traps.c 19.4 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

59
#include <asm/mach_traps.h>
60

61
#ifdef CONFIG_X86_64
62
#include <asm/x86_init.h>
63 64 65
#include <asm/pgalloc.h>
#include <asm/proto.h>
#else
66
#include <asm/processor-flags.h>
I
Ingo Molnar 已提交
67
#include <asm/setup.h>
L
Linus Torvalds 已提交
68 69 70 71

asmlinkage int system_call(void);

/* Do we ignore FPU interrupts ? */
I
Ingo Molnar 已提交
72
char ignore_fpu_irq;
L
Linus Torvalds 已提交
73 74 75

/*
 * The IDT has to be page-aligned to simplify the Pentium
76
 * F0 0F bug workaround.
L
Linus Torvalds 已提交
77
 */
78
gate_desc idt_table[NR_VECTORS] __page_aligned_data = { { { { 0, 0 } } }, };
79
#endif
L
Linus Torvalds 已提交
80

81 82 83
DECLARE_BITMAP(used_vectors, NR_VECTORS);
EXPORT_SYMBOL_GPL(used_vectors);

84 85 86 87 88 89
static inline void conditional_sti(struct pt_regs *regs)
{
	if (regs->flags & X86_EFLAGS_IF)
		local_irq_enable();
}

90 91 92 93 94 95 96
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
97 98 99 100 101 102
static inline void conditional_cli(struct pt_regs *regs)
{
	if (regs->flags & X86_EFLAGS_IF)
		local_irq_disable();
}

103 104 105 106 107 108 109
static inline void preempt_conditional_cli(struct pt_regs *regs)
{
	if (regs->flags & X86_EFLAGS_IF)
		local_irq_disable();
	dec_preempt_count();
}

I
Ingo Molnar 已提交
110
static void __kprobes
111
do_trap(int trapnr, int signr, char *str, struct pt_regs *regs,
I
Ingo Molnar 已提交
112
	long error_code, siginfo_t *info)
L
Linus Torvalds 已提交
113
{
114 115
	struct task_struct *tsk = current;

116
#ifdef CONFIG_X86_32
117
	if (regs->flags & X86_VM_MASK) {
118 119 120 121
		/*
		 * traps 0, 1, 3, 4, and 5 should be forwarded to vm86.
		 * On nmi (interrupt 2), do_trap should not be called.
		 */
122
		if (trapnr < X86_TRAP_UD)
L
Linus Torvalds 已提交
123 124 125
			goto vm86_trap;
		goto trap_signal;
	}
126
#endif
L
Linus Torvalds 已提交
127

128
	if (!user_mode(regs))
L
Linus Torvalds 已提交
129 130
		goto kernel_trap;

131
#ifdef CONFIG_X86_32
I
Ingo Molnar 已提交
132
trap_signal:
133
#endif
I
Ingo Molnar 已提交
134
	/*
135
	 * We want error_code and trap_nr set for userspace faults and
I
Ingo Molnar 已提交
136 137 138 139 140 141 142 143
	 * 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;
144
	tsk->thread.trap_nr = trapnr;
145

146 147 148
#ifdef CONFIG_X86_64
	if (show_unhandled_signals && unhandled_signal(tsk, signr) &&
	    printk_ratelimit()) {
149 150 151
		pr_info("%s[%d] trap %s ip:%lx sp:%lx error:%lx",
			tsk->comm, tsk->pid, str,
			regs->ip, regs->sp, error_code);
152
		print_vma_addr(" in ", regs->ip);
153
		pr_cont("\n");
154 155 156
	}
#endif

I
Ingo Molnar 已提交
157 158 159 160 161
	if (info)
		force_sig_info(signr, info, tsk);
	else
		force_sig(signr, tsk);
	return;
L
Linus Torvalds 已提交
162

I
Ingo Molnar 已提交
163 164 165
kernel_trap:
	if (!fixup_exception(regs)) {
		tsk->thread.error_code = error_code;
166
		tsk->thread.trap_nr = trapnr;
I
Ingo Molnar 已提交
167
		die(str, regs, error_code);
L
Linus Torvalds 已提交
168
	}
I
Ingo Molnar 已提交
169
	return;
L
Linus Torvalds 已提交
170

171
#ifdef CONFIG_X86_32
I
Ingo Molnar 已提交
172 173 174 175 176
vm86_trap:
	if (handle_vm86_trap((struct kernel_vm86_regs *) regs,
						error_code, trapnr))
		goto trap_signal;
	return;
177
#endif
L
Linus Torvalds 已提交
178 179
}

I
Ingo Molnar 已提交
180
#define DO_ERROR(trapnr, signr, str, name)				\
181
dotraplinkage void do_##name(struct pt_regs *regs, long error_code)	\
I
Ingo Molnar 已提交
182 183
{									\
	if (notify_die(DIE_TRAP, str, regs, error_code, trapnr, signr)	\
184
							== NOTIFY_STOP)	\
I
Ingo Molnar 已提交
185
		return;							\
186
	conditional_sti(regs);						\
187
	do_trap(trapnr, signr, str, regs, error_code, NULL);		\
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
192 193 194 195 196 197 198
{									\
	siginfo_t info;							\
	info.si_signo = signr;						\
	info.si_errno = 0;						\
	info.si_code = sicode;						\
	info.si_addr = (void __user *)siaddr;				\
	if (notify_die(DIE_TRAP, str, regs, error_code, trapnr, signr)	\
199
							== NOTIFY_STOP)	\
I
Ingo Molnar 已提交
200
		return;							\
201
	conditional_sti(regs);						\
202
	do_trap(trapnr, signr, str, regs, error_code, &info);		\
L
Linus Torvalds 已提交
203 204
}

205 206 207 208 209 210 211 212 213 214
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)
215
#ifdef CONFIG_X86_32
216
DO_ERROR(X86_TRAP_SS, SIGBUS, "stack segment", stack_segment)
217
#endif
218 219
DO_ERROR_INFO(X86_TRAP_AC, SIGBUS, "alignment check", alignment_check,
		BUS_ADRALN, 0)
L
Linus Torvalds 已提交
220

221 222 223 224 225
#ifdef CONFIG_X86_64
/* Runs on IST stack */
dotraplinkage void do_stack_segment(struct pt_regs *regs, long error_code)
{
	if (notify_die(DIE_TRAP, "stack segment", regs, error_code,
226
			X86_TRAP_SS, SIGBUS) == NOTIFY_STOP)
227 228
		return;
	preempt_conditional_sti(regs);
229
	do_trap(X86_TRAP_SS, SIGBUS, "stack segment", regs, error_code, NULL);
230 231 232 233 234 235 236 237 238
	preempt_conditional_cli(regs);
}

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

	/* Return not checked because double check cannot be ignored */
239
	notify_die(DIE_TRAP, str, regs, error_code, X86_TRAP_DF, SIGSEGV);
240 241

	tsk->thread.error_code = error_code;
242
	tsk->thread.trap_nr = X86_TRAP_DF;
243

244 245 246 247
	/*
	 * This is always a kernel trap and never fixable (and thus must
	 * never return).
	 */
248 249 250 251 252
	for (;;)
		die(str, regs, error_code);
}
#endif

253
dotraplinkage void __kprobes
254
do_general_protection(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
255
{
256
	struct task_struct *tsk;
I
Ingo Molnar 已提交
257

258 259
	conditional_sti(regs);

260
#ifdef CONFIG_X86_32
261
	if (regs->flags & X86_VM_MASK)
L
Linus Torvalds 已提交
262
		goto gp_in_vm86;
263
#endif
L
Linus Torvalds 已提交
264

265
	tsk = current;
266
	if (!user_mode(regs))
L
Linus Torvalds 已提交
267 268
		goto gp_in_kernel;

269
	tsk->thread.error_code = error_code;
270
	tsk->thread.trap_nr = X86_TRAP_GP;
I
Ingo Molnar 已提交
271

272 273
	if (show_unhandled_signals && unhandled_signal(tsk, SIGSEGV) &&
			printk_ratelimit()) {
274
		pr_info("%s[%d] general protection ip:%lx sp:%lx error:%lx",
275 276
			tsk->comm, task_pid_nr(tsk),
			regs->ip, regs->sp, error_code);
277
		print_vma_addr(" in ", regs->ip);
278
		pr_cont("\n");
279
	}
280

281
	force_sig(SIGSEGV, tsk);
L
Linus Torvalds 已提交
282 283
	return;

284
#ifdef CONFIG_X86_32
L
Linus Torvalds 已提交
285 286 287 288
gp_in_vm86:
	local_irq_enable();
	handle_vm86_fault((struct kernel_vm86_regs *) regs, error_code);
	return;
289
#endif
L
Linus Torvalds 已提交
290 291

gp_in_kernel:
292 293 294 295
	if (fixup_exception(regs))
		return;

	tsk->thread.error_code = error_code;
296
	tsk->thread.trap_nr = X86_TRAP_GP;
297 298
	if (notify_die(DIE_GPF, "general protection fault", regs, error_code,
			X86_TRAP_GP, SIGSEGV) == NOTIFY_STOP)
299 300
		return;
	die("general protection fault", regs, error_code);
L
Linus Torvalds 已提交
301 302
}

303
/* May run on IST stack. */
304
dotraplinkage void __kprobes notrace do_int3(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
305
{
306
#ifdef CONFIG_DYNAMIC_FTRACE
307 308 309 310 311 312
	/*
	 * 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))
313 314
		return;
#endif
315
#ifdef CONFIG_KGDB_LOW_LEVEL_TRAP
316 317
	if (kgdb_ll_trap(DIE_INT3, "int3", regs, error_code, X86_TRAP_BP,
				SIGTRAP) == NOTIFY_STOP)
318 319
		return;
#endif /* CONFIG_KGDB_LOW_LEVEL_TRAP */
320

321 322
	if (notify_die(DIE_INT3, "int3", regs, error_code, X86_TRAP_BP,
			SIGTRAP) == NOTIFY_STOP)
323
		return;
I
Ingo Molnar 已提交
324

325 326 327 328 329
	/*
	 * Let others (NMI) know that the debug stack is in use
	 * as we may switch to the interrupt stack.
	 */
	debug_stack_usage_inc();
330
	preempt_conditional_sti(regs);
331
	do_trap(X86_TRAP_BP, SIGTRAP, "int3", regs, error_code, NULL);
332
	preempt_conditional_cli(regs);
333
	debug_stack_usage_dec();
L
Linus Torvalds 已提交
334 335
}

336
#ifdef CONFIG_X86_64
337 338 339 340 341
/*
 * 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
 */
342 343 344 345 346 347 348 349 350
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);
351 352 353 354
	/*
	 * Exception from kernel and interrupts are enabled. Move to
	 * kernel process stack.
	 */
355 356 357 358 359 360 361 362
	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
363 364 365 366 367 368 369 370 371 372
/*
 * 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
373
 *
L
Linus Torvalds 已提交
374 375 376 377 378 379 380 381 382 383
 * 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)
384 385
 *
 * May run on IST stack.
L
Linus Torvalds 已提交
386
 */
387
dotraplinkage void __kprobes do_debug(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
388 389
{
	struct task_struct *tsk = current;
390
	int user_icebp = 0;
391
	unsigned long dr6;
392
	int si_code;
L
Linus Torvalds 已提交
393

394
	get_debugreg(dr6, 6);
L
Linus Torvalds 已提交
395

396 397 398
	/* Filter out all the reserved bits which are preset to 1 */
	dr6 &= ~DR6_RESERVED;

399 400 401 402 403 404 405 406
	/*
	 * 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
407
	/* Catch kmemcheck conditions first of all! */
408
	if ((dr6 & DR_STEP) && kmemcheck_trap(regs))
V
Vegard Nossum 已提交
409 410
		return;

411 412
	/* DR6 may or may not be cleared by the CPU */
	set_debugreg(0, 6);
413

P
Peter Zijlstra 已提交
414 415 416 417 418
	/*
	 * The processor cleared BTF, so don't mark that we need it set.
	 */
	clear_tsk_thread_flag(tsk, TIF_BLOCKSTEP);

419 420 421
	/* Store the virtualized DR6 value */
	tsk->thread.debugreg6 = dr6;

422 423
	if (notify_die(DIE_DEBUG, "debug", regs, PTR_ERR(&dr6), error_code,
							SIGTRAP) == NOTIFY_STOP)
L
Linus Torvalds 已提交
424
		return;
425

426 427 428 429 430 431
	/*
	 * 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
432
	/* It's safe to allow irq's after DR6 has been saved */
433
	preempt_conditional_sti(regs);
L
Linus Torvalds 已提交
434

435
	if (regs->flags & X86_VM_MASK) {
436 437
		handle_vm86_trap((struct kernel_vm86_regs *) regs, error_code,
					X86_TRAP_DB);
438
		preempt_conditional_cli(regs);
439
		debug_stack_usage_dec();
440
		return;
L
Linus Torvalds 已提交
441 442 443
	}

	/*
444 445 446 447 448
	 * 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
449
	 */
450 451 452 453
	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
454
	}
455
	si_code = get_si_code(tsk->thread.debugreg6);
456
	if (tsk->thread.debugreg6 & (DR_STEP | DR_TRAP_BITS) || user_icebp)
457
		send_sigtrap(tsk, regs, error_code, si_code);
458
	preempt_conditional_cli(regs);
459
	debug_stack_usage_dec();
L
Linus Torvalds 已提交
460 461 462 463 464 465 466 467 468

	return;
}

/*
 * 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
 */
469
void math_error(struct pt_regs *regs, int error_code, int trapnr)
L
Linus Torvalds 已提交
470
{
471
	struct task_struct *task = current;
L
Linus Torvalds 已提交
472
	siginfo_t info;
473
	unsigned short err;
474 475
	char *str = (trapnr == X86_TRAP_MF) ? "fpu exception" :
						"simd exception";
476 477 478 479 480 481 482 483 484

	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;
485
			task->thread.trap_nr = trapnr;
486 487 488 489
			die(str, regs, error_code);
		}
		return;
	}
L
Linus Torvalds 已提交
490 491 492 493 494

	/*
	 * Save the info for the exception handler and clear the error.
	 */
	save_init_fpu(task);
495
	task->thread.trap_nr = trapnr;
496
	task->thread.error_code = error_code;
L
Linus Torvalds 已提交
497 498
	info.si_signo = SIGFPE;
	info.si_errno = 0;
499
	info.si_addr = (void __user *)regs->ip;
500
	if (trapnr == X86_TRAP_MF) {
501 502 503 504 505 506 507 508 509 510 511 512 513
		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);
514

515 516 517 518 519 520 521 522 523 524 525
		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;
	}
526 527

	if (err & 0x001) {	/* Invalid op */
I
Ingo Molnar 已提交
528 529 530 531 532 533
		/*
		 * swd & 0x240 == 0x040: Stack Underflow
		 * swd & 0x240 == 0x240: Stack Overflow
		 * User must clear the SF bit (0x40) if set
		 */
		info.si_code = FPE_FLTINV;
534
	} else if (err & 0x004) { /* Divide by Zero */
I
Ingo Molnar 已提交
535
		info.si_code = FPE_FLTDIV;
536
	} else if (err & 0x008) { /* Overflow */
I
Ingo Molnar 已提交
537
		info.si_code = FPE_FLTOVF;
538 539 540
	} else if (err & 0x012) { /* Denormal, Underflow */
		info.si_code = FPE_FLTUND;
	} else if (err & 0x020) { /* Precision */
I
Ingo Molnar 已提交
541
		info.si_code = FPE_FLTRES;
542
	} else {
543
		/*
544 545 546
		 * 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.
547
		 */
548
		return;
L
Linus Torvalds 已提交
549 550 551 552
	}
	force_sig_info(SIGFPE, &info, task);
}

553
dotraplinkage void do_coprocessor_error(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
554
{
555
#ifdef CONFIG_X86_32
L
Linus Torvalds 已提交
556
	ignore_fpu_irq = 1;
557 558
#endif

559
	math_error(regs, error_code, X86_TRAP_MF);
L
Linus Torvalds 已提交
560 561
}

562 563
dotraplinkage void
do_simd_coprocessor_error(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
564
{
565
	math_error(regs, error_code, X86_TRAP_XF);
L
Linus Torvalds 已提交
566 567
}

568 569
dotraplinkage void
do_spurious_interrupt_bug(struct pt_regs *regs, long error_code)
L
Linus Torvalds 已提交
570
{
571
	conditional_sti(regs);
L
Linus Torvalds 已提交
572 573
#if 0
	/* No need to warn about this any longer. */
574
	pr_info("Ignoring P6 Local APIC Spurious Interrupt Bug...\n");
L
Linus Torvalds 已提交
575 576 577
#endif
}

578
asmlinkage void __attribute__((weak)) smp_thermal_interrupt(void)
L
Linus Torvalds 已提交
579 580
{
}
581

582
asmlinkage void __attribute__((weak)) smp_threshold_interrupt(void)
583 584 585
{
}

L
Linus Torvalds 已提交
586
/*
I
Ingo Molnar 已提交
587
 * 'math_state_restore()' saves the current math information in the
L
Linus Torvalds 已提交
588 589 590 591 592
 * 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.
 *
593 594
 * 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
595
 */
596
void math_state_restore(void)
L
Linus Torvalds 已提交
597
{
598
	struct task_struct *tsk = current;
L
Linus Torvalds 已提交
599

600 601 602 603 604 605 606 607 608 609 610 611 612 613 614
	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();
	}

615
	__thread_fpu_begin(tsk);
616

617 618 619 620
	/*
	 * Paranoid restore. send a SIGSEGV if we fail to restore the state.
	 */
	if (unlikely(restore_fpu_checking(tsk))) {
621
		drop_init_fpu(tsk);
622 623 624
		force_sig(SIGSEGV, tsk);
		return;
	}
625 626

	tsk->fpu_counter++;
L
Linus Torvalds 已提交
627
}
628
EXPORT_SYMBOL_GPL(math_state_restore);
L
Linus Torvalds 已提交
629

630
dotraplinkage void __kprobes
631
do_device_not_available(struct pt_regs *regs, long error_code)
632
{
633
	BUG_ON(use_eager_fpu());
634

635
#ifdef CONFIG_MATH_EMULATION
636
	if (read_cr0() & X86_CR0_EM) {
637 638
		struct math_emu_info info = { };

639
		conditional_sti(regs);
640

641
		info.regs = regs;
642
		math_emulate(&info);
643
		return;
644
	}
645 646 647 648
#endif
	math_state_restore(); /* interrupts still off */
#ifdef CONFIG_X86_32
	conditional_sti(regs);
649
#endif
650 651
}

652
#ifdef CONFIG_X86_32
653
dotraplinkage void do_iret_error(struct pt_regs *regs, long error_code)
654 655 656 657 658 659 660
{
	siginfo_t info;
	local_irq_enable();

	info.si_signo = SIGILL;
	info.si_errno = 0;
	info.si_code = ILL_BADSTK;
661
	info.si_addr = NULL;
662 663
	if (notify_die(DIE_TRAP, "iret exception", regs, error_code,
			X86_TRAP_IRET, SIGILL) == NOTIFY_STOP)
664
		return;
665 666
	do_trap(X86_TRAP_IRET, SIGILL, "iret exception", regs, error_code,
		&info);
667
}
668
#endif
669

670 671 672
/* Set of traps needed for early debugging. */
void __init early_trap_init(void)
{
673
	set_intr_gate_ist(X86_TRAP_DB, &debug, DEBUG_STACK);
674
	/* int3 can be called from all */
675 676
	set_system_intr_gate_ist(X86_TRAP_BP, &int3, DEBUG_STACK);
	set_intr_gate(X86_TRAP_PF, &page_fault);
677 678 679
	load_idt(&idt_descr);
}

L
Linus Torvalds 已提交
680 681
void __init trap_init(void)
{
682 683
	int i;

L
Linus Torvalds 已提交
684
#ifdef CONFIG_EISA
I
Ingo Molnar 已提交
685
	void __iomem *p = early_ioremap(0x0FFFD9, 4);
I
Ingo Molnar 已提交
686 687

	if (readl(p) == 'E' + ('I'<<8) + ('S'<<16) + ('A'<<24))
L
Linus Torvalds 已提交
688
		EISA_bus = 1;
I
Ingo Molnar 已提交
689
	early_iounmap(p, 4);
L
Linus Torvalds 已提交
690 691
#endif

692 693
	set_intr_gate(X86_TRAP_DE, &divide_error);
	set_intr_gate_ist(X86_TRAP_NMI, &nmi, NMI_STACK);
694
	/* int4 can be called from all */
695 696 697 698
	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);
699
#ifdef CONFIG_X86_32
700
	set_task_gate(X86_TRAP_DF, GDT_ENTRY_DOUBLEFAULT_TSS);
701
#else
702
	set_intr_gate_ist(X86_TRAP_DF, &double_fault, DOUBLEFAULT_STACK);
703
#endif
704 705 706 707 708 709 710 711
	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
712
#ifdef CONFIG_X86_MCE
713
	set_intr_gate_ist(X86_TRAP_MC, &machine_check, MCE_STACK);
L
Linus Torvalds 已提交
714
#endif
715
	set_intr_gate(X86_TRAP_XF, &simd_coprocessor_error);
L
Linus Torvalds 已提交
716

717 718 719 720
	/* Reserve all the builtin and the syscall vector: */
	for (i = 0; i < FIRST_EXTERNAL_VECTOR; i++)
		set_bit(i, used_vectors);

721 722
#ifdef CONFIG_IA32_EMULATION
	set_system_intr_gate(IA32_SYSCALL_VECTOR, ia32_syscall);
723
	set_bit(IA32_SYSCALL_VECTOR, used_vectors);
724 725 726
#endif

#ifdef CONFIG_X86_32
727
	set_system_trap_gate(SYSCALL_VECTOR, &system_call);
728
	set_bit(SYSCALL_VECTOR, used_vectors);
729
#endif
730

L
Linus Torvalds 已提交
731
	/*
I
Ingo Molnar 已提交
732
	 * Should be a barrier for any external CPU state:
L
Linus Torvalds 已提交
733 734 735
	 */
	cpu_init();

736
	x86_init.irqs.trap_init();
737 738 739

#ifdef CONFIG_X86_64
	memcpy(&nmi_idt_table, &idt_table, IDT_ENTRIES * 16);
740 741
	set_nmi_gate(X86_TRAP_DB, &debug);
	set_nmi_gate(X86_TRAP_BP, &int3);
742
#endif
L
Linus Torvalds 已提交
743
}