 434
		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;
435 436
	if (numpages < cpa->numpages)
		cpa->numpages = numpages;
437 438 439 440 441

	/*
	 * We are safe now. Check whether the new pgprot is the same:
	 */
	old_pte = *kpte;
442
	old_prot = new_prot = req_prot = pte_pgprot(old_pte);
443

444 445
	pgprot_val(req_prot) &= ~pgprot_val(cpa->mask_clr);
	pgprot_val(req_prot) |= pgprot_val(cpa->mask_set);
T
Thomas Gleixner 已提交
446 447 448 449 450 451 452 453

	/*
	 * 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;

454
	new_prot = static_protections(req_prot, address, pfn);
455

456 457 458 459 460
	/*
	 * 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:
	 */
461 462 463 464
	addr = address & pmask;
	pfn = pte_pfn(old_pte);
	for (i = 0; i < (psize >> PAGE_SHIFT); i++, addr += PAGE_SIZE, pfn++) {
		pgprot_t chk_prot = static_protections(req_prot, addr, pfn);
465 466 467 468 469

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

470 471 472 473 474
	/*
	 * 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
475
		do_split = 0;
476 477 478 479 480 481 482 483 484 485 486
		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.
	 */
487
	if (address == (address & pmask) && cpa->numpages == (psize >> PAGE_SHIFT)) {
488 489 490 491 492 493
		/*
		 * 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);
494
		cpa->flags |= CPA_FLUSHTLB;
I
Ingo Molnar 已提交
495
		do_split = 0;
496 497 498
	}

out_unlock:
A
Andrea Arcangeli 已提交
499
	spin_unlock(&pgd_lock);
I
Ingo Molnar 已提交
500

I
Ingo Molnar 已提交
501
	return do_split;
502 503
}

504
static int split_large_page(pte_t *kpte, unsigned long address)
505
{
A
Andrea Arcangeli 已提交
506
	unsigned long pfn, pfninc = 1;
I
Ingo Molnar 已提交
507
	unsigned int i, level;
508
	pte_t *pbase, *tmp;
I
Ingo Molnar 已提交
509
	pgprot_t ref_prot;
510 511 512 513
	struct page *base;

	if (!debug_pagealloc)
		spin_unlock(&cpa_lock);
514
	base = alloc_pages(GFP_KERNEL | __GFP_NOTRACK, 0);
515 516
	if (!debug_pagealloc)
		spin_lock(&cpa_lock);
S
Suresh Siddha 已提交
517 518
	if (!base)
		return -ENOMEM;
519

A
Andrea Arcangeli 已提交
520
	spin_lock(&pgd_lock);
521 522 523 524 525
	/*
	 * 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
526
	if (tmp != kpte)
527 528 529
		goto out_unlock;

	pbase = (pte_t *)page_address(base);
530
	paravirt_alloc_pte(&init_mm, page_to_pfn(base));
T
Thomas Gleixner 已提交
531
	ref_prot = pte_pgprot(pte_clrhuge(*kpte));
532 533 534 535 536 537 538
	/*
	 * 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);
539

540 541 542 543 544 545 546
#ifdef CONFIG_X86_64
	if (level == PG_LEVEL_1G) {
		pfninc = PMD_PAGE_SIZE >> PAGE_SHIFT;
		pgprot_val(ref_prot) |= _PAGE_PSE;
	}
#endif

547 548 549 550
	/*
	 * Get the target pfn from the original entry:
	 */
	pfn = pte_pfn(*kpte);
551
	for (i = 0; i < PTRS_PER_PTE; i++, pfn += pfninc)
552
		set_pte(&pbase[i], pfn_pte(pfn, ref_prot));
553

554
	if (address >= (unsigned long)__va(0) &&
555 556 557 558 559
		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) &&
560 561
		address < (unsigned long)__va(max_pfn_mapped << PAGE_SHIFT))
		split_page_count(level);
562
#endif
563

564
	/*
565
	 * Install the new, split up pagetable.
566
	 *
567 568 569
	 * We use the standard kernel pagetable protections for the new
	 * pagetable protections, the actual ptes set above control the
	 * primary protection behavior:
570
	 */
571
	__set_pmd_pte(kpte, address, mk_pte(base, __pgprot(_KERNPG_TABLE)));
572 573 574 575 576 577 578 579 580 581 582

	/*
	 * 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();

583 584 585
	base = NULL;

out_unlock:
T
Thomas Gleixner 已提交
586 587 588 589
	/*
	 * 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
590 591
	if (base)
		__free_page(base);
A
Andrea Arcangeli 已提交
592
	spin_unlock(&pgd_lock);
593 594 595 596

	return 0;
}

597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626
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
627
static int __change_page_attr(struct cpa_data *cpa, int primary)
628
{
629
	unsigned long address;
630 631
	int do_split, err;
	unsigned int level;
T
Thomas Gleixner 已提交
632
	pte_t *kpte, old_pte;
L
Linus Torvalds 已提交
633

634 635 636 637 638 639
	if (cpa->flags & CPA_PAGES_ARRAY) {
		struct page *page = cpa->pages[cpa->curpage];
		if (unlikely(PageHighMem(page)))
			return 0;
		address = (unsigned long)page_address(page);
	} else if (cpa->flags & CPA_ARRAY)
640 641 642
		address = cpa->vaddr[cpa->curpage];
	else
		address = *cpa->vaddr;
643
repeat:
644
	kpte = lookup_address(address, &level);
L
Linus Torvalds 已提交
645
	if (!kpte)
646
		return __cpa_process_fault(cpa, address, primary);
T
Thomas Gleixner 已提交
647 648

	old_pte = *kpte;
649 650
	if (!pte_val(old_pte))
		return __cpa_process_fault(cpa, address, primary);
651

T
Thomas Gleixner 已提交
652
	if (level == PG_LEVEL_4K) {
T
Thomas Gleixner 已提交
653
		pte_t new_pte;
654
		pgprot_t new_prot = pte_pgprot(old_pte);
T
Thomas Gleixner 已提交
655
		unsigned long pfn = pte_pfn(old_pte);
I
Ingo Molnar 已提交
656

T
Thomas Gleixner 已提交
657 658
		pgprot_val(new_prot) &= ~pgprot_val(cpa->mask_clr);
		pgprot_val(new_prot) |= pgprot_val(cpa->mask_set);
I
Ingo Molnar 已提交
659

T
Thomas Gleixner 已提交
660
		new_prot = static_protections(new_prot, address, pfn);
I
Ingo Molnar 已提交
661

662 663 664 665 666
		/*
		 * 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
667 668
		new_pte = pfn_pte(pfn, canon_pgprot(new_prot));
		cpa->pfn = pfn;
669 670 671 672 673
		/*
		 * Do we really change anything ?
		 */
		if (pte_val(old_pte) != pte_val(new_pte)) {
			set_pte_atomic(kpte, new_pte);
674
			cpa->flags |= CPA_FLUSHTLB;
675
		}
676
		cpa->numpages = 1;
677
		return 0;
L
Linus Torvalds 已提交
678
	}
679 680 681 682 683

	/*
	 * Check, whether we can keep the large page intact
	 * and just change the pte:
	 */
I
Ingo Molnar 已提交
684
	do_split = try_preserve_large_page(kpte, address, cpa);
685 686
	/*
	 * When the range fits into the existing large page,
687
	 * return. cp->numpages and cpa->tlbflush have been updated in
688 689
	 * try_large_page:
	 */
I
Ingo Molnar 已提交
690 691
	if (do_split <= 0)
		return do_split;
692 693 694 695

	/*
	 * We have to split the large page:
	 */
I
Ingo Molnar 已提交
696 697
	err = split_large_page(kpte, address);
	if (!err) {
698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716
		/*
	 	 * 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
717 718
		goto repeat;
	}
I
Ingo Molnar 已提交
719

I
Ingo Molnar 已提交
720
	return err;
721
}
L
Linus Torvalds 已提交
722

T
Thomas Gleixner 已提交
723 724 725
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
726
{
T
Thomas Gleixner 已提交
727
	struct cpa_data alias_cpa;
T
Tejun Heo 已提交
728
	unsigned long laddr = (unsigned long)__va(cpa->pfn << PAGE_SHIFT);
729
	unsigned long vaddr;
T
Tejun Heo 已提交
730
	int ret;
731

Y
Yinghai Lu 已提交
732
	if (cpa->pfn >= max_pfn_mapped)
T
Thomas Gleixner 已提交
733
		return 0;
734

735
#ifdef CONFIG_X86_64
Y
Yinghai Lu 已提交
736
	if (cpa->pfn >= max_low_pfn_mapped && cpa->pfn < (1UL<<(32-PAGE_SHIFT)))
737 738
		return 0;
#endif
739 740 741 742
	/*
	 * No need to redo, when the primary call touched the direct
	 * mapping already:
	 */
743 744 745 746 747 748
	if (cpa->flags & CPA_PAGES_ARRAY) {
		struct page *page = cpa->pages[cpa->curpage];
		if (unlikely(PageHighMem(page)))
			return 0;
		vaddr = (unsigned long)page_address(page);
	} else if (cpa->flags & CPA_ARRAY)
749 750 751 752 753
		vaddr = cpa->vaddr[cpa->curpage];
	else
		vaddr = *cpa->vaddr;

	if (!(within(vaddr, PAGE_OFFSET,
754
		    PAGE_OFFSET + (max_pfn_mapped << PAGE_SHIFT)))) {
755

756
		alias_cpa = *cpa;
T
Tejun Heo 已提交
757
		alias_cpa.vaddr = &laddr;
758
		alias_cpa.flags &= ~(CPA_PAGES_ARRAY | CPA_ARRAY);
759

760
		ret = __change_page_attr_set_clr(&alias_cpa, 0);
T
Tejun Heo 已提交
761 762
		if (ret)
			return ret;
763
	}
764 765

#ifdef CONFIG_X86_64
A
Arjan van de Ven 已提交
766
	/*
T
Tejun Heo 已提交
767 768
	 * If the primary call didn't touch the high mapping already
	 * and the physical address is inside the kernel map, we need
769
	 * to touch the high mapped kernel as well:
A
Arjan van de Ven 已提交
770
	 */
T
Tejun Heo 已提交
771 772 773 774 775 776 777
	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
778

T
Tejun Heo 已提交
779 780 781 782 783 784
		/*
		 * 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
785
#endif
T
Tejun Heo 已提交
786 787

	return 0;
L
Linus Torvalds 已提交
788 789
}

T
Thomas Gleixner 已提交
790
static int __change_page_attr_set_clr(struct cpa_data *cpa, int checkalias)
791
{
792
	int ret, numpages = cpa->numpages;
793

794 795 796 797 798
	while (numpages) {
		/*
		 * Store the remaining nr of pages for the large page
		 * preservation check.
		 */
799
		cpa->numpages = numpages;
800
		/* for array changes, we can't use large page */
801
		if (cpa->flags & (CPA_ARRAY | CPA_PAGES_ARRAY))
802
			cpa->numpages = 1;
T
Thomas Gleixner 已提交
803

804 805
		if (!debug_pagealloc)
			spin_lock(&cpa_lock);
T
Thomas Gleixner 已提交
806
		ret = __change_page_attr(cpa, checkalias);
807 808
		if (!debug_pagealloc)
			spin_unlock(&cpa_lock);
809 810 811
		if (ret)
			return ret;

T
Thomas Gleixner 已提交
812 813 814 815 816 817
		if (checkalias) {
			ret = cpa_process_alias(cpa);
			if (ret)
				return ret;
		}

818 819 820 821 822
		/*
		 * 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.
		 */
823 824
		BUG_ON(cpa->numpages > numpages);
		numpages -= cpa->numpages;
825
		if (cpa->flags & (CPA_PAGES_ARRAY | CPA_ARRAY))
826 827 828 829
			cpa->curpage++;
		else
			*cpa->vaddr += cpa->numpages * PAGE_SIZE;

830
	}
831 832 833
	return 0;
}

834 835 836 837 838 839
static inline int cache_attr(pgprot_t attr)
{
	return pgprot_val(attr) &
		(_PAGE_PAT | _PAGE_PAT_LARGE | _PAGE_PWT | _PAGE_PCD);
}

840
static int change_page_attr_set_clr(unsigned long *addr, int numpages,
841
				    pgprot_t mask_set, pgprot_t mask_clr,
842 843
				    int force_split, int in_flag,
				    struct page **pages)
844
{
T
Thomas Gleixner 已提交
845
	struct cpa_data cpa;
846
	int ret, cache, checkalias;
847
	unsigned long baddr = 0;
848 849 850 851 852 853 854

	/*
	 * Check, if we are requested to change a not supported
	 * feature:
	 */
	mask_set = canon_pgprot(mask_set);
	mask_clr = canon_pgprot(mask_clr);
855
	if (!pgprot_val(mask_set) && !pgprot_val(mask_clr) && !force_split)
856 857
		return 0;

858
	/* Ensure we are PAGE_SIZE aligned */
859
	if (in_flag & CPA_ARRAY) {
860 861 862 863 864 865 866
		int i;
		for (i = 0; i < numpages; i++) {
			if (addr[i] & ~PAGE_MASK) {
				addr[i] &= PAGE_MASK;
				WARN_ON_ONCE(1);
			}
		}
867 868 869 870 871 872 873 874 875 876 877 878
	} 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);
		}
879 880 881 882 883
		/*
		 * Save address for cache flush. *addr is modified in the call
		 * to __change_page_attr_set_clr() below.
		 */
		baddr = *addr;
884 885
	}

886 887 888
	/* Must avoid aliasing mappings in the highmem code */
	kmap_flush_unused();

N
Nick Piggin 已提交
889 890
	vm_unmap_aliases();

T
Thomas Gleixner 已提交
891
	cpa.vaddr = addr;
892
	cpa.pages = pages;
T
Thomas Gleixner 已提交
893 894 895
	cpa.numpages = numpages;
	cpa.mask_set = mask_set;
	cpa.mask_clr = mask_clr;
896 897
	cpa.flags = 0;
	cpa.curpage = 0;
898
	cpa.force_split = force_split;
T
Thomas Gleixner 已提交
899

900 901
	if (in_flag & (CPA_ARRAY | CPA_PAGES_ARRAY))
		cpa.flags |= in_flag;
902

903 904 905 906
	/* 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);
907

908 909 910
	/*
	 * Check whether we really changed something:
	 */
911
	if (!(cpa.flags & CPA_FLUSHTLB))
912
		goto out;
913

914 915 916 917 918 919
	/*
	 * No need to flush, when we did not set any of the caching
	 * attributes:
	 */
	cache = cache_attr(mask_set);

920 921
	/*
	 * On success we use clflush, when the CPU supports it to
922 923 924
	 * avoid the wbindv. If the CPU does not support it, in the
	 * error case, and during early boot (for EFI) we fall back
	 * to cpa_flush_all (which uses wbinvd):
925
	 */
926 927 928
	if (early_boot_irqs_disabled)
		__cpa_flush_all((void *)(long)cache);
	else if (!ret && cpu_has_clflush) {
929 930 931 932
		if (cpa.flags & (CPA_PAGES_ARRAY | CPA_ARRAY)) {
			cpa_flush_array(addr, numpages, cache,
					cpa.flags, pages);
		} else
933
			cpa_flush_range(baddr, numpages, cache);
934
	} else
935
		cpa_flush_all(cache);
936

937
out:
938 939 940
	return ret;
}

941 942
static inline int change_page_attr_set(unsigned long *addr, int numpages,
				       pgprot_t mask, int array)
943
{
944
	return change_page_attr_set_clr(addr, numpages, mask, __pgprot(0), 0,
945
		(array ? CPA_ARRAY : 0), NULL);
946 947
}

948 949
static inline int change_page_attr_clear(unsigned long *addr, int numpages,
					 pgprot_t mask, int array)
950
{
951
	return change_page_attr_set_clr(addr, numpages, __pgprot(0), mask, 0,
952
		(array ? CPA_ARRAY : 0), NULL);
953 954
}

955 956 957 958 959 960 961 962 963 964 965 966 967 968
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);
}

969
int _set_memory_uc(unsigned long addr, int numpages)
970
{
971 972 973
	/*
	 * for now UC MINUS. see comments in ioremap_nocache()
	 */
974 975
	return change_page_attr_set(&addr, numpages,
				    __pgprot(_PAGE_CACHE_UC_MINUS), 0);
976
}
977 978 979

int set_memory_uc(unsigned long addr, int numpages)
{
980 981
	int ret;

982 983 984
	/*
	 * for now UC MINUS. see comments in ioremap_nocache()
	 */
985 986 987 988 989 990 991 992 993 994
	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;
995

996 997 998 999
out_free:
	free_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE);
out_err:
	return ret;
1000
}
1001 1002
EXPORT_SYMBOL(set_memory_uc);

1003
static int _set_memory_array(unsigned long *addr, int addrinarray,
1004
		unsigned long new_type)
1005
{
1006 1007 1008
	int i, j;
	int ret;

1009 1010 1011 1012
	/*
	 * for now UC MINUS. see comments in ioremap_nocache()
	 */
	for (i = 0; i < addrinarray; i++) {
1013
		ret = reserve_memtype(__pa(addr[i]), __pa(addr[i]) + PAGE_SIZE,
1014
					new_type, NULL);
1015 1016
		if (ret)
			goto out_free;
1017 1018
	}

1019
	ret = change_page_attr_set(addr, addrinarray,
1020
				    __pgprot(_PAGE_CACHE_UC_MINUS), 1);
1021 1022 1023 1024 1025 1026

	if (!ret && new_type == _PAGE_CACHE_WC)
		ret = change_page_attr_set_clr(addr, addrinarray,
					       __pgprot(_PAGE_CACHE_WC),
					       __pgprot(_PAGE_CACHE_MASK),
					       0, CPA_ARRAY, NULL);
1027 1028 1029 1030 1031 1032 1033 1034 1035 1036
	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;
1037
}
1038 1039 1040 1041 1042

int set_memory_array_uc(unsigned long *addr, int addrinarray)
{
	return _set_memory_array(addr, addrinarray, _PAGE_CACHE_UC_MINUS);
}
1043 1044
EXPORT_SYMBOL(set_memory_array_uc);

1045 1046 1047 1048 1049 1050
int set_memory_array_wc(unsigned long *addr, int addrinarray)
{
	return _set_memory_array(addr, addrinarray, _PAGE_CACHE_WC);
}
EXPORT_SYMBOL(set_memory_array_wc);

1051 1052
int _set_memory_wc(unsigned long addr, int numpages)
{
1053
	int ret;
1054 1055
	unsigned long addr_copy = addr;

1056 1057 1058
	ret = change_page_attr_set(&addr, numpages,
				    __pgprot(_PAGE_CACHE_UC_MINUS), 0);
	if (!ret) {
1059 1060 1061 1062
		ret = change_page_attr_set_clr(&addr_copy, numpages,
					       __pgprot(_PAGE_CACHE_WC),
					       __pgprot(_PAGE_CACHE_MASK),
					       0, 0, NULL);
1063 1064
	}
	return ret;
1065 1066 1067 1068
}

int set_memory_wc(unsigned long addr, int numpages)
{
1069 1070
	int ret;

1071
	if (!pat_enabled)
1072 1073
		return set_memory_uc(addr, numpages);

1074 1075 1076 1077
	ret = reserve_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE,
		_PAGE_CACHE_WC, NULL);
	if (ret)
		goto out_err;
1078

1079 1080 1081 1082 1083 1084 1085 1086 1087 1088
	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;
1089 1090 1091
}
EXPORT_SYMBOL(set_memory_wc);

1092
int _set_memory_wb(unsigned long addr, int numpages)
1093
{
1094 1095
	return change_page_attr_clear(&addr, numpages,
				      __pgprot(_PAGE_CACHE_MASK), 0);
1096
}
1097 1098 1099

int set_memory_wb(unsigned long addr, int numpages)
{
1100 1101 1102 1103 1104 1105
	int ret;

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

1106
	free_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE);
1107
	return 0;
1108
}
1109 1110
EXPORT_SYMBOL(set_memory_wb);

1111 1112 1113
int set_memory_array_wb(unsigned long *addr, int addrinarray)
{
	int i;
1114 1115 1116 1117
	int ret;

	ret = change_page_attr_clear(addr, addrinarray,
				      __pgprot(_PAGE_CACHE_MASK), 1);
1118 1119
	if (ret)
		return ret;
1120

1121 1122
	for (i = 0; i < addrinarray; i++)
		free_memtype(__pa(addr[i]), __pa(addr[i]) + PAGE_SIZE);
1123

1124
	return 0;
1125 1126 1127
}
EXPORT_SYMBOL(set_memory_array_wb);

1128 1129
int set_memory_x(unsigned long addr, int numpages)
{
1130 1131 1132
	if (!(__supported_pte_mask & _PAGE_NX))
		return 0;

1133
	return change_page_attr_clear(&addr, numpages, __pgprot(_PAGE_NX), 0);
1134 1135 1136 1137 1138
}
EXPORT_SYMBOL(set_memory_x);

int set_memory_nx(unsigned long addr, int numpages)
{
1139 1140 1141
	if (!(__supported_pte_mask & _PAGE_NX))
		return 0;

1142
	return change_page_attr_set(&addr, numpages, __pgprot(_PAGE_NX), 0);
1143 1144 1145 1146 1147
}
EXPORT_SYMBOL(set_memory_nx);

int set_memory_ro(unsigned long addr, int numpages)
{
1148
	return change_page_attr_clear(&addr, numpages, __pgprot(_PAGE_RW), 0);
1149
}
1150
EXPORT_SYMBOL_GPL(set_memory_ro);
1151 1152 1153

int set_memory_rw(unsigned long addr, int numpages)
{
1154
	return change_page_attr_set(&addr, numpages, __pgprot(_PAGE_RW), 0);
1155
}
1156
EXPORT_SYMBOL_GPL(set_memory_rw);
I
Ingo Molnar 已提交
1157 1158 1159

int set_memory_np(unsigned long addr, int numpages)
{
1160
	return change_page_attr_clear(&addr, numpages, __pgprot(_PAGE_PRESENT), 0);
I
Ingo Molnar 已提交
1161
}
1162

1163 1164
int set_memory_4k(unsigned long addr, int numpages)
{
1165
	return change_page_attr_set_clr(&addr, numpages, __pgprot(0),
1166
					__pgprot(0), 1, 0, NULL);
1167 1168
}

1169 1170 1171 1172
int set_pages_uc(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);

T
Thomas Gleixner 已提交
1173
	return set_memory_uc(addr, numpages);
1174 1175 1176
}
EXPORT_SYMBOL(set_pages_uc);

1177 1178
static int _set_pages_array(struct page **pages, int addrinarray,
		unsigned long new_type)
1179 1180 1181 1182 1183
{
	unsigned long start;
	unsigned long end;
	int i;
	int free_idx;
1184
	int ret;
1185 1186

	for (i = 0; i < addrinarray; i++) {
1187 1188 1189
		if (PageHighMem(pages[i]))
			continue;
		start = page_to_pfn(pages[i]) << PAGE_SHIFT;
1190
		end = start + PAGE_SIZE;
1191
		if (reserve_memtype(start, end, new_type, NULL))
1192 1193 1194
			goto err_out;
	}

1195 1196 1197 1198 1199 1200 1201 1202 1203 1204
	ret = cpa_set_pages_array(pages, addrinarray,
			__pgprot(_PAGE_CACHE_UC_MINUS));
	if (!ret && new_type == _PAGE_CACHE_WC)
		ret = change_page_attr_set_clr(NULL, addrinarray,
					       __pgprot(_PAGE_CACHE_WC),
					       __pgprot(_PAGE_CACHE_MASK),
					       0, CPA_PAGES_ARRAY, pages);
	if (ret)
		goto err_out;
	return 0; /* Success */
1205 1206 1207
err_out:
	free_idx = i;
	for (i = 0; i < free_idx; i++) {
1208 1209 1210
		if (PageHighMem(pages[i]))
			continue;
		start = page_to_pfn(pages[i]) << PAGE_SHIFT;
1211 1212 1213 1214 1215
		end = start + PAGE_SIZE;
		free_memtype(start, end);
	}
	return -EINVAL;
}
1216 1217 1218 1219 1220

int set_pages_array_uc(struct page **pages, int addrinarray)
{
	return _set_pages_array(pages, addrinarray, _PAGE_CACHE_UC_MINUS);
}
1221 1222
EXPORT_SYMBOL(set_pages_array_uc);

1223 1224 1225 1226 1227 1228
int set_pages_array_wc(struct page **pages, int addrinarray)
{
	return _set_pages_array(pages, addrinarray, _PAGE_CACHE_WC);
}
EXPORT_SYMBOL(set_pages_array_wc);

1229 1230 1231 1232
int set_pages_wb(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);

T
Thomas Gleixner 已提交
1233
	return set_memory_wb(addr, numpages);
1234 1235 1236
}
EXPORT_SYMBOL(set_pages_wb);

1237 1238 1239 1240 1241 1242 1243 1244 1245
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));
1246 1247
	if (retval)
		return retval;
1248 1249

	for (i = 0; i < addrinarray; i++) {
1250 1251 1252
		if (PageHighMem(pages[i]))
			continue;
		start = page_to_pfn(pages[i]) << PAGE_SHIFT;
1253 1254 1255 1256
		end = start + PAGE_SIZE;
		free_memtype(start, end);
	}

1257
	return 0;
1258 1259 1260
}
EXPORT_SYMBOL(set_pages_array_wb);

1261 1262 1263 1264
int set_pages_x(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);

T
Thomas Gleixner 已提交
1265
	return set_memory_x(addr, numpages);
1266 1267 1268 1269 1270 1271 1272
}
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
1273
	return set_memory_nx(addr, numpages);
1274 1275 1276 1277 1278 1279 1280
}
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
1281
	return set_memory_ro(addr, numpages);
1282 1283 1284 1285 1286
}

int set_pages_rw(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);
1287

T
Thomas Gleixner 已提交
1288
	return set_memory_rw(addr, numpages);
I
Ingo Molnar 已提交
1289 1290
}

L
Linus Torvalds 已提交
1291
#ifdef CONFIG_DEBUG_PAGEALLOC
I
Ingo Molnar 已提交
1292 1293 1294

static int __set_pages_p(struct page *page, int numpages)
{
1295 1296
	unsigned long tempaddr = (unsigned long) page_address(page);
	struct cpa_data cpa = { .vaddr = &tempaddr,
T
Thomas Gleixner 已提交
1297 1298
				.numpages = numpages,
				.mask_set = __pgprot(_PAGE_PRESENT | _PAGE_RW),
1299 1300
				.mask_clr = __pgprot(0),
				.flags = 0};
1301

1302 1303 1304 1305 1306 1307 1308
	/*
	 * 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
1309 1310 1311 1312
}

static int __set_pages_np(struct page *page, int numpages)
{
1313 1314
	unsigned long tempaddr = (unsigned long) page_address(page);
	struct cpa_data cpa = { .vaddr = &tempaddr,
T
Thomas Gleixner 已提交
1315 1316
				.numpages = numpages,
				.mask_set = __pgprot(0),
1317 1318
				.mask_clr = __pgprot(_PAGE_PRESENT | _PAGE_RW),
				.flags = 0};
1319

1320 1321 1322 1323 1324 1325 1326
	/*
	 * 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
1327 1328
}

L
Linus Torvalds 已提交
1329 1330 1331 1332
void kernel_map_pages(struct page *page, int numpages, int enable)
{
	if (PageHighMem(page))
		return;
1333
	if (!enable) {
1334 1335
		debug_check_no_locks_freed(page_address(page),
					   numpages * PAGE_SIZE);
1336
	}
1337

1338
	/*
I
Ingo Molnar 已提交
1339
	 * The return value is ignored as the calls cannot fail.
1340 1341
	 * 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
1342
	 */
I
Ingo Molnar 已提交
1343 1344 1345 1346
	if (enable)
		__set_pages_p(page, numpages);
	else
		__set_pages_np(page, numpages);
1347 1348

	/*
1349 1350
	 * We should perform an IPI and flush all tlbs,
	 * but that can deadlock->flush only current cpu:
L
Linus Torvalds 已提交
1351 1352
	 */
	__flush_tlb_all();
1353 1354
}

1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371
#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 */
1372 1373 1374 1375 1376 1377 1378 1379

/*
 * 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