ftrace.c 10.1 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13
/*
 * 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>
14
#include <linux/uaccess.h>
15 16
#include <linux/ftrace.h>
#include <linux/percpu.h>
17
#include <linux/sched.h>
18 19 20
#include <linux/init.h>
#include <linux/list.h>

21
#include <asm/ftrace.h>
22
#include <linux/ftrace.h>
23
#include <asm/nops.h>
24
#include <asm/nmi.h>
25 26


27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51

#ifdef CONFIG_FUNCTION_RET_TRACER

/*
 * These functions are picked from those used on
 * this page for dynamic ftrace. They have been
 * simplified to ignore all traces in NMI context.
 */
static atomic_t in_nmi;

void ftrace_nmi_enter(void)
{
	atomic_inc(&in_nmi);
}

void ftrace_nmi_exit(void)
{
	atomic_dec(&in_nmi);
}

/* Add a function return address to the trace stack on thread info.*/
static int push_return_trace(unsigned long ret, unsigned long long time,
				unsigned long func)
{
	int index;
52
	struct thread_info *ti = current_thread_info();
53 54

	/* The return trace stack is full */
55 56
	if (ti->curr_ret_stack == FTRACE_RET_STACK_SIZE - 1)
		return -EBUSY;
57 58 59 60 61 62

	index = ++ti->curr_ret_stack;
	ti->ret_stack[index].ret = ret;
	ti->ret_stack[index].func = func;
	ti->ret_stack[index].calltime = time;

63
	return 0;
64 65 66 67 68 69 70 71
}

/* Retrieve a function return address to the trace stack on thread info.*/
static void pop_return_trace(unsigned long *ret, unsigned long long *time,
				unsigned long *func)
{
	int index;

72
	struct thread_info *ti = current_thread_info();
73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128
	index = ti->curr_ret_stack;
	*ret = ti->ret_stack[index].ret;
	*func = ti->ret_stack[index].func;
	*time = ti->ret_stack[index].calltime;
	ti->curr_ret_stack--;
}

/*
 * Send the trace to the ring-buffer.
 * @return the original return address.
 */
unsigned long ftrace_return_to_handler(void)
{
	struct ftrace_retfunc trace;
	pop_return_trace(&trace.ret, &trace.calltime, &trace.func);
	trace.rettime = cpu_clock(raw_smp_processor_id());
	ftrace_function_return(&trace);

	return trace.ret;
}

/*
 * 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;
	unsigned long return_hooker = (unsigned long)
				&return_to_handler;

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

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

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

		".section __ex_table, \"a\"\n"
		"   .long 1b, 3b\n"
		"   .long 2b, 3b\n"
		".previous\n"

129
		: [parent_replaced] "=r" (parent), [old] "=r" (old),
130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154
		  [faulted] "=r" (faulted)
		: [parent_old] "0" (parent), [return_hooker] "r" (return_hooker)
		: "memory"
	);

	if (WARN_ON(faulted)) {
		unregister_ftrace_return();
		return;
	}

	if (WARN_ON(!__kernel_text_address(old))) {
		unregister_ftrace_return();
		*parent = old;
		return;
	}

	calltime = cpu_clock(raw_smp_processor_id());

	if (push_return_trace(old, calltime, self_addr) == -EBUSY)
		*parent = old;
}

#endif

#ifdef CONFIG_DYNAMIC_FTRACE
155 156

union ftrace_code_union {
157
	char code[MCOUNT_INSN_SIZE];
158 159 160 161 162 163
	struct {
		char e8;
		int offset;
	} __attribute__((packed));
};

164
static int ftrace_calc_offset(long ip, long addr)
165 166 167
{
	return (int)(addr - ip);
}
168

169
unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr)
170 171
{
	static union ftrace_code_union calc;
172

173
	calc.e8		= 0xe8;
174
	calc.offset	= ftrace_calc_offset(ip + MCOUNT_INSN_SIZE, addr);
175 176 177 178 179 180

	/*
	 * No locking needed, this must be called via kstop_machine
	 * which in essence is like running on a uniprocessor machine.
	 */
	return calc.code;
181 182
}

183 184 185 186 187 188 189 190 191 192 193
/*
 * 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.
 *
S
Steven Rostedt 已提交
194
 * 1) Put the instruction pointer into the IP buffer
195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211
 *    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.
 */

S
Steven Rostedt 已提交
212 213 214 215 216
static atomic_t in_nmi = 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 */
217

S
Steven Rostedt 已提交
218 219
static unsigned nmi_wait_count;
static atomic_t nmi_update_count = ATOMIC_INIT(0);
S
Steven Rostedt 已提交
220 221 222 223 224 225 226 227 228 229 230

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;
}

231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248
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);

}

void ftrace_nmi_enter(void)
{
	atomic_inc(&in_nmi);
	/* Must have in_nmi seen before reading write flag */
	smp_mb();
S
Steven Rostedt 已提交
249
	if (mod_code_write) {
250
		ftrace_mod_code();
S
Steven Rostedt 已提交
251 252
		atomic_inc(&nmi_update_count);
	}
253 254 255 256 257 258 259 260 261 262 263
}

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

static void wait_for_nmi(void)
{
S
Steven Rostedt 已提交
264 265 266 267
	int waited = 0;

	while (atomic_read(&in_nmi)) {
		waited = 1;
268
		cpu_relax();
S
Steven Rostedt 已提交
269 270 271 272
	}

	if (waited)
		nmi_wait_count++;
273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309
}

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;
}


310 311 312 313 314 315 316 317 318


static unsigned char ftrace_nop[MCOUNT_INSN_SIZE];

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

319
int
320 321 322
ftrace_modify_code(unsigned long ip, unsigned char *old_code,
		   unsigned char *new_code)
{
323
	unsigned char replaced[MCOUNT_INSN_SIZE];
324 325 326 327

	/*
	 * Note: Due to modules and __init, code can
	 *  disappear and change, we need to protect against faulting
S
Steven Rostedt 已提交
328
	 *  as well as code changing. We do this by using the
S
Steven Rostedt 已提交
329
	 *  probe_kernel_* functions.
330 331
	 *
	 * No real locking needed, this code is run through
332
	 * kstop_machine, or before SMP starts.
333
	 */
S
Steven Rostedt 已提交
334 335

	/* read the text we want to modify */
S
Steven Rostedt 已提交
336
	if (probe_kernel_read(replaced, (void *)ip, MCOUNT_INSN_SIZE))
337
		return -EFAULT;
338

S
Steven Rostedt 已提交
339
	/* Make sure it is what we expect it to be */
340
	if (memcmp(replaced, old_code, MCOUNT_INSN_SIZE) != 0)
341
		return -EINVAL;
342

S
Steven Rostedt 已提交
343
	/* replace the text with the new text */
344
	if (do_ftrace_mod_code(ip, new_code))
345
		return -EPERM;
346 347

	sync_core();
348

349
	return 0;
350 351
}

352
int ftrace_update_ftrace_func(ftrace_func_t func)
353 354
{
	unsigned long ip = (unsigned long)(&ftrace_call);
355
	unsigned char old[MCOUNT_INSN_SIZE], *new;
356 357
	int ret;

358
	memcpy(old, &ftrace_call, MCOUNT_INSN_SIZE);
359 360 361 362 363 364 365
	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)
366
{
367 368 369 370
	extern const unsigned char ftrace_test_p6nop[];
	extern const unsigned char ftrace_test_nop5[];
	extern const unsigned char ftrace_test_jmp[];
	int faulted = 0;
371

372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388
	/*
	 * 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"
389 390 391
		"nop\n"
		"nop\n"
		"nop\n"  /* 2 byte jmp + 3 bytes */
392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410
		"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");
411
		memcpy(ftrace_nop, ftrace_test_p6nop, MCOUNT_INSN_SIZE);
412 413 414
		break;
	case 1:
		pr_info("ftrace: converting mcount calls to 66 66 66 66 90\n");
415
		memcpy(ftrace_nop, ftrace_test_nop5, MCOUNT_INSN_SIZE);
416 417
		break;
	case 2:
418
		pr_info("ftrace: converting mcount calls to jmp . + 5\n");
419
		memcpy(ftrace_nop, ftrace_test_jmp, MCOUNT_INSN_SIZE);
420 421 422 423 424
		break;
	}

	/* The return code is retured via data */
	*(unsigned long *)data = 0;
S
Steven Rostedt 已提交
425

426 427
	return 0;
}
428
#endif