traps.c 26.0 KB
Newer Older
L
Linus Torvalds 已提交
1 2
/*
 *  Copyright (C) 1991, 1992  Linus Torvalds
3
 *  Copyright (C) 2000, 2001, 2002 Andi Kleen, SuSE Labs
L
Linus Torvalds 已提交
4 5 6 7 8 9
 *
 *  Pentium III FXSR, SSE support
 *	Gareth Hughes <gareth@valinux.com>, May 2000
 */

/*
10
 * Handle hardware traps and faults.
L
Linus Torvalds 已提交
11
 */
12 13 14

#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

15
#include <linux/context_tracking.h>
I
Ingo Molnar 已提交
16 17 18 19 20 21
#include <linux/interrupt.h>
#include <linux/kallsyms.h>
#include <linux/spinlock.h>
#include <linux/kprobes.h>
#include <linux/uaccess.h>
#include <linux/kdebug.h>
22
#include <linux/kgdb.h>
L
Linus Torvalds 已提交
23
#include <linux/kernel.h>
I
Ingo Molnar 已提交
24 25
#include <linux/module.h>
#include <linux/ptrace.h>
26
#include <linux/uprobes.h>
L
Linus Torvalds 已提交
27
#include <linux/string.h>
I
Ingo Molnar 已提交
28
#include <linux/delay.h>
L
Linus Torvalds 已提交
29
#include <linux/errno.h>
I
Ingo Molnar 已提交
30 31
#include <linux/kexec.h>
#include <linux/sched.h>
L
Linus Torvalds 已提交
32 33
#include <linux/timer.h>
#include <linux/init.h>
J
Jeremy Fitzhardinge 已提交
34
#include <linux/bug.h>
I
Ingo Molnar 已提交
35 36
#include <linux/nmi.h>
#include <linux/mm.h>
37 38
#include <linux/smp.h>
#include <linux/io.h>
L
Linus Torvalds 已提交
39 40 41 42 43 44

#ifdef CONFIG_EISA
#include <linux/ioport.h>
#include <linux/eisa.h>
#endif

D
Dave Jiang 已提交
45 46 47 48
#if defined(CONFIG_EDAC)
#include <linux/edac.h>
#endif

V
Vegard Nossum 已提交
49
#include <asm/kmemcheck.h>
I
Ingo Molnar 已提交
50
#include <asm/stacktrace.h>
L
Linus Torvalds 已提交
51 52
#include <asm/processor.h>
#include <asm/debugreg.h>
A
Arun Sharma 已提交
53
#include <linux/atomic.h>
54
#include <asm/ftrace.h>
55
#include <asm/traps.h>
L
Linus Torvalds 已提交
56
#include <asm/desc.h>
57
#include <asm/fpu/internal.h>
H
Hidetoshi Seto 已提交
58
#include <asm/mce.h>
59
#include <asm/fixmap.h>
60
#include <asm/mach_traps.h>
61
#include <asm/alternative.h>
62
#include <asm/fpu/xstate.h>
D
Dave Hansen 已提交
63
#include <asm/trace/mpx.h>
64
#include <asm/mpx.h>
B
Brian Gerst 已提交
65
#include <asm/vm86.h>
66

67
#ifdef CONFIG_X86_64
68
#include <asm/x86_init.h>
69 70
#include <asm/pgalloc.h>
#include <asm/proto.h>
K
Kees Cook 已提交
71 72 73

/* No need to be aligned, but done to keep all IDTs defined the same way. */
gate_desc debug_idt_table[NR_VECTORS] __page_aligned_bss;
74
#else
75
#include <asm/processor-flags.h>
I
Ingo Molnar 已提交
76
#include <asm/setup.h>
77
#include <asm/proto.h>
78
#endif
L
Linus Torvalds 已提交
79

K
Kees Cook 已提交
80 81 82
/* Must be page-aligned because the real IDT is used in a fixmap. */
gate_desc idt_table[NR_VECTORS] __page_aligned_bss;

83 84 85
DECLARE_BITMAP(used_vectors, NR_VECTORS);
EXPORT_SYMBOL_GPL(used_vectors);

86
static inline void cond_local_irq_enable(struct pt_regs *regs)
87 88 89 90 91
{
	if (regs->flags & X86_EFLAGS_IF)
		local_irq_enable();
}

92
static inline void cond_local_irq_disable(struct pt_regs *regs)
93 94 95 96 97
{
	if (regs->flags & X86_EFLAGS_IF)
		local_irq_disable();
}

98
void ist_enter(struct pt_regs *regs)
99
{
100
	if (user_mode(regs)) {
101
		RCU_LOCKDEP_WARN(!rcu_is_watching(), "entry code didn't wake RCU");
102 103 104 105 106 107 108 109 110
	} else {
		/*
		 * We might have interrupted pretty much anything.  In
		 * fact, if we're a machine check, we can even interrupt
		 * NMI processing.  We don't want in_nmi() to return true,
		 * but we need to notify RCU.
		 */
		rcu_nmi_enter();
	}
111 112

	/*
113 114 115 116
	 * We are atomic because we're on the IST stack; or we're on
	 * x86_32, in which case we still shouldn't schedule; or we're
	 * on x86_64 and entered from user mode, in which case we're
	 * still atomic unless ist_begin_non_atomic is called.
117 118 119 120
	 */
	preempt_count_add(HARDIRQ_OFFSET);

	/* This code is a bit fragile.  Test it. */
121
	RCU_LOCKDEP_WARN(!rcu_is_watching(), "ist_enter didn't work");
122 123
}

124
void ist_exit(struct pt_regs *regs)
125 126 127
{
	preempt_count_sub(HARDIRQ_OFFSET);

128
	if (!user_mode(regs))
129 130 131
		rcu_nmi_exit();
}

132 133 134 135 136 137
/**
 * ist_begin_non_atomic() - begin a non-atomic section in an IST exception
 * @regs:	regs passed to the IST exception handler
 *
 * IST exception handlers normally cannot schedule.  As a special
 * exception, if the exception interrupted userspace code (i.e.
138
 * user_mode(regs) would return true) and the exception was not
139 140 141
 * a double fault, it can be safe to schedule.  ist_begin_non_atomic()
 * begins a non-atomic section within an ist_enter()/ist_exit() region.
 * Callers are responsible for enabling interrupts themselves inside
142
 * the non-atomic section, and callers must call ist_end_non_atomic()
143 144 145 146
 * before ist_exit().
 */
void ist_begin_non_atomic(struct pt_regs *regs)
{
147
	BUG_ON(!user_mode(regs));
148 149 150 151 152 153

	/*
	 * Sanity check: we need to be on the normal thread stack.  This
	 * will catch asm bugs and any attempt to use ist_preempt_enable
	 * from double_fault.
	 */
154 155
	BUG_ON((unsigned long)(current_top_of_stack() -
			       current_stack_pointer()) >= THREAD_SIZE);
156 157 158 159 160 161 162 163 164 165 166 167 168 169

	preempt_count_sub(HARDIRQ_OFFSET);
}

/**
 * ist_end_non_atomic() - begin a non-atomic section in an IST exception
 *
 * Ends a non-atomic section started with ist_begin_non_atomic().
 */
void ist_end_non_atomic(void)
{
	preempt_count_add(HARDIRQ_OFFSET);
}

170
static nokprobe_inline int
171 172
do_trap_no_signal(struct task_struct *tsk, int trapnr, char *str,
		  struct pt_regs *regs,	long error_code)
L
Linus Torvalds 已提交
173
{
174
	if (v8086_mode(regs)) {
175
		/*
176
		 * Traps 0, 1, 3, 4, and 5 should be forwarded to vm86.
177 178
		 * On nmi (interrupt 2), do_trap should not be called.
		 */
179 180 181 182 183 184
		if (trapnr < X86_TRAP_UD) {
			if (!handle_vm86_trap((struct kernel_vm86_regs *) regs,
						error_code, trapnr))
				return 0;
		}
		return -1;
L
Linus Torvalds 已提交
185
	}
186

187
	if (!user_mode(regs)) {
188 189 190 191 192 193 194
		if (!fixup_exception(regs)) {
			tsk->thread.error_code = error_code;
			tsk->thread.trap_nr = trapnr;
			die(str, regs, error_code);
		}
		return 0;
	}
L
Linus Torvalds 已提交
195

196 197
	return -1;
}
L
Linus Torvalds 已提交
198

199 200
static siginfo_t *fill_trap_info(struct pt_regs *regs, int signr, int trapnr,
				siginfo_t *info)
201 202 203 204 205
{
	unsigned long siaddr;
	int sicode;

	switch (trapnr) {
206 207 208
	default:
		return SEND_SIG_PRIV;

209 210
	case X86_TRAP_DE:
		sicode = FPE_INTDIV;
211
		siaddr = uprobe_get_trap_addr(regs);
212 213 214
		break;
	case X86_TRAP_UD:
		sicode = ILL_ILLOPN;
215
		siaddr = uprobe_get_trap_addr(regs);
216 217 218 219 220 221 222 223 224 225 226
		break;
	case X86_TRAP_AC:
		sicode = BUS_ADRALN;
		siaddr = 0;
		break;
	}

	info->si_signo = signr;
	info->si_errno = 0;
	info->si_code = sicode;
	info->si_addr = (void __user *)siaddr;
227
	return info;
228 229
}

230
static void
231 232 233 234 235 236 237 238
do_trap(int trapnr, int signr, char *str, struct pt_regs *regs,
	long error_code, siginfo_t *info)
{
	struct task_struct *tsk = current;


	if (!do_trap_no_signal(tsk, trapnr, str, regs, error_code))
		return;
I
Ingo Molnar 已提交
239
	/*
240
	 * We want error_code and trap_nr set for userspace faults and
I
Ingo Molnar 已提交
241 242 243 244 245 246 247 248
	 * kernelspace faults which result in die(), but not
	 * kernelspace faults which are fixed up.  die() gives the
	 * process no chance to handle the signal and notice the
	 * kernel fault information, so that won't result in polluting
	 * the information about previously queued, but not yet
	 * delivered, faults.  See also do_general_protection below.
	 */
	tsk->thread.error_code = error_code;
249
	tsk->thread.trap_nr = trapnr;
250

251 252 253
#ifdef CONFIG_X86_64
	if (show_unhandled_signals && unhandled_signal(tsk, signr) &&
	    printk_ratelimit()) {
254 255 256
		pr_info("%s[%d] trap %s ip:%lx sp:%lx error:%lx",
			tsk->comm, tsk->pid, str,
			regs->ip, regs->sp, error_code);
257
		print_vma_addr(" in ", regs->ip);
258
		pr_cont("\n");
259 260 261
	}
#endif

262
	force_sig_info(signr, info ?: SEND_SIG_PRIV, tsk);
L
Linus Torvalds 已提交
263
}
264
NOKPROBE_SYMBOL(do_trap);
L
Linus Torvalds 已提交
265

266
static void do_error_trap(struct pt_regs *regs, long error_code, char *str,
267
			  unsigned long trapnr, int signr)
268
{
269
	siginfo_t info;
270

271
	RCU_LOCKDEP_WARN(!rcu_is_watching(), "entry code didn't wake RCU");
272

273 274
	if (notify_die(DIE_TRAP, str, regs, error_code, trapnr, signr) !=
			NOTIFY_STOP) {
275
		cond_local_irq_enable(regs);
276 277
		do_trap(trapnr, signr, str, regs, error_code,
			fill_trap_info(regs, signr, trapnr, &info));
278 279 280
	}
}

I
Ingo Molnar 已提交
281
#define DO_ERROR(trapnr, signr, str, name)				\
282
dotraplinkage void do_##name(struct pt_regs *regs, long error_code)	\
I
Ingo Molnar 已提交
283
{									\
284
	do_error_trap(regs, error_code, str, trapnr, signr);		\
L
Linus Torvalds 已提交
285 286
}

O
Oleg Nesterov 已提交
287 288 289 290 291 292 293 294
DO_ERROR(X86_TRAP_DE,     SIGFPE,  "divide error",		divide_error)
DO_ERROR(X86_TRAP_OF,     SIGSEGV, "overflow",			overflow)
DO_ERROR(X86_TRAP_UD,     SIGILL,  "invalid opcode",		invalid_op)
DO_ERROR(X86_TRAP_OLD_MF, SIGFPE,  "coprocessor segment overrun",coprocessor_segment_overrun)
DO_ERROR(X86_TRAP_TS,     SIGSEGV, "invalid TSS",		invalid_TSS)
DO_ERROR(X86_TRAP_NP,     SIGBUS,  "segment not present",	segment_not_present)
DO_ERROR(X86_TRAP_SS,     SIGBUS,  "stack segment",		stack_segment)
DO_ERROR(X86_TRAP_AC,     SIGBUS,  "alignment check",		alignment_check)
L
Linus Torvalds 已提交
295

296 297 298 299 300 301 302
#ifdef CONFIG_X86_64
/* Runs on IST stack */
dotraplinkage void do_double_fault(struct pt_regs *regs, long error_code)
{
	static const char str[] = "double fault";
	struct task_struct *tsk = current;

303 304 305 306 307 308 309 310
#ifdef CONFIG_X86_ESPFIX64
	extern unsigned char native_irq_return_iret[];

	/*
	 * If IRET takes a non-IST fault on the espfix64 stack, then we
	 * end up promoting it to a doublefault.  In that case, modify
	 * the stack to make it look like we just entered the #GP
	 * handler from user space, similar to bad_iret.
311 312
	 *
	 * No need for ist_enter here because we don't use RCU.
313 314 315 316 317 318 319 320 321 322 323 324
	 */
	if (((long)regs->sp >> PGDIR_SHIFT) == ESPFIX_PGD_ENTRY &&
		regs->cs == __KERNEL_CS &&
		regs->ip == (unsigned long)native_irq_return_iret)
	{
		struct pt_regs *normal_regs = task_pt_regs(current);

		/* Fake a #GP(0) from userspace. */
		memmove(&normal_regs->ip, (void *)regs->sp, 5*8);
		normal_regs->orig_ax = 0;  /* Missing (lost) #GP error code */
		regs->ip = (unsigned long)general_protection;
		regs->sp = (unsigned long)&normal_regs->orig_ax;
325

326 327 328 329
		return;
	}
#endif

330
	ist_enter(regs);
331
	notify_die(DIE_TRAP, str, regs, error_code, X86_TRAP_DF, SIGSEGV);
332 333

	tsk->thread.error_code = error_code;
334
	tsk->thread.trap_nr = X86_TRAP_DF;
335

336 337 338
#ifdef CONFIG_DOUBLEFAULT
	df_debug(regs, error_code);
#endif
339 340 341 342
	/*
	 * This is always a kernel trap and never fixable (and thus must
	 * never return).
	 */
343 344 345 346 347
	for (;;)
		die(str, regs, error_code);
}
#endif

348 349
dotraplinkage void do_bounds(struct pt_regs *regs, long error_code)
{
350
	const struct mpx_bndcsr *bndcsr;
351 352
	siginfo_t *info;

353
	RCU_LOCKDEP_WARN(!rcu_is_watching(), "entry code didn't wake RCU");
354 355
	if (notify_die(DIE_TRAP, "bounds", regs, error_code,
			X86_TRAP_BR, SIGSEGV) == NOTIFY_STOP)
356
		return;
357
	cond_local_irq_enable(regs);
358

359
	if (!user_mode(regs))
360 361 362 363 364 365 366 367 368
		die("bounds", regs, error_code);

	if (!cpu_feature_enabled(X86_FEATURE_MPX)) {
		/* The exception is not from Intel MPX */
		goto exit_trap;
	}

	/*
	 * We need to look at BNDSTATUS to resolve this exception.
369 370 371
	 * A NULL here might mean that it is in its 'init state',
	 * which is all zeros which indicates MPX was not
	 * responsible for the exception.
372
	 */
D
Dave Hansen 已提交
373
	bndcsr = get_xsave_field_ptr(XFEATURE_MASK_BNDCSR);
374 375 376
	if (!bndcsr)
		goto exit_trap;

D
Dave Hansen 已提交
377
	trace_bounds_exception_mpx(bndcsr);
378 379 380 381 382 383 384
	/*
	 * The error code field of the BNDSTATUS register communicates status
	 * information of a bound range exception #BR or operation involving
	 * bound directory.
	 */
	switch (bndcsr->bndstatus & MPX_BNDSTA_ERROR_CODE) {
	case 2:	/* Bound directory has invalid entry. */
385
		if (mpx_handle_bd_fault())
386 387 388
			goto exit_trap;
		break; /* Success, it was handled */
	case 1: /* Bound violation. */
389
		info = mpx_generate_siginfo(regs);
390
		if (IS_ERR(info)) {
391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413
			/*
			 * We failed to decode the MPX instruction.  Act as if
			 * the exception was not caused by MPX.
			 */
			goto exit_trap;
		}
		/*
		 * Success, we decoded the instruction and retrieved
		 * an 'info' containing the address being accessed
		 * which caused the exception.  This information
		 * allows and application to possibly handle the
		 * #BR exception itself.
		 */
		do_trap(X86_TRAP_BR, SIGSEGV, "bounds", regs, error_code, info);
		kfree(info);
		break;
	case 0: /* No exception caused by Intel MPX operations. */
		goto exit_trap;
	default:
		die("bounds", regs, error_code);
	}

	return;
414

415 416 417 418 419 420 421 422 423 424 425
exit_trap:
	/*
	 * This path out is for all the cases where we could not
	 * handle the exception in some way (like allocating a
	 * table or telling userspace about it.  We will also end
	 * up here if the kernel has MPX turned off at compile
	 * time..
	 */
	do_trap(X86_TRAP_BR, SIGSEGV, "bounds", regs, error_code, NULL);
}

426
dotraplinkage void
427
do_general_protection(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
428
{
429
	struct task_struct *tsk;
I
Ingo Molnar 已提交
430

431
	RCU_LOCKDEP_WARN(!rcu_is_watching(), "entry code didn't wake RCU");
432
	cond_local_irq_enable(regs);
433

434
	if (v8086_mode(regs)) {
435 436
		local_irq_enable();
		handle_vm86_fault((struct kernel_vm86_regs *) regs, error_code);
437
		return;
438
	}
L
Linus Torvalds 已提交
439

440
	tsk = current;
441
	if (!user_mode(regs)) {
442
		if (fixup_exception(regs))
443
			return;
444 445 446

		tsk->thread.error_code = error_code;
		tsk->thread.trap_nr = X86_TRAP_GP;
447 448
		if (notify_die(DIE_GPF, "general protection fault", regs, error_code,
			       X86_TRAP_GP, SIGSEGV) != NOTIFY_STOP)
449
			die("general protection fault", regs, error_code);
450
		return;
451
	}
L
Linus Torvalds 已提交
452

453
	tsk->thread.error_code = error_code;
454
	tsk->thread.trap_nr = X86_TRAP_GP;
I
Ingo Molnar 已提交
455

456 457
	if (show_unhandled_signals && unhandled_signal(tsk, SIGSEGV) &&
			printk_ratelimit()) {
458
		pr_info("%s[%d] general protection ip:%lx sp:%lx error:%lx",
459 460
			tsk->comm, task_pid_nr(tsk),
			regs->ip, regs->sp, error_code);
461
		print_vma_addr(" in ", regs->ip);
462
		pr_cont("\n");
463
	}
464

465
	force_sig_info(SIGSEGV, SEND_SIG_PRIV, tsk);
L
Linus Torvalds 已提交
466
}
467
NOKPROBE_SYMBOL(do_general_protection);
L
Linus Torvalds 已提交
468

469
/* May run on IST stack. */
470
dotraplinkage void notrace do_int3(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
471
{
472
#ifdef CONFIG_DYNAMIC_FTRACE
473 474 475 476 477 478
	/*
	 * ftrace must be first, everything else may cause a recursive crash.
	 * See note by declaration of modifying_ftrace_code in ftrace.c
	 */
	if (unlikely(atomic_read(&modifying_ftrace_code)) &&
	    ftrace_int3_handler(regs))
479 480
		return;
#endif
481 482 483
	if (poke_int3_handler(regs))
		return;

484
	ist_enter(regs);
485
	RCU_LOCKDEP_WARN(!rcu_is_watching(), "entry code didn't wake RCU");
486
#ifdef CONFIG_KGDB_LOW_LEVEL_TRAP
487 488
	if (kgdb_ll_trap(DIE_INT3, "int3", regs, error_code, X86_TRAP_BP,
				SIGTRAP) == NOTIFY_STOP)
489
		goto exit;
490
#endif /* CONFIG_KGDB_LOW_LEVEL_TRAP */
491

492 493
#ifdef CONFIG_KPROBES
	if (kprobe_int3_handler(regs))
494
		goto exit;
495 496
#endif

497 498
	if (notify_die(DIE_INT3, "int3", regs, error_code, X86_TRAP_BP,
			SIGTRAP) == NOTIFY_STOP)
499
		goto exit;
I
Ingo Molnar 已提交
500

501 502 503 504 505
	/*
	 * Let others (NMI) know that the debug stack is in use
	 * as we may switch to the interrupt stack.
	 */
	debug_stack_usage_inc();
506 507
	preempt_disable();
	cond_local_irq_enable(regs);
508
	do_trap(X86_TRAP_BP, SIGTRAP, "int3", regs, error_code, NULL);
509 510
	cond_local_irq_disable(regs);
	preempt_enable_no_resched();
511
	debug_stack_usage_dec();
512
exit:
513
	ist_exit(regs);
L
Linus Torvalds 已提交
514
}
515
NOKPROBE_SYMBOL(do_int3);
L
Linus Torvalds 已提交
516

517
#ifdef CONFIG_X86_64
518
/*
519 520 521
 * Help handler running on IST stack to switch off the IST stack if the
 * interrupted code was in user mode. The actual stack switch is done in
 * entry_64.S
522
 */
523
asmlinkage __visible notrace struct pt_regs *sync_regs(struct pt_regs *eregs)
524
{
525 526
	struct pt_regs *regs = task_pt_regs(current);
	*regs = *eregs;
527 528
	return regs;
}
529
NOKPROBE_SYMBOL(sync_regs);
A
Andy Lutomirski 已提交
530 531 532 533 534 535

struct bad_iret_stack {
	void *error_entry_ret;
	struct pt_regs regs;
};

536
asmlinkage __visible notrace
A
Andy Lutomirski 已提交
537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555
struct bad_iret_stack *fixup_bad_iret(struct bad_iret_stack *s)
{
	/*
	 * This is called from entry_64.S early in handling a fault
	 * caused by a bad iret to user mode.  To handle the fault
	 * correctly, we want move our stack frame to task_pt_regs
	 * and we want to pretend that the exception came from the
	 * iret target.
	 */
	struct bad_iret_stack *new_stack =
		container_of(task_pt_regs(current),
			     struct bad_iret_stack, regs);

	/* Copy the IRET target to the new stack. */
	memmove(&new_stack->regs.ip, (void *)s->regs.sp, 5*8);

	/* Copy the remainder of the stack from the current stack. */
	memmove(new_stack, s, offsetof(struct bad_iret_stack, regs.ip));

556
	BUG_ON(!user_mode(&new_stack->regs));
A
Andy Lutomirski 已提交
557 558
	return new_stack;
}
559
NOKPROBE_SYMBOL(fixup_bad_iret);
560 561
#endif

562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584
static bool is_sysenter_singlestep(struct pt_regs *regs)
{
	/*
	 * We don't try for precision here.  If we're anywhere in the region of
	 * code that can be single-stepped in the SYSENTER entry path, then
	 * assume that this is a useless single-step trap due to SYSENTER
	 * being invoked with TF set.  (We don't know in advance exactly
	 * which instructions will be hit because BTF could plausibly
	 * be set.)
	 */
#ifdef CONFIG_X86_32
	return (regs->ip - (unsigned long)__begin_SYSENTER_singlestep_region) <
		(unsigned long)__end_SYSENTER_singlestep_region -
		(unsigned long)__begin_SYSENTER_singlestep_region;
#elif defined(CONFIG_IA32_EMULATION)
	return (regs->ip - (unsigned long)entry_SYSENTER_compat) <
		(unsigned long)__end_entry_SYSENTER_compat -
		(unsigned long)entry_SYSENTER_compat;
#else
	return false;
#endif
}

L
Linus Torvalds 已提交
585 586 587 588 589 590 591 592 593 594
/*
 * Our handling of the processor debug registers is non-trivial.
 * We do not clear them on entry and exit from the kernel. Therefore
 * it is possible to get a watchpoint trap here from inside the kernel.
 * However, the code in ./ptrace.c has ensured that the user can
 * only set watchpoints on userspace addresses. Therefore the in-kernel
 * watchpoint trap can only occur in code which is reading/writing
 * from user space. Such code must not hold kernel locks (since it
 * can equally take a page fault), therefore it is safe to call
 * force_sig_info even though that claims and releases locks.
I
Ingo Molnar 已提交
595
 *
L
Linus Torvalds 已提交
596 597 598 599 600 601 602 603 604 605
 * Code in ./signal.c ensures that the debug control register
 * is restored before we deliver any signal, and therefore that
 * user code runs with the correct debug control register even though
 * we clear it here.
 *
 * Being careful here means that we don't have to be as careful in a
 * lot of more complicated places (task switching can be a bit lazy
 * about restoring all the debug state, and ptrace doesn't have to
 * find every occurrence of the TF bit that could be saved away even
 * by user code)
606 607
 *
 * May run on IST stack.
L
Linus Torvalds 已提交
608
 */
609
dotraplinkage void do_debug(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
610 611
{
	struct task_struct *tsk = current;
612
	int user_icebp = 0;
613
	unsigned long dr6;
614
	int si_code;
L
Linus Torvalds 已提交
615

616
	ist_enter(regs);
617

618
	get_debugreg(dr6, 6);
619 620 621 622 623 624 625 626 627 628 629 630
	/*
	 * The Intel SDM says:
	 *
	 *   Certain debug exceptions may clear bits 0-3. The remaining
	 *   contents of the DR6 register are never cleared by the
	 *   processor. To avoid confusion in identifying debug
	 *   exceptions, debug handlers should clear the register before
	 *   returning to the interrupted task.
	 *
	 * Keep it simple: clear DR6 immediately.
	 */
	set_debugreg(0, 6);
L
Linus Torvalds 已提交
631

632 633 634
	/* Filter out all the reserved bits which are preset to 1 */
	dr6 &= ~DR6_RESERVED;

635 636 637 638 639 640 641
	/*
	 * The SDM says "The processor clears the BTF flag when it
	 * generates a debug exception."  Clear TIF_BLOCKSTEP to keep
	 * TIF_BLOCKSTEP in sync with the hardware BTF flag.
	 */
	clear_tsk_thread_flag(tsk, TIF_BLOCKSTEP);

642 643 644 645 646 647 648 649 650 651 652 653
	if (unlikely(!user_mode(regs) && (dr6 & DR_STEP) &&
		     is_sysenter_singlestep(regs))) {
		dr6 &= ~DR_STEP;
		if (!dr6)
			goto exit;
		/*
		 * else we might have gotten a single-step trap and hit a
		 * watchpoint at the same time, in which case we should fall
		 * through and handle the watchpoint.
		 */
	}

654 655 656 657 658
	/*
	 * If dr6 has no reason to give us about the origin of this trap,
	 * then it's very likely the result of an icebp/int01 trap.
	 * User wants a sigtrap for that.
	 */
659
	if (!dr6 && user_mode(regs))
660 661
		user_icebp = 1;

662
	/* Catch kmemcheck conditions! */
663
	if ((dr6 & DR_STEP) && kmemcheck_trap(regs))
664
		goto exit;
V
Vegard Nossum 已提交
665

666 667 668
	/* Store the virtualized DR6 value */
	tsk->thread.debugreg6 = dr6;

669 670 671 672 673
#ifdef CONFIG_KPROBES
	if (kprobe_debug_handler(regs))
		goto exit;
#endif

674
	if (notify_die(DIE_DEBUG, "debug", regs, (long)&dr6, error_code,
675
							SIGTRAP) == NOTIFY_STOP)
676
		goto exit;
677

678 679 680 681 682 683
	/*
	 * Let others (NMI) know that the debug stack is in use
	 * as we may switch to the interrupt stack.
	 */
	debug_stack_usage_inc();

L
Linus Torvalds 已提交
684
	/* It's safe to allow irq's after DR6 has been saved */
685 686
	preempt_disable();
	cond_local_irq_enable(regs);
L
Linus Torvalds 已提交
687

688
	if (v8086_mode(regs)) {
689 690
		handle_vm86_trap((struct kernel_vm86_regs *) regs, error_code,
					X86_TRAP_DB);
691 692
		cond_local_irq_disable(regs);
		preempt_enable_no_resched();
693
		debug_stack_usage_dec();
694
		goto exit;
L
Linus Torvalds 已提交
695 696
	}

697 698 699 700 701 702 703
	if (WARN_ON_ONCE((dr6 & DR_STEP) && !user_mode(regs))) {
		/*
		 * Historical junk that used to handle SYSENTER single-stepping.
		 * This should be unreachable now.  If we survive for a while
		 * without anyone hitting this warning, we'll turn this into
		 * an oops.
		 */
704 705 706
		tsk->thread.debugreg6 &= ~DR_STEP;
		set_tsk_thread_flag(tsk, TIF_SINGLESTEP);
		regs->flags &= ~X86_EFLAGS_TF;
L
Linus Torvalds 已提交
707
	}
708
	si_code = get_si_code(tsk->thread.debugreg6);
709
	if (tsk->thread.debugreg6 & (DR_STEP | DR_TRAP_BITS) || user_icebp)
710
		send_sigtrap(tsk, regs, error_code, si_code);
711 712
	cond_local_irq_disable(regs);
	preempt_enable_no_resched();
713
	debug_stack_usage_dec();
L
Linus Torvalds 已提交
714

715
exit:
716 717 718 719 720 721 722 723
#if defined(CONFIG_X86_32)
	/*
	 * This is the most likely code path that involves non-trivial use
	 * of the SYSENTER stack.  Check that we haven't overrun it.
	 */
	WARN(this_cpu_read(cpu_tss.SYSENTER_stack_canary) != STACK_END_MAGIC,
	     "Overran or corrupted SYSENTER stack\n");
#endif
724
	ist_exit(regs);
L
Linus Torvalds 已提交
725
}
726
NOKPROBE_SYMBOL(do_debug);
L
Linus Torvalds 已提交
727 728 729 730 731 732

/*
 * Note that we play around with the 'TS' bit in an attempt to get
 * the correct behaviour even in the presence of the asynchronous
 * IRQ13 behaviour
 */
733
static void math_error(struct pt_regs *regs, int error_code, int trapnr)
L
Linus Torvalds 已提交
734
{
735
	struct task_struct *task = current;
736
	struct fpu *fpu = &task->thread.fpu;
L
Linus Torvalds 已提交
737
	siginfo_t info;
738 739
	char *str = (trapnr == X86_TRAP_MF) ? "fpu exception" :
						"simd exception";
740 741 742

	if (notify_die(DIE_TRAP, str, regs, error_code, trapnr, SIGFPE) == NOTIFY_STOP)
		return;
743
	cond_local_irq_enable(regs);
744

745
	if (!user_mode(regs)) {
746 747
		if (!fixup_exception(regs)) {
			task->thread.error_code = error_code;
748
			task->thread.trap_nr = trapnr;
749 750 751 752
			die(str, regs, error_code);
		}
		return;
	}
L
Linus Torvalds 已提交
753 754 755 756

	/*
	 * Save the info for the exception handler and clear the error.
	 */
757 758 759
	fpu__save(fpu);

	task->thread.trap_nr	= trapnr;
760
	task->thread.error_code = error_code;
761 762 763
	info.si_signo		= SIGFPE;
	info.si_errno		= 0;
	info.si_addr		= (void __user *)uprobe_get_trap_addr(regs);
764

765
	info.si_code = fpu__exception_code(fpu, trapnr);
766

767 768
	/* Retry when we get spurious exceptions: */
	if (!info.si_code)
769
		return;
770

L
Linus Torvalds 已提交
771 772 773
	force_sig_info(SIGFPE, &info, task);
}

774
dotraplinkage void do_coprocessor_error(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
775
{
776
	RCU_LOCKDEP_WARN(!rcu_is_watching(), "entry code didn't wake RCU");
777
	math_error(regs, error_code, X86_TRAP_MF);
L
Linus Torvalds 已提交
778 779
}

780 781
dotraplinkage void
do_simd_coprocessor_error(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
782
{
783
	RCU_LOCKDEP_WARN(!rcu_is_watching(), "entry code didn't wake RCU");
784
	math_error(regs, error_code, X86_TRAP_XF);
L
Linus Torvalds 已提交
785 786
}

787 788
dotraplinkage void
do_spurious_interrupt_bug(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
789
{
790
	cond_local_irq_enable(regs);
791 792
}

793
dotraplinkage void
794
do_device_not_available(struct pt_regs *regs, long error_code)
795
{
796
	RCU_LOCKDEP_WARN(!rcu_is_watching(), "entry code didn't wake RCU");
797
	BUG_ON(use_eager_fpu());
798

799
#ifdef CONFIG_MATH_EMULATION
800
	if (read_cr0() & X86_CR0_EM) {
801 802
		struct math_emu_info info = { };

803
		cond_local_irq_enable(regs);
804

805
		info.regs = regs;
806
		math_emulate(&info);
807
		return;
808
	}
809
#endif
810
	fpu__restore(&current->thread.fpu); /* interrupts still off */
811
#ifdef CONFIG_X86_32
812
	cond_local_irq_enable(regs);
813
#endif
814
}
815
NOKPROBE_SYMBOL(do_device_not_available);
816

817
#ifdef CONFIG_X86_32
818
dotraplinkage void do_iret_error(struct pt_regs *regs, long error_code)
819 820
{
	siginfo_t info;
821

822
	RCU_LOCKDEP_WARN(!rcu_is_watching(), "entry code didn't wake RCU");
823 824 825 826 827
	local_irq_enable();

	info.si_signo = SIGILL;
	info.si_errno = 0;
	info.si_code = ILL_BADSTK;
828
	info.si_addr = NULL;
829
	if (notify_die(DIE_TRAP, "iret exception", regs, error_code,
830 831 832 833
			X86_TRAP_IRET, SIGILL) != NOTIFY_STOP) {
		do_trap(X86_TRAP_IRET, SIGILL, "iret exception", regs, error_code,
			&info);
	}
834
}
835
#endif
836

837 838 839
/* Set of traps needed for early debugging. */
void __init early_trap_init(void)
{
840
	/*
841 842 843 844 845
	 * Don't use IST to set DEBUG_STACK as it doesn't work until TSS
	 * is ready in cpu_init() <-- trap_init(). Before trap_init(),
	 * CPU runs at ring 0 so it is impossible to hit an invalid
	 * stack.  Using the original stack works well enough at this
	 * early stage. DEBUG_STACK will be equipped after cpu_init() in
846
	 * trap_init().
847 848 849 850
	 *
	 * We don't need to set trace_idt_table like set_intr_gate(),
	 * since we don't have trace_debug and it will be reset to
	 * 'debug' in trap_init() by set_intr_gate_ist().
851
	 */
852
	set_intr_gate_notrace(X86_TRAP_DB, debug);
853
	/* int3 can be called from all */
854
	set_system_intr_gate(X86_TRAP_BP, &int3);
855
#ifdef CONFIG_X86_32
856
	set_intr_gate(X86_TRAP_PF, page_fault);
857
#endif
858 859 860
	load_idt(&idt_descr);
}

861 862 863
void __init early_trap_pf_init(void)
{
#ifdef CONFIG_X86_64
864
	set_intr_gate(X86_TRAP_PF, page_fault);
865 866 867
#endif
}

L
Linus Torvalds 已提交
868 869
void __init trap_init(void)
{
870 871
	int i;

L
Linus Torvalds 已提交
872
#ifdef CONFIG_EISA
I
Ingo Molnar 已提交
873
	void __iomem *p = early_ioremap(0x0FFFD9, 4);
I
Ingo Molnar 已提交
874 875

	if (readl(p) == 'E' + ('I'<<8) + ('S'<<16) + ('A'<<24))
L
Linus Torvalds 已提交
876
		EISA_bus = 1;
I
Ingo Molnar 已提交
877
	early_iounmap(p, 4);
L
Linus Torvalds 已提交
878 879
#endif

880
	set_intr_gate(X86_TRAP_DE, divide_error);
881
	set_intr_gate_ist(X86_TRAP_NMI, &nmi, NMI_STACK);
882
	/* int4 can be called from all */
883
	set_system_intr_gate(X86_TRAP_OF, &overflow);
884 885 886
	set_intr_gate(X86_TRAP_BR, bounds);
	set_intr_gate(X86_TRAP_UD, invalid_op);
	set_intr_gate(X86_TRAP_NM, device_not_available);
887
#ifdef CONFIG_X86_32
888
	set_task_gate(X86_TRAP_DF, GDT_ENTRY_DOUBLEFAULT_TSS);
889
#else
890
	set_intr_gate_ist(X86_TRAP_DF, &double_fault, DOUBLEFAULT_STACK);
891
#endif
892 893 894
	set_intr_gate(X86_TRAP_OLD_MF, coprocessor_segment_overrun);
	set_intr_gate(X86_TRAP_TS, invalid_TSS);
	set_intr_gate(X86_TRAP_NP, segment_not_present);
895
	set_intr_gate(X86_TRAP_SS, stack_segment);
896 897 898 899
	set_intr_gate(X86_TRAP_GP, general_protection);
	set_intr_gate(X86_TRAP_SPURIOUS, spurious_interrupt_bug);
	set_intr_gate(X86_TRAP_MF, coprocessor_error);
	set_intr_gate(X86_TRAP_AC, alignment_check);
L
Linus Torvalds 已提交
900
#ifdef CONFIG_X86_MCE
901
	set_intr_gate_ist(X86_TRAP_MC, &machine_check, MCE_STACK);
L
Linus Torvalds 已提交
902
#endif
903
	set_intr_gate(X86_TRAP_XF, simd_coprocessor_error);
L
Linus Torvalds 已提交
904

905 906 907 908
	/* Reserve all the builtin and the syscall vector: */
	for (i = 0; i < FIRST_EXTERNAL_VECTOR; i++)
		set_bit(i, used_vectors);

909
#ifdef CONFIG_IA32_EMULATION
910
	set_system_intr_gate(IA32_SYSCALL_VECTOR, entry_INT80_compat);
911
	set_bit(IA32_SYSCALL_VECTOR, used_vectors);
912 913 914
#endif

#ifdef CONFIG_X86_32
915
	set_system_trap_gate(IA32_SYSCALL_VECTOR, entry_INT80_32);
916
	set_bit(IA32_SYSCALL_VECTOR, used_vectors);
917
#endif
918

919 920 921 922 923 924 925 926
	/*
	 * Set the IDT descriptor to a fixed read-only location, so that the
	 * "sidt" instruction will not leak the location of the kernel, and
	 * to defend the IDT against arbitrary memory write vulnerabilities.
	 * It will be reloaded in cpu_init() */
	__set_fixmap(FIX_RO_IDT, __pa_symbol(idt_table), PAGE_KERNEL_RO);
	idt_descr.address = fix_to_virt(FIX_RO_IDT);

L
Linus Torvalds 已提交
927
	/*
I
Ingo Molnar 已提交
928
	 * Should be a barrier for any external CPU state:
L
Linus Torvalds 已提交
929 930 931
	 */
	cpu_init();

932 933
	/*
	 * X86_TRAP_DB and X86_TRAP_BP have been set
934
	 * in early_trap_init(). However, ITS works only after
935 936 937 938 939 940
	 * cpu_init() loads TSS. See comments in early_trap_init().
	 */
	set_intr_gate_ist(X86_TRAP_DB, &debug, DEBUG_STACK);
	/* int3 can be called from all */
	set_system_intr_gate_ist(X86_TRAP_BP, &int3, DEBUG_STACK);

941
	x86_init.irqs.trap_init();
942 943

#ifdef CONFIG_X86_64
944
	memcpy(&debug_idt_table, &idt_table, IDT_ENTRIES * 16);
945 946
	set_nmi_gate(X86_TRAP_DB, &debug);
	set_nmi_gate(X86_TRAP_BP, &int3);
947
#endif
L
Linus Torvalds 已提交
948
}