[mirror_ubuntu-bionic-kernel.git] / arch / x86 / kernel / ftrace.c

/*
 * Code for replacing ftrace calls with jumps.
 *
 * Copyright (C) 2007-2008 Steven Rostedt <srostedt@redhat.com>
 *
 * Thanks goes to Ingo Molnar, for suggesting the idea.
 * Mathieu Desnoyers, for suggesting postponing the modifications.
 * Arjan van de Ven, for keeping me straight, and explaining to me
 * the dangers of modifying code on the run.
 */

#include <linux/spinlock.h>
#include <linux/hardirq.h>
#include <linux/uaccess.h>
#include <linux/ftrace.h>
#include <linux/percpu.h>
#include <linux/sched.h>
#include <linux/init.h>
#include <linux/list.h>

#include <asm/cacheflush.h>
#include <asm/ftrace.h>
#include <linux/ftrace.h>
#include <asm/nops.h>
#include <asm/nmi.h>


#ifdef CONFIG_DYNAMIC_FTRACE

int ftrace_arch_code_modify_prepare(void)
{
	set_kernel_text_rw();
	return 0;
}

int ftrace_arch_code_modify_post_process(void)
{
	set_kernel_text_ro();
	return 0;
}

union ftrace_code_union {
	char code[MCOUNT_INSN_SIZE];
	struct {
		char e8;
		int offset;
	} __attribute__((packed));
};

static int ftrace_calc_offset(long ip, long addr)
{
	return (int)(addr - ip);
}

static unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr)
{
	static union ftrace_code_union calc;

	calc.e8		= 0xe8;
	calc.offset	= ftrace_calc_offset(ip + MCOUNT_INSN_SIZE, addr);

	/*
	 * No locking needed, this must be called via kstop_machine
	 * which in essence is like running on a uniprocessor machine.
	 */
	return calc.code;
}

/*
 * Modifying code must take extra care. On an SMP machine, if
 * the code being modified is also being executed on another CPU
 * that CPU will have undefined results and possibly take a GPF.
 * We use kstop_machine to stop other CPUS from exectuing code.
 * But this does not stop NMIs from happening. We still need
 * to protect against that. We separate out the modification of
 * the code to take care of this.
 *
 * Two buffers are added: An IP buffer and a "code" buffer.
 *
 * 1) Put the instruction pointer into the IP buffer
 *    and the new code into the "code" buffer.
 * 2) Set a flag that says we are modifying code
 * 3) Wait for any running NMIs to finish.
 * 4) Write the code
 * 5) clear the flag.
 * 6) Wait for any running NMIs to finish.
 *
 * If an NMI is executed, the first thing it does is to call
 * "ftrace_nmi_enter". This will check if the flag is set to write
 * and if it is, it will write what is in the IP and "code" buffers.
 *
 * The trick is, it does not matter if everyone is writing the same
 * content to the code location. Also, if a CPU is executing code
 * it is OK to write to that code location if the contents being written
 * are the same as what exists.
 */

static atomic_t nmi_running = ATOMIC_INIT(0);
static int mod_code_status;		/* holds return value of text write */
static int mod_code_write;		/* set when NMI should do the write */
static void *mod_code_ip;		/* holds the IP to write to */
static void *mod_code_newcode;		/* holds the text to write to the IP */

static unsigned nmi_wait_count;
static atomic_t nmi_update_count = ATOMIC_INIT(0);

int ftrace_arch_read_dyn_info(char *buf, int size)
{
	int r;

	r = snprintf(buf, size, "%u %u",
		     nmi_wait_count,
		     atomic_read(&nmi_update_count));
	return r;
}

static void ftrace_mod_code(void)
{
	/*
	 * Yes, more than one CPU process can be writing to mod_code_status.
	 *    (and the code itself)
	 * But if one were to fail, then they all should, and if one were
	 * to succeed, then they all should.
	 */
	mod_code_status = probe_kernel_write(mod_code_ip, mod_code_newcode,
					     MCOUNT_INSN_SIZE);

	/* if we fail, then kill any new writers */
	if (mod_code_status)
		mod_code_write = 0;
}

void ftrace_nmi_enter(void)
{
	atomic_inc(&nmi_running);
	/* Must have nmi_running seen before reading write flag */
	smp_mb();
	if (mod_code_write) {
		ftrace_mod_code();
		atomic_inc(&nmi_update_count);
	}
}

void ftrace_nmi_exit(void)
{
	/* Finish all executions before clearing nmi_running */
	smp_wmb();
	atomic_dec(&nmi_running);
}

static void wait_for_nmi(void)
{
	if (!atomic_read(&nmi_running))
		return;

	do {
		cpu_relax();
	} while (atomic_read(&nmi_running));

	nmi_wait_count++;
}

static int
do_ftrace_mod_code(unsigned long ip, void *new_code)
{
	mod_code_ip = (void *)ip;
	mod_code_newcode = new_code;

	/* The buffers need to be visible before we let NMIs write them */
	smp_wmb();

	mod_code_write = 1;

	/* Make sure write bit is visible before we wait on NMIs */
	smp_mb();

	wait_for_nmi();

	/* Make sure all running NMIs have finished before we write the code */
	smp_mb();

	ftrace_mod_code();

	/* Make sure the write happens before clearing the bit */
	smp_wmb();

	mod_code_write = 0;

	/* make sure NMIs see the cleared bit */
	smp_mb();

	wait_for_nmi();

	return mod_code_status;
}


static unsigned char ftrace_nop[MCOUNT_INSN_SIZE];

static unsigned char *ftrace_nop_replace(void)
{
	return ftrace_nop;
}

static int
ftrace_modify_code(unsigned long ip, unsigned char *old_code,
		   unsigned char *new_code)
{
	unsigned char replaced[MCOUNT_INSN_SIZE];

	/*
	 * Note: Due to modules and __init, code can
	 *  disappear and change, we need to protect against faulting
	 *  as well as code changing. We do this by using the
	 *  probe_kernel_* functions.
	 *
	 * No real locking needed, this code is run through
	 * kstop_machine, or before SMP starts.
	 */

	/* read the text we want to modify */
	if (probe_kernel_read(replaced, (void *)ip, MCOUNT_INSN_SIZE))
		return -EFAULT;

	/* Make sure it is what we expect it to be */
	if (memcmp(replaced, old_code, MCOUNT_INSN_SIZE) != 0)
		return -EINVAL;

	/* replace the text with the new text */
	if (do_ftrace_mod_code(ip, new_code))
		return -EPERM;

	sync_core();

	return 0;
}

int ftrace_make_nop(struct module *mod,
		    struct dyn_ftrace *rec, unsigned long addr)
{
	unsigned char *new, *old;
	unsigned long ip = rec->ip;

	old = ftrace_call_replace(ip, addr);
	new = ftrace_nop_replace();

	return ftrace_modify_code(rec->ip, old, new);
}

int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)
{
	unsigned char *new, *old;
	unsigned long ip = rec->ip;

	old = ftrace_nop_replace();
	new = ftrace_call_replace(ip, addr);

	return ftrace_modify_code(rec->ip, old, new);
}

int ftrace_update_ftrace_func(ftrace_func_t func)
{
	unsigned long ip = (unsigned long)(&ftrace_call);
	unsigned char old[MCOUNT_INSN_SIZE], *new;
	int ret;

	memcpy(old, &ftrace_call, MCOUNT_INSN_SIZE);
	new = ftrace_call_replace(ip, (unsigned long)func);
	ret = ftrace_modify_code(ip, old, new);

	return ret;
}

int __init ftrace_dyn_arch_init(void *data)
{
	extern const unsigned char ftrace_test_p6nop[];
	extern const unsigned char ftrace_test_nop5[];
	extern const unsigned char ftrace_test_jmp[];
	int faulted = 0;

	/*
	 * There is no good nop for all x86 archs.
	 * We will default to using the P6_NOP5, but first we
	 * will test to make sure that the nop will actually
	 * work on this CPU. If it faults, we will then
	 * go to a lesser efficient 5 byte nop. If that fails
	 * we then just use a jmp as our nop. This isn't the most
	 * efficient nop, but we can not use a multi part nop
	 * since we would then risk being preempted in the middle
	 * of that nop, and if we enabled tracing then, it might
	 * cause a system crash.
	 *
	 * TODO: check the cpuid to determine the best nop.
	 */
	asm volatile (
		"ftrace_test_jmp:"
		"jmp ftrace_test_p6nop\n"
		"nop\n"
		"nop\n"
		"nop\n"  /* 2 byte jmp + 3 bytes */
		"ftrace_test_p6nop:"
		P6_NOP5
		"jmp 1f\n"
		"ftrace_test_nop5:"
		".byte 0x66,0x66,0x66,0x66,0x90\n"
		"1:"
		".section .fixup, \"ax\"\n"
		"2:	movl $1, %0\n"
		"	jmp ftrace_test_nop5\n"
		"3:	movl $2, %0\n"
		"	jmp 1b\n"
		".previous\n"
		_ASM_EXTABLE(ftrace_test_p6nop, 2b)
		_ASM_EXTABLE(ftrace_test_nop5, 3b)
		: "=r"(faulted) : "0" (faulted));

	switch (faulted) {
	case 0:
		pr_info("ftrace: converting mcount calls to 0f 1f 44 00 00\n");
		memcpy(ftrace_nop, ftrace_test_p6nop, MCOUNT_INSN_SIZE);
		break;
	case 1:
		pr_info("ftrace: converting mcount calls to 66 66 66 66 90\n");
		memcpy(ftrace_nop, ftrace_test_nop5, MCOUNT_INSN_SIZE);
		break;
	case 2:
		pr_info("ftrace: converting mcount calls to jmp . + 5\n");
		memcpy(ftrace_nop, ftrace_test_jmp, MCOUNT_INSN_SIZE);
		break;
	}

	/* The return code is retured via data */
	*(unsigned long *)data = 0;

	return 0;
}
#endif

#ifdef CONFIG_FUNCTION_GRAPH_TRACER

#ifdef CONFIG_DYNAMIC_FTRACE
extern void ftrace_graph_call(void);

static int ftrace_mod_jmp(unsigned long ip,
			  int old_offset, int new_offset)
{
	unsigned char code[MCOUNT_INSN_SIZE];

	if (probe_kernel_read(code, (void *)ip, MCOUNT_INSN_SIZE))
		return -EFAULT;

	if (code[0] != 0xe9 || old_offset != *(int *)(&code[1]))
		return -EINVAL;

	*(int *)(&code[1]) = new_offset;

	if (do_ftrace_mod_code(ip, &code))
		return -EPERM;

	return 0;
}

int ftrace_enable_ftrace_graph_caller(void)
{
	unsigned long ip = (unsigned long)(&ftrace_graph_call);
	int old_offset, new_offset;

	old_offset = (unsigned long)(&ftrace_stub) - (ip + MCOUNT_INSN_SIZE);
	new_offset = (unsigned long)(&ftrace_graph_caller) - (ip + MCOUNT_INSN_SIZE);

	return ftrace_mod_jmp(ip, old_offset, new_offset);
}

int ftrace_disable_ftrace_graph_caller(void)
{
	unsigned long ip = (unsigned long)(&ftrace_graph_call);
	int old_offset, new_offset;

	old_offset = (unsigned long)(&ftrace_graph_caller) - (ip + MCOUNT_INSN_SIZE);
	new_offset = (unsigned long)(&ftrace_stub) - (ip + MCOUNT_INSN_SIZE);

	return ftrace_mod_jmp(ip, old_offset, new_offset);
}

#endif /* !CONFIG_DYNAMIC_FTRACE */

/*
 * Hook the return address and push it in the stack of return addrs
 * in current thread info.
 */
void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr)
{
	unsigned long old;
	unsigned long long calltime;
	int faulted;
	struct ftrace_graph_ent trace;
	unsigned long return_hooker = (unsigned long)
				&return_to_handler;

	/* Nmi's are currently unsupported */
	if (unlikely(in_nmi()))
		return;

	if (unlikely(atomic_read(&current->tracing_graph_pause)))
		return;

	/*
	 * Protect against fault, even if it shouldn't
	 * happen. This tool is too much intrusive to
	 * ignore such a protection.
	 */
	asm volatile(
		"1: " _ASM_MOV " (%[parent]), %[old]\n"
		"2: " _ASM_MOV " %[return_hooker], (%[parent])\n"
		"   movl $0, %[faulted]\n"
		"3:\n"

		".section .fixup, \"ax\"\n"
		"4: movl $1, %[faulted]\n"
		"   jmp 3b\n"
		".previous\n"

		_ASM_EXTABLE(1b, 4b)
		_ASM_EXTABLE(2b, 4b)

		: [old] "=r" (old), [faulted] "=r" (faulted)
		: [parent] "r" (parent), [return_hooker] "r" (return_hooker)
		: "memory"
	);

	if (unlikely(faulted)) {
		ftrace_graph_stop();
		WARN_ON(1);
		return;
	}

	calltime = trace_clock_local();

	if (ftrace_push_return_trace(old, calltime,
				self_addr, &trace.depth) == -EBUSY) {
		*parent = old;
		return;
	}

	trace.func = self_addr;

	/* Only trace if the calling function expects to */
	if (!ftrace_graph_entry(&trace)) {
		current->curr_ret_stack--;
		*parent = old;
	}
}
#endif /* CONFIG_FUNCTION_GRAPH_TRACER */
Commit	Line	Data
3d083395 SR	1	/*
	2	* Code for replacing ftrace calls with jumps.
	3	*
	4	* Copyright (C) 2007-2008 Steven Rostedt <srostedt@redhat.com>
	5	*
	6	* Thanks goes to Ingo Molnar, for suggesting the idea.
	7	* Mathieu Desnoyers, for suggesting postponing the modifications.
	8	* Arjan van de Ven, for keeping me straight, and explaining to me
	9	* the dangers of modifying code on the run.
	10	*/
	11
	12	#include <linux/spinlock.h>
	13	#include <linux/hardirq.h>
6f93fc07	14	#include <linux/uaccess.h>
3d083395 SR	15	#include <linux/ftrace.h>
3d083395 SR	16	#include <linux/percpu.h>
19b3e967	17	#include <linux/sched.h>
3d083395 SR	18	#include <linux/init.h>
	19	#include <linux/list.h>
	20
16239630	21	#include <asm/cacheflush.h>
395a59d0	22	#include <asm/ftrace.h>
caf4b323	23	#include <linux/ftrace.h>
732f3ca7	24	#include <asm/nops.h>
caf4b323	25	#include <asm/nmi.h>
3d083395	26
3d083395	27
caf4b323	28	#ifdef CONFIG_DYNAMIC_FTRACE
3d083395	29
16239630 SR	30	int ftrace_arch_code_modify_prepare(void)
	31	{
	32	set_kernel_text_rw();
	33	return 0;
	34	}
	35
	36	int ftrace_arch_code_modify_post_process(void)
	37	{
	38	set_kernel_text_ro();
	39	return 0;
	40	}
	41
3d083395	42	union ftrace_code_union {
395a59d0	43	char code[MCOUNT_INSN_SIZE];
3d083395 SR	44	struct {
	45	char e8;
	46	int offset;
	47	} __attribute__((packed));
	48	};
	49
15adc048	50	static int ftrace_calc_offset(long ip, long addr)
3c1720f0 SR	51	{
	52	return (int)(addr - ip);
	53	}
3d083395	54
31e88909	55	static unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr)
3c1720f0 SR	56	{
3c1720f0 SR	57	static union ftrace_code_union calc;
3d083395	58
3c1720f0	59	calc.e8 = 0xe8;
395a59d0	60	calc.offset = ftrace_calc_offset(ip + MCOUNT_INSN_SIZE, addr);
3c1720f0 SR	61
	62	/*
	63	* No locking needed, this must be called via kstop_machine
	64	* which in essence is like running on a uniprocessor machine.
	65	*/
	66	return calc.code;
3d083395 SR	67	}
3d083395 SR	68
17666f02 SR	69	/*
	70	* Modifying code must take extra care. On an SMP machine, if
	71	* the code being modified is also being executed on another CPU
	72	* that CPU will have undefined results and possibly take a GPF.
	73	* We use kstop_machine to stop other CPUS from exectuing code.
	74	* But this does not stop NMIs from happening. We still need
	75	* to protect against that. We separate out the modification of
	76	* the code to take care of this.
	77	*
	78	* Two buffers are added: An IP buffer and a "code" buffer.
	79	*
a26a2a27	80	* 1) Put the instruction pointer into the IP buffer
17666f02 SR	81	* and the new code into the "code" buffer.
	82	* 2) Set a flag that says we are modifying code
	83	* 3) Wait for any running NMIs to finish.
	84	* 4) Write the code
	85	* 5) clear the flag.
	86	* 6) Wait for any running NMIs to finish.
	87	*
	88	* If an NMI is executed, the first thing it does is to call
	89	* "ftrace_nmi_enter". This will check if the flag is set to write
	90	* and if it is, it will write what is in the IP and "code" buffers.
	91	*
	92	* The trick is, it does not matter if everyone is writing the same
	93	* content to the code location. Also, if a CPU is executing code
	94	* it is OK to write to that code location if the contents being written
	95	* are the same as what exists.
	96	*/
	97
4e6ea144	98	static atomic_t nmi_running = ATOMIC_INIT(0);
a26a2a27 SR	99	static int mod_code_status; /* holds return value of text write */
	100	static int mod_code_write; /* set when NMI should do the write */
	101	static void mod_code_ip; / holds the IP to write to */
	102	static void mod_code_newcode; / holds the text to write to the IP */
17666f02	103
a26a2a27 SR	104	static unsigned nmi_wait_count;
a26a2a27 SR	105	static atomic_t nmi_update_count = ATOMIC_INIT(0);
b807c3d0 SR	106
	107	int ftrace_arch_read_dyn_info(char *buf, int size)
	108	{
	109	int r;
	110
	111	r = snprintf(buf, size, "%u %u",
	112	nmi_wait_count,
	113	atomic_read(&nmi_update_count));
	114	return r;
	115	}
	116
17666f02 SR	117	static void ftrace_mod_code(void)
	118	{
	119	/*
	120	* Yes, more than one CPU process can be writing to mod_code_status.
	121	* (and the code itself)
	122	* But if one were to fail, then they all should, and if one were
	123	* to succeed, then they all should.
	124	*/
	125	mod_code_status = probe_kernel_write(mod_code_ip, mod_code_newcode,
	126	MCOUNT_INSN_SIZE);
90c7ac49 SR	127
	128	/* if we fail, then kill any new writers */
	129	if (mod_code_status)
	130	mod_code_write = 0;
17666f02 SR	131	}
17666f02 SR	132
a81bd80a	133	void ftrace_nmi_enter(void)
17666f02	134	{
4e6ea144 SR	135	atomic_inc(&nmi_running);
4e6ea144 SR	136	/* Must have nmi_running seen before reading write flag */
17666f02	137	smp_mb();
b807c3d0	138	if (mod_code_write) {
17666f02	139	ftrace_mod_code();
b807c3d0 SR	140	atomic_inc(&nmi_update_count);
b807c3d0 SR	141	}
17666f02 SR	142	}
17666f02 SR	143
a81bd80a	144	void ftrace_nmi_exit(void)
17666f02	145	{
4e6ea144	146	/* Finish all executions before clearing nmi_running */
17666f02	147	smp_wmb();
4e6ea144	148	atomic_dec(&nmi_running);
17666f02 SR	149	}
	150
	151	static void wait_for_nmi(void)
	152	{
4e6ea144	153	if (!atomic_read(&nmi_running))
89025282	154	return;
b807c3d0	155
89025282	156	do {
17666f02	157	cpu_relax();
4e6ea144	158	} while (atomic_read(&nmi_running));
b807c3d0	159
89025282	160	nmi_wait_count++;
17666f02 SR	161	}
	162
	163	static int
	164	do_ftrace_mod_code(unsigned long ip, void *new_code)
	165	{
	166	mod_code_ip = (void *)ip;
	167	mod_code_newcode = new_code;
	168
	169	/* The buffers need to be visible before we let NMIs write them */
	170	smp_wmb();
	171
	172	mod_code_write = 1;
	173
	174	/* Make sure write bit is visible before we wait on NMIs */
	175	smp_mb();
	176
	177	wait_for_nmi();
	178
	179	/* Make sure all running NMIs have finished before we write the code */
	180	smp_mb();
	181
	182	ftrace_mod_code();
	183
	184	/* Make sure the write happens before clearing the bit */
	185	smp_wmb();
	186
	187	mod_code_write = 0;
	188
	189	/* make sure NMIs see the cleared bit */
	190	smp_mb();
	191
	192	wait_for_nmi();
	193
	194	return mod_code_status;
	195	}
	196
	197
caf4b323 FW	198
	199
	200	static unsigned char ftrace_nop[MCOUNT_INSN_SIZE];
	201
31e88909	202	static unsigned char *ftrace_nop_replace(void)
caf4b323 FW	203	{
	204	return ftrace_nop;
	205	}
	206
31e88909	207	static int
3d083395 SR	208	ftrace_modify_code(unsigned long ip, unsigned char *old_code,
	209	unsigned char *new_code)
	210	{
6f93fc07	211	unsigned char replaced[MCOUNT_INSN_SIZE];
3d083395 SR	212
	213	/*
	214	* Note: Due to modules and __init, code can
	215	* disappear and change, we need to protect against faulting
76aefee5	216	* as well as code changing. We do this by using the
ab9a0918	217	* probe_kernel_* functions.
3d083395 SR	218	*
3d083395 SR	219	* No real locking needed, this code is run through
6f93fc07	220	* kstop_machine, or before SMP starts.
3d083395	221	*/
76aefee5 SR	222
76aefee5 SR	223	/* read the text we want to modify */
ab9a0918	224	if (probe_kernel_read(replaced, (void *)ip, MCOUNT_INSN_SIZE))
593eb8a2	225	return -EFAULT;
6f93fc07	226
76aefee5	227	/* Make sure it is what we expect it to be */
6f93fc07	228	if (memcmp(replaced, old_code, MCOUNT_INSN_SIZE) != 0)
593eb8a2	229	return -EINVAL;
3d083395	230
76aefee5	231	/* replace the text with the new text */
17666f02	232	if (do_ftrace_mod_code(ip, new_code))
593eb8a2	233	return -EPERM;
6f93fc07 SR	234
6f93fc07 SR	235	sync_core();
3d083395	236
6f93fc07	237	return 0;
3d083395 SR	238	}
3d083395 SR	239
31e88909 SR	240	int ftrace_make_nop(struct module *mod,
	241	struct dyn_ftrace *rec, unsigned long addr)
	242	{
	243	unsigned char new, old;
	244	unsigned long ip = rec->ip;
	245
	246	old = ftrace_call_replace(ip, addr);
	247	new = ftrace_nop_replace();
	248
	249	return ftrace_modify_code(rec->ip, old, new);
	250	}
	251
	252	int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)
	253	{
	254	unsigned char new, old;
	255	unsigned long ip = rec->ip;
	256
	257	old = ftrace_nop_replace();
	258	new = ftrace_call_replace(ip, addr);
	259
	260	return ftrace_modify_code(rec->ip, old, new);
	261	}
	262
15adc048	263	int ftrace_update_ftrace_func(ftrace_func_t func)
d61f82d0 SR	264	{
d61f82d0 SR	265	unsigned long ip = (unsigned long)(&ftrace_call);
395a59d0	266	unsigned char old[MCOUNT_INSN_SIZE], *new;
d61f82d0 SR	267	int ret;
d61f82d0 SR	268
395a59d0	269	memcpy(old, &ftrace_call, MCOUNT_INSN_SIZE);
d61f82d0 SR	270	new = ftrace_call_replace(ip, (unsigned long)func);
	271	ret = ftrace_modify_code(ip, old, new);
	272
	273	return ret;
	274	}
	275
d61f82d0	276	int __init ftrace_dyn_arch_init(void *data)
3d083395	277	{
732f3ca7 SR	278	extern const unsigned char ftrace_test_p6nop[];
	279	extern const unsigned char ftrace_test_nop5[];
	280	extern const unsigned char ftrace_test_jmp[];
	281	int faulted = 0;
d61f82d0	282
732f3ca7 SR	283	/*
	284	* There is no good nop for all x86 archs.
	285	* We will default to using the P6_NOP5, but first we
	286	* will test to make sure that the nop will actually
	287	* work on this CPU. If it faults, we will then
	288	* go to a lesser efficient 5 byte nop. If that fails
	289	* we then just use a jmp as our nop. This isn't the most
	290	* efficient nop, but we can not use a multi part nop
	291	* since we would then risk being preempted in the middle
	292	* of that nop, and if we enabled tracing then, it might
	293	* cause a system crash.
	294	*
	295	* TODO: check the cpuid to determine the best nop.
	296	*/
	297	asm volatile (
732f3ca7 SR	298	"ftrace_test_jmp:"
732f3ca7 SR	299	"jmp ftrace_test_p6nop\n"
8b27386a AK	300	"nop\n"
	301	"nop\n"
	302	"nop\n" /* 2 byte jmp + 3 bytes */
732f3ca7 SR	303	"ftrace_test_p6nop:"
	304	P6_NOP5
	305	"jmp 1f\n"
	306	"ftrace_test_nop5:"
	307	".byte 0x66,0x66,0x66,0x66,0x90\n"
732f3ca7 SR	308	"1:"
	309	".section .fixup, \"ax\"\n"
	310	"2: movl $1, %0\n"
	311	" jmp ftrace_test_nop5\n"
	312	"3: movl $2, %0\n"
	313	" jmp 1b\n"
	314	".previous\n"
	315	_ASM_EXTABLE(ftrace_test_p6nop, 2b)
	316	_ASM_EXTABLE(ftrace_test_nop5, 3b)
	317	: "=r"(faulted) : "0" (faulted));
	318
	319	switch (faulted) {
	320	case 0:
	321	pr_info("ftrace: converting mcount calls to 0f 1f 44 00 00\n");
8115f3f0	322	memcpy(ftrace_nop, ftrace_test_p6nop, MCOUNT_INSN_SIZE);
732f3ca7 SR	323	break;
	324	case 1:
	325	pr_info("ftrace: converting mcount calls to 66 66 66 66 90\n");
8115f3f0	326	memcpy(ftrace_nop, ftrace_test_nop5, MCOUNT_INSN_SIZE);
732f3ca7 SR	327	break;
732f3ca7 SR	328	case 2:
8b27386a	329	pr_info("ftrace: converting mcount calls to jmp . + 5\n");
8115f3f0	330	memcpy(ftrace_nop, ftrace_test_jmp, MCOUNT_INSN_SIZE);
732f3ca7 SR	331	break;
	332	}
	333
	334	/* The return code is retured via data */
	335	(unsigned long )data = 0;
dfa60aba	336
3d083395 SR	337	return 0;
3d083395 SR	338	}
caf4b323	339	#endif
e7d3737e	340
fb52607a	341	#ifdef CONFIG_FUNCTION_GRAPH_TRACER
e7d3737e	342
5a45cfe1 SR	343	#ifdef CONFIG_DYNAMIC_FTRACE
	344	extern void ftrace_graph_call(void);
	345
	346	static int ftrace_mod_jmp(unsigned long ip,
	347	int old_offset, int new_offset)
	348	{
	349	unsigned char code[MCOUNT_INSN_SIZE];
	350
	351	if (probe_kernel_read(code, (void *)ip, MCOUNT_INSN_SIZE))
	352	return -EFAULT;
	353
	354	if (code[0] != 0xe9 \|\| old_offset != (int )(&code[1]))
	355	return -EINVAL;
	356
	357	(int )(&code[1]) = new_offset;
	358
	359	if (do_ftrace_mod_code(ip, &code))
	360	return -EPERM;
	361
	362	return 0;
	363	}
	364
	365	int ftrace_enable_ftrace_graph_caller(void)
	366	{
	367	unsigned long ip = (unsigned long)(&ftrace_graph_call);
	368	int old_offset, new_offset;
	369
	370	old_offset = (unsigned long)(&ftrace_stub) - (ip + MCOUNT_INSN_SIZE);
	371	new_offset = (unsigned long)(&ftrace_graph_caller) - (ip + MCOUNT_INSN_SIZE);
	372
	373	return ftrace_mod_jmp(ip, old_offset, new_offset);
	374	}
	375
	376	int ftrace_disable_ftrace_graph_caller(void)
	377	{
	378	unsigned long ip = (unsigned long)(&ftrace_graph_call);
	379	int old_offset, new_offset;
	380
	381	old_offset = (unsigned long)(&ftrace_graph_caller) - (ip + MCOUNT_INSN_SIZE);
	382	new_offset = (unsigned long)(&ftrace_stub) - (ip + MCOUNT_INSN_SIZE);
	383
	384	return ftrace_mod_jmp(ip, old_offset, new_offset);
	385	}
	386
e7d3737e FW	387	#endif /* !CONFIG_DYNAMIC_FTRACE */
e7d3737e FW	388
e7d3737e FW	389	/*
	390	* Hook the return address and push it in the stack of return addrs
	391	* in current thread info.
	392	*/
	393	void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr)
	394	{
	395	unsigned long old;
	396	unsigned long long calltime;
	397	int faulted;
287b6e68	398	struct ftrace_graph_ent trace;
e7d3737e FW	399	unsigned long return_hooker = (unsigned long)
	400	&return_to_handler;
	401
	402	/* Nmi's are currently unsupported */
9a5fd902	403	if (unlikely(in_nmi()))
380c4b14 FW	404	return;
	405
	406	if (unlikely(atomic_read(&current->tracing_graph_pause)))
e7d3737e FW	407	return;
	408
	409	/*
	410	* Protect against fault, even if it shouldn't
	411	* happen. This tool is too much intrusive to
	412	* ignore such a protection.
	413	*/
	414	asm volatile(
96665788 SR	415	"1: " _ASM_MOV " (%[parent]), %[old]\n"
96665788 SR	416	"2: " _ASM_MOV " %[return_hooker], (%[parent])\n"
e7d3737e	417	" movl $0, %[faulted]\n"
e3944bfa	418	"3:\n"
e7d3737e FW	419
e7d3737e FW	420	".section .fixup, \"ax\"\n"
e3944bfa SR	421	"4: movl $1, %[faulted]\n"
e3944bfa SR	422	" jmp 3b\n"
e7d3737e FW	423	".previous\n"
e7d3737e FW	424
e3944bfa SR	425	_ASM_EXTABLE(1b, 4b)
e3944bfa SR	426	_ASM_EXTABLE(2b, 4b)
e7d3737e	427
96665788 SR	428	: [old] "=r" (old), [faulted] "=r" (faulted)
96665788 SR	429	: [parent] "r" (parent), [return_hooker] "r" (return_hooker)
e7d3737e FW	430	: "memory"
	431	);
	432
14a866c5 SR	433	if (unlikely(faulted)) {
	434	ftrace_graph_stop();
	435	WARN_ON(1);
e7d3737e FW	436	return;
	437	}
	438
0012693a	439	calltime = trace_clock_local();
e7d3737e	440
712406a6	441	if (ftrace_push_return_trace(old, calltime,
287b6e68	442	self_addr, &trace.depth) == -EBUSY) {
e7d3737e	443	*parent = old;
287b6e68 FW	444	return;
	445	}
	446
	447	trace.func = self_addr;
287b6e68	448
e49dc19c SR	449	/* Only trace if the calling function expects to */
	450	if (!ftrace_graph_entry(&trace)) {
	451	current->curr_ret_stack--;
	452	*parent = old;
	453	}
e7d3737e	454	}
fb52607a	455	#endif /* CONFIG_FUNCTION_GRAPH_TRACER */