traps.c 20.1 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

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

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

D
Dave Jiang 已提交
43 44 45 46
#if defined(CONFIG_EDAC)
#include <linux/edac.h>
#endif

V
Vegard Nossum 已提交
47
#include <asm/kmemcheck.h>
I
Ingo Molnar 已提交
48
#include <asm/stacktrace.h>
L
Linus Torvalds 已提交
49 50
#include <asm/processor.h>
#include <asm/debugreg.h>
A
Arun Sharma 已提交
51
#include <linux/atomic.h>
52
#include <asm/ftrace.h>
53
#include <asm/traps.h>
L
Linus Torvalds 已提交
54 55
#include <asm/desc.h>
#include <asm/i387.h>
56
#include <asm/fpu-internal.h>
H
Hidetoshi Seto 已提交
57
#include <asm/mce.h>
58
#include <asm/context_tracking.h>
59

60
#include <asm/mach_traps.h>
61

62
#ifdef CONFIG_X86_64
63
#include <asm/x86_init.h>
64 65 66
#include <asm/pgalloc.h>
#include <asm/proto.h>
#else
67
#include <asm/processor-flags.h>
I
Ingo Molnar 已提交
68
#include <asm/setup.h>
L
Linus Torvalds 已提交
69 70 71 72 73

asmlinkage int system_call(void);

/*
 * The IDT has to be page-aligned to simplify the Pentium
74
 * F0 0F bug workaround.
L
Linus Torvalds 已提交
75
 */
76
gate_desc idt_table[NR_VECTORS] __page_aligned_data = { { { { 0, 0 } } }, };
77
#endif
L
Linus Torvalds 已提交
78

79 80 81
DECLARE_BITMAP(used_vectors, NR_VECTORS);
EXPORT_SYMBOL_GPL(used_vectors);

82 83 84 85 86 87
static inline void conditional_sti(struct pt_regs *regs)
{
	if (regs->flags & X86_EFLAGS_IF)
		local_irq_enable();
}

88 89 90 91 92 93 94
static inline void preempt_conditional_sti(struct pt_regs *regs)
{
	inc_preempt_count();
	if (regs->flags & X86_EFLAGS_IF)
		local_irq_enable();
}

T
Thomas Gleixner 已提交
95 96 97 98 99 100
static inline void conditional_cli(struct pt_regs *regs)
{
	if (regs->flags & X86_EFLAGS_IF)
		local_irq_disable();
}

101 102 103 104 105 106 107
static inline void preempt_conditional_cli(struct pt_regs *regs)
{
	if (regs->flags & X86_EFLAGS_IF)
		local_irq_disable();
	dec_preempt_count();
}

108 109 110
static int __kprobes
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
111
{
112
#ifdef CONFIG_X86_32
113
	if (regs->flags & X86_VM_MASK) {
114
		/*
115
		 * Traps 0, 1, 3, 4, and 5 should be forwarded to vm86.
116 117
		 * On nmi (interrupt 2), do_trap should not be called.
		 */
118 119 120 121 122 123
		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
124
	}
125
#endif
126 127 128 129 130 131 132 133
	if (!user_mode(regs)) {
		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
134

135 136
	return -1;
}
L
Linus Torvalds 已提交
137

138 139 140 141 142 143 144 145 146
static void __kprobes
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
147
	/*
148
	 * We want error_code and trap_nr set for userspace faults and
I
Ingo Molnar 已提交
149 150 151 152 153 154 155 156
	 * 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;
157
	tsk->thread.trap_nr = trapnr;
158

159 160 161
#ifdef CONFIG_X86_64
	if (show_unhandled_signals && unhandled_signal(tsk, signr) &&
	    printk_ratelimit()) {
162 163 164
		pr_info("%s[%d] trap %s ip:%lx sp:%lx error:%lx",
			tsk->comm, tsk->pid, str,
			regs->ip, regs->sp, error_code);
165
		print_vma_addr(" in ", regs->ip);
166
		pr_cont("\n");
167 168 169
	}
#endif

I
Ingo Molnar 已提交
170 171 172 173
	if (info)
		force_sig_info(signr, info, tsk);
	else
		force_sig(signr, tsk);
L
Linus Torvalds 已提交
174 175
}

I
Ingo Molnar 已提交
176
#define DO_ERROR(trapnr, signr, str, name)				\
177
dotraplinkage void do_##name(struct pt_regs *regs, long error_code)	\
I
Ingo Molnar 已提交
178
{									\
179 180 181 182
	exception_enter(regs);						\
	if (notify_die(DIE_TRAP, str, regs, error_code,			\
			trapnr, signr) == NOTIFY_STOP) {		\
		exception_exit(regs);					\
I
Ingo Molnar 已提交
183
		return;							\
184
	}								\
185
	conditional_sti(regs);						\
186
	do_trap(trapnr, signr, str, regs, error_code, NULL);		\
187
	exception_exit(regs);						\
L
Linus Torvalds 已提交
188 189
}

190
#define DO_ERROR_INFO(trapnr, signr, str, name, sicode, siaddr)		\
191
dotraplinkage void do_##name(struct pt_regs *regs, long error_code)	\
I
Ingo Molnar 已提交
192 193 194 195 196 197
{									\
	siginfo_t info;							\
	info.si_signo = signr;						\
	info.si_errno = 0;						\
	info.si_code = sicode;						\
	info.si_addr = (void __user *)siaddr;				\
198 199 200 201
	exception_enter(regs);						\
	if (notify_die(DIE_TRAP, str, regs, error_code,			\
			trapnr, signr) == NOTIFY_STOP) {		\
		exception_exit(regs);					\
I
Ingo Molnar 已提交
202
		return;							\
203
	}								\
204
	conditional_sti(regs);						\
205
	do_trap(trapnr, signr, str, regs, error_code, &info);		\
206
	exception_exit(regs);						\
L
Linus Torvalds 已提交
207 208
}

209 210 211 212 213 214 215 216 217 218
DO_ERROR_INFO(X86_TRAP_DE, SIGFPE, "divide error", divide_error, FPE_INTDIV,
		regs->ip)
DO_ERROR(X86_TRAP_OF, SIGSEGV, "overflow", overflow)
DO_ERROR(X86_TRAP_BR, SIGSEGV, "bounds", bounds)
DO_ERROR_INFO(X86_TRAP_UD, SIGILL, "invalid opcode", invalid_op, ILL_ILLOPN,
		regs->ip)
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)
219
#ifdef CONFIG_X86_32
220
DO_ERROR(X86_TRAP_SS, SIGBUS, "stack segment", stack_segment)
221
#endif
222 223
DO_ERROR_INFO(X86_TRAP_AC, SIGBUS, "alignment check", alignment_check,
		BUS_ADRALN, 0)
L
Linus Torvalds 已提交
224

225 226 227 228
#ifdef CONFIG_X86_64
/* Runs on IST stack */
dotraplinkage void do_stack_segment(struct pt_regs *regs, long error_code)
{
229
	exception_enter(regs);
230
	if (notify_die(DIE_TRAP, "stack segment", regs, error_code,
231 232 233 234 235 236
		       X86_TRAP_SS, SIGBUS) != NOTIFY_STOP) {
		preempt_conditional_sti(regs);
		do_trap(X86_TRAP_SS, SIGBUS, "stack segment", regs, error_code, NULL);
		preempt_conditional_cli(regs);
	}
	exception_exit(regs);
237 238 239 240 241 242 243
}

dotraplinkage void do_double_fault(struct pt_regs *regs, long error_code)
{
	static const char str[] = "double fault";
	struct task_struct *tsk = current;

244
	exception_enter(regs);
245
	/* Return not checked because double check cannot be ignored */
246
	notify_die(DIE_TRAP, str, regs, error_code, X86_TRAP_DF, SIGSEGV);
247 248

	tsk->thread.error_code = error_code;
249
	tsk->thread.trap_nr = X86_TRAP_DF;
250

251 252 253 254
	/*
	 * This is always a kernel trap and never fixable (and thus must
	 * never return).
	 */
255 256 257 258 259
	for (;;)
		die(str, regs, error_code);
}
#endif

260
dotraplinkage void __kprobes
261
do_general_protection(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
262
{
263
	struct task_struct *tsk;
I
Ingo Molnar 已提交
264

265
	exception_enter(regs);
266 267
	conditional_sti(regs);

268
#ifdef CONFIG_X86_32
269 270 271
	if (regs->flags & X86_VM_MASK) {
		local_irq_enable();
		handle_vm86_fault((struct kernel_vm86_regs *) regs, error_code);
272
		goto exit;
273
	}
274
#endif
L
Linus Torvalds 已提交
275

276
	tsk = current;
277 278
	if (!user_mode(regs)) {
		if (fixup_exception(regs))
279
			goto exit;
280 281 282

		tsk->thread.error_code = error_code;
		tsk->thread.trap_nr = X86_TRAP_GP;
283 284
		if (notify_die(DIE_GPF, "general protection fault", regs, error_code,
			       X86_TRAP_GP, SIGSEGV) != NOTIFY_STOP)
285
			die("general protection fault", regs, error_code);
286
		goto exit;
287
	}
L
Linus Torvalds 已提交
288

289
	tsk->thread.error_code = error_code;
290
	tsk->thread.trap_nr = X86_TRAP_GP;
I
Ingo Molnar 已提交
291

292 293
	if (show_unhandled_signals && unhandled_signal(tsk, SIGSEGV) &&
			printk_ratelimit()) {
294
		pr_info("%s[%d] general protection ip:%lx sp:%lx error:%lx",
295 296
			tsk->comm, task_pid_nr(tsk),
			regs->ip, regs->sp, error_code);
297
		print_vma_addr(" in ", regs->ip);
298
		pr_cont("\n");
299
	}
300

301
	force_sig(SIGSEGV, tsk);
302 303
exit:
	exception_exit(regs);
L
Linus Torvalds 已提交
304 305
}

306
/* May run on IST stack. */
307
dotraplinkage void __kprobes notrace do_int3(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
308
{
309
#ifdef CONFIG_DYNAMIC_FTRACE
310 311 312 313 314 315
	/*
	 * 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))
316 317
		return;
#endif
318
	exception_enter(regs);
319
#ifdef CONFIG_KGDB_LOW_LEVEL_TRAP
320 321
	if (kgdb_ll_trap(DIE_INT3, "int3", regs, error_code, X86_TRAP_BP,
				SIGTRAP) == NOTIFY_STOP)
322
		goto exit;
323
#endif /* CONFIG_KGDB_LOW_LEVEL_TRAP */
324

325 326
	if (notify_die(DIE_INT3, "int3", regs, error_code, X86_TRAP_BP,
			SIGTRAP) == NOTIFY_STOP)
327
		goto exit;
I
Ingo Molnar 已提交
328

329 330 331 332 333
	/*
	 * Let others (NMI) know that the debug stack is in use
	 * as we may switch to the interrupt stack.
	 */
	debug_stack_usage_inc();
334
	preempt_conditional_sti(regs);
335
	do_trap(X86_TRAP_BP, SIGTRAP, "int3", regs, error_code, NULL);
336
	preempt_conditional_cli(regs);
337
	debug_stack_usage_dec();
338 339
exit:
	exception_exit(regs);
L
Linus Torvalds 已提交
340 341
}

342
#ifdef CONFIG_X86_64
343 344 345 346 347
/*
 * Help handler running on IST stack to switch back to user stack
 * for scheduling or signal handling. The actual stack switch is done in
 * entry.S
 */
348 349 350 351 352 353 354 355 356
asmlinkage __kprobes struct pt_regs *sync_regs(struct pt_regs *eregs)
{
	struct pt_regs *regs = eregs;
	/* Did already sync */
	if (eregs == (struct pt_regs *)eregs->sp)
		;
	/* Exception from user space */
	else if (user_mode(eregs))
		regs = task_pt_regs(current);
357 358 359 360
	/*
	 * Exception from kernel and interrupts are enabled. Move to
	 * kernel process stack.
	 */
361 362 363 364 365 366 367 368
	else if (eregs->flags & X86_EFLAGS_IF)
		regs = (struct pt_regs *)(eregs->sp -= sizeof(struct pt_regs));
	if (eregs != regs)
		*regs = *eregs;
	return regs;
}
#endif

L
Linus Torvalds 已提交
369 370 371 372 373 374 375 376 377 378
/*
 * 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
379
 *
L
Linus Torvalds 已提交
380 381 382 383 384 385 386 387 388 389
 * 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)
390 391
 *
 * May run on IST stack.
L
Linus Torvalds 已提交
392
 */
393
dotraplinkage void __kprobes do_debug(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
394 395
{
	struct task_struct *tsk = current;
396
	int user_icebp = 0;
397
	unsigned long dr6;
398
	int si_code;
L
Linus Torvalds 已提交
399

400 401
	exception_enter(regs);

402
	get_debugreg(dr6, 6);
L
Linus Torvalds 已提交
403

404 405 406
	/* Filter out all the reserved bits which are preset to 1 */
	dr6 &= ~DR6_RESERVED;

407 408 409 410 411 412 413 414
	/*
	 * 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.
	 */
	if (!dr6 && user_mode(regs))
		user_icebp = 1;

V
Vegard Nossum 已提交
415
	/* Catch kmemcheck conditions first of all! */
416
	if ((dr6 & DR_STEP) && kmemcheck_trap(regs))
417
		goto exit;
V
Vegard Nossum 已提交
418

419 420
	/* DR6 may or may not be cleared by the CPU */
	set_debugreg(0, 6);
421

P
Peter Zijlstra 已提交
422 423 424 425 426
	/*
	 * The processor cleared BTF, so don't mark that we need it set.
	 */
	clear_tsk_thread_flag(tsk, TIF_BLOCKSTEP);

427 428 429
	/* Store the virtualized DR6 value */
	tsk->thread.debugreg6 = dr6;

430 431
	if (notify_die(DIE_DEBUG, "debug", regs, PTR_ERR(&dr6), error_code,
							SIGTRAP) == NOTIFY_STOP)
432
		goto exit;
433

434 435 436 437 438 439
	/*
	 * 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
440
	/* It's safe to allow irq's after DR6 has been saved */
441
	preempt_conditional_sti(regs);
L
Linus Torvalds 已提交
442

443
	if (regs->flags & X86_VM_MASK) {
444 445
		handle_vm86_trap((struct kernel_vm86_regs *) regs, error_code,
					X86_TRAP_DB);
446
		preempt_conditional_cli(regs);
447
		debug_stack_usage_dec();
448
		goto exit;
L
Linus Torvalds 已提交
449 450 451
	}

	/*
452 453 454 455 456
	 * Single-stepping through system calls: ignore any exceptions in
	 * kernel space, but re-enable TF when returning to user mode.
	 *
	 * We already checked v86 mode above, so we can check for kernel mode
	 * by just checking the CPL of CS.
L
Linus Torvalds 已提交
457
	 */
458 459 460 461
	if ((dr6 & DR_STEP) && !user_mode(regs)) {
		tsk->thread.debugreg6 &= ~DR_STEP;
		set_tsk_thread_flag(tsk, TIF_SINGLESTEP);
		regs->flags &= ~X86_EFLAGS_TF;
L
Linus Torvalds 已提交
462
	}
463
	si_code = get_si_code(tsk->thread.debugreg6);
464
	if (tsk->thread.debugreg6 & (DR_STEP | DR_TRAP_BITS) || user_icebp)
465
		send_sigtrap(tsk, regs, error_code, si_code);
466
	preempt_conditional_cli(regs);
467
	debug_stack_usage_dec();
L
Linus Torvalds 已提交
468

469 470
exit:
	exception_exit(regs);
L
Linus Torvalds 已提交
471 472 473 474 475 476 477
}

/*
 * 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
 */
478
void math_error(struct pt_regs *regs, int error_code, int trapnr)
L
Linus Torvalds 已提交
479
{
480
	struct task_struct *task = current;
L
Linus Torvalds 已提交
481
	siginfo_t info;
482
	unsigned short err;
483 484
	char *str = (trapnr == X86_TRAP_MF) ? "fpu exception" :
						"simd exception";
485 486 487 488 489 490 491 492 493

	if (notify_die(DIE_TRAP, str, regs, error_code, trapnr, SIGFPE) == NOTIFY_STOP)
		return;
	conditional_sti(regs);

	if (!user_mode_vm(regs))
	{
		if (!fixup_exception(regs)) {
			task->thread.error_code = error_code;
494
			task->thread.trap_nr = trapnr;
495 496 497 498
			die(str, regs, error_code);
		}
		return;
	}
L
Linus Torvalds 已提交
499 500 501 502 503

	/*
	 * Save the info for the exception handler and clear the error.
	 */
	save_init_fpu(task);
504
	task->thread.trap_nr = trapnr;
505
	task->thread.error_code = error_code;
L
Linus Torvalds 已提交
506 507
	info.si_signo = SIGFPE;
	info.si_errno = 0;
508
	info.si_addr = (void __user *)regs->ip;
509
	if (trapnr == X86_TRAP_MF) {
510 511 512 513 514 515 516 517 518 519 520 521 522
		unsigned short cwd, swd;
		/*
		 * (~cwd & swd) will mask out exceptions that are not set to unmasked
		 * status.  0x3f is the exception bits in these regs, 0x200 is the
		 * C1 reg you need in case of a stack fault, 0x040 is the stack
		 * fault bit.  We should only be taking one exception at a time,
		 * so if this combination doesn't produce any single exception,
		 * then we have a bad program that isn't synchronizing its FPU usage
		 * and it will suffer the consequences since we won't be able to
		 * fully reproduce the context of the exception
		 */
		cwd = get_fpu_cwd(task);
		swd = get_fpu_swd(task);
523

524 525 526 527 528 529 530 531 532 533 534
		err = swd & ~cwd;
	} else {
		/*
		 * The SIMD FPU exceptions are handled a little differently, as there
		 * is only a single status/control register.  Thus, to determine which
		 * unmasked exception was caught we must mask the exception mask bits
		 * at 0x1f80, and then use these to mask the exception bits at 0x3f.
		 */
		unsigned short mxcsr = get_fpu_mxcsr(task);
		err = ~(mxcsr >> 7) & mxcsr;
	}
535 536

	if (err & 0x001) {	/* Invalid op */
I
Ingo Molnar 已提交
537 538 539 540 541 542
		/*
		 * swd & 0x240 == 0x040: Stack Underflow
		 * swd & 0x240 == 0x240: Stack Overflow
		 * User must clear the SF bit (0x40) if set
		 */
		info.si_code = FPE_FLTINV;
543
	} else if (err & 0x004) { /* Divide by Zero */
I
Ingo Molnar 已提交
544
		info.si_code = FPE_FLTDIV;
545
	} else if (err & 0x008) { /* Overflow */
I
Ingo Molnar 已提交
546
		info.si_code = FPE_FLTOVF;
547 548 549
	} else if (err & 0x012) { /* Denormal, Underflow */
		info.si_code = FPE_FLTUND;
	} else if (err & 0x020) { /* Precision */
I
Ingo Molnar 已提交
550
		info.si_code = FPE_FLTRES;
551
	} else {
552
		/*
553 554 555
		 * If we're using IRQ 13, or supposedly even some trap
		 * X86_TRAP_MF implementations, it's possible
		 * we get a spurious trap, which is not an error.
556
		 */
557
		return;
L
Linus Torvalds 已提交
558 559 560 561
	}
	force_sig_info(SIGFPE, &info, task);
}

562
dotraplinkage void do_coprocessor_error(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
563
{
564
	exception_enter(regs);
565
	math_error(regs, error_code, X86_TRAP_MF);
566
	exception_exit(regs);
L
Linus Torvalds 已提交
567 568
}

569 570
dotraplinkage void
do_simd_coprocessor_error(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
571
{
572
	exception_enter(regs);
573
	math_error(regs, error_code, X86_TRAP_XF);
574
	exception_exit(regs);
L
Linus Torvalds 已提交
575 576
}

577 578
dotraplinkage void
do_spurious_interrupt_bug(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
579
{
580
	conditional_sti(regs);
L
Linus Torvalds 已提交
581 582
#if 0
	/* No need to warn about this any longer. */
583
	pr_info("Ignoring P6 Local APIC Spurious Interrupt Bug...\n");
L
Linus Torvalds 已提交
584 585 586
#endif
}

587
asmlinkage void __attribute__((weak)) smp_thermal_interrupt(void)
L
Linus Torvalds 已提交
588 589
{
}
590

591
asmlinkage void __attribute__((weak)) smp_threshold_interrupt(void)
592 593 594
{
}

L
Linus Torvalds 已提交
595
/*
I
Ingo Molnar 已提交
596
 * 'math_state_restore()' saves the current math information in the
L
Linus Torvalds 已提交
597 598 599 600 601
 * old math state array, and gets the new ones from the current task
 *
 * Careful.. There are problems with IBM-designed IRQ13 behaviour.
 * Don't touch unless you *really* know how it works.
 *
602 603
 * Must be called with kernel preemption disabled (eg with local
 * local interrupts as in the case of do_device_not_available).
L
Linus Torvalds 已提交
604
 */
605
void math_state_restore(void)
L
Linus Torvalds 已提交
606
{
607
	struct task_struct *tsk = current;
L
Linus Torvalds 已提交
608

609 610 611 612 613 614 615 616 617 618 619 620 621 622 623
	if (!tsk_used_math(tsk)) {
		local_irq_enable();
		/*
		 * does a slab alloc which can sleep
		 */
		if (init_fpu(tsk)) {
			/*
			 * ran out of memory!
			 */
			do_group_exit(SIGKILL);
			return;
		}
		local_irq_disable();
	}

624
	__thread_fpu_begin(tsk);
625

626 627 628 629
	/*
	 * Paranoid restore. send a SIGSEGV if we fail to restore the state.
	 */
	if (unlikely(restore_fpu_checking(tsk))) {
630
		drop_init_fpu(tsk);
631 632 633
		force_sig(SIGSEGV, tsk);
		return;
	}
634 635

	tsk->fpu_counter++;
L
Linus Torvalds 已提交
636
}
637
EXPORT_SYMBOL_GPL(math_state_restore);
L
Linus Torvalds 已提交
638

639
dotraplinkage void __kprobes
640
do_device_not_available(struct pt_regs *regs, long error_code)
641
{
642
	exception_enter(regs);
643
	BUG_ON(use_eager_fpu());
644

645
#ifdef CONFIG_MATH_EMULATION
646
	if (read_cr0() & X86_CR0_EM) {
647 648
		struct math_emu_info info = { };

649
		conditional_sti(regs);
650

651
		info.regs = regs;
652
		math_emulate(&info);
653
		exception_exit(regs);
654
		return;
655
	}
656 657 658 659
#endif
	math_state_restore(); /* interrupts still off */
#ifdef CONFIG_X86_32
	conditional_sti(regs);
660
#endif
661
	exception_exit(regs);
662 663
}

664
#ifdef CONFIG_X86_32
665
dotraplinkage void do_iret_error(struct pt_regs *regs, long error_code)
666 667
{
	siginfo_t info;
668 669

	exception_enter(regs);
670 671 672 673 674
	local_irq_enable();

	info.si_signo = SIGILL;
	info.si_errno = 0;
	info.si_code = ILL_BADSTK;
675
	info.si_addr = NULL;
676
	if (notify_die(DIE_TRAP, "iret exception", regs, error_code,
677 678 679 680 681
			X86_TRAP_IRET, SIGILL) != NOTIFY_STOP) {
		do_trap(X86_TRAP_IRET, SIGILL, "iret exception", regs, error_code,
			&info);
	}
	exception_exit(regs);
682
}
683
#endif
684

685 686 687
/* Set of traps needed for early debugging. */
void __init early_trap_init(void)
{
688
	set_intr_gate_ist(X86_TRAP_DB, &debug, DEBUG_STACK);
689
	/* int3 can be called from all */
690
	set_system_intr_gate_ist(X86_TRAP_BP, &int3, DEBUG_STACK);
691
#ifdef CONFIG_X86_32
692
	set_intr_gate(X86_TRAP_PF, &page_fault);
693
#endif
694 695 696
	load_idt(&idt_descr);
}

697 698 699 700 701 702 703
void __init early_trap_pf_init(void)
{
#ifdef CONFIG_X86_64
	set_intr_gate(X86_TRAP_PF, &page_fault);
#endif
}

L
Linus Torvalds 已提交
704 705
void __init trap_init(void)
{
706 707
	int i;

L
Linus Torvalds 已提交
708
#ifdef CONFIG_EISA
I
Ingo Molnar 已提交
709
	void __iomem *p = early_ioremap(0x0FFFD9, 4);
I
Ingo Molnar 已提交
710 711

	if (readl(p) == 'E' + ('I'<<8) + ('S'<<16) + ('A'<<24))
L
Linus Torvalds 已提交
712
		EISA_bus = 1;
I
Ingo Molnar 已提交
713
	early_iounmap(p, 4);
L
Linus Torvalds 已提交
714 715
#endif

716 717
	set_intr_gate(X86_TRAP_DE, &divide_error);
	set_intr_gate_ist(X86_TRAP_NMI, &nmi, NMI_STACK);
718
	/* int4 can be called from all */
719 720 721 722
	set_system_intr_gate(X86_TRAP_OF, &overflow);
	set_intr_gate(X86_TRAP_BR, &bounds);
	set_intr_gate(X86_TRAP_UD, &invalid_op);
	set_intr_gate(X86_TRAP_NM, &device_not_available);
723
#ifdef CONFIG_X86_32
724
	set_task_gate(X86_TRAP_DF, GDT_ENTRY_DOUBLEFAULT_TSS);
725
#else
726
	set_intr_gate_ist(X86_TRAP_DF, &double_fault, DOUBLEFAULT_STACK);
727
#endif
728 729 730 731 732 733 734 735
	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);
	set_intr_gate_ist(X86_TRAP_SS, &stack_segment, STACKFAULT_STACK);
	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
736
#ifdef CONFIG_X86_MCE
737
	set_intr_gate_ist(X86_TRAP_MC, &machine_check, MCE_STACK);
L
Linus Torvalds 已提交
738
#endif
739
	set_intr_gate(X86_TRAP_XF, &simd_coprocessor_error);
L
Linus Torvalds 已提交
740

741 742 743 744
	/* Reserve all the builtin and the syscall vector: */
	for (i = 0; i < FIRST_EXTERNAL_VECTOR; i++)
		set_bit(i, used_vectors);

745 746
#ifdef CONFIG_IA32_EMULATION
	set_system_intr_gate(IA32_SYSCALL_VECTOR, ia32_syscall);
747
	set_bit(IA32_SYSCALL_VECTOR, used_vectors);
748 749 750
#endif

#ifdef CONFIG_X86_32
751
	set_system_trap_gate(SYSCALL_VECTOR, &system_call);
752
	set_bit(SYSCALL_VECTOR, used_vectors);
753
#endif
754

L
Linus Torvalds 已提交
755
	/*
I
Ingo Molnar 已提交
756
	 * Should be a barrier for any external CPU state:
L
Linus Torvalds 已提交
757 758 759
	 */
	cpu_init();

760
	x86_init.irqs.trap_init();
761 762 763

#ifdef CONFIG_X86_64
	memcpy(&nmi_idt_table, &idt_table, IDT_ENTRIES * 16);
764 765
	set_nmi_gate(X86_TRAP_DB, &debug);
	set_nmi_gate(X86_TRAP_BP, &int3);
766
#endif
L
Linus Torvalds 已提交
767
}