xref: /linux-6.15/include/linux/kernel.h (revision cead1855)
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_BUILD
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_resched(const char *file, int line, unsigned int offsets);
112 extern void __might_sleep(const char *file, int line);
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__); 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_resched(const char *file, int line,
169 				     unsigned int offsets) { }
170 static inline void __might_sleep(const char *file, int line) { }
171 # define might_sleep() do { might_resched(); } while (0)
172 # define cant_sleep() do { } while (0)
173 # define cant_migrate()		do { } while (0)
174 # define sched_annotate_sleep() do { } while (0)
175 # define non_block_start() do { } while (0)
176 # define non_block_end() do { } while (0)
177 #endif
178 
179 #define might_sleep_if(cond) do { if (cond) might_sleep(); } while (0)
180 
181 #if defined(CONFIG_MMU) && \
182 	(defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_DEBUG_ATOMIC_SLEEP))
183 #define might_fault() __might_fault(__FILE__, __LINE__)
184 void __might_fault(const char *file, int line);
185 #else
186 static inline void might_fault(void) { }
187 #endif
188 
189 void do_exit(long error_code) __noreturn;
190 
191 extern int num_to_str(char *buf, int size,
192 		      unsigned long long num, unsigned int width);
193 
194 /* lib/printf utilities */
195 
196 extern __printf(2, 3) int sprintf(char *buf, const char * fmt, ...);
197 extern __printf(2, 0) int vsprintf(char *buf, const char *, va_list);
198 extern __printf(3, 4)
199 int snprintf(char *buf, size_t size, const char *fmt, ...);
200 extern __printf(3, 0)
201 int vsnprintf(char *buf, size_t size, const char *fmt, va_list args);
202 extern __printf(3, 4)
203 int scnprintf(char *buf, size_t size, const char *fmt, ...);
204 extern __printf(3, 0)
205 int vscnprintf(char *buf, size_t size, const char *fmt, va_list args);
206 extern __printf(2, 3) __malloc
207 char *kasprintf(gfp_t gfp, const char *fmt, ...);
208 extern __printf(2, 0) __malloc
209 char *kvasprintf(gfp_t gfp, const char *fmt, va_list args);
210 extern __printf(2, 0)
211 const char *kvasprintf_const(gfp_t gfp, const char *fmt, va_list args);
212 
213 extern __scanf(2, 3)
214 int sscanf(const char *, const char *, ...);
215 extern __scanf(2, 0)
216 int vsscanf(const char *, const char *, va_list);
217 
218 extern int no_hash_pointers_enable(char *str);
219 
220 extern int get_option(char **str, int *pint);
221 extern char *get_options(const char *str, int nints, int *ints);
222 extern unsigned long long memparse(const char *ptr, char **retptr);
223 extern bool parse_option_str(const char *str, const char *option);
224 extern char *next_arg(char *args, char **param, char **val);
225 
226 extern int core_kernel_text(unsigned long addr);
227 extern int __kernel_text_address(unsigned long addr);
228 extern int kernel_text_address(unsigned long addr);
229 extern int func_ptr_is_kernel_text(void *ptr);
230 
231 extern void bust_spinlocks(int yes);
232 
233 extern int root_mountflags;
234 
235 extern bool early_boot_irqs_disabled;
236 
237 /*
238  * Values used for system_state. Ordering of the states must not be changed
239  * as code checks for <, <=, >, >= STATE.
240  */
241 extern enum system_states {
242 	SYSTEM_BOOTING,
243 	SYSTEM_SCHEDULING,
244 	SYSTEM_FREEING_INITMEM,
245 	SYSTEM_RUNNING,
246 	SYSTEM_HALT,
247 	SYSTEM_POWER_OFF,
248 	SYSTEM_RESTART,
249 	SYSTEM_SUSPEND,
250 } system_state;
251 
252 extern const char hex_asc[];
253 #define hex_asc_lo(x)	hex_asc[((x) & 0x0f)]
254 #define hex_asc_hi(x)	hex_asc[((x) & 0xf0) >> 4]
255 
256 static inline char *hex_byte_pack(char *buf, u8 byte)
257 {
258 	*buf++ = hex_asc_hi(byte);
259 	*buf++ = hex_asc_lo(byte);
260 	return buf;
261 }
262 
263 extern const char hex_asc_upper[];
264 #define hex_asc_upper_lo(x)	hex_asc_upper[((x) & 0x0f)]
265 #define hex_asc_upper_hi(x)	hex_asc_upper[((x) & 0xf0) >> 4]
266 
267 static inline char *hex_byte_pack_upper(char *buf, u8 byte)
268 {
269 	*buf++ = hex_asc_upper_hi(byte);
270 	*buf++ = hex_asc_upper_lo(byte);
271 	return buf;
272 }
273 
274 extern int hex_to_bin(char ch);
275 extern int __must_check hex2bin(u8 *dst, const char *src, size_t count);
276 extern char *bin2hex(char *dst, const void *src, size_t count);
277 
278 bool mac_pton(const char *s, u8 *mac);
279 
280 /*
281  * General tracing related utility functions - trace_printk(),
282  * tracing_on/tracing_off and tracing_start()/tracing_stop
283  *
284  * Use tracing_on/tracing_off when you want to quickly turn on or off
285  * tracing. It simply enables or disables the recording of the trace events.
286  * This also corresponds to the user space /sys/kernel/debug/tracing/tracing_on
287  * file, which gives a means for the kernel and userspace to interact.
288  * Place a tracing_off() in the kernel where you want tracing to end.
289  * From user space, examine the trace, and then echo 1 > tracing_on
290  * to continue tracing.
291  *
292  * tracing_stop/tracing_start has slightly more overhead. It is used
293  * by things like suspend to ram where disabling the recording of the
294  * trace is not enough, but tracing must actually stop because things
295  * like calling smp_processor_id() may crash the system.
296  *
297  * Most likely, you want to use tracing_on/tracing_off.
298  */
299 
300 enum ftrace_dump_mode {
301 	DUMP_NONE,
302 	DUMP_ALL,
303 	DUMP_ORIG,
304 };
305 
306 #ifdef CONFIG_TRACING
307 void tracing_on(void);
308 void tracing_off(void);
309 int tracing_is_on(void);
310 void tracing_snapshot(void);
311 void tracing_snapshot_alloc(void);
312 
313 extern void tracing_start(void);
314 extern void tracing_stop(void);
315 
316 static inline __printf(1, 2)
317 void ____trace_printk_check_format(const char *fmt, ...)
318 {
319 }
320 #define __trace_printk_check_format(fmt, args...)			\
321 do {									\
322 	if (0)								\
323 		____trace_printk_check_format(fmt, ##args);		\
324 } while (0)
325 
326 /**
327  * trace_printk - printf formatting in the ftrace buffer
328  * @fmt: the printf format for printing
329  *
330  * Note: __trace_printk is an internal function for trace_printk() and
331  *       the @ip is passed in via the trace_printk() macro.
332  *
333  * This function allows a kernel developer to debug fast path sections
334  * that printk is not appropriate for. By scattering in various
335  * printk like tracing in the code, a developer can quickly see
336  * where problems are occurring.
337  *
338  * This is intended as a debugging tool for the developer only.
339  * Please refrain from leaving trace_printks scattered around in
340  * your code. (Extra memory is used for special buffers that are
341  * allocated when trace_printk() is used.)
342  *
343  * A little optimization trick is done here. If there's only one
344  * argument, there's no need to scan the string for printf formats.
345  * The trace_puts() will suffice. But how can we take advantage of
346  * using trace_puts() when trace_printk() has only one argument?
347  * By stringifying the args and checking the size we can tell
348  * whether or not there are args. __stringify((__VA_ARGS__)) will
349  * turn into "()\0" with a size of 3 when there are no args, anything
350  * else will be bigger. All we need to do is define a string to this,
351  * and then take its size and compare to 3. If it's bigger, use
352  * do_trace_printk() otherwise, optimize it to trace_puts(). Then just
353  * let gcc optimize the rest.
354  */
355 
356 #define trace_printk(fmt, ...)				\
357 do {							\
358 	char _______STR[] = __stringify((__VA_ARGS__));	\
359 	if (sizeof(_______STR) > 3)			\
360 		do_trace_printk(fmt, ##__VA_ARGS__);	\
361 	else						\
362 		trace_puts(fmt);			\
363 } while (0)
364 
365 #define do_trace_printk(fmt, args...)					\
366 do {									\
367 	static const char *trace_printk_fmt __used			\
368 		__section("__trace_printk_fmt") =			\
369 		__builtin_constant_p(fmt) ? fmt : NULL;			\
370 									\
371 	__trace_printk_check_format(fmt, ##args);			\
372 									\
373 	if (__builtin_constant_p(fmt))					\
374 		__trace_bprintk(_THIS_IP_, trace_printk_fmt, ##args);	\
375 	else								\
376 		__trace_printk(_THIS_IP_, fmt, ##args);			\
377 } while (0)
378 
379 extern __printf(2, 3)
380 int __trace_bprintk(unsigned long ip, const char *fmt, ...);
381 
382 extern __printf(2, 3)
383 int __trace_printk(unsigned long ip, const char *fmt, ...);
384 
385 /**
386  * trace_puts - write a string into the ftrace buffer
387  * @str: the string to record
388  *
389  * Note: __trace_bputs is an internal function for trace_puts and
390  *       the @ip is passed in via the trace_puts macro.
391  *
392  * This is similar to trace_printk() but is made for those really fast
393  * paths that a developer wants the least amount of "Heisenbug" effects,
394  * where the processing of the print format is still too much.
395  *
396  * This function allows a kernel developer to debug fast path sections
397  * that printk is not appropriate for. By scattering in various
398  * printk like tracing in the code, a developer can quickly see
399  * where problems are occurring.
400  *
401  * This is intended as a debugging tool for the developer only.
402  * Please refrain from leaving trace_puts scattered around in
403  * your code. (Extra memory is used for special buffers that are
404  * allocated when trace_puts() is used.)
405  *
406  * Returns: 0 if nothing was written, positive # if string was.
407  *  (1 when __trace_bputs is used, strlen(str) when __trace_puts is used)
408  */
409 
410 #define trace_puts(str) ({						\
411 	static const char *trace_printk_fmt __used			\
412 		__section("__trace_printk_fmt") =			\
413 		__builtin_constant_p(str) ? str : NULL;			\
414 									\
415 	if (__builtin_constant_p(str))					\
416 		__trace_bputs(_THIS_IP_, trace_printk_fmt);		\
417 	else								\
418 		__trace_puts(_THIS_IP_, str, strlen(str));		\
419 })
420 extern int __trace_bputs(unsigned long ip, const char *str);
421 extern int __trace_puts(unsigned long ip, const char *str, int size);
422 
423 extern void trace_dump_stack(int skip);
424 
425 /*
426  * The double __builtin_constant_p is because gcc will give us an error
427  * if we try to allocate the static variable to fmt if it is not a
428  * constant. Even with the outer if statement.
429  */
430 #define ftrace_vprintk(fmt, vargs)					\
431 do {									\
432 	if (__builtin_constant_p(fmt)) {				\
433 		static const char *trace_printk_fmt __used		\
434 		  __section("__trace_printk_fmt") =			\
435 			__builtin_constant_p(fmt) ? fmt : NULL;		\
436 									\
437 		__ftrace_vbprintk(_THIS_IP_, trace_printk_fmt, vargs);	\
438 	} else								\
439 		__ftrace_vprintk(_THIS_IP_, fmt, vargs);		\
440 } while (0)
441 
442 extern __printf(2, 0) int
443 __ftrace_vbprintk(unsigned long ip, const char *fmt, va_list ap);
444 
445 extern __printf(2, 0) int
446 __ftrace_vprintk(unsigned long ip, const char *fmt, va_list ap);
447 
448 extern void ftrace_dump(enum ftrace_dump_mode oops_dump_mode);
449 #else
450 static inline void tracing_start(void) { }
451 static inline void tracing_stop(void) { }
452 static inline void trace_dump_stack(int skip) { }
453 
454 static inline void tracing_on(void) { }
455 static inline void tracing_off(void) { }
456 static inline int tracing_is_on(void) { return 0; }
457 static inline void tracing_snapshot(void) { }
458 static inline void tracing_snapshot_alloc(void) { }
459 
460 static inline __printf(1, 2)
461 int trace_printk(const char *fmt, ...)
462 {
463 	return 0;
464 }
465 static __printf(1, 0) inline int
466 ftrace_vprintk(const char *fmt, va_list ap)
467 {
468 	return 0;
469 }
470 static inline void ftrace_dump(enum ftrace_dump_mode oops_dump_mode) { }
471 #endif /* CONFIG_TRACING */
472 
473 /* This counts to 12. Any more, it will return 13th argument. */
474 #define __COUNT_ARGS(_0, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _n, X...) _n
475 #define COUNT_ARGS(X...) __COUNT_ARGS(, ##X, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0)
476 
477 #define __CONCAT(a, b) a ## b
478 #define CONCATENATE(a, b) __CONCAT(a, b)
479 
480 /* Rebuild everything on CONFIG_FTRACE_MCOUNT_RECORD */
481 #ifdef CONFIG_FTRACE_MCOUNT_RECORD
482 # define REBUILD_DUE_TO_FTRACE_MCOUNT_RECORD
483 #endif
484 
485 /* Permissions on a sysfs file: you didn't miss the 0 prefix did you? */
486 #define VERIFY_OCTAL_PERMISSIONS(perms)						\
487 	(BUILD_BUG_ON_ZERO((perms) < 0) +					\
488 	 BUILD_BUG_ON_ZERO((perms) > 0777) +					\
489 	 /* USER_READABLE >= GROUP_READABLE >= OTHER_READABLE */		\
490 	 BUILD_BUG_ON_ZERO((((perms) >> 6) & 4) < (((perms) >> 3) & 4)) +	\
491 	 BUILD_BUG_ON_ZERO((((perms) >> 3) & 4) < ((perms) & 4)) +		\
492 	 /* USER_WRITABLE >= GROUP_WRITABLE */					\
493 	 BUILD_BUG_ON_ZERO((((perms) >> 6) & 2) < (((perms) >> 3) & 2)) +	\
494 	 /* OTHER_WRITABLE?  Generally considered a bad idea. */		\
495 	 BUILD_BUG_ON_ZERO((perms) & 2) +					\
496 	 (perms))
497 #endif
498