traps_64.c 29.4 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
/*
 *  Copyright (C) 1991, 1992  Linus Torvalds
 *  Copyright (C) 2000, 2001, 2002 Andi Kleen, SuSE Labs
 *
 *  Pentium III FXSR, SSE support
 *	Gareth Hughes <gareth@valinux.com>, May 2000
 */

/*
 * 'Traps.c' handles hardware traps and faults after we have saved some
 * state in 'entry.S'.
 */
#include <linux/sched.h>
#include <linux/kernel.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/ptrace.h>
#include <linux/timer.h>
#include <linux/mm.h>
#include <linux/init.h>
#include <linux/delay.h>
#include <linux/spinlock.h>
#include <linux/interrupt.h>
24
#include <linux/kallsyms.h>
L
Linus Torvalds 已提交
25 26
#include <linux/module.h>
#include <linux/moduleparam.h>
27
#include <linux/nmi.h>
28
#include <linux/kprobes.h>
29
#include <linux/kexec.h>
30
#include <linux/unwind.h>
31
#include <linux/uaccess.h>
32
#include <linux/bug.h>
33
#include <linux/kdebug.h>
34
#include <linux/utsname.h>
L
Linus Torvalds 已提交
35

D
Dave Jiang 已提交
36 37 38 39
#if defined(CONFIG_EDAC)
#include <linux/edac.h>
#endif

L
Linus Torvalds 已提交
40 41 42 43 44 45 46
#include <asm/system.h>
#include <asm/io.h>
#include <asm/atomic.h>
#include <asm/debugreg.h>
#include <asm/desc.h>
#include <asm/i387.h>
#include <asm/processor.h>
47
#include <asm/unwind.h>
L
Linus Torvalds 已提交
48 49 50 51 52
#include <asm/smp.h>
#include <asm/pgalloc.h>
#include <asm/pda.h>
#include <asm/proto.h>
#include <asm/nmi.h>
53
#include <asm/stacktrace.h>
L
Linus Torvalds 已提交
54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76

asmlinkage void divide_error(void);
asmlinkage void debug(void);
asmlinkage void nmi(void);
asmlinkage void int3(void);
asmlinkage void overflow(void);
asmlinkage void bounds(void);
asmlinkage void invalid_op(void);
asmlinkage void device_not_available(void);
asmlinkage void double_fault(void);
asmlinkage void coprocessor_segment_overrun(void);
asmlinkage void invalid_TSS(void);
asmlinkage void segment_not_present(void);
asmlinkage void stack_segment(void);
asmlinkage void general_protection(void);
asmlinkage void page_fault(void);
asmlinkage void coprocessor_error(void);
asmlinkage void simd_coprocessor_error(void);
asmlinkage void reserved(void);
asmlinkage void alignment_check(void);
asmlinkage void machine_check(void);
asmlinkage void spurious_interrupt_bug(void);

77 78
static unsigned int code_bytes = 64;

L
Linus Torvalds 已提交
79 80
static inline void conditional_sti(struct pt_regs *regs)
{
81
	if (regs->flags & X86_EFLAGS_IF)
L
Linus Torvalds 已提交
82 83 84
		local_irq_enable();
}

85 86 87
static inline void preempt_conditional_sti(struct pt_regs *regs)
{
	preempt_disable();
88
	if (regs->flags & X86_EFLAGS_IF)
89 90 91 92 93
		local_irq_enable();
}

static inline void preempt_conditional_cli(struct pt_regs *regs)
{
94
	if (regs->flags & X86_EFLAGS_IF)
95
		local_irq_disable();
96 97
	/* Make sure to not schedule here because we could be running
	   on an exception stack. */
98 99 100
	preempt_enable_no_resched();
}

101
int kstack_depth_to_print = 12;
L
Linus Torvalds 已提交
102

103
void printk_address(unsigned long address, int reliable)
104
{
105
#ifdef CONFIG_KALLSYMS
L
Linus Torvalds 已提交
106 107 108
	unsigned long offset = 0, symsize;
	const char *symname;
	char *modname;
109
	char *delim = ":";
C
Cyrill Gorcunov 已提交
110
	char namebuf[KSYM_NAME_LEN];
111
	char reliab[4] = "";
L
Linus Torvalds 已提交
112

113 114 115 116 117 118
	symname = kallsyms_lookup(address, &symsize, &offset,
					&modname, namebuf);
	if (!symname) {
		printk(" [<%016lx>]\n", address);
		return;
	}
119 120 121
	if (!reliable)
		strcpy(reliab, "? ");

122
	if (!modname)
123
		modname = delim = "";
124 125
	printk(" [<%016lx>] %s%s%s%s%s+0x%lx/0x%lx\n",
		address, reliab, delim, modname, delim, symname, offset, symsize);
L
Linus Torvalds 已提交
126
#else
127
	printk(" [<%016lx>]\n", address);
L
Linus Torvalds 已提交
128
#endif
129
}
L
Linus Torvalds 已提交
130

131
static unsigned long *in_exception_stack(unsigned cpu, unsigned long stack,
132
					unsigned *usedp, char **idp)
133
{
134
	static char ids[][8] = {
135 136 137 138 139
		[DEBUG_STACK - 1] = "#DB",
		[NMI_STACK - 1] = "NMI",
		[DOUBLEFAULT_STACK - 1] = "#DF",
		[STACKFAULT_STACK - 1] = "#SS",
		[MCE_STACK - 1] = "#MC",
140 141 142
#if DEBUG_STKSZ > EXCEPTION_STKSZ
		[N_EXCEPTION_STACKS ... N_EXCEPTION_STACKS + DEBUG_STKSZ / EXCEPTION_STKSZ - 2] = "#DB[?]"
#endif
143 144
	};
	unsigned k;
L
Linus Torvalds 已提交
145

146 147 148 149
	/*
	 * Iterate over all exception stacks, and figure out whether
	 * 'stack' is in one of them:
	 */
150
	for (k = 0; k < N_EXCEPTION_STACKS; k++) {
151
		unsigned long end = per_cpu(orig_ist, cpu).ist[k];
152 153 154 155
		/*
		 * Is 'stack' above this exception frame's end?
		 * If yes then skip to the next frame.
		 */
156 157
		if (stack >= end)
			continue;
158 159 160 161
		/*
		 * Is 'stack' above this exception frame's start address?
		 * If yes then we found the right frame.
		 */
162
		if (stack >= end - EXCEPTION_STKSZ) {
163 164 165 166 167 168
			/*
			 * Make sure we only iterate through an exception
			 * stack once. If it comes up for the second time
			 * then there's something wrong going on - just
			 * break out and return NULL:
			 */
169 170 171 172 173 174
			if (*usedp & (1U << k))
				break;
			*usedp |= 1U << k;
			*idp = ids[k];
			return (unsigned long *)end;
		}
175 176 177 178 179
		/*
		 * If this is a debug stack, and if it has a larger size than
		 * the usual exception stacks, then 'stack' might still
		 * be within the lower portion of the debug stack:
		 */
180 181 182 183
#if DEBUG_STKSZ > EXCEPTION_STKSZ
		if (k == DEBUG_STACK - 1 && stack >= end - DEBUG_STKSZ) {
			unsigned j = N_EXCEPTION_STACKS - 1;

184 185 186 187 188
			/*
			 * Black magic. A large debug stack is composed of
			 * multiple exception stack entries, which we
			 * iterate through now. Dont look:
			 */
189 190 191 192 193 194 195 196 197 198 199 200
			do {
				++j;
				end -= EXCEPTION_STKSZ;
				ids[j][4] = '1' + (j - N_EXCEPTION_STACKS);
			} while (stack < end - EXCEPTION_STKSZ);
			if (*usedp & (1U << j))
				break;
			*usedp |= 1U << j;
			*idp = ids[j];
			return (unsigned long *)end;
		}
#endif
L
Linus Torvalds 已提交
201 202
	}
	return NULL;
203
}
L
Linus Torvalds 已提交
204

205 206
#define MSG(txt) ops->warning(data, txt)

L
Linus Torvalds 已提交
207
/*
S
Simon Arlott 已提交
208
 * x86-64 can have up to three kernel stacks: 
L
Linus Torvalds 已提交
209 210
 * process stack
 * interrupt stack
211
 * severe exception (double fault, nmi, stack fault, debug, mce) hardware stack
L
Linus Torvalds 已提交
212 213
 */

214 215
static inline int valid_stack_ptr(struct thread_info *tinfo,
			void *p, unsigned int size, void *end)
216
{
J
Jan Engelhardt 已提交
217
	void *t = tinfo;
218 219 220 221 222 223 224 225 226
	if (end) {
		if (p < end && p >= (end-THREAD_SIZE))
			return 1;
		else
			return 0;
	}
	return p > t && p < t + THREAD_SIZE - size;
}

227 228 229 230 231 232 233
/* The form of the top of the frame on the stack */
struct stack_frame {
	struct stack_frame *next_frame;
	unsigned long return_address;
};


234 235 236 237 238
static inline unsigned long print_context_stack(struct thread_info *tinfo,
				unsigned long *stack, unsigned long bp,
				const struct stacktrace_ops *ops, void *data,
				unsigned long *end)
{
239 240 241 242 243 244
	struct stack_frame *frame = (struct stack_frame *)bp;

	while (valid_stack_ptr(tinfo, stack, sizeof(*stack), end)) {
		unsigned long addr;

		addr = *stack;
245
		if (__kernel_text_address(addr)) {
246 247 248 249 250 251 252
			if ((unsigned long) stack == bp + 8) {
				ops->address(data, addr, 1);
				frame = frame->next_frame;
				bp = (unsigned long) frame;
			} else {
				ops->address(data, addr, bp == 0);
			}
253
		}
254
		stack++;
255 256
	}
	return bp;
257 258
}

259
void dump_trace(struct task_struct *tsk, struct pt_regs *regs,
260
		unsigned long *stack, unsigned long bp,
J
Jan Beulich 已提交
261
		const struct stacktrace_ops *ops, void *data)
L
Linus Torvalds 已提交
262
{
263
	const unsigned cpu = get_cpu();
264
	unsigned long *irqstack_end = (unsigned long*)cpu_pda(cpu)->irqstackptr;
265
	unsigned used = 0;
266
	struct thread_info *tinfo;
L
Linus Torvalds 已提交
267

268 269
	if (!tsk)
		tsk = current;
270
	tinfo = task_thread_info(tsk);
271

272 273 274 275
	if (!stack) {
		unsigned long dummy;
		stack = &dummy;
		if (tsk && tsk != current)
276
			stack = (unsigned long *)tsk->thread.sp;
277 278
	}

279 280 281 282 283 284 285 286 287 288 289 290 291
#ifdef CONFIG_FRAME_POINTER
	if (!bp) {
		if (tsk == current) {
			/* Grab bp right from our regs */
			asm("movq %%rbp, %0" : "=r" (bp):);
		} else {
			/* bp is the last reg pushed by switch_to */
			bp = *(unsigned long *) tsk->thread.sp;
		}
	}
#endif


292

293 294 295 296 297
	/*
	 * Print function call entries in all stacks, starting at the
	 * current stack address. If the stacks consist of nested
	 * exceptions
	 */
298 299
	for (;;) {
		char *id;
300 301 302 303 304
		unsigned long *estack_end;
		estack_end = in_exception_stack(cpu, (unsigned long)stack,
						&used, &id);

		if (estack_end) {
305 306
			if (ops->stack(data, id) < 0)
				break;
307

308 309
			bp = print_context_stack(tinfo, stack, bp, ops,
							data, estack_end);
310
			ops->stack(data, "<EOE>");
311 312 313 314 315
			/*
			 * We link to the next stack via the
			 * second-to-last pointer (index -2 to end) in the
			 * exception stack:
			 */
316 317
			stack = (unsigned long *) estack_end[-2];
			continue;
L
Linus Torvalds 已提交
318
		}
319 320 321 322 323 324
		if (irqstack_end) {
			unsigned long *irqstack;
			irqstack = irqstack_end -
				(IRQSTACKSIZE - 64) / sizeof(*irqstack);

			if (stack >= irqstack && stack < irqstack_end) {
325 326
				if (ops->stack(data, "IRQ") < 0)
					break;
327 328
				bp = print_context_stack(tinfo, stack, bp,
						ops, data, irqstack_end);
329 330 331 332 333
				/*
				 * We link to the next stack (which would be
				 * the process stack normally) the last
				 * pointer (index -1 to end) in the IRQ stack:
				 */
334 335
				stack = (unsigned long *) (irqstack_end[-1]);
				irqstack_end = NULL;
336
				ops->stack(data, "EOI");
337
				continue;
L
Linus Torvalds 已提交
338 339
			}
		}
340
		break;
L
Linus Torvalds 已提交
341
	}
342

343
	/*
344
	 * This handles the process stack:
345
	 */
346
	bp = print_context_stack(tinfo, stack, bp, ops, data, NULL);
347
	put_cpu();
348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367
}
EXPORT_SYMBOL(dump_trace);

static void
print_trace_warning_symbol(void *data, char *msg, unsigned long symbol)
{
	print_symbol(msg, symbol);
	printk("\n");
}

static void print_trace_warning(void *data, char *msg)
{
	printk("%s\n", msg);
}

static int print_trace_stack(void *data, char *name)
{
	printk(" <%s> ", name);
	return 0;
}
368

369
static void print_trace_address(void *data, unsigned long addr, int reliable)
370
{
371
	touch_nmi_watchdog();
372
	printk_address(addr, reliable);
373 374
}

J
Jan Beulich 已提交
375
static const struct stacktrace_ops print_trace_ops = {
376 377 378 379 380 381 382
	.warning = print_trace_warning,
	.warning_symbol = print_trace_warning_symbol,
	.stack = print_trace_stack,
	.address = print_trace_address,
};

void
383 384
show_trace(struct task_struct *tsk, struct pt_regs *regs, unsigned long *stack,
		unsigned long bp)
385 386
{
	printk("\nCall Trace:\n");
387
	dump_trace(tsk, regs, stack, bp, &print_trace_ops, NULL);
L
Linus Torvalds 已提交
388 389 390
	printk("\n");
}

391
static void
392 393
_show_stack(struct task_struct *tsk, struct pt_regs *regs, unsigned long *sp,
							unsigned long bp)
L
Linus Torvalds 已提交
394 395 396
{
	unsigned long *stack;
	int i;
397
	const int cpu = smp_processor_id();
398 399
	unsigned long *irqstack_end = (unsigned long *) (cpu_pda(cpu)->irqstackptr);
	unsigned long *irqstack = (unsigned long *) (cpu_pda(cpu)->irqstackptr - IRQSTACKSIZE);
L
Linus Torvalds 已提交
400 401 402 403

	// debugging aid: "show_stack(NULL, NULL);" prints the
	// back trace for this cpu.

404
	if (sp == NULL) {
L
Linus Torvalds 已提交
405
		if (tsk)
406
			sp = (unsigned long *)tsk->thread.sp;
L
Linus Torvalds 已提交
407
		else
408
			sp = (unsigned long *)&sp;
L
Linus Torvalds 已提交
409 410
	}

411
	stack = sp;
L
Linus Torvalds 已提交
412 413 414 415 416 417 418 419 420 421 422
	for(i=0; i < kstack_depth_to_print; i++) {
		if (stack >= irqstack && stack <= irqstack_end) {
			if (stack == irqstack_end) {
				stack = (unsigned long *) (irqstack_end[-1]);
				printk(" <EOI> ");
			}
		} else {
		if (((long) stack & (THREAD_SIZE-1)) == 0)
			break;
		}
		if (i && ((i % 4) == 0))
423 424
			printk("\n");
		printk(" %016lx", *stack++);
425
		touch_nmi_watchdog();
L
Linus Torvalds 已提交
426
	}
427
	show_trace(tsk, regs, sp, bp);
428 429
}

430
void show_stack(struct task_struct *tsk, unsigned long * sp)
431
{
432
	_show_stack(tsk, NULL, sp, 0);
L
Linus Torvalds 已提交
433 434 435 436 437 438 439 440
}

/*
 * The architecture-independent dump_stack generator
 */
void dump_stack(void)
{
	unsigned long dummy;
441
	unsigned long bp = 0;
442

443 444 445 446 447
#ifdef CONFIG_FRAME_POINTER
	if (!bp)
		asm("movq %%rbp, %0" : "=r" (bp):);
#endif

448 449 450 451 452
	printk("Pid: %d, comm: %.20s %s %s %.*s\n",
		current->pid, current->comm, print_tainted(),
		init_utsname()->release,
		(int)strcspn(init_utsname()->version, " "),
		init_utsname()->version);
453
	show_trace(NULL, NULL, &dummy, bp);
L
Linus Torvalds 已提交
454 455 456 457 458 459 460
}

EXPORT_SYMBOL(dump_stack);

void show_registers(struct pt_regs *regs)
{
	int i;
461
	unsigned long sp;
462
	const int cpu = smp_processor_id();
463
	struct task_struct *cur = cpu_pda(cpu)->pcurrent;
464 465 466
	u8 *ip;
	unsigned int code_prologue = code_bytes * 43 / 64;
	unsigned int code_len = code_bytes;
L
Linus Torvalds 已提交
467

468
	sp = regs->sp;
469
	ip = (u8 *) regs->ip - code_prologue;
L
Linus Torvalds 已提交
470 471 472
	printk("CPU %d ", cpu);
	__show_regs(regs);
	printk("Process %s (pid: %d, threadinfo %p, task %p)\n",
A
Al Viro 已提交
473
		cur->comm, cur->pid, task_thread_info(cur), cur);
L
Linus Torvalds 已提交
474 475 476 477 478

	/*
	 * When in-kernel, we also print out the stack and code at the
	 * time of the fault..
	 */
479 480
	if (!user_mode(regs)) {
		unsigned char c;
L
Linus Torvalds 已提交
481
		printk("Stack: ");
482
		_show_stack(NULL, regs, (unsigned long *)sp, regs->bp);
483
		printk("\n");
L
Linus Torvalds 已提交
484

485 486 487 488 489 490 491 492 493
		printk(KERN_EMERG "Code: ");
		if (ip < (u8 *)PAGE_OFFSET || probe_kernel_address(ip, c)) {
			/* try starting at RIP */
			ip = (u8 *) regs->ip;
			code_len = code_len - code_prologue + 1;
		}
		for (i = 0; i < code_len; i++, ip++) {
			if (ip < (u8 *)PAGE_OFFSET ||
					probe_kernel_address(ip, c)) {
L
Linus Torvalds 已提交
494 495 496
				printk(" Bad RIP value.");
				break;
			}
497 498 499 500
			if (ip == (u8 *)regs->ip)
				printk("<%02x> ", c);
			else
				printk("%02x ", c);
L
Linus Torvalds 已提交
501 502 503 504 505
		}
	}
	printk("\n");
}	

506
int is_valid_bugaddr(unsigned long ip)
507 508 509
{
	unsigned short ud2;

510
	if (__copy_from_user(&ud2, (const void __user *) ip, sizeof(ud2)))
511 512 513 514
		return 0;

	return ud2 == 0x0b0f;
}
L
Linus Torvalds 已提交
515

A
Andi Kleen 已提交
516
static raw_spinlock_t die_lock = __RAW_SPIN_LOCK_UNLOCKED;
L
Linus Torvalds 已提交
517
static int die_owner = -1;
518
static unsigned int die_nest_count;
L
Linus Torvalds 已提交
519

520
unsigned __kprobes long oops_begin(void)
L
Linus Torvalds 已提交
521
{
A
Andrew Morton 已提交
522
	int cpu;
523 524
	unsigned long flags;

525 526
	oops_enter();

527
	/* racy, but better than risking deadlock. */
A
Andi Kleen 已提交
528
	raw_local_irq_save(flags);
A
Andrew Morton 已提交
529
	cpu = smp_processor_id();
A
Andi Kleen 已提交
530
	if (!__raw_spin_trylock(&die_lock)) {
L
Linus Torvalds 已提交
531 532 533
		if (cpu == die_owner) 
			/* nested oops. should stop eventually */;
		else
A
Andi Kleen 已提交
534
			__raw_spin_lock(&die_lock);
L
Linus Torvalds 已提交
535
	}
536
	die_nest_count++;
537
	die_owner = cpu;
L
Linus Torvalds 已提交
538
	console_verbose();
539 540
	bust_spinlocks(1);
	return flags;
L
Linus Torvalds 已提交
541 542
}

543
void __kprobes oops_end(unsigned long flags, struct pt_regs *regs, int signr)
L
Linus Torvalds 已提交
544 545
{ 
	die_owner = -1;
546
	bust_spinlocks(0);
547
	die_nest_count--;
A
Andi Kleen 已提交
548
	if (!die_nest_count)
549
		/* Nest count reaches zero, release the lock. */
A
Andi Kleen 已提交
550 551
		__raw_spin_unlock(&die_lock);
	raw_local_irq_restore(flags);
552 553 554 555
	if (!regs) {
		oops_exit();
		return;
	}
L
Linus Torvalds 已提交
556
	if (panic_on_oops)
557
		panic("Fatal exception");
558
	oops_exit();
559
	do_exit(signr);
560
}
L
Linus Torvalds 已提交
561

562
int __kprobes __die(const char * str, struct pt_regs * regs, long err)
L
Linus Torvalds 已提交
563 564 565 566 567 568 569 570 571 572 573 574 575
{
	static int die_counter;
	printk(KERN_EMERG "%s: %04lx [%u] ", str, err & 0xffff,++die_counter);
#ifdef CONFIG_PREEMPT
	printk("PREEMPT ");
#endif
#ifdef CONFIG_SMP
	printk("SMP ");
#endif
#ifdef CONFIG_DEBUG_PAGEALLOC
	printk("DEBUG_PAGEALLOC");
#endif
	printk("\n");
576 577
	if (notify_die(DIE_OOPS, str, regs, err, current->thread.trap_no, SIGSEGV) == NOTIFY_STOP)
		return 1;
L
Linus Torvalds 已提交
578
	show_registers(regs);
579
	add_taint(TAINT_DIE);
L
Linus Torvalds 已提交
580 581
	/* Executive summary in case the oops scrolled away */
	printk(KERN_ALERT "RIP ");
582
	printk_address(regs->ip, 1);
583
	printk(" RSP <%016lx>\n", regs->sp);
584 585
	if (kexec_should_crash(current))
		crash_kexec(regs);
586
	return 0;
L
Linus Torvalds 已提交
587 588 589 590
}

void die(const char * str, struct pt_regs * regs, long err)
{
591 592
	unsigned long flags = oops_begin();

593
	if (!user_mode(regs))
594
		report_bug(regs->ip, regs);
595

596 597 598
	if (__die(str, regs, err))
		regs = NULL;
	oops_end(flags, regs, SIGSEGV);
L
Linus Torvalds 已提交
599 600
}

601
void __kprobes die_nmi(char *str, struct pt_regs *regs, int do_panic)
L
Linus Torvalds 已提交
602
{
603 604
	unsigned long flags = oops_begin();

L
Linus Torvalds 已提交
605 606 607 608
	/*
	 * We are in trouble anyway, lets at least try
	 * to get a message out.
	 */
609
	printk(str, smp_processor_id());
L
Linus Torvalds 已提交
610
	show_registers(regs);
611 612
	if (kexec_should_crash(current))
		crash_kexec(regs);
613 614
	if (do_panic || panic_on_oops)
		panic("Non maskable interrupt");
615
	oops_end(flags, NULL, SIGBUS);
616 617
	nmi_exit();
	local_irq_enable();
618
	do_exit(SIGBUS);
L
Linus Torvalds 已提交
619 620
}

621 622 623
static void __kprobes do_trap(int trapnr, int signr, char *str,
			      struct pt_regs * regs, long error_code,
			      siginfo_t *info)
L
Linus Torvalds 已提交
624
{
625 626 627
	struct task_struct *tsk = current;

	if (user_mode(regs)) {
628 629 630 631 632 633 634 635 636 637 638 639 640
		/*
		 * We want error_code and trap_no set for userspace
		 * faults and 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;
		tsk->thread.trap_no = trapnr;

641
		if (show_unhandled_signals && unhandled_signal(tsk, signr) &&
642
		    printk_ratelimit()) {
L
Linus Torvalds 已提交
643
			printk(KERN_INFO
644
			       "%s[%d] trap %s ip:%lx sp:%lx error:%lx",
L
Linus Torvalds 已提交
645
			       tsk->comm, tsk->pid, str,
646
			       regs->ip, regs->sp, error_code);
647 648 649
			print_vma_addr(" in ", regs->ip);
			printk("\n");
		}
L
Linus Torvalds 已提交
650 651 652 653 654 655 656 657 658

		if (info)
			force_sig_info(signr, info, tsk);
		else
			force_sig(signr, tsk);
		return;
	}


659 660 661 662
	if (!fixup_exception(regs)) {
		tsk->thread.error_code = error_code;
		tsk->thread.trap_no = trapnr;
		die(str, regs, error_code);
L
Linus Torvalds 已提交
663
	}
664
	return;
L
Linus Torvalds 已提交
665 666 667 668 669 670 671 672
}

#define DO_ERROR(trapnr, signr, str, name) \
asmlinkage void do_##name(struct pt_regs * regs, long error_code) \
{ \
	if (notify_die(DIE_TRAP, str, regs, error_code, trapnr, signr) \
							== NOTIFY_STOP) \
		return; \
673
	conditional_sti(regs);						\
L
Linus Torvalds 已提交
674 675 676 677 678 679 680 681 682 683 684
	do_trap(trapnr, signr, str, regs, error_code, NULL); \
}

#define DO_ERROR_INFO(trapnr, signr, str, name, sicode, siaddr) \
asmlinkage void do_##name(struct pt_regs * regs, long error_code) \
{ \
	siginfo_t info; \
	info.si_signo = signr; \
	info.si_errno = 0; \
	info.si_code = sicode; \
	info.si_addr = (void __user *)siaddr; \
685
	trace_hardirqs_fixup(); \
L
Linus Torvalds 已提交
686 687 688
	if (notify_die(DIE_TRAP, str, regs, error_code, trapnr, signr) \
							== NOTIFY_STOP) \
		return; \
689
	conditional_sti(regs);						\
L
Linus Torvalds 已提交
690 691 692
	do_trap(trapnr, signr, str, regs, error_code, &info); \
}

693
DO_ERROR_INFO( 0, SIGFPE,  "divide error", divide_error, FPE_INTDIV, regs->ip)
L
Linus Torvalds 已提交
694 695
DO_ERROR( 4, SIGSEGV, "overflow", overflow)
DO_ERROR( 5, SIGSEGV, "bounds", bounds)
696
DO_ERROR_INFO( 6, SIGILL,  "invalid opcode", invalid_op, ILL_ILLOPN, regs->ip)
L
Linus Torvalds 已提交
697 698 699 700 701 702
DO_ERROR( 7, SIGSEGV, "device not available", device_not_available)
DO_ERROR( 9, SIGFPE,  "coprocessor segment overrun", coprocessor_segment_overrun)
DO_ERROR(10, SIGSEGV, "invalid TSS", invalid_TSS)
DO_ERROR(11, SIGBUS,  "segment not present", segment_not_present)
DO_ERROR_INFO(17, SIGBUS, "alignment check", alignment_check, BUS_ADRALN, 0)
DO_ERROR(18, SIGSEGV, "reserved", reserved)
703 704 705 706 707 708 709 710 711 712 713

/* Runs on IST stack */
asmlinkage void do_stack_segment(struct pt_regs *regs, long error_code)
{
	if (notify_die(DIE_TRAP, "stack segment", regs, error_code,
			12, SIGBUS) == NOTIFY_STOP)
		return;
	preempt_conditional_sti(regs);
	do_trap(12, SIGBUS, "stack segment", regs, error_code, NULL);
	preempt_conditional_cli(regs);
}
714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730

asmlinkage 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 */
	notify_die(DIE_TRAP, str, regs, error_code, 8, SIGSEGV);

	tsk->thread.error_code = error_code;
	tsk->thread.trap_no = 8;

	/* This is always a kernel trap and never fixable (and thus must
	   never return). */
	for (;;)
		die(str, regs, error_code);
}
L
Linus Torvalds 已提交
731

732 733
asmlinkage void __kprobes do_general_protection(struct pt_regs * regs,
						long error_code)
L
Linus Torvalds 已提交
734
{
735 736
	struct task_struct *tsk = current;

L
Linus Torvalds 已提交
737 738
	conditional_sti(regs);

739
	if (user_mode(regs)) {
740 741 742
		tsk->thread.error_code = error_code;
		tsk->thread.trap_no = 13;

743
		if (show_unhandled_signals && unhandled_signal(tsk, SIGSEGV) &&
744
		    printk_ratelimit()) {
L
Linus Torvalds 已提交
745
			printk(KERN_INFO
746
		       "%s[%d] general protection ip:%lx sp:%lx error:%lx",
L
Linus Torvalds 已提交
747
			       tsk->comm, tsk->pid,
748
			       regs->ip, regs->sp, error_code);
749 750 751
			print_vma_addr(" in ", regs->ip);
			printk("\n");
		}
L
Linus Torvalds 已提交
752 753 754 755 756

		force_sig(SIGSEGV, tsk);
		return;
	} 

757 758
	if (fixup_exception(regs))
		return;
759

760 761 762 763 764 765
	tsk->thread.error_code = error_code;
	tsk->thread.trap_no = 13;
	if (notify_die(DIE_GPF, "general protection fault", regs,
				error_code, 13, SIGSEGV) == NOTIFY_STOP)
		return;
	die("general protection fault", regs, error_code);
L
Linus Torvalds 已提交
766 767
}

768 769
static __kprobes void
mem_parity_error(unsigned char reason, struct pt_regs * regs)
L
Linus Torvalds 已提交
770
{
771 772
	printk(KERN_EMERG "Uhhuh. NMI received for unknown reason %02x.\n",
		reason);
773
	printk(KERN_EMERG "You have some hardware problem, likely on the PCI bus.\n");
774

D
Dave Jiang 已提交
775 776 777 778 779 780 781
#if defined(CONFIG_EDAC)
	if(edac_handler_set()) {
		edac_atomic_assert_error();
		return;
	}
#endif

782
	if (panic_on_unrecovered_nmi)
783 784 785
		panic("NMI: Not continuing");

	printk(KERN_EMERG "Dazed and confused, but trying to continue\n");
L
Linus Torvalds 已提交
786 787 788 789 790 791

	/* Clear and disable the memory parity error line. */
	reason = (reason & 0xf) | 4;
	outb(reason, 0x61);
}

792 793
static __kprobes void
io_check_error(unsigned char reason, struct pt_regs * regs)
L
Linus Torvalds 已提交
794 795 796 797 798 799 800 801 802 803 804 805
{
	printk("NMI: IOCK error (debug interrupt?)\n");
	show_registers(regs);

	/* Re-enable the IOCK line, wait for a few seconds */
	reason = (reason & 0xf) | 8;
	outb(reason, 0x61);
	mdelay(2000);
	reason &= ~8;
	outb(reason, 0x61);
}

806 807
static __kprobes void
unknown_nmi_error(unsigned char reason, struct pt_regs * regs)
808 809 810 811
{
	printk(KERN_EMERG "Uhhuh. NMI received for unknown reason %02x.\n",
		reason);
	printk(KERN_EMERG "Do you have a strange power saving mode enabled?\n");
812 813

	if (panic_on_unrecovered_nmi)
814
		panic("NMI: Not continuing");
815

816
	printk(KERN_EMERG "Dazed and confused, but trying to continue\n");
L
Linus Torvalds 已提交
817 818
}

819 820
/* Runs on IST stack. This code must keep interrupts off all the time.
   Nested NMIs are prevented by the CPU. */
821
asmlinkage __kprobes void default_do_nmi(struct pt_regs *regs)
L
Linus Torvalds 已提交
822 823
{
	unsigned char reason = 0;
A
Ashok Raj 已提交
824 825 826
	int cpu;

	cpu = smp_processor_id();
L
Linus Torvalds 已提交
827 828

	/* Only the BSP gets external NMIs from the system.  */
A
Ashok Raj 已提交
829
	if (!cpu)
L
Linus Torvalds 已提交
830 831 832
		reason = get_nmi_reason();

	if (!(reason & 0xc0)) {
833
		if (notify_die(DIE_NMI_IPI, "nmi_ipi", regs, reason, 2, SIGINT)
L
Linus Torvalds 已提交
834 835 836 837 838 839
								== NOTIFY_STOP)
			return;
		/*
		 * Ok, so this is none of the documented NMI sources,
		 * so it must be the NMI watchdog.
		 */
840
		if (nmi_watchdog_tick(regs,reason))
L
Linus Torvalds 已提交
841
			return;
842 843 844
		if (!do_nmi_callback(regs,cpu))
			unknown_nmi_error(reason, regs);

L
Linus Torvalds 已提交
845 846
		return;
	}
847
	if (notify_die(DIE_NMI, "nmi", regs, reason, 2, SIGINT) == NOTIFY_STOP)
L
Linus Torvalds 已提交
848 849 850 851 852 853 854 855 856 857
		return; 

	/* AK: following checks seem to be broken on modern chipsets. FIXME */

	if (reason & 0x80)
		mem_parity_error(reason, regs);
	if (reason & 0x40)
		io_check_error(reason, regs);
}

858
/* runs on IST stack. */
859
asmlinkage void __kprobes do_int3(struct pt_regs * regs, long error_code)
L
Linus Torvalds 已提交
860
{
P
Peter Zijlstra 已提交
861 862
	trace_hardirqs_fixup();

L
Linus Torvalds 已提交
863 864 865
	if (notify_die(DIE_INT3, "int3", regs, error_code, 3, SIGTRAP) == NOTIFY_STOP) {
		return;
	}
866
	preempt_conditional_sti(regs);
L
Linus Torvalds 已提交
867
	do_trap(3, SIGTRAP, "int3", regs, error_code, NULL);
868
	preempt_conditional_cli(regs);
L
Linus Torvalds 已提交
869 870
}

871 872 873
/* 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 */
874
asmlinkage __kprobes struct pt_regs *sync_regs(struct pt_regs *eregs)
875 876 877
{
	struct pt_regs *regs = eregs;
	/* Did already sync */
878
	if (eregs == (struct pt_regs *)eregs->sp)
879 880
		;
	/* Exception from user space */
881
	else if (user_mode(eregs))
A
Al Viro 已提交
882
		regs = task_pt_regs(current);
883 884
	/* Exception from kernel and interrupts are enabled. Move to
 	   kernel process stack. */
885 886
	else if (eregs->flags & X86_EFLAGS_IF)
		regs = (struct pt_regs *)(eregs->sp -= sizeof(struct pt_regs));
887 888 889 890 891
	if (eregs != regs)
		*regs = *eregs;
	return regs;
}

L
Linus Torvalds 已提交
892
/* runs on IST stack. */
893 894
asmlinkage void __kprobes do_debug(struct pt_regs * regs,
				   unsigned long error_code)
L
Linus Torvalds 已提交
895 896 897 898 899
{
	unsigned long condition;
	struct task_struct *tsk = current;
	siginfo_t info;

900 901
	trace_hardirqs_fixup();

902
	get_debugreg(condition, 6);
L
Linus Torvalds 已提交
903

904 905 906 907 908 909
	/*
	 * The processor cleared BTF, so don't mark that we need it set.
	 */
	clear_tsk_thread_flag(tsk, TIF_DEBUGCTLMSR);
	tsk->thread.debugctlmsr = 0;

L
Linus Torvalds 已提交
910
	if (notify_die(DIE_DEBUG, "debug", regs, condition, error_code,
911
						SIGTRAP) == NOTIFY_STOP)
912
		return;
913

914
	preempt_conditional_sti(regs);
L
Linus Torvalds 已提交
915 916 917 918 919 920 921 922 923 924

	/* Mask out spurious debug traps due to lazy DR7 setting */
	if (condition & (DR_TRAP0|DR_TRAP1|DR_TRAP2|DR_TRAP3)) {
		if (!tsk->thread.debugreg7) { 
			goto clear_dr7;
		}
	}

	tsk->thread.debugreg6 = condition;

R
Roland McGrath 已提交
925 926 927 928 929

	/*
	 * Single-stepping through TF: make sure we ignore any events in
	 * kernel space (but re-enable TF when returning to user mode).
	 */
930
	if (condition & DR_STEP) {
931
                if (!user_mode(regs))
L
Linus Torvalds 已提交
932 933 934 935 936 937 938 939 940
                       goto clear_TF_reenable;
	}

	/* Ok, finally something we can handle */
	tsk->thread.trap_no = 1;
	tsk->thread.error_code = error_code;
	info.si_signo = SIGTRAP;
	info.si_errno = 0;
	info.si_code = TRAP_BRKPT;
941
	info.si_addr = user_mode(regs) ? (void __user *)regs->ip : NULL;
942
	force_sig_info(SIGTRAP, &info, tsk);
L
Linus Torvalds 已提交
943 944

clear_dr7:
945
	set_debugreg(0UL, 7);
946
	preempt_conditional_cli(regs);
947
	return;
L
Linus Torvalds 已提交
948 949 950

clear_TF_reenable:
	set_tsk_thread_flag(tsk, TIF_SINGLESTEP);
951
	regs->flags &= ~X86_EFLAGS_TF;
952
	preempt_conditional_cli(regs);
L
Linus Torvalds 已提交
953 954
}

955
static int kernel_math_error(struct pt_regs *regs, const char *str, int trapnr)
L
Linus Torvalds 已提交
956
{
957
	if (fixup_exception(regs))
L
Linus Torvalds 已提交
958
		return 1;
959

960
	notify_die(DIE_GPF, str, regs, 0, trapnr, SIGFPE);
961
	/* Illegal floating point operation in the kernel */
962
	current->thread.trap_no = trapnr;
L
Linus Torvalds 已提交
963 964 965 966 967 968 969 970 971 972 973
	die(str, regs, 0);
	return 0;
}

/*
 * 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
 */
asmlinkage void do_coprocessor_error(struct pt_regs *regs)
{
974
	void __user *ip = (void __user *)(regs->ip);
L
Linus Torvalds 已提交
975 976 977 978 979
	struct task_struct * task;
	siginfo_t info;
	unsigned short cwd, swd;

	conditional_sti(regs);
980
	if (!user_mode(regs) &&
981
	    kernel_math_error(regs, "kernel x87 math error", 16))
L
Linus Torvalds 已提交
982 983 984 985 986 987 988 989 990 991 992 993
		return;

	/*
	 * Save the info for the exception handler and clear the error.
	 */
	task = current;
	save_init_fpu(task);
	task->thread.trap_no = 16;
	task->thread.error_code = 0;
	info.si_signo = SIGFPE;
	info.si_errno = 0;
	info.si_code = __SI_FAULT;
994
	info.si_addr = ip;
L
Linus Torvalds 已提交
995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006
	/*
	 * (~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);
1007
	switch (swd & ~cwd & 0x3f) {
L
Linus Torvalds 已提交
1008 1009 1010 1011
		case 0x000:
		default:
			break;
		case 0x001: /* Invalid Op */
1012 1013 1014 1015 1016
			/*
			 * swd & 0x240 == 0x040: Stack Underflow
			 * swd & 0x240 == 0x240: Stack Overflow
			 * User must clear the SF bit (0x40) if set
			 */
L
Linus Torvalds 已提交
1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042
			info.si_code = FPE_FLTINV;
			break;
		case 0x002: /* Denormalize */
		case 0x010: /* Underflow */
			info.si_code = FPE_FLTUND;
			break;
		case 0x004: /* Zero Divide */
			info.si_code = FPE_FLTDIV;
			break;
		case 0x008: /* Overflow */
			info.si_code = FPE_FLTOVF;
			break;
		case 0x020: /* Precision */
			info.si_code = FPE_FLTRES;
			break;
	}
	force_sig_info(SIGFPE, &info, task);
}

asmlinkage void bad_intr(void)
{
	printk("bad interrupt"); 
}

asmlinkage void do_simd_coprocessor_error(struct pt_regs *regs)
{
1043
	void __user *ip = (void __user *)(regs->ip);
L
Linus Torvalds 已提交
1044 1045 1046 1047 1048
	struct task_struct * task;
	siginfo_t info;
	unsigned short mxcsr;

	conditional_sti(regs);
1049
	if (!user_mode(regs) &&
1050
        	kernel_math_error(regs, "kernel simd math error", 19))
L
Linus Torvalds 已提交
1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062
		return;

	/*
	 * Save the info for the exception handler and clear the error.
	 */
	task = current;
	save_init_fpu(task);
	task->thread.trap_no = 19;
	task->thread.error_code = 0;
	info.si_signo = SIGFPE;
	info.si_errno = 0;
	info.si_code = __SI_FAULT;
1063
	info.si_addr = ip;
L
Linus Torvalds 已提交
1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099
	/*
	 * 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.
	 */
	mxcsr = get_fpu_mxcsr(task);
	switch (~((mxcsr & 0x1f80) >> 7) & (mxcsr & 0x3f)) {
		case 0x000:
		default:
			break;
		case 0x001: /* Invalid Op */
			info.si_code = FPE_FLTINV;
			break;
		case 0x002: /* Denormalize */
		case 0x010: /* Underflow */
			info.si_code = FPE_FLTUND;
			break;
		case 0x004: /* Zero Divide */
			info.si_code = FPE_FLTDIV;
			break;
		case 0x008: /* Overflow */
			info.si_code = FPE_FLTOVF;
			break;
		case 0x020: /* Precision */
			info.si_code = FPE_FLTRES;
			break;
	}
	force_sig_info(SIGFPE, &info, task);
}

asmlinkage void do_spurious_interrupt_bug(struct pt_regs * regs)
{
}

asmlinkage void __attribute__((weak)) smp_thermal_interrupt(void)
1100 1101 1102 1103
{
}

asmlinkage void __attribute__((weak)) mce_threshold_interrupt(void)
L
Linus Torvalds 已提交
1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121
{
}

/*
 *  'math_state_restore()' saves the current math information in the
 * 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.
 */
asmlinkage void math_state_restore(void)
{
	struct task_struct *me = current;
	clts();			/* Allow maths ops (or we recurse) */

	if (!used_math())
		init_fpu(me);
	restore_fpu_checking(&me->thread.i387.fxsave);
A
Al Viro 已提交
1122
	task_thread_info(me)->status |= TS_USEDFPU;
1123
	me->fpu_counter++;
L
Linus Torvalds 已提交
1124
}
1125
EXPORT_SYMBOL_GPL(math_state_restore);
L
Linus Torvalds 已提交
1126 1127 1128 1129 1130 1131

void __init trap_init(void)
{
	set_intr_gate(0,&divide_error);
	set_intr_gate_ist(1,&debug,DEBUG_STACK);
	set_intr_gate_ist(2,&nmi,NMI_STACK);
1132
 	set_system_gate_ist(3,&int3,DEBUG_STACK); /* int3 can be called from all */
1133 1134
	set_system_gate(4,&overflow);	/* int4 can be called from all */
	set_intr_gate(5,&bounds);
L
Linus Torvalds 已提交
1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162
	set_intr_gate(6,&invalid_op);
	set_intr_gate(7,&device_not_available);
	set_intr_gate_ist(8,&double_fault, DOUBLEFAULT_STACK);
	set_intr_gate(9,&coprocessor_segment_overrun);
	set_intr_gate(10,&invalid_TSS);
	set_intr_gate(11,&segment_not_present);
	set_intr_gate_ist(12,&stack_segment,STACKFAULT_STACK);
	set_intr_gate(13,&general_protection);
	set_intr_gate(14,&page_fault);
	set_intr_gate(15,&spurious_interrupt_bug);
	set_intr_gate(16,&coprocessor_error);
	set_intr_gate(17,&alignment_check);
#ifdef CONFIG_X86_MCE
	set_intr_gate_ist(18,&machine_check, MCE_STACK); 
#endif
	set_intr_gate(19,&simd_coprocessor_error);

#ifdef CONFIG_IA32_EMULATION
	set_system_gate(IA32_SYSCALL_VECTOR, ia32_syscall);
#endif
       
	/*
	 * Should be a barrier for any external CPU state.
	 */
	cpu_init();
}


1163
static int __init oops_setup(char *s)
L
Linus Torvalds 已提交
1164
{ 
1165 1166 1167 1168 1169
	if (!s)
		return -EINVAL;
	if (!strcmp(s, "panic"))
		panic_on_oops = 1;
	return 0;
L
Linus Torvalds 已提交
1170
} 
1171
early_param("oops", oops_setup);
L
Linus Torvalds 已提交
1172 1173 1174

static int __init kstack_setup(char *s)
{
1175 1176
	if (!s)
		return -EINVAL;
L
Linus Torvalds 已提交
1177
	kstack_depth_to_print = simple_strtoul(s,NULL,0);
1178
	return 0;
L
Linus Torvalds 已提交
1179
}
1180
early_param("kstack", kstack_setup);
1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191


static int __init code_bytes_setup(char *s)
{
	code_bytes = simple_strtoul(s, NULL, 0);
	if (code_bytes > 8192)
		code_bytes = 8192;

	return 1;
}
__setup("code_bytes=", code_bytes_setup);