xref: /linux-6.15/include/linux/kernel.h (revision 439e7271)
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 <asm/byteorder.h>
15 #include <uapi/linux/kernel.h>
16 
17 #define USHRT_MAX	((u16)(~0U))
18 #define SHRT_MAX	((s16)(USHRT_MAX>>1))
19 #define SHRT_MIN	((s16)(-SHRT_MAX - 1))
20 #define INT_MAX		((int)(~0U>>1))
21 #define INT_MIN		(-INT_MAX - 1)
22 #define UINT_MAX	(~0U)
23 #define LONG_MAX	((long)(~0UL>>1))
24 #define LONG_MIN	(-LONG_MAX - 1)
25 #define ULONG_MAX	(~0UL)
26 #define LLONG_MAX	((long long)(~0ULL>>1))
27 #define LLONG_MIN	(-LLONG_MAX - 1)
28 #define ULLONG_MAX	(~0ULL)
29 #define SIZE_MAX	(~(size_t)0)
30 
31 #define U8_MAX		((u8)~0U)
32 #define S8_MAX		((s8)(U8_MAX>>1))
33 #define S8_MIN		((s8)(-S8_MAX - 1))
34 #define U16_MAX		((u16)~0U)
35 #define S16_MAX		((s16)(U16_MAX>>1))
36 #define S16_MIN		((s16)(-S16_MAX - 1))
37 #define U32_MAX		((u32)~0U)
38 #define S32_MAX		((s32)(U32_MAX>>1))
39 #define S32_MIN		((s32)(-S32_MAX - 1))
40 #define U64_MAX		((u64)~0ULL)
41 #define S64_MAX		((s64)(U64_MAX>>1))
42 #define S64_MIN		((s64)(-S64_MAX - 1))
43 
44 #define STACK_MAGIC	0xdeadbeef
45 
46 #define REPEAT_BYTE(x)	((~0ul / 0xff) * (x))
47 
48 /* @a is a power of 2 value */
49 #define ALIGN(x, a)		__ALIGN_KERNEL((x), (a))
50 #define ALIGN_DOWN(x, a)	__ALIGN_KERNEL((x) - ((a) - 1), (a))
51 #define __ALIGN_MASK(x, mask)	__ALIGN_KERNEL_MASK((x), (mask))
52 #define PTR_ALIGN(p, a)		((typeof(p))ALIGN((unsigned long)(p), (a)))
53 #define IS_ALIGNED(x, a)		(((x) & ((typeof(x))(a) - 1)) == 0)
54 
55 /* generic data direction definitions */
56 #define READ			0
57 #define WRITE			1
58 
59 #define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]) + __must_be_array(arr))
60 
61 #define u64_to_user_ptr(x) (		\
62 {					\
63 	typecheck(u64, x);		\
64 	(void __user *)(uintptr_t)x;	\
65 }					\
66 )
67 
68 /*
69  * This looks more complex than it should be. But we need to
70  * get the type for the ~ right in round_down (it needs to be
71  * as wide as the result!), and we want to evaluate the macro
72  * arguments just once each.
73  */
74 #define __round_mask(x, y) ((__typeof__(x))((y)-1))
75 #define round_up(x, y) ((((x)-1) | __round_mask(x, y))+1)
76 #define round_down(x, y) ((x) & ~__round_mask(x, y))
77 
78 #define FIELD_SIZEOF(t, f) (sizeof(((t*)0)->f))
79 #define DIV_ROUND_UP __KERNEL_DIV_ROUND_UP
80 #define DIV_ROUND_UP_ULL(ll,d) \
81 	({ unsigned long long _tmp = (ll)+(d)-1; do_div(_tmp, d); _tmp; })
82 
83 #if BITS_PER_LONG == 32
84 # define DIV_ROUND_UP_SECTOR_T(ll,d) DIV_ROUND_UP_ULL(ll, d)
85 #else
86 # define DIV_ROUND_UP_SECTOR_T(ll,d) DIV_ROUND_UP(ll,d)
87 #endif
88 
89 /* The `const' in roundup() prevents gcc-3.3 from calling __divdi3 */
90 #define roundup(x, y) (					\
91 {							\
92 	const typeof(y) __y = y;			\
93 	(((x) + (__y - 1)) / __y) * __y;		\
94 }							\
95 )
96 #define rounddown(x, y) (				\
97 {							\
98 	typeof(x) __x = (x);				\
99 	__x - (__x % (y));				\
100 }							\
101 )
102 
103 /*
104  * Divide positive or negative dividend by positive or negative divisor
105  * and round to closest integer. Result is undefined for negative
106  * divisors if he dividend variable type is unsigned and for negative
107  * dividends if the divisor variable type is unsigned.
108  */
109 #define DIV_ROUND_CLOSEST(x, divisor)(			\
110 {							\
111 	typeof(x) __x = x;				\
112 	typeof(divisor) __d = divisor;			\
113 	(((typeof(x))-1) > 0 ||				\
114 	 ((typeof(divisor))-1) > 0 ||			\
115 	 (((__x) > 0) == ((__d) > 0))) ?		\
116 		(((__x) + ((__d) / 2)) / (__d)) :	\
117 		(((__x) - ((__d) / 2)) / (__d));	\
118 }							\
119 )
120 /*
121  * Same as above but for u64 dividends. divisor must be a 32-bit
122  * number.
123  */
124 #define DIV_ROUND_CLOSEST_ULL(x, divisor)(		\
125 {							\
126 	typeof(divisor) __d = divisor;			\
127 	unsigned long long _tmp = (x) + (__d) / 2;	\
128 	do_div(_tmp, __d);				\
129 	_tmp;						\
130 }							\
131 )
132 
133 /*
134  * Multiplies an integer by a fraction, while avoiding unnecessary
135  * overflow or loss of precision.
136  */
137 #define mult_frac(x, numer, denom)(			\
138 {							\
139 	typeof(x) quot = (x) / (denom);			\
140 	typeof(x) rem  = (x) % (denom);			\
141 	(quot * (numer)) + ((rem * (numer)) / (denom));	\
142 }							\
143 )
144 
145 
146 #define _RET_IP_		(unsigned long)__builtin_return_address(0)
147 #define _THIS_IP_  ({ __label__ __here; __here: (unsigned long)&&__here; })
148 
149 #ifdef CONFIG_LBDAF
150 # include <asm/div64.h>
151 # define sector_div(a, b) do_div(a, b)
152 #else
153 # define sector_div(n, b)( \
154 { \
155 	int _res; \
156 	_res = (n) % (b); \
157 	(n) /= (b); \
158 	_res; \
159 } \
160 )
161 #endif
162 
163 /**
164  * upper_32_bits - return bits 32-63 of a number
165  * @n: the number we're accessing
166  *
167  * A basic shift-right of a 64- or 32-bit quantity.  Use this to suppress
168  * the "right shift count >= width of type" warning when that quantity is
169  * 32-bits.
170  */
171 #define upper_32_bits(n) ((u32)(((n) >> 16) >> 16))
172 
173 /**
174  * lower_32_bits - return bits 0-31 of a number
175  * @n: the number we're accessing
176  */
177 #define lower_32_bits(n) ((u32)(n))
178 
179 struct completion;
180 struct pt_regs;
181 struct user;
182 
183 #ifdef CONFIG_PREEMPT_VOLUNTARY
184 extern int _cond_resched(void);
185 # define might_resched() _cond_resched()
186 #else
187 # define might_resched() do { } while (0)
188 #endif
189 
190 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
191   void ___might_sleep(const char *file, int line, int preempt_offset);
192   void __might_sleep(const char *file, int line, int preempt_offset);
193 /**
194  * might_sleep - annotation for functions that can sleep
195  *
196  * this macro will print a stack trace if it is executed in an atomic
197  * context (spinlock, irq-handler, ...).
198  *
199  * This is a useful debugging help to be able to catch problems early and not
200  * be bitten later when the calling function happens to sleep when it is not
201  * supposed to.
202  */
203 # define might_sleep() \
204 	do { __might_sleep(__FILE__, __LINE__, 0); might_resched(); } while (0)
205 # define sched_annotate_sleep()	(current->task_state_change = 0)
206 #else
207   static inline void ___might_sleep(const char *file, int line,
208 				   int preempt_offset) { }
209   static inline void __might_sleep(const char *file, int line,
210 				   int preempt_offset) { }
211 # define might_sleep() do { might_resched(); } while (0)
212 # define sched_annotate_sleep() do { } while (0)
213 #endif
214 
215 #define might_sleep_if(cond) do { if (cond) might_sleep(); } while (0)
216 
217 /**
218  * abs - return absolute value of an argument
219  * @x: the value.  If it is unsigned type, it is converted to signed type first.
220  *     char is treated as if it was signed (regardless of whether it really is)
221  *     but the macro's return type is preserved as char.
222  *
223  * Return: an absolute value of x.
224  */
225 #define abs(x)	__abs_choose_expr(x, long long,				\
226 		__abs_choose_expr(x, long,				\
227 		__abs_choose_expr(x, int,				\
228 		__abs_choose_expr(x, short,				\
229 		__abs_choose_expr(x, char,				\
230 		__builtin_choose_expr(					\
231 			__builtin_types_compatible_p(typeof(x), char),	\
232 			(char)({ signed char __x = (x); __x<0?-__x:__x; }), \
233 			((void)0)))))))
234 
235 #define __abs_choose_expr(x, type, other) __builtin_choose_expr(	\
236 	__builtin_types_compatible_p(typeof(x),   signed type) ||	\
237 	__builtin_types_compatible_p(typeof(x), unsigned type),		\
238 	({ signed type __x = (x); __x < 0 ? -__x : __x; }), other)
239 
240 /**
241  * reciprocal_scale - "scale" a value into range [0, ep_ro)
242  * @val: value
243  * @ep_ro: right open interval endpoint
244  *
245  * Perform a "reciprocal multiplication" in order to "scale" a value into
246  * range [0, ep_ro), where the upper interval endpoint is right-open.
247  * This is useful, e.g. for accessing a index of an array containing
248  * ep_ro elements, for example. Think of it as sort of modulus, only that
249  * the result isn't that of modulo. ;) Note that if initial input is a
250  * small value, then result will return 0.
251  *
252  * Return: a result based on val in interval [0, ep_ro).
253  */
254 static inline u32 reciprocal_scale(u32 val, u32 ep_ro)
255 {
256 	return (u32)(((u64) val * ep_ro) >> 32);
257 }
258 
259 #if defined(CONFIG_MMU) && \
260 	(defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_DEBUG_ATOMIC_SLEEP))
261 #define might_fault() __might_fault(__FILE__, __LINE__)
262 void __might_fault(const char *file, int line);
263 #else
264 static inline void might_fault(void) { }
265 #endif
266 
267 extern struct atomic_notifier_head panic_notifier_list;
268 extern long (*panic_blink)(int state);
269 __printf(1, 2)
270 void panic(const char *fmt, ...) __noreturn __cold;
271 void nmi_panic(struct pt_regs *regs, const char *msg);
272 extern void oops_enter(void);
273 extern void oops_exit(void);
274 void print_oops_end_marker(void);
275 extern int oops_may_print(void);
276 void do_exit(long error_code) __noreturn;
277 void complete_and_exit(struct completion *, long) __noreturn;
278 
279 /* Internal, do not use. */
280 int __must_check _kstrtoul(const char *s, unsigned int base, unsigned long *res);
281 int __must_check _kstrtol(const char *s, unsigned int base, long *res);
282 
283 int __must_check kstrtoull(const char *s, unsigned int base, unsigned long long *res);
284 int __must_check kstrtoll(const char *s, unsigned int base, long long *res);
285 
286 /**
287  * kstrtoul - convert a string to an unsigned long
288  * @s: The start of the string. The string must be null-terminated, and may also
289  *  include a single newline before its terminating null. The first character
290  *  may also be a plus sign, but not a minus sign.
291  * @base: The number base to use. The maximum supported base is 16. If base is
292  *  given as 0, then the base of the string is automatically detected with the
293  *  conventional semantics - If it begins with 0x the number will be parsed as a
294  *  hexadecimal (case insensitive), if it otherwise begins with 0, it will be
295  *  parsed as an octal number. Otherwise it will be parsed as a decimal.
296  * @res: Where to write the result of the conversion on success.
297  *
298  * Returns 0 on success, -ERANGE on overflow and -EINVAL on parsing error.
299  * Used as a replacement for the obsolete simple_strtoull. Return code must
300  * be checked.
301 */
302 static inline int __must_check kstrtoul(const char *s, unsigned int base, unsigned long *res)
303 {
304 	/*
305 	 * We want to shortcut function call, but
306 	 * __builtin_types_compatible_p(unsigned long, unsigned long long) = 0.
307 	 */
308 	if (sizeof(unsigned long) == sizeof(unsigned long long) &&
309 	    __alignof__(unsigned long) == __alignof__(unsigned long long))
310 		return kstrtoull(s, base, (unsigned long long *)res);
311 	else
312 		return _kstrtoul(s, base, res);
313 }
314 
315 /**
316  * kstrtol - convert a string to a long
317  * @s: The start of the string. The string must be null-terminated, and may also
318  *  include a single newline before its terminating null. The first character
319  *  may also be a plus sign or a minus sign.
320  * @base: The number base to use. The maximum supported base is 16. If base is
321  *  given as 0, then the base of the string is automatically detected with the
322  *  conventional semantics - If it begins with 0x the number will be parsed as a
323  *  hexadecimal (case insensitive), if it otherwise begins with 0, it will be
324  *  parsed as an octal number. Otherwise it will be parsed as a decimal.
325  * @res: Where to write the result of the conversion on success.
326  *
327  * Returns 0 on success, -ERANGE on overflow and -EINVAL on parsing error.
328  * Used as a replacement for the obsolete simple_strtoull. Return code must
329  * be checked.
330  */
331 static inline int __must_check kstrtol(const char *s, unsigned int base, long *res)
332 {
333 	/*
334 	 * We want to shortcut function call, but
335 	 * __builtin_types_compatible_p(long, long long) = 0.
336 	 */
337 	if (sizeof(long) == sizeof(long long) &&
338 	    __alignof__(long) == __alignof__(long long))
339 		return kstrtoll(s, base, (long long *)res);
340 	else
341 		return _kstrtol(s, base, res);
342 }
343 
344 int __must_check kstrtouint(const char *s, unsigned int base, unsigned int *res);
345 int __must_check kstrtoint(const char *s, unsigned int base, int *res);
346 
347 static inline int __must_check kstrtou64(const char *s, unsigned int base, u64 *res)
348 {
349 	return kstrtoull(s, base, res);
350 }
351 
352 static inline int __must_check kstrtos64(const char *s, unsigned int base, s64 *res)
353 {
354 	return kstrtoll(s, base, res);
355 }
356 
357 static inline int __must_check kstrtou32(const char *s, unsigned int base, u32 *res)
358 {
359 	return kstrtouint(s, base, res);
360 }
361 
362 static inline int __must_check kstrtos32(const char *s, unsigned int base, s32 *res)
363 {
364 	return kstrtoint(s, base, res);
365 }
366 
367 int __must_check kstrtou16(const char *s, unsigned int base, u16 *res);
368 int __must_check kstrtos16(const char *s, unsigned int base, s16 *res);
369 int __must_check kstrtou8(const char *s, unsigned int base, u8 *res);
370 int __must_check kstrtos8(const char *s, unsigned int base, s8 *res);
371 int __must_check kstrtobool(const char *s, bool *res);
372 
373 int __must_check kstrtoull_from_user(const char __user *s, size_t count, unsigned int base, unsigned long long *res);
374 int __must_check kstrtoll_from_user(const char __user *s, size_t count, unsigned int base, long long *res);
375 int __must_check kstrtoul_from_user(const char __user *s, size_t count, unsigned int base, unsigned long *res);
376 int __must_check kstrtol_from_user(const char __user *s, size_t count, unsigned int base, long *res);
377 int __must_check kstrtouint_from_user(const char __user *s, size_t count, unsigned int base, unsigned int *res);
378 int __must_check kstrtoint_from_user(const char __user *s, size_t count, unsigned int base, int *res);
379 int __must_check kstrtou16_from_user(const char __user *s, size_t count, unsigned int base, u16 *res);
380 int __must_check kstrtos16_from_user(const char __user *s, size_t count, unsigned int base, s16 *res);
381 int __must_check kstrtou8_from_user(const char __user *s, size_t count, unsigned int base, u8 *res);
382 int __must_check kstrtos8_from_user(const char __user *s, size_t count, unsigned int base, s8 *res);
383 int __must_check kstrtobool_from_user(const char __user *s, size_t count, bool *res);
384 
385 static inline int __must_check kstrtou64_from_user(const char __user *s, size_t count, unsigned int base, u64 *res)
386 {
387 	return kstrtoull_from_user(s, count, base, res);
388 }
389 
390 static inline int __must_check kstrtos64_from_user(const char __user *s, size_t count, unsigned int base, s64 *res)
391 {
392 	return kstrtoll_from_user(s, count, base, res);
393 }
394 
395 static inline int __must_check kstrtou32_from_user(const char __user *s, size_t count, unsigned int base, u32 *res)
396 {
397 	return kstrtouint_from_user(s, count, base, res);
398 }
399 
400 static inline int __must_check kstrtos32_from_user(const char __user *s, size_t count, unsigned int base, s32 *res)
401 {
402 	return kstrtoint_from_user(s, count, base, res);
403 }
404 
405 /* Obsolete, do not use.  Use kstrto<foo> instead */
406 
407 extern unsigned long simple_strtoul(const char *,char **,unsigned int);
408 extern long simple_strtol(const char *,char **,unsigned int);
409 extern unsigned long long simple_strtoull(const char *,char **,unsigned int);
410 extern long long simple_strtoll(const char *,char **,unsigned int);
411 
412 extern int num_to_str(char *buf, int size, unsigned long long num);
413 
414 /* lib/printf utilities */
415 
416 extern __printf(2, 3) int sprintf(char *buf, const char * fmt, ...);
417 extern __printf(2, 0) int vsprintf(char *buf, const char *, va_list);
418 extern __printf(3, 4)
419 int snprintf(char *buf, size_t size, const char *fmt, ...);
420 extern __printf(3, 0)
421 int vsnprintf(char *buf, size_t size, const char *fmt, va_list args);
422 extern __printf(3, 4)
423 int scnprintf(char *buf, size_t size, const char *fmt, ...);
424 extern __printf(3, 0)
425 int vscnprintf(char *buf, size_t size, const char *fmt, va_list args);
426 extern __printf(2, 3) __malloc
427 char *kasprintf(gfp_t gfp, const char *fmt, ...);
428 extern __printf(2, 0) __malloc
429 char *kvasprintf(gfp_t gfp, const char *fmt, va_list args);
430 extern __printf(2, 0)
431 const char *kvasprintf_const(gfp_t gfp, const char *fmt, va_list args);
432 
433 extern __scanf(2, 3)
434 int sscanf(const char *, const char *, ...);
435 extern __scanf(2, 0)
436 int vsscanf(const char *, const char *, va_list);
437 
438 extern int get_option(char **str, int *pint);
439 extern char *get_options(const char *str, int nints, int *ints);
440 extern unsigned long long memparse(const char *ptr, char **retptr);
441 extern bool parse_option_str(const char *str, const char *option);
442 extern char *next_arg(char *args, char **param, char **val);
443 
444 extern int core_kernel_text(unsigned long addr);
445 extern int core_kernel_data(unsigned long addr);
446 extern int __kernel_text_address(unsigned long addr);
447 extern int kernel_text_address(unsigned long addr);
448 extern int func_ptr_is_kernel_text(void *ptr);
449 
450 unsigned long int_sqrt(unsigned long);
451 
452 extern void bust_spinlocks(int yes);
453 extern int oops_in_progress;		/* If set, an oops, panic(), BUG() or die() is in progress */
454 extern int panic_timeout;
455 extern int panic_on_oops;
456 extern int panic_on_unrecovered_nmi;
457 extern int panic_on_io_nmi;
458 extern int panic_on_warn;
459 extern int sysctl_panic_on_rcu_stall;
460 extern int sysctl_panic_on_stackoverflow;
461 
462 extern bool crash_kexec_post_notifiers;
463 
464 /*
465  * panic_cpu is used for synchronizing panic() and crash_kexec() execution. It
466  * holds a CPU number which is executing panic() currently. A value of
467  * PANIC_CPU_INVALID means no CPU has entered panic() or crash_kexec().
468  */
469 extern atomic_t panic_cpu;
470 #define PANIC_CPU_INVALID	-1
471 
472 /*
473  * Only to be used by arch init code. If the user over-wrote the default
474  * CONFIG_PANIC_TIMEOUT, honor it.
475  */
476 static inline void set_arch_panic_timeout(int timeout, int arch_default_timeout)
477 {
478 	if (panic_timeout == arch_default_timeout)
479 		panic_timeout = timeout;
480 }
481 extern const char *print_tainted(void);
482 enum lockdep_ok {
483 	LOCKDEP_STILL_OK,
484 	LOCKDEP_NOW_UNRELIABLE
485 };
486 extern void add_taint(unsigned flag, enum lockdep_ok);
487 extern int test_taint(unsigned flag);
488 extern unsigned long get_taint(void);
489 extern int root_mountflags;
490 
491 extern bool early_boot_irqs_disabled;
492 
493 /* Values used for system_state */
494 extern enum system_states {
495 	SYSTEM_BOOTING,
496 	SYSTEM_RUNNING,
497 	SYSTEM_HALT,
498 	SYSTEM_POWER_OFF,
499 	SYSTEM_RESTART,
500 } system_state;
501 
502 #define TAINT_PROPRIETARY_MODULE	0
503 #define TAINT_FORCED_MODULE		1
504 #define TAINT_CPU_OUT_OF_SPEC		2
505 #define TAINT_FORCED_RMMOD		3
506 #define TAINT_MACHINE_CHECK		4
507 #define TAINT_BAD_PAGE			5
508 #define TAINT_USER			6
509 #define TAINT_DIE			7
510 #define TAINT_OVERRIDDEN_ACPI_TABLE	8
511 #define TAINT_WARN			9
512 #define TAINT_CRAP			10
513 #define TAINT_FIRMWARE_WORKAROUND	11
514 #define TAINT_OOT_MODULE		12
515 #define TAINT_UNSIGNED_MODULE		13
516 #define TAINT_SOFTLOCKUP		14
517 #define TAINT_LIVEPATCH			15
518 #define TAINT_FLAGS_COUNT		16
519 
520 struct taint_flag {
521 	char c_true;	/* character printed when tainted */
522 	char c_false;	/* character printed when not tainted */
523 	bool module;	/* also show as a per-module taint flag */
524 };
525 
526 extern const struct taint_flag taint_flags[TAINT_FLAGS_COUNT];
527 
528 extern const char hex_asc[];
529 #define hex_asc_lo(x)	hex_asc[((x) & 0x0f)]
530 #define hex_asc_hi(x)	hex_asc[((x) & 0xf0) >> 4]
531 
532 static inline char *hex_byte_pack(char *buf, u8 byte)
533 {
534 	*buf++ = hex_asc_hi(byte);
535 	*buf++ = hex_asc_lo(byte);
536 	return buf;
537 }
538 
539 extern const char hex_asc_upper[];
540 #define hex_asc_upper_lo(x)	hex_asc_upper[((x) & 0x0f)]
541 #define hex_asc_upper_hi(x)	hex_asc_upper[((x) & 0xf0) >> 4]
542 
543 static inline char *hex_byte_pack_upper(char *buf, u8 byte)
544 {
545 	*buf++ = hex_asc_upper_hi(byte);
546 	*buf++ = hex_asc_upper_lo(byte);
547 	return buf;
548 }
549 
550 extern int hex_to_bin(char ch);
551 extern int __must_check hex2bin(u8 *dst, const char *src, size_t count);
552 extern char *bin2hex(char *dst, const void *src, size_t count);
553 
554 bool mac_pton(const char *s, u8 *mac);
555 
556 /*
557  * General tracing related utility functions - trace_printk(),
558  * tracing_on/tracing_off and tracing_start()/tracing_stop
559  *
560  * Use tracing_on/tracing_off when you want to quickly turn on or off
561  * tracing. It simply enables or disables the recording of the trace events.
562  * This also corresponds to the user space /sys/kernel/debug/tracing/tracing_on
563  * file, which gives a means for the kernel and userspace to interact.
564  * Place a tracing_off() in the kernel where you want tracing to end.
565  * From user space, examine the trace, and then echo 1 > tracing_on
566  * to continue tracing.
567  *
568  * tracing_stop/tracing_start has slightly more overhead. It is used
569  * by things like suspend to ram where disabling the recording of the
570  * trace is not enough, but tracing must actually stop because things
571  * like calling smp_processor_id() may crash the system.
572  *
573  * Most likely, you want to use tracing_on/tracing_off.
574  */
575 
576 enum ftrace_dump_mode {
577 	DUMP_NONE,
578 	DUMP_ALL,
579 	DUMP_ORIG,
580 };
581 
582 #ifdef CONFIG_TRACING
583 void tracing_on(void);
584 void tracing_off(void);
585 int tracing_is_on(void);
586 void tracing_snapshot(void);
587 void tracing_snapshot_alloc(void);
588 
589 extern void tracing_start(void);
590 extern void tracing_stop(void);
591 
592 static inline __printf(1, 2)
593 void ____trace_printk_check_format(const char *fmt, ...)
594 {
595 }
596 #define __trace_printk_check_format(fmt, args...)			\
597 do {									\
598 	if (0)								\
599 		____trace_printk_check_format(fmt, ##args);		\
600 } while (0)
601 
602 /**
603  * trace_printk - printf formatting in the ftrace buffer
604  * @fmt: the printf format for printing
605  *
606  * Note: __trace_printk is an internal function for trace_printk and
607  *       the @ip is passed in via the trace_printk macro.
608  *
609  * This function allows a kernel developer to debug fast path sections
610  * that printk is not appropriate for. By scattering in various
611  * printk like tracing in the code, a developer can quickly see
612  * where problems are occurring.
613  *
614  * This is intended as a debugging tool for the developer only.
615  * Please refrain from leaving trace_printks scattered around in
616  * your code. (Extra memory is used for special buffers that are
617  * allocated when trace_printk() is used)
618  *
619  * A little optization trick is done here. If there's only one
620  * argument, there's no need to scan the string for printf formats.
621  * The trace_puts() will suffice. But how can we take advantage of
622  * using trace_puts() when trace_printk() has only one argument?
623  * By stringifying the args and checking the size we can tell
624  * whether or not there are args. __stringify((__VA_ARGS__)) will
625  * turn into "()\0" with a size of 3 when there are no args, anything
626  * else will be bigger. All we need to do is define a string to this,
627  * and then take its size and compare to 3. If it's bigger, use
628  * do_trace_printk() otherwise, optimize it to trace_puts(). Then just
629  * let gcc optimize the rest.
630  */
631 
632 #define trace_printk(fmt, ...)				\
633 do {							\
634 	char _______STR[] = __stringify((__VA_ARGS__));	\
635 	if (sizeof(_______STR) > 3)			\
636 		do_trace_printk(fmt, ##__VA_ARGS__);	\
637 	else						\
638 		trace_puts(fmt);			\
639 } while (0)
640 
641 #define do_trace_printk(fmt, args...)					\
642 do {									\
643 	static const char *trace_printk_fmt __used			\
644 		__attribute__((section("__trace_printk_fmt"))) =	\
645 		__builtin_constant_p(fmt) ? fmt : NULL;			\
646 									\
647 	__trace_printk_check_format(fmt, ##args);			\
648 									\
649 	if (__builtin_constant_p(fmt))					\
650 		__trace_bprintk(_THIS_IP_, trace_printk_fmt, ##args);	\
651 	else								\
652 		__trace_printk(_THIS_IP_, fmt, ##args);			\
653 } while (0)
654 
655 extern __printf(2, 3)
656 int __trace_bprintk(unsigned long ip, const char *fmt, ...);
657 
658 extern __printf(2, 3)
659 int __trace_printk(unsigned long ip, const char *fmt, ...);
660 
661 /**
662  * trace_puts - write a string into the ftrace buffer
663  * @str: the string to record
664  *
665  * Note: __trace_bputs is an internal function for trace_puts and
666  *       the @ip is passed in via the trace_puts macro.
667  *
668  * This is similar to trace_printk() but is made for those really fast
669  * paths that a developer wants the least amount of "Heisenbug" affects,
670  * where the processing of the print format is still too much.
671  *
672  * This function allows a kernel developer to debug fast path sections
673  * that printk is not appropriate for. By scattering in various
674  * printk like tracing in the code, a developer can quickly see
675  * where problems are occurring.
676  *
677  * This is intended as a debugging tool for the developer only.
678  * Please refrain from leaving trace_puts scattered around in
679  * your code. (Extra memory is used for special buffers that are
680  * allocated when trace_puts() is used)
681  *
682  * Returns: 0 if nothing was written, positive # if string was.
683  *  (1 when __trace_bputs is used, strlen(str) when __trace_puts is used)
684  */
685 
686 #define trace_puts(str) ({						\
687 	static const char *trace_printk_fmt __used			\
688 		__attribute__((section("__trace_printk_fmt"))) =	\
689 		__builtin_constant_p(str) ? str : NULL;			\
690 									\
691 	if (__builtin_constant_p(str))					\
692 		__trace_bputs(_THIS_IP_, trace_printk_fmt);		\
693 	else								\
694 		__trace_puts(_THIS_IP_, str, strlen(str));		\
695 })
696 extern int __trace_bputs(unsigned long ip, const char *str);
697 extern int __trace_puts(unsigned long ip, const char *str, int size);
698 
699 extern void trace_dump_stack(int skip);
700 
701 /*
702  * The double __builtin_constant_p is because gcc will give us an error
703  * if we try to allocate the static variable to fmt if it is not a
704  * constant. Even with the outer if statement.
705  */
706 #define ftrace_vprintk(fmt, vargs)					\
707 do {									\
708 	if (__builtin_constant_p(fmt)) {				\
709 		static const char *trace_printk_fmt __used		\
710 		  __attribute__((section("__trace_printk_fmt"))) =	\
711 			__builtin_constant_p(fmt) ? fmt : NULL;		\
712 									\
713 		__ftrace_vbprintk(_THIS_IP_, trace_printk_fmt, vargs);	\
714 	} else								\
715 		__ftrace_vprintk(_THIS_IP_, fmt, vargs);		\
716 } while (0)
717 
718 extern __printf(2, 0) int
719 __ftrace_vbprintk(unsigned long ip, const char *fmt, va_list ap);
720 
721 extern __printf(2, 0) int
722 __ftrace_vprintk(unsigned long ip, const char *fmt, va_list ap);
723 
724 extern void ftrace_dump(enum ftrace_dump_mode oops_dump_mode);
725 #else
726 static inline void tracing_start(void) { }
727 static inline void tracing_stop(void) { }
728 static inline void trace_dump_stack(int skip) { }
729 
730 static inline void tracing_on(void) { }
731 static inline void tracing_off(void) { }
732 static inline int tracing_is_on(void) { return 0; }
733 static inline void tracing_snapshot(void) { }
734 static inline void tracing_snapshot_alloc(void) { }
735 
736 static inline __printf(1, 2)
737 int trace_printk(const char *fmt, ...)
738 {
739 	return 0;
740 }
741 static __printf(1, 0) inline int
742 ftrace_vprintk(const char *fmt, va_list ap)
743 {
744 	return 0;
745 }
746 static inline void ftrace_dump(enum ftrace_dump_mode oops_dump_mode) { }
747 #endif /* CONFIG_TRACING */
748 
749 /*
750  * min()/max()/clamp() macros that also do
751  * strict type-checking.. See the
752  * "unnecessary" pointer comparison.
753  */
754 #define __min(t1, t2, min1, min2, x, y) ({		\
755 	t1 min1 = (x);					\
756 	t2 min2 = (y);					\
757 	(void) (&min1 == &min2);			\
758 	min1 < min2 ? min1 : min2; })
759 #define min(x, y)					\
760 	__min(typeof(x), typeof(y),			\
761 	      __UNIQUE_ID(min1_), __UNIQUE_ID(min2_),	\
762 	      x, y)
763 
764 #define __max(t1, t2, max1, max2, x, y) ({		\
765 	t1 max1 = (x);					\
766 	t2 max2 = (y);					\
767 	(void) (&max1 == &max2);			\
768 	max1 > max2 ? max1 : max2; })
769 #define max(x, y)					\
770 	__max(typeof(x), typeof(y),			\
771 	      __UNIQUE_ID(max1_), __UNIQUE_ID(max2_),	\
772 	      x, y)
773 
774 #define min3(x, y, z) min((typeof(x))min(x, y), z)
775 #define max3(x, y, z) max((typeof(x))max(x, y), z)
776 
777 /**
778  * min_not_zero - return the minimum that is _not_ zero, unless both are zero
779  * @x: value1
780  * @y: value2
781  */
782 #define min_not_zero(x, y) ({			\
783 	typeof(x) __x = (x);			\
784 	typeof(y) __y = (y);			\
785 	__x == 0 ? __y : ((__y == 0) ? __x : min(__x, __y)); })
786 
787 /**
788  * clamp - return a value clamped to a given range with strict typechecking
789  * @val: current value
790  * @lo: lowest allowable value
791  * @hi: highest allowable value
792  *
793  * This macro does strict typechecking of lo/hi to make sure they are of the
794  * same type as val.  See the unnecessary pointer comparisons.
795  */
796 #define clamp(val, lo, hi) min((typeof(val))max(val, lo), hi)
797 
798 /*
799  * ..and if you can't take the strict
800  * types, you can specify one yourself.
801  *
802  * Or not use min/max/clamp at all, of course.
803  */
804 #define min_t(type, x, y)				\
805 	__min(type, type,				\
806 	      __UNIQUE_ID(min1_), __UNIQUE_ID(min2_),	\
807 	      x, y)
808 
809 #define max_t(type, x, y)				\
810 	__max(type, type,				\
811 	      __UNIQUE_ID(min1_), __UNIQUE_ID(min2_),	\
812 	      x, y)
813 
814 /**
815  * clamp_t - return a value clamped to a given range using a given type
816  * @type: the type of variable to use
817  * @val: current value
818  * @lo: minimum allowable value
819  * @hi: maximum allowable value
820  *
821  * This macro does no typechecking and uses temporary variables of type
822  * 'type' to make all the comparisons.
823  */
824 #define clamp_t(type, val, lo, hi) min_t(type, max_t(type, val, lo), hi)
825 
826 /**
827  * clamp_val - return a value clamped to a given range using val's type
828  * @val: current value
829  * @lo: minimum allowable value
830  * @hi: maximum allowable value
831  *
832  * This macro does no typechecking and uses temporary variables of whatever
833  * type the input argument 'val' is.  This is useful when val is an unsigned
834  * type and min and max are literals that will otherwise be assigned a signed
835  * integer type.
836  */
837 #define clamp_val(val, lo, hi) clamp_t(typeof(val), val, lo, hi)
838 
839 
840 /*
841  * swap - swap value of @a and @b
842  */
843 #define swap(a, b) \
844 	do { typeof(a) __tmp = (a); (a) = (b); (b) = __tmp; } while (0)
845 
846 /**
847  * container_of - cast a member of a structure out to the containing structure
848  * @ptr:	the pointer to the member.
849  * @type:	the type of the container struct this is embedded in.
850  * @member:	the name of the member within the struct.
851  *
852  */
853 #define container_of(ptr, type, member) ({			\
854 	const typeof( ((type *)0)->member ) *__mptr = (ptr);	\
855 	(type *)( (char *)__mptr - offsetof(type,member) );})
856 
857 /* Rebuild everything on CONFIG_FTRACE_MCOUNT_RECORD */
858 #ifdef CONFIG_FTRACE_MCOUNT_RECORD
859 # define REBUILD_DUE_TO_FTRACE_MCOUNT_RECORD
860 #endif
861 
862 /* Permissions on a sysfs file: you didn't miss the 0 prefix did you? */
863 #define VERIFY_OCTAL_PERMISSIONS(perms)						\
864 	(BUILD_BUG_ON_ZERO((perms) < 0) +					\
865 	 BUILD_BUG_ON_ZERO((perms) > 0777) +					\
866 	 /* USER_READABLE >= GROUP_READABLE >= OTHER_READABLE */		\
867 	 BUILD_BUG_ON_ZERO((((perms) >> 6) & 4) < (((perms) >> 3) & 4)) +	\
868 	 BUILD_BUG_ON_ZERO((((perms) >> 3) & 4) < ((perms) & 4)) +		\
869 	 /* USER_WRITABLE >= GROUP_WRITABLE */					\
870 	 BUILD_BUG_ON_ZERO((((perms) >> 6) & 2) < (((perms) >> 3) & 2)) +	\
871 	 /* OTHER_WRITABLE?  Generally considered a bad idea. */		\
872 	 BUILD_BUG_ON_ZERO((perms) & 2) +					\
873 	 (perms))
874 #endif
875