xref: /linux-6.15/include/linux/kernel.h (revision bbb03029)
1 #ifndef _LINUX_KERNEL_H
2 #define _LINUX_KERNEL_H
3 
4 
5 #include <stdarg.h>
6 #include <linux/linkage.h>
7 #include <linux/stddef.h>
8 #include <linux/types.h>
9 #include <linux/compiler.h>
10 #include <linux/bitops.h>
11 #include <linux/log2.h>
12 #include <linux/typecheck.h>
13 #include <linux/printk.h>
14 #include <linux/build_bug.h>
15 #include <asm/byteorder.h>
16 #include <uapi/linux/kernel.h>
17 
18 #define USHRT_MAX	((u16)(~0U))
19 #define SHRT_MAX	((s16)(USHRT_MAX>>1))
20 #define SHRT_MIN	((s16)(-SHRT_MAX - 1))
21 #define INT_MAX		((int)(~0U>>1))
22 #define INT_MIN		(-INT_MAX - 1)
23 #define UINT_MAX	(~0U)
24 #define LONG_MAX	((long)(~0UL>>1))
25 #define LONG_MIN	(-LONG_MAX - 1)
26 #define ULONG_MAX	(~0UL)
27 #define LLONG_MAX	((long long)(~0ULL>>1))
28 #define LLONG_MIN	(-LLONG_MAX - 1)
29 #define ULLONG_MAX	(~0ULL)
30 #define SIZE_MAX	(~(size_t)0)
31 
32 #define U8_MAX		((u8)~0U)
33 #define S8_MAX		((s8)(U8_MAX>>1))
34 #define S8_MIN		((s8)(-S8_MAX - 1))
35 #define U16_MAX		((u16)~0U)
36 #define S16_MAX		((s16)(U16_MAX>>1))
37 #define S16_MIN		((s16)(-S16_MAX - 1))
38 #define U32_MAX		((u32)~0U)
39 #define S32_MAX		((s32)(U32_MAX>>1))
40 #define S32_MIN		((s32)(-S32_MAX - 1))
41 #define U64_MAX		((u64)~0ULL)
42 #define S64_MAX		((s64)(U64_MAX>>1))
43 #define S64_MIN		((s64)(-S64_MAX - 1))
44 
45 #define STACK_MAGIC	0xdeadbeef
46 
47 #define REPEAT_BYTE(x)	((~0ul / 0xff) * (x))
48 
49 /* @a is a power of 2 value */
50 #define ALIGN(x, a)		__ALIGN_KERNEL((x), (a))
51 #define ALIGN_DOWN(x, a)	__ALIGN_KERNEL((x) - ((a) - 1), (a))
52 #define __ALIGN_MASK(x, mask)	__ALIGN_KERNEL_MASK((x), (mask))
53 #define PTR_ALIGN(p, a)		((typeof(p))ALIGN((unsigned long)(p), (a)))
54 #define IS_ALIGNED(x, a)		(((x) & ((typeof(x))(a) - 1)) == 0)
55 
56 /* generic data direction definitions */
57 #define READ			0
58 #define WRITE			1
59 
60 #define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]) + __must_be_array(arr))
61 
62 #define u64_to_user_ptr(x) (		\
63 {					\
64 	typecheck(u64, x);		\
65 	(void __user *)(uintptr_t)x;	\
66 }					\
67 )
68 
69 /*
70  * This looks more complex than it should be. But we need to
71  * get the type for the ~ right in round_down (it needs to be
72  * as wide as the result!), and we want to evaluate the macro
73  * arguments just once each.
74  */
75 #define __round_mask(x, y) ((__typeof__(x))((y)-1))
76 #define round_up(x, y) ((((x)-1) | __round_mask(x, y))+1)
77 #define round_down(x, y) ((x) & ~__round_mask(x, y))
78 
79 #define FIELD_SIZEOF(t, f) (sizeof(((t*)0)->f))
80 #define DIV_ROUND_UP __KERNEL_DIV_ROUND_UP
81 #define DIV_ROUND_UP_ULL(ll,d) \
82 	({ unsigned long long _tmp = (ll)+(d)-1; do_div(_tmp, d); _tmp; })
83 
84 #if BITS_PER_LONG == 32
85 # define DIV_ROUND_UP_SECTOR_T(ll,d) DIV_ROUND_UP_ULL(ll, d)
86 #else
87 # define DIV_ROUND_UP_SECTOR_T(ll,d) DIV_ROUND_UP(ll,d)
88 #endif
89 
90 /* The `const' in roundup() prevents gcc-3.3 from calling __divdi3 */
91 #define roundup(x, y) (					\
92 {							\
93 	const typeof(y) __y = y;			\
94 	(((x) + (__y - 1)) / __y) * __y;		\
95 }							\
96 )
97 #define rounddown(x, y) (				\
98 {							\
99 	typeof(x) __x = (x);				\
100 	__x - (__x % (y));				\
101 }							\
102 )
103 
104 /*
105  * Divide positive or negative dividend by positive or negative divisor
106  * and round to closest integer. Result is undefined for negative
107  * divisors if he dividend variable type is unsigned and for negative
108  * dividends if the divisor variable type is unsigned.
109  */
110 #define DIV_ROUND_CLOSEST(x, divisor)(			\
111 {							\
112 	typeof(x) __x = x;				\
113 	typeof(divisor) __d = divisor;			\
114 	(((typeof(x))-1) > 0 ||				\
115 	 ((typeof(divisor))-1) > 0 ||			\
116 	 (((__x) > 0) == ((__d) > 0))) ?		\
117 		(((__x) + ((__d) / 2)) / (__d)) :	\
118 		(((__x) - ((__d) / 2)) / (__d));	\
119 }							\
120 )
121 /*
122  * Same as above but for u64 dividends. divisor must be a 32-bit
123  * number.
124  */
125 #define DIV_ROUND_CLOSEST_ULL(x, divisor)(		\
126 {							\
127 	typeof(divisor) __d = divisor;			\
128 	unsigned long long _tmp = (x) + (__d) / 2;	\
129 	do_div(_tmp, __d);				\
130 	_tmp;						\
131 }							\
132 )
133 
134 /*
135  * Multiplies an integer by a fraction, while avoiding unnecessary
136  * overflow or loss of precision.
137  */
138 #define mult_frac(x, numer, denom)(			\
139 {							\
140 	typeof(x) quot = (x) / (denom);			\
141 	typeof(x) rem  = (x) % (denom);			\
142 	(quot * (numer)) + ((rem * (numer)) / (denom));	\
143 }							\
144 )
145 
146 
147 #define _RET_IP_		(unsigned long)__builtin_return_address(0)
148 #define _THIS_IP_  ({ __label__ __here; __here: (unsigned long)&&__here; })
149 
150 #ifdef CONFIG_LBDAF
151 # include <asm/div64.h>
152 # define sector_div(a, b) do_div(a, b)
153 #else
154 # define sector_div(n, b)( \
155 { \
156 	int _res; \
157 	_res = (n) % (b); \
158 	(n) /= (b); \
159 	_res; \
160 } \
161 )
162 #endif
163 
164 /**
165  * upper_32_bits - return bits 32-63 of a number
166  * @n: the number we're accessing
167  *
168  * A basic shift-right of a 64- or 32-bit quantity.  Use this to suppress
169  * the "right shift count >= width of type" warning when that quantity is
170  * 32-bits.
171  */
172 #define upper_32_bits(n) ((u32)(((n) >> 16) >> 16))
173 
174 /**
175  * lower_32_bits - return bits 0-31 of a number
176  * @n: the number we're accessing
177  */
178 #define lower_32_bits(n) ((u32)(n))
179 
180 struct completion;
181 struct pt_regs;
182 struct user;
183 
184 #ifdef CONFIG_PREEMPT_VOLUNTARY
185 extern int _cond_resched(void);
186 # define might_resched() _cond_resched()
187 #else
188 # define might_resched() do { } while (0)
189 #endif
190 
191 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
192   void ___might_sleep(const char *file, int line, int preempt_offset);
193   void __might_sleep(const char *file, int line, int preempt_offset);
194 /**
195  * might_sleep - annotation for functions that can sleep
196  *
197  * this macro will print a stack trace if it is executed in an atomic
198  * context (spinlock, irq-handler, ...).
199  *
200  * This is a useful debugging help to be able to catch problems early and not
201  * be bitten later when the calling function happens to sleep when it is not
202  * supposed to.
203  */
204 # define might_sleep() \
205 	do { __might_sleep(__FILE__, __LINE__, 0); might_resched(); } while (0)
206 # define sched_annotate_sleep()	(current->task_state_change = 0)
207 #else
208   static inline void ___might_sleep(const char *file, int line,
209 				   int preempt_offset) { }
210   static inline void __might_sleep(const char *file, int line,
211 				   int preempt_offset) { }
212 # define might_sleep() do { might_resched(); } while (0)
213 # define sched_annotate_sleep() do { } while (0)
214 #endif
215 
216 #define might_sleep_if(cond) do { if (cond) might_sleep(); } while (0)
217 
218 /**
219  * abs - return absolute value of an argument
220  * @x: the value.  If it is unsigned type, it is converted to signed type first.
221  *     char is treated as if it was signed (regardless of whether it really is)
222  *     but the macro's return type is preserved as char.
223  *
224  * Return: an absolute value of x.
225  */
226 #define abs(x)	__abs_choose_expr(x, long long,				\
227 		__abs_choose_expr(x, long,				\
228 		__abs_choose_expr(x, int,				\
229 		__abs_choose_expr(x, short,				\
230 		__abs_choose_expr(x, char,				\
231 		__builtin_choose_expr(					\
232 			__builtin_types_compatible_p(typeof(x), char),	\
233 			(char)({ signed char __x = (x); __x<0?-__x:__x; }), \
234 			((void)0)))))))
235 
236 #define __abs_choose_expr(x, type, other) __builtin_choose_expr(	\
237 	__builtin_types_compatible_p(typeof(x),   signed type) ||	\
238 	__builtin_types_compatible_p(typeof(x), unsigned type),		\
239 	({ signed type __x = (x); __x < 0 ? -__x : __x; }), other)
240 
241 /**
242  * reciprocal_scale - "scale" a value into range [0, ep_ro)
243  * @val: value
244  * @ep_ro: right open interval endpoint
245  *
246  * Perform a "reciprocal multiplication" in order to "scale" a value into
247  * range [0, ep_ro), where the upper interval endpoint is right-open.
248  * This is useful, e.g. for accessing a index of an array containing
249  * ep_ro elements, for example. Think of it as sort of modulus, only that
250  * the result isn't that of modulo. ;) Note that if initial input is a
251  * small value, then result will return 0.
252  *
253  * Return: a result based on val in interval [0, ep_ro).
254  */
255 static inline u32 reciprocal_scale(u32 val, u32 ep_ro)
256 {
257 	return (u32)(((u64) val * ep_ro) >> 32);
258 }
259 
260 #if defined(CONFIG_MMU) && \
261 	(defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_DEBUG_ATOMIC_SLEEP))
262 #define might_fault() __might_fault(__FILE__, __LINE__)
263 void __might_fault(const char *file, int line);
264 #else
265 static inline void might_fault(void) { }
266 #endif
267 
268 extern struct atomic_notifier_head panic_notifier_list;
269 extern long (*panic_blink)(int state);
270 __printf(1, 2)
271 void panic(const char *fmt, ...) __noreturn __cold;
272 void nmi_panic(struct pt_regs *regs, const char *msg);
273 extern void oops_enter(void);
274 extern void oops_exit(void);
275 void print_oops_end_marker(void);
276 extern int oops_may_print(void);
277 void do_exit(long error_code) __noreturn;
278 void complete_and_exit(struct completion *, long) __noreturn;
279 
280 /* Internal, do not use. */
281 int __must_check _kstrtoul(const char *s, unsigned int base, unsigned long *res);
282 int __must_check _kstrtol(const char *s, unsigned int base, long *res);
283 
284 int __must_check kstrtoull(const char *s, unsigned int base, unsigned long long *res);
285 int __must_check kstrtoll(const char *s, unsigned int base, long long *res);
286 
287 /**
288  * kstrtoul - convert a string to an unsigned long
289  * @s: The start of the string. The string must be null-terminated, and may also
290  *  include a single newline before its terminating null. The first character
291  *  may also be a plus sign, but not a minus sign.
292  * @base: The number base to use. The maximum supported base is 16. If base is
293  *  given as 0, then the base of the string is automatically detected with the
294  *  conventional semantics - If it begins with 0x the number will be parsed as a
295  *  hexadecimal (case insensitive), if it otherwise begins with 0, it will be
296  *  parsed as an octal number. Otherwise it will be parsed as a decimal.
297  * @res: Where to write the result of the conversion on success.
298  *
299  * Returns 0 on success, -ERANGE on overflow and -EINVAL on parsing error.
300  * Used as a replacement for the obsolete simple_strtoull. Return code must
301  * be checked.
302 */
303 static inline int __must_check kstrtoul(const char *s, unsigned int base, unsigned long *res)
304 {
305 	/*
306 	 * We want to shortcut function call, but
307 	 * __builtin_types_compatible_p(unsigned long, unsigned long long) = 0.
308 	 */
309 	if (sizeof(unsigned long) == sizeof(unsigned long long) &&
310 	    __alignof__(unsigned long) == __alignof__(unsigned long long))
311 		return kstrtoull(s, base, (unsigned long long *)res);
312 	else
313 		return _kstrtoul(s, base, res);
314 }
315 
316 /**
317  * kstrtol - convert a string to a long
318  * @s: The start of the string. The string must be null-terminated, and may also
319  *  include a single newline before its terminating null. The first character
320  *  may also be a plus sign or a minus sign.
321  * @base: The number base to use. The maximum supported base is 16. If base is
322  *  given as 0, then the base of the string is automatically detected with the
323  *  conventional semantics - If it begins with 0x the number will be parsed as a
324  *  hexadecimal (case insensitive), if it otherwise begins with 0, it will be
325  *  parsed as an octal number. Otherwise it will be parsed as a decimal.
326  * @res: Where to write the result of the conversion on success.
327  *
328  * Returns 0 on success, -ERANGE on overflow and -EINVAL on parsing error.
329  * Used as a replacement for the obsolete simple_strtoull. Return code must
330  * be checked.
331  */
332 static inline int __must_check kstrtol(const char *s, unsigned int base, long *res)
333 {
334 	/*
335 	 * We want to shortcut function call, but
336 	 * __builtin_types_compatible_p(long, long long) = 0.
337 	 */
338 	if (sizeof(long) == sizeof(long long) &&
339 	    __alignof__(long) == __alignof__(long long))
340 		return kstrtoll(s, base, (long long *)res);
341 	else
342 		return _kstrtol(s, base, res);
343 }
344 
345 int __must_check kstrtouint(const char *s, unsigned int base, unsigned int *res);
346 int __must_check kstrtoint(const char *s, unsigned int base, int *res);
347 
348 static inline int __must_check kstrtou64(const char *s, unsigned int base, u64 *res)
349 {
350 	return kstrtoull(s, base, res);
351 }
352 
353 static inline int __must_check kstrtos64(const char *s, unsigned int base, s64 *res)
354 {
355 	return kstrtoll(s, base, res);
356 }
357 
358 static inline int __must_check kstrtou32(const char *s, unsigned int base, u32 *res)
359 {
360 	return kstrtouint(s, base, res);
361 }
362 
363 static inline int __must_check kstrtos32(const char *s, unsigned int base, s32 *res)
364 {
365 	return kstrtoint(s, base, res);
366 }
367 
368 int __must_check kstrtou16(const char *s, unsigned int base, u16 *res);
369 int __must_check kstrtos16(const char *s, unsigned int base, s16 *res);
370 int __must_check kstrtou8(const char *s, unsigned int base, u8 *res);
371 int __must_check kstrtos8(const char *s, unsigned int base, s8 *res);
372 int __must_check kstrtobool(const char *s, bool *res);
373 
374 int __must_check kstrtoull_from_user(const char __user *s, size_t count, unsigned int base, unsigned long long *res);
375 int __must_check kstrtoll_from_user(const char __user *s, size_t count, unsigned int base, long long *res);
376 int __must_check kstrtoul_from_user(const char __user *s, size_t count, unsigned int base, unsigned long *res);
377 int __must_check kstrtol_from_user(const char __user *s, size_t count, unsigned int base, long *res);
378 int __must_check kstrtouint_from_user(const char __user *s, size_t count, unsigned int base, unsigned int *res);
379 int __must_check kstrtoint_from_user(const char __user *s, size_t count, unsigned int base, int *res);
380 int __must_check kstrtou16_from_user(const char __user *s, size_t count, unsigned int base, u16 *res);
381 int __must_check kstrtos16_from_user(const char __user *s, size_t count, unsigned int base, s16 *res);
382 int __must_check kstrtou8_from_user(const char __user *s, size_t count, unsigned int base, u8 *res);
383 int __must_check kstrtos8_from_user(const char __user *s, size_t count, unsigned int base, s8 *res);
384 int __must_check kstrtobool_from_user(const char __user *s, size_t count, bool *res);
385 
386 static inline int __must_check kstrtou64_from_user(const char __user *s, size_t count, unsigned int base, u64 *res)
387 {
388 	return kstrtoull_from_user(s, count, base, res);
389 }
390 
391 static inline int __must_check kstrtos64_from_user(const char __user *s, size_t count, unsigned int base, s64 *res)
392 {
393 	return kstrtoll_from_user(s, count, base, res);
394 }
395 
396 static inline int __must_check kstrtou32_from_user(const char __user *s, size_t count, unsigned int base, u32 *res)
397 {
398 	return kstrtouint_from_user(s, count, base, res);
399 }
400 
401 static inline int __must_check kstrtos32_from_user(const char __user *s, size_t count, unsigned int base, s32 *res)
402 {
403 	return kstrtoint_from_user(s, count, base, res);
404 }
405 
406 /* Obsolete, do not use.  Use kstrto<foo> instead */
407 
408 extern unsigned long simple_strtoul(const char *,char **,unsigned int);
409 extern long simple_strtol(const char *,char **,unsigned int);
410 extern unsigned long long simple_strtoull(const char *,char **,unsigned int);
411 extern long long simple_strtoll(const char *,char **,unsigned int);
412 
413 extern int num_to_str(char *buf, int size, unsigned long long num);
414 
415 /* lib/printf utilities */
416 
417 extern __printf(2, 3) int sprintf(char *buf, const char * fmt, ...);
418 extern __printf(2, 0) int vsprintf(char *buf, const char *, va_list);
419 extern __printf(3, 4)
420 int snprintf(char *buf, size_t size, const char *fmt, ...);
421 extern __printf(3, 0)
422 int vsnprintf(char *buf, size_t size, const char *fmt, va_list args);
423 extern __printf(3, 4)
424 int scnprintf(char *buf, size_t size, const char *fmt, ...);
425 extern __printf(3, 0)
426 int vscnprintf(char *buf, size_t size, const char *fmt, va_list args);
427 extern __printf(2, 3) __malloc
428 char *kasprintf(gfp_t gfp, const char *fmt, ...);
429 extern __printf(2, 0) __malloc
430 char *kvasprintf(gfp_t gfp, const char *fmt, va_list args);
431 extern __printf(2, 0)
432 const char *kvasprintf_const(gfp_t gfp, const char *fmt, va_list args);
433 
434 extern __scanf(2, 3)
435 int sscanf(const char *, const char *, ...);
436 extern __scanf(2, 0)
437 int vsscanf(const char *, const char *, va_list);
438 
439 extern int get_option(char **str, int *pint);
440 extern char *get_options(const char *str, int nints, int *ints);
441 extern unsigned long long memparse(const char *ptr, char **retptr);
442 extern bool parse_option_str(const char *str, const char *option);
443 extern char *next_arg(char *args, char **param, char **val);
444 
445 extern int core_kernel_text(unsigned long addr);
446 extern int core_kernel_data(unsigned long addr);
447 extern int __kernel_text_address(unsigned long addr);
448 extern int kernel_text_address(unsigned long addr);
449 extern int func_ptr_is_kernel_text(void *ptr);
450 
451 unsigned long int_sqrt(unsigned long);
452 
453 extern void bust_spinlocks(int yes);
454 extern int oops_in_progress;		/* If set, an oops, panic(), BUG() or die() is in progress */
455 extern int panic_timeout;
456 extern int panic_on_oops;
457 extern int panic_on_unrecovered_nmi;
458 extern int panic_on_io_nmi;
459 extern int panic_on_warn;
460 extern int sysctl_panic_on_rcu_stall;
461 extern int sysctl_panic_on_stackoverflow;
462 
463 extern bool crash_kexec_post_notifiers;
464 
465 /*
466  * panic_cpu is used for synchronizing panic() and crash_kexec() execution. It
467  * holds a CPU number which is executing panic() currently. A value of
468  * PANIC_CPU_INVALID means no CPU has entered panic() or crash_kexec().
469  */
470 extern atomic_t panic_cpu;
471 #define PANIC_CPU_INVALID	-1
472 
473 /*
474  * Only to be used by arch init code. If the user over-wrote the default
475  * CONFIG_PANIC_TIMEOUT, honor it.
476  */
477 static inline void set_arch_panic_timeout(int timeout, int arch_default_timeout)
478 {
479 	if (panic_timeout == arch_default_timeout)
480 		panic_timeout = timeout;
481 }
482 extern const char *print_tainted(void);
483 enum lockdep_ok {
484 	LOCKDEP_STILL_OK,
485 	LOCKDEP_NOW_UNRELIABLE
486 };
487 extern void add_taint(unsigned flag, enum lockdep_ok);
488 extern int test_taint(unsigned flag);
489 extern unsigned long get_taint(void);
490 extern int root_mountflags;
491 
492 extern bool early_boot_irqs_disabled;
493 
494 /*
495  * Values used for system_state. Ordering of the states must not be changed
496  * as code checks for <, <=, >, >= STATE.
497  */
498 extern enum system_states {
499 	SYSTEM_BOOTING,
500 	SYSTEM_SCHEDULING,
501 	SYSTEM_RUNNING,
502 	SYSTEM_HALT,
503 	SYSTEM_POWER_OFF,
504 	SYSTEM_RESTART,
505 } system_state;
506 
507 #define TAINT_PROPRIETARY_MODULE	0
508 #define TAINT_FORCED_MODULE		1
509 #define TAINT_CPU_OUT_OF_SPEC		2
510 #define TAINT_FORCED_RMMOD		3
511 #define TAINT_MACHINE_CHECK		4
512 #define TAINT_BAD_PAGE			5
513 #define TAINT_USER			6
514 #define TAINT_DIE			7
515 #define TAINT_OVERRIDDEN_ACPI_TABLE	8
516 #define TAINT_WARN			9
517 #define TAINT_CRAP			10
518 #define TAINT_FIRMWARE_WORKAROUND	11
519 #define TAINT_OOT_MODULE		12
520 #define TAINT_UNSIGNED_MODULE		13
521 #define TAINT_SOFTLOCKUP		14
522 #define TAINT_LIVEPATCH			15
523 #define TAINT_FLAGS_COUNT		16
524 
525 struct taint_flag {
526 	char c_true;	/* character printed when tainted */
527 	char c_false;	/* character printed when not tainted */
528 	bool module;	/* also show as a per-module taint flag */
529 };
530 
531 extern const struct taint_flag taint_flags[TAINT_FLAGS_COUNT];
532 
533 extern const char hex_asc[];
534 #define hex_asc_lo(x)	hex_asc[((x) & 0x0f)]
535 #define hex_asc_hi(x)	hex_asc[((x) & 0xf0) >> 4]
536 
537 static inline char *hex_byte_pack(char *buf, u8 byte)
538 {
539 	*buf++ = hex_asc_hi(byte);
540 	*buf++ = hex_asc_lo(byte);
541 	return buf;
542 }
543 
544 extern const char hex_asc_upper[];
545 #define hex_asc_upper_lo(x)	hex_asc_upper[((x) & 0x0f)]
546 #define hex_asc_upper_hi(x)	hex_asc_upper[((x) & 0xf0) >> 4]
547 
548 static inline char *hex_byte_pack_upper(char *buf, u8 byte)
549 {
550 	*buf++ = hex_asc_upper_hi(byte);
551 	*buf++ = hex_asc_upper_lo(byte);
552 	return buf;
553 }
554 
555 extern int hex_to_bin(char ch);
556 extern int __must_check hex2bin(u8 *dst, const char *src, size_t count);
557 extern char *bin2hex(char *dst, const void *src, size_t count);
558 
559 bool mac_pton(const char *s, u8 *mac);
560 
561 /*
562  * General tracing related utility functions - trace_printk(),
563  * tracing_on/tracing_off and tracing_start()/tracing_stop
564  *
565  * Use tracing_on/tracing_off when you want to quickly turn on or off
566  * tracing. It simply enables or disables the recording of the trace events.
567  * This also corresponds to the user space /sys/kernel/debug/tracing/tracing_on
568  * file, which gives a means for the kernel and userspace to interact.
569  * Place a tracing_off() in the kernel where you want tracing to end.
570  * From user space, examine the trace, and then echo 1 > tracing_on
571  * to continue tracing.
572  *
573  * tracing_stop/tracing_start has slightly more overhead. It is used
574  * by things like suspend to ram where disabling the recording of the
575  * trace is not enough, but tracing must actually stop because things
576  * like calling smp_processor_id() may crash the system.
577  *
578  * Most likely, you want to use tracing_on/tracing_off.
579  */
580 
581 enum ftrace_dump_mode {
582 	DUMP_NONE,
583 	DUMP_ALL,
584 	DUMP_ORIG,
585 };
586 
587 #ifdef CONFIG_TRACING
588 void tracing_on(void);
589 void tracing_off(void);
590 int tracing_is_on(void);
591 void tracing_snapshot(void);
592 void tracing_snapshot_alloc(void);
593 
594 extern void tracing_start(void);
595 extern void tracing_stop(void);
596 
597 static inline __printf(1, 2)
598 void ____trace_printk_check_format(const char *fmt, ...)
599 {
600 }
601 #define __trace_printk_check_format(fmt, args...)			\
602 do {									\
603 	if (0)								\
604 		____trace_printk_check_format(fmt, ##args);		\
605 } while (0)
606 
607 /**
608  * trace_printk - printf formatting in the ftrace buffer
609  * @fmt: the printf format for printing
610  *
611  * Note: __trace_printk is an internal function for trace_printk and
612  *       the @ip is passed in via the trace_printk macro.
613  *
614  * This function allows a kernel developer to debug fast path sections
615  * that printk is not appropriate for. By scattering in various
616  * printk like tracing in the code, a developer can quickly see
617  * where problems are occurring.
618  *
619  * This is intended as a debugging tool for the developer only.
620  * Please refrain from leaving trace_printks scattered around in
621  * your code. (Extra memory is used for special buffers that are
622  * allocated when trace_printk() is used)
623  *
624  * A little optization trick is done here. If there's only one
625  * argument, there's no need to scan the string for printf formats.
626  * The trace_puts() will suffice. But how can we take advantage of
627  * using trace_puts() when trace_printk() has only one argument?
628  * By stringifying the args and checking the size we can tell
629  * whether or not there are args. __stringify((__VA_ARGS__)) will
630  * turn into "()\0" with a size of 3 when there are no args, anything
631  * else will be bigger. All we need to do is define a string to this,
632  * and then take its size and compare to 3. If it's bigger, use
633  * do_trace_printk() otherwise, optimize it to trace_puts(). Then just
634  * let gcc optimize the rest.
635  */
636 
637 #define trace_printk(fmt, ...)				\
638 do {							\
639 	char _______STR[] = __stringify((__VA_ARGS__));	\
640 	if (sizeof(_______STR) > 3)			\
641 		do_trace_printk(fmt, ##__VA_ARGS__);	\
642 	else						\
643 		trace_puts(fmt);			\
644 } while (0)
645 
646 #define do_trace_printk(fmt, args...)					\
647 do {									\
648 	static const char *trace_printk_fmt __used			\
649 		__attribute__((section("__trace_printk_fmt"))) =	\
650 		__builtin_constant_p(fmt) ? fmt : NULL;			\
651 									\
652 	__trace_printk_check_format(fmt, ##args);			\
653 									\
654 	if (__builtin_constant_p(fmt))					\
655 		__trace_bprintk(_THIS_IP_, trace_printk_fmt, ##args);	\
656 	else								\
657 		__trace_printk(_THIS_IP_, fmt, ##args);			\
658 } while (0)
659 
660 extern __printf(2, 3)
661 int __trace_bprintk(unsigned long ip, const char *fmt, ...);
662 
663 extern __printf(2, 3)
664 int __trace_printk(unsigned long ip, const char *fmt, ...);
665 
666 /**
667  * trace_puts - write a string into the ftrace buffer
668  * @str: the string to record
669  *
670  * Note: __trace_bputs is an internal function for trace_puts and
671  *       the @ip is passed in via the trace_puts macro.
672  *
673  * This is similar to trace_printk() but is made for those really fast
674  * paths that a developer wants the least amount of "Heisenbug" affects,
675  * where the processing of the print format is still too much.
676  *
677  * This function allows a kernel developer to debug fast path sections
678  * that printk is not appropriate for. By scattering in various
679  * printk like tracing in the code, a developer can quickly see
680  * where problems are occurring.
681  *
682  * This is intended as a debugging tool for the developer only.
683  * Please refrain from leaving trace_puts scattered around in
684  * your code. (Extra memory is used for special buffers that are
685  * allocated when trace_puts() is used)
686  *
687  * Returns: 0 if nothing was written, positive # if string was.
688  *  (1 when __trace_bputs is used, strlen(str) when __trace_puts is used)
689  */
690 
691 #define trace_puts(str) ({						\
692 	static const char *trace_printk_fmt __used			\
693 		__attribute__((section("__trace_printk_fmt"))) =	\
694 		__builtin_constant_p(str) ? str : NULL;			\
695 									\
696 	if (__builtin_constant_p(str))					\
697 		__trace_bputs(_THIS_IP_, trace_printk_fmt);		\
698 	else								\
699 		__trace_puts(_THIS_IP_, str, strlen(str));		\
700 })
701 extern int __trace_bputs(unsigned long ip, const char *str);
702 extern int __trace_puts(unsigned long ip, const char *str, int size);
703 
704 extern void trace_dump_stack(int skip);
705 
706 /*
707  * The double __builtin_constant_p is because gcc will give us an error
708  * if we try to allocate the static variable to fmt if it is not a
709  * constant. Even with the outer if statement.
710  */
711 #define ftrace_vprintk(fmt, vargs)					\
712 do {									\
713 	if (__builtin_constant_p(fmt)) {				\
714 		static const char *trace_printk_fmt __used		\
715 		  __attribute__((section("__trace_printk_fmt"))) =	\
716 			__builtin_constant_p(fmt) ? fmt : NULL;		\
717 									\
718 		__ftrace_vbprintk(_THIS_IP_, trace_printk_fmt, vargs);	\
719 	} else								\
720 		__ftrace_vprintk(_THIS_IP_, fmt, vargs);		\
721 } while (0)
722 
723 extern __printf(2, 0) int
724 __ftrace_vbprintk(unsigned long ip, const char *fmt, va_list ap);
725 
726 extern __printf(2, 0) int
727 __ftrace_vprintk(unsigned long ip, const char *fmt, va_list ap);
728 
729 extern void ftrace_dump(enum ftrace_dump_mode oops_dump_mode);
730 #else
731 static inline void tracing_start(void) { }
732 static inline void tracing_stop(void) { }
733 static inline void trace_dump_stack(int skip) { }
734 
735 static inline void tracing_on(void) { }
736 static inline void tracing_off(void) { }
737 static inline int tracing_is_on(void) { return 0; }
738 static inline void tracing_snapshot(void) { }
739 static inline void tracing_snapshot_alloc(void) { }
740 
741 static inline __printf(1, 2)
742 int trace_printk(const char *fmt, ...)
743 {
744 	return 0;
745 }
746 static __printf(1, 0) inline int
747 ftrace_vprintk(const char *fmt, va_list ap)
748 {
749 	return 0;
750 }
751 static inline void ftrace_dump(enum ftrace_dump_mode oops_dump_mode) { }
752 #endif /* CONFIG_TRACING */
753 
754 /*
755  * min()/max()/clamp() macros that also do
756  * strict type-checking.. See the
757  * "unnecessary" pointer comparison.
758  */
759 #define __min(t1, t2, min1, min2, x, y) ({		\
760 	t1 min1 = (x);					\
761 	t2 min2 = (y);					\
762 	(void) (&min1 == &min2);			\
763 	min1 < min2 ? min1 : min2; })
764 #define min(x, y)					\
765 	__min(typeof(x), typeof(y),			\
766 	      __UNIQUE_ID(min1_), __UNIQUE_ID(min2_),	\
767 	      x, y)
768 
769 #define __max(t1, t2, max1, max2, x, y) ({		\
770 	t1 max1 = (x);					\
771 	t2 max2 = (y);					\
772 	(void) (&max1 == &max2);			\
773 	max1 > max2 ? max1 : max2; })
774 #define max(x, y)					\
775 	__max(typeof(x), typeof(y),			\
776 	      __UNIQUE_ID(max1_), __UNIQUE_ID(max2_),	\
777 	      x, y)
778 
779 #define min3(x, y, z) min((typeof(x))min(x, y), z)
780 #define max3(x, y, z) max((typeof(x))max(x, y), z)
781 
782 /**
783  * min_not_zero - return the minimum that is _not_ zero, unless both are zero
784  * @x: value1
785  * @y: value2
786  */
787 #define min_not_zero(x, y) ({			\
788 	typeof(x) __x = (x);			\
789 	typeof(y) __y = (y);			\
790 	__x == 0 ? __y : ((__y == 0) ? __x : min(__x, __y)); })
791 
792 /**
793  * clamp - return a value clamped to a given range with strict typechecking
794  * @val: current value
795  * @lo: lowest allowable value
796  * @hi: highest allowable value
797  *
798  * This macro does strict typechecking of lo/hi to make sure they are of the
799  * same type as val.  See the unnecessary pointer comparisons.
800  */
801 #define clamp(val, lo, hi) min((typeof(val))max(val, lo), hi)
802 
803 /*
804  * ..and if you can't take the strict
805  * types, you can specify one yourself.
806  *
807  * Or not use min/max/clamp at all, of course.
808  */
809 #define min_t(type, x, y)				\
810 	__min(type, type,				\
811 	      __UNIQUE_ID(min1_), __UNIQUE_ID(min2_),	\
812 	      x, y)
813 
814 #define max_t(type, x, y)				\
815 	__max(type, type,				\
816 	      __UNIQUE_ID(min1_), __UNIQUE_ID(min2_),	\
817 	      x, y)
818 
819 /**
820  * clamp_t - return a value clamped to a given range using a given type
821  * @type: the type of variable to use
822  * @val: current value
823  * @lo: minimum allowable value
824  * @hi: maximum allowable value
825  *
826  * This macro does no typechecking and uses temporary variables of type
827  * 'type' to make all the comparisons.
828  */
829 #define clamp_t(type, val, lo, hi) min_t(type, max_t(type, val, lo), hi)
830 
831 /**
832  * clamp_val - return a value clamped to a given range using val's type
833  * @val: current value
834  * @lo: minimum allowable value
835  * @hi: maximum allowable value
836  *
837  * This macro does no typechecking and uses temporary variables of whatever
838  * type the input argument 'val' is.  This is useful when val is an unsigned
839  * type and min and max are literals that will otherwise be assigned a signed
840  * integer type.
841  */
842 #define clamp_val(val, lo, hi) clamp_t(typeof(val), val, lo, hi)
843 
844 
845 /*
846  * swap - swap value of @a and @b
847  */
848 #define swap(a, b) \
849 	do { typeof(a) __tmp = (a); (a) = (b); (b) = __tmp; } while (0)
850 
851 /**
852  * container_of - cast a member of a structure out to the containing structure
853  * @ptr:	the pointer to the member.
854  * @type:	the type of the container struct this is embedded in.
855  * @member:	the name of the member within the struct.
856  *
857  */
858 #define container_of(ptr, type, member) ({				\
859 	void *__mptr = (void *)(ptr);					\
860 	BUILD_BUG_ON_MSG(!__same_type(*(ptr), ((type *)0)->member) &&	\
861 			 !__same_type(*(ptr), void),			\
862 			 "pointer type mismatch in container_of()");	\
863 	((type *)(__mptr - offsetof(type, member))); })
864 
865 /* Rebuild everything on CONFIG_FTRACE_MCOUNT_RECORD */
866 #ifdef CONFIG_FTRACE_MCOUNT_RECORD
867 # define REBUILD_DUE_TO_FTRACE_MCOUNT_RECORD
868 #endif
869 
870 /* Permissions on a sysfs file: you didn't miss the 0 prefix did you? */
871 #define VERIFY_OCTAL_PERMISSIONS(perms)						\
872 	(BUILD_BUG_ON_ZERO((perms) < 0) +					\
873 	 BUILD_BUG_ON_ZERO((perms) > 0777) +					\
874 	 /* USER_READABLE >= GROUP_READABLE >= OTHER_READABLE */		\
875 	 BUILD_BUG_ON_ZERO((((perms) >> 6) & 4) < (((perms) >> 3) & 4)) +	\
876 	 BUILD_BUG_ON_ZERO((((perms) >> 3) & 4) < ((perms) & 4)) +		\
877 	 /* USER_WRITABLE >= GROUP_WRITABLE */					\
878 	 BUILD_BUG_ON_ZERO((((perms) >> 6) & 2) < (((perms) >> 3) & 2)) +	\
879 	 /* OTHER_WRITABLE?  Generally considered a bad idea. */		\
880 	 BUILD_BUG_ON_ZERO((perms) & 2) +					\
881 	 (perms))
882 #endif
883