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/rcu.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

asmlinkage int system_call(void);

/* Do we ignore FPU interrupts ? */
I
Ingo Molnar 已提交
73
char ignore_fpu_irq;
L
Linus Torvalds 已提交
74 75 76

/*
 * The IDT has to be page-aligned to simplify the Pentium
77
 * F0 0F bug workaround.
L
Linus Torvalds 已提交
78
 */
79
gate_desc idt_table[NR_VECTORS] __page_aligned_data = { { { { 0, 0 } } }, };
80
#endif
L
Linus Torvalds 已提交
81

82 83 84
DECLARE_BITMAP(used_vectors, NR_VECTORS);
EXPORT_SYMBOL_GPL(used_vectors);

85 86 87 88 89 90
static inline void conditional_sti(struct pt_regs *regs)
{
	if (regs->flags & X86_EFLAGS_IF)
		local_irq_enable();
}

91 92 93 94 95 96 97
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
98 99 100 101 102 103
static inline void conditional_cli(struct pt_regs *regs)
{
	if (regs->flags & X86_EFLAGS_IF)
		local_irq_disable();
}

104 105 106 107 108 109 110
static inline void preempt_conditional_cli(struct pt_regs *regs)
{
	if (regs->flags & X86_EFLAGS_IF)
		local_irq_disable();
	dec_preempt_count();
}

111 112 113
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
114
{
115
#ifdef CONFIG_X86_32
116
	if (regs->flags & X86_VM_MASK) {
117
		/*
118
		 * Traps 0, 1, 3, 4, and 5 should be forwarded to vm86.
119 120
		 * On nmi (interrupt 2), do_trap should not be called.
		 */
121 122 123 124 125 126
		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
127
	}
128
#endif
129 130 131 132 133 134 135 136
	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
137

138 139
	return -1;
}
L
Linus Torvalds 已提交
140

141 142 143 144 145 146 147 148 149
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
150
	/*
151
	 * We want error_code and trap_nr set for userspace faults and
I
Ingo Molnar 已提交
152 153 154 155 156 157 158 159
	 * 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;
160
	tsk->thread.trap_nr = trapnr;
161

162 163 164
#ifdef CONFIG_X86_64
	if (show_unhandled_signals && unhandled_signal(tsk, signr) &&
	    printk_ratelimit()) {
165 166 167
		pr_info("%s[%d] trap %s ip:%lx sp:%lx error:%lx",
			tsk->comm, tsk->pid, str,
			regs->ip, regs->sp, error_code);
168
		print_vma_addr(" in ", regs->ip);
169
		pr_cont("\n");
170 171 172
	}
#endif

I
Ingo Molnar 已提交
173 174 175 176
	if (info)
		force_sig_info(signr, info, tsk);
	else
		force_sig(signr, tsk);
L
Linus Torvalds 已提交
177 178
}

I
Ingo Molnar 已提交
179
#define DO_ERROR(trapnr, signr, str, name)				\
180
dotraplinkage void do_##name(struct pt_regs *regs, long error_code)	\
I
Ingo Molnar 已提交
181
{									\
182 183 184 185
	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
186
		return;							\
187
	}								\
188
	conditional_sti(regs);						\
189
	do_trap(trapnr, signr, str, regs, error_code, NULL);		\
190
	exception_exit(regs);						\
L
Linus Torvalds 已提交
191 192
}

193
#define DO_ERROR_INFO(trapnr, signr, str, name, sicode, siaddr)		\
194
dotraplinkage void do_##name(struct pt_regs *regs, long error_code)	\
I
Ingo Molnar 已提交
195 196 197 198 199 200
{									\
	siginfo_t info;							\
	info.si_signo = signr;						\
	info.si_errno = 0;						\
	info.si_code = sicode;						\
	info.si_addr = (void __user *)siaddr;				\
201 202 203 204
	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
205
		return;							\
206
	}								\
207
	conditional_sti(regs);						\
208
	do_trap(trapnr, signr, str, regs, error_code, &info);		\
209
	exception_exit(regs);						\
L
Linus Torvalds 已提交
210 211
}

212 213 214 215 216 217 218 219 220 221
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)
222
#ifdef CONFIG_X86_32
223
DO_ERROR(X86_TRAP_SS, SIGBUS, "stack segment", stack_segment)
224
#endif
225 226
DO_ERROR_INFO(X86_TRAP_AC, SIGBUS, "alignment check", alignment_check,
		BUS_ADRALN, 0)
L
Linus Torvalds 已提交
227

228 229 230 231
#ifdef CONFIG_X86_64
/* Runs on IST stack */
dotraplinkage void do_stack_segment(struct pt_regs *regs, long error_code)
{
232
	exception_enter(regs);
233
	if (notify_die(DIE_TRAP, "stack segment", regs, error_code,
234 235 236 237 238 239
		       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);
240 241 242 243 244 245 246
}

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

247
	exception_enter(regs);
248
	/* Return not checked because double check cannot be ignored */
249
	notify_die(DIE_TRAP, str, regs, error_code, X86_TRAP_DF, SIGSEGV);
250 251

	tsk->thread.error_code = error_code;
252
	tsk->thread.trap_nr = X86_TRAP_DF;
253

254 255 256 257
	/*
	 * This is always a kernel trap and never fixable (and thus must
	 * never return).
	 */
258 259 260 261 262
	for (;;)
		die(str, regs, error_code);
}
#endif

263
dotraplinkage void __kprobes
264
do_general_protection(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
265
{
266
	struct task_struct *tsk;
I
Ingo Molnar 已提交
267

268
	exception_enter(regs);
269 270
	conditional_sti(regs);

271
#ifdef CONFIG_X86_32
272 273 274
	if (regs->flags & X86_VM_MASK) {
		local_irq_enable();
		handle_vm86_fault((struct kernel_vm86_regs *) regs, error_code);
275
		goto exit;
276
	}
277
#endif
L
Linus Torvalds 已提交
278

279
	tsk = current;
280 281
	if (!user_mode(regs)) {
		if (fixup_exception(regs))
282
			goto exit;
283 284 285

		tsk->thread.error_code = error_code;
		tsk->thread.trap_nr = X86_TRAP_GP;
286 287
		if (notify_die(DIE_GPF, "general protection fault", regs, error_code,
			       X86_TRAP_GP, SIGSEGV) != NOTIFY_STOP)
288
			die("general protection fault", regs, error_code);
289
		goto exit;
290
	}
L
Linus Torvalds 已提交
291

292
	tsk->thread.error_code = error_code;
293
	tsk->thread.trap_nr = X86_TRAP_GP;
I
Ingo Molnar 已提交
294

295 296
	if (show_unhandled_signals && unhandled_signal(tsk, SIGSEGV) &&
			printk_ratelimit()) {
297
		pr_info("%s[%d] general protection ip:%lx sp:%lx error:%lx",
298 299
			tsk->comm, task_pid_nr(tsk),
			regs->ip, regs->sp, error_code);
300
		print_vma_addr(" in ", regs->ip);
301
		pr_cont("\n");
302
	}
303

304
	force_sig(SIGSEGV, tsk);
305 306
exit:
	exception_exit(regs);
L
Linus Torvalds 已提交
307 308
}

309
/* May run on IST stack. */
310
dotraplinkage void __kprobes notrace do_int3(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
311
{
312
#ifdef CONFIG_DYNAMIC_FTRACE
313 314 315 316 317 318
	/*
	 * 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))
319 320
		return;
#endif
321
	exception_enter(regs);
322
#ifdef CONFIG_KGDB_LOW_LEVEL_TRAP
323 324
	if (kgdb_ll_trap(DIE_INT3, "int3", regs, error_code, X86_TRAP_BP,
				SIGTRAP) == NOTIFY_STOP)
325
		goto exit;
326
#endif /* CONFIG_KGDB_LOW_LEVEL_TRAP */
327

328 329
	if (notify_die(DIE_INT3, "int3", regs, error_code, X86_TRAP_BP,
			SIGTRAP) == NOTIFY_STOP)
330
		goto exit;
I
Ingo Molnar 已提交
331

332 333 334 335 336
	/*
	 * Let others (NMI) know that the debug stack is in use
	 * as we may switch to the interrupt stack.
	 */
	debug_stack_usage_inc();
337
	preempt_conditional_sti(regs);
338
	do_trap(X86_TRAP_BP, SIGTRAP, "int3", regs, error_code, NULL);
339
	preempt_conditional_cli(regs);
340
	debug_stack_usage_dec();
341 342
exit:
	exception_exit(regs);
L
Linus Torvalds 已提交
343 344
}

345
#ifdef CONFIG_X86_64
346 347 348 349 350
/*
 * 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
 */
351 352 353 354 355 356 357 358 359
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);
360 361 362 363
	/*
	 * Exception from kernel and interrupts are enabled. Move to
	 * kernel process stack.
	 */
364 365 366 367 368 369 370 371
	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
372 373 374 375 376 377 378 379 380 381
/*
 * 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
382
 *
L
Linus Torvalds 已提交
383 384 385 386 387 388 389 390 391 392
 * 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)
393 394
 *
 * May run on IST stack.
L
Linus Torvalds 已提交
395
 */
396
dotraplinkage void __kprobes do_debug(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
397 398
{
	struct task_struct *tsk = current;
399
	int user_icebp = 0;
400
	unsigned long dr6;
401
	int si_code;
L
Linus Torvalds 已提交
402

403 404
	exception_enter(regs);

405
	get_debugreg(dr6, 6);
L
Linus Torvalds 已提交
406

407 408 409
	/* Filter out all the reserved bits which are preset to 1 */
	dr6 &= ~DR6_RESERVED;

410 411 412 413 414 415 416 417
	/*
	 * 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
418
	/* Catch kmemcheck conditions first of all! */
419
	if ((dr6 & DR_STEP) && kmemcheck_trap(regs))
420
		goto exit;
V
Vegard Nossum 已提交
421

422 423
	/* DR6 may or may not be cleared by the CPU */
	set_debugreg(0, 6);
424

P
Peter Zijlstra 已提交
425 426 427 428 429
	/*
	 * The processor cleared BTF, so don't mark that we need it set.
	 */
	clear_tsk_thread_flag(tsk, TIF_BLOCKSTEP);

430 431 432
	/* Store the virtualized DR6 value */
	tsk->thread.debugreg6 = dr6;

433 434
	if (notify_die(DIE_DEBUG, "debug", regs, PTR_ERR(&dr6), error_code,
							SIGTRAP) == NOTIFY_STOP)
435
		goto exit;
436

437 438 439 440 441 442
	/*
	 * 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
443
	/* It's safe to allow irq's after DR6 has been saved */
444
	preempt_conditional_sti(regs);
L
Linus Torvalds 已提交
445

446
	if (regs->flags & X86_VM_MASK) {
447 448
		handle_vm86_trap((struct kernel_vm86_regs *) regs, error_code,
					X86_TRAP_DB);
449
		preempt_conditional_cli(regs);
450
		debug_stack_usage_dec();
451
		goto exit;
L
Linus Torvalds 已提交
452 453 454
	}

	/*
455 456 457 458 459
	 * 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
460
	 */
461 462 463 464
	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
465
	}
466
	si_code = get_si_code(tsk->thread.debugreg6);
467
	if (tsk->thread.debugreg6 & (DR_STEP | DR_TRAP_BITS) || user_icebp)
468
		send_sigtrap(tsk, regs, error_code, si_code);
469
	preempt_conditional_cli(regs);
470
	debug_stack_usage_dec();
L
Linus Torvalds 已提交
471

472 473
exit:
	exception_exit(regs);
L
Linus Torvalds 已提交
474 475 476 477 478 479 480
}

/*
 * 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
 */
481
void math_error(struct pt_regs *regs, int error_code, int trapnr)
L
Linus Torvalds 已提交
482
{
483
	struct task_struct *task = current;
L
Linus Torvalds 已提交
484
	siginfo_t info;
485
	unsigned short err;
486 487
	char *str = (trapnr == X86_TRAP_MF) ? "fpu exception" :
						"simd exception";
488 489 490 491 492 493 494 495 496

	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;
497
			task->thread.trap_nr = trapnr;
498 499 500 501
			die(str, regs, error_code);
		}
		return;
	}
L
Linus Torvalds 已提交
502 503 504 505 506

	/*
	 * Save the info for the exception handler and clear the error.
	 */
	save_init_fpu(task);
507
	task->thread.trap_nr = trapnr;
508
	task->thread.error_code = error_code;
L
Linus Torvalds 已提交
509 510
	info.si_signo = SIGFPE;
	info.si_errno = 0;
511
	info.si_addr = (void __user *)regs->ip;
512
	if (trapnr == X86_TRAP_MF) {
513 514 515 516 517 518 519 520 521 522 523 524 525
		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);
526

527 528 529 530 531 532 533 534 535 536 537
		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;
	}
538 539

	if (err & 0x001) {	/* Invalid op */
I
Ingo Molnar 已提交
540 541 542 543 544 545
		/*
		 * swd & 0x240 == 0x040: Stack Underflow
		 * swd & 0x240 == 0x240: Stack Overflow
		 * User must clear the SF bit (0x40) if set
		 */
		info.si_code = FPE_FLTINV;
546
	} else if (err & 0x004) { /* Divide by Zero */
I
Ingo Molnar 已提交
547
		info.si_code = FPE_FLTDIV;
548
	} else if (err & 0x008) { /* Overflow */
I
Ingo Molnar 已提交
549
		info.si_code = FPE_FLTOVF;
550 551 552
	} else if (err & 0x012) { /* Denormal, Underflow */
		info.si_code = FPE_FLTUND;
	} else if (err & 0x020) { /* Precision */
I
Ingo Molnar 已提交
553
		info.si_code = FPE_FLTRES;
554
	} else {
555
		/*
556 557 558
		 * 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.
559
		 */
560
		return;
L
Linus Torvalds 已提交
561 562 563 564
	}
	force_sig_info(SIGFPE, &info, task);
}

565
dotraplinkage void do_coprocessor_error(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
566
{
567
#ifdef CONFIG_X86_32
L
Linus Torvalds 已提交
568
	ignore_fpu_irq = 1;
569
#endif
570
	exception_enter(regs);
571
	math_error(regs, error_code, X86_TRAP_MF);
572
	exception_exit(regs);
L
Linus Torvalds 已提交
573 574
}

575 576
dotraplinkage void
do_simd_coprocessor_error(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
577
{
578
	exception_enter(regs);
579
	math_error(regs, error_code, X86_TRAP_XF);
580
	exception_exit(regs);
L
Linus Torvalds 已提交
581 582
}

583 584
dotraplinkage void
do_spurious_interrupt_bug(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
585
{
586
	conditional_sti(regs);
L
Linus Torvalds 已提交
587 588
#if 0
	/* No need to warn about this any longer. */
589
	pr_info("Ignoring P6 Local APIC Spurious Interrupt Bug...\n");
L
Linus Torvalds 已提交
590 591 592
#endif
}

593
asmlinkage void __attribute__((weak)) smp_thermal_interrupt(void)
L
Linus Torvalds 已提交
594 595
{
}
596

597
asmlinkage void __attribute__((weak)) smp_threshold_interrupt(void)
598 599 600
{
}

L
Linus Torvalds 已提交
601
/*
I
Ingo Molnar 已提交
602
 * 'math_state_restore()' saves the current math information in the
L
Linus Torvalds 已提交
603 604 605 606 607
 * 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.
 *
608 609
 * 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
610
 */
611
void math_state_restore(void)
L
Linus Torvalds 已提交
612
{
613
	struct task_struct *tsk = current;
L
Linus Torvalds 已提交
614

615 616 617 618 619 620 621 622 623 624 625 626 627 628 629
	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();
	}

630
	__thread_fpu_begin(tsk);
631

632 633 634 635
	/*
	 * Paranoid restore. send a SIGSEGV if we fail to restore the state.
	 */
	if (unlikely(restore_fpu_checking(tsk))) {
636
		drop_init_fpu(tsk);
637 638 639
		force_sig(SIGSEGV, tsk);
		return;
	}
640 641

	tsk->fpu_counter++;
L
Linus Torvalds 已提交
642
}
643
EXPORT_SYMBOL_GPL(math_state_restore);
L
Linus Torvalds 已提交
644

645
dotraplinkage void __kprobes
646
do_device_not_available(struct pt_regs *regs, long error_code)
647
{
648
	exception_enter(regs);
649
	BUG_ON(use_eager_fpu());
650

651
#ifdef CONFIG_MATH_EMULATION
652
	if (read_cr0() & X86_CR0_EM) {
653 654
		struct math_emu_info info = { };

655
		conditional_sti(regs);
656

657
		info.regs = regs;
658
		math_emulate(&info);
659
		exception_exit(regs);
660
		return;
661
	}
662 663 664 665
#endif
	math_state_restore(); /* interrupts still off */
#ifdef CONFIG_X86_32
	conditional_sti(regs);
666
#endif
667
	exception_exit(regs);
668 669
}

670
#ifdef CONFIG_X86_32
671
dotraplinkage void do_iret_error(struct pt_regs *regs, long error_code)
672 673
{
	siginfo_t info;
674 675

	exception_enter(regs);
676 677 678 679 680
	local_irq_enable();

	info.si_signo = SIGILL;
	info.si_errno = 0;
	info.si_code = ILL_BADSTK;
681
	info.si_addr = NULL;
682
	if (notify_die(DIE_TRAP, "iret exception", regs, error_code,
683 684 685 686 687
			X86_TRAP_IRET, SIGILL) != NOTIFY_STOP) {
		do_trap(X86_TRAP_IRET, SIGILL, "iret exception", regs, error_code,
			&info);
	}
	exception_exit(regs);
688
}
689
#endif
690

691 692 693
/* Set of traps needed for early debugging. */
void __init early_trap_init(void)
{
694
	set_intr_gate_ist(X86_TRAP_DB, &debug, DEBUG_STACK);
695
	/* int3 can be called from all */
696 697
	set_system_intr_gate_ist(X86_TRAP_BP, &int3, DEBUG_STACK);
	set_intr_gate(X86_TRAP_PF, &page_fault);
698 699 700
	load_idt(&idt_descr);
}

L
Linus Torvalds 已提交
701 702
void __init trap_init(void)
{
703 704
	int i;

L
Linus Torvalds 已提交
705
#ifdef CONFIG_EISA
I
Ingo Molnar 已提交
706
	void __iomem *p = early_ioremap(0x0FFFD9, 4);
I
Ingo Molnar 已提交
707 708

	if (readl(p) == 'E' + ('I'<<8) + ('S'<<16) + ('A'<<24))
L
Linus Torvalds 已提交
709
		EISA_bus = 1;
I
Ingo Molnar 已提交
710
	early_iounmap(p, 4);
L
Linus Torvalds 已提交
711 712
#endif

713 714
	set_intr_gate(X86_TRAP_DE, &divide_error);
	set_intr_gate_ist(X86_TRAP_NMI, &nmi, NMI_STACK);
715
	/* int4 can be called from all */
716 717 718 719
	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);
720
#ifdef CONFIG_X86_32
721
	set_task_gate(X86_TRAP_DF, GDT_ENTRY_DOUBLEFAULT_TSS);
722
#else
723
	set_intr_gate_ist(X86_TRAP_DF, &double_fault, DOUBLEFAULT_STACK);
724
#endif
725 726 727 728 729 730 731 732
	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
733
#ifdef CONFIG_X86_MCE
734
	set_intr_gate_ist(X86_TRAP_MC, &machine_check, MCE_STACK);
L
Linus Torvalds 已提交
735
#endif
736
	set_intr_gate(X86_TRAP_XF, &simd_coprocessor_error);
L
Linus Torvalds 已提交
737

738 739 740 741
	/* Reserve all the builtin and the syscall vector: */
	for (i = 0; i < FIRST_EXTERNAL_VECTOR; i++)
		set_bit(i, used_vectors);

742 743
#ifdef CONFIG_IA32_EMULATION
	set_system_intr_gate(IA32_SYSCALL_VECTOR, ia32_syscall);
744
	set_bit(IA32_SYSCALL_VECTOR, used_vectors);
745 746 747
#endif

#ifdef CONFIG_X86_32
748
	set_system_trap_gate(SYSCALL_VECTOR, &system_call);
749
	set_bit(SYSCALL_VECTOR, used_vectors);
750
#endif
751

L
Linus Torvalds 已提交
752
	/*
I
Ingo Molnar 已提交
753
	 * Should be a barrier for any external CPU state:
L
Linus Torvalds 已提交
754 755 756
	 */
	cpu_init();

757
	x86_init.irqs.trap_init();
758 759 760

#ifdef CONFIG_X86_64
	memcpy(&nmi_idt_table, &idt_table, IDT_ENTRIES * 16);
761 762
	set_nmi_gate(X86_TRAP_DB, &debug);
	set_nmi_gate(X86_TRAP_BP, &int3);
763
#endif
L
Linus Torvalds 已提交
764
}