pageattr.c 35.4 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
#ifdef CONFIG_X86_64

static inline unsigned long highmap_start_pfn(void)
{
97
	return __pa_symbol(_text) >> PAGE_SHIFT;
T
Thomas Gleixner 已提交
98 99 100 101
}

static inline unsigned long highmap_end_pfn(void)
{
102
	return __pa_symbol(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
	 */
279 280
	if (within(pfn, __pa_symbol(__start_rodata) >> PAGE_SHIFT,
		   __pa_symbol(__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

366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396
/*
 * This is necessary because __pa() does not work on some
 * kinds of memory, like vmalloc() or the alloc_remap()
 * areas on 32-bit NUMA systems.  The percpu areas can
 * end up in this kind of memory, for instance.
 *
 * This could be optimized, but it is only intended to be
 * used at inititalization time, and keeping it
 * unoptimized should increase the testing coverage for
 * the more obscure platforms.
 */
phys_addr_t slow_virt_to_phys(void *__virt_addr)
{
	unsigned long virt_addr = (unsigned long)__virt_addr;
	phys_addr_t phys_addr;
	unsigned long offset;
	enum pg_level level;
	unsigned long psize;
	unsigned long pmask;
	pte_t *pte;

	pte = lookup_address(virt_addr, &level);
	BUG_ON(!pte);
	psize = page_level_size(level);
	pmask = page_level_mask(level);
	offset = virt_addr & ~pmask;
	phys_addr = pte_pfn(*pte) << PAGE_SHIFT;
	return (phys_addr | offset);
}
EXPORT_SYMBOL_GPL(slow_virt_to_phys);

I
Ingo Molnar 已提交
397 398 399
/*
 * Set the new pmd in all the pgds we know about:
 */
I
Ingo Molnar 已提交
400
static void __set_pmd_pte(pte_t *kpte, unsigned long address, pte_t pte)
401 402 403
{
	/* change init_mm */
	set_pte_atomic(kpte, pte);
404
#ifdef CONFIG_X86_32
405
	if (!SHARED_KERNEL_PMD) {
406 407
		struct page *page;

408
		list_for_each_entry(page, &pgd_list, lru) {
409 410 411 412 413 414 415 416 417
			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
418
	}
419
#endif
L
Linus Torvalds 已提交
420 421
}

I
Ingo Molnar 已提交
422 423 424
static int
try_preserve_large_page(pte_t *kpte, unsigned long address,
			struct cpa_data *cpa)
425
{
A
Andrea Arcangeli 已提交
426
	unsigned long nextpage_addr, numpages, pmask, psize, addr, pfn;
427
	pte_t new_pte, old_pte, *tmp;
428
	pgprot_t old_prot, new_prot, req_prot;
429
	int i, do_split = 1;
430
	enum pg_level level;
431

432 433 434
	if (cpa->force_split)
		return 1;

A
Andrea Arcangeli 已提交
435
	spin_lock(&pgd_lock);
436 437 438 439 440 441 442 443 444 445
	/*
	 * 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:
446
#ifdef CONFIG_X86_64
447
	case PG_LEVEL_1G:
448
#endif
449 450 451
		psize = page_level_size(level);
		pmask = page_level_mask(level);
		break;
452
	default:
I
Ingo Molnar 已提交
453
		do_split = -EINVAL;
454 455 456 457 458 459 460 461 462
		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;
463 464
	if (numpages < cpa->numpages)
		cpa->numpages = numpages;
465 466 467 468 469

	/*
	 * We are safe now. Check whether the new pgprot is the same:
	 */
	old_pte = *kpte;
470
	old_prot = new_prot = req_prot = pte_pgprot(old_pte);
471

472 473
	pgprot_val(req_prot) &= ~pgprot_val(cpa->mask_clr);
	pgprot_val(req_prot) |= pgprot_val(cpa->mask_set);
T
Thomas Gleixner 已提交
474

475 476 477 478 479 480 481 482 483 484 485 486 487
	/*
	 * Set the PSE and GLOBAL flags only if the PRESENT flag is
	 * set otherwise pmd_present/pmd_huge will return true even on
	 * a non present pmd. The canon_pgprot will clear _PAGE_GLOBAL
	 * for the ancient hardware that doesn't support it.
	 */
	if (pgprot_val(new_prot) & _PAGE_PRESENT)
		pgprot_val(new_prot) |= _PAGE_PSE | _PAGE_GLOBAL;
	else
		pgprot_val(new_prot) &= ~(_PAGE_PSE | _PAGE_GLOBAL);

	new_prot = canon_pgprot(new_prot);

T
Thomas Gleixner 已提交
488 489 490 491 492 493 494
	/*
	 * 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;

495
	new_prot = static_protections(req_prot, address, pfn);
496

497 498 499 500 501
	/*
	 * 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:
	 */
502 503 504 505
	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);
506 507 508 509 510

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

511 512 513 514 515
	/*
	 * 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
516
		do_split = 0;
517 518 519 520 521 522 523 524 525 526 527
		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.
	 */
528
	if (address == (address & pmask) && cpa->numpages == (psize >> PAGE_SHIFT)) {
529 530 531 532
		/*
		 * The address is aligned and the number of pages
		 * covers the full page.
		 */
533
		new_pte = pfn_pte(pte_pfn(old_pte), new_prot);
534
		__set_pmd_pte(kpte, address, new_pte);
535
		cpa->flags |= CPA_FLUSHTLB;
I
Ingo Molnar 已提交
536
		do_split = 0;
537 538 539
	}

out_unlock:
A
Andrea Arcangeli 已提交
540
	spin_unlock(&pgd_lock);
I
Ingo Molnar 已提交
541

I
Ingo Molnar 已提交
542
	return do_split;
543 544
}

545
int __split_large_page(pte_t *kpte, unsigned long address, pte_t *pbase)
546
{
A
Andrea Arcangeli 已提交
547
	unsigned long pfn, pfninc = 1;
I
Ingo Molnar 已提交
548
	unsigned int i, level;
549
	pte_t *tmp;
I
Ingo Molnar 已提交
550
	pgprot_t ref_prot;
551
	struct page *base = virt_to_page(pbase);
552

A
Andrea Arcangeli 已提交
553
	spin_lock(&pgd_lock);
554 555 556 557 558
	/*
	 * Check for races, another CPU might have split this page
	 * up for us already:
	 */
	tmp = lookup_address(address, &level);
559 560 561 562
	if (tmp != kpte) {
		spin_unlock(&pgd_lock);
		return 1;
	}
563

564
	paravirt_alloc_pte(&init_mm, page_to_pfn(base));
T
Thomas Gleixner 已提交
565
	ref_prot = pte_pgprot(pte_clrhuge(*kpte));
566 567 568 569 570 571 572
	/*
	 * 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);
573

574 575 576
#ifdef CONFIG_X86_64
	if (level == PG_LEVEL_1G) {
		pfninc = PMD_PAGE_SIZE >> PAGE_SHIFT;
577 578 579 580 581 582 583 584 585
		/*
		 * Set the PSE flags only if the PRESENT flag is set
		 * otherwise pmd_present/pmd_huge will return true
		 * even on a non present pmd.
		 */
		if (pgprot_val(ref_prot) & _PAGE_PRESENT)
			pgprot_val(ref_prot) |= _PAGE_PSE;
		else
			pgprot_val(ref_prot) &= ~_PAGE_PSE;
586 587 588
	}
#endif

589 590 591 592 593 594 595 596 597 598 599
	/*
	 * Set the GLOBAL flags only if the PRESENT flag is set
	 * otherwise pmd/pte_present will return true even on a non
	 * present pmd/pte. The canon_pgprot will clear _PAGE_GLOBAL
	 * for the ancient hardware that doesn't support it.
	 */
	if (pgprot_val(ref_prot) & _PAGE_PRESENT)
		pgprot_val(ref_prot) |= _PAGE_GLOBAL;
	else
		pgprot_val(ref_prot) &= ~_PAGE_GLOBAL;

600 601 602 603
	/*
	 * Get the target pfn from the original entry:
	 */
	pfn = pte_pfn(*kpte);
604
	for (i = 0; i < PTRS_PER_PTE; i++, pfn += pfninc)
605
		set_pte(&pbase[i], pfn_pte(pfn, canon_pgprot(ref_prot)));
606

607 608
	if (pfn_range_is_mapped(PFN_DOWN(__pa(address)),
				PFN_DOWN(__pa(address)) + 1))
609 610
		split_page_count(level);

611
	/*
612
	 * Install the new, split up pagetable.
613
	 *
614 615 616
	 * We use the standard kernel pagetable protections for the new
	 * pagetable protections, the actual ptes set above control the
	 * primary protection behavior:
617
	 */
618
	__set_pmd_pte(kpte, address, mk_pte(base, __pgprot(_KERNPG_TABLE)));
619 620 621 622 623 624 625 626 627 628

	/*
	 * 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();
629
	spin_unlock(&pgd_lock);
630

631 632
	return 0;
}
633

634 635 636 637 638 639 640 641 642 643 644 645 646 647 648
static int split_large_page(pte_t *kpte, unsigned long address)
{
	pte_t *pbase;
	struct page *base;

	if (!debug_pagealloc)
		spin_unlock(&cpa_lock);
	base = alloc_pages(GFP_KERNEL | __GFP_NOTRACK, 0);
	if (!debug_pagealloc)
		spin_lock(&cpa_lock);
	if (!base)
		return -ENOMEM;

	pbase = (pte_t *)page_address(base);
	if (__split_large_page(kpte, address, pbase))
S
Suresh Siddha 已提交
649
		__free_page(base);
650 651 652 653

	return 0;
}

654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683
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
684
static int __change_page_attr(struct cpa_data *cpa, int primary)
685
{
686
	unsigned long address;
687 688
	int do_split, err;
	unsigned int level;
T
Thomas Gleixner 已提交
689
	pte_t *kpte, old_pte;
L
Linus Torvalds 已提交
690

691 692 693 694 695 696
	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)
697 698 699
		address = cpa->vaddr[cpa->curpage];
	else
		address = *cpa->vaddr;
700
repeat:
701
	kpte = lookup_address(address, &level);
L
Linus Torvalds 已提交
702
	if (!kpte)
703
		return __cpa_process_fault(cpa, address, primary);
T
Thomas Gleixner 已提交
704 705

	old_pte = *kpte;
706 707
	if (!pte_val(old_pte))
		return __cpa_process_fault(cpa, address, primary);
708

T
Thomas Gleixner 已提交
709
	if (level == PG_LEVEL_4K) {
T
Thomas Gleixner 已提交
710
		pte_t new_pte;
711
		pgprot_t new_prot = pte_pgprot(old_pte);
T
Thomas Gleixner 已提交
712
		unsigned long pfn = pte_pfn(old_pte);
I
Ingo Molnar 已提交
713

T
Thomas Gleixner 已提交
714 715
		pgprot_val(new_prot) &= ~pgprot_val(cpa->mask_clr);
		pgprot_val(new_prot) |= pgprot_val(cpa->mask_set);
I
Ingo Molnar 已提交
716

T
Thomas Gleixner 已提交
717
		new_prot = static_protections(new_prot, address, pfn);
I
Ingo Molnar 已提交
718

719 720 721 722 723 724 725 726 727 728 729 730
		/*
		 * Set the GLOBAL flags only if the PRESENT flag is
		 * set otherwise pte_present will return true even on
		 * a non present pte. The canon_pgprot will clear
		 * _PAGE_GLOBAL for the ancient hardware that doesn't
		 * support it.
		 */
		if (pgprot_val(new_prot) & _PAGE_PRESENT)
			pgprot_val(new_prot) |= _PAGE_GLOBAL;
		else
			pgprot_val(new_prot) &= ~_PAGE_GLOBAL;

731 732 733 734 735
		/*
		 * 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
736 737
		new_pte = pfn_pte(pfn, canon_pgprot(new_prot));
		cpa->pfn = pfn;
738 739 740 741 742
		/*
		 * Do we really change anything ?
		 */
		if (pte_val(old_pte) != pte_val(new_pte)) {
			set_pte_atomic(kpte, new_pte);
743
			cpa->flags |= CPA_FLUSHTLB;
744
		}
745
		cpa->numpages = 1;
746
		return 0;
L
Linus Torvalds 已提交
747
	}
748 749 750 751 752

	/*
	 * Check, whether we can keep the large page intact
	 * and just change the pte:
	 */
I
Ingo Molnar 已提交
753
	do_split = try_preserve_large_page(kpte, address, cpa);
754 755
	/*
	 * When the range fits into the existing large page,
756
	 * return. cp->numpages and cpa->tlbflush have been updated in
757 758
	 * try_large_page:
	 */
I
Ingo Molnar 已提交
759 760
	if (do_split <= 0)
		return do_split;
761 762 763 764

	/*
	 * We have to split the large page:
	 */
I
Ingo Molnar 已提交
765 766
	err = split_large_page(kpte, address);
	if (!err) {
767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785
		/*
	 	 * 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
786 787
		goto repeat;
	}
I
Ingo Molnar 已提交
788

I
Ingo Molnar 已提交
789
	return err;
790
}
L
Linus Torvalds 已提交
791

T
Thomas Gleixner 已提交
792 793 794
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
795
{
T
Thomas Gleixner 已提交
796
	struct cpa_data alias_cpa;
T
Tejun Heo 已提交
797
	unsigned long laddr = (unsigned long)__va(cpa->pfn << PAGE_SHIFT);
798
	unsigned long vaddr;
T
Tejun Heo 已提交
799
	int ret;
800

801
	if (!pfn_range_is_mapped(cpa->pfn, cpa->pfn + 1))
T
Thomas Gleixner 已提交
802
		return 0;
803

804 805 806 807
	/*
	 * No need to redo, when the primary call touched the direct
	 * mapping already:
	 */
808 809 810 811 812 813
	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)
814 815 816 817 818
		vaddr = cpa->vaddr[cpa->curpage];
	else
		vaddr = *cpa->vaddr;

	if (!(within(vaddr, PAGE_OFFSET,
819
		    PAGE_OFFSET + (max_pfn_mapped << PAGE_SHIFT)))) {
820

821
		alias_cpa = *cpa;
T
Tejun Heo 已提交
822
		alias_cpa.vaddr = &laddr;
823
		alias_cpa.flags &= ~(CPA_PAGES_ARRAY | CPA_ARRAY);
824

825
		ret = __change_page_attr_set_clr(&alias_cpa, 0);
T
Tejun Heo 已提交
826 827
		if (ret)
			return ret;
828
	}
829 830

#ifdef CONFIG_X86_64
A
Arjan van de Ven 已提交
831
	/*
T
Tejun Heo 已提交
832 833
	 * If the primary call didn't touch the high mapping already
	 * and the physical address is inside the kernel map, we need
834
	 * to touch the high mapped kernel as well:
A
Arjan van de Ven 已提交
835
	 */
T
Tejun Heo 已提交
836 837 838 839 840 841 842
	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
843

T
Tejun Heo 已提交
844 845 846 847 848 849
		/*
		 * 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
850
#endif
T
Tejun Heo 已提交
851 852

	return 0;
L
Linus Torvalds 已提交
853 854
}

T
Thomas Gleixner 已提交
855
static int __change_page_attr_set_clr(struct cpa_data *cpa, int checkalias)
856
{
857
	int ret, numpages = cpa->numpages;
858

859 860 861 862 863
	while (numpages) {
		/*
		 * Store the remaining nr of pages for the large page
		 * preservation check.
		 */
864
		cpa->numpages = numpages;
865
		/* for array changes, we can't use large page */
866
		if (cpa->flags & (CPA_ARRAY | CPA_PAGES_ARRAY))
867
			cpa->numpages = 1;
T
Thomas Gleixner 已提交
868

869 870
		if (!debug_pagealloc)
			spin_lock(&cpa_lock);
T
Thomas Gleixner 已提交
871
		ret = __change_page_attr(cpa, checkalias);
872 873
		if (!debug_pagealloc)
			spin_unlock(&cpa_lock);
874 875 876
		if (ret)
			return ret;

T
Thomas Gleixner 已提交
877 878 879 880 881 882
		if (checkalias) {
			ret = cpa_process_alias(cpa);
			if (ret)
				return ret;
		}

883 884 885 886 887
		/*
		 * 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.
		 */
888 889
		BUG_ON(cpa->numpages > numpages);
		numpages -= cpa->numpages;
890
		if (cpa->flags & (CPA_PAGES_ARRAY | CPA_ARRAY))
891 892 893 894
			cpa->curpage++;
		else
			*cpa->vaddr += cpa->numpages * PAGE_SIZE;

895
	}
896 897 898
	return 0;
}

899 900 901 902 903 904
static inline int cache_attr(pgprot_t attr)
{
	return pgprot_val(attr) &
		(_PAGE_PAT | _PAGE_PAT_LARGE | _PAGE_PWT | _PAGE_PCD);
}

905
static int change_page_attr_set_clr(unsigned long *addr, int numpages,
906
				    pgprot_t mask_set, pgprot_t mask_clr,
907 908
				    int force_split, int in_flag,
				    struct page **pages)
909
{
T
Thomas Gleixner 已提交
910
	struct cpa_data cpa;
911
	int ret, cache, checkalias;
912
	unsigned long baddr = 0;
913 914 915 916 917 918 919

	/*
	 * Check, if we are requested to change a not supported
	 * feature:
	 */
	mask_set = canon_pgprot(mask_set);
	mask_clr = canon_pgprot(mask_clr);
920
	if (!pgprot_val(mask_set) && !pgprot_val(mask_clr) && !force_split)
921 922
		return 0;

923
	/* Ensure we are PAGE_SIZE aligned */
924
	if (in_flag & CPA_ARRAY) {
925 926 927 928 929 930 931
		int i;
		for (i = 0; i < numpages; i++) {
			if (addr[i] & ~PAGE_MASK) {
				addr[i] &= PAGE_MASK;
				WARN_ON_ONCE(1);
			}
		}
932 933 934 935 936 937 938 939 940 941 942 943
	} 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);
		}
944 945 946 947 948
		/*
		 * Save address for cache flush. *addr is modified in the call
		 * to __change_page_attr_set_clr() below.
		 */
		baddr = *addr;
949 950
	}

951 952 953
	/* Must avoid aliasing mappings in the highmem code */
	kmap_flush_unused();

N
Nick Piggin 已提交
954 955
	vm_unmap_aliases();

T
Thomas Gleixner 已提交
956
	cpa.vaddr = addr;
957
	cpa.pages = pages;
T
Thomas Gleixner 已提交
958 959 960
	cpa.numpages = numpages;
	cpa.mask_set = mask_set;
	cpa.mask_clr = mask_clr;
961 962
	cpa.flags = 0;
	cpa.curpage = 0;
963
	cpa.force_split = force_split;
T
Thomas Gleixner 已提交
964

965 966
	if (in_flag & (CPA_ARRAY | CPA_PAGES_ARRAY))
		cpa.flags |= in_flag;
967

968 969 970 971
	/* 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);
972

973 974 975
	/*
	 * Check whether we really changed something:
	 */
976
	if (!(cpa.flags & CPA_FLUSHTLB))
977
		goto out;
978

979 980 981 982 983 984
	/*
	 * No need to flush, when we did not set any of the caching
	 * attributes:
	 */
	cache = cache_attr(mask_set);

985 986
	/*
	 * On success we use clflush, when the CPU supports it to
987 988 989
	 * avoid the wbindv. If the CPU does not support it and in the
	 * error case we fall back to cpa_flush_all (which uses
	 * wbindv):
990
	 */
991
	if (!ret && cpu_has_clflush) {
992 993 994 995
		if (cpa.flags & (CPA_PAGES_ARRAY | CPA_ARRAY)) {
			cpa_flush_array(addr, numpages, cache,
					cpa.flags, pages);
		} else
996
			cpa_flush_range(baddr, numpages, cache);
997
	} else
998
		cpa_flush_all(cache);
999

1000
out:
1001 1002 1003
	return ret;
}

1004 1005
static inline int change_page_attr_set(unsigned long *addr, int numpages,
				       pgprot_t mask, int array)
1006
{
1007
	return change_page_attr_set_clr(addr, numpages, mask, __pgprot(0), 0,
1008
		(array ? CPA_ARRAY : 0), NULL);
1009 1010
}

1011 1012
static inline int change_page_attr_clear(unsigned long *addr, int numpages,
					 pgprot_t mask, int array)
1013
{
1014
	return change_page_attr_set_clr(addr, numpages, __pgprot(0), mask, 0,
1015
		(array ? CPA_ARRAY : 0), NULL);
1016 1017
}

1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031
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);
}

1032
int _set_memory_uc(unsigned long addr, int numpages)
1033
{
1034 1035 1036
	/*
	 * for now UC MINUS. see comments in ioremap_nocache()
	 */
1037 1038
	return change_page_attr_set(&addr, numpages,
				    __pgprot(_PAGE_CACHE_UC_MINUS), 0);
1039
}
1040 1041 1042

int set_memory_uc(unsigned long addr, int numpages)
{
1043 1044
	int ret;

1045 1046 1047
	/*
	 * for now UC MINUS. see comments in ioremap_nocache()
	 */
1048 1049 1050 1051 1052 1053 1054 1055 1056 1057
	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;
1058

1059 1060 1061 1062
out_free:
	free_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE);
out_err:
	return ret;
1063
}
1064 1065
EXPORT_SYMBOL(set_memory_uc);

1066
static int _set_memory_array(unsigned long *addr, int addrinarray,
1067
		unsigned long new_type)
1068
{
1069 1070 1071
	int i, j;
	int ret;

1072 1073 1074 1075
	/*
	 * for now UC MINUS. see comments in ioremap_nocache()
	 */
	for (i = 0; i < addrinarray; i++) {
1076
		ret = reserve_memtype(__pa(addr[i]), __pa(addr[i]) + PAGE_SIZE,
1077
					new_type, NULL);
1078 1079
		if (ret)
			goto out_free;
1080 1081
	}

1082
	ret = change_page_attr_set(addr, addrinarray,
1083
				    __pgprot(_PAGE_CACHE_UC_MINUS), 1);
1084 1085 1086 1087 1088 1089

	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);
1090 1091 1092 1093 1094 1095 1096 1097 1098 1099
	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;
1100
}
1101 1102 1103 1104 1105

int set_memory_array_uc(unsigned long *addr, int addrinarray)
{
	return _set_memory_array(addr, addrinarray, _PAGE_CACHE_UC_MINUS);
}
1106 1107
EXPORT_SYMBOL(set_memory_array_uc);

1108 1109 1110 1111 1112 1113
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);

1114 1115
int _set_memory_wc(unsigned long addr, int numpages)
{
1116
	int ret;
1117 1118
	unsigned long addr_copy = addr;

1119 1120 1121
	ret = change_page_attr_set(&addr, numpages,
				    __pgprot(_PAGE_CACHE_UC_MINUS), 0);
	if (!ret) {
1122 1123 1124 1125
		ret = change_page_attr_set_clr(&addr_copy, numpages,
					       __pgprot(_PAGE_CACHE_WC),
					       __pgprot(_PAGE_CACHE_MASK),
					       0, 0, NULL);
1126 1127
	}
	return ret;
1128 1129 1130 1131
}

int set_memory_wc(unsigned long addr, int numpages)
{
1132 1133
	int ret;

1134
	if (!pat_enabled)
1135 1136
		return set_memory_uc(addr, numpages);

1137 1138 1139 1140
	ret = reserve_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE,
		_PAGE_CACHE_WC, NULL);
	if (ret)
		goto out_err;
1141

1142 1143 1144 1145 1146 1147 1148 1149 1150 1151
	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;
1152 1153 1154
}
EXPORT_SYMBOL(set_memory_wc);

1155
int _set_memory_wb(unsigned long addr, int numpages)
1156
{
1157 1158
	return change_page_attr_clear(&addr, numpages,
				      __pgprot(_PAGE_CACHE_MASK), 0);
1159
}
1160 1161 1162

int set_memory_wb(unsigned long addr, int numpages)
{
1163 1164 1165 1166 1167 1168
	int ret;

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

1169
	free_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE);
1170
	return 0;
1171
}
1172 1173
EXPORT_SYMBOL(set_memory_wb);

1174 1175 1176
int set_memory_array_wb(unsigned long *addr, int addrinarray)
{
	int i;
1177 1178 1179 1180
	int ret;

	ret = change_page_attr_clear(addr, addrinarray,
				      __pgprot(_PAGE_CACHE_MASK), 1);
1181 1182
	if (ret)
		return ret;
1183

1184 1185
	for (i = 0; i < addrinarray; i++)
		free_memtype(__pa(addr[i]), __pa(addr[i]) + PAGE_SIZE);
1186

1187
	return 0;
1188 1189 1190
}
EXPORT_SYMBOL(set_memory_array_wb);

1191 1192
int set_memory_x(unsigned long addr, int numpages)
{
1193 1194 1195
	if (!(__supported_pte_mask & _PAGE_NX))
		return 0;

1196
	return change_page_attr_clear(&addr, numpages, __pgprot(_PAGE_NX), 0);
1197 1198 1199 1200 1201
}
EXPORT_SYMBOL(set_memory_x);

int set_memory_nx(unsigned long addr, int numpages)
{
1202 1203 1204
	if (!(__supported_pte_mask & _PAGE_NX))
		return 0;

1205
	return change_page_attr_set(&addr, numpages, __pgprot(_PAGE_NX), 0);
1206 1207 1208 1209 1210
}
EXPORT_SYMBOL(set_memory_nx);

int set_memory_ro(unsigned long addr, int numpages)
{
1211
	return change_page_attr_clear(&addr, numpages, __pgprot(_PAGE_RW), 0);
1212
}
1213
EXPORT_SYMBOL_GPL(set_memory_ro);
1214 1215 1216

int set_memory_rw(unsigned long addr, int numpages)
{
1217
	return change_page_attr_set(&addr, numpages, __pgprot(_PAGE_RW), 0);
1218
}
1219
EXPORT_SYMBOL_GPL(set_memory_rw);
I
Ingo Molnar 已提交
1220 1221 1222

int set_memory_np(unsigned long addr, int numpages)
{
1223
	return change_page_attr_clear(&addr, numpages, __pgprot(_PAGE_PRESENT), 0);
I
Ingo Molnar 已提交
1224
}
1225

1226 1227
int set_memory_4k(unsigned long addr, int numpages)
{
1228
	return change_page_attr_set_clr(&addr, numpages, __pgprot(0),
1229
					__pgprot(0), 1, 0, NULL);
1230 1231
}

1232 1233 1234 1235
int set_pages_uc(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);

T
Thomas Gleixner 已提交
1236
	return set_memory_uc(addr, numpages);
1237 1238 1239
}
EXPORT_SYMBOL(set_pages_uc);

1240 1241
static int _set_pages_array(struct page **pages, int addrinarray,
		unsigned long new_type)
1242 1243 1244 1245 1246
{
	unsigned long start;
	unsigned long end;
	int i;
	int free_idx;
1247
	int ret;
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
		end = start + PAGE_SIZE;
1254
		if (reserve_memtype(start, end, new_type, NULL))
1255 1256 1257
			goto err_out;
	}

1258 1259 1260 1261 1262 1263 1264 1265 1266 1267
	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 */
1268 1269 1270
err_out:
	free_idx = i;
	for (i = 0; i < free_idx; i++) {
1271 1272 1273
		if (PageHighMem(pages[i]))
			continue;
		start = page_to_pfn(pages[i]) << PAGE_SHIFT;
1274 1275 1276 1277 1278
		end = start + PAGE_SIZE;
		free_memtype(start, end);
	}
	return -EINVAL;
}
1279 1280 1281 1282 1283

int set_pages_array_uc(struct page **pages, int addrinarray)
{
	return _set_pages_array(pages, addrinarray, _PAGE_CACHE_UC_MINUS);
}
1284 1285
EXPORT_SYMBOL(set_pages_array_uc);

1286 1287 1288 1289 1290 1291
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);

1292 1293 1294 1295
int set_pages_wb(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);

T
Thomas Gleixner 已提交
1296
	return set_memory_wb(addr, numpages);
1297 1298 1299
}
EXPORT_SYMBOL(set_pages_wb);

1300 1301 1302 1303 1304 1305 1306 1307 1308
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));
1309 1310
	if (retval)
		return retval;
1311 1312

	for (i = 0; i < addrinarray; i++) {
1313 1314 1315
		if (PageHighMem(pages[i]))
			continue;
		start = page_to_pfn(pages[i]) << PAGE_SHIFT;
1316 1317 1318 1319
		end = start + PAGE_SIZE;
		free_memtype(start, end);
	}

1320
	return 0;
1321 1322 1323
}
EXPORT_SYMBOL(set_pages_array_wb);

1324 1325 1326 1327
int set_pages_x(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);

T
Thomas Gleixner 已提交
1328
	return set_memory_x(addr, numpages);
1329 1330 1331 1332 1333 1334 1335
}
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
1336
	return set_memory_nx(addr, numpages);
1337 1338 1339 1340 1341 1342 1343
}
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
1344
	return set_memory_ro(addr, numpages);
1345 1346 1347 1348 1349
}

int set_pages_rw(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);
1350

T
Thomas Gleixner 已提交
1351
	return set_memory_rw(addr, numpages);
I
Ingo Molnar 已提交
1352 1353
}

L
Linus Torvalds 已提交
1354
#ifdef CONFIG_DEBUG_PAGEALLOC
I
Ingo Molnar 已提交
1355 1356 1357

static int __set_pages_p(struct page *page, int numpages)
{
1358 1359
	unsigned long tempaddr = (unsigned long) page_address(page);
	struct cpa_data cpa = { .vaddr = &tempaddr,
T
Thomas Gleixner 已提交
1360 1361
				.numpages = numpages,
				.mask_set = __pgprot(_PAGE_PRESENT | _PAGE_RW),
1362 1363
				.mask_clr = __pgprot(0),
				.flags = 0};
1364

1365 1366 1367 1368 1369 1370 1371
	/*
	 * 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
1372 1373 1374 1375
}

static int __set_pages_np(struct page *page, int numpages)
{
1376 1377
	unsigned long tempaddr = (unsigned long) page_address(page);
	struct cpa_data cpa = { .vaddr = &tempaddr,
T
Thomas Gleixner 已提交
1378 1379
				.numpages = numpages,
				.mask_set = __pgprot(0),
1380 1381
				.mask_clr = __pgprot(_PAGE_PRESENT | _PAGE_RW),
				.flags = 0};
1382

1383 1384 1385 1386 1387 1388 1389
	/*
	 * 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
1390 1391
}

L
Linus Torvalds 已提交
1392 1393 1394 1395
void kernel_map_pages(struct page *page, int numpages, int enable)
{
	if (PageHighMem(page))
		return;
1396
	if (!enable) {
1397 1398
		debug_check_no_locks_freed(page_address(page),
					   numpages * PAGE_SIZE);
1399
	}
1400

1401
	/*
I
Ingo Molnar 已提交
1402
	 * The return value is ignored as the calls cannot fail.
1403 1404
	 * 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
1405
	 */
I
Ingo Molnar 已提交
1406 1407 1408 1409
	if (enable)
		__set_pages_p(page, numpages);
	else
		__set_pages_np(page, numpages);
1410 1411

	/*
1412 1413
	 * We should perform an IPI and flush all tlbs,
	 * but that can deadlock->flush only current cpu:
L
Linus Torvalds 已提交
1414 1415
	 */
	__flush_tlb_all();
1416 1417
}

1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434
#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 */
1435 1436 1437 1438 1439 1440 1441 1442

/*
 * 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