pageattr.c 47.6 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>
7 8
#include <linux/sched.h>
#include <linux/mm.h>
9
#include <linux/interrupt.h>
10 11
#include <linux/seq_file.h>
#include <linux/debugfs.h>
12
#include <linux/pfn.h>
13
#include <linux/percpu.h>
14
#include <linux/gfp.h>
15
#include <linux/pci.h>
16
#include <linux/vmalloc.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;
33
	pgd_t		*pgd;
T
Thomas Gleixner 已提交
34 35
	pgprot_t	mask_set;
	pgprot_t	mask_clr;
36
	unsigned long	numpages;
37
	int		flags;
T
Thomas Gleixner 已提交
38
	unsigned long	pfn;
39
	unsigned	force_split : 1;
40
	int		curpage;
41
	struct page	**pages;
T
Thomas Gleixner 已提交
42 43
};

44 45 46 47 48 49 50 51
/*
 * 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);

52 53
#define CPA_FLUSHTLB 1
#define CPA_ARRAY 2
54
#define CPA_PAGES_ARRAY 4
55

56
#ifdef CONFIG_PROC_FS
57 58
static unsigned long direct_pages_count[PG_LEVEL_NUM];

59
void update_page_count(int level, unsigned long pages)
60 61
{
	/* Protect against CPA */
A
Andrea Arcangeli 已提交
62
	spin_lock(&pgd_lock);
63
	direct_pages_count[level] += pages;
A
Andrea Arcangeli 已提交
64
	spin_unlock(&pgd_lock);
65 66 67 68
}

static void split_page_count(int level)
{
69 70 71
	if (direct_pages_count[level] == 0)
		return;

72 73 74 75
	direct_pages_count[level]--;
	direct_pages_count[level - 1] += PTRS_PER_PTE;
}

76
void arch_report_meminfo(struct seq_file *m)
77
{
78
	seq_printf(m, "DirectMap4k:    %8lu kB\n",
H
Hugh Dickins 已提交
79 80
			direct_pages_count[PG_LEVEL_4K] << 2);
#if defined(CONFIG_X86_64) || defined(CONFIG_X86_PAE)
81
	seq_printf(m, "DirectMap2M:    %8lu kB\n",
H
Hugh Dickins 已提交
82 83
			direct_pages_count[PG_LEVEL_2M] << 11);
#else
84
	seq_printf(m, "DirectMap4M:    %8lu kB\n",
H
Hugh Dickins 已提交
85 86 87
			direct_pages_count[PG_LEVEL_2M] << 12);
#endif
	if (direct_gbpages)
88
		seq_printf(m, "DirectMap1G:    %8lu kB\n",
H
Hugh Dickins 已提交
89
			direct_pages_count[PG_LEVEL_1G] << 20);
90
}
91 92 93
#else
static inline void split_page_count(int level) { }
#endif
94

T
Thomas Gleixner 已提交
95 96 97 98
#ifdef CONFIG_X86_64

static inline unsigned long highmap_start_pfn(void)
{
99
	return __pa_symbol(_text) >> PAGE_SHIFT;
T
Thomas Gleixner 已提交
100 101 102 103
}

static inline unsigned long highmap_end_pfn(void)
{
104
	return __pa_symbol(roundup(_brk_end, PMD_SIZE)) >> PAGE_SHIFT;
T
Thomas Gleixner 已提交
105 106 107 108
}

#endif

109 110
static inline int
within(unsigned long addr, unsigned long start, unsigned long end)
I
Ingo Molnar 已提交
111
{
112 113 114
	return addr >= start && addr < end;
}

T
Thomas Gleixner 已提交
115 116 117
/*
 * Flushing functions
 */
118 119 120

/**
 * clflush_cache_range - flush a cache range with clflush
121
 * @vaddr:	virtual start address
122 123
 * @size:	number of bytes to flush
 *
124 125
 * clflushopt is an unordered instruction which needs fencing with mfence or
 * sfence to avoid ordering issues.
126
 */
I
Ingo Molnar 已提交
127
void clflush_cache_range(void *vaddr, unsigned int size)
T
Thomas Gleixner 已提交
128
{
129 130
	const unsigned long clflush_size = boot_cpu_data.x86_clflush_size;
	void *p = (void *)((unsigned long)vaddr & ~(clflush_size - 1));
131
	void *vend = vaddr + size;
132 133 134

	if (p >= vend)
		return;
T
Thomas Gleixner 已提交
135

136
	mb();
I
Ingo Molnar 已提交
137

138
	for (; p < vend; p += clflush_size)
139
		clflushopt(p);
I
Ingo Molnar 已提交
140

141
	mb();
T
Thomas Gleixner 已提交
142
}
143
EXPORT_SYMBOL_GPL(clflush_cache_range);
T
Thomas Gleixner 已提交
144

145
static void __cpa_flush_all(void *arg)
T
Thomas Gleixner 已提交
146
{
147 148
	unsigned long cache = (unsigned long)arg;

T
Thomas Gleixner 已提交
149 150 151 152 153 154
	/*
	 * Flush all to work around Errata in early athlons regarding
	 * large page flushing.
	 */
	__flush_tlb_all();

155
	if (cache && boot_cpu_data.x86 >= 4)
T
Thomas Gleixner 已提交
156 157 158
		wbinvd();
}

159
static void cpa_flush_all(unsigned long cache)
T
Thomas Gleixner 已提交
160 161 162
{
	BUG_ON(irqs_disabled());

163
	on_each_cpu(__cpa_flush_all, (void *) cache, 1);
T
Thomas Gleixner 已提交
164 165
}

166 167 168 169 170 171 172 173 174 175
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();
}

176
static void cpa_flush_range(unsigned long start, int numpages, int cache)
177
{
I
Ingo Molnar 已提交
178 179 180
	unsigned int i, level;
	unsigned long addr;

181
	BUG_ON(irqs_disabled());
I
Ingo Molnar 已提交
182
	WARN_ON(PAGE_ALIGN(start) != start);
183

184
	on_each_cpu(__cpa_flush_range, NULL, 1);
185

186 187 188
	if (!cache)
		return;

T
Thomas Gleixner 已提交
189 190 191 192 193 194
	/*
	 * 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
195 196 197 198 199 200
	for (i = 0, addr = start; i < numpages; i++, addr += PAGE_SIZE) {
		pte_t *pte = lookup_address(addr, &level);

		/*
		 * Only flush present addresses:
		 */
201
		if (pte && (pte_val(*pte) & _PAGE_PRESENT))
I
Ingo Molnar 已提交
202 203
			clflush_cache_range((void *) addr, PAGE_SIZE);
	}
204 205
}

206 207
static void cpa_flush_array(unsigned long *start, int numpages, int cache,
			    int in_flags, struct page **pages)
208 209
{
	unsigned int i, level;
210
	unsigned long do_wbinvd = cache && numpages >= 1024; /* 4M threshold */
211 212 213

	BUG_ON(irqs_disabled());

214
	on_each_cpu(__cpa_flush_all, (void *) do_wbinvd, 1);
215

216
	if (!cache || do_wbinvd)
217 218 219 220 221 222 223 224
		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:
	 */
225 226 227 228 229 230 231 232 233 234
	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);
235 236 237 238 239

		/*
		 * Only flush present addresses:
		 */
		if (pte && (pte_val(*pte) & _PAGE_PRESENT))
240
			clflush_cache_range((void *)addr, PAGE_SIZE);
241 242 243
	}
}

244 245 246 247 248 249
/*
 * 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
250 251
static inline pgprot_t static_protections(pgprot_t prot, unsigned long address,
				   unsigned long pfn)
252 253 254
{
	pgprot_t forbidden = __pgprot(0);

I
Ingo Molnar 已提交
255
	/*
256 257
	 * 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
258
	 */
259 260
#ifdef CONFIG_PCI_BIOS
	if (pcibios_enabled && within(pfn, BIOS_BEGIN >> PAGE_SHIFT, BIOS_END >> PAGE_SHIFT))
261
		pgprot_val(forbidden) |= _PAGE_NX;
262
#endif
263 264 265

	/*
	 * 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
	 */
276 277
	if (within(pfn, __pa_symbol(__start_rodata) >> PAGE_SHIFT,
		   __pa_symbol(__end_rodata) >> PAGE_SHIFT))
278
		pgprot_val(forbidden) |= _PAGE_RW;
279

280
#if defined(CONFIG_X86_64)
281
	/*
282 283 284 285
	 * 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.
286 287 288 289
	 *
	 * This will preserve the large page mappings for kernel text/data
	 * at no extra cost.
	 */
290 291
	if (kernel_set_to_readonly &&
	    within(address, (unsigned long)_text,
292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309
		   (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
310
		 * as well.
311 312 313 314
		 */
		if (lookup_address(address, &level) && (level != PG_LEVEL_4K))
			pgprot_val(forbidden) |= _PAGE_RW;
	}
315 316
#endif

317
	prot = __pgprot(pgprot_val(prot) & ~pgprot_val(forbidden));
I
Ingo Molnar 已提交
318 319 320 321

	return prot;
}

322 323 324 325 326 327
/*
 * Lookup the page table entry for a virtual address in a specific pgd.
 * Return a pointer to the entry and the level of the mapping.
 */
pte_t *lookup_address_in_pgd(pgd_t *pgd, unsigned long address,
			     unsigned int *level)
328
{
L
Linus Torvalds 已提交
329 330
	pud_t *pud;
	pmd_t *pmd;
331

T
Thomas Gleixner 已提交
332 333
	*level = PG_LEVEL_NONE;

L
Linus Torvalds 已提交
334 335
	if (pgd_none(*pgd))
		return NULL;
I
Ingo Molnar 已提交
336

L
Linus Torvalds 已提交
337 338 339
	pud = pud_offset(pgd, address);
	if (pud_none(*pud))
		return NULL;
340 341 342 343 344

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

L
Linus Torvalds 已提交
345 346 347
	pmd = pmd_offset(pud, address);
	if (pmd_none(*pmd))
		return NULL;
T
Thomas Gleixner 已提交
348 349

	*level = PG_LEVEL_2M;
T
Thomas Gleixner 已提交
350
	if (pmd_large(*pmd) || !pmd_present(*pmd))
L
Linus Torvalds 已提交
351 352
		return (pte_t *)pmd;

T
Thomas Gleixner 已提交
353
	*level = PG_LEVEL_4K;
I
Ingo Molnar 已提交
354

355 356
	return pte_offset_kernel(pmd, address);
}
357 358 359 360 361 362 363 364 365 366 367

/*
 * 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.
 */
pte_t *lookup_address(unsigned long address, unsigned int *level)
{
368
        return lookup_address_in_pgd(pgd_offset_k(address), address, level);
369
}
370
EXPORT_SYMBOL_GPL(lookup_address);
371

372 373 374 375
static pte_t *_lookup_address_cpa(struct cpa_data *cpa, unsigned long address,
				  unsigned int *level)
{
        if (cpa->pgd)
376
		return lookup_address_in_pgd(cpa->pgd + pgd_index(address),
377 378 379 380 381
					       address, level);

        return lookup_address(address, level);
}

382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401
/*
 * Lookup the PMD entry for a virtual address. Return a pointer to the entry
 * or NULL if not present.
 */
pmd_t *lookup_pmd_address(unsigned long address)
{
	pgd_t *pgd;
	pud_t *pud;

	pgd = pgd_offset_k(address);
	if (pgd_none(*pgd))
		return NULL;

	pud = pud_offset(pgd, address);
	if (pud_none(*pud) || pud_large(*pud) || !pud_present(*pud))
		return NULL;

	return pmd_offset(pud, address);
}

402 403 404 405 406 407 408 409 410 411 412 413 414 415
/*
 * 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;
416 417
	phys_addr_t phys_addr;
	unsigned long offset;
418 419 420 421 422
	enum pg_level level;
	pte_t *pte;

	pte = lookup_address(virt_addr, &level);
	BUG_ON(!pte);
423

424 425 426 427 428
	/*
	 * pXX_pfn() returns unsigned long, which must be cast to phys_addr_t
	 * before being left-shifted PAGE_SHIFT bits -- this trick is to
	 * make 32-PAE kernel work correctly.
	 */
429 430
	switch (level) {
	case PG_LEVEL_1G:
431
		phys_addr = (phys_addr_t)pud_pfn(*(pud_t *)pte) << PAGE_SHIFT;
432 433 434
		offset = virt_addr & ~PUD_PAGE_MASK;
		break;
	case PG_LEVEL_2M:
435
		phys_addr = (phys_addr_t)pmd_pfn(*(pmd_t *)pte) << PAGE_SHIFT;
436 437 438
		offset = virt_addr & ~PMD_PAGE_MASK;
		break;
	default:
439
		phys_addr = (phys_addr_t)pte_pfn(*pte) << PAGE_SHIFT;
440 441 442 443
		offset = virt_addr & ~PAGE_MASK;
	}

	return (phys_addr_t)(phys_addr | offset);
444 445 446
}
EXPORT_SYMBOL_GPL(slow_virt_to_phys);

I
Ingo Molnar 已提交
447 448 449
/*
 * Set the new pmd in all the pgds we know about:
 */
I
Ingo Molnar 已提交
450
static void __set_pmd_pte(pte_t *kpte, unsigned long address, pte_t pte)
451 452 453
{
	/* change init_mm */
	set_pte_atomic(kpte, pte);
454
#ifdef CONFIG_X86_32
455
	if (!SHARED_KERNEL_PMD) {
456 457
		struct page *page;

458
		list_for_each_entry(page, &pgd_list, lru) {
459 460 461 462 463 464 465 466 467
			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
468
	}
469
#endif
L
Linus Torvalds 已提交
470 471
}

I
Ingo Molnar 已提交
472 473 474
static int
try_preserve_large_page(pte_t *kpte, unsigned long address,
			struct cpa_data *cpa)
475
{
476
	unsigned long nextpage_addr, numpages, pmask, psize, addr, pfn, old_pfn;
477
	pte_t new_pte, old_pte, *tmp;
478
	pgprot_t old_prot, new_prot, req_prot;
479
	int i, do_split = 1;
480
	enum pg_level level;
481

482 483 484
	if (cpa->force_split)
		return 1;

A
Andrea Arcangeli 已提交
485
	spin_lock(&pgd_lock);
486 487 488 489
	/*
	 * Check for races, another CPU might have split this page
	 * up already:
	 */
490
	tmp = _lookup_address_cpa(cpa, address, &level);
491 492 493 494 495
	if (tmp != kpte)
		goto out_unlock;

	switch (level) {
	case PG_LEVEL_2M:
496 497 498
		old_prot = pmd_pgprot(*(pmd_t *)kpte);
		old_pfn = pmd_pfn(*(pmd_t *)kpte);
		break;
499
	case PG_LEVEL_1G:
500 501
		old_prot = pud_pgprot(*(pud_t *)kpte);
		old_pfn = pud_pfn(*(pud_t *)kpte);
502
		break;
503
	default:
I
Ingo Molnar 已提交
504
		do_split = -EINVAL;
505 506 507
		goto out_unlock;
	}

508 509 510
	psize = page_level_size(level);
	pmask = page_level_mask(level);

511 512 513 514 515 516
	/*
	 * 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;
517 518
	if (numpages < cpa->numpages)
		cpa->numpages = numpages;
519 520 521

	/*
	 * We are safe now. Check whether the new pgprot is the same:
522 523
	 * Convert protection attributes to 4k-format, as cpa->mask* are set
	 * up accordingly.
524 525
	 */
	old_pte = *kpte;
526
	req_prot = pgprot_large_2_4k(old_prot);
527

528 529
	pgprot_val(req_prot) &= ~pgprot_val(cpa->mask_clr);
	pgprot_val(req_prot) |= pgprot_val(cpa->mask_set);
T
Thomas Gleixner 已提交
530

531 532 533 534 535 536 537
	/*
	 * req_prot is in format of 4k pages. It must be converted to large
	 * page format: the caching mode includes the PAT bit located at
	 * different bit positions in the two formats.
	 */
	req_prot = pgprot_4k_2_large(req_prot);

538 539 540 541 542 543
	/*
	 * 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.
	 */
544 545
	if (pgprot_val(req_prot) & _PAGE_PRESENT)
		pgprot_val(req_prot) |= _PAGE_PSE | _PAGE_GLOBAL;
546
	else
547
		pgprot_val(req_prot) &= ~(_PAGE_PSE | _PAGE_GLOBAL);
548

549
	req_prot = canon_pgprot(req_prot);
550

T
Thomas Gleixner 已提交
551
	/*
552
	 * old_pfn points to the large page base pfn. So we need
T
Thomas Gleixner 已提交
553 554
	 * to add the offset of the virtual address:
	 */
555
	pfn = old_pfn + ((address & (psize - 1)) >> PAGE_SHIFT);
T
Thomas Gleixner 已提交
556 557
	cpa->pfn = pfn;

558
	new_prot = static_protections(req_prot, address, pfn);
559

560 561 562 563 564
	/*
	 * 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:
	 */
565
	addr = address & pmask;
566
	pfn = old_pfn;
567 568
	for (i = 0; i < (psize >> PAGE_SHIFT); i++, addr += PAGE_SIZE, pfn++) {
		pgprot_t chk_prot = static_protections(req_prot, addr, pfn);
569 570 571 572 573

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

574 575 576 577 578
	/*
	 * 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
579
		do_split = 0;
580 581 582 583 584 585 586 587 588 589 590
		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.
	 */
591
	if (address == (address & pmask) && cpa->numpages == (psize >> PAGE_SHIFT)) {
592 593 594 595
		/*
		 * The address is aligned and the number of pages
		 * covers the full page.
		 */
596
		new_pte = pfn_pte(old_pfn, new_prot);
597
		__set_pmd_pte(kpte, address, new_pte);
598
		cpa->flags |= CPA_FLUSHTLB;
I
Ingo Molnar 已提交
599
		do_split = 0;
600 601 602
	}

out_unlock:
A
Andrea Arcangeli 已提交
603
	spin_unlock(&pgd_lock);
I
Ingo Molnar 已提交
604

I
Ingo Molnar 已提交
605
	return do_split;
606 607
}

608
static int
609 610
__split_large_page(struct cpa_data *cpa, pte_t *kpte, unsigned long address,
		   struct page *base)
611
{
612
	pte_t *pbase = (pte_t *)page_address(base);
613
	unsigned long ref_pfn, pfn, pfninc = 1;
I
Ingo Molnar 已提交
614
	unsigned int i, level;
615
	pte_t *tmp;
I
Ingo Molnar 已提交
616
	pgprot_t ref_prot;
617

A
Andrea Arcangeli 已提交
618
	spin_lock(&pgd_lock);
619 620 621 622
	/*
	 * Check for races, another CPU might have split this page
	 * up for us already:
	 */
623
	tmp = _lookup_address_cpa(cpa, address, &level);
624 625 626 627
	if (tmp != kpte) {
		spin_unlock(&pgd_lock);
		return 1;
	}
628

629
	paravirt_alloc_pte(&init_mm, page_to_pfn(base));
630

631 632 633 634
	switch (level) {
	case PG_LEVEL_2M:
		ref_prot = pmd_pgprot(*(pmd_t *)kpte);
		/* clear PSE and promote PAT bit to correct position */
635
		ref_prot = pgprot_large_2_4k(ref_prot);
636 637
		ref_pfn = pmd_pfn(*(pmd_t *)kpte);
		break;
638

639 640 641
	case PG_LEVEL_1G:
		ref_prot = pud_pgprot(*(pud_t *)kpte);
		ref_pfn = pud_pfn(*(pud_t *)kpte);
642
		pfninc = PMD_PAGE_SIZE >> PAGE_SHIFT;
643

644
		/*
645
		 * Clear the PSE flags if the PRESENT flag is not set
646 647 648
		 * otherwise pmd_present/pmd_huge will return true
		 * even on a non present pmd.
		 */
649
		if (!(pgprot_val(ref_prot) & _PAGE_PRESENT))
650
			pgprot_val(ref_prot) &= ~_PAGE_PSE;
651 652 653 654 655
		break;

	default:
		spin_unlock(&pgd_lock);
		return 1;
656 657
	}

658 659 660 661 662 663 664 665 666 667 668
	/*
	 * 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;

669 670 671
	/*
	 * Get the target pfn from the original entry:
	 */
672
	pfn = ref_pfn;
673
	for (i = 0; i < PTRS_PER_PTE; i++, pfn += pfninc)
674
		set_pte(&pbase[i], pfn_pte(pfn, canon_pgprot(ref_prot)));
675

676 677 678 679 680 681
	if (virt_addr_valid(address)) {
		unsigned long pfn = PFN_DOWN(__pa(address));

		if (pfn_range_is_mapped(pfn, pfn + 1))
			split_page_count(level);
	}
682

683
	/*
684
	 * Install the new, split up pagetable.
685
	 *
686 687 688
	 * We use the standard kernel pagetable protections for the new
	 * pagetable protections, the actual ptes set above control the
	 * primary protection behavior:
689
	 */
690
	__set_pmd_pte(kpte, address, mk_pte(base, __pgprot(_KERNPG_TABLE)));
691 692 693 694 695 696 697 698 699 700

	/*
	 * 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();
701
	spin_unlock(&pgd_lock);
702

703 704
	return 0;
}
705

706 707
static int split_large_page(struct cpa_data *cpa, pte_t *kpte,
			    unsigned long address)
708 709 710
{
	struct page *base;

711
	if (!debug_pagealloc_enabled())
712 713
		spin_unlock(&cpa_lock);
	base = alloc_pages(GFP_KERNEL | __GFP_NOTRACK, 0);
714
	if (!debug_pagealloc_enabled())
715 716 717 718
		spin_lock(&cpa_lock);
	if (!base)
		return -ENOMEM;

719
	if (__split_large_page(cpa, kpte, address, base))
S
Suresh Siddha 已提交
720
		__free_page(base);
721 722 723 724

	return 0;
}

725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748
static bool try_to_free_pte_page(pte_t *pte)
{
	int i;

	for (i = 0; i < PTRS_PER_PTE; i++)
		if (!pte_none(pte[i]))
			return false;

	free_page((unsigned long)pte);
	return true;
}

static bool try_to_free_pmd_page(pmd_t *pmd)
{
	int i;

	for (i = 0; i < PTRS_PER_PMD; i++)
		if (!pmd_none(pmd[i]))
			return false;

	free_page((unsigned long)pmd);
	return true;
}

749 750 751 752 753 754 755 756 757 758 759 760
static bool try_to_free_pud_page(pud_t *pud)
{
	int i;

	for (i = 0; i < PTRS_PER_PUD; i++)
		if (!pud_none(pud[i]))
			return false;

	free_page((unsigned long)pud);
	return true;
}

761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829
static bool unmap_pte_range(pmd_t *pmd, unsigned long start, unsigned long end)
{
	pte_t *pte = pte_offset_kernel(pmd, start);

	while (start < end) {
		set_pte(pte, __pte(0));

		start += PAGE_SIZE;
		pte++;
	}

	if (try_to_free_pte_page((pte_t *)pmd_page_vaddr(*pmd))) {
		pmd_clear(pmd);
		return true;
	}
	return false;
}

static void __unmap_pmd_range(pud_t *pud, pmd_t *pmd,
			      unsigned long start, unsigned long end)
{
	if (unmap_pte_range(pmd, start, end))
		if (try_to_free_pmd_page((pmd_t *)pud_page_vaddr(*pud)))
			pud_clear(pud);
}

static void unmap_pmd_range(pud_t *pud, unsigned long start, unsigned long end)
{
	pmd_t *pmd = pmd_offset(pud, start);

	/*
	 * Not on a 2MB page boundary?
	 */
	if (start & (PMD_SIZE - 1)) {
		unsigned long next_page = (start + PMD_SIZE) & PMD_MASK;
		unsigned long pre_end = min_t(unsigned long, end, next_page);

		__unmap_pmd_range(pud, pmd, start, pre_end);

		start = pre_end;
		pmd++;
	}

	/*
	 * Try to unmap in 2M chunks.
	 */
	while (end - start >= PMD_SIZE) {
		if (pmd_large(*pmd))
			pmd_clear(pmd);
		else
			__unmap_pmd_range(pud, pmd, start, start + PMD_SIZE);

		start += PMD_SIZE;
		pmd++;
	}

	/*
	 * 4K leftovers?
	 */
	if (start < end)
		return __unmap_pmd_range(pud, pmd, start, end);

	/*
	 * Try again to free the PMD page if haven't succeeded above.
	 */
	if (!pud_none(*pud))
		if (try_to_free_pmd_page((pmd_t *)pud_page_vaddr(*pud)))
			pud_clear(pud);
}
830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873

static void unmap_pud_range(pgd_t *pgd, unsigned long start, unsigned long end)
{
	pud_t *pud = pud_offset(pgd, start);

	/*
	 * Not on a GB page boundary?
	 */
	if (start & (PUD_SIZE - 1)) {
		unsigned long next_page = (start + PUD_SIZE) & PUD_MASK;
		unsigned long pre_end	= min_t(unsigned long, end, next_page);

		unmap_pmd_range(pud, start, pre_end);

		start = pre_end;
		pud++;
	}

	/*
	 * Try to unmap in 1G chunks?
	 */
	while (end - start >= PUD_SIZE) {

		if (pud_large(*pud))
			pud_clear(pud);
		else
			unmap_pmd_range(pud, start, start + PUD_SIZE);

		start += PUD_SIZE;
		pud++;
	}

	/*
	 * 2M leftovers?
	 */
	if (start < end)
		unmap_pmd_range(pud, start, end);

	/*
	 * No need to try to free the PUD page because we'll free it in
	 * populate_pgd's error path
	 */
}

874 875 876 877 878 879 880 881 882 883
static void unmap_pgd_range(pgd_t *root, unsigned long addr, unsigned long end)
{
	pgd_t *pgd_entry = root + pgd_index(addr);

	unmap_pud_range(pgd_entry, addr, end);

	if (try_to_free_pud_page((pud_t *)pgd_page_vaddr(*pgd_entry)))
		pgd_clear(pgd_entry);
}

884 885 886 887 888 889 890 891 892 893
static int alloc_pte_page(pmd_t *pmd)
{
	pte_t *pte = (pte_t *)get_zeroed_page(GFP_KERNEL | __GFP_NOTRACK);
	if (!pte)
		return -1;

	set_pmd(pmd, __pmd(__pa(pte) | _KERNPG_TABLE));
	return 0;
}

894 895 896 897 898 899 900 901 902 903
static int alloc_pmd_page(pud_t *pud)
{
	pmd_t *pmd = (pmd_t *)get_zeroed_page(GFP_KERNEL | __GFP_NOTRACK);
	if (!pmd)
		return -1;

	set_pud(pud, __pud(__pa(pmd) | _KERNPG_TABLE));
	return 0;
}

904 905 906 907 908 909 910 911
static void populate_pte(struct cpa_data *cpa,
			 unsigned long start, unsigned long end,
			 unsigned num_pages, pmd_t *pmd, pgprot_t pgprot)
{
	pte_t *pte;

	pte = pte_offset_kernel(pmd, start);

912 913 914 915 916 917 918 919 920 921 922
	/*
	 * 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(pgprot) & _PAGE_PRESENT)
		pgprot_val(pgprot) |= _PAGE_GLOBAL;
	else
		pgprot_val(pgprot) &= ~_PAGE_GLOBAL;
923

924
	pgprot = canon_pgprot(pgprot);
925 926

	while (num_pages-- && start < end) {
927
		set_pte(pte, pfn_pte(cpa->pfn, pgprot));
928 929

		start	 += PAGE_SIZE;
930
		cpa->pfn++;
931 932 933
		pte++;
	}
}
934 935 936 937 938 939 940

static int populate_pmd(struct cpa_data *cpa,
			unsigned long start, unsigned long end,
			unsigned num_pages, pud_t *pud, pgprot_t pgprot)
{
	unsigned int cur_pages = 0;
	pmd_t *pmd;
941
	pgprot_t pmd_pgprot;
942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972

	/*
	 * Not on a 2M boundary?
	 */
	if (start & (PMD_SIZE - 1)) {
		unsigned long pre_end = start + (num_pages << PAGE_SHIFT);
		unsigned long next_page = (start + PMD_SIZE) & PMD_MASK;

		pre_end   = min_t(unsigned long, pre_end, next_page);
		cur_pages = (pre_end - start) >> PAGE_SHIFT;
		cur_pages = min_t(unsigned int, num_pages, cur_pages);

		/*
		 * Need a PTE page?
		 */
		pmd = pmd_offset(pud, start);
		if (pmd_none(*pmd))
			if (alloc_pte_page(pmd))
				return -1;

		populate_pte(cpa, start, pre_end, cur_pages, pmd, pgprot);

		start = pre_end;
	}

	/*
	 * We mapped them all?
	 */
	if (num_pages == cur_pages)
		return cur_pages;

973 974
	pmd_pgprot = pgprot_4k_2_large(pgprot);

975 976 977 978 979 980 981 982 983 984 985
	while (end - start >= PMD_SIZE) {

		/*
		 * We cannot use a 1G page so allocate a PMD page if needed.
		 */
		if (pud_none(*pud))
			if (alloc_pmd_page(pud))
				return -1;

		pmd = pmd_offset(pud, start);

986
		set_pmd(pmd, __pmd(cpa->pfn << PAGE_SHIFT | _PAGE_PSE |
987
				   massage_pgprot(pmd_pgprot)));
988 989

		start	  += PMD_SIZE;
990
		cpa->pfn  += PMD_SIZE >> PAGE_SHIFT;
991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007
		cur_pages += PMD_SIZE >> PAGE_SHIFT;
	}

	/*
	 * Map trailing 4K pages.
	 */
	if (start < end) {
		pmd = pmd_offset(pud, start);
		if (pmd_none(*pmd))
			if (alloc_pte_page(pmd))
				return -1;

		populate_pte(cpa, start, end, num_pages - cur_pages,
			     pmd, pgprot);
	}
	return num_pages;
}
1008 1009 1010 1011 1012 1013 1014

static int populate_pud(struct cpa_data *cpa, unsigned long start, pgd_t *pgd,
			pgprot_t pgprot)
{
	pud_t *pud;
	unsigned long end;
	int cur_pages = 0;
1015
	pgprot_t pud_pgprot;
1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052

	end = start + (cpa->numpages << PAGE_SHIFT);

	/*
	 * Not on a Gb page boundary? => map everything up to it with
	 * smaller pages.
	 */
	if (start & (PUD_SIZE - 1)) {
		unsigned long pre_end;
		unsigned long next_page = (start + PUD_SIZE) & PUD_MASK;

		pre_end   = min_t(unsigned long, end, next_page);
		cur_pages = (pre_end - start) >> PAGE_SHIFT;
		cur_pages = min_t(int, (int)cpa->numpages, cur_pages);

		pud = pud_offset(pgd, start);

		/*
		 * Need a PMD page?
		 */
		if (pud_none(*pud))
			if (alloc_pmd_page(pud))
				return -1;

		cur_pages = populate_pmd(cpa, start, pre_end, cur_pages,
					 pud, pgprot);
		if (cur_pages < 0)
			return cur_pages;

		start = pre_end;
	}

	/* We mapped them all? */
	if (cpa->numpages == cur_pages)
		return cur_pages;

	pud = pud_offset(pgd, start);
1053
	pud_pgprot = pgprot_4k_2_large(pgprot);
1054 1055 1056 1057

	/*
	 * Map everything starting from the Gb boundary, possibly with 1G pages
	 */
1058
	while (boot_cpu_has(X86_FEATURE_GBPAGES) && end - start >= PUD_SIZE) {
1059
		set_pud(pud, __pud(cpa->pfn << PAGE_SHIFT | _PAGE_PSE |
1060
				   massage_pgprot(pud_pgprot)));
1061 1062

		start	  += PUD_SIZE;
1063
		cpa->pfn  += PUD_SIZE >> PAGE_SHIFT;
1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085
		cur_pages += PUD_SIZE >> PAGE_SHIFT;
		pud++;
	}

	/* Map trailing leftover */
	if (start < end) {
		int tmp;

		pud = pud_offset(pgd, start);
		if (pud_none(*pud))
			if (alloc_pmd_page(pud))
				return -1;

		tmp = populate_pmd(cpa, start, end, cpa->numpages - cur_pages,
				   pud, pgprot);
		if (tmp < 0)
			return cur_pages;

		cur_pages += tmp;
	}
	return cur_pages;
}
1086 1087 1088 1089 1090 1091 1092 1093 1094

/*
 * Restrictions for kernel page table do not necessarily apply when mapping in
 * an alternate PGD.
 */
static int populate_pgd(struct cpa_data *cpa, unsigned long addr)
{
	pgprot_t pgprot = __pgprot(_KERNPG_TABLE);
	pud_t *pud = NULL;	/* shut up gcc */
1095
	pgd_t *pgd_entry;
1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114
	int ret;

	pgd_entry = cpa->pgd + pgd_index(addr);

	/*
	 * Allocate a PUD page and hand it down for mapping.
	 */
	if (pgd_none(*pgd_entry)) {
		pud = (pud_t *)get_zeroed_page(GFP_KERNEL | __GFP_NOTRACK);
		if (!pud)
			return -1;

		set_pgd(pgd_entry, __pgd(__pa(pud) | _KERNPG_TABLE));
	}

	pgprot_val(pgprot) &= ~pgprot_val(cpa->mask_clr);
	pgprot_val(pgprot) |=  pgprot_val(cpa->mask_set);

	ret = populate_pud(cpa, addr, pgd_entry, pgprot);
1115
	if (ret < 0) {
1116
		unmap_pgd_range(cpa->pgd, addr,
1117
				addr + (cpa->numpages << PAGE_SHIFT));
1118
		return ret;
1119
	}
1120

1121 1122 1123 1124
	cpa->numpages = ret;
	return 0;
}

1125 1126 1127
static int __cpa_process_fault(struct cpa_data *cpa, unsigned long vaddr,
			       int primary)
{
1128 1129 1130
	if (cpa->pgd)
		return populate_pgd(cpa, vaddr);

1131 1132 1133
	/*
	 * Ignore all non primary paths.
	 */
1134 1135
	if (!primary) {
		cpa->numpages = 1;
1136
		return 0;
1137
	}
1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159

	/*
	 * 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
1160
static int __change_page_attr(struct cpa_data *cpa, int primary)
1161
{
1162
	unsigned long address;
1163 1164
	int do_split, err;
	unsigned int level;
T
Thomas Gleixner 已提交
1165
	pte_t *kpte, old_pte;
L
Linus Torvalds 已提交
1166

1167 1168 1169 1170 1171 1172
	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)
1173 1174 1175
		address = cpa->vaddr[cpa->curpage];
	else
		address = *cpa->vaddr;
1176
repeat:
1177
	kpte = _lookup_address_cpa(cpa, address, &level);
L
Linus Torvalds 已提交
1178
	if (!kpte)
1179
		return __cpa_process_fault(cpa, address, primary);
T
Thomas Gleixner 已提交
1180 1181

	old_pte = *kpte;
1182 1183
	if (!pte_val(old_pte))
		return __cpa_process_fault(cpa, address, primary);
1184

T
Thomas Gleixner 已提交
1185
	if (level == PG_LEVEL_4K) {
T
Thomas Gleixner 已提交
1186
		pte_t new_pte;
1187
		pgprot_t new_prot = pte_pgprot(old_pte);
T
Thomas Gleixner 已提交
1188
		unsigned long pfn = pte_pfn(old_pte);
I
Ingo Molnar 已提交
1189

T
Thomas Gleixner 已提交
1190 1191
		pgprot_val(new_prot) &= ~pgprot_val(cpa->mask_clr);
		pgprot_val(new_prot) |= pgprot_val(cpa->mask_set);
I
Ingo Molnar 已提交
1192

T
Thomas Gleixner 已提交
1193
		new_prot = static_protections(new_prot, address, pfn);
I
Ingo Molnar 已提交
1194

1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206
		/*
		 * 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;

1207 1208 1209 1210 1211
		/*
		 * 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
1212 1213
		new_pte = pfn_pte(pfn, canon_pgprot(new_prot));
		cpa->pfn = pfn;
1214 1215 1216 1217 1218
		/*
		 * Do we really change anything ?
		 */
		if (pte_val(old_pte) != pte_val(new_pte)) {
			set_pte_atomic(kpte, new_pte);
1219
			cpa->flags |= CPA_FLUSHTLB;
1220
		}
1221
		cpa->numpages = 1;
1222
		return 0;
L
Linus Torvalds 已提交
1223
	}
1224 1225 1226 1227 1228

	/*
	 * Check, whether we can keep the large page intact
	 * and just change the pte:
	 */
I
Ingo Molnar 已提交
1229
	do_split = try_preserve_large_page(kpte, address, cpa);
1230 1231
	/*
	 * When the range fits into the existing large page,
1232
	 * return. cp->numpages and cpa->tlbflush have been updated in
1233 1234
	 * try_large_page:
	 */
I
Ingo Molnar 已提交
1235 1236
	if (do_split <= 0)
		return do_split;
1237 1238 1239 1240

	/*
	 * We have to split the large page:
	 */
1241
	err = split_large_page(cpa, kpte, address);
I
Ingo Molnar 已提交
1242
	if (!err) {
1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261
		/*
	 	 * 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
1262 1263
		goto repeat;
	}
I
Ingo Molnar 已提交
1264

I
Ingo Molnar 已提交
1265
	return err;
1266
}
L
Linus Torvalds 已提交
1267

T
Thomas Gleixner 已提交
1268 1269 1270
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
1271
{
T
Thomas Gleixner 已提交
1272
	struct cpa_data alias_cpa;
T
Tejun Heo 已提交
1273
	unsigned long laddr = (unsigned long)__va(cpa->pfn << PAGE_SHIFT);
1274
	unsigned long vaddr;
T
Tejun Heo 已提交
1275
	int ret;
1276

1277
	if (!pfn_range_is_mapped(cpa->pfn, cpa->pfn + 1))
T
Thomas Gleixner 已提交
1278
		return 0;
1279

1280 1281 1282 1283
	/*
	 * No need to redo, when the primary call touched the direct
	 * mapping already:
	 */
1284 1285 1286 1287 1288 1289
	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)
1290 1291 1292 1293 1294
		vaddr = cpa->vaddr[cpa->curpage];
	else
		vaddr = *cpa->vaddr;

	if (!(within(vaddr, PAGE_OFFSET,
1295
		    PAGE_OFFSET + (max_pfn_mapped << PAGE_SHIFT)))) {
1296

1297
		alias_cpa = *cpa;
T
Tejun Heo 已提交
1298
		alias_cpa.vaddr = &laddr;
1299
		alias_cpa.flags &= ~(CPA_PAGES_ARRAY | CPA_ARRAY);
1300

1301
		ret = __change_page_attr_set_clr(&alias_cpa, 0);
T
Tejun Heo 已提交
1302 1303
		if (ret)
			return ret;
1304
	}
1305 1306

#ifdef CONFIG_X86_64
A
Arjan van de Ven 已提交
1307
	/*
T
Tejun Heo 已提交
1308 1309
	 * If the primary call didn't touch the high mapping already
	 * and the physical address is inside the kernel map, we need
1310
	 * to touch the high mapped kernel as well:
A
Arjan van de Ven 已提交
1311
	 */
T
Tejun Heo 已提交
1312 1313 1314 1315 1316 1317 1318
	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
1319

T
Tejun Heo 已提交
1320 1321 1322 1323 1324 1325
		/*
		 * 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
1326
#endif
T
Tejun Heo 已提交
1327 1328

	return 0;
L
Linus Torvalds 已提交
1329 1330
}

T
Thomas Gleixner 已提交
1331
static int __change_page_attr_set_clr(struct cpa_data *cpa, int checkalias)
1332
{
1333
	int ret, numpages = cpa->numpages;
1334

1335 1336 1337 1338 1339
	while (numpages) {
		/*
		 * Store the remaining nr of pages for the large page
		 * preservation check.
		 */
1340
		cpa->numpages = numpages;
1341
		/* for array changes, we can't use large page */
1342
		if (cpa->flags & (CPA_ARRAY | CPA_PAGES_ARRAY))
1343
			cpa->numpages = 1;
T
Thomas Gleixner 已提交
1344

1345
		if (!debug_pagealloc_enabled())
1346
			spin_lock(&cpa_lock);
T
Thomas Gleixner 已提交
1347
		ret = __change_page_attr(cpa, checkalias);
1348
		if (!debug_pagealloc_enabled())
1349
			spin_unlock(&cpa_lock);
1350 1351 1352
		if (ret)
			return ret;

T
Thomas Gleixner 已提交
1353 1354 1355 1356 1357 1358
		if (checkalias) {
			ret = cpa_process_alias(cpa);
			if (ret)
				return ret;
		}

1359 1360 1361 1362 1363
		/*
		 * 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.
		 */
1364
		BUG_ON(cpa->numpages > numpages || !cpa->numpages);
1365
		numpages -= cpa->numpages;
1366
		if (cpa->flags & (CPA_PAGES_ARRAY | CPA_ARRAY))
1367 1368 1369 1370
			cpa->curpage++;
		else
			*cpa->vaddr += cpa->numpages * PAGE_SIZE;

1371
	}
1372 1373 1374
	return 0;
}

1375
static int change_page_attr_set_clr(unsigned long *addr, int numpages,
1376
				    pgprot_t mask_set, pgprot_t mask_clr,
1377 1378
				    int force_split, int in_flag,
				    struct page **pages)
1379
{
T
Thomas Gleixner 已提交
1380
	struct cpa_data cpa;
1381
	int ret, cache, checkalias;
1382
	unsigned long baddr = 0;
1383

1384 1385
	memset(&cpa, 0, sizeof(cpa));

1386 1387 1388 1389 1390 1391
	/*
	 * Check, if we are requested to change a not supported
	 * feature:
	 */
	mask_set = canon_pgprot(mask_set);
	mask_clr = canon_pgprot(mask_clr);
1392
	if (!pgprot_val(mask_set) && !pgprot_val(mask_clr) && !force_split)
1393 1394
		return 0;

1395
	/* Ensure we are PAGE_SIZE aligned */
1396
	if (in_flag & CPA_ARRAY) {
1397 1398 1399 1400 1401 1402 1403
		int i;
		for (i = 0; i < numpages; i++) {
			if (addr[i] & ~PAGE_MASK) {
				addr[i] &= PAGE_MASK;
				WARN_ON_ONCE(1);
			}
		}
1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415
	} 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);
		}
1416 1417 1418 1419 1420
		/*
		 * Save address for cache flush. *addr is modified in the call
		 * to __change_page_attr_set_clr() below.
		 */
		baddr = *addr;
1421 1422
	}

1423 1424 1425
	/* Must avoid aliasing mappings in the highmem code */
	kmap_flush_unused();

N
Nick Piggin 已提交
1426 1427
	vm_unmap_aliases();

T
Thomas Gleixner 已提交
1428
	cpa.vaddr = addr;
1429
	cpa.pages = pages;
T
Thomas Gleixner 已提交
1430 1431 1432
	cpa.numpages = numpages;
	cpa.mask_set = mask_set;
	cpa.mask_clr = mask_clr;
1433 1434
	cpa.flags = 0;
	cpa.curpage = 0;
1435
	cpa.force_split = force_split;
T
Thomas Gleixner 已提交
1436

1437 1438
	if (in_flag & (CPA_ARRAY | CPA_PAGES_ARRAY))
		cpa.flags |= in_flag;
1439

1440 1441 1442 1443
	/* 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);
1444

1445 1446 1447
	/*
	 * Check whether we really changed something:
	 */
1448
	if (!(cpa.flags & CPA_FLUSHTLB))
1449
		goto out;
1450

1451 1452 1453 1454
	/*
	 * No need to flush, when we did not set any of the caching
	 * attributes:
	 */
1455
	cache = !!pgprot2cachemode(mask_set);
1456

1457
	/*
1458 1459
	 * On success we use CLFLUSH, when the CPU supports it to
	 * avoid the WBINVD. If the CPU does not support it and in the
1460
	 * error case we fall back to cpa_flush_all (which uses
1461
	 * WBINVD):
1462
	 */
1463
	if (!ret && cpu_has_clflush) {
1464 1465 1466 1467
		if (cpa.flags & (CPA_PAGES_ARRAY | CPA_ARRAY)) {
			cpa_flush_array(addr, numpages, cache,
					cpa.flags, pages);
		} else
1468
			cpa_flush_range(baddr, numpages, cache);
1469
	} else
1470
		cpa_flush_all(cache);
1471

1472
out:
1473 1474 1475
	return ret;
}

1476 1477
static inline int change_page_attr_set(unsigned long *addr, int numpages,
				       pgprot_t mask, int array)
1478
{
1479
	return change_page_attr_set_clr(addr, numpages, mask, __pgprot(0), 0,
1480
		(array ? CPA_ARRAY : 0), NULL);
1481 1482
}

1483 1484
static inline int change_page_attr_clear(unsigned long *addr, int numpages,
					 pgprot_t mask, int array)
1485
{
1486
	return change_page_attr_set_clr(addr, numpages, __pgprot(0), mask, 0,
1487
		(array ? CPA_ARRAY : 0), NULL);
1488 1489
}

1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503
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);
}

1504
int _set_memory_uc(unsigned long addr, int numpages)
1505
{
1506 1507
	/*
	 * for now UC MINUS. see comments in ioremap_nocache()
1508 1509 1510
	 * If you really need strong UC use ioremap_uc(), but note
	 * that you cannot override IO areas with set_memory_*() as
	 * these helpers cannot work with IO memory.
1511
	 */
1512
	return change_page_attr_set(&addr, numpages,
1513 1514
				    cachemode2pgprot(_PAGE_CACHE_MODE_UC_MINUS),
				    0);
1515
}
1516 1517 1518

int set_memory_uc(unsigned long addr, int numpages)
{
1519 1520
	int ret;

1521 1522 1523
	/*
	 * for now UC MINUS. see comments in ioremap_nocache()
	 */
1524
	ret = reserve_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE,
1525
			      _PAGE_CACHE_MODE_UC_MINUS, NULL);
1526 1527 1528 1529 1530 1531 1532 1533
	if (ret)
		goto out_err;

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

	return 0;
1534

1535 1536 1537 1538
out_free:
	free_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE);
out_err:
	return ret;
1539
}
1540 1541
EXPORT_SYMBOL(set_memory_uc);

1542
static int _set_memory_array(unsigned long *addr, int addrinarray,
1543
		enum page_cache_mode new_type)
1544
{
1545
	enum page_cache_mode set_type;
1546 1547 1548
	int i, j;
	int ret;

1549
	for (i = 0; i < addrinarray; i++) {
1550
		ret = reserve_memtype(__pa(addr[i]), __pa(addr[i]) + PAGE_SIZE,
1551
					new_type, NULL);
1552 1553
		if (ret)
			goto out_free;
1554 1555
	}

1556 1557 1558 1559
	/* If WC, set to UC- first and then WC */
	set_type = (new_type == _PAGE_CACHE_MODE_WC) ?
				_PAGE_CACHE_MODE_UC_MINUS : new_type;

1560
	ret = change_page_attr_set(addr, addrinarray,
1561
				   cachemode2pgprot(set_type), 1);
1562

1563
	if (!ret && new_type == _PAGE_CACHE_MODE_WC)
1564
		ret = change_page_attr_set_clr(addr, addrinarray,
1565 1566
					       cachemode2pgprot(
						_PAGE_CACHE_MODE_WC),
1567 1568
					       __pgprot(_PAGE_CACHE_MASK),
					       0, CPA_ARRAY, NULL);
1569 1570 1571 1572 1573 1574 1575 1576 1577 1578
	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;
1579
}
1580 1581 1582

int set_memory_array_uc(unsigned long *addr, int addrinarray)
{
1583
	return _set_memory_array(addr, addrinarray, _PAGE_CACHE_MODE_UC_MINUS);
1584
}
1585 1586
EXPORT_SYMBOL(set_memory_array_uc);

1587 1588
int set_memory_array_wc(unsigned long *addr, int addrinarray)
{
1589
	return _set_memory_array(addr, addrinarray, _PAGE_CACHE_MODE_WC);
1590 1591 1592
}
EXPORT_SYMBOL(set_memory_array_wc);

1593 1594 1595 1596 1597 1598
int set_memory_array_wt(unsigned long *addr, int addrinarray)
{
	return _set_memory_array(addr, addrinarray, _PAGE_CACHE_MODE_WT);
}
EXPORT_SYMBOL_GPL(set_memory_array_wt);

1599 1600
int _set_memory_wc(unsigned long addr, int numpages)
{
1601
	int ret;
1602 1603
	unsigned long addr_copy = addr;

1604
	ret = change_page_attr_set(&addr, numpages,
1605 1606
				   cachemode2pgprot(_PAGE_CACHE_MODE_UC_MINUS),
				   0);
1607
	if (!ret) {
1608
		ret = change_page_attr_set_clr(&addr_copy, numpages,
1609 1610
					       cachemode2pgprot(
						_PAGE_CACHE_MODE_WC),
1611 1612
					       __pgprot(_PAGE_CACHE_MASK),
					       0, 0, NULL);
1613 1614
	}
	return ret;
1615 1616 1617 1618
}

int set_memory_wc(unsigned long addr, int numpages)
{
1619 1620 1621
	int ret;

	ret = reserve_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE,
1622
		_PAGE_CACHE_MODE_WC, NULL);
1623
	if (ret)
1624
		return ret;
1625

1626 1627
	ret = _set_memory_wc(addr, numpages);
	if (ret)
1628
		free_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE);
1629 1630

	return ret;
1631 1632 1633
}
EXPORT_SYMBOL(set_memory_wc);

1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656
int _set_memory_wt(unsigned long addr, int numpages)
{
	return change_page_attr_set(&addr, numpages,
				    cachemode2pgprot(_PAGE_CACHE_MODE_WT), 0);
}

int set_memory_wt(unsigned long addr, int numpages)
{
	int ret;

	ret = reserve_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE,
			      _PAGE_CACHE_MODE_WT, NULL);
	if (ret)
		return ret;

	ret = _set_memory_wt(addr, numpages);
	if (ret)
		free_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE);

	return ret;
}
EXPORT_SYMBOL_GPL(set_memory_wt);

1657
int _set_memory_wb(unsigned long addr, int numpages)
1658
{
1659
	/* WB cache mode is hard wired to all cache attribute bits being 0 */
1660 1661
	return change_page_attr_clear(&addr, numpages,
				      __pgprot(_PAGE_CACHE_MASK), 0);
1662
}
1663 1664 1665

int set_memory_wb(unsigned long addr, int numpages)
{
1666 1667 1668 1669 1670 1671
	int ret;

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

1672
	free_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE);
1673
	return 0;
1674
}
1675 1676
EXPORT_SYMBOL(set_memory_wb);

1677 1678 1679
int set_memory_array_wb(unsigned long *addr, int addrinarray)
{
	int i;
1680 1681
	int ret;

1682
	/* WB cache mode is hard wired to all cache attribute bits being 0 */
1683 1684
	ret = change_page_attr_clear(addr, addrinarray,
				      __pgprot(_PAGE_CACHE_MASK), 1);
1685 1686
	if (ret)
		return ret;
1687

1688 1689
	for (i = 0; i < addrinarray; i++)
		free_memtype(__pa(addr[i]), __pa(addr[i]) + PAGE_SIZE);
1690

1691
	return 0;
1692 1693 1694
}
EXPORT_SYMBOL(set_memory_array_wb);

1695 1696
int set_memory_x(unsigned long addr, int numpages)
{
1697 1698 1699
	if (!(__supported_pte_mask & _PAGE_NX))
		return 0;

1700
	return change_page_attr_clear(&addr, numpages, __pgprot(_PAGE_NX), 0);
1701 1702 1703 1704 1705
}
EXPORT_SYMBOL(set_memory_x);

int set_memory_nx(unsigned long addr, int numpages)
{
1706 1707 1708
	if (!(__supported_pte_mask & _PAGE_NX))
		return 0;

1709
	return change_page_attr_set(&addr, numpages, __pgprot(_PAGE_NX), 0);
1710 1711 1712 1713 1714
}
EXPORT_SYMBOL(set_memory_nx);

int set_memory_ro(unsigned long addr, int numpages)
{
1715
	return change_page_attr_clear(&addr, numpages, __pgprot(_PAGE_RW), 0);
1716 1717 1718 1719
}

int set_memory_rw(unsigned long addr, int numpages)
{
1720
	return change_page_attr_set(&addr, numpages, __pgprot(_PAGE_RW), 0);
1721
}
I
Ingo Molnar 已提交
1722 1723 1724

int set_memory_np(unsigned long addr, int numpages)
{
1725
	return change_page_attr_clear(&addr, numpages, __pgprot(_PAGE_PRESENT), 0);
I
Ingo Molnar 已提交
1726
}
1727

1728 1729
int set_memory_4k(unsigned long addr, int numpages)
{
1730
	return change_page_attr_set_clr(&addr, numpages, __pgprot(0),
1731
					__pgprot(0), 1, 0, NULL);
1732 1733
}

1734 1735 1736 1737
int set_pages_uc(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);

T
Thomas Gleixner 已提交
1738
	return set_memory_uc(addr, numpages);
1739 1740 1741
}
EXPORT_SYMBOL(set_pages_uc);

1742
static int _set_pages_array(struct page **pages, int addrinarray,
1743
		enum page_cache_mode new_type)
1744 1745 1746
{
	unsigned long start;
	unsigned long end;
1747
	enum page_cache_mode set_type;
1748 1749
	int i;
	int free_idx;
1750
	int ret;
1751 1752

	for (i = 0; i < addrinarray; i++) {
1753 1754 1755
		if (PageHighMem(pages[i]))
			continue;
		start = page_to_pfn(pages[i]) << PAGE_SHIFT;
1756
		end = start + PAGE_SIZE;
1757
		if (reserve_memtype(start, end, new_type, NULL))
1758 1759 1760
			goto err_out;
	}

1761 1762 1763 1764
	/* If WC, set to UC- first and then WC */
	set_type = (new_type == _PAGE_CACHE_MODE_WC) ?
				_PAGE_CACHE_MODE_UC_MINUS : new_type;

1765
	ret = cpa_set_pages_array(pages, addrinarray,
1766
				  cachemode2pgprot(set_type));
1767
	if (!ret && new_type == _PAGE_CACHE_MODE_WC)
1768
		ret = change_page_attr_set_clr(NULL, addrinarray,
1769 1770
					       cachemode2pgprot(
						_PAGE_CACHE_MODE_WC),
1771 1772 1773 1774 1775
					       __pgprot(_PAGE_CACHE_MASK),
					       0, CPA_PAGES_ARRAY, pages);
	if (ret)
		goto err_out;
	return 0; /* Success */
1776 1777 1778
err_out:
	free_idx = i;
	for (i = 0; i < free_idx; i++) {
1779 1780 1781
		if (PageHighMem(pages[i]))
			continue;
		start = page_to_pfn(pages[i]) << PAGE_SHIFT;
1782 1783 1784 1785 1786
		end = start + PAGE_SIZE;
		free_memtype(start, end);
	}
	return -EINVAL;
}
1787 1788 1789

int set_pages_array_uc(struct page **pages, int addrinarray)
{
1790
	return _set_pages_array(pages, addrinarray, _PAGE_CACHE_MODE_UC_MINUS);
1791
}
1792 1793
EXPORT_SYMBOL(set_pages_array_uc);

1794 1795
int set_pages_array_wc(struct page **pages, int addrinarray)
{
1796
	return _set_pages_array(pages, addrinarray, _PAGE_CACHE_MODE_WC);
1797 1798 1799
}
EXPORT_SYMBOL(set_pages_array_wc);

1800 1801 1802 1803 1804 1805
int set_pages_array_wt(struct page **pages, int addrinarray)
{
	return _set_pages_array(pages, addrinarray, _PAGE_CACHE_MODE_WT);
}
EXPORT_SYMBOL_GPL(set_pages_array_wt);

1806 1807 1808 1809
int set_pages_wb(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);

T
Thomas Gleixner 已提交
1810
	return set_memory_wb(addr, numpages);
1811 1812 1813
}
EXPORT_SYMBOL(set_pages_wb);

1814 1815 1816 1817 1818 1819 1820
int set_pages_array_wb(struct page **pages, int addrinarray)
{
	int retval;
	unsigned long start;
	unsigned long end;
	int i;

1821
	/* WB cache mode is hard wired to all cache attribute bits being 0 */
1822 1823
	retval = cpa_clear_pages_array(pages, addrinarray,
			__pgprot(_PAGE_CACHE_MASK));
1824 1825
	if (retval)
		return retval;
1826 1827

	for (i = 0; i < addrinarray; i++) {
1828 1829 1830
		if (PageHighMem(pages[i]))
			continue;
		start = page_to_pfn(pages[i]) << PAGE_SHIFT;
1831 1832 1833 1834
		end = start + PAGE_SIZE;
		free_memtype(start, end);
	}

1835
	return 0;
1836 1837 1838
}
EXPORT_SYMBOL(set_pages_array_wb);

1839 1840 1841 1842
int set_pages_x(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);

T
Thomas Gleixner 已提交
1843
	return set_memory_x(addr, numpages);
1844 1845 1846 1847 1848 1849 1850
}
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
1851
	return set_memory_nx(addr, numpages);
1852 1853 1854 1855 1856 1857 1858
}
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
1859
	return set_memory_ro(addr, numpages);
1860 1861 1862 1863 1864
}

int set_pages_rw(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);
1865

T
Thomas Gleixner 已提交
1866
	return set_memory_rw(addr, numpages);
I
Ingo Molnar 已提交
1867 1868
}

L
Linus Torvalds 已提交
1869
#ifdef CONFIG_DEBUG_PAGEALLOC
I
Ingo Molnar 已提交
1870 1871 1872

static int __set_pages_p(struct page *page, int numpages)
{
1873 1874
	unsigned long tempaddr = (unsigned long) page_address(page);
	struct cpa_data cpa = { .vaddr = &tempaddr,
1875
				.pgd = NULL,
T
Thomas Gleixner 已提交
1876 1877
				.numpages = numpages,
				.mask_set = __pgprot(_PAGE_PRESENT | _PAGE_RW),
1878 1879
				.mask_clr = __pgprot(0),
				.flags = 0};
1880

1881 1882 1883 1884 1885 1886 1887
	/*
	 * 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
1888 1889 1890 1891
}

static int __set_pages_np(struct page *page, int numpages)
{
1892 1893
	unsigned long tempaddr = (unsigned long) page_address(page);
	struct cpa_data cpa = { .vaddr = &tempaddr,
1894
				.pgd = NULL,
T
Thomas Gleixner 已提交
1895 1896
				.numpages = numpages,
				.mask_set = __pgprot(0),
1897 1898
				.mask_clr = __pgprot(_PAGE_PRESENT | _PAGE_RW),
				.flags = 0};
1899

1900 1901 1902 1903 1904 1905 1906
	/*
	 * 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
1907 1908
}

1909
void __kernel_map_pages(struct page *page, int numpages, int enable)
L
Linus Torvalds 已提交
1910 1911 1912
{
	if (PageHighMem(page))
		return;
1913
	if (!enable) {
1914 1915
		debug_check_no_locks_freed(page_address(page),
					   numpages * PAGE_SIZE);
1916
	}
1917

1918
	/*
I
Ingo Molnar 已提交
1919
	 * The return value is ignored as the calls cannot fail.
1920 1921
	 * 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
1922
	 */
I
Ingo Molnar 已提交
1923 1924 1925 1926
	if (enable)
		__set_pages_p(page, numpages);
	else
		__set_pages_np(page, numpages);
1927 1928

	/*
1929 1930
	 * We should perform an IPI and flush all tlbs,
	 * but that can deadlock->flush only current cpu:
L
Linus Torvalds 已提交
1931 1932
	 */
	__flush_tlb_all();
1933 1934

	arch_flush_lazy_mmu_mode();
1935 1936
}

1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953
#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 */
1954

1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975
int kernel_map_pages_in_pgd(pgd_t *pgd, u64 pfn, unsigned long address,
			    unsigned numpages, unsigned long page_flags)
{
	int retval = -EINVAL;

	struct cpa_data cpa = {
		.vaddr = &address,
		.pfn = pfn,
		.pgd = pgd,
		.numpages = numpages,
		.mask_set = __pgprot(0),
		.mask_clr = __pgprot(0),
		.flags = 0,
	};

	if (!(__supported_pte_mask & _PAGE_NX))
		goto out;

	if (!(page_flags & _PAGE_NX))
		cpa.mask_clr = __pgprot(_PAGE_NX);

1976 1977 1978
	if (!(page_flags & _PAGE_RW))
		cpa.mask_clr = __pgprot(_PAGE_RW);

1979 1980 1981 1982 1983 1984 1985 1986 1987
	cpa.mask_set = __pgprot(_PAGE_PRESENT | page_flags);

	retval = __change_page_attr_set_clr(&cpa, 0);
	__flush_tlb_all();

out:
	return retval;
}

1988 1989 1990 1991 1992 1993
void kernel_unmap_pages_in_pgd(pgd_t *root, unsigned long address,
			       unsigned numpages)
{
	unmap_pgd_range(root, address, address + (numpages << PAGE_SHIFT));
}

1994 1995 1996 1997 1998 1999 2000
/*
 * 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