pageattr.c 58.5 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>
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/api.h>
L
Linus Torvalds 已提交
19 20
#include <asm/processor.h>
#include <asm/tlbflush.h>
21
#include <asm/sections.h>
22
#include <asm/setup.h>
23
#include <linux/uaccess.h>
24
#include <asm/pgalloc.h>
25
#include <asm/proto.h>
26
#include <asm/pat.h>
27
#include <asm/set_memory.h>
L
Linus Torvalds 已提交
28

I
Ingo Molnar 已提交
29 30 31
/*
 * The current flushing context - we pass it instead of 5 arguments:
 */
32
struct cpa_data {
33
	unsigned long	*vaddr;
34
	pgd_t		*pgd;
35 36
	pgprot_t	mask_set;
	pgprot_t	mask_clr;
37
	unsigned long	numpages;
38
	int		flags;
39
	unsigned long	pfn;
40 41
	unsigned	force_split		: 1,
			force_static_prot	: 1;
42
	int		curpage;
43
	struct page	**pages;
44 45
};

46
enum cpa_warn {
47
	CPA_CONFLICT,
48 49 50 51 52 53
	CPA_PROTECT,
	CPA_DETECT,
};

static const int cpa_warn_level = CPA_PROTECT;

54 55 56 57 58 59 60 61
/*
 * 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);

62 63
#define CPA_FLUSHTLB 1
#define CPA_ARRAY 2
64
#define CPA_PAGES_ARRAY 4
65
#define CPA_NO_CHECK_ALIAS 8 /* Do not search for aliases */
66

67
#ifdef CONFIG_PROC_FS
68 69
static unsigned long direct_pages_count[PG_LEVEL_NUM];

70
void update_page_count(int level, unsigned long pages)
71 72
{
	/* Protect against CPA */
73
	spin_lock(&pgd_lock);
74
	direct_pages_count[level] += pages;
75
	spin_unlock(&pgd_lock);
76 77 78 79
}

static void split_page_count(int level)
{
80 81 82
	if (direct_pages_count[level] == 0)
		return;

83 84 85 86
	direct_pages_count[level]--;
	direct_pages_count[level - 1] += PTRS_PER_PTE;
}

87
void arch_report_meminfo(struct seq_file *m)
88
{
89
	seq_printf(m, "DirectMap4k:    %8lu kB\n",
90 91
			direct_pages_count[PG_LEVEL_4K] << 2);
#if defined(CONFIG_X86_64) || defined(CONFIG_X86_PAE)
92
	seq_printf(m, "DirectMap2M:    %8lu kB\n",
93 94
			direct_pages_count[PG_LEVEL_2M] << 11);
#else
95
	seq_printf(m, "DirectMap4M:    %8lu kB\n",
96 97 98
			direct_pages_count[PG_LEVEL_2M] << 12);
#endif
	if (direct_gbpages)
99
		seq_printf(m, "DirectMap1G:    %8lu kB\n",
100
			direct_pages_count[PG_LEVEL_1G] << 20);
101
}
102 103 104
#else
static inline void split_page_count(int level) { }
#endif
105

106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186
#ifdef CONFIG_X86_CPA_STATISTICS

static unsigned long cpa_1g_checked;
static unsigned long cpa_1g_sameprot;
static unsigned long cpa_1g_preserved;
static unsigned long cpa_2m_checked;
static unsigned long cpa_2m_sameprot;
static unsigned long cpa_2m_preserved;
static unsigned long cpa_4k_install;

static inline void cpa_inc_1g_checked(void)
{
	cpa_1g_checked++;
}

static inline void cpa_inc_2m_checked(void)
{
	cpa_2m_checked++;
}

static inline void cpa_inc_4k_install(void)
{
	cpa_4k_install++;
}

static inline void cpa_inc_lp_sameprot(int level)
{
	if (level == PG_LEVEL_1G)
		cpa_1g_sameprot++;
	else
		cpa_2m_sameprot++;
}

static inline void cpa_inc_lp_preserved(int level)
{
	if (level == PG_LEVEL_1G)
		cpa_1g_preserved++;
	else
		cpa_2m_preserved++;
}

static int cpastats_show(struct seq_file *m, void *p)
{
	seq_printf(m, "1G pages checked:     %16lu\n", cpa_1g_checked);
	seq_printf(m, "1G pages sameprot:    %16lu\n", cpa_1g_sameprot);
	seq_printf(m, "1G pages preserved:   %16lu\n", cpa_1g_preserved);
	seq_printf(m, "2M pages checked:     %16lu\n", cpa_2m_checked);
	seq_printf(m, "2M pages sameprot:    %16lu\n", cpa_2m_sameprot);
	seq_printf(m, "2M pages preserved:   %16lu\n", cpa_2m_preserved);
	seq_printf(m, "4K pages set-checked: %16lu\n", cpa_4k_install);
	return 0;
}

static int cpastats_open(struct inode *inode, struct file *file)
{
	return single_open(file, cpastats_show, NULL);
}

static const struct file_operations cpastats_fops = {
	.open		= cpastats_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
	.release	= single_release,
};

static int __init cpa_stats_init(void)
{
	debugfs_create_file("cpa_stats", S_IRUSR, arch_debugfs_dir, NULL,
			    &cpastats_fops);
	return 0;
}
late_initcall(cpa_stats_init);
#else
static inline void cpa_inc_1g_checked(void) { }
static inline void cpa_inc_2m_checked(void) { }
static inline void cpa_inc_4k_install(void) { }
static inline void cpa_inc_lp_sameprot(int level) { }
static inline void cpa_inc_lp_preserved(int level) { }
#endif


187 188 189 190 191 192 193 194 195 196 197 198
static inline int
within(unsigned long addr, unsigned long start, unsigned long end)
{
	return addr >= start && addr < end;
}

static inline int
within_inclusive(unsigned long addr, unsigned long start, unsigned long end)
{
	return addr >= start && addr <= end;
}

199 200 201 202
#ifdef CONFIG_X86_64

static inline unsigned long highmap_start_pfn(void)
{
203
	return __pa_symbol(_text) >> PAGE_SHIFT;
204 205 206 207
}

static inline unsigned long highmap_end_pfn(void)
{
208 209
	/* Do not reference physical address outside the kernel. */
	return __pa_symbol(roundup(_brk_end, PMD_SIZE) - 1) >> PAGE_SHIFT;
210 211
}

212
static bool __cpa_pfn_in_highmap(unsigned long pfn)
213
{
214 215 216 217 218
	/*
	 * Kernel text has an alias mapping at a high address, known
	 * here as "highmap".
	 */
	return within_inclusive(pfn, highmap_start_pfn(), highmap_end_pfn());
219 220
}

221 222 223
#else

static bool __cpa_pfn_in_highmap(unsigned long pfn)
224
{
225 226
	/* There is no highmap on 32-bit */
	return false;
227 228
}

229 230
#endif

231 232 233
/*
 * Flushing functions
 */
234 235 236

/**
 * clflush_cache_range - flush a cache range with clflush
237
 * @vaddr:	virtual start address
238 239
 * @size:	number of bytes to flush
 *
240 241
 * clflushopt is an unordered instruction which needs fencing with mfence or
 * sfence to avoid ordering issues.
242
 */
243
void clflush_cache_range(void *vaddr, unsigned int size)
244
{
245 246
	const unsigned long clflush_size = boot_cpu_data.x86_clflush_size;
	void *p = (void *)((unsigned long)vaddr & ~(clflush_size - 1));
247
	void *vend = vaddr + size;
248 249 250

	if (p >= vend)
		return;
251

252
	mb();
253

254
	for (; p < vend; p += clflush_size)
255
		clflushopt(p);
256

257
	mb();
258
}
259
EXPORT_SYMBOL_GPL(clflush_cache_range);
260

261 262 263 264 265 266
void arch_invalidate_pmem(void *addr, size_t size)
{
	clflush_cache_range(addr, size);
}
EXPORT_SYMBOL_GPL(arch_invalidate_pmem);

267
static void __cpa_flush_all(void *arg)
268
{
269 270
	unsigned long cache = (unsigned long)arg;

271 272 273 274 275 276
	/*
	 * Flush all to work around Errata in early athlons regarding
	 * large page flushing.
	 */
	__flush_tlb_all();

277
	if (cache && boot_cpu_data.x86 >= 4)
278 279 280
		wbinvd();
}

281
static void cpa_flush_all(unsigned long cache)
282
{
283
	BUG_ON(irqs_disabled() && !early_boot_irqs_disabled);
284

285
	on_each_cpu(__cpa_flush_all, (void *) cache, 1);
286 287
}

288 289 290 291 292 293 294 295 296 297
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();
}

298
static void cpa_flush_range(unsigned long start, int numpages, int cache)
299
{
300 301 302
	unsigned int i, level;
	unsigned long addr;

303
	BUG_ON(irqs_disabled() && !early_boot_irqs_disabled);
304
	WARN_ON(PAGE_ALIGN(start) != start);
305

306
	on_each_cpu(__cpa_flush_range, NULL, 1);
307

308 309 310
	if (!cache)
		return;

T
Thomas Gleixner 已提交
311 312 313 314 315 316
	/*
	 * 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:
	 */
317 318 319 320 321 322
	for (i = 0, addr = start; i < numpages; i++, addr += PAGE_SIZE) {
		pte_t *pte = lookup_address(addr, &level);

		/*
		 * Only flush present addresses:
		 */
323
		if (pte && (pte_val(*pte) & _PAGE_PRESENT))
324 325
			clflush_cache_range((void *) addr, PAGE_SIZE);
	}
326 327
}

328 329
static void cpa_flush_array(unsigned long *start, int numpages, int cache,
			    int in_flags, struct page **pages)
330 331
{
	unsigned int i, level;
332 333 334 335 336 337 338 339 340 341 342 343
#ifdef CONFIG_PREEMPT
	/*
	 * Avoid wbinvd() because it causes latencies on all CPUs,
	 * regardless of any CPU isolation that may be in effect.
	 *
	 * This should be extended for CAT enabled systems independent of
	 * PREEMPT because wbinvd() does not respect the CAT partitions and
	 * this is exposed to unpriviledged users through the graphics
	 * subsystem.
	 */
	unsigned long do_wbinvd = 0;
#else
344
	unsigned long do_wbinvd = cache && numpages >= 1024; /* 4M threshold */
345
#endif
346

347
	BUG_ON(irqs_disabled() && !early_boot_irqs_disabled);
348

349
	on_each_cpu(__cpa_flush_all, (void *) do_wbinvd, 1);
350

351
	if (!cache || do_wbinvd)
352 353 354 355 356 357 358 359
		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:
	 */
360 361 362 363 364 365 366 367 368 369
	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);
370 371 372 373 374

		/*
		 * Only flush present addresses:
		 */
		if (pte && (pte_val(*pte) & _PAGE_PRESENT))
375
			clflush_cache_range((void *)addr, PAGE_SIZE);
376 377 378
	}
}

379 380 381 382 383 384 385
static bool overlaps(unsigned long r1_start, unsigned long r1_end,
		     unsigned long r2_start, unsigned long r2_end)
{
	return (r1_start <= r2_end && r1_end >= r2_start) ||
		(r2_start <= r1_end && r2_end >= r1_start);
}

386
#ifdef CONFIG_PCI_BIOS
387
/*
388 389
 * The BIOS area between 640k and 1Mb needs to be executable for PCI BIOS
 * based config access (CONFIG_PCI_GOBIOS) support.
390
 */
391
#define BIOS_PFN	PFN_DOWN(BIOS_BEGIN)
392
#define BIOS_PFN_END	PFN_DOWN(BIOS_END - 1)
393

394
static pgprotval_t protect_pci_bios(unsigned long spfn, unsigned long epfn)
395
{
396
	if (pcibios_enabled && overlaps(spfn, epfn, BIOS_PFN, BIOS_PFN_END))
397 398 399 400
		return _PAGE_NX;
	return 0;
}
#else
401
static pgprotval_t protect_pci_bios(unsigned long spfn, unsigned long epfn)
402 403 404
{
	return 0;
}
405
#endif
406

407 408 409 410 411
/*
 * The .rodata section needs to be read-only. Using the pfn catches all
 * aliases.  This also includes __ro_after_init, so do not enforce until
 * kernel_set_to_readonly is true.
 */
412
static pgprotval_t protect_rodata(unsigned long spfn, unsigned long epfn)
413
{
414 415 416 417 418 419 420
	unsigned long epfn_ro, spfn_ro = PFN_DOWN(__pa_symbol(__start_rodata));

	/*
	 * Note: __end_rodata is at page aligned and not inclusive, so
	 * subtract 1 to get the last enforced PFN in the rodata area.
	 */
	epfn_ro = PFN_DOWN(__pa_symbol(__end_rodata)) - 1;
421

422
	if (kernel_set_to_readonly && overlaps(spfn, epfn, spfn_ro, epfn_ro))
423 424 425 426 427 428 429 430 431 432 433 434
		return _PAGE_RW;
	return 0;
}

/*
 * Protect kernel text against becoming non executable by forbidding
 * _PAGE_NX.  This protects only the high kernel mapping (_text -> _etext)
 * out of which the kernel actually executes.  Do not protect the low
 * mapping.
 *
 * This does not cover __inittext since that is gone after boot.
 */
435
static pgprotval_t protect_kernel_text(unsigned long start, unsigned long end)
436
{
437 438 439 440
	unsigned long t_end = (unsigned long)_etext - 1;
	unsigned long t_start = (unsigned long)_text;

	if (overlaps(start, end, t_start, t_end))
441 442 443
		return _PAGE_NX;
	return 0;
}
444

445
#if defined(CONFIG_X86_64)
446 447 448 449 450 451 452 453 454
/*
 * 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.
 *
 * This will preserve the large page mappings for kernel text/data at no
 * extra cost.
 */
455 456
static pgprotval_t protect_kernel_text_ro(unsigned long start,
					  unsigned long end)
457
{
458 459
	unsigned long t_end = (unsigned long)__end_rodata_hpage_align - 1;
	unsigned long t_start = (unsigned long)_text;
460 461
	unsigned int level;

462
	if (!kernel_set_to_readonly || !overlaps(start, end, t_start, t_end))
463
		return 0;
464
	/*
465 466 467
	 * 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.
468
	 *
469 470 471 472 473 474
	 * This also fixes the Linux Xen paravirt guest boot failure caused
	 * by 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,
	 * so the protections for kernel text and identity mappings have to
	 * be the same.
475
	 */
476
	if (lookup_address(start, &level) && (level != PG_LEVEL_4K))
477 478 479 480
		return _PAGE_RW;
	return 0;
}
#else
481 482
static pgprotval_t protect_kernel_text_ro(unsigned long start,
					  unsigned long end)
483 484 485
{
	return 0;
}
486 487
#endif

488 489 490 491 492 493 494 495 496 497
static inline bool conflicts(pgprot_t prot, pgprotval_t val)
{
	return (pgprot_val(prot) & ~val) != pgprot_val(prot);
}

static inline void check_conflict(int warnlvl, pgprot_t prot, pgprotval_t val,
				  unsigned long start, unsigned long end,
				  unsigned long pfn, const char *txt)
{
	static const char *lvltxt[] = {
498
		[CPA_CONFLICT]	= "conflict",
499 500 501 502 503 504 505 506 507 508 509 510
		[CPA_PROTECT]	= "protect",
		[CPA_DETECT]	= "detect",
	};

	if (warnlvl > cpa_warn_level || !conflicts(prot, val))
		return;

	pr_warn("CPA %8s %10s: 0x%016lx - 0x%016lx PFN %lx req %016llx prevent %016llx\n",
		lvltxt[warnlvl], txt, start, end, pfn, (unsigned long long)pgprot_val(prot),
		(unsigned long long)val);
}

511 512 513 514 515 516
/*
 * 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.
 */
517
static inline pgprot_t static_protections(pgprot_t prot, unsigned long start,
518 519
					  unsigned long pfn, unsigned long npg,
					  int warnlvl)
520
{
521
	pgprotval_t forbidden, res;
522
	unsigned long end;
523

524 525 526 527 528 529 530
	/*
	 * There is no point in checking RW/NX conflicts when the requested
	 * mapping is setting the page !PRESENT.
	 */
	if (!(pgprot_val(prot) & _PAGE_PRESENT))
		return prot;

531
	/* Operate on the virtual address */
532
	end = start + npg * PAGE_SIZE - 1;
533 534 535 536 537 538 539 540

	res = protect_kernel_text(start, end);
	check_conflict(warnlvl, prot, res, start, end, pfn, "Text NX");
	forbidden = res;

	res = protect_kernel_text_ro(start, end);
	check_conflict(warnlvl, prot, res, start, end, pfn, "Text RO");
	forbidden |= res;
541 542

	/* Check the PFN directly */
543 544 545 546 547 548 549
	res = protect_pci_bios(pfn, pfn + npg - 1);
	check_conflict(warnlvl, prot, res, start, end, pfn, "PCIBIOS NX");
	forbidden |= res;

	res = protect_rodata(pfn, pfn + npg - 1);
	check_conflict(warnlvl, prot, res, start, end, pfn, "Rodata RO");
	forbidden |= res;
550

551
	return __pgprot(pgprot_val(prot) & ~forbidden);
552 553
}

554 555 556 557 558 559
/*
 * 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)
560
{
561
	p4d_t *p4d;
L
Linus Torvalds 已提交
562 563
	pud_t *pud;
	pmd_t *pmd;
564

T
Thomas Gleixner 已提交
565 566
	*level = PG_LEVEL_NONE;

L
Linus Torvalds 已提交
567 568
	if (pgd_none(*pgd))
		return NULL;
I
Ingo Molnar 已提交
569

570 571 572 573 574 575 576 577 578
	p4d = p4d_offset(pgd, address);
	if (p4d_none(*p4d))
		return NULL;

	*level = PG_LEVEL_512G;
	if (p4d_large(*p4d) || !p4d_present(*p4d))
		return (pte_t *)p4d;

	pud = pud_offset(p4d, address);
L
Linus Torvalds 已提交
579 580
	if (pud_none(*pud))
		return NULL;
581 582 583 584 585

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

L
Linus Torvalds 已提交
586 587 588
	pmd = pmd_offset(pud, address);
	if (pmd_none(*pmd))
		return NULL;
T
Thomas Gleixner 已提交
589 590

	*level = PG_LEVEL_2M;
591
	if (pmd_large(*pmd) || !pmd_present(*pmd))
L
Linus Torvalds 已提交
592 593
		return (pte_t *)pmd;

T
Thomas Gleixner 已提交
594
	*level = PG_LEVEL_4K;
I
Ingo Molnar 已提交
595

596 597
	return pte_offset_kernel(pmd, address);
}
598 599 600 601 602 603 604 605 606 607 608

/*
 * 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)
{
609
	return lookup_address_in_pgd(pgd_offset_k(address), address, level);
610
}
611
EXPORT_SYMBOL_GPL(lookup_address);
612

613 614 615
static pte_t *_lookup_address_cpa(struct cpa_data *cpa, unsigned long address,
				  unsigned int *level)
{
616
	if (cpa->pgd)
617
		return lookup_address_in_pgd(cpa->pgd + pgd_index(address),
618 619
					       address, level);

620
	return lookup_address(address, level);
621 622
}

623 624 625 626 627 628 629
/*
 * 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;
630
	p4d_t *p4d;
631 632 633 634 635 636
	pud_t *pud;

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

637 638 639 640 641
	p4d = p4d_offset(pgd, address);
	if (p4d_none(*p4d) || p4d_large(*p4d) || !p4d_present(*p4d))
		return NULL;

	pud = pud_offset(p4d, address);
642 643 644 645 646 647
	if (pud_none(*pud) || pud_large(*pud) || !pud_present(*pud))
		return NULL;

	return pmd_offset(pud, address);
}

648 649 650 651 652 653 654 655 656 657 658 659 660 661
/*
 * 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;
662 663
	phys_addr_t phys_addr;
	unsigned long offset;
664 665 666 667 668
	enum pg_level level;
	pte_t *pte;

	pte = lookup_address(virt_addr, &level);
	BUG_ON(!pte);
669

670 671 672 673 674
	/*
	 * 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.
	 */
675 676
	switch (level) {
	case PG_LEVEL_1G:
677
		phys_addr = (phys_addr_t)pud_pfn(*(pud_t *)pte) << PAGE_SHIFT;
678 679 680
		offset = virt_addr & ~PUD_PAGE_MASK;
		break;
	case PG_LEVEL_2M:
681
		phys_addr = (phys_addr_t)pmd_pfn(*(pmd_t *)pte) << PAGE_SHIFT;
682 683 684
		offset = virt_addr & ~PMD_PAGE_MASK;
		break;
	default:
685
		phys_addr = (phys_addr_t)pte_pfn(*pte) << PAGE_SHIFT;
686 687 688 689
		offset = virt_addr & ~PAGE_MASK;
	}

	return (phys_addr_t)(phys_addr | offset);
690 691 692
}
EXPORT_SYMBOL_GPL(slow_virt_to_phys);

I
Ingo Molnar 已提交
693 694 695
/*
 * Set the new pmd in all the pgds we know about:
 */
I
Ingo Molnar 已提交
696
static void __set_pmd_pte(pte_t *kpte, unsigned long address, pte_t pte)
697 698 699
{
	/* change init_mm */
	set_pte_atomic(kpte, pte);
700
#ifdef CONFIG_X86_32
701
	if (!SHARED_KERNEL_PMD) {
702 703
		struct page *page;

704
		list_for_each_entry(page, &pgd_list, lru) {
705
			pgd_t *pgd;
706
			p4d_t *p4d;
707 708 709 710
			pud_t *pud;
			pmd_t *pmd;

			pgd = (pgd_t *)page_address(page) + pgd_index(address);
711 712
			p4d = p4d_offset(pgd, address);
			pud = pud_offset(p4d, address);
713 714 715
			pmd = pmd_offset(pud, address);
			set_pte_atomic((pte_t *)pmd, pte);
		}
L
Linus Torvalds 已提交
716
	}
717
#endif
L
Linus Torvalds 已提交
718 719
}

720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736
static pgprot_t pgprot_clear_protnone_bits(pgprot_t prot)
{
	/*
	 * _PAGE_GLOBAL means "global page" for present PTEs.
	 * But, it is also used to indicate _PAGE_PROTNONE
	 * for non-present PTEs.
	 *
	 * This ensures that a _PAGE_GLOBAL PTE going from
	 * present to non-present is not confused as
	 * _PAGE_PROTNONE.
	 */
	if (!(pgprot_val(prot) & _PAGE_PRESENT))
		pgprot_val(prot) &= ~_PAGE_GLOBAL;

	return prot;
}

737 738
static int __should_split_large_page(pte_t *kpte, unsigned long address,
				     struct cpa_data *cpa)
739
{
740
	unsigned long numpages, pmask, psize, lpaddr, pfn, old_pfn;
741
	pgprot_t old_prot, new_prot, req_prot, chk_prot;
742
	pte_t new_pte, *tmp;
743
	enum pg_level level;
744 745 746 747 748

	/*
	 * Check for races, another CPU might have split this page
	 * up already:
	 */
749
	tmp = _lookup_address_cpa(cpa, address, &level);
750
	if (tmp != kpte)
751
		return 1;
752 753 754

	switch (level) {
	case PG_LEVEL_2M:
755 756
		old_prot = pmd_pgprot(*(pmd_t *)kpte);
		old_pfn = pmd_pfn(*(pmd_t *)kpte);
757
		cpa_inc_2m_checked();
758
		break;
759
	case PG_LEVEL_1G:
760 761
		old_prot = pud_pgprot(*(pud_t *)kpte);
		old_pfn = pud_pfn(*(pud_t *)kpte);
762
		cpa_inc_1g_checked();
763
		break;
764
	default:
765
		return -EINVAL;
766 767
	}

768 769 770
	psize = page_level_size(level);
	pmask = page_level_mask(level);

771 772 773 774
	/*
	 * Calculate the number of pages, which fit into this large
	 * page starting at address:
	 */
775 776
	lpaddr = (address + psize) & pmask;
	numpages = (lpaddr - address) >> PAGE_SHIFT;
777 778
	if (numpages < cpa->numpages)
		cpa->numpages = numpages;
779 780 781

	/*
	 * We are safe now. Check whether the new pgprot is the same:
782 783
	 * Convert protection attributes to 4k-format, as cpa->mask* are set
	 * up accordingly.
784
	 */
785

786
	/* Clear PSE (aka _PAGE_PAT) and move PAT bit to correct position */
787
	req_prot = pgprot_large_2_4k(old_prot);
788

789 790
	pgprot_val(req_prot) &= ~pgprot_val(cpa->mask_clr);
	pgprot_val(req_prot) |= pgprot_val(cpa->mask_set);
791

792 793 794 795 796 797
	/*
	 * 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);
798
	req_prot = pgprot_clear_protnone_bits(req_prot);
799
	if (pgprot_val(req_prot) & _PAGE_PRESENT)
800
		pgprot_val(req_prot) |= _PAGE_PSE;
801

802
	/*
803 804
	 * old_pfn points to the large page base pfn. So we need to add the
	 * offset of the virtual address:
805
	 */
806
	pfn = old_pfn + ((address & (psize - 1)) >> PAGE_SHIFT);
807 808
	cpa->pfn = pfn;

809 810 811 812 813 814
	/*
	 * Calculate the large page base address and the number of 4K pages
	 * in the large page
	 */
	lpaddr = address & pmask;
	numpages = psize >> PAGE_SHIFT;
815

816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832
	/*
	 * Sanity check that the existing mapping is correct versus the static
	 * protections. static_protections() guards against !PRESENT, so no
	 * extra conditional required here.
	 */
	chk_prot = static_protections(old_prot, lpaddr, old_pfn, numpages,
				      CPA_CONFLICT);

	if (WARN_ON_ONCE(pgprot_val(chk_prot) != pgprot_val(old_prot))) {
		/*
		 * Split the large page and tell the split code to
		 * enforce static protections.
		 */
		cpa->force_static_prot = 1;
		return 1;
	}

833 834 835 836 837 838 839 840 841 842 843 844 845 846
	/*
	 * Optimization: If the requested pgprot is the same as the current
	 * pgprot, then the large page can be preserved and no updates are
	 * required independent of alignment and length of the requested
	 * range. The above already established that the current pgprot is
	 * correct, which in consequence makes the requested pgprot correct
	 * as well if it is the same. The static protection scan below will
	 * not come to a different conclusion.
	 */
	if (pgprot_val(req_prot) == pgprot_val(old_prot)) {
		cpa_inc_lp_sameprot(level);
		return 0;
	}

847
	/*
848
	 * If the requested range does not cover the full page, split it up
849
	 */
850 851
	if (address != lpaddr || cpa->numpages != numpages)
		return 1;
852 853

	/*
854 855
	 * Check whether the requested pgprot is conflicting with a static
	 * protection requirement in the large page.
856
	 */
857 858
	new_prot = static_protections(req_prot, lpaddr, old_pfn, numpages,
				      CPA_DETECT);
859 860

	/*
861 862 863 864 865 866 867
	 * If there is a conflict, split the large page.
	 *
	 * There used to be a 4k wise evaluation trying really hard to
	 * preserve the large pages, but experimentation has shown, that this
	 * does not help at all. There might be corner cases which would
	 * preserve one large page occasionally, but it's really not worth the
	 * extra code and cycles for the common case.
868
	 */
869
	if (pgprot_val(req_prot) != pgprot_val(new_prot))
870 871 872 873 874 875
		return 1;

	/* All checks passed. Update the large page mapping. */
	new_pte = pfn_pte(old_pfn, new_prot);
	__set_pmd_pte(kpte, address, new_pte);
	cpa->flags |= CPA_FLUSHTLB;
876
	cpa_inc_lp_preserved(level);
877 878 879 880 881 882 883 884 885 886
	return 0;
}

static int should_split_large_page(pte_t *kpte, unsigned long address,
				   struct cpa_data *cpa)
{
	int do_split;

	if (cpa->force_split)
		return 1;
887

888 889
	spin_lock(&pgd_lock);
	do_split = __should_split_large_page(kpte, address, cpa);
890
	spin_unlock(&pgd_lock);
I
Ingo Molnar 已提交
891

892
	return do_split;
893 894
}

895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929
static void split_set_pte(struct cpa_data *cpa, pte_t *pte, unsigned long pfn,
			  pgprot_t ref_prot, unsigned long address,
			  unsigned long size)
{
	unsigned int npg = PFN_DOWN(size);
	pgprot_t prot;

	/*
	 * If should_split_large_page() discovered an inconsistent mapping,
	 * remove the invalid protection in the split mapping.
	 */
	if (!cpa->force_static_prot)
		goto set;

	prot = static_protections(ref_prot, address, pfn, npg, CPA_PROTECT);

	if (pgprot_val(prot) == pgprot_val(ref_prot))
		goto set;

	/*
	 * If this is splitting a PMD, fix it up. PUD splits cannot be
	 * fixed trivially as that would require to rescan the newly
	 * installed PMD mappings after returning from split_large_page()
	 * so an eventual further split can allocate the necessary PTE
	 * pages. Warn for now and revisit it in case this actually
	 * happens.
	 */
	if (size == PAGE_SIZE)
		ref_prot = prot;
	else
		pr_warn_once("CPA: Cannot fixup static protections for PUD split\n");
set:
	set_pte(pte, pfn_pte(pfn, ref_prot));
}

930
static int
931 932
__split_large_page(struct cpa_data *cpa, pte_t *kpte, unsigned long address,
		   struct page *base)
933
{
934
	unsigned long lpaddr, lpinc, ref_pfn, pfn, pfninc = 1;
935
	pte_t *pbase = (pte_t *)page_address(base);
I
Ingo Molnar 已提交
936 937
	unsigned int i, level;
	pgprot_t ref_prot;
938
	pte_t *tmp;
939

940
	spin_lock(&pgd_lock);
941 942 943 944
	/*
	 * Check for races, another CPU might have split this page
	 * up for us already:
	 */
945
	tmp = _lookup_address_cpa(cpa, address, &level);
946 947 948 949
	if (tmp != kpte) {
		spin_unlock(&pgd_lock);
		return 1;
	}
950

951
	paravirt_alloc_pte(&init_mm, page_to_pfn(base));
952

953 954 955
	switch (level) {
	case PG_LEVEL_2M:
		ref_prot = pmd_pgprot(*(pmd_t *)kpte);
956 957 958 959
		/*
		 * Clear PSE (aka _PAGE_PAT) and move
		 * PAT bit to correct position.
		 */
960
		ref_prot = pgprot_large_2_4k(ref_prot);
961
		ref_pfn = pmd_pfn(*(pmd_t *)kpte);
962 963
		lpaddr = address & PMD_MASK;
		lpinc = PAGE_SIZE;
964
		break;
965

966 967 968
	case PG_LEVEL_1G:
		ref_prot = pud_pgprot(*(pud_t *)kpte);
		ref_pfn = pud_pfn(*(pud_t *)kpte);
969
		pfninc = PMD_PAGE_SIZE >> PAGE_SHIFT;
970 971
		lpaddr = address & PUD_MASK;
		lpinc = PMD_SIZE;
972
		/*
973
		 * Clear the PSE flags if the PRESENT flag is not set
974 975 976
		 * otherwise pmd_present/pmd_huge will return true
		 * even on a non present pmd.
		 */
977
		if (!(pgprot_val(ref_prot) & _PAGE_PRESENT))
978
			pgprot_val(ref_prot) &= ~_PAGE_PSE;
979 980 981 982 983
		break;

	default:
		spin_unlock(&pgd_lock);
		return 1;
984 985
	}

986
	ref_prot = pgprot_clear_protnone_bits(ref_prot);
987

988 989 990
	/*
	 * Get the target pfn from the original entry:
	 */
991
	pfn = ref_pfn;
992 993
	for (i = 0; i < PTRS_PER_PTE; i++, pfn += pfninc, lpaddr += lpinc)
		split_set_pte(cpa, pbase + i, pfn, ref_prot, lpaddr, lpinc);
994

995 996 997 998 999 1000
	if (virt_addr_valid(address)) {
		unsigned long pfn = PFN_DOWN(__pa(address));

		if (pfn_range_is_mapped(pfn, pfn + 1))
			split_page_count(level);
	}
1001

1002
	/*
1003
	 * Install the new, split up pagetable.
1004
	 *
1005 1006 1007
	 * We use the standard kernel pagetable protections for the new
	 * pagetable protections, the actual ptes set above control the
	 * primary protection behavior:
1008
	 */
1009
	__set_pmd_pte(kpte, address, mk_pte(base, __pgprot(_KERNPG_TABLE)));
1010 1011 1012 1013 1014 1015 1016 1017 1018 1019

	/*
	 * 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();
1020
	spin_unlock(&pgd_lock);
1021

1022 1023
	return 0;
}
1024

1025 1026
static int split_large_page(struct cpa_data *cpa, pte_t *kpte,
			    unsigned long address)
1027 1028 1029
{
	struct page *base;

1030
	if (!debug_pagealloc_enabled())
1031
		spin_unlock(&cpa_lock);
1032
	base = alloc_pages(GFP_KERNEL, 0);
1033
	if (!debug_pagealloc_enabled())
1034 1035 1036 1037
		spin_lock(&cpa_lock);
	if (!base)
		return -ENOMEM;

1038
	if (__split_large_page(cpa, kpte, address, base))
1039
		__free_page(base);
1040 1041 1042 1043

	return 0;
}

1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136
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;
}

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);
}
1137

1138
static void unmap_pud_range(p4d_t *p4d, unsigned long start, unsigned long end)
1139
{
1140
	pud_t *pud = pud_offset(p4d, start);
1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180

	/*
	 * 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
	 */
}

1181 1182
static int alloc_pte_page(pmd_t *pmd)
{
1183
	pte_t *pte = (pte_t *)get_zeroed_page(GFP_KERNEL);
1184 1185 1186 1187 1188 1189 1190
	if (!pte)
		return -1;

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

1191 1192
static int alloc_pmd_page(pud_t *pud)
{
1193
	pmd_t *pmd = (pmd_t *)get_zeroed_page(GFP_KERNEL);
1194 1195 1196 1197 1198 1199 1200
	if (!pmd)
		return -1;

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

1201 1202 1203 1204 1205 1206 1207 1208
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);

1209
	pgprot = pgprot_clear_protnone_bits(pgprot);
1210 1211

	while (num_pages-- && start < end) {
1212
		set_pte(pte, pfn_pte(cpa->pfn, pgprot));
1213 1214

		start	 += PAGE_SIZE;
1215
		cpa->pfn++;
1216 1217 1218
		pte++;
	}
}
1219

1220 1221 1222
static long populate_pmd(struct cpa_data *cpa,
			 unsigned long start, unsigned long end,
			 unsigned num_pages, pud_t *pud, pgprot_t pgprot)
1223
{
1224
	long cur_pages = 0;
1225
	pmd_t *pmd;
1226
	pgprot_t pmd_pgprot;
1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257

	/*
	 * 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;

1258 1259
	pmd_pgprot = pgprot_4k_2_large(pgprot);

1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270
	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);

1271 1272
		set_pmd(pmd, pmd_mkhuge(pfn_pmd(cpa->pfn,
					canon_pgprot(pmd_pgprot))));
1273 1274

		start	  += PMD_SIZE;
1275
		cpa->pfn  += PMD_SIZE >> PAGE_SHIFT;
1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292
		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;
}
1293

1294 1295
static int populate_pud(struct cpa_data *cpa, unsigned long start, p4d_t *p4d,
			pgprot_t pgprot)
1296 1297 1298
{
	pud_t *pud;
	unsigned long end;
1299
	long cur_pages = 0;
1300
	pgprot_t pud_pgprot;
1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315

	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);

1316
		pud = pud_offset(p4d, start);
1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336

		/*
		 * 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;

1337
	pud = pud_offset(p4d, start);
1338
	pud_pgprot = pgprot_4k_2_large(pgprot);
1339 1340 1341 1342

	/*
	 * Map everything starting from the Gb boundary, possibly with 1G pages
	 */
1343
	while (boot_cpu_has(X86_FEATURE_GBPAGES) && end - start >= PUD_SIZE) {
1344 1345
		set_pud(pud, pud_mkhuge(pfn_pud(cpa->pfn,
				   canon_pgprot(pud_pgprot))));
1346 1347

		start	  += PUD_SIZE;
1348
		cpa->pfn  += PUD_SIZE >> PAGE_SHIFT;
1349 1350 1351 1352 1353 1354
		cur_pages += PUD_SIZE >> PAGE_SHIFT;
		pud++;
	}

	/* Map trailing leftover */
	if (start < end) {
1355
		long tmp;
1356

1357
		pud = pud_offset(p4d, start);
1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370
		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;
}
1371 1372 1373 1374 1375 1376 1377 1378 1379

/*
 * 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 */
1380
	p4d_t *p4d;
1381
	pgd_t *pgd_entry;
1382
	long ret;
1383 1384 1385

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

1386
	if (pgd_none(*pgd_entry)) {
1387
		p4d = (p4d_t *)get_zeroed_page(GFP_KERNEL);
1388 1389 1390 1391 1392 1393
		if (!p4d)
			return -1;

		set_pgd(pgd_entry, __pgd(__pa(p4d) | _KERNPG_TABLE));
	}

1394 1395 1396
	/*
	 * Allocate a PUD page and hand it down for mapping.
	 */
1397 1398
	p4d = p4d_offset(pgd_entry, addr);
	if (p4d_none(*p4d)) {
1399
		pud = (pud_t *)get_zeroed_page(GFP_KERNEL);
1400 1401
		if (!pud)
			return -1;
1402

1403
		set_p4d(p4d, __p4d(__pa(pud) | _KERNPG_TABLE));
1404 1405 1406 1407 1408
	}

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

1409
	ret = populate_pud(cpa, addr, p4d, pgprot);
1410
	if (ret < 0) {
1411 1412 1413 1414 1415
		/*
		 * Leave the PUD page in place in case some other CPU or thread
		 * already found it, but remove any useless entries we just
		 * added to it.
		 */
1416
		unmap_pud_range(p4d, addr,
1417
				addr + (cpa->numpages << PAGE_SHIFT));
1418
		return ret;
1419
	}
1420

1421 1422 1423 1424
	cpa->numpages = ret;
	return 0;
}

1425 1426 1427
static int __cpa_process_fault(struct cpa_data *cpa, unsigned long vaddr,
			       int primary)
{
1428 1429 1430 1431 1432 1433
	if (cpa->pgd) {
		/*
		 * Right now, we only execute this code path when mapping
		 * the EFI virtual memory map regions, no other users
		 * provide a ->pgd value. This may change in the future.
		 */
1434
		return populate_pgd(cpa, vaddr);
1435
	}
1436

1437 1438 1439
	/*
	 * Ignore all non primary paths.
	 */
1440 1441
	if (!primary) {
		cpa->numpages = 1;
1442
		return 0;
1443
	}
1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456

	/*
	 * 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;
1457 1458 1459 1460

	} else if (__cpa_pfn_in_highmap(cpa->pfn)) {
		/* Faults in the highmap are OK, so do not warn: */
		return -EFAULT;
1461 1462 1463 1464 1465 1466 1467 1468 1469
	} else {
		WARN(1, KERN_WARNING "CPA: called for zero pte. "
			"vaddr = %lx cpa->vaddr = %lx\n", vaddr,
			*cpa->vaddr);

		return -EFAULT;
	}
}

1470
static int __change_page_attr(struct cpa_data *cpa, int primary)
1471
{
1472
	unsigned long address;
1473 1474
	int do_split, err;
	unsigned int level;
1475
	pte_t *kpte, old_pte;
L
Linus Torvalds 已提交
1476

1477 1478 1479 1480 1481 1482
	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)
1483 1484 1485
		address = cpa->vaddr[cpa->curpage];
	else
		address = *cpa->vaddr;
1486
repeat:
1487
	kpte = _lookup_address_cpa(cpa, address, &level);
L
Linus Torvalds 已提交
1488
	if (!kpte)
1489
		return __cpa_process_fault(cpa, address, primary);
1490 1491

	old_pte = *kpte;
1492
	if (pte_none(old_pte))
1493
		return __cpa_process_fault(cpa, address, primary);
1494

T
Thomas Gleixner 已提交
1495
	if (level == PG_LEVEL_4K) {
1496
		pte_t new_pte;
1497
		pgprot_t new_prot = pte_pgprot(old_pte);
1498
		unsigned long pfn = pte_pfn(old_pte);
I
Ingo Molnar 已提交
1499

1500 1501
		pgprot_val(new_prot) &= ~pgprot_val(cpa->mask_clr);
		pgprot_val(new_prot) |= pgprot_val(cpa->mask_set);
I
Ingo Molnar 已提交
1502

1503
		cpa_inc_4k_install();
1504 1505
		new_prot = static_protections(new_prot, address, pfn, 1,
					      CPA_PROTECT);
I
Ingo Molnar 已提交
1506

1507
		new_prot = pgprot_clear_protnone_bits(new_prot);
1508

1509 1510 1511 1512 1513
		/*
		 * 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
		 */
1514
		new_pte = pfn_pte(pfn, new_prot);
1515
		cpa->pfn = pfn;
1516 1517 1518 1519 1520
		/*
		 * Do we really change anything ?
		 */
		if (pte_val(old_pte) != pte_val(new_pte)) {
			set_pte_atomic(kpte, new_pte);
1521
			cpa->flags |= CPA_FLUSHTLB;
1522
		}
1523
		cpa->numpages = 1;
1524
		return 0;
L
Linus Torvalds 已提交
1525
	}
1526 1527 1528 1529 1530

	/*
	 * Check, whether we can keep the large page intact
	 * and just change the pte:
	 */
1531
	do_split = should_split_large_page(kpte, address, cpa);
1532 1533
	/*
	 * When the range fits into the existing large page,
1534
	 * return. cp->numpages and cpa->tlbflush have been updated in
1535 1536
	 * try_large_page:
	 */
1537 1538
	if (do_split <= 0)
		return do_split;
1539 1540 1541 1542

	/*
	 * We have to split the large page:
	 */
1543
	err = split_large_page(cpa, kpte, address);
1544
	if (!err) {
1545
		/*
1546 1547 1548 1549 1550
		 * 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
1551 1552 1553 1554 1555 1556
		 *  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."
1557 1558
		 *
		 * We do this global tlb flush inside the cpa_lock, so that we
1559 1560 1561
		 * 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.
1562
		 */
1563
		flush_tlb_all();
1564 1565
		goto repeat;
	}
1566

1567
	return err;
1568
}
L
Linus Torvalds 已提交
1569

1570 1571 1572
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
1573
{
1574
	struct cpa_data alias_cpa;
1575
	unsigned long laddr = (unsigned long)__va(cpa->pfn << PAGE_SHIFT);
1576
	unsigned long vaddr;
1577
	int ret;
1578

1579
	if (!pfn_range_is_mapped(cpa->pfn, cpa->pfn + 1))
1580
		return 0;
1581

1582 1583 1584 1585
	/*
	 * No need to redo, when the primary call touched the direct
	 * mapping already:
	 */
1586 1587 1588 1589 1590 1591
	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)
1592 1593 1594 1595 1596
		vaddr = cpa->vaddr[cpa->curpage];
	else
		vaddr = *cpa->vaddr;

	if (!(within(vaddr, PAGE_OFFSET,
1597
		    PAGE_OFFSET + (max_pfn_mapped << PAGE_SHIFT)))) {
1598

1599
		alias_cpa = *cpa;
1600
		alias_cpa.vaddr = &laddr;
1601
		alias_cpa.flags &= ~(CPA_PAGES_ARRAY | CPA_ARRAY);
1602

1603
		ret = __change_page_attr_set_clr(&alias_cpa, 0);
1604 1605
		if (ret)
			return ret;
1606
	}
1607 1608

#ifdef CONFIG_X86_64
1609
	/*
1610 1611
	 * If the primary call didn't touch the high mapping already
	 * and the physical address is inside the kernel map, we need
1612
	 * to touch the high mapped kernel as well:
1613
	 */
1614
	if (!within(vaddr, (unsigned long)_text, _brk_end) &&
1615
	    __cpa_pfn_in_highmap(cpa->pfn)) {
1616 1617 1618 1619 1620
		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);
1621

1622 1623 1624 1625 1626 1627
		/*
		 * The high mapping range is imprecise, so ignore the
		 * return value.
		 */
		__change_page_attr_set_clr(&alias_cpa, 0);
	}
1628
#endif
1629 1630

	return 0;
L
Linus Torvalds 已提交
1631 1632
}

1633
static int __change_page_attr_set_clr(struct cpa_data *cpa, int checkalias)
1634
{
1635 1636
	unsigned long numpages = cpa->numpages;
	int ret;
1637

1638 1639 1640 1641 1642
	while (numpages) {
		/*
		 * Store the remaining nr of pages for the large page
		 * preservation check.
		 */
1643
		cpa->numpages = numpages;
1644
		/* for array changes, we can't use large page */
1645
		if (cpa->flags & (CPA_ARRAY | CPA_PAGES_ARRAY))
1646
			cpa->numpages = 1;
1647

1648
		if (!debug_pagealloc_enabled())
1649
			spin_lock(&cpa_lock);
1650
		ret = __change_page_attr(cpa, checkalias);
1651
		if (!debug_pagealloc_enabled())
1652
			spin_unlock(&cpa_lock);
1653 1654 1655
		if (ret)
			return ret;

1656 1657 1658 1659 1660 1661
		if (checkalias) {
			ret = cpa_process_alias(cpa);
			if (ret)
				return ret;
		}

1662 1663 1664 1665 1666
		/*
		 * 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.
		 */
1667
		BUG_ON(cpa->numpages > numpages || !cpa->numpages);
1668
		numpages -= cpa->numpages;
1669
		if (cpa->flags & (CPA_PAGES_ARRAY | CPA_ARRAY))
1670 1671 1672 1673
			cpa->curpage++;
		else
			*cpa->vaddr += cpa->numpages * PAGE_SIZE;

1674
	}
1675 1676 1677
	return 0;
}

1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700
/*
 * Machine check recovery code needs to change cache mode of poisoned
 * pages to UC to avoid speculative access logging another error. But
 * passing the address of the 1:1 mapping to set_memory_uc() is a fine
 * way to encourage a speculative access. So we cheat and flip the top
 * bit of the address. This works fine for the code that updates the
 * page tables. But at the end of the process we need to flush the cache
 * and the non-canonical address causes a #GP fault when used by the
 * CLFLUSH instruction.
 *
 * But in the common case we already have a canonical address. This code
 * will fix the top bit if needed and is a no-op otherwise.
 */
static inline unsigned long make_addr_canonical_again(unsigned long addr)
{
#ifdef CONFIG_X86_64
	return (long)(addr << 1) >> 1;
#else
	return addr;
#endif
}


1701
static int change_page_attr_set_clr(unsigned long *addr, int numpages,
1702
				    pgprot_t mask_set, pgprot_t mask_clr,
1703 1704
				    int force_split, int in_flag,
				    struct page **pages)
1705
{
1706
	struct cpa_data cpa;
1707
	int ret, cache, checkalias;
1708
	unsigned long baddr = 0;
1709

1710 1711
	memset(&cpa, 0, sizeof(cpa));

1712
	/*
1713 1714
	 * Check, if we are requested to set a not supported
	 * feature.  Clearing non-supported features is OK.
1715 1716
	 */
	mask_set = canon_pgprot(mask_set);
1717

1718
	if (!pgprot_val(mask_set) && !pgprot_val(mask_clr) && !force_split)
1719 1720
		return 0;

1721
	/* Ensure we are PAGE_SIZE aligned */
1722
	if (in_flag & CPA_ARRAY) {
1723 1724 1725 1726 1727 1728 1729
		int i;
		for (i = 0; i < numpages; i++) {
			if (addr[i] & ~PAGE_MASK) {
				addr[i] &= PAGE_MASK;
				WARN_ON_ONCE(1);
			}
		}
1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741
	} 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);
		}
1742 1743 1744 1745
		/*
		 * Save address for cache flush. *addr is modified in the call
		 * to __change_page_attr_set_clr() below.
		 */
1746
		baddr = make_addr_canonical_again(*addr);
1747 1748
	}

1749 1750 1751
	/* Must avoid aliasing mappings in the highmem code */
	kmap_flush_unused();

N
Nick Piggin 已提交
1752 1753
	vm_unmap_aliases();

1754
	cpa.vaddr = addr;
1755
	cpa.pages = pages;
1756 1757 1758
	cpa.numpages = numpages;
	cpa.mask_set = mask_set;
	cpa.mask_clr = mask_clr;
1759 1760
	cpa.flags = 0;
	cpa.curpage = 0;
1761
	cpa.force_split = force_split;
1762

1763 1764
	if (in_flag & (CPA_ARRAY | CPA_PAGES_ARRAY))
		cpa.flags |= in_flag;
1765

1766 1767
	/* No alias checking for _NX bit modifications */
	checkalias = (pgprot_val(mask_set) | pgprot_val(mask_clr)) != _PAGE_NX;
1768 1769 1770
	/* Has caller explicitly disabled alias checking? */
	if (in_flag & CPA_NO_CHECK_ALIAS)
		checkalias = 0;
1771 1772

	ret = __change_page_attr_set_clr(&cpa, checkalias);
1773

1774 1775 1776
	/*
	 * Check whether we really changed something:
	 */
1777
	if (!(cpa.flags & CPA_FLUSHTLB))
1778
		goto out;
1779

1780 1781 1782 1783
	/*
	 * No need to flush, when we did not set any of the caching
	 * attributes:
	 */
1784
	cache = !!pgprot2cachemode(mask_set);
1785

1786
	/*
1787 1788
	 * On success we use CLFLUSH, when the CPU supports it to
	 * avoid the WBINVD. If the CPU does not support it and in the
1789
	 * error case we fall back to cpa_flush_all (which uses
1790
	 * WBINVD):
1791
	 */
1792
	if (!ret && boot_cpu_has(X86_FEATURE_CLFLUSH)) {
1793 1794 1795 1796
		if (cpa.flags & (CPA_PAGES_ARRAY | CPA_ARRAY)) {
			cpa_flush_array(addr, numpages, cache,
					cpa.flags, pages);
		} else
1797
			cpa_flush_range(baddr, numpages, cache);
1798
	} else
1799
		cpa_flush_all(cache);
1800

1801
out:
1802 1803 1804
	return ret;
}

1805 1806
static inline int change_page_attr_set(unsigned long *addr, int numpages,
				       pgprot_t mask, int array)
1807
{
1808
	return change_page_attr_set_clr(addr, numpages, mask, __pgprot(0), 0,
1809
		(array ? CPA_ARRAY : 0), NULL);
1810 1811
}

1812 1813
static inline int change_page_attr_clear(unsigned long *addr, int numpages,
					 pgprot_t mask, int array)
1814
{
1815
	return change_page_attr_set_clr(addr, numpages, __pgprot(0), mask, 0,
1816
		(array ? CPA_ARRAY : 0), NULL);
1817 1818
}

1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832
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);
}

1833
int _set_memory_uc(unsigned long addr, int numpages)
1834
{
1835 1836
	/*
	 * for now UC MINUS. see comments in ioremap_nocache()
1837 1838 1839
	 * 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.
1840
	 */
1841
	return change_page_attr_set(&addr, numpages,
1842 1843
				    cachemode2pgprot(_PAGE_CACHE_MODE_UC_MINUS),
				    0);
1844
}
1845 1846 1847

int set_memory_uc(unsigned long addr, int numpages)
{
1848 1849
	int ret;

1850 1851 1852
	/*
	 * for now UC MINUS. see comments in ioremap_nocache()
	 */
1853
	ret = reserve_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE,
1854
			      _PAGE_CACHE_MODE_UC_MINUS, NULL);
1855 1856 1857 1858 1859 1860 1861 1862
	if (ret)
		goto out_err;

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

	return 0;
1863

1864 1865 1866 1867
out_free:
	free_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE);
out_err:
	return ret;
1868
}
1869 1870
EXPORT_SYMBOL(set_memory_uc);

1871
static int _set_memory_array(unsigned long *addr, int addrinarray,
1872
		enum page_cache_mode new_type)
1873
{
1874
	enum page_cache_mode set_type;
1875 1876 1877
	int i, j;
	int ret;

1878
	for (i = 0; i < addrinarray; i++) {
1879
		ret = reserve_memtype(__pa(addr[i]), __pa(addr[i]) + PAGE_SIZE,
1880
					new_type, NULL);
1881 1882
		if (ret)
			goto out_free;
1883 1884
	}

1885 1886 1887 1888
	/* If WC, set to UC- first and then WC */
	set_type = (new_type == _PAGE_CACHE_MODE_WC) ?
				_PAGE_CACHE_MODE_UC_MINUS : new_type;

1889
	ret = change_page_attr_set(addr, addrinarray,
1890
				   cachemode2pgprot(set_type), 1);
1891

1892
	if (!ret && new_type == _PAGE_CACHE_MODE_WC)
1893
		ret = change_page_attr_set_clr(addr, addrinarray,
1894 1895
					       cachemode2pgprot(
						_PAGE_CACHE_MODE_WC),
1896 1897
					       __pgprot(_PAGE_CACHE_MASK),
					       0, CPA_ARRAY, NULL);
1898 1899 1900 1901 1902 1903 1904 1905 1906 1907
	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;
1908
}
1909 1910 1911

int set_memory_array_uc(unsigned long *addr, int addrinarray)
{
1912
	return _set_memory_array(addr, addrinarray, _PAGE_CACHE_MODE_UC_MINUS);
1913
}
1914 1915
EXPORT_SYMBOL(set_memory_array_uc);

1916 1917
int set_memory_array_wc(unsigned long *addr, int addrinarray)
{
1918
	return _set_memory_array(addr, addrinarray, _PAGE_CACHE_MODE_WC);
1919 1920 1921
}
EXPORT_SYMBOL(set_memory_array_wc);

1922 1923 1924 1925 1926 1927
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);

1928 1929
int _set_memory_wc(unsigned long addr, int numpages)
{
1930
	int ret;
1931 1932
	unsigned long addr_copy = addr;

1933
	ret = change_page_attr_set(&addr, numpages,
1934 1935
				   cachemode2pgprot(_PAGE_CACHE_MODE_UC_MINUS),
				   0);
1936
	if (!ret) {
1937
		ret = change_page_attr_set_clr(&addr_copy, numpages,
1938 1939
					       cachemode2pgprot(
						_PAGE_CACHE_MODE_WC),
1940 1941
					       __pgprot(_PAGE_CACHE_MASK),
					       0, 0, NULL);
1942 1943
	}
	return ret;
1944 1945 1946 1947
}

int set_memory_wc(unsigned long addr, int numpages)
{
1948 1949 1950
	int ret;

	ret = reserve_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE,
1951
		_PAGE_CACHE_MODE_WC, NULL);
1952
	if (ret)
1953
		return ret;
1954

1955 1956
	ret = _set_memory_wc(addr, numpages);
	if (ret)
1957
		free_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE);
1958 1959

	return ret;
1960 1961 1962
}
EXPORT_SYMBOL(set_memory_wc);

1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985
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);

1986
int _set_memory_wb(unsigned long addr, int numpages)
1987
{
1988
	/* WB cache mode is hard wired to all cache attribute bits being 0 */
1989 1990
	return change_page_attr_clear(&addr, numpages,
				      __pgprot(_PAGE_CACHE_MASK), 0);
1991
}
1992 1993 1994

int set_memory_wb(unsigned long addr, int numpages)
{
1995 1996 1997 1998 1999 2000
	int ret;

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

2001
	free_memtype(__pa(addr), __pa(addr) + numpages * PAGE_SIZE);
2002
	return 0;
2003
}
2004 2005
EXPORT_SYMBOL(set_memory_wb);

2006 2007 2008
int set_memory_array_wb(unsigned long *addr, int addrinarray)
{
	int i;
2009 2010
	int ret;

2011
	/* WB cache mode is hard wired to all cache attribute bits being 0 */
2012 2013
	ret = change_page_attr_clear(addr, addrinarray,
				      __pgprot(_PAGE_CACHE_MASK), 1);
2014 2015
	if (ret)
		return ret;
2016

2017 2018
	for (i = 0; i < addrinarray; i++)
		free_memtype(__pa(addr[i]), __pa(addr[i]) + PAGE_SIZE);
2019

2020
	return 0;
2021 2022 2023
}
EXPORT_SYMBOL(set_memory_array_wb);

2024 2025
int set_memory_x(unsigned long addr, int numpages)
{
2026 2027 2028
	if (!(__supported_pte_mask & _PAGE_NX))
		return 0;

2029
	return change_page_attr_clear(&addr, numpages, __pgprot(_PAGE_NX), 0);
2030 2031 2032 2033 2034
}
EXPORT_SYMBOL(set_memory_x);

int set_memory_nx(unsigned long addr, int numpages)
{
2035 2036 2037
	if (!(__supported_pte_mask & _PAGE_NX))
		return 0;

2038
	return change_page_attr_set(&addr, numpages, __pgprot(_PAGE_NX), 0);
2039 2040 2041 2042 2043
}
EXPORT_SYMBOL(set_memory_nx);

int set_memory_ro(unsigned long addr, int numpages)
{
2044
	return change_page_attr_clear(&addr, numpages, __pgprot(_PAGE_RW), 0);
2045 2046 2047 2048
}

int set_memory_rw(unsigned long addr, int numpages)
{
2049
	return change_page_attr_set(&addr, numpages, __pgprot(_PAGE_RW), 0);
2050
}
2051 2052 2053

int set_memory_np(unsigned long addr, int numpages)
{
2054
	return change_page_attr_clear(&addr, numpages, __pgprot(_PAGE_PRESENT), 0);
2055
}
2056

2057 2058 2059 2060 2061 2062 2063 2064 2065
int set_memory_np_noalias(unsigned long addr, int numpages)
{
	int cpa_flags = CPA_NO_CHECK_ALIAS;

	return change_page_attr_set_clr(&addr, numpages, __pgprot(0),
					__pgprot(_PAGE_PRESENT), 0,
					cpa_flags, NULL);
}

2066 2067
int set_memory_4k(unsigned long addr, int numpages)
{
2068
	return change_page_attr_set_clr(&addr, numpages, __pgprot(0),
2069
					__pgprot(0), 1, 0, NULL);
2070 2071
}

2072 2073 2074 2075 2076 2077
int set_memory_nonglobal(unsigned long addr, int numpages)
{
	return change_page_attr_clear(&addr, numpages,
				      __pgprot(_PAGE_GLOBAL), 0);
}

2078 2079 2080 2081 2082 2083
int set_memory_global(unsigned long addr, int numpages)
{
	return change_page_attr_set(&addr, numpages,
				    __pgprot(_PAGE_GLOBAL), 0);
}

2084 2085 2086 2087 2088 2089
static int __set_memory_enc_dec(unsigned long addr, int numpages, bool enc)
{
	struct cpa_data cpa;
	unsigned long start;
	int ret;

2090 2091
	/* Nothing to do if memory encryption is not active */
	if (!mem_encrypt_active())
2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139
		return 0;

	/* Should not be working on unaligned addresses */
	if (WARN_ONCE(addr & ~PAGE_MASK, "misaligned address: %#lx\n", addr))
		addr &= PAGE_MASK;

	start = addr;

	memset(&cpa, 0, sizeof(cpa));
	cpa.vaddr = &addr;
	cpa.numpages = numpages;
	cpa.mask_set = enc ? __pgprot(_PAGE_ENC) : __pgprot(0);
	cpa.mask_clr = enc ? __pgprot(0) : __pgprot(_PAGE_ENC);
	cpa.pgd = init_mm.pgd;

	/* Must avoid aliasing mappings in the highmem code */
	kmap_flush_unused();
	vm_unmap_aliases();

	/*
	 * Before changing the encryption attribute, we need to flush caches.
	 */
	if (static_cpu_has(X86_FEATURE_CLFLUSH))
		cpa_flush_range(start, numpages, 1);
	else
		cpa_flush_all(1);

	ret = __change_page_attr_set_clr(&cpa, 1);

	/*
	 * After changing the encryption attribute, we need to flush TLBs
	 * again in case any speculative TLB caching occurred (but no need
	 * to flush caches again).  We could just use cpa_flush_all(), but
	 * in case TLB flushing gets optimized in the cpa_flush_range()
	 * path use the same logic as above.
	 */
	if (static_cpu_has(X86_FEATURE_CLFLUSH))
		cpa_flush_range(start, numpages, 0);
	else
		cpa_flush_all(0);

	return ret;
}

int set_memory_encrypted(unsigned long addr, int numpages)
{
	return __set_memory_enc_dec(addr, numpages, true);
}
2140
EXPORT_SYMBOL_GPL(set_memory_encrypted);
2141 2142 2143 2144 2145

int set_memory_decrypted(unsigned long addr, int numpages)
{
	return __set_memory_enc_dec(addr, numpages, false);
}
2146
EXPORT_SYMBOL_GPL(set_memory_decrypted);
2147

2148 2149 2150 2151
int set_pages_uc(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);

2152
	return set_memory_uc(addr, numpages);
2153 2154 2155
}
EXPORT_SYMBOL(set_pages_uc);

2156
static int _set_pages_array(struct page **pages, int addrinarray,
2157
		enum page_cache_mode new_type)
2158 2159 2160
{
	unsigned long start;
	unsigned long end;
2161
	enum page_cache_mode set_type;
2162 2163
	int i;
	int free_idx;
2164
	int ret;
2165 2166

	for (i = 0; i < addrinarray; i++) {
2167 2168 2169
		if (PageHighMem(pages[i]))
			continue;
		start = page_to_pfn(pages[i]) << PAGE_SHIFT;
2170
		end = start + PAGE_SIZE;
2171
		if (reserve_memtype(start, end, new_type, NULL))
2172 2173 2174
			goto err_out;
	}

2175 2176 2177 2178
	/* If WC, set to UC- first and then WC */
	set_type = (new_type == _PAGE_CACHE_MODE_WC) ?
				_PAGE_CACHE_MODE_UC_MINUS : new_type;

2179
	ret = cpa_set_pages_array(pages, addrinarray,
2180
				  cachemode2pgprot(set_type));
2181
	if (!ret && new_type == _PAGE_CACHE_MODE_WC)
2182
		ret = change_page_attr_set_clr(NULL, addrinarray,
2183 2184
					       cachemode2pgprot(
						_PAGE_CACHE_MODE_WC),
2185 2186 2187 2188 2189
					       __pgprot(_PAGE_CACHE_MASK),
					       0, CPA_PAGES_ARRAY, pages);
	if (ret)
		goto err_out;
	return 0; /* Success */
2190 2191 2192
err_out:
	free_idx = i;
	for (i = 0; i < free_idx; i++) {
2193 2194 2195
		if (PageHighMem(pages[i]))
			continue;
		start = page_to_pfn(pages[i]) << PAGE_SHIFT;
2196 2197 2198 2199 2200
		end = start + PAGE_SIZE;
		free_memtype(start, end);
	}
	return -EINVAL;
}
2201 2202 2203

int set_pages_array_uc(struct page **pages, int addrinarray)
{
2204
	return _set_pages_array(pages, addrinarray, _PAGE_CACHE_MODE_UC_MINUS);
2205
}
2206 2207
EXPORT_SYMBOL(set_pages_array_uc);

2208 2209
int set_pages_array_wc(struct page **pages, int addrinarray)
{
2210
	return _set_pages_array(pages, addrinarray, _PAGE_CACHE_MODE_WC);
2211 2212 2213
}
EXPORT_SYMBOL(set_pages_array_wc);

2214 2215 2216 2217 2218 2219
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);

2220 2221 2222 2223
int set_pages_wb(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);

2224
	return set_memory_wb(addr, numpages);
2225 2226 2227
}
EXPORT_SYMBOL(set_pages_wb);

2228 2229 2230 2231 2232 2233 2234
int set_pages_array_wb(struct page **pages, int addrinarray)
{
	int retval;
	unsigned long start;
	unsigned long end;
	int i;

2235
	/* WB cache mode is hard wired to all cache attribute bits being 0 */
2236 2237
	retval = cpa_clear_pages_array(pages, addrinarray,
			__pgprot(_PAGE_CACHE_MASK));
2238 2239
	if (retval)
		return retval;
2240 2241

	for (i = 0; i < addrinarray; i++) {
2242 2243 2244
		if (PageHighMem(pages[i]))
			continue;
		start = page_to_pfn(pages[i]) << PAGE_SHIFT;
2245 2246 2247 2248
		end = start + PAGE_SIZE;
		free_memtype(start, end);
	}

2249
	return 0;
2250 2251 2252
}
EXPORT_SYMBOL(set_pages_array_wb);

2253 2254 2255 2256
int set_pages_x(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);

2257
	return set_memory_x(addr, numpages);
2258 2259 2260 2261 2262 2263 2264
}
EXPORT_SYMBOL(set_pages_x);

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

2265
	return set_memory_nx(addr, numpages);
2266 2267 2268 2269 2270 2271 2272
}
EXPORT_SYMBOL(set_pages_nx);

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

2273
	return set_memory_ro(addr, numpages);
2274 2275 2276 2277 2278
}

int set_pages_rw(struct page *page, int numpages)
{
	unsigned long addr = (unsigned long)page_address(page);
2279

2280
	return set_memory_rw(addr, numpages);
2281 2282
}

L
Linus Torvalds 已提交
2283
#ifdef CONFIG_DEBUG_PAGEALLOC
2284 2285 2286

static int __set_pages_p(struct page *page, int numpages)
{
2287 2288
	unsigned long tempaddr = (unsigned long) page_address(page);
	struct cpa_data cpa = { .vaddr = &tempaddr,
2289
				.pgd = NULL,
2290 2291
				.numpages = numpages,
				.mask_set = __pgprot(_PAGE_PRESENT | _PAGE_RW),
2292 2293
				.mask_clr = __pgprot(0),
				.flags = 0};
2294

2295 2296 2297 2298 2299 2300 2301
	/*
	 * 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);
2302 2303 2304 2305
}

static int __set_pages_np(struct page *page, int numpages)
{
2306 2307
	unsigned long tempaddr = (unsigned long) page_address(page);
	struct cpa_data cpa = { .vaddr = &tempaddr,
2308
				.pgd = NULL,
2309 2310
				.numpages = numpages,
				.mask_set = __pgprot(0),
2311 2312
				.mask_clr = __pgprot(_PAGE_PRESENT | _PAGE_RW),
				.flags = 0};
2313

2314 2315 2316 2317 2318 2319 2320
	/*
	 * 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);
2321 2322
}

2323
void __kernel_map_pages(struct page *page, int numpages, int enable)
L
Linus Torvalds 已提交
2324 2325 2326
{
	if (PageHighMem(page))
		return;
2327
	if (!enable) {
2328 2329
		debug_check_no_locks_freed(page_address(page),
					   numpages * PAGE_SIZE);
2330
	}
2331

2332
	/*
2333
	 * The return value is ignored as the calls cannot fail.
2334 2335
	 * 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
2336
	 */
2337 2338 2339 2340
	if (enable)
		__set_pages_p(page, numpages);
	else
		__set_pages_np(page, numpages);
2341 2342

	/*
2343
	 * We should perform an IPI and flush all tlbs,
2344 2345 2346
	 * but that can deadlock->flush only current cpu.
	 * Preemption needs to be disabled around __flush_tlb_all() due to
	 * CR3 reload in __native_flush_tlb().
L
Linus Torvalds 已提交
2347
	 */
2348
	preempt_disable();
L
Linus Torvalds 已提交
2349
	__flush_tlb_all();
2350
	preempt_enable();
2351 2352

	arch_flush_lazy_mmu_mode();
2353 2354
}

2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371
#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 */
2372

2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393
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);

2394 2395 2396
	if (!(page_flags & _PAGE_RW))
		cpa.mask_clr = __pgprot(_PAGE_RW);

2397 2398 2399
	if (!(page_flags & _PAGE_ENC))
		cpa.mask_clr = pgprot_encrypted(cpa.mask_clr);

2400 2401 2402 2403 2404 2405 2406 2407 2408
	cpa.mask_set = __pgprot(_PAGE_PRESENT | page_flags);

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

out:
	return retval;
}

2409 2410 2411 2412 2413 2414 2415
/*
 * 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
新手
引导
客服 返回
顶部