xref: /linux-6.15/include/linux/kernel.h (revision a20deb3a)
1 /* SPDX-License-Identifier: GPL-2.0 */
2 #ifndef _LINUX_KERNEL_H
3 #define _LINUX_KERNEL_H
4 
5 #include <linux/stdarg.h>
6 #include <linux/align.h>
7 #include <linux/limits.h>
8 #include <linux/linkage.h>
9 #include <linux/stddef.h>
10 #include <linux/types.h>
11 #include <linux/compiler.h>
12 #include <linux/container_of.h>
13 #include <linux/bitops.h>
14 #include <linux/kstrtox.h>
15 #include <linux/log2.h>
16 #include <linux/math.h>
17 #include <linux/minmax.h>
18 #include <linux/typecheck.h>
19 #include <linux/panic.h>
20 #include <linux/printk.h>
21 #include <linux/build_bug.h>
22 #include <linux/static_call_types.h>
23 #include <linux/instruction_pointer.h>
24 #include <asm/byteorder.h>
25 
26 #include <uapi/linux/kernel.h>
27 
28 #define STACK_MAGIC	0xdeadbeef
29 
30 /**
31  * REPEAT_BYTE - repeat the value @x multiple times as an unsigned long value
32  * @x: value to repeat
33  *
34  * NOTE: @x is not checked for > 0xff; larger values produce odd results.
35  */
36 #define REPEAT_BYTE(x)	((~0ul / 0xff) * (x))
37 
38 /* generic data direction definitions */
39 #define READ			0
40 #define WRITE			1
41 
42 /**
43  * ARRAY_SIZE - get the number of elements in array @arr
44  * @arr: array to be sized
45  */
46 #define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]) + __must_be_array(arr))
47 
48 #define PTR_IF(cond, ptr)	((cond) ? (ptr) : NULL)
49 
50 #define u64_to_user_ptr(x) (		\
51 {					\
52 	typecheck(u64, (x));		\
53 	(void __user *)(uintptr_t)(x);	\
54 }					\
55 )
56 
57 /**
58  * upper_32_bits - return bits 32-63 of a number
59  * @n: the number we're accessing
60  *
61  * A basic shift-right of a 64- or 32-bit quantity.  Use this to suppress
62  * the "right shift count >= width of type" warning when that quantity is
63  * 32-bits.
64  */
65 #define upper_32_bits(n) ((u32)(((n) >> 16) >> 16))
66 
67 /**
68  * lower_32_bits - return bits 0-31 of a number
69  * @n: the number we're accessing
70  */
71 #define lower_32_bits(n) ((u32)((n) & 0xffffffff))
72 
73 /**
74  * upper_16_bits - return bits 16-31 of a number
75  * @n: the number we're accessing
76  */
77 #define upper_16_bits(n) ((u16)((n) >> 16))
78 
79 /**
80  * lower_16_bits - return bits 0-15 of a number
81  * @n: the number we're accessing
82  */
83 #define lower_16_bits(n) ((u16)((n) & 0xffff))
84 
85 struct completion;
86 struct user;
87 
88 #ifdef CONFIG_PREEMPT_VOLUNTARY
89 
90 extern int __cond_resched(void);
91 # define might_resched() __cond_resched()
92 
93 #elif defined(CONFIG_PREEMPT_DYNAMIC)
94 
95 extern int __cond_resched(void);
96 
97 DECLARE_STATIC_CALL(might_resched, __cond_resched);
98 
99 static __always_inline void might_resched(void)
100 {
101 	static_call_mod(might_resched)();
102 }
103 
104 #else
105 
106 # define might_resched() do { } while (0)
107 
108 #endif /* CONFIG_PREEMPT_* */
109 
110 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
111 extern void ___might_sleep(const char *file, int line, int preempt_offset);
112 extern void __might_sleep(const char *file, int line, int preempt_offset);
113 extern void __cant_sleep(const char *file, int line, int preempt_offset);
114 extern void __cant_migrate(const char *file, int line);
115 
116 /**
117  * might_sleep - annotation for functions that can sleep
118  *
119  * this macro will print a stack trace if it is executed in an atomic
120  * context (spinlock, irq-handler, ...). Additional sections where blocking is
121  * not allowed can be annotated with non_block_start() and non_block_end()
122  * pairs.
123  *
124  * This is a useful debugging help to be able to catch problems early and not
125  * be bitten later when the calling function happens to sleep when it is not
126  * supposed to.
127  */
128 # define might_sleep() \
129 	do { __might_sleep(__FILE__, __LINE__, 0); might_resched(); } while (0)
130 /**
131  * cant_sleep - annotation for functions that cannot sleep
132  *
133  * this macro will print a stack trace if it is executed with preemption enabled
134  */
135 # define cant_sleep() \
136 	do { __cant_sleep(__FILE__, __LINE__, 0); } while (0)
137 # define sched_annotate_sleep()	(current->task_state_change = 0)
138 
139 /**
140  * cant_migrate - annotation for functions that cannot migrate
141  *
142  * Will print a stack trace if executed in code which is migratable
143  */
144 # define cant_migrate()							\
145 	do {								\
146 		if (IS_ENABLED(CONFIG_SMP))				\
147 			__cant_migrate(__FILE__, __LINE__);		\
148 	} while (0)
149 
150 /**
151  * non_block_start - annotate the start of section where sleeping is prohibited
152  *
153  * This is on behalf of the oom reaper, specifically when it is calling the mmu
154  * notifiers. The problem is that if the notifier were to block on, for example,
155  * mutex_lock() and if the process which holds that mutex were to perform a
156  * sleeping memory allocation, the oom reaper is now blocked on completion of
157  * that memory allocation. Other blocking calls like wait_event() pose similar
158  * issues.
159  */
160 # define non_block_start() (current->non_block_count++)
161 /**
162  * non_block_end - annotate the end of section where sleeping is prohibited
163  *
164  * Closes a section opened by non_block_start().
165  */
166 # define non_block_end() WARN_ON(current->non_block_count-- == 0)
167 #else
168   static inline void ___might_sleep(const char *file, int line,
169 				   int preempt_offset) { }
170   static inline void __might_sleep(const char *file, int line,
171 				   int preempt_offset) { }
172 # define might_sleep() do { might_resched(); } while (0)
173 # define cant_sleep() do { } while (0)
174 # define cant_migrate()		do { } while (0)
175 # define sched_annotate_sleep() do { } while (0)
176 # define non_block_start() do { } while (0)
177 # define non_block_end() do { } while (0)
178 #endif
179 
180 #define might_sleep_if(cond) do { if (cond) might_sleep(); } while (0)
181 
182 #if defined(CONFIG_MMU) && \
183 	(defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_DEBUG_ATOMIC_SLEEP))
184 #define might_fault() __might_fault(__FILE__, __LINE__)
185 void __might_fault(const char *file, int line);
186 #else
187 static inline void might_fault(void) { }
188 #endif
189 
190 void do_exit(long error_code) __noreturn;
191 void complete_and_exit(struct completion *, long) __noreturn;
192 
193 extern int num_to_str(char *buf, int size,
194 		      unsigned long long num, unsigned int width);
195 
196 /* lib/printf utilities */
197 
198 extern __printf(2, 3) int sprintf(char *buf, const char * fmt, ...);
199 extern __printf(2, 0) int vsprintf(char *buf, const char *, va_list);
200 extern __printf(3, 4)
201 int snprintf(char *buf, size_t size, const char *fmt, ...);
202 extern __printf(3, 0)
203 int vsnprintf(char *buf, size_t size, const char *fmt, va_list args);
204 extern __printf(3, 4)
205 int scnprintf(char *buf, size_t size, const char *fmt, ...);
206 extern __printf(3, 0)
207 int vscnprintf(char *buf, size_t size, const char *fmt, va_list args);
208 extern __printf(2, 3) __malloc
209 char *kasprintf(gfp_t gfp, const char *fmt, ...);
210 extern __printf(2, 0) __malloc
211 char *kvasprintf(gfp_t gfp, const char *fmt, va_list args);
212 extern __printf(2, 0)
213 const char *kvasprintf_const(gfp_t gfp, const char *fmt, va_list args);
214 
215 extern __scanf(2, 3)
216 int sscanf(const char *, const char *, ...);
217 extern __scanf(2, 0)
218 int vsscanf(const char *, const char *, va_list);
219 
220 extern int no_hash_pointers_enable(char *str);
221 
222 extern int get_option(char **str, int *pint);
223 extern char *get_options(const char *str, int nints, int *ints);
224 extern unsigned long long memparse(const char *ptr, char **retptr);
225 extern bool parse_option_str(const char *str, const char *option);
226 extern char *next_arg(char *args, char **param, char **val);
227 
228 extern int core_kernel_text(unsigned long addr);
229 extern int init_kernel_text(unsigned long addr);
230 extern int __kernel_text_address(unsigned long addr);
231 extern int kernel_text_address(unsigned long addr);
232 extern int func_ptr_is_kernel_text(void *ptr);
233 
234 extern void bust_spinlocks(int yes);
235 
236 extern int root_mountflags;
237 
238 extern bool early_boot_irqs_disabled;
239 
240 /*
241  * Values used for system_state. Ordering of the states must not be changed
242  * as code checks for <, <=, >, >= STATE.
243  */
244 extern enum system_states {
245 	SYSTEM_BOOTING,
246 	SYSTEM_SCHEDULING,
247 	SYSTEM_FREEING_INITMEM,
248 	SYSTEM_RUNNING,
249 	SYSTEM_HALT,
250 	SYSTEM_POWER_OFF,
251 	SYSTEM_RESTART,
252 	SYSTEM_SUSPEND,
253 } system_state;
254 
255 extern const char hex_asc[];
256 #define hex_asc_lo(x)	hex_asc[((x) & 0x0f)]
257 #define hex_asc_hi(x)	hex_asc[((x) & 0xf0) >> 4]
258 
259 static inline char *hex_byte_pack(char *buf, u8 byte)
260 {
261 	*buf++ = hex_asc_hi(byte);
262 	*buf++ = hex_asc_lo(byte);
263 	return buf;
264 }
265 
266 extern const char hex_asc_upper[];
267 #define hex_asc_upper_lo(x)	hex_asc_upper[((x) & 0x0f)]
268 #define hex_asc_upper_hi(x)	hex_asc_upper[((x) & 0xf0) >> 4]
269 
270 static inline char *hex_byte_pack_upper(char *buf, u8 byte)
271 {
272 	*buf++ = hex_asc_upper_hi(byte);
273 	*buf++ = hex_asc_upper_lo(byte);
274 	return buf;
275 }
276 
277 extern int hex_to_bin(char ch);
278 extern int __must_check hex2bin(u8 *dst, const char *src, size_t count);
279 extern char *bin2hex(char *dst, const void *src, size_t count);
280 
281 bool mac_pton(const char *s, u8 *mac);
282 
283 /*
284  * General tracing related utility functions - trace_printk(),
285  * tracing_on/tracing_off and tracing_start()/tracing_stop
286  *
287  * Use tracing_on/tracing_off when you want to quickly turn on or off
288  * tracing. It simply enables or disables the recording of the trace events.
289  * This also corresponds to the user space /sys/kernel/debug/tracing/tracing_on
290  * file, which gives a means for the kernel and userspace to interact.
291  * Place a tracing_off() in the kernel where you want tracing to end.
292  * From user space, examine the trace, and then echo 1 > tracing_on
293  * to continue tracing.
294  *
295  * tracing_stop/tracing_start has slightly more overhead. It is used
296  * by things like suspend to ram where disabling the recording of the
297  * trace is not enough, but tracing must actually stop because things
298  * like calling smp_processor_id() may crash the system.
299  *
300  * Most likely, you want to use tracing_on/tracing_off.
301  */
302 
303 enum ftrace_dump_mode {
304 	DUMP_NONE,
305 	DUMP_ALL,
306 	DUMP_ORIG,
307 };
308 
309 #ifdef CONFIG_TRACING
310 void tracing_on(void);
311 void tracing_off(void);
312 int tracing_is_on(void);
313 void tracing_snapshot(void);
314 void tracing_snapshot_alloc(void);
315 
316 extern void tracing_start(void);
317 extern void tracing_stop(void);
318 
319 static inline __printf(1, 2)
320 void ____trace_printk_check_format(const char *fmt, ...)
321 {
322 }
323 #define __trace_printk_check_format(fmt, args...)			\
324 do {									\
325 	if (0)								\
326 		____trace_printk_check_format(fmt, ##args);		\
327 } while (0)
328 
329 /**
330  * trace_printk - printf formatting in the ftrace buffer
331  * @fmt: the printf format for printing
332  *
333  * Note: __trace_printk is an internal function for trace_printk() and
334  *       the @ip is passed in via the trace_printk() macro.
335  *
336  * This function allows a kernel developer to debug fast path sections
337  * that printk is not appropriate for. By scattering in various
338  * printk like tracing in the code, a developer can quickly see
339  * where problems are occurring.
340  *
341  * This is intended as a debugging tool for the developer only.
342  * Please refrain from leaving trace_printks scattered around in
343  * your code. (Extra memory is used for special buffers that are
344  * allocated when trace_printk() is used.)
345  *
346  * A little optimization trick is done here. If there's only one
347  * argument, there's no need to scan the string for printf formats.
348  * The trace_puts() will suffice. But how can we take advantage of
349  * using trace_puts() when trace_printk() has only one argument?
350  * By stringifying the args and checking the size we can tell
351  * whether or not there are args. __stringify((__VA_ARGS__)) will
352  * turn into "()\0" with a size of 3 when there are no args, anything
353  * else will be bigger. All we need to do is define a string to this,
354  * and then take its size and compare to 3. If it's bigger, use
355  * do_trace_printk() otherwise, optimize it to trace_puts(). Then just
356  * let gcc optimize the rest.
357  */
358 
359 #define trace_printk(fmt, ...)				\
360 do {							\
361 	char _______STR[] = __stringify((__VA_ARGS__));	\
362 	if (sizeof(_______STR) > 3)			\
363 		do_trace_printk(fmt, ##__VA_ARGS__);	\
364 	else						\
365 		trace_puts(fmt);			\
366 } while (0)
367 
368 #define do_trace_printk(fmt, args...)					\
369 do {									\
370 	static const char *trace_printk_fmt __used			\
371 		__section("__trace_printk_fmt") =			\
372 		__builtin_constant_p(fmt) ? fmt : NULL;			\
373 									\
374 	__trace_printk_check_format(fmt, ##args);			\
375 									\
376 	if (__builtin_constant_p(fmt))					\
377 		__trace_bprintk(_THIS_IP_, trace_printk_fmt, ##args);	\
378 	else								\
379 		__trace_printk(_THIS_IP_, fmt, ##args);			\
380 } while (0)
381 
382 extern __printf(2, 3)
383 int __trace_bprintk(unsigned long ip, const char *fmt, ...);
384 
385 extern __printf(2, 3)
386 int __trace_printk(unsigned long ip, const char *fmt, ...);
387 
388 /**
389  * trace_puts - write a string into the ftrace buffer
390  * @str: the string to record
391  *
392  * Note: __trace_bputs is an internal function for trace_puts and
393  *       the @ip is passed in via the trace_puts macro.
394  *
395  * This is similar to trace_printk() but is made for those really fast
396  * paths that a developer wants the least amount of "Heisenbug" effects,
397  * where the processing of the print format is still too much.
398  *
399  * This function allows a kernel developer to debug fast path sections
400  * that printk is not appropriate for. By scattering in various
401  * printk like tracing in the code, a developer can quickly see
402  * where problems are occurring.
403  *
404  * This is intended as a debugging tool for the developer only.
405  * Please refrain from leaving trace_puts scattered around in
406  * your code. (Extra memory is used for special buffers that are
407  * allocated when trace_puts() is used.)
408  *
409  * Returns: 0 if nothing was written, positive # if string was.
410  *  (1 when __trace_bputs is used, strlen(str) when __trace_puts is used)
411  */
412 
413 #define trace_puts(str) ({						\
414 	static const char *trace_printk_fmt __used			\
415 		__section("__trace_printk_fmt") =			\
416 		__builtin_constant_p(str) ? str : NULL;			\
417 									\
418 	if (__builtin_constant_p(str))					\
419 		__trace_bputs(_THIS_IP_, trace_printk_fmt);		\
420 	else								\
421 		__trace_puts(_THIS_IP_, str, strlen(str));		\
422 })
423 extern int __trace_bputs(unsigned long ip, const char *str);
424 extern int __trace_puts(unsigned long ip, const char *str, int size);
425 
426 extern void trace_dump_stack(int skip);
427 
428 /*
429  * The double __builtin_constant_p is because gcc will give us an error
430  * if we try to allocate the static variable to fmt if it is not a
431  * constant. Even with the outer if statement.
432  */
433 #define ftrace_vprintk(fmt, vargs)					\
434 do {									\
435 	if (__builtin_constant_p(fmt)) {				\
436 		static const char *trace_printk_fmt __used		\
437 		  __section("__trace_printk_fmt") =			\
438 			__builtin_constant_p(fmt) ? fmt : NULL;		\
439 									\
440 		__ftrace_vbprintk(_THIS_IP_, trace_printk_fmt, vargs);	\
441 	} else								\
442 		__ftrace_vprintk(_THIS_IP_, fmt, vargs);		\
443 } while (0)
444 
445 extern __printf(2, 0) int
446 __ftrace_vbprintk(unsigned long ip, const char *fmt, va_list ap);
447 
448 extern __printf(2, 0) int
449 __ftrace_vprintk(unsigned long ip, const char *fmt, va_list ap);
450 
451 extern void ftrace_dump(enum ftrace_dump_mode oops_dump_mode);
452 #else
453 static inline void tracing_start(void) { }
454 static inline void tracing_stop(void) { }
455 static inline void trace_dump_stack(int skip) { }
456 
457 static inline void tracing_on(void) { }
458 static inline void tracing_off(void) { }
459 static inline int tracing_is_on(void) { return 0; }
460 static inline void tracing_snapshot(void) { }
461 static inline void tracing_snapshot_alloc(void) { }
462 
463 static inline __printf(1, 2)
464 int trace_printk(const char *fmt, ...)
465 {
466 	return 0;
467 }
468 static __printf(1, 0) inline int
469 ftrace_vprintk(const char *fmt, va_list ap)
470 {
471 	return 0;
472 }
473 static inline void ftrace_dump(enum ftrace_dump_mode oops_dump_mode) { }
474 #endif /* CONFIG_TRACING */
475 
476 /* This counts to 12. Any more, it will return 13th argument. */
477 #define __COUNT_ARGS(_0, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _n, X...) _n
478 #define COUNT_ARGS(X...) __COUNT_ARGS(, ##X, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0)
479 
480 #define __CONCAT(a, b) a ## b
481 #define CONCATENATE(a, b) __CONCAT(a, b)
482 
483 /* Rebuild everything on CONFIG_FTRACE_MCOUNT_RECORD */
484 #ifdef CONFIG_FTRACE_MCOUNT_RECORD
485 # define REBUILD_DUE_TO_FTRACE_MCOUNT_RECORD
486 #endif
487 
488 /* Permissions on a sysfs file: you didn't miss the 0 prefix did you? */
489 #define VERIFY_OCTAL_PERMISSIONS(perms)						\
490 	(BUILD_BUG_ON_ZERO((perms) < 0) +					\
491 	 BUILD_BUG_ON_ZERO((perms) > 0777) +					\
492 	 /* USER_READABLE >= GROUP_READABLE >= OTHER_READABLE */		\
493 	 BUILD_BUG_ON_ZERO((((perms) >> 6) & 4) < (((perms) >> 3) & 4)) +	\
494 	 BUILD_BUG_ON_ZERO((((perms) >> 3) & 4) < ((perms) & 4)) +		\
495 	 /* USER_WRITABLE >= GROUP_WRITABLE */					\
496 	 BUILD_BUG_ON_ZERO((((perms) >> 6) & 2) < (((perms) >> 3) & 2)) +	\
497 	 /* OTHER_WRITABLE?  Generally considered a bad idea. */		\
498 	 BUILD_BUG_ON_ZERO((perms) & 2) +					\
499 	 (perms))
500 #endif
501