[linux-2.6-block.git] / arch / x86 / kernel / dumpstack.c

/*
 *  Copyright (C) 1991, 1992  Linus Torvalds
 *  Copyright (C) 2000, 2001, 2002 Andi Kleen, SuSE Labs
 */
#include <linux/kallsyms.h>
#include <linux/kprobes.h>
#include <linux/uaccess.h>
#include <linux/utsname.h>
#include <linux/hardirq.h>
#include <linux/kdebug.h>
#include <linux/module.h>
#include <linux/ptrace.h>
#include <linux/ftrace.h>
#include <linux/kexec.h>
#include <linux/bug.h>
#include <linux/nmi.h>
#include <linux/sysfs.h>

#include <asm/stacktrace.h>


int panic_on_unrecovered_nmi;
int panic_on_io_nmi;
unsigned int code_bytes = 64;
int kstack_depth_to_print = 3 * STACKSLOTS_PER_LINE;
static int die_counter;

bool in_task_stack(unsigned long *stack, struct task_struct *task,
		   struct stack_info *info)
{
	unsigned long *begin = task_stack_page(task);
	unsigned long *end   = task_stack_page(task) + THREAD_SIZE;

	if (stack < begin || stack >= end)
		return false;

	info->type	= STACK_TYPE_TASK;
	info->begin	= begin;
	info->end	= end;
	info->next_sp	= NULL;

	return true;
}

static void printk_stack_address(unsigned long address, int reliable,
				 char *log_lvl)
{
	touch_nmi_watchdog();
	printk("%s [<%p>] %s%pB\n",
		log_lvl, (void *)address, reliable ? "" : "? ",
		(void *)address);
}

void printk_address(unsigned long address)
{
	pr_cont(" [<%p>] %pS\n", (void *)address, (void *)address);
}

/*
 * x86-64 can have up to three kernel stacks:
 * process stack
 * interrupt stack
 * severe exception (double fault, nmi, stack fault, debug, mce) hardware stack
 */

unsigned long
print_context_stack(struct task_struct *task,
		unsigned long *stack, unsigned long bp,
		const struct stacktrace_ops *ops, void *data,
		struct stack_info *info, int *graph)
{
	struct stack_frame *frame = (struct stack_frame *)bp;

	/*
	 * If we overflowed the stack into a guard page, jump back to the
	 * bottom of the usable stack.
	 */
	if ((unsigned long)task_stack_page(task) - (unsigned long)stack <
	    PAGE_SIZE)
		stack = (unsigned long *)task_stack_page(task);

	while (on_stack(info, stack, sizeof(*stack))) {
		unsigned long addr = *stack;

		if (__kernel_text_address(addr)) {
			unsigned long real_addr;
			int reliable = 0;

			if ((unsigned long) stack == bp + sizeof(long)) {
				reliable = 1;
				frame = frame->next_frame;
				bp = (unsigned long) frame;
			}

			/*
			 * When function graph tracing is enabled for a
			 * function, its return address on the stack is
			 * replaced with the address of an ftrace handler
			 * (return_to_handler).  In that case, before printing
			 * the "real" address, we want to print the handler
			 * address as an "unreliable" hint that function graph
			 * tracing was involved.
			 */
			real_addr = ftrace_graph_ret_addr(task, graph, addr,
							  stack);
			if (real_addr != addr)
				ops->address(data, addr, 0);

			ops->address(data, real_addr, reliable);
		}
		stack++;
	}
	return bp;
}
EXPORT_SYMBOL_GPL(print_context_stack);

unsigned long
print_context_stack_bp(struct task_struct *task,
		       unsigned long *stack, unsigned long bp,
		       const struct stacktrace_ops *ops, void *data,
		       struct stack_info *info, int *graph)
{
	struct stack_frame *frame = (struct stack_frame *)bp;
	unsigned long *retp = &frame->return_address;

	while (on_stack(info, stack, sizeof(*stack) * 2)) {
		unsigned long addr = *retp;
		unsigned long real_addr;

		if (!__kernel_text_address(addr))
			break;

		real_addr = ftrace_graph_ret_addr(task, graph, addr, retp);
		if (ops->address(data, real_addr, 1))
			break;

		frame = frame->next_frame;
		retp = &frame->return_address;
	}

	return (unsigned long)frame;
}
EXPORT_SYMBOL_GPL(print_context_stack_bp);

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

/*
 * Print one address/symbol entries per line.
 */
static int print_trace_address(void *data, unsigned long addr, int reliable)
{
	printk_stack_address(addr, reliable, data);
	return 0;
}

static const struct stacktrace_ops print_trace_ops = {
	.stack			= print_trace_stack,
	.address		= print_trace_address,
	.walk_stack		= print_context_stack,
};

void
show_trace_log_lvl(struct task_struct *task, struct pt_regs *regs,
		unsigned long *stack, unsigned long bp, char *log_lvl)
{
	printk("%sCall Trace:\n", log_lvl);
	dump_trace(task, regs, stack, bp, &print_trace_ops, log_lvl);
}

void show_stack(struct task_struct *task, unsigned long *sp)
{
	unsigned long bp = 0;

	task = task ? : current;

	/*
	 * Stack frames below this one aren't interesting.  Don't show them
	 * if we're printing for %current.
	 */
	if (!sp && task == current) {
		sp = get_stack_pointer(current, NULL);
		bp = (unsigned long)get_frame_pointer(current, NULL);
	}

	show_stack_log_lvl(task, NULL, sp, bp, "");
}

void show_stack_regs(struct pt_regs *regs)
{
	show_stack_log_lvl(current, regs, NULL, 0, "");
}

static arch_spinlock_t die_lock = __ARCH_SPIN_LOCK_UNLOCKED;
static int die_owner = -1;
static unsigned int die_nest_count;

unsigned long oops_begin(void)
{
	int cpu;
	unsigned long flags;

	oops_enter();

	/* racy, but better than risking deadlock. */
	raw_local_irq_save(flags);
	cpu = smp_processor_id();
	if (!arch_spin_trylock(&die_lock)) {
		if (cpu == die_owner)
			/* nested oops. should stop eventually */;
		else
			arch_spin_lock(&die_lock);
	}
	die_nest_count++;
	die_owner = cpu;
	console_verbose();
	bust_spinlocks(1);
	return flags;
}
EXPORT_SYMBOL_GPL(oops_begin);
NOKPROBE_SYMBOL(oops_begin);

void __noreturn rewind_stack_do_exit(int signr);

void oops_end(unsigned long flags, struct pt_regs *regs, int signr)
{
	if (regs && kexec_should_crash(current))
		crash_kexec(regs);

	bust_spinlocks(0);
	die_owner = -1;
	add_taint(TAINT_DIE, LOCKDEP_NOW_UNRELIABLE);
	die_nest_count--;
	if (!die_nest_count)
		/* Nest count reaches zero, release the lock. */
		arch_spin_unlock(&die_lock);
	raw_local_irq_restore(flags);
	oops_exit();

	if (!signr)
		return;
	if (in_interrupt())
		panic("Fatal exception in interrupt");
	if (panic_on_oops)
		panic("Fatal exception");

	/*
	 * We're not going to return, but we might be on an IST stack or
	 * have very little stack space left.  Rewind the stack and kill
	 * the task.
	 */
	rewind_stack_do_exit(signr);
}
NOKPROBE_SYMBOL(oops_end);

int __die(const char *str, struct pt_regs *regs, long err)
{
#ifdef CONFIG_X86_32
	unsigned short ss;
	unsigned long sp;
#endif
	printk(KERN_DEFAULT
	       "%s: %04lx [#%d]%s%s%s%s\n", str, err & 0xffff, ++die_counter,
	       IS_ENABLED(CONFIG_PREEMPT) ? " PREEMPT"         : "",
	       IS_ENABLED(CONFIG_SMP)     ? " SMP"             : "",
	       debug_pagealloc_enabled()  ? " DEBUG_PAGEALLOC" : "",
	       IS_ENABLED(CONFIG_KASAN)   ? " KASAN"           : "");

	if (notify_die(DIE_OOPS, str, regs, err,
			current->thread.trap_nr, SIGSEGV) == NOTIFY_STOP)
		return 1;

	print_modules();
	show_regs(regs);
#ifdef CONFIG_X86_32
	if (user_mode(regs)) {
		sp = regs->sp;
		ss = regs->ss & 0xffff;
	} else {
		sp = kernel_stack_pointer(regs);
		savesegment(ss, ss);
	}
	printk(KERN_EMERG "EIP: [<%08lx>] ", regs->ip);
	print_symbol("%s", regs->ip);
	printk(" SS:ESP %04x:%08lx\n", ss, sp);
#else
	/* Executive summary in case the oops scrolled away */
	printk(KERN_ALERT "RIP ");
	printk_address(regs->ip);
	printk(" RSP <%016lx>\n", regs->sp);
#endif
	return 0;
}
NOKPROBE_SYMBOL(__die);

/*
 * This is gone through when something in the kernel has done something bad
 * and is about to be terminated:
 */
void die(const char *str, struct pt_regs *regs, long err)
{
	unsigned long flags = oops_begin();
	int sig = SIGSEGV;

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

	if (__die(str, regs, err))
		sig = 0;
	oops_end(flags, regs, sig);
}

static int __init kstack_setup(char *s)
{
	ssize_t ret;
	unsigned long val;

	if (!s)
		return -EINVAL;

	ret = kstrtoul(s, 0, &val);
	if (ret)
		return ret;
	kstack_depth_to_print = val;
	return 0;
}
early_param("kstack", kstack_setup);

static int __init code_bytes_setup(char *s)
{
	ssize_t ret;
	unsigned long val;

	if (!s)
		return -EINVAL;

	ret = kstrtoul(s, 0, &val);
	if (ret)
		return ret;

	code_bytes = val;
	if (code_bytes > 8192)
		code_bytes = 8192;

	return 1;
}
__setup("code_bytes=", code_bytes_setup);
Commit	Line	Data
878719e8 NH	1	/*
	2	* Copyright (C) 1991, 1992 Linus Torvalds
	3	* Copyright (C) 2000, 2001, 2002 Andi Kleen, SuSE Labs
	4	*/
	5	#include <linux/kallsyms.h>
	6	#include <linux/kprobes.h>
	7	#include <linux/uaccess.h>
	8	#include <linux/utsname.h>
	9	#include <linux/hardirq.h>
	10	#include <linux/kdebug.h>
	11	#include <linux/module.h>
	12	#include <linux/ptrace.h>
712406a6	13	#include <linux/ftrace.h>
878719e8 NH	14	#include <linux/kexec.h>
	15	#include <linux/bug.h>
	16	#include <linux/nmi.h>
	17	#include <linux/sysfs.h>
	18
	19	#include <asm/stacktrace.h>
	20
878719e8 NH	21
878719e8 NH	22	int panic_on_unrecovered_nmi;
5211a242	23	int panic_on_io_nmi;
878719e8 NH	24	unsigned int code_bytes = 64;
	25	int kstack_depth_to_print = 3 * STACKSLOTS_PER_LINE;
	26	static int die_counter;
	27
cb76c939 JP	28	bool in_task_stack(unsigned long stack, struct task_struct task,
	29	struct stack_info *info)
	30	{
	31	unsigned long *begin = task_stack_page(task);
	32	unsigned long *end = task_stack_page(task) + THREAD_SIZE;
	33
	34	if (stack < begin \|\| stack >= end)
	35	return false;
	36
	37	info->type = STACK_TYPE_TASK;
	38	info->begin = begin;
	39	info->end = end;
	40	info->next_sp = NULL;
	41
	42	return true;
	43	}
	44
1fc7f61c	45	static void printk_stack_address(unsigned long address, int reliable,
d438f5fd	46	char *log_lvl)
878719e8	47	{
d438f5fd	48	touch_nmi_watchdog();
1fc7f61c	49	printk("%s [<%p>] %s%pB\n",
d438f5fd	50	log_lvl, (void *)address, reliable ? "" : "? ",
1fc7f61c	51	(void *)address);
878719e8 NH	52	}
878719e8 NH	53
5f01c988 JS	54	void printk_address(unsigned long address)
	55	{
	56	pr_cont(" [<%p>] %pS\n", (void )address, (void )address);
	57	}
	58
878719e8 NH	59	/*
	60	* x86-64 can have up to three kernel stacks:
	61	* process stack
	62	* interrupt stack
	63	* severe exception (double fault, nmi, stack fault, debug, mce) hardware stack
	64	*/
	65
878719e8	66	unsigned long
da01e18a	67	print_context_stack(struct task_struct *task,
878719e8 NH	68	unsigned long *stack, unsigned long bp,
878719e8 NH	69	const struct stacktrace_ops ops, void data,
cb76c939	70	struct stack_info info, int graph)
878719e8 NH	71	{
	72	struct stack_frame frame = (struct stack_frame )bp;
	73
9a2e9da3 AL	74	/*
	75	* If we overflowed the stack into a guard page, jump back to the
	76	* bottom of the usable stack.
	77	*/
	78	if ((unsigned long)task_stack_page(task) - (unsigned long)stack <
	79	PAGE_SIZE)
	80	stack = (unsigned long *)task_stack_page(task);
	81
cb76c939	82	while (on_stack(info, stack, sizeof(*stack))) {
408fe5de	83	unsigned long addr = *stack;
878719e8	84
878719e8	85	if (__kernel_text_address(addr)) {
408fe5de JP	86	unsigned long real_addr;
	87	int reliable = 0;
	88
878719e8	89	if ((unsigned long) stack == bp + sizeof(long)) {
408fe5de	90	reliable = 1;
878719e8 NH	91	frame = frame->next_frame;
878719e8 NH	92	bp = (unsigned long) frame;
878719e8	93	}
408fe5de	94
6f727b84 JP	95	/*
	96	* When function graph tracing is enabled for a
	97	* function, its return address on the stack is
	98	* replaced with the address of an ftrace handler
	99	* (return_to_handler). In that case, before printing
	100	* the "real" address, we want to print the handler
	101	* address as an "unreliable" hint that function graph
	102	* tracing was involved.
	103	*/
408fe5de JP	104	real_addr = ftrace_graph_ret_addr(task, graph, addr,
	105	stack);
	106	if (real_addr != addr)
6f727b84 JP	107	ops->address(data, addr, 0);
	108
	109	ops->address(data, real_addr, reliable);
878719e8 NH	110	}
	111	stack++;
	112	}
	113	return bp;
	114	}
06d65bda FW	115	EXPORT_SYMBOL_GPL(print_context_stack);
	116
	117	unsigned long
da01e18a	118	print_context_stack_bp(struct task_struct *task,
06d65bda FW	119	unsigned long *stack, unsigned long bp,
06d65bda FW	120	const struct stacktrace_ops ops, void data,
cb76c939	121	struct stack_info info, int graph)
06d65bda FW	122	{
06d65bda FW	123	struct stack_frame frame = (struct stack_frame )bp;
408fe5de	124	unsigned long *retp = &frame->return_address;
06d65bda	125
cb76c939	126	while (on_stack(info, stack, sizeof(stack) 2)) {
408fe5de JP	127	unsigned long addr = *retp;
408fe5de JP	128	unsigned long real_addr;
06d65bda	129
c2c5d45d FW	130	if (!__kernel_text_address(addr))
	131	break;
	132
408fe5de	133	real_addr = ftrace_graph_ret_addr(task, graph, addr, retp);
6f727b84 JP	134	if (ops->address(data, real_addr, 1))
6f727b84 JP	135	break;
408fe5de	136
c2c5d45d	137	frame = frame->next_frame;
408fe5de	138	retp = &frame->return_address;
06d65bda	139	}
c2c5d45d	140
06d65bda FW	141	return (unsigned long)frame;
	142	}
	143	EXPORT_SYMBOL_GPL(print_context_stack_bp);
878719e8	144
cb76c939	145	static int print_trace_stack(void data, const char name)
878719e8 NH	146	{
	147	printk("%s <%s> ", (char *)data, name);
	148	return 0;
	149	}
	150
	151	/*
	152	* Print one address/symbol entries per line.
	153	*/
568b329a	154	static int print_trace_address(void *data, unsigned long addr, int reliable)
878719e8	155	{
1fc7f61c	156	printk_stack_address(addr, reliable, data);
568b329a	157	return 0;
878719e8 NH	158	}
	159
	160	static const struct stacktrace_ops print_trace_ops = {
06d65bda FW	161	.stack = print_trace_stack,
06d65bda FW	162	.address = print_trace_address,
61c1917f	163	.walk_stack = print_context_stack,
878719e8 NH	164	};
	165
	166	void
	167	show_trace_log_lvl(struct task_struct task, struct pt_regs regs,
e8e999cf	168	unsigned long stack, unsigned long bp, char log_lvl)
878719e8 NH	169	{
878719e8 NH	170	printk("%sCall Trace:\n", log_lvl);
e8e999cf	171	dump_trace(task, regs, stack, bp, &print_trace_ops, log_lvl);
878719e8 NH	172	}
878719e8 NH	173
878719e8 NH	174	void show_stack(struct task_struct task, unsigned long sp)
878719e8 NH	175	{
a77f2a4e	176	unsigned long bp = 0;
a77f2a4e	177
81539169 JP	178	task = task ? : current;
81539169 JP	179
a77f2a4e TH	180	/*
	181	* Stack frames below this one aren't interesting. Don't show them
	182	* if we're printing for %current.
	183	*/
81539169	184	if (!sp && task == current) {
4b8afafb JP	185	sp = get_stack_pointer(current, NULL);
4b8afafb JP	186	bp = (unsigned long)get_frame_pointer(current, NULL);
a77f2a4e TH	187	}
	188
	189	show_stack_log_lvl(task, NULL, sp, bp, "");
878719e8 NH	190	}
878719e8 NH	191
81c2949f BP	192	void show_stack_regs(struct pt_regs *regs)
81c2949f BP	193	{
5a8ff54c	194	show_stack_log_lvl(current, regs, NULL, 0, "");
81c2949f BP	195	}
81c2949f BP	196
edc35bd7	197	static arch_spinlock_t die_lock = __ARCH_SPIN_LOCK_UNLOCKED;
878719e8 NH	198	static int die_owner = -1;
	199	static unsigned int die_nest_count;
	200
9326638c	201	unsigned long oops_begin(void)
878719e8 NH	202	{
	203	int cpu;
	204	unsigned long flags;
	205
	206	oops_enter();
	207
	208	/* racy, but better than risking deadlock. */
	209	raw_local_irq_save(flags);
	210	cpu = smp_processor_id();
0199c4e6	211	if (!arch_spin_trylock(&die_lock)) {
878719e8 NH	212	if (cpu == die_owner)
	213	/* nested oops. should stop eventually */;
	214	else
0199c4e6	215	arch_spin_lock(&die_lock);
878719e8 NH	216	}
	217	die_nest_count++;
	218	die_owner = cpu;
	219	console_verbose();
	220	bust_spinlocks(1);
	221	return flags;
	222	}
81e88fdc	223	EXPORT_SYMBOL_GPL(oops_begin);
9326638c	224	NOKPROBE_SYMBOL(oops_begin);
878719e8	225
2deb4be2 AL	226	void __noreturn rewind_stack_do_exit(int signr);
2deb4be2 AL	227
9326638c	228	void oops_end(unsigned long flags, struct pt_regs *regs, int signr)
878719e8 NH	229	{
	230	if (regs && kexec_should_crash(current))
	231	crash_kexec(regs);
	232
	233	bust_spinlocks(0);
	234	die_owner = -1;
373d4d09	235	add_taint(TAINT_DIE, LOCKDEP_NOW_UNRELIABLE);
878719e8 NH	236	die_nest_count--;
	237	if (!die_nest_count)
	238	/* Nest count reaches zero, release the lock. */
0199c4e6	239	arch_spin_unlock(&die_lock);
878719e8 NH	240	raw_local_irq_restore(flags);
	241	oops_exit();
	242
	243	if (!signr)
	244	return;
	245	if (in_interrupt())
	246	panic("Fatal exception in interrupt");
	247	if (panic_on_oops)
	248	panic("Fatal exception");
2deb4be2 AL	249
	250	/*
	251	* We're not going to return, but we might be on an IST stack or
	252	* have very little stack space left. Rewind the stack and kill
	253	* the task.
	254	*/
	255	rewind_stack_do_exit(signr);
878719e8	256	}
9326638c	257	NOKPROBE_SYMBOL(oops_end);
878719e8	258
9326638c	259	int __die(const char str, struct pt_regs regs, long err)
878719e8 NH	260	{
	261	#ifdef CONFIG_X86_32
	262	unsigned short ss;
	263	unsigned long sp;
	264	#endif
b0f4c4b3	265	printk(KERN_DEFAULT
8fad7ec5 RV	266	"%s: %04lx [#%d]%s%s%s%s\n", str, err & 0xffff, ++die_counter,
	267	IS_ENABLED(CONFIG_PREEMPT) ? " PREEMPT" : "",
	268	IS_ENABLED(CONFIG_SMP) ? " SMP" : "",
	269	debug_pagealloc_enabled() ? " DEBUG_PAGEALLOC" : "",
	270	IS_ENABLED(CONFIG_KASAN) ? " KASAN" : "");
	271
878719e8	272	if (notify_die(DIE_OOPS, str, regs, err,
51e7dc70	273	current->thread.trap_nr, SIGSEGV) == NOTIFY_STOP)
878719e8 NH	274	return 1;
878719e8 NH	275
0fa0e2f0	276	print_modules();
57da8b96	277	show_regs(regs);
878719e8	278	#ifdef CONFIG_X86_32
f39b6f0e	279	if (user_mode(regs)) {
878719e8 NH	280	sp = regs->sp;
878719e8 NH	281	ss = regs->ss & 0xffff;
a343c75d PA	282	} else {
	283	sp = kernel_stack_pointer(regs);
	284	savesegment(ss, ss);
878719e8 NH	285	}
	286	printk(KERN_EMERG "EIP: [<%08lx>] ", regs->ip);
	287	print_symbol("%s", regs->ip);
	288	printk(" SS:ESP %04x:%08lx\n", ss, sp);
	289	#else
	290	/* Executive summary in case the oops scrolled away */
	291	printk(KERN_ALERT "RIP ");
5f01c988	292	printk_address(regs->ip);
878719e8 NH	293	printk(" RSP <%016lx>\n", regs->sp);
	294	#endif
	295	return 0;
	296	}
9326638c	297	NOKPROBE_SYMBOL(__die);
878719e8 NH	298
	299	/*
	300	* This is gone through when something in the kernel has done something bad
	301	* and is about to be terminated:
	302	*/
	303	void die(const char str, struct pt_regs regs, long err)
	304	{
	305	unsigned long flags = oops_begin();
	306	int sig = SIGSEGV;
	307
f39b6f0e	308	if (!user_mode(regs))
878719e8 NH	309	report_bug(regs->ip, regs);
	310
	311	if (__die(str, regs, err))
	312	sig = 0;
	313	oops_end(flags, regs, sig);
	314	}
	315
878719e8 NH	316	static int __init kstack_setup(char *s)
878719e8 NH	317	{
363f7ce3 SK	318	ssize_t ret;
	319	unsigned long val;
	320
878719e8 NH	321	if (!s)
878719e8 NH	322	return -EINVAL;
363f7ce3 SK	323
	324	ret = kstrtoul(s, 0, &val);
	325	if (ret)
	326	return ret;
	327	kstack_depth_to_print = val;
878719e8 NH	328	return 0;
	329	}
	330	early_param("kstack", kstack_setup);
	331
	332	static int __init code_bytes_setup(char *s)
	333	{
363f7ce3 SK	334	ssize_t ret;
	335	unsigned long val;
	336
	337	if (!s)
	338	return -EINVAL;
	339
	340	ret = kstrtoul(s, 0, &val);
	341	if (ret)
	342	return ret;
	343
	344	code_bytes = val;
878719e8 NH	345	if (code_bytes > 8192)
	346	code_bytes = 8192;
	347
	348	return 1;
	349	}
	350	__setup("code_bytes=", code_bytes_setup);