xref: /linux-6.15/include/linux/kernel.h (revision 6ae0a628)
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_MASK(x, mask)	__ALIGN_KERNEL_MASK((x), (mask))
51 #define PTR_ALIGN(p, a)		((typeof(p))ALIGN((unsigned long)(p), (a)))
52 #define IS_ALIGNED(x, a)		(((x) & ((typeof(x))(a) - 1)) == 0)
53 
54 /* generic data direction definitions */
55 #define READ			0
56 #define WRITE			1
57 
58 #define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]) + __must_be_array(arr))
59 
60 #define u64_to_user_ptr(x) (		\
61 {					\
62 	typecheck(u64, x);		\
63 	(void __user *)(uintptr_t)x;	\
64 }					\
65 )
66 
67 /*
68  * This looks more complex than it should be. But we need to
69  * get the type for the ~ right in round_down (it needs to be
70  * as wide as the result!), and we want to evaluate the macro
71  * arguments just once each.
72  */
73 #define __round_mask(x, y) ((__typeof__(x))((y)-1))
74 #define round_up(x, y) ((((x)-1) | __round_mask(x, y))+1)
75 #define round_down(x, y) ((x) & ~__round_mask(x, y))
76 
77 #define FIELD_SIZEOF(t, f) (sizeof(((t*)0)->f))
78 #define DIV_ROUND_UP __KERNEL_DIV_ROUND_UP
79 #define DIV_ROUND_UP_ULL(ll,d) \
80 	({ unsigned long long _tmp = (ll)+(d)-1; do_div(_tmp, d); _tmp; })
81 
82 #if BITS_PER_LONG == 32
83 # define DIV_ROUND_UP_SECTOR_T(ll,d) DIV_ROUND_UP_ULL(ll, d)
84 #else
85 # define DIV_ROUND_UP_SECTOR_T(ll,d) DIV_ROUND_UP(ll,d)
86 #endif
87 
88 /* The `const' in roundup() prevents gcc-3.3 from calling __divdi3 */
89 #define roundup(x, y) (					\
90 {							\
91 	const typeof(y) __y = y;			\
92 	(((x) + (__y - 1)) / __y) * __y;		\
93 }							\
94 )
95 #define rounddown(x, y) (				\
96 {							\
97 	typeof(x) __x = (x);				\
98 	__x - (__x % (y));				\
99 }							\
100 )
101 
102 /*
103  * Divide positive or negative dividend by positive divisor and round
104  * to closest integer. Result is undefined for negative divisors and
105  * for negative dividends if the divisor variable type is unsigned.
106  */
107 #define DIV_ROUND_CLOSEST(x, divisor)(			\
108 {							\
109 	typeof(x) __x = x;				\
110 	typeof(divisor) __d = divisor;			\
111 	(((typeof(x))-1) > 0 ||				\
112 	 ((typeof(divisor))-1) > 0 || (__x) > 0) ?	\
113 		(((__x) + ((__d) / 2)) / (__d)) :	\
114 		(((__x) - ((__d) / 2)) / (__d));	\
115 }							\
116 )
117 /*
118  * Same as above but for u64 dividends. divisor must be a 32-bit
119  * number.
120  */
121 #define DIV_ROUND_CLOSEST_ULL(x, divisor)(		\
122 {							\
123 	typeof(divisor) __d = divisor;			\
124 	unsigned long long _tmp = (x) + (__d) / 2;	\
125 	do_div(_tmp, __d);				\
126 	_tmp;						\
127 }							\
128 )
129 
130 /*
131  * Multiplies an integer by a fraction, while avoiding unnecessary
132  * overflow or loss of precision.
133  */
134 #define mult_frac(x, numer, denom)(			\
135 {							\
136 	typeof(x) quot = (x) / (denom);			\
137 	typeof(x) rem  = (x) % (denom);			\
138 	(quot * (numer)) + ((rem * (numer)) / (denom));	\
139 }							\
140 )
141 
142 
143 #define _RET_IP_		(unsigned long)__builtin_return_address(0)
144 #define _THIS_IP_  ({ __label__ __here; __here: (unsigned long)&&__here; })
145 
146 #ifdef CONFIG_LBDAF
147 # include <asm/div64.h>
148 # define sector_div(a, b) do_div(a, b)
149 #else
150 # define sector_div(n, b)( \
151 { \
152 	int _res; \
153 	_res = (n) % (b); \
154 	(n) /= (b); \
155 	_res; \
156 } \
157 )
158 #endif
159 
160 /**
161  * upper_32_bits - return bits 32-63 of a number
162  * @n: the number we're accessing
163  *
164  * A basic shift-right of a 64- or 32-bit quantity.  Use this to suppress
165  * the "right shift count >= width of type" warning when that quantity is
166  * 32-bits.
167  */
168 #define upper_32_bits(n) ((u32)(((n) >> 16) >> 16))
169 
170 /**
171  * lower_32_bits - return bits 0-31 of a number
172  * @n: the number we're accessing
173  */
174 #define lower_32_bits(n) ((u32)(n))
175 
176 struct completion;
177 struct pt_regs;
178 struct user;
179 
180 #ifdef CONFIG_PREEMPT_VOLUNTARY
181 extern int _cond_resched(void);
182 # define might_resched() _cond_resched()
183 #else
184 # define might_resched() do { } while (0)
185 #endif
186 
187 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
188   void ___might_sleep(const char *file, int line, int preempt_offset);
189   void __might_sleep(const char *file, int line, int preempt_offset);
190 /**
191  * might_sleep - annotation for functions that can sleep
192  *
193  * this macro will print a stack trace if it is executed in an atomic
194  * context (spinlock, irq-handler, ...).
195  *
196  * This is a useful debugging help to be able to catch problems early and not
197  * be bitten later when the calling function happens to sleep when it is not
198  * supposed to.
199  */
200 # define might_sleep() \
201 	do { __might_sleep(__FILE__, __LINE__, 0); might_resched(); } while (0)
202 # define sched_annotate_sleep()	(current->task_state_change = 0)
203 #else
204   static inline void ___might_sleep(const char *file, int line,
205 				   int preempt_offset) { }
206   static inline void __might_sleep(const char *file, int line,
207 				   int preempt_offset) { }
208 # define might_sleep() do { might_resched(); } while (0)
209 # define sched_annotate_sleep() do { } while (0)
210 #endif
211 
212 #define might_sleep_if(cond) do { if (cond) might_sleep(); } while (0)
213 
214 /**
215  * abs - return absolute value of an argument
216  * @x: the value.  If it is unsigned type, it is converted to signed type first.
217  *     char is treated as if it was signed (regardless of whether it really is)
218  *     but the macro's return type is preserved as char.
219  *
220  * Return: an absolute value of x.
221  */
222 #define abs(x)	__abs_choose_expr(x, long long,				\
223 		__abs_choose_expr(x, long,				\
224 		__abs_choose_expr(x, int,				\
225 		__abs_choose_expr(x, short,				\
226 		__abs_choose_expr(x, char,				\
227 		__builtin_choose_expr(					\
228 			__builtin_types_compatible_p(typeof(x), char),	\
229 			(char)({ signed char __x = (x); __x<0?-__x:__x; }), \
230 			((void)0)))))))
231 
232 #define __abs_choose_expr(x, type, other) __builtin_choose_expr(	\
233 	__builtin_types_compatible_p(typeof(x),   signed type) ||	\
234 	__builtin_types_compatible_p(typeof(x), unsigned type),		\
235 	({ signed type __x = (x); __x < 0 ? -__x : __x; }), other)
236 
237 /**
238  * reciprocal_scale - "scale" a value into range [0, ep_ro)
239  * @val: value
240  * @ep_ro: right open interval endpoint
241  *
242  * Perform a "reciprocal multiplication" in order to "scale" a value into
243  * range [0, ep_ro), where the upper interval endpoint is right-open.
244  * This is useful, e.g. for accessing a index of an array containing
245  * ep_ro elements, for example. Think of it as sort of modulus, only that
246  * the result isn't that of modulo. ;) Note that if initial input is a
247  * small value, then result will return 0.
248  *
249  * Return: a result based on val in interval [0, ep_ro).
250  */
251 static inline u32 reciprocal_scale(u32 val, u32 ep_ro)
252 {
253 	return (u32)(((u64) val * ep_ro) >> 32);
254 }
255 
256 #if defined(CONFIG_MMU) && \
257 	(defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_DEBUG_ATOMIC_SLEEP))
258 #define might_fault() __might_fault(__FILE__, __LINE__)
259 void __might_fault(const char *file, int line);
260 #else
261 static inline void might_fault(void) { }
262 #endif
263 
264 extern struct atomic_notifier_head panic_notifier_list;
265 extern long (*panic_blink)(int state);
266 __printf(1, 2)
267 void panic(const char *fmt, ...) __noreturn __cold;
268 void nmi_panic(struct pt_regs *regs, const char *msg);
269 extern void oops_enter(void);
270 extern void oops_exit(void);
271 void print_oops_end_marker(void);
272 extern int oops_may_print(void);
273 void do_exit(long error_code) __noreturn;
274 void complete_and_exit(struct completion *, long) __noreturn;
275 
276 /* Internal, do not use. */
277 int __must_check _kstrtoul(const char *s, unsigned int base, unsigned long *res);
278 int __must_check _kstrtol(const char *s, unsigned int base, long *res);
279 
280 int __must_check kstrtoull(const char *s, unsigned int base, unsigned long long *res);
281 int __must_check kstrtoll(const char *s, unsigned int base, long long *res);
282 
283 /**
284  * kstrtoul - convert a string to an unsigned long
285  * @s: The start of the string. The string must be null-terminated, and may also
286  *  include a single newline before its terminating null. The first character
287  *  may also be a plus sign, but not a minus sign.
288  * @base: The number base to use. The maximum supported base is 16. If base is
289  *  given as 0, then the base of the string is automatically detected with the
290  *  conventional semantics - If it begins with 0x the number will be parsed as a
291  *  hexadecimal (case insensitive), if it otherwise begins with 0, it will be
292  *  parsed as an octal number. Otherwise it will be parsed as a decimal.
293  * @res: Where to write the result of the conversion on success.
294  *
295  * Returns 0 on success, -ERANGE on overflow and -EINVAL on parsing error.
296  * Used as a replacement for the obsolete simple_strtoull. Return code must
297  * be checked.
298 */
299 static inline int __must_check kstrtoul(const char *s, unsigned int base, unsigned long *res)
300 {
301 	/*
302 	 * We want to shortcut function call, but
303 	 * __builtin_types_compatible_p(unsigned long, unsigned long long) = 0.
304 	 */
305 	if (sizeof(unsigned long) == sizeof(unsigned long long) &&
306 	    __alignof__(unsigned long) == __alignof__(unsigned long long))
307 		return kstrtoull(s, base, (unsigned long long *)res);
308 	else
309 		return _kstrtoul(s, base, res);
310 }
311 
312 /**
313  * kstrtol - convert a string to a long
314  * @s: The start of the string. The string must be null-terminated, and may also
315  *  include a single newline before its terminating null. The first character
316  *  may also be a plus sign or a minus sign.
317  * @base: The number base to use. The maximum supported base is 16. If base is
318  *  given as 0, then the base of the string is automatically detected with the
319  *  conventional semantics - If it begins with 0x the number will be parsed as a
320  *  hexadecimal (case insensitive), if it otherwise begins with 0, it will be
321  *  parsed as an octal number. Otherwise it will be parsed as a decimal.
322  * @res: Where to write the result of the conversion on success.
323  *
324  * Returns 0 on success, -ERANGE on overflow and -EINVAL on parsing error.
325  * Used as a replacement for the obsolete simple_strtoull. Return code must
326  * be checked.
327  */
328 static inline int __must_check kstrtol(const char *s, unsigned int base, long *res)
329 {
330 	/*
331 	 * We want to shortcut function call, but
332 	 * __builtin_types_compatible_p(long, long long) = 0.
333 	 */
334 	if (sizeof(long) == sizeof(long long) &&
335 	    __alignof__(long) == __alignof__(long long))
336 		return kstrtoll(s, base, (long long *)res);
337 	else
338 		return _kstrtol(s, base, res);
339 }
340 
341 int __must_check kstrtouint(const char *s, unsigned int base, unsigned int *res);
342 int __must_check kstrtoint(const char *s, unsigned int base, int *res);
343 
344 static inline int __must_check kstrtou64(const char *s, unsigned int base, u64 *res)
345 {
346 	return kstrtoull(s, base, res);
347 }
348 
349 static inline int __must_check kstrtos64(const char *s, unsigned int base, s64 *res)
350 {
351 	return kstrtoll(s, base, res);
352 }
353 
354 static inline int __must_check kstrtou32(const char *s, unsigned int base, u32 *res)
355 {
356 	return kstrtouint(s, base, res);
357 }
358 
359 static inline int __must_check kstrtos32(const char *s, unsigned int base, s32 *res)
360 {
361 	return kstrtoint(s, base, res);
362 }
363 
364 int __must_check kstrtou16(const char *s, unsigned int base, u16 *res);
365 int __must_check kstrtos16(const char *s, unsigned int base, s16 *res);
366 int __must_check kstrtou8(const char *s, unsigned int base, u8 *res);
367 int __must_check kstrtos8(const char *s, unsigned int base, s8 *res);
368 int __must_check kstrtobool(const char *s, bool *res);
369 
370 int __must_check kstrtoull_from_user(const char __user *s, size_t count, unsigned int base, unsigned long long *res);
371 int __must_check kstrtoll_from_user(const char __user *s, size_t count, unsigned int base, long long *res);
372 int __must_check kstrtoul_from_user(const char __user *s, size_t count, unsigned int base, unsigned long *res);
373 int __must_check kstrtol_from_user(const char __user *s, size_t count, unsigned int base, long *res);
374 int __must_check kstrtouint_from_user(const char __user *s, size_t count, unsigned int base, unsigned int *res);
375 int __must_check kstrtoint_from_user(const char __user *s, size_t count, unsigned int base, int *res);
376 int __must_check kstrtou16_from_user(const char __user *s, size_t count, unsigned int base, u16 *res);
377 int __must_check kstrtos16_from_user(const char __user *s, size_t count, unsigned int base, s16 *res);
378 int __must_check kstrtou8_from_user(const char __user *s, size_t count, unsigned int base, u8 *res);
379 int __must_check kstrtos8_from_user(const char __user *s, size_t count, unsigned int base, s8 *res);
380 int __must_check kstrtobool_from_user(const char __user *s, size_t count, bool *res);
381 
382 static inline int __must_check kstrtou64_from_user(const char __user *s, size_t count, unsigned int base, u64 *res)
383 {
384 	return kstrtoull_from_user(s, count, base, res);
385 }
386 
387 static inline int __must_check kstrtos64_from_user(const char __user *s, size_t count, unsigned int base, s64 *res)
388 {
389 	return kstrtoll_from_user(s, count, base, res);
390 }
391 
392 static inline int __must_check kstrtou32_from_user(const char __user *s, size_t count, unsigned int base, u32 *res)
393 {
394 	return kstrtouint_from_user(s, count, base, res);
395 }
396 
397 static inline int __must_check kstrtos32_from_user(const char __user *s, size_t count, unsigned int base, s32 *res)
398 {
399 	return kstrtoint_from_user(s, count, base, res);
400 }
401 
402 /* Obsolete, do not use.  Use kstrto<foo> instead */
403 
404 extern unsigned long simple_strtoul(const char *,char **,unsigned int);
405 extern long simple_strtol(const char *,char **,unsigned int);
406 extern unsigned long long simple_strtoull(const char *,char **,unsigned int);
407 extern long long simple_strtoll(const char *,char **,unsigned int);
408 
409 extern int num_to_str(char *buf, int size, unsigned long long num);
410 
411 /* lib/printf utilities */
412 
413 extern __printf(2, 3) int sprintf(char *buf, const char * fmt, ...);
414 extern __printf(2, 0) int vsprintf(char *buf, const char *, va_list);
415 extern __printf(3, 4)
416 int snprintf(char *buf, size_t size, const char *fmt, ...);
417 extern __printf(3, 0)
418 int vsnprintf(char *buf, size_t size, const char *fmt, va_list args);
419 extern __printf(3, 4)
420 int scnprintf(char *buf, size_t size, const char *fmt, ...);
421 extern __printf(3, 0)
422 int vscnprintf(char *buf, size_t size, const char *fmt, va_list args);
423 extern __printf(2, 3) __malloc
424 char *kasprintf(gfp_t gfp, const char *fmt, ...);
425 extern __printf(2, 0) __malloc
426 char *kvasprintf(gfp_t gfp, const char *fmt, va_list args);
427 extern __printf(2, 0)
428 const char *kvasprintf_const(gfp_t gfp, const char *fmt, va_list args);
429 
430 extern __scanf(2, 3)
431 int sscanf(const char *, const char *, ...);
432 extern __scanf(2, 0)
433 int vsscanf(const char *, const char *, va_list);
434 
435 extern int get_option(char **str, int *pint);
436 extern char *get_options(const char *str, int nints, int *ints);
437 extern unsigned long long memparse(const char *ptr, char **retptr);
438 extern bool parse_option_str(const char *str, const char *option);
439 
440 extern int core_kernel_text(unsigned long addr);
441 extern int core_kernel_data(unsigned long addr);
442 extern int __kernel_text_address(unsigned long addr);
443 extern int kernel_text_address(unsigned long addr);
444 extern int func_ptr_is_kernel_text(void *ptr);
445 
446 unsigned long int_sqrt(unsigned long);
447 
448 extern void bust_spinlocks(int yes);
449 extern int oops_in_progress;		/* If set, an oops, panic(), BUG() or die() is in progress */
450 extern int panic_timeout;
451 extern int panic_on_oops;
452 extern int panic_on_unrecovered_nmi;
453 extern int panic_on_io_nmi;
454 extern int panic_on_warn;
455 extern int sysctl_panic_on_rcu_stall;
456 extern int sysctl_panic_on_stackoverflow;
457 
458 extern bool crash_kexec_post_notifiers;
459 
460 /*
461  * panic_cpu is used for synchronizing panic() and crash_kexec() execution. It
462  * holds a CPU number which is executing panic() currently. A value of
463  * PANIC_CPU_INVALID means no CPU has entered panic() or crash_kexec().
464  */
465 extern atomic_t panic_cpu;
466 #define PANIC_CPU_INVALID	-1
467 
468 /*
469  * Only to be used by arch init code. If the user over-wrote the default
470  * CONFIG_PANIC_TIMEOUT, honor it.
471  */
472 static inline void set_arch_panic_timeout(int timeout, int arch_default_timeout)
473 {
474 	if (panic_timeout == arch_default_timeout)
475 		panic_timeout = timeout;
476 }
477 extern const char *print_tainted(void);
478 enum lockdep_ok {
479 	LOCKDEP_STILL_OK,
480 	LOCKDEP_NOW_UNRELIABLE
481 };
482 extern void add_taint(unsigned flag, enum lockdep_ok);
483 extern int test_taint(unsigned flag);
484 extern unsigned long get_taint(void);
485 extern int root_mountflags;
486 
487 extern bool early_boot_irqs_disabled;
488 
489 /* Values used for system_state */
490 extern enum system_states {
491 	SYSTEM_BOOTING,
492 	SYSTEM_RUNNING,
493 	SYSTEM_HALT,
494 	SYSTEM_POWER_OFF,
495 	SYSTEM_RESTART,
496 } system_state;
497 
498 #define TAINT_PROPRIETARY_MODULE	0
499 #define TAINT_FORCED_MODULE		1
500 #define TAINT_CPU_OUT_OF_SPEC		2
501 #define TAINT_FORCED_RMMOD		3
502 #define TAINT_MACHINE_CHECK		4
503 #define TAINT_BAD_PAGE			5
504 #define TAINT_USER			6
505 #define TAINT_DIE			7
506 #define TAINT_OVERRIDDEN_ACPI_TABLE	8
507 #define TAINT_WARN			9
508 #define TAINT_CRAP			10
509 #define TAINT_FIRMWARE_WORKAROUND	11
510 #define TAINT_OOT_MODULE		12
511 #define TAINT_UNSIGNED_MODULE		13
512 #define TAINT_SOFTLOCKUP		14
513 #define TAINT_LIVEPATCH			15
514 #define TAINT_FLAGS_COUNT		16
515 
516 struct taint_flag {
517 	char true;	/* character printed when tainted */
518 	char false;	/* character printed when not tainted */
519 	bool module;	/* also show as a per-module taint flag */
520 };
521 
522 extern const struct taint_flag taint_flags[TAINT_FLAGS_COUNT];
523 
524 extern const char hex_asc[];
525 #define hex_asc_lo(x)	hex_asc[((x) & 0x0f)]
526 #define hex_asc_hi(x)	hex_asc[((x) & 0xf0) >> 4]
527 
528 static inline char *hex_byte_pack(char *buf, u8 byte)
529 {
530 	*buf++ = hex_asc_hi(byte);
531 	*buf++ = hex_asc_lo(byte);
532 	return buf;
533 }
534 
535 extern const char hex_asc_upper[];
536 #define hex_asc_upper_lo(x)	hex_asc_upper[((x) & 0x0f)]
537 #define hex_asc_upper_hi(x)	hex_asc_upper[((x) & 0xf0) >> 4]
538 
539 static inline char *hex_byte_pack_upper(char *buf, u8 byte)
540 {
541 	*buf++ = hex_asc_upper_hi(byte);
542 	*buf++ = hex_asc_upper_lo(byte);
543 	return buf;
544 }
545 
546 extern int hex_to_bin(char ch);
547 extern int __must_check hex2bin(u8 *dst, const char *src, size_t count);
548 extern char *bin2hex(char *dst, const void *src, size_t count);
549 
550 bool mac_pton(const char *s, u8 *mac);
551 
552 /*
553  * General tracing related utility functions - trace_printk(),
554  * tracing_on/tracing_off and tracing_start()/tracing_stop
555  *
556  * Use tracing_on/tracing_off when you want to quickly turn on or off
557  * tracing. It simply enables or disables the recording of the trace events.
558  * This also corresponds to the user space /sys/kernel/debug/tracing/tracing_on
559  * file, which gives a means for the kernel and userspace to interact.
560  * Place a tracing_off() in the kernel where you want tracing to end.
561  * From user space, examine the trace, and then echo 1 > tracing_on
562  * to continue tracing.
563  *
564  * tracing_stop/tracing_start has slightly more overhead. It is used
565  * by things like suspend to ram where disabling the recording of the
566  * trace is not enough, but tracing must actually stop because things
567  * like calling smp_processor_id() may crash the system.
568  *
569  * Most likely, you want to use tracing_on/tracing_off.
570  */
571 
572 enum ftrace_dump_mode {
573 	DUMP_NONE,
574 	DUMP_ALL,
575 	DUMP_ORIG,
576 };
577 
578 #ifdef CONFIG_TRACING
579 void tracing_on(void);
580 void tracing_off(void);
581 int tracing_is_on(void);
582 void tracing_snapshot(void);
583 void tracing_snapshot_alloc(void);
584 
585 extern void tracing_start(void);
586 extern void tracing_stop(void);
587 
588 static inline __printf(1, 2)
589 void ____trace_printk_check_format(const char *fmt, ...)
590 {
591 }
592 #define __trace_printk_check_format(fmt, args...)			\
593 do {									\
594 	if (0)								\
595 		____trace_printk_check_format(fmt, ##args);		\
596 } while (0)
597 
598 /**
599  * trace_printk - printf formatting in the ftrace buffer
600  * @fmt: the printf format for printing
601  *
602  * Note: __trace_printk is an internal function for trace_printk and
603  *       the @ip is passed in via the trace_printk macro.
604  *
605  * This function allows a kernel developer to debug fast path sections
606  * that printk is not appropriate for. By scattering in various
607  * printk like tracing in the code, a developer can quickly see
608  * where problems are occurring.
609  *
610  * This is intended as a debugging tool for the developer only.
611  * Please refrain from leaving trace_printks scattered around in
612  * your code. (Extra memory is used for special buffers that are
613  * allocated when trace_printk() is used)
614  *
615  * A little optization trick is done here. If there's only one
616  * argument, there's no need to scan the string for printf formats.
617  * The trace_puts() will suffice. But how can we take advantage of
618  * using trace_puts() when trace_printk() has only one argument?
619  * By stringifying the args and checking the size we can tell
620  * whether or not there are args. __stringify((__VA_ARGS__)) will
621  * turn into "()\0" with a size of 3 when there are no args, anything
622  * else will be bigger. All we need to do is define a string to this,
623  * and then take its size and compare to 3. If it's bigger, use
624  * do_trace_printk() otherwise, optimize it to trace_puts(). Then just
625  * let gcc optimize the rest.
626  */
627 
628 #define trace_printk(fmt, ...)				\
629 do {							\
630 	char _______STR[] = __stringify((__VA_ARGS__));	\
631 	if (sizeof(_______STR) > 3)			\
632 		do_trace_printk(fmt, ##__VA_ARGS__);	\
633 	else						\
634 		trace_puts(fmt);			\
635 } while (0)
636 
637 #define do_trace_printk(fmt, args...)					\
638 do {									\
639 	static const char *trace_printk_fmt __used			\
640 		__attribute__((section("__trace_printk_fmt"))) =	\
641 		__builtin_constant_p(fmt) ? fmt : NULL;			\
642 									\
643 	__trace_printk_check_format(fmt, ##args);			\
644 									\
645 	if (__builtin_constant_p(fmt))					\
646 		__trace_bprintk(_THIS_IP_, trace_printk_fmt, ##args);	\
647 	else								\
648 		__trace_printk(_THIS_IP_, fmt, ##args);			\
649 } while (0)
650 
651 extern __printf(2, 3)
652 int __trace_bprintk(unsigned long ip, const char *fmt, ...);
653 
654 extern __printf(2, 3)
655 int __trace_printk(unsigned long ip, const char *fmt, ...);
656 
657 /**
658  * trace_puts - write a string into the ftrace buffer
659  * @str: the string to record
660  *
661  * Note: __trace_bputs is an internal function for trace_puts and
662  *       the @ip is passed in via the trace_puts macro.
663  *
664  * This is similar to trace_printk() but is made for those really fast
665  * paths that a developer wants the least amount of "Heisenbug" affects,
666  * where the processing of the print format is still too much.
667  *
668  * This function allows a kernel developer to debug fast path sections
669  * that printk is not appropriate for. By scattering in various
670  * printk like tracing in the code, a developer can quickly see
671  * where problems are occurring.
672  *
673  * This is intended as a debugging tool for the developer only.
674  * Please refrain from leaving trace_puts scattered around in
675  * your code. (Extra memory is used for special buffers that are
676  * allocated when trace_puts() is used)
677  *
678  * Returns: 0 if nothing was written, positive # if string was.
679  *  (1 when __trace_bputs is used, strlen(str) when __trace_puts is used)
680  */
681 
682 #define trace_puts(str) ({						\
683 	static const char *trace_printk_fmt __used			\
684 		__attribute__((section("__trace_printk_fmt"))) =	\
685 		__builtin_constant_p(str) ? str : NULL;			\
686 									\
687 	if (__builtin_constant_p(str))					\
688 		__trace_bputs(_THIS_IP_, trace_printk_fmt);		\
689 	else								\
690 		__trace_puts(_THIS_IP_, str, strlen(str));		\
691 })
692 extern int __trace_bputs(unsigned long ip, const char *str);
693 extern int __trace_puts(unsigned long ip, const char *str, int size);
694 
695 extern void trace_dump_stack(int skip);
696 
697 /*
698  * The double __builtin_constant_p is because gcc will give us an error
699  * if we try to allocate the static variable to fmt if it is not a
700  * constant. Even with the outer if statement.
701  */
702 #define ftrace_vprintk(fmt, vargs)					\
703 do {									\
704 	if (__builtin_constant_p(fmt)) {				\
705 		static const char *trace_printk_fmt __used		\
706 		  __attribute__((section("__trace_printk_fmt"))) =	\
707 			__builtin_constant_p(fmt) ? fmt : NULL;		\
708 									\
709 		__ftrace_vbprintk(_THIS_IP_, trace_printk_fmt, vargs);	\
710 	} else								\
711 		__ftrace_vprintk(_THIS_IP_, fmt, vargs);		\
712 } while (0)
713 
714 extern __printf(2, 0) int
715 __ftrace_vbprintk(unsigned long ip, const char *fmt, va_list ap);
716 
717 extern __printf(2, 0) int
718 __ftrace_vprintk(unsigned long ip, const char *fmt, va_list ap);
719 
720 extern void ftrace_dump(enum ftrace_dump_mode oops_dump_mode);
721 #else
722 static inline void tracing_start(void) { }
723 static inline void tracing_stop(void) { }
724 static inline void trace_dump_stack(int skip) { }
725 
726 static inline void tracing_on(void) { }
727 static inline void tracing_off(void) { }
728 static inline int tracing_is_on(void) { return 0; }
729 static inline void tracing_snapshot(void) { }
730 static inline void tracing_snapshot_alloc(void) { }
731 
732 static inline __printf(1, 2)
733 int trace_printk(const char *fmt, ...)
734 {
735 	return 0;
736 }
737 static __printf(1, 0) inline int
738 ftrace_vprintk(const char *fmt, va_list ap)
739 {
740 	return 0;
741 }
742 static inline void ftrace_dump(enum ftrace_dump_mode oops_dump_mode) { }
743 #endif /* CONFIG_TRACING */
744 
745 /*
746  * min()/max()/clamp() macros that also do
747  * strict type-checking.. See the
748  * "unnecessary" pointer comparison.
749  */
750 #define __min(t1, t2, min1, min2, x, y) ({		\
751 	t1 min1 = (x);					\
752 	t2 min2 = (y);					\
753 	(void) (&min1 == &min2);			\
754 	min1 < min2 ? min1 : min2; })
755 #define min(x, y)					\
756 	__min(typeof(x), typeof(y),			\
757 	      __UNIQUE_ID(min1_), __UNIQUE_ID(min2_),	\
758 	      x, y)
759 
760 #define __max(t1, t2, max1, max2, x, y) ({		\
761 	t1 max1 = (x);					\
762 	t2 max2 = (y);					\
763 	(void) (&max1 == &max2);			\
764 	max1 > max2 ? max1 : max2; })
765 #define max(x, y)					\
766 	__max(typeof(x), typeof(y),			\
767 	      __UNIQUE_ID(max1_), __UNIQUE_ID(max2_),	\
768 	      x, y)
769 
770 #define min3(x, y, z) min((typeof(x))min(x, y), z)
771 #define max3(x, y, z) max((typeof(x))max(x, y), z)
772 
773 /**
774  * min_not_zero - return the minimum that is _not_ zero, unless both are zero
775  * @x: value1
776  * @y: value2
777  */
778 #define min_not_zero(x, y) ({			\
779 	typeof(x) __x = (x);			\
780 	typeof(y) __y = (y);			\
781 	__x == 0 ? __y : ((__y == 0) ? __x : min(__x, __y)); })
782 
783 /**
784  * clamp - return a value clamped to a given range with strict typechecking
785  * @val: current value
786  * @lo: lowest allowable value
787  * @hi: highest allowable value
788  *
789  * This macro does strict typechecking of lo/hi to make sure they are of the
790  * same type as val.  See the unnecessary pointer comparisons.
791  */
792 #define clamp(val, lo, hi) min((typeof(val))max(val, lo), hi)
793 
794 /*
795  * ..and if you can't take the strict
796  * types, you can specify one yourself.
797  *
798  * Or not use min/max/clamp at all, of course.
799  */
800 #define min_t(type, x, y)				\
801 	__min(type, type,				\
802 	      __UNIQUE_ID(min1_), __UNIQUE_ID(min2_),	\
803 	      x, y)
804 
805 #define max_t(type, x, y)				\
806 	__max(type, type,				\
807 	      __UNIQUE_ID(min1_), __UNIQUE_ID(min2_),	\
808 	      x, y)
809 
810 /**
811  * clamp_t - return a value clamped to a given range using a given type
812  * @type: the type of variable to use
813  * @val: current value
814  * @lo: minimum allowable value
815  * @hi: maximum allowable value
816  *
817  * This macro does no typechecking and uses temporary variables of type
818  * 'type' to make all the comparisons.
819  */
820 #define clamp_t(type, val, lo, hi) min_t(type, max_t(type, val, lo), hi)
821 
822 /**
823  * clamp_val - return a value clamped to a given range using val's type
824  * @val: current value
825  * @lo: minimum allowable value
826  * @hi: maximum allowable value
827  *
828  * This macro does no typechecking and uses temporary variables of whatever
829  * type the input argument 'val' is.  This is useful when val is an unsigned
830  * type and min and max are literals that will otherwise be assigned a signed
831  * integer type.
832  */
833 #define clamp_val(val, lo, hi) clamp_t(typeof(val), val, lo, hi)
834 
835 
836 /*
837  * swap - swap value of @a and @b
838  */
839 #define swap(a, b) \
840 	do { typeof(a) __tmp = (a); (a) = (b); (b) = __tmp; } while (0)
841 
842 /**
843  * container_of - cast a member of a structure out to the containing structure
844  * @ptr:	the pointer to the member.
845  * @type:	the type of the container struct this is embedded in.
846  * @member:	the name of the member within the struct.
847  *
848  */
849 #define container_of(ptr, type, member) ({			\
850 	const typeof( ((type *)0)->member ) *__mptr = (ptr);	\
851 	(type *)( (char *)__mptr - offsetof(type,member) );})
852 
853 /* Rebuild everything on CONFIG_FTRACE_MCOUNT_RECORD */
854 #ifdef CONFIG_FTRACE_MCOUNT_RECORD
855 # define REBUILD_DUE_TO_FTRACE_MCOUNT_RECORD
856 #endif
857 
858 /* Permissions on a sysfs file: you didn't miss the 0 prefix did you? */
859 #define VERIFY_OCTAL_PERMISSIONS(perms)						\
860 	(BUILD_BUG_ON_ZERO((perms) < 0) +					\
861 	 BUILD_BUG_ON_ZERO((perms) > 0777) +					\
862 	 /* USER_READABLE >= GROUP_READABLE >= OTHER_READABLE */		\
863 	 BUILD_BUG_ON_ZERO((((perms) >> 6) & 4) < (((perms) >> 3) & 4)) +	\
864 	 BUILD_BUG_ON_ZERO((((perms) >> 3) & 4) < ((perms) & 4)) +		\
865 	 /* USER_WRITABLE >= GROUP_WRITABLE */					\
866 	 BUILD_BUG_ON_ZERO((((perms) >> 6) & 2) < (((perms) >> 3) & 2)) +	\
867 	 /* OTHER_WRITABLE?  Generally considered a bad idea. */		\
868 	 BUILD_BUG_ON_ZERO((perms) & 2) +					\
869 	 (perms))
870 #endif
871