fault.c 26.2 KB
Newer Older
L
Linus Torvalds 已提交
1 2
/*
 *  Copyright (C) 1995  Linus Torvalds
I
Ingo Molnar 已提交
3
 *  Copyright (C) 2001, 2002 Andi Kleen, SuSE Labs.
4
 *  Copyright (C) 2008-2009, Red Hat Inc., Ingo Molnar
L
Linus Torvalds 已提交
5 6
 */
#include <linux/interrupt.h>
I
Ingo Molnar 已提交
7 8
#include <linux/mmiotrace.h>
#include <linux/bootmem.h>
L
Linus Torvalds 已提交
9
#include <linux/compiler.h>
H
Harvey Harrison 已提交
10
#include <linux/highmem.h>
11
#include <linux/kprobes.h>
12
#include <linux/uaccess.h>
I
Ingo Molnar 已提交
13 14 15 16 17 18 19
#include <linux/vmalloc.h>
#include <linux/vt_kern.h>
#include <linux/signal.h>
#include <linux/kernel.h>
#include <linux/ptrace.h>
#include <linux/string.h>
#include <linux/module.h>
20
#include <linux/kdebug.h>
I
Ingo Molnar 已提交
21
#include <linux/errno.h>
22
#include <linux/magic.h>
I
Ingo Molnar 已提交
23 24 25 26 27 28 29 30 31
#include <linux/sched.h>
#include <linux/types.h>
#include <linux/init.h>
#include <linux/mman.h>
#include <linux/tty.h>
#include <linux/smp.h>
#include <linux/mm.h>

#include <asm-generic/sections.h>
L
Linus Torvalds 已提交
32 33

#include <asm/tlbflush.h>
I
Ingo Molnar 已提交
34 35 36
#include <asm/pgalloc.h>
#include <asm/segment.h>
#include <asm/system.h>
L
Linus Torvalds 已提交
37
#include <asm/proto.h>
38
#include <asm/traps.h>
I
Ingo Molnar 已提交
39
#include <asm/desc.h>
L
Linus Torvalds 已提交
40

41
/*
I
Ingo Molnar 已提交
42 43 44 45 46 47 48
 * Page fault error code bits:
 *
 *   bit 0 ==	 0: no page found	1: protection fault
 *   bit 1 ==	 0: read access		1: write access
 *   bit 2 ==	 0: kernel-mode access	1: user-mode access
 *   bit 3 ==				1: use of reserved bit detected
 *   bit 4 ==				1: fault was an instruction fetch
49
 */
I
Ingo Molnar 已提交
50 51 52 53 54 55 56 57
enum x86_pf_error_code {

	PF_PROT		=		1 << 0,
	PF_WRITE	=		1 << 1,
	PF_USER		=		1 << 2,
	PF_RSVD		=		1 << 3,
	PF_INSTR	=		1 << 4,
};
58

59 60 61
/*
 * (returns 0 if mmiotrace is disabled)
 */
62
static inline int kmmio_fault(struct pt_regs *regs, unsigned long addr)
63
{
64 65 66 67
	if (unlikely(is_kmmio_active()))
		if (kmmio_handler(regs, addr) == 1)
			return -1;
	return 0;
68 69
}

70
static inline int notify_page_fault(struct pt_regs *regs)
71
{
72 73 74
	int ret = 0;

	/* kprobe_running() needs smp_processor_id() */
75
	if (kprobes_built_in() && !user_mode_vm(regs)) {
76 77 78 79 80
		preempt_disable();
		if (kprobe_running() && kprobe_fault_handler(regs, 14))
			ret = 1;
		preempt_enable();
	}
81

82
	return ret;
83
}
84

85
/*
I
Ingo Molnar 已提交
86 87 88 89 90 91
 * Prefetch quirks:
 *
 * 32-bit mode:
 *
 *   Sometimes AMD Athlon/Opteron CPUs report invalid exceptions on prefetch.
 *   Check that here and ignore it.
92
 *
I
Ingo Molnar 已提交
93
 * 64-bit mode:
94
 *
I
Ingo Molnar 已提交
95 96 97 98
 *   Sometimes the CPU reports invalid exceptions on prefetch.
 *   Check that here and ignore it.
 *
 * Opcode checker based on code by Richard Brunner.
99
 */
100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146
static inline int
check_prefetch_opcode(struct pt_regs *regs, unsigned char *instr,
		      unsigned char opcode, int *prefetch)
{
	unsigned char instr_hi = opcode & 0xf0;
	unsigned char instr_lo = opcode & 0x0f;

	switch (instr_hi) {
	case 0x20:
	case 0x30:
		/*
		 * Values 0x26,0x2E,0x36,0x3E are valid x86 prefixes.
		 * In X86_64 long mode, the CPU will signal invalid
		 * opcode if some of these prefixes are present so
		 * X86_64 will never get here anyway
		 */
		return ((instr_lo & 7) == 0x6);
#ifdef CONFIG_X86_64
	case 0x40:
		/*
		 * In AMD64 long mode 0x40..0x4F are valid REX prefixes
		 * Need to figure out under what instruction mode the
		 * instruction was issued. Could check the LDT for lm,
		 * but for now it's good enough to assume that long
		 * mode only uses well known segments or kernel.
		 */
		return (!user_mode(regs)) || (regs->cs == __USER_CS);
#endif
	case 0x60:
		/* 0x64 thru 0x67 are valid prefixes in all modes. */
		return (instr_lo & 0xC) == 0x4;
	case 0xF0:
		/* 0xF0, 0xF2, 0xF3 are valid prefixes in all modes. */
		return !instr_lo || (instr_lo>>1) == 1;
	case 0x00:
		/* Prefetch instruction is 0x0F0D or 0x0F18 */
		if (probe_kernel_address(instr, opcode))
			return 0;

		*prefetch = (instr_lo == 0xF) &&
			(opcode == 0x0D || opcode == 0x18);
		return 0;
	default:
		return 0;
	}
}

I
Ingo Molnar 已提交
147 148
static int
is_prefetch(struct pt_regs *regs, unsigned long error_code, unsigned long addr)
149
{
I
Ingo Molnar 已提交
150
	unsigned char *max_instr;
151
	unsigned char *instr;
152
	int prefetch = 0;
L
Linus Torvalds 已提交
153

I
Ingo Molnar 已提交
154 155 156 157
	/*
	 * If it was a exec (instruction fetch) fault on NX page, then
	 * do not ignore the fault:
	 */
158
	if (error_code & PF_INSTR)
L
Linus Torvalds 已提交
159
		return 0;
160

161
	instr = (void *)convert_ip_to_linear(current, regs);
162
	max_instr = instr + 15;
L
Linus Torvalds 已提交
163

164
	if (user_mode(regs) && instr >= (unsigned char *)TASK_SIZE)
L
Linus Torvalds 已提交
165 166
		return 0;

167
	while (instr < max_instr) {
I
Ingo Molnar 已提交
168
		unsigned char opcode;
L
Linus Torvalds 已提交
169

170
		if (probe_kernel_address(instr, opcode))
171
			break;
L
Linus Torvalds 已提交
172 173 174

		instr++;

175
		if (!check_prefetch_opcode(regs, instr, opcode, &prefetch))
L
Linus Torvalds 已提交
176 177 178 179 180
			break;
	}
	return prefetch;
}

I
Ingo Molnar 已提交
181 182 183
static void
force_sig_info_fault(int si_signo, int si_code, unsigned long address,
		     struct task_struct *tsk)
184 185 186
{
	siginfo_t info;

I
Ingo Molnar 已提交
187 188 189 190 191
	info.si_signo	= si_signo;
	info.si_errno	= 0;
	info.si_code	= si_code;
	info.si_addr	= (void __user *)address;

192 193 194
	force_sig_info(si_signo, &info, tsk);
}

195 196 197 198 199
DEFINE_SPINLOCK(pgd_lock);
LIST_HEAD(pgd_list);

#ifdef CONFIG_X86_32
static inline pmd_t *vmalloc_sync_one(pgd_t *pgd, unsigned long address)
200
{
201 202 203 204
	unsigned index = pgd_index(address);
	pgd_t *pgd_k;
	pud_t *pud, *pud_k;
	pmd_t *pmd, *pmd_k;
I
Ingo Molnar 已提交
205

206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308
	pgd += index;
	pgd_k = init_mm.pgd + index;

	if (!pgd_present(*pgd_k))
		return NULL;

	/*
	 * set_pgd(pgd, *pgd_k); here would be useless on PAE
	 * and redundant with the set_pmd() on non-PAE. As would
	 * set_pud.
	 */
	pud = pud_offset(pgd, address);
	pud_k = pud_offset(pgd_k, address);
	if (!pud_present(*pud_k))
		return NULL;

	pmd = pmd_offset(pud, address);
	pmd_k = pmd_offset(pud_k, address);
	if (!pmd_present(*pmd_k))
		return NULL;

	if (!pmd_present(*pmd)) {
		set_pmd(pmd, *pmd_k);
		arch_flush_lazy_mmu_mode();
	} else {
		BUG_ON(pmd_page(*pmd) != pmd_page(*pmd_k));
	}

	return pmd_k;
}

void vmalloc_sync_all(void)
{
	unsigned long address;

	if (SHARED_KERNEL_PMD)
		return;

	for (address = VMALLOC_START & PMD_MASK;
	     address >= TASK_SIZE && address < FIXADDR_TOP;
	     address += PMD_SIZE) {

		unsigned long flags;
		struct page *page;

		spin_lock_irqsave(&pgd_lock, flags);
		list_for_each_entry(page, &pgd_list, lru) {
			if (!vmalloc_sync_one(page_address(page), address))
				break;
		}
		spin_unlock_irqrestore(&pgd_lock, flags);
	}
}

/*
 * 32-bit:
 *
 *   Handle a fault on the vmalloc or module mapping area
 */
static noinline int vmalloc_fault(unsigned long address)
{
	unsigned long pgd_paddr;
	pmd_t *pmd_k;
	pte_t *pte_k;

	/* Make sure we are in vmalloc area: */
	if (!(address >= VMALLOC_START && address < VMALLOC_END))
		return -1;

	/*
	 * Synchronize this task's top level page-table
	 * with the 'reference' page table.
	 *
	 * Do _not_ use "current" here. We might be inside
	 * an interrupt in the middle of a task switch..
	 */
	pgd_paddr = read_cr3();
	pmd_k = vmalloc_sync_one(__va(pgd_paddr), address);
	if (!pmd_k)
		return -1;

	pte_k = pte_offset_kernel(pmd_k, address);
	if (!pte_present(*pte_k))
		return -1;

	return 0;
}

/*
 * Did it hit the DOS screen memory VA from vm86 mode?
 */
static inline void
check_v8086_mode(struct pt_regs *regs, unsigned long address,
		 struct task_struct *tsk)
{
	unsigned long bit;

	if (!v8086_mode(regs))
		return;

	bit = (address - 0xA0000) >> PAGE_SHIFT;
	if (bit < 32)
		tsk->thread.screen_bitmap |= 1 << bit;
309
}
L
Linus Torvalds 已提交
310

A
Adrian Bunk 已提交
311
static void dump_pagetable(unsigned long address)
L
Linus Torvalds 已提交
312
{
313 314 315 316
	__typeof__(pte_val(__pte(0))) page;

	page = read_cr3();
	page = ((__typeof__(page) *) __va(page))[address >> PGDIR_SHIFT];
I
Ingo Molnar 已提交
317

318 319 320 321 322 323
#ifdef CONFIG_X86_PAE
	printk("*pdpt = %016Lx ", page);
	if ((page >> PAGE_SHIFT) < max_low_pfn
	    && page & _PAGE_PRESENT) {
		page &= PAGE_MASK;
		page = ((__typeof__(page) *) __va(page))[(address >> PMD_SHIFT)
I
Ingo Molnar 已提交
324
							& (PTRS_PER_PMD - 1)];
325 326 327 328 329 330 331 332 333 334 335
		printk(KERN_CONT "*pde = %016Lx ", page);
		page &= ~_PAGE_NX;
	}
#else
	printk("*pde = %08lx ", page);
#endif

	/*
	 * We must not directly access the pte in the highpte
	 * case if the page table is located in highmem.
	 * And let's rather not kmap-atomic the pte, just in case
I
Ingo Molnar 已提交
336
	 * it's allocated already:
337 338 339 340
	 */
	if ((page >> PAGE_SHIFT) < max_low_pfn
	    && (page & _PAGE_PRESENT)
	    && !(page & _PAGE_PSE)) {
I
Ingo Molnar 已提交
341

342 343
		page &= PAGE_MASK;
		page = ((__typeof__(page) *) __va(page))[(address >> PAGE_SHIFT)
I
Ingo Molnar 已提交
344
							& (PTRS_PER_PTE - 1)];
345 346 347 348
		printk("*pte = %0*Lx ", sizeof(page)*2, (u64)page);
	}

	printk("\n");
349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474
}

#else /* CONFIG_X86_64: */

void vmalloc_sync_all(void)
{
	unsigned long address;

	for (address = VMALLOC_START & PGDIR_MASK; address <= VMALLOC_END;
	     address += PGDIR_SIZE) {

		const pgd_t *pgd_ref = pgd_offset_k(address);
		unsigned long flags;
		struct page *page;

		if (pgd_none(*pgd_ref))
			continue;

		spin_lock_irqsave(&pgd_lock, flags);
		list_for_each_entry(page, &pgd_list, lru) {
			pgd_t *pgd;
			pgd = (pgd_t *)page_address(page) + pgd_index(address);
			if (pgd_none(*pgd))
				set_pgd(pgd, *pgd_ref);
			else
				BUG_ON(pgd_page_vaddr(*pgd) != pgd_page_vaddr(*pgd_ref));
		}
		spin_unlock_irqrestore(&pgd_lock, flags);
	}
}

/*
 * 64-bit:
 *
 *   Handle a fault on the vmalloc area
 *
 * This assumes no large pages in there.
 */
static noinline int vmalloc_fault(unsigned long address)
{
	pgd_t *pgd, *pgd_ref;
	pud_t *pud, *pud_ref;
	pmd_t *pmd, *pmd_ref;
	pte_t *pte, *pte_ref;

	/* Make sure we are in vmalloc area: */
	if (!(address >= VMALLOC_START && address < VMALLOC_END))
		return -1;

	/*
	 * Copy kernel mappings over when needed. This can also
	 * happen within a race in page table update. In the later
	 * case just flush:
	 */
	pgd = pgd_offset(current->active_mm, address);
	pgd_ref = pgd_offset_k(address);
	if (pgd_none(*pgd_ref))
		return -1;

	if (pgd_none(*pgd))
		set_pgd(pgd, *pgd_ref);
	else
		BUG_ON(pgd_page_vaddr(*pgd) != pgd_page_vaddr(*pgd_ref));

	/*
	 * Below here mismatches are bugs because these lower tables
	 * are shared:
	 */

	pud = pud_offset(pgd, address);
	pud_ref = pud_offset(pgd_ref, address);
	if (pud_none(*pud_ref))
		return -1;

	if (pud_none(*pud) || pud_page_vaddr(*pud) != pud_page_vaddr(*pud_ref))
		BUG();

	pmd = pmd_offset(pud, address);
	pmd_ref = pmd_offset(pud_ref, address);
	if (pmd_none(*pmd_ref))
		return -1;

	if (pmd_none(*pmd) || pmd_page(*pmd) != pmd_page(*pmd_ref))
		BUG();

	pte_ref = pte_offset_kernel(pmd_ref, address);
	if (!pte_present(*pte_ref))
		return -1;

	pte = pte_offset_kernel(pmd, address);

	/*
	 * Don't use pte_page here, because the mappings can point
	 * outside mem_map, and the NUMA hash lookup cannot handle
	 * that:
	 */
	if (!pte_present(*pte) || pte_pfn(*pte) != pte_pfn(*pte_ref))
		BUG();

	return 0;
}

static const char errata93_warning[] =
KERN_ERR "******* Your BIOS seems to not contain a fix for K8 errata #93\n"
KERN_ERR "******* Working around it, but it may cause SEGVs or burn power.\n"
KERN_ERR "******* Please consider a BIOS update.\n"
KERN_ERR "******* Disabling USB legacy in the BIOS may also help.\n";

/*
 * No vm86 mode in 64-bit mode:
 */
static inline void
check_v8086_mode(struct pt_regs *regs, unsigned long address,
		 struct task_struct *tsk)
{
}

static int bad_address(void *p)
{
	unsigned long dummy;

	return probe_kernel_address((unsigned long *)p, dummy);
}

static void dump_pagetable(unsigned long address)
{
L
Linus Torvalds 已提交
475 476 477 478 479
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *pte;

480
	pgd = (pgd_t *)read_cr3();
L
Linus Torvalds 已提交
481

482
	pgd = __va((unsigned long)pgd & PHYSICAL_PAGE_MASK);
I
Ingo Molnar 已提交
483

L
Linus Torvalds 已提交
484
	pgd += pgd_index(address);
I
Ingo Molnar 已提交
485 486 487
	if (bad_address(pgd))
		goto bad;

488
	printk("PGD %lx ", pgd_val(*pgd));
I
Ingo Molnar 已提交
489 490 491

	if (!pgd_present(*pgd))
		goto out;
L
Linus Torvalds 已提交
492

493
	pud = pud_offset(pgd, address);
I
Ingo Molnar 已提交
494 495 496
	if (bad_address(pud))
		goto bad;

L
Linus Torvalds 已提交
497
	printk("PUD %lx ", pud_val(*pud));
498
	if (!pud_present(*pud) || pud_large(*pud))
I
Ingo Molnar 已提交
499
		goto out;
L
Linus Torvalds 已提交
500 501

	pmd = pmd_offset(pud, address);
I
Ingo Molnar 已提交
502 503 504
	if (bad_address(pmd))
		goto bad;

L
Linus Torvalds 已提交
505
	printk("PMD %lx ", pmd_val(*pmd));
I
Ingo Molnar 已提交
506 507
	if (!pmd_present(*pmd) || pmd_large(*pmd))
		goto out;
L
Linus Torvalds 已提交
508 509

	pte = pte_offset_kernel(pmd, address);
I
Ingo Molnar 已提交
510 511 512
	if (bad_address(pte))
		goto bad;

513
	printk("PTE %lx", pte_val(*pte));
I
Ingo Molnar 已提交
514
out:
L
Linus Torvalds 已提交
515 516 517 518
	printk("\n");
	return;
bad:
	printk("BAD\n");
519 520
}

521
#endif /* CONFIG_X86_64 */
L
Linus Torvalds 已提交
522

I
Ingo Molnar 已提交
523 524 525 526 527 528 529 530 531 532 533 534 535
/*
 * Workaround for K8 erratum #93 & buggy BIOS.
 *
 * BIOS SMM functions are required to use a specific workaround
 * to avoid corruption of the 64bit RIP register on C stepping K8.
 *
 * A lot of BIOS that didn't get tested properly miss this.
 *
 * The OS sees this as a page fault with the upper 32bits of RIP cleared.
 * Try to work around it here.
 *
 * Note we only handle faults in kernel here.
 * Does nothing on 32-bit.
536
 */
537
static int is_errata93(struct pt_regs *regs, unsigned long address)
L
Linus Torvalds 已提交
538
{
539
#ifdef CONFIG_X86_64
I
Ingo Molnar 已提交
540 541
	static int once;

542
	if (address != regs->ip)
L
Linus Torvalds 已提交
543
		return 0;
I
Ingo Molnar 已提交
544

545
	if ((address >> 32) != 0)
L
Linus Torvalds 已提交
546
		return 0;
I
Ingo Molnar 已提交
547

L
Linus Torvalds 已提交
548
	address |= 0xffffffffUL << 32;
549 550
	if ((address >= (u64)_stext && address <= (u64)_etext) ||
	    (address >= MODULES_VADDR && address <= MODULES_END)) {
I
Ingo Molnar 已提交
551
		if (!once) {
552
			printk(errata93_warning);
I
Ingo Molnar 已提交
553
			once = 1;
L
Linus Torvalds 已提交
554
		}
555
		regs->ip = address;
L
Linus Torvalds 已提交
556 557
		return 1;
	}
558
#endif
L
Linus Torvalds 已提交
559
	return 0;
560
}
L
Linus Torvalds 已提交
561

562
/*
I
Ingo Molnar 已提交
563 564 565 566 567
 * Work around K8 erratum #100 K8 in compat mode occasionally jumps
 * to illegal addresses >4GB.
 *
 * We catch this in the page fault handler because these addresses
 * are not reachable. Just detect this case and return.  Any code
568 569 570 571 572
 * segment in LDT is compatibility mode.
 */
static int is_errata100(struct pt_regs *regs, unsigned long address)
{
#ifdef CONFIG_X86_64
I
Ingo Molnar 已提交
573
	if ((regs->cs == __USER32_CS || (regs->cs & (1<<2))) && (address >> 32))
574 575 576 577 578
		return 1;
#endif
	return 0;
}

579 580 581 582
static int is_f00f_bug(struct pt_regs *regs, unsigned long address)
{
#ifdef CONFIG_X86_F00F_BUG
	unsigned long nr;
I
Ingo Molnar 已提交
583

584
	/*
I
Ingo Molnar 已提交
585
	 * Pentium F0 0F C7 C8 bug workaround:
586 587 588 589 590 591 592 593 594 595 596 597 598
	 */
	if (boot_cpu_data.f00f_bug) {
		nr = (address - idt_descr.address) >> 3;

		if (nr == 6) {
			do_invalid_op(regs, 0);
			return 1;
		}
	}
#endif
	return 0;
}

599 600 601
static const char nx_warning[] = KERN_CRIT
"kernel tried to execute NX-protected page - exploit attempt? (uid: %d)\n";

I
Ingo Molnar 已提交
602 603 604
static void
show_fault_oops(struct pt_regs *regs, unsigned long error_code,
		unsigned long address)
605
{
606 607 608 609
	if (!oops_may_print())
		return;

	if (error_code & PF_INSTR) {
610
		unsigned int level;
I
Ingo Molnar 已提交
611

612 613
		pte_t *pte = lookup_address(address, &level);

614 615
		if (pte && pte_present(*pte) && !pte_exec(*pte))
			printk(nx_warning, current_uid());
616 617
	}

618
	printk(KERN_ALERT "BUG: unable to handle kernel ");
619
	if (address < PAGE_SIZE)
620
		printk(KERN_CONT "NULL pointer dereference");
621
	else
622
		printk(KERN_CONT "paging request");
I
Ingo Molnar 已提交
623

624
	printk(KERN_CONT " at %p\n", (void *) address);
625
	printk(KERN_ALERT "IP:");
626
	printk_address(regs->ip, 1);
I
Ingo Molnar 已提交
627

628 629 630
	dump_pagetable(address);
}

I
Ingo Molnar 已提交
631 632 633
static noinline void
pgtable_bad(struct pt_regs *regs, unsigned long error_code,
	    unsigned long address)
L
Linus Torvalds 已提交
634
{
I
Ingo Molnar 已提交
635 636 637 638 639 640 641
	struct task_struct *tsk;
	unsigned long flags;
	int sig;

	flags = oops_begin();
	tsk = current;
	sig = SIGKILL;
642

L
Linus Torvalds 已提交
643
	printk(KERN_ALERT "%s: Corrupted page table at address %lx\n",
644
	       tsk->comm, address);
L
Linus Torvalds 已提交
645
	dump_pagetable(address);
I
Ingo Molnar 已提交
646 647 648 649 650

	tsk->thread.cr2		= address;
	tsk->thread.trap_no	= 14;
	tsk->thread.error_code	= error_code;

651
	if (__die("Bad pagetable", regs, error_code))
652
		sig = 0;
I
Ingo Molnar 已提交
653

654
	oops_end(flags, regs, sig);
L
Linus Torvalds 已提交
655 656
}

I
Ingo Molnar 已提交
657 658 659
static noinline void
no_context(struct pt_regs *regs, unsigned long error_code,
	   unsigned long address)
660 661
{
	struct task_struct *tsk = current;
662
	unsigned long *stackend;
663 664 665
	unsigned long flags;
	int sig;

I
Ingo Molnar 已提交
666
	/* Are we prepared to handle this kernel fault? */
667 668 669 670
	if (fixup_exception(regs))
		return;

	/*
I
Ingo Molnar 已提交
671 672 673 674 675 676 677
	 * 32-bit:
	 *
	 *   Valid to do another page fault here, because if this fault
	 *   had been triggered by is_prefetch fixup_exception would have
	 *   handled it.
	 *
	 * 64-bit:
678
	 *
I
Ingo Molnar 已提交
679
	 *   Hall of shame of CPU/BIOS bugs.
680 681 682 683 684 685 686 687 688
	 */
	if (is_prefetch(regs, error_code, address))
		return;

	if (is_errata93(regs, address))
		return;

	/*
	 * Oops. The kernel tried to access some bad page. We'll have to
I
Ingo Molnar 已提交
689
	 * terminate things with extreme prejudice:
690 691 692 693 694
	 */
	flags = oops_begin();

	show_fault_oops(regs, error_code, address);

I
Ingo Molnar 已提交
695
	stackend = end_of_stack(tsk);
696 697 698
	if (*stackend != STACK_END_MAGIC)
		printk(KERN_ALERT "Thread overran stack, or stack corrupted\n");

699 700 701
	tsk->thread.cr2		= address;
	tsk->thread.trap_no	= 14;
	tsk->thread.error_code	= error_code;
702 703 704 705

	sig = SIGKILL;
	if (__die("Oops", regs, error_code))
		sig = 0;
I
Ingo Molnar 已提交
706

707 708
	/* Executive summary in case the body of the oops scrolled away */
	printk(KERN_EMERG "CR2: %016lx\n", address);
I
Ingo Molnar 已提交
709

710 711 712
	oops_end(flags, regs, sig);
}

I
Ingo Molnar 已提交
713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739
/*
 * Print out info about fatal segfaults, if the show_unhandled_signals
 * sysctl is set:
 */
static inline void
show_signal_msg(struct pt_regs *regs, unsigned long error_code,
		unsigned long address, struct task_struct *tsk)
{
	if (!unhandled_signal(tsk, SIGSEGV))
		return;

	if (!printk_ratelimit())
		return;

	printk(KERN_CONT "%s%s[%d]: segfault at %lx ip %p sp %p error %lx",
		task_pid_nr(tsk) > 1 ? KERN_INFO : KERN_EMERG,
		tsk->comm, task_pid_nr(tsk), address,
		(void *)regs->ip, (void *)regs->sp, error_code);

	print_vma_addr(KERN_CONT " in ", regs->ip);

	printk(KERN_CONT "\n");
}

static void
__bad_area_nosemaphore(struct pt_regs *regs, unsigned long error_code,
		       unsigned long address, int si_code)
740 741 742 743 744 745
{
	struct task_struct *tsk = current;

	/* User mode accesses just cause a SIGSEGV */
	if (error_code & PF_USER) {
		/*
I
Ingo Molnar 已提交
746
		 * It's possible to have interrupts off here:
747 748 749 750 751
		 */
		local_irq_enable();

		/*
		 * Valid to do another page fault here because this one came
I
Ingo Molnar 已提交
752
		 * from user space:
753 754 755 756 757 758 759
		 */
		if (is_prefetch(regs, error_code, address))
			return;

		if (is_errata100(regs, address))
			return;

I
Ingo Molnar 已提交
760 761 762 763 764 765 766
		if (unlikely(show_unhandled_signals))
			show_signal_msg(regs, error_code, address, tsk);

		/* Kernel addresses are always protection faults: */
		tsk->thread.cr2		= address;
		tsk->thread.error_code	= error_code | (address >= TASK_SIZE);
		tsk->thread.trap_no	= 14;
767 768

		force_sig_info_fault(SIGSEGV, si_code, address, tsk);
I
Ingo Molnar 已提交
769

770 771 772 773 774 775 776 777 778
		return;
	}

	if (is_f00f_bug(regs, address))
		return;

	no_context(regs, error_code, address);
}

I
Ingo Molnar 已提交
779 780 781
static noinline void
bad_area_nosemaphore(struct pt_regs *regs, unsigned long error_code,
		     unsigned long address)
782 783 784 785
{
	__bad_area_nosemaphore(regs, error_code, address, SEGV_MAPERR);
}

I
Ingo Molnar 已提交
786 787 788
static void
__bad_area(struct pt_regs *regs, unsigned long error_code,
	   unsigned long address, int si_code)
789 790 791 792 793 794 795 796 797 798 799 800
{
	struct mm_struct *mm = current->mm;

	/*
	 * Something tried to access memory that isn't in our memory map..
	 * Fix it, but check if it's kernel or user first..
	 */
	up_read(&mm->mmap_sem);

	__bad_area_nosemaphore(regs, error_code, address, si_code);
}

I
Ingo Molnar 已提交
801 802
static noinline void
bad_area(struct pt_regs *regs, unsigned long error_code, unsigned long address)
803 804 805 806
{
	__bad_area(regs, error_code, address, SEGV_MAPERR);
}

I
Ingo Molnar 已提交
807 808 809
static noinline void
bad_area_access_error(struct pt_regs *regs, unsigned long error_code,
		      unsigned long address)
810 811 812 813 814
{
	__bad_area(regs, error_code, address, SEGV_ACCERR);
}

/* TODO: fixup for "mm-invoke-oom-killer-from-page-fault.patch" */
I
Ingo Molnar 已提交
815 816 817
static void
out_of_memory(struct pt_regs *regs, unsigned long error_code,
	      unsigned long address)
818 819 820
{
	/*
	 * We ran out of memory, call the OOM killer, and return the userspace
I
Ingo Molnar 已提交
821
	 * (which will retry the fault, or kill us if we got oom-killed):
822 823
	 */
	up_read(&current->mm->mmap_sem);
I
Ingo Molnar 已提交
824

825 826 827
	pagefault_out_of_memory();
}

I
Ingo Molnar 已提交
828 829
static void
do_sigbus(struct pt_regs *regs, unsigned long error_code, unsigned long address)
830 831 832 833 834 835
{
	struct task_struct *tsk = current;
	struct mm_struct *mm = tsk->mm;

	up_read(&mm->mmap_sem);

I
Ingo Molnar 已提交
836
	/* Kernel mode? Handle exceptions or die: */
837 838
	if (!(error_code & PF_USER))
		no_context(regs, error_code, address);
I
Ingo Molnar 已提交
839

840
	/* User-space => ok to do another page fault: */
841 842
	if (is_prefetch(regs, error_code, address))
		return;
I
Ingo Molnar 已提交
843 844 845 846 847

	tsk->thread.cr2		= address;
	tsk->thread.error_code	= error_code;
	tsk->thread.trap_no	= 14;

848 849 850
	force_sig_info_fault(SIGBUS, BUS_ADRERR, address, tsk);
}

I
Ingo Molnar 已提交
851 852 853
static noinline void
mm_fault_error(struct pt_regs *regs, unsigned long error_code,
	       unsigned long address, unsigned int fault)
854
{
I
Ingo Molnar 已提交
855
	if (fault & VM_FAULT_OOM) {
856
		out_of_memory(regs, error_code, address);
I
Ingo Molnar 已提交
857 858 859 860 861 862
	} else {
		if (fault & VM_FAULT_SIGBUS)
			do_sigbus(regs, error_code, address);
		else
			BUG();
	}
863 864
}

865 866 867 868
static int spurious_fault_check(unsigned long error_code, pte_t *pte)
{
	if ((error_code & PF_WRITE) && !pte_write(*pte))
		return 0;
I
Ingo Molnar 已提交
869

870 871 872 873 874 875
	if ((error_code & PF_INSTR) && !pte_exec(*pte))
		return 0;

	return 1;
}

876
/*
I
Ingo Molnar 已提交
877 878 879 880 881 882 883 884
 * Handle a spurious fault caused by a stale TLB entry.
 *
 * This allows us to lazily refresh the TLB when increasing the
 * permissions of a kernel page (RO -> RW or NX -> X).  Doing it
 * eagerly is very expensive since that implies doing a full
 * cross-processor TLB flush, even if no stale TLB entries exist
 * on other processors.
 *
885 886 887
 * There are no security implications to leaving a stale TLB when
 * increasing the permissions on a page.
 */
I
Ingo Molnar 已提交
888 889
static noinline int
spurious_fault(unsigned long error_code, unsigned long address)
890 891 892 893 894
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *pte;
895
	int ret;
896 897 898 899 900 901 902 903 904 905 906 907 908

	/* Reserved-bit violation or user access to kernel space? */
	if (error_code & (PF_USER | PF_RSVD))
		return 0;

	pgd = init_mm.pgd + pgd_index(address);
	if (!pgd_present(*pgd))
		return 0;

	pud = pud_offset(pgd, address);
	if (!pud_present(*pud))
		return 0;

909 910 911
	if (pud_large(*pud))
		return spurious_fault_check(error_code, (pte_t *) pud);

912 913 914 915
	pmd = pmd_offset(pud, address);
	if (!pmd_present(*pmd))
		return 0;

916 917 918
	if (pmd_large(*pmd))
		return spurious_fault_check(error_code, (pte_t *) pmd);

919 920 921 922
	pte = pte_offset_kernel(pmd, address);
	if (!pte_present(*pte))
		return 0;

923 924 925 926 927
	ret = spurious_fault_check(error_code, pte);
	if (!ret)
		return 0;

	/*
I
Ingo Molnar 已提交
928 929
	 * Make sure we have permissions in PMD.
	 * If not, then there's a bug in the page tables:
930 931 932
	 */
	ret = spurious_fault_check(error_code, (pte_t *) pmd);
	WARN_ONCE(!ret, "PMD has incorrect permission bits\n");
I
Ingo Molnar 已提交
933

934
	return ret;
935 936
}

937
int show_unhandled_signals = 1;
L
Linus Torvalds 已提交
938

I
Ingo Molnar 已提交
939 940
static inline int
access_error(unsigned long error_code, int write, struct vm_area_struct *vma)
941 942
{
	if (write) {
I
Ingo Molnar 已提交
943
		/* write, present and write, not present: */
944 945
		if (unlikely(!(vma->vm_flags & VM_WRITE)))
			return 1;
I
Ingo Molnar 已提交
946
		return 0;
947 948
	}

I
Ingo Molnar 已提交
949 950 951 952 953 954 955 956
	/* read, present: */
	if (unlikely(error_code & PF_PROT))
		return 1;

	/* read, not present: */
	if (unlikely(!(vma->vm_flags & (VM_READ | VM_EXEC | VM_WRITE))))
		return 1;

957 958 959
	return 0;
}

960 961
static int fault_in_kernel_space(unsigned long address)
{
962
	return address >= TASK_SIZE_MAX;
963 964
}

L
Linus Torvalds 已提交
965 966 967 968 969
/*
 * This routine handles page faults.  It determines the address,
 * and the problem, and then passes it off to one of the appropriate
 * routines.
 */
970 971
dotraplinkage void __kprobes
do_page_fault(struct pt_regs *regs, unsigned long error_code)
L
Linus Torvalds 已提交
972
{
I
Ingo Molnar 已提交
973
	struct vm_area_struct *vma;
L
Linus Torvalds 已提交
974
	struct task_struct *tsk;
I
Ingo Molnar 已提交
975
	unsigned long address;
L
Linus Torvalds 已提交
976
	struct mm_struct *mm;
977
	int write;
978
	int fault;
L
Linus Torvalds 已提交
979

980 981
	tsk = current;
	mm = tsk->mm;
I
Ingo Molnar 已提交
982

983 984
	prefetchw(&mm->mmap_sem);

I
Ingo Molnar 已提交
985
	/* Get the faulting address: */
986
	address = read_cr2();
L
Linus Torvalds 已提交
987

988
	if (unlikely(kmmio_fault(regs, address)))
989
		return;
L
Linus Torvalds 已提交
990 991 992 993 994 995 996 997 998 999 1000 1001

	/*
	 * We fault-in kernel-space virtual memory on-demand. The
	 * 'reference' page table is init_mm.pgd.
	 *
	 * NOTE! We MUST NOT take any locks for this case. We may
	 * be in an interrupt or a critical region, and should
	 * only copy the information from the master page table,
	 * nothing more.
	 *
	 * This verifies that the fault happens in kernel space
	 * (error_code & 4) == 0, and that the fault was not a
1002
	 * protection error (error_code & 9) == 0.
L
Linus Torvalds 已提交
1003
	 */
1004
	if (unlikely(fault_in_kernel_space(address))) {
1005 1006 1007
		if (!(error_code & (PF_RSVD|PF_USER|PF_PROT)) &&
		    vmalloc_fault(address) >= 0)
			return;
1008

I
Ingo Molnar 已提交
1009
		/* Can handle a stale RO->RW TLB: */
1010
		if (spurious_fault(error_code, address))
1011 1012
			return;

I
Ingo Molnar 已提交
1013
		/* kprobes don't want to hook the spurious faults: */
1014 1015
		if (notify_page_fault(regs))
			return;
1016 1017
		/*
		 * Don't take the mm semaphore here. If we fixup a prefetch
I
Ingo Molnar 已提交
1018
		 * fault we could otherwise deadlock:
1019
		 */
1020
		bad_area_nosemaphore(regs, error_code, address);
I
Ingo Molnar 已提交
1021

1022
		return;
1023 1024
	}

I
Ingo Molnar 已提交
1025
	/* kprobes don't want to hook the spurious faults: */
I
Ingo Molnar 已提交
1026
	if (unlikely(notify_page_fault(regs)))
1027
		return;
1028
	/*
1029 1030 1031 1032
	 * It's safe to allow irq's after cr2 has been saved and the
	 * vmalloc fault has been handled.
	 *
	 * User-mode registers count as a user access even for any
I
Ingo Molnar 已提交
1033
	 * potential system fault or CPU buglet:
1034
	 */
1035 1036 1037
	if (user_mode_vm(regs)) {
		local_irq_enable();
		error_code |= PF_USER;
I
Ingo Molnar 已提交
1038 1039 1040 1041
	} else {
		if (regs->flags & X86_EFLAGS_IF)
			local_irq_enable();
	}
1042

1043
	if (unlikely(error_code & PF_RSVD))
1044
		pgtable_bad(regs, error_code, address);
L
Linus Torvalds 已提交
1045 1046

	/*
I
Ingo Molnar 已提交
1047 1048
	 * If we're in an interrupt, have no user context or are running
	 * in an atomic region then we must not take the fault:
L
Linus Torvalds 已提交
1049
	 */
1050 1051 1052 1053
	if (unlikely(in_atomic() || !mm)) {
		bad_area_nosemaphore(regs, error_code, address);
		return;
	}
L
Linus Torvalds 已提交
1054

I
Ingo Molnar 已提交
1055 1056
	/*
	 * When running in the kernel we expect faults to occur only to
I
Ingo Molnar 已提交
1057 1058 1059 1060 1061 1062 1063
	 * addresses in user space.  All other faults represent errors in
	 * the kernel and should generate an OOPS.  Unfortunately, in the
	 * case of an erroneous fault occurring in a code path which already
	 * holds mmap_sem we will deadlock attempting to validate the fault
	 * against the address space.  Luckily the kernel only validly
	 * references user space from well defined areas of code, which are
	 * listed in the exceptions table.
L
Linus Torvalds 已提交
1064 1065
	 *
	 * As the vast majority of faults will be valid we will only perform
I
Ingo Molnar 已提交
1066 1067 1068 1069
	 * the source reference check when there is a possibility of a
	 * deadlock. Attempt to lock the address space, if we cannot we then
	 * validate the source. If this is invalid we can skip the address
	 * space check, thus avoiding the deadlock:
L
Linus Torvalds 已提交
1070
	 */
1071
	if (unlikely(!down_read_trylock(&mm->mmap_sem))) {
1072
		if ((error_code & PF_USER) == 0 &&
1073 1074 1075 1076
		    !search_exception_tables(regs->ip)) {
			bad_area_nosemaphore(regs, error_code, address);
			return;
		}
L
Linus Torvalds 已提交
1077
		down_read(&mm->mmap_sem);
1078 1079
	} else {
		/*
I
Ingo Molnar 已提交
1080 1081 1082
		 * The above down_read_trylock() might have succeeded in
		 * which case we'll have missed the might_sleep() from
		 * down_read():
1083 1084
		 */
		might_sleep();
L
Linus Torvalds 已提交
1085 1086 1087
	}

	vma = find_vma(mm, address);
1088 1089 1090 1091 1092
	if (unlikely(!vma)) {
		bad_area(regs, error_code, address);
		return;
	}
	if (likely(vma->vm_start <= address))
L
Linus Torvalds 已提交
1093
		goto good_area;
1094 1095 1096 1097
	if (unlikely(!(vma->vm_flags & VM_GROWSDOWN))) {
		bad_area(regs, error_code, address);
		return;
	}
1098
	if (error_code & PF_USER) {
1099 1100 1101
		/*
		 * Accessing the stack below %sp is always a bug.
		 * The large cushion allows instructions like enter
I
Ingo Molnar 已提交
1102
		 * and pusha to work. ("enter $65535, $31" pushes
1103
		 * 32 pointers and then decrements %sp by 65535.)
1104
		 */
1105 1106 1107 1108
		if (unlikely(address + 65536 + 32 * sizeof(unsigned long) < regs->sp)) {
			bad_area(regs, error_code, address);
			return;
		}
L
Linus Torvalds 已提交
1109
	}
1110 1111 1112 1113 1114 1115 1116 1117 1118
	if (unlikely(expand_stack(vma, address))) {
		bad_area(regs, error_code, address);
		return;
	}

	/*
	 * Ok, we have a good vm_area for this memory access, so
	 * we can handle it..
	 */
L
Linus Torvalds 已提交
1119
good_area:
1120
	write = error_code & PF_WRITE;
I
Ingo Molnar 已提交
1121

1122 1123 1124
	if (unlikely(access_error(error_code, write, vma))) {
		bad_area_access_error(regs, error_code, address);
		return;
L
Linus Torvalds 已提交
1125 1126 1127 1128 1129
	}

	/*
	 * If for any reason at all we couldn't handle the fault,
	 * make sure we exit gracefully rather than endlessly redo
I
Ingo Molnar 已提交
1130
	 * the fault:
L
Linus Torvalds 已提交
1131
	 */
N
Nick Piggin 已提交
1132
	fault = handle_mm_fault(mm, vma, address, write);
I
Ingo Molnar 已提交
1133

N
Nick Piggin 已提交
1134
	if (unlikely(fault & VM_FAULT_ERROR)) {
1135 1136
		mm_fault_error(regs, error_code, address, fault);
		return;
L
Linus Torvalds 已提交
1137
	}
I
Ingo Molnar 已提交
1138

N
Nick Piggin 已提交
1139 1140 1141 1142
	if (fault & VM_FAULT_MAJOR)
		tsk->maj_flt++;
	else
		tsk->min_flt++;
1143

1144 1145
	check_v8086_mode(regs, address, tsk);

L
Linus Torvalds 已提交
1146 1147
	up_read(&mm->mmap_sem);
}