fault.c 23.6 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1 2 3 4 5 6 7 8 9 10 11 12
/*
 *  Copyright (C) 1995  Linus Torvalds
 *  Copyright (C) 2001,2002 Andi Kleen, SuSE Labs.
 */

#include <linux/signal.h>
#include <linux/sched.h>
#include <linux/kernel.h>
#include <linux/errno.h>
#include <linux/string.h>
#include <linux/types.h>
#include <linux/ptrace.h>
13
#include <linux/mmiotrace.h>
Linus Torvalds's avatar
Linus Torvalds committed
14 15 16 17 18 19 20 21
#include <linux/mman.h>
#include <linux/mm.h>
#include <linux/smp.h>
#include <linux/interrupt.h>
#include <linux/init.h>
#include <linux/tty.h>
#include <linux/vt_kern.h>		/* For unblank_screen() */
#include <linux/compiler.h>
Harvey Harrison's avatar
Harvey Harrison committed
22 23
#include <linux/highmem.h>
#include <linux/bootmem.h>		/* for max_low_pfn */
24
#include <linux/vmalloc.h>
Linus Torvalds's avatar
Linus Torvalds committed
25
#include <linux/module.h>
26
#include <linux/kprobes.h>
27
#include <linux/uaccess.h>
28
#include <linux/kdebug.h>
Linus Torvalds's avatar
Linus Torvalds committed
29 30

#include <asm/system.h>
Harvey Harrison's avatar
Harvey Harrison committed
31 32
#include <asm/desc.h>
#include <asm/segment.h>
Linus Torvalds's avatar
Linus Torvalds committed
33 34 35 36 37 38
#include <asm/pgalloc.h>
#include <asm/smp.h>
#include <asm/tlbflush.h>
#include <asm/proto.h>
#include <asm-generic/sections.h>

39 40 41 42 43 44 45 46
/*
 * Page fault error code bits
 *	bit 0 == 0 means no page found, 1 means protection fault
 *	bit 1 == 0 means read, 1 means write
 *	bit 2 == 0 means kernel, 1 means user-mode
 *	bit 3 == 1 means use of reserved bit detected
 *	bit 4 == 1 means fault was an instruction fetch
 */
47
#define PF_PROT		(1<<0)
48
#define PF_WRITE	(1<<1)
49 50
#define PF_USER		(1<<2)
#define PF_RSVD		(1<<3)
51 52
#define PF_INSTR	(1<<4)

53
static inline int kmmio_fault(struct pt_regs *regs, unsigned long addr)
54
{
55
#ifdef CONFIG_MMIOTRACE_HOOKS
56 57 58
	if (unlikely(is_kmmio_active()))
		if (kmmio_handler(regs, addr) == 1)
			return -1;
59
#endif
60
	return 0;
61 62
}

63
static inline int notify_page_fault(struct pt_regs *regs)
64
{
65
#ifdef CONFIG_KPROBES
66 67 68
	int ret = 0;

	/* kprobe_running() needs smp_processor_id() */
69
	if (!user_mode_vm(regs)) {
70 71 72 73 74
		preempt_disable();
		if (kprobe_running() && kprobe_fault_handler(regs, 14))
			ret = 1;
		preempt_enable();
	}
75

76 77 78 79
	return ret;
#else
	return 0;
#endif
80
}
81

82 83 84 85 86 87 88 89 90 91 92 93 94
/*
 * X86_32
 * Sometimes AMD Athlon/Opteron CPUs report invalid exceptions on prefetch.
 * Check that here and ignore it.
 *
 * X86_64
 * Sometimes the CPU reports invalid exceptions on prefetch.
 * Check that here and ignore it.
 *
 * Opcode checker based on code by Richard Brunner
 */
static int is_prefetch(struct pt_regs *regs, unsigned long addr,
		       unsigned long error_code)
95
{
96
	unsigned char *instr;
Linus Torvalds's avatar
Linus Torvalds committed
97
	int scan_more = 1;
98
	int prefetch = 0;
99
	unsigned char *max_instr;
Linus Torvalds's avatar
Linus Torvalds committed
100

Ingo Molnar's avatar
Ingo Molnar committed
101 102 103 104
	/*
	 * If it was a exec (instruction fetch) fault on NX page, then
	 * do not ignore the fault:
	 */
105
	if (error_code & PF_INSTR)
Linus Torvalds's avatar
Linus Torvalds committed
106
		return 0;
107

108
	instr = (unsigned char *)convert_ip_to_linear(current, regs);
109
	max_instr = instr + 15;
Linus Torvalds's avatar
Linus Torvalds committed
110

111
	if (user_mode(regs) && instr >= (unsigned char *)TASK_SIZE)
Linus Torvalds's avatar
Linus Torvalds committed
112 113
		return 0;

114
	while (scan_more && instr < max_instr) {
Linus Torvalds's avatar
Linus Torvalds committed
115 116 117 118
		unsigned char opcode;
		unsigned char instr_hi;
		unsigned char instr_lo;

119
		if (probe_kernel_address(instr, opcode))
120
			break;
Linus Torvalds's avatar
Linus Torvalds committed
121

122 123
		instr_hi = opcode & 0xf0;
		instr_lo = opcode & 0x0f;
Linus Torvalds's avatar
Linus Torvalds committed
124 125
		instr++;

126
		switch (instr_hi) {
Linus Torvalds's avatar
Linus Torvalds committed
127 128
		case 0x20:
		case 0x30:
129 130 131 132 133 134
			/*
			 * 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
			 */
Linus Torvalds's avatar
Linus Torvalds committed
135 136
			scan_more = ((instr_lo & 7) == 0x6);
			break;
137
#ifdef CONFIG_X86_64
Linus Torvalds's avatar
Linus Torvalds committed
138
		case 0x40:
139 140 141 142 143 144 145
			/*
			 * 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.
			 */
146
			scan_more = (!user_mode(regs)) || (regs->cs == __USER_CS);
Linus Torvalds's avatar
Linus Torvalds committed
147
			break;
148
#endif
Linus Torvalds's avatar
Linus Torvalds committed
149 150 151
		case 0x60:
			/* 0x64 thru 0x67 are valid prefixes in all modes. */
			scan_more = (instr_lo & 0xC) == 0x4;
152
			break;
Linus Torvalds's avatar
Linus Torvalds committed
153
		case 0xF0:
154
			/* 0xF0, 0xF2, 0xF3 are valid prefixes in all modes. */
Linus Torvalds's avatar
Linus Torvalds committed
155
			scan_more = !instr_lo || (instr_lo>>1) == 1;
156
			break;
Linus Torvalds's avatar
Linus Torvalds committed
157 158 159
		case 0x00:
			/* Prefetch instruction is 0x0F0D or 0x0F18 */
			scan_more = 0;
160

161
			if (probe_kernel_address(instr, opcode))
Linus Torvalds's avatar
Linus Torvalds committed
162 163 164
				break;
			prefetch = (instr_lo == 0xF) &&
				(opcode == 0x0D || opcode == 0x18);
165
			break;
Linus Torvalds's avatar
Linus Torvalds committed
166 167 168
		default:
			scan_more = 0;
			break;
169
		}
Linus Torvalds's avatar
Linus Torvalds committed
170 171 172 173
	}
	return prefetch;
}

174 175 176 177 178 179 180 181 182 183 184 185
static void force_sig_info_fault(int si_signo, int si_code,
	unsigned long address, struct task_struct *tsk)
{
	siginfo_t info;

	info.si_signo = si_signo;
	info.si_errno = 0;
	info.si_code = si_code;
	info.si_addr = (void __user *)address;
	force_sig_info(si_signo, &info, tsk);
}

186
#ifdef CONFIG_X86_64
187 188
static int bad_address(void *p)
{
Linus Torvalds's avatar
Linus Torvalds committed
189
	unsigned long dummy;
190
	return probe_kernel_address((unsigned long *)p, dummy);
191
}
192
#endif
Linus Torvalds's avatar
Linus Torvalds committed
193

194
static void dump_pagetable(unsigned long address)
Linus Torvalds's avatar
Linus Torvalds committed
195
{
196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231
#ifdef CONFIG_X86_32
	__typeof__(pte_val(__pte(0))) page;

	page = read_cr3();
	page = ((__typeof__(page) *) __va(page))[address >> PGDIR_SHIFT];
#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)
		                                         & (PTRS_PER_PMD - 1)];
		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
	 * it's allocated already.
	 */
	if ((page >> PAGE_SHIFT) < max_low_pfn
	    && (page & _PAGE_PRESENT)
	    && !(page & _PAGE_PSE)) {
		page &= PAGE_MASK;
		page = ((__typeof__(page) *) __va(page))[(address >> PAGE_SHIFT)
		                                         & (PTRS_PER_PTE - 1)];
		printk("*pte = %0*Lx ", sizeof(page)*2, (u64)page);
	}

	printk("\n");
#else /* CONFIG_X86_64 */
Linus Torvalds's avatar
Linus Torvalds committed
232 233 234 235 236
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *pte;

237
	pgd = (pgd_t *)read_cr3();
Linus Torvalds's avatar
Linus Torvalds committed
238

239
	pgd = __va((unsigned long)pgd & PHYSICAL_PAGE_MASK);
Linus Torvalds's avatar
Linus Torvalds committed
240 241
	pgd += pgd_index(address);
	if (bad_address(pgd)) goto bad;
242
	printk("PGD %lx ", pgd_val(*pgd));
243
	if (!pgd_present(*pgd)) goto ret;
Linus Torvalds's avatar
Linus Torvalds committed
244

245
	pud = pud_offset(pgd, address);
Linus Torvalds's avatar
Linus Torvalds committed
246 247
	if (bad_address(pud)) goto bad;
	printk("PUD %lx ", pud_val(*pud));
248 249
	if (!pud_present(*pud) || pud_large(*pud))
		goto ret;
Linus Torvalds's avatar
Linus Torvalds committed
250 251 252 253

	pmd = pmd_offset(pud, address);
	if (bad_address(pmd)) goto bad;
	printk("PMD %lx ", pmd_val(*pmd));
254
	if (!pmd_present(*pmd) || pmd_large(*pmd)) goto ret;
Linus Torvalds's avatar
Linus Torvalds committed
255 256 257

	pte = pte_offset_kernel(pmd, address);
	if (bad_address(pte)) goto bad;
258
	printk("PTE %lx", pte_val(*pte));
Linus Torvalds's avatar
Linus Torvalds committed
259 260 261 262 263
ret:
	printk("\n");
	return;
bad:
	printk("BAD\n");
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
#endif
}

#ifdef CONFIG_X86_32
static inline pmd_t *vmalloc_sync_one(pgd_t *pgd, unsigned long address)
{
	unsigned index = pgd_index(address);
	pgd_t *pgd_k;
	pud_t *pud, *pud_k;
	pmd_t *pmd, *pmd_k;

	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;
Linus Torvalds's avatar
Linus Torvalds committed
302
}
303
#endif
Linus Torvalds's avatar
Linus Torvalds committed
304

305
#ifdef CONFIG_X86_64
306
static const char errata93_warning[] =
Linus Torvalds's avatar
Linus Torvalds committed
307 308 309 310
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";
311
#endif
Linus Torvalds's avatar
Linus Torvalds committed
312 313 314

/* Workaround for K8 erratum #93 & buggy BIOS.
   BIOS SMM functions are required to use a specific workaround
315 316
   to avoid corruption of the 64bit RIP register on C stepping K8.
   A lot of BIOS that didn't get tested properly miss this.
Linus Torvalds's avatar
Linus Torvalds committed
317 318
   The OS sees this as a page fault with the upper 32bits of RIP cleared.
   Try to work around it here.
319 320 321
   Note we only handle faults in kernel here.
   Does nothing for X86_32
 */
322
static int is_errata93(struct pt_regs *regs, unsigned long address)
Linus Torvalds's avatar
Linus Torvalds committed
323
{
324
#ifdef CONFIG_X86_64
Linus Torvalds's avatar
Linus Torvalds committed
325
	static int warned;
326
	if (address != regs->ip)
Linus Torvalds's avatar
Linus Torvalds committed
327
		return 0;
328
	if ((address >> 32) != 0)
Linus Torvalds's avatar
Linus Torvalds committed
329 330
		return 0;
	address |= 0xffffffffUL << 32;
331 332
	if ((address >= (u64)_stext && address <= (u64)_etext) ||
	    (address >= MODULES_VADDR && address <= MODULES_END)) {
Linus Torvalds's avatar
Linus Torvalds committed
333
		if (!warned) {
334
			printk(errata93_warning);
Linus Torvalds's avatar
Linus Torvalds committed
335 336
			warned = 1;
		}
337
		regs->ip = address;
Linus Torvalds's avatar
Linus Torvalds committed
338 339
		return 1;
	}
340
#endif
Linus Torvalds's avatar
Linus Torvalds committed
341
	return 0;
342
}
Linus Torvalds's avatar
Linus Torvalds committed
343

344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359
/*
 * 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
 * segment in LDT is compatibility mode.
 */
static int is_errata100(struct pt_regs *regs, unsigned long address)
{
#ifdef CONFIG_X86_64
	if ((regs->cs == __USER32_CS || (regs->cs & (1<<2))) &&
	    (address >> 32))
		return 1;
#endif
	return 0;
}

360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380
void do_invalid_op(struct pt_regs *, unsigned long);

static int is_f00f_bug(struct pt_regs *regs, unsigned long address)
{
#ifdef CONFIG_X86_F00F_BUG
	unsigned long nr;
	/*
	 * Pentium F0 0F C7 C8 bug workaround.
	 */
	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;
}

381 382 383
static void show_fault_oops(struct pt_regs *regs, unsigned long error_code,
			    unsigned long address)
{
384 385 386
#ifdef CONFIG_X86_32
	if (!oops_may_print())
		return;
387
#endif
388 389 390

#ifdef CONFIG_X86_PAE
	if (error_code & PF_INSTR) {
391
		unsigned int level;
392 393 394 395 396 397 398 399 400
		pte_t *pte = lookup_address(address, &level);

		if (pte && pte_present(*pte) && !pte_exec(*pte))
			printk(KERN_CRIT "kernel tried to execute "
				"NX-protected page - exploit attempt? "
				"(uid: %d)\n", current->uid);
	}
#endif

401
	printk(KERN_ALERT "BUG: unable to handle kernel ");
402
	if (address < PAGE_SIZE)
403
		printk(KERN_CONT "NULL pointer dereference");
404
	else
405
		printk(KERN_CONT "paging request");
406
	printk(KERN_CONT " at %p\n", (void *) address);
407
	printk(KERN_ALERT "IP:");
408 409 410 411
	printk_address(regs->ip, 1);
	dump_pagetable(address);
}

412
#ifdef CONFIG_X86_64
Linus Torvalds's avatar
Linus Torvalds committed
413 414 415
static noinline void pgtable_bad(unsigned long address, struct pt_regs *regs,
				 unsigned long error_code)
{
416
	unsigned long flags = oops_begin();
417
	struct task_struct *tsk;
418

Linus Torvalds's avatar
Linus Torvalds committed
419 420 421
	printk(KERN_ALERT "%s: Corrupted page table at address %lx\n",
	       current->comm, address);
	dump_pagetable(address);
422 423 424 425
	tsk = current;
	tsk->thread.cr2 = address;
	tsk->thread.trap_no = 14;
	tsk->thread.error_code = error_code;
426 427 428
	if (__die("Bad pagetable", regs, error_code))
		regs = NULL;
	oops_end(flags, regs, SIGKILL);
Linus Torvalds's avatar
Linus Torvalds committed
429
}
430
#endif
Linus Torvalds's avatar
Linus Torvalds committed
431

432 433 434 435 436 437 438 439 440 441
static int spurious_fault_check(unsigned long error_code, pte_t *pte)
{
	if ((error_code & PF_WRITE) && !pte_write(*pte))
		return 0;
	if ((error_code & PF_INSTR) && !pte_exec(*pte))
		return 0;

	return 1;
}

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
/*
 * 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.
 * There are no security implications to leaving a stale TLB when
 * increasing the permissions on a page.
 */
static int spurious_fault(unsigned long address,
			  unsigned long error_code)
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *pte;

	/* 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;

471 472 473
	if (pud_large(*pud))
		return spurious_fault_check(error_code, (pte_t *) pud);

474 475 476 477
	pmd = pmd_offset(pud, address);
	if (!pmd_present(*pmd))
		return 0;

478 479 480
	if (pmd_large(*pmd))
		return spurious_fault_check(error_code, (pte_t *) pmd);

481 482 483 484
	pte = pte_offset_kernel(pmd, address);
	if (!pte_present(*pte))
		return 0;

485
	return spurious_fault_check(error_code, pte);
486 487
}

Linus Torvalds's avatar
Linus Torvalds committed
488
/*
489 490 491 492
 * X86_32
 * Handle a fault on the vmalloc or module mapping area
 *
 * X86_64
493
 * Handle a fault on the vmalloc area
494 495
 *
 * This assumes no large pages in there.
Linus Torvalds's avatar
Linus Torvalds committed
496 497 498
 */
static int vmalloc_fault(unsigned long address)
{
499 500 501 502
#ifdef CONFIG_X86_32
	unsigned long pgd_paddr;
	pmd_t *pmd_k;
	pte_t *pte_k;
503 504 505 506 507

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

508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523
	/*
	 * 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;
#else
Linus Torvalds's avatar
Linus Torvalds committed
524 525 526 527 528
	pgd_t *pgd, *pgd_ref;
	pud_t *pud, *pud_ref;
	pmd_t *pmd, *pmd_ref;
	pte_t *pte, *pte_ref;

529 530 531 532
	/* Make sure we are in vmalloc area */
	if (!(address >= VMALLOC_START && address < VMALLOC_END))
		return -1;

Linus Torvalds's avatar
Linus Torvalds committed
533 534 535 536 537 538 539 540 541 542
	/* 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->mm ?: &init_mm, address);
	pgd_ref = pgd_offset_k(address);
	if (pgd_none(*pgd_ref))
		return -1;
	if (pgd_none(*pgd))
		set_pgd(pgd, *pgd_ref);
543
	else
544
		BUG_ON(pgd_page_vaddr(*pgd) != pgd_page_vaddr(*pgd_ref));
Linus Torvalds's avatar
Linus Torvalds committed
545 546 547 548 549 550 551 552

	/* 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;
553
	if (pud_none(*pud) || pud_page_vaddr(*pud) != pud_page_vaddr(*pud_ref))
Linus Torvalds's avatar
Linus Torvalds committed
554 555 556 557 558 559 560 561 562 563 564
		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);
565 566 567 568
	/* 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))
Linus Torvalds's avatar
Linus Torvalds committed
569 570
		BUG();
	return 0;
571
#endif
Linus Torvalds's avatar
Linus Torvalds committed
572 573
}

574
int show_unhandled_signals = 1;
Linus Torvalds's avatar
Linus Torvalds committed
575 576 577 578 579 580

/*
 * This routine handles page faults.  It determines the address,
 * and the problem, and then passes it off to one of the appropriate
 * routines.
 */
581 582 583 584
#ifdef CONFIG_X86_64
asmlinkage
#endif
void __kprobes do_page_fault(struct pt_regs *regs, unsigned long error_code)
Linus Torvalds's avatar
Linus Torvalds committed
585 586 587
{
	struct task_struct *tsk;
	struct mm_struct *mm;
588
	struct vm_area_struct *vma;
Linus Torvalds's avatar
Linus Torvalds committed
589
	unsigned long address;
590 591 592
	int write, si_code;
	int fault;
#ifdef CONFIG_X86_64
593
	unsigned long flags;
594
#endif
Linus Torvalds's avatar
Linus Torvalds committed
595

Peter Zijlstra's avatar
Peter Zijlstra committed
596 597 598 599 600
	/*
	 * We can fault from pretty much anywhere, with unknown IRQ state.
	 */
	trace_hardirqs_fixup();

601 602 603 604
	tsk = current;
	mm = tsk->mm;
	prefetchw(&mm->mmap_sem);

Linus Torvalds's avatar
Linus Torvalds committed
605
	/* get the address */
606
	address = read_cr2();
Linus Torvalds's avatar
Linus Torvalds committed
607

608
	si_code = SEGV_MAPERR;
Linus Torvalds's avatar
Linus Torvalds committed
609

610 611
	if (notify_page_fault(regs))
		return;
612
	if (unlikely(kmmio_fault(regs, address)))
613
		return;
Linus Torvalds's avatar
Linus Torvalds committed
614 615 616 617 618 619 620 621 622 623 624 625

	/*
	 * 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
626
	 * protection error (error_code & 9) == 0.
Linus Torvalds's avatar
Linus Torvalds committed
627
	 */
628 629
#ifdef CONFIG_X86_32
	if (unlikely(address >= TASK_SIZE)) {
630 631 632
#else
	if (unlikely(address >= TASK_SIZE64)) {
#endif
633 634 635
		if (!(error_code & (PF_RSVD|PF_USER|PF_PROT)) &&
		    vmalloc_fault(address) >= 0)
			return;
636 637 638 639 640

		/* Can handle a stale RO->RW TLB */
		if (spurious_fault(address, error_code))
			return;

641 642 643 644 645 646 647
		/*
		 * Don't take the mm semaphore here. If we fixup a prefetch
		 * fault we could otherwise deadlock.
		 */
		goto bad_area_nosemaphore;
	}

648 649

#ifdef CONFIG_X86_32
650 651
	/* It's safe to allow irq's after cr2 has been saved and the vmalloc
	   fault has been handled. */
652
	if (regs->flags & (X86_EFLAGS_IF | X86_VM_MASK))
653 654 655 656 657 658 659 660 661
		local_irq_enable();

	/*
	 * If we're in an interrupt, have no user context or are running in an
	 * atomic region then we must not take the fault.
	 */
	if (in_atomic() || !mm)
		goto bad_area_nosemaphore;
#else /* CONFIG_X86_64 */
662
	if (likely(regs->flags & X86_EFLAGS_IF))
663 664
		local_irq_enable();

665
	if (unlikely(error_code & PF_RSVD))
Linus Torvalds's avatar
Linus Torvalds committed
666 667 668
		pgtable_bad(address, regs, error_code);

	/*
669 670
	 * If we're in an interrupt, have no user context or are running in an
	 * atomic region then we must not take the fault.
Linus Torvalds's avatar
Linus Torvalds committed
671 672 673 674
	 */
	if (unlikely(in_atomic() || !mm))
		goto bad_area_nosemaphore;

675 676 677 678 679 680
	/*
	 * User-mode registers count as a user access even for any
	 * potential system fault or CPU buglet.
	 */
	if (user_mode_vm(regs))
		error_code |= PF_USER;
681 682
again:
#endif
Linus Torvalds's avatar
Linus Torvalds committed
683 684
	/* When running in the kernel we expect faults to occur only to
	 * addresses in user space.  All other faults represent errors in the
Simon Arlott's avatar
Simon Arlott committed
685
	 * kernel and should generate an OOPS.  Unfortunately, in the case of an
686
	 * erroneous fault occurring in a code path which already holds mmap_sem
Linus Torvalds's avatar
Linus Torvalds committed
687 688 689 690 691 692
	 * 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.
	 *
	 * As the vast majority of faults will be valid we will only perform
Simon Arlott's avatar
Simon Arlott committed
693
	 * the source reference check when there is a possibility of a deadlock.
Linus Torvalds's avatar
Linus Torvalds committed
694 695 696 697 698
	 * 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.
	 */
	if (!down_read_trylock(&mm->mmap_sem)) {
699
		if ((error_code & PF_USER) == 0 &&
700
		    !search_exception_tables(regs->ip))
Linus Torvalds's avatar
Linus Torvalds committed
701 702 703 704 705 706 707
			goto bad_area_nosemaphore;
		down_read(&mm->mmap_sem);
	}

	vma = find_vma(mm, address);
	if (!vma)
		goto bad_area;
708
	if (vma->vm_start <= address)
Linus Torvalds's avatar
Linus Torvalds committed
709 710 711
		goto good_area;
	if (!(vma->vm_flags & VM_GROWSDOWN))
		goto bad_area;
712
	if (error_code & PF_USER) {
713 714 715 716 717
		/*
		 * Accessing the stack below %sp is always a bug.
		 * The large cushion allows instructions like enter
		 * and pusha to work.  ("enter $65535,$31" pushes
		 * 32 pointers and then decrements %sp by 65535.)
718
		 */
719
		if (address + 65536 + 32 * sizeof(unsigned long) < regs->sp)
Linus Torvalds's avatar
Linus Torvalds committed
720 721 722 723 724 725 726 727 728
			goto bad_area;
	}
	if (expand_stack(vma, address))
		goto bad_area;
/*
 * Ok, we have a good vm_area for this memory access, so
 * we can handle it..
 */
good_area:
729
	si_code = SEGV_ACCERR;
Linus Torvalds's avatar
Linus Torvalds committed
730
	write = 0;
731
	switch (error_code & (PF_PROT|PF_WRITE)) {
732 733 734 735 736 737 738 739 740 741 742
	default:	/* 3: write, present */
		/* fall through */
	case PF_WRITE:		/* write, not present */
		if (!(vma->vm_flags & VM_WRITE))
			goto bad_area;
		write++;
		break;
	case PF_PROT:		/* read, present */
		goto bad_area;
	case 0:			/* read, not present */
		if (!(vma->vm_flags & (VM_READ | VM_EXEC | VM_WRITE)))
Linus Torvalds's avatar
Linus Torvalds committed
743 744 745
			goto bad_area;
	}

746 747 748
#ifdef CONFIG_X86_32
survive:
#endif
Linus Torvalds's avatar
Linus Torvalds committed
749 750 751 752 753
	/*
	 * If for any reason at all we couldn't handle the fault,
	 * make sure we exit gracefully rather than endlessly redo
	 * the fault.
	 */
Nick Piggin's avatar
Nick Piggin committed
754 755 756 757 758 759 760
	fault = handle_mm_fault(mm, vma, address, write);
	if (unlikely(fault & VM_FAULT_ERROR)) {
		if (fault & VM_FAULT_OOM)
			goto out_of_memory;
		else if (fault & VM_FAULT_SIGBUS)
			goto do_sigbus;
		BUG();
Linus Torvalds's avatar
Linus Torvalds committed
761
	}
Nick Piggin's avatar
Nick Piggin committed
762 763 764 765
	if (fault & VM_FAULT_MAJOR)
		tsk->maj_flt++;
	else
		tsk->min_flt++;
766 767 768 769 770 771 772 773 774 775 776

#ifdef CONFIG_X86_32
	/*
	 * Did it hit the DOS screen memory VA from vm86 mode?
	 */
	if (v8086_mode(regs)) {
		unsigned long bit = (address - 0xA0000) >> PAGE_SHIFT;
		if (bit < 32)
			tsk->thread.screen_bitmap |= 1 << bit;
	}
#endif
Linus Torvalds's avatar
Linus Torvalds committed
777 778 779 780 781 782 783 784 785 786 787 788
	up_read(&mm->mmap_sem);
	return;

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

bad_area_nosemaphore:
	/* User mode accesses just cause a SIGSEGV */
789
	if (error_code & PF_USER) {
790 791 792 793 794
		/*
		 * It's possible to have interrupts off here.
		 */
		local_irq_enable();

795 796 797 798
		/*
		 * Valid to do another page fault here because this one came
		 * from user space.
		 */
Linus Torvalds's avatar
Linus Torvalds committed
799 800 801
		if (is_prefetch(regs, address, error_code))
			return;

802
		if (is_errata100(regs, address))
Linus Torvalds's avatar
Linus Torvalds committed
803 804
			return;

805 806
		if (show_unhandled_signals && unhandled_signal(tsk, SIGSEGV) &&
		    printk_ratelimit()) {
Linus Torvalds's avatar
Linus Torvalds committed
807
			printk(
808
			"%s%s[%d]: segfault at %lx ip %p sp %p error %lx",
809
			task_pid_nr(tsk) > 1 ? KERN_INFO : KERN_EMERG,
810 811
			tsk->comm, task_pid_nr(tsk), address,
			(void *) regs->ip, (void *) regs->sp, error_code);
812 813
			print_vma_addr(" in ", regs->ip);
			printk("\n");
Linus Torvalds's avatar
Linus Torvalds committed
814
		}
815

Linus Torvalds's avatar
Linus Torvalds committed
816 817 818 819
		tsk->thread.cr2 = address;
		/* Kernel addresses are always protection faults */
		tsk->thread.error_code = error_code | (address >= TASK_SIZE);
		tsk->thread.trap_no = 14;
820
		force_sig_info_fault(SIGSEGV, si_code, address, tsk);
Linus Torvalds's avatar
Linus Torvalds committed
821 822 823
		return;
	}

824 825 826
	if (is_f00f_bug(regs, address))
		return;

Linus Torvalds's avatar
Linus Torvalds committed
827 828
no_context:
	/* Are we prepared to handle this kernel fault?  */
829
	if (fixup_exception(regs))
Linus Torvalds's avatar
Linus Torvalds committed
830 831
		return;

832
	/*
833 834 835 836 837 838
	 * X86_32
	 * Valid to do another page fault here, because if this fault
	 * had been triggered by is_prefetch fixup_exception would have
	 * handled it.
	 *
	 * X86_64
Linus Torvalds's avatar
Linus Torvalds committed
839 840
	 * Hall of shame of CPU/BIOS bugs.
	 */
841 842
	if (is_prefetch(regs, address, error_code))
		return;
Linus Torvalds's avatar
Linus Torvalds committed
843 844

	if (is_errata93(regs, address))
845
		return;
Linus Torvalds's avatar
Linus Torvalds committed
846 847 848 849 850

/*
 * Oops. The kernel tried to access some bad page. We'll have to
 * terminate things with extreme prejudice.
 */
851 852
#ifdef CONFIG_X86_32
	bust_spinlocks(1);
853 854 855
#else
	flags = oops_begin();
#endif
856 857

	show_fault_oops(regs, error_code, address);
Linus Torvalds's avatar
Linus Torvalds committed
858

859 860 861
	tsk->thread.cr2 = address;
	tsk->thread.trap_no = 14;
	tsk->thread.error_code = error_code;
862 863

#ifdef CONFIG_X86_32
864 865 866
	die("Oops", regs, error_code);
	bust_spinlocks(0);
	do_exit(SIGKILL);
867
#else
868 869
	if (__die("Oops", regs, error_code))
		regs = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
870 871
	/* Executive summary in case the body of the oops scrolled away */
	printk(KERN_EMERG "CR2: %016lx\n", address);
872
	oops_end(flags, regs, SIGKILL);
873
#endif
Linus Torvalds's avatar
Linus Torvalds committed
874 875 876 877 878 879 880

/*
 * We ran out of memory, or some other thing happened to us that made
 * us unable to handle the page fault gracefully.
 */
out_of_memory:
	up_read(&mm->mmap_sem);
881 882
	if (is_global_init(tsk)) {
		yield();
883
#ifdef CONFIG_X86_32
884 885 886
		down_read(&mm->mmap_sem);
		goto survive;
#else
Linus Torvalds's avatar
Linus Torvalds committed
887
		goto again;
888
#endif
889 890
	}

Linus Torvalds's avatar
Linus Torvalds committed
891
	printk("VM: killing process %s\n", tsk->comm);
892
	if (error_code & PF_USER)
893
		do_group_exit(SIGKILL);
Linus Torvalds's avatar
Linus Torvalds committed
894 895 896 897 898 899
	goto no_context;

do_sigbus:
	up_read(&mm->mmap_sem);

	/* Kernel mode? Handle exceptions or die */
900
	if (!(error_code & PF_USER))
Linus Torvalds's avatar
Linus Torvalds committed
901
		goto no_context;
902 903 904 905 906
#ifdef CONFIG_X86_32
	/* User space => ok to do another page fault */
	if (is_prefetch(regs, address, error_code))
		return;
#endif
Linus Torvalds's avatar
Linus Torvalds committed
907 908 909
	tsk->thread.cr2 = address;
	tsk->thread.error_code = error_code;
	tsk->thread.trap_no = 14;
910
	force_sig_info_fault(SIGBUS, BUS_ADRERR, address, tsk);
Linus Torvalds's avatar
Linus Torvalds committed
911
}
912

913
DEFINE_SPINLOCK(pgd_lock);
914
LIST_HEAD(pgd_list);
915 916 917

void vmalloc_sync_all(void)
{
918
#ifdef CONFIG_X86_32
919
	unsigned long start = VMALLOC_START & PGDIR_MASK;
920 921 922 923 924 925 926
	unsigned long address;

	if (SHARED_KERNEL_PMD)
		return;

	BUILD_BUG_ON(TASK_SIZE & ~PGDIR_MASK);
	for (address = start; address >= TASK_SIZE; address += PGDIR_SIZE) {
927 928 929 930 931 932 933 934
		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;
935
		}
936
		spin_unlock_irqrestore(&pgd_lock, flags);
937 938
	}
#else /* CONFIG_X86_64 */
939
	unsigned long start = VMALLOC_START & PGDIR_MASK;
940 941 942
	unsigned long address;

	for (address = start; address <= VMALLOC_END; address += PGDIR_SIZE) {
943 944 945 946 947 948 949 950 951 952 953 954 955 956
		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));
957
		}
958
		spin_unlock_irqrestore(&pgd_lock, flags);
959
	}
960
#endif
961
}