1/* SPDX-License-Identifier: GPL-2.0 */
2/*
3 * NOTE:
4 *
5 * This header has combined a lot of unrelated to each other stuff.
6 * The process of splitting its content is in progress while keeping
7 * backward compatibility. That's why it's highly recommended NOT to
8 * include this header inside another header file, especially under
9 * generic or architectural include/ directory.
10 */
11#ifndef _LINUX_KERNEL_H
12#define _LINUX_KERNEL_H
13
14#include <linux/stdarg.h>
15#include <linux/align.h>
16#include <linux/limits.h>
17#include <linux/linkage.h>
18#include <linux/stddef.h>
19#include <linux/types.h>
20#include <linux/compiler.h>
21#include <linux/container_of.h>
22#include <linux/bitops.h>
23#include <linux/kstrtox.h>
24#include <linux/log2.h>
25#include <linux/math.h>
26#include <linux/minmax.h>
27#include <linux/typecheck.h>
28#include <linux/panic.h>
29#include <linux/printk.h>
30#include <linux/build_bug.h>
31#include <linux/static_call_types.h>
32#include <linux/instruction_pointer.h>
33#include <asm/byteorder.h>
34
35#include <uapi/linux/kernel.h>
36
37#define STACK_MAGIC 0xdeadbeef
38
39/**
40 * REPEAT_BYTE - repeat the value @x multiple times as an unsigned long value
41 * @x: value to repeat
42 *
43 * NOTE: @x is not checked for > 0xff; larger values produce odd results.
44 */
45#define REPEAT_BYTE(x) ((~0ul / 0xff) * (x))
46
47/* generic data direction definitions */
48#define READ 0
49#define WRITE 1
50
51/**
52 * ARRAY_SIZE - get the number of elements in array @arr
53 * @arr: array to be sized
54 */
55#define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]) + __must_be_array(arr))
56
57#define PTR_IF(cond, ptr) ((cond) ? (ptr) : NULL)
58
59#define u64_to_user_ptr(x) ( \
60{ \
61 typecheck(u64, (x)); \
62 (void __user *)(uintptr_t)(x); \
63} \
64)
65
66/**
67 * upper_32_bits - return bits 32-63 of a number
68 * @n: the number we're accessing
69 *
70 * A basic shift-right of a 64- or 32-bit quantity. Use this to suppress
71 * the "right shift count >= width of type" warning when that quantity is
72 * 32-bits.
73 */
74#define upper_32_bits(n) ((u32)(((n) >> 16) >> 16))
75
76/**
77 * lower_32_bits - return bits 0-31 of a number
78 * @n: the number we're accessing
79 */
80#define lower_32_bits(n) ((u32)((n) & 0xffffffff))
81
82/**
83 * upper_16_bits - return bits 16-31 of a number
84 * @n: the number we're accessing
85 */
86#define upper_16_bits(n) ((u16)((n) >> 16))
87
88/**
89 * lower_16_bits - return bits 0-15 of a number
90 * @n: the number we're accessing
91 */
92#define lower_16_bits(n) ((u16)((n) & 0xffff))
93
94struct completion;
95struct user;
96
97#ifdef CONFIG_PREEMPT_VOLUNTARY_BUILD
98
99extern int __cond_resched(void);
100# define might_resched() __cond_resched()
101
102#elif defined(CONFIG_PREEMPT_DYNAMIC) && defined(CONFIG_HAVE_PREEMPT_DYNAMIC_CALL)
103
104extern int __cond_resched(void);
105
106DECLARE_STATIC_CALL(might_resched, __cond_resched);
107
108static __always_inline void might_resched(void)
109{
110 static_call_mod(might_resched)();
111}
112
113#elif defined(CONFIG_PREEMPT_DYNAMIC) && defined(CONFIG_HAVE_PREEMPT_DYNAMIC_KEY)
114
115extern int dynamic_might_resched(void);
116# define might_resched() dynamic_might_resched()
117
118#else
119
120# define might_resched() do { } while (0)
121
122#endif /* CONFIG_PREEMPT_* */
123
124#ifdef CONFIG_DEBUG_ATOMIC_SLEEP
125extern void __might_resched(const char *file, int line, unsigned int offsets);
126extern void __might_sleep(const char *file, int line);
127extern void __cant_sleep(const char *file, int line, int preempt_offset);
128extern void __cant_migrate(const char *file, int line);
129
130/**
131 * might_sleep - annotation for functions that can sleep
132 *
133 * this macro will print a stack trace if it is executed in an atomic
134 * context (spinlock, irq-handler, ...). Additional sections where blocking is
135 * not allowed can be annotated with non_block_start() and non_block_end()
136 * pairs.
137 *
138 * This is a useful debugging help to be able to catch problems early and not
139 * be bitten later when the calling function happens to sleep when it is not
140 * supposed to.
141 */
142# define might_sleep() \
143 do { __might_sleep(__FILE__, __LINE__); might_resched(); } while (0)
144/**
145 * cant_sleep - annotation for functions that cannot sleep
146 *
147 * this macro will print a stack trace if it is executed with preemption enabled
148 */
149# define cant_sleep() \
150 do { __cant_sleep(__FILE__, __LINE__, 0); } while (0)
151# define sched_annotate_sleep() (current->task_state_change = 0)
152
153/**
154 * cant_migrate - annotation for functions that cannot migrate
155 *
156 * Will print a stack trace if executed in code which is migratable
157 */
158# define cant_migrate() \
159 do { \
160 if (IS_ENABLED(CONFIG_SMP)) \
161 __cant_migrate(__FILE__, __LINE__); \
162 } while (0)
163
164/**
165 * non_block_start - annotate the start of section where sleeping is prohibited
166 *
167 * This is on behalf of the oom reaper, specifically when it is calling the mmu
168 * notifiers. The problem is that if the notifier were to block on, for example,
169 * mutex_lock() and if the process which holds that mutex were to perform a
170 * sleeping memory allocation, the oom reaper is now blocked on completion of
171 * that memory allocation. Other blocking calls like wait_event() pose similar
172 * issues.
173 */
174# define non_block_start() (current->non_block_count++)
175/**
176 * non_block_end - annotate the end of section where sleeping is prohibited
177 *
178 * Closes a section opened by non_block_start().
179 */
180# define non_block_end() WARN_ON(current->non_block_count-- == 0)
181#else
182 static inline void __might_resched(const char *file, int line,
183 unsigned int offsets) { }
184static inline void __might_sleep(const char *file, int line) { }
185# define might_sleep() do { might_resched(); } while (0)
186# define cant_sleep() do { } while (0)
187# define cant_migrate() do { } while (0)
188# define sched_annotate_sleep() do { } while (0)
189# define non_block_start() do { } while (0)
190# define non_block_end() do { } while (0)
191#endif
192
193#define might_sleep_if(cond) do { if (cond) might_sleep(); } while (0)
194
195#if defined(CONFIG_MMU) && \
196 (defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_DEBUG_ATOMIC_SLEEP))
197#define might_fault() __might_fault(__FILE__, __LINE__)
198void __might_fault(const char *file, int line);
199#else
200static inline void might_fault(void) { }
201#endif
202
203void do_exit(long error_code) __noreturn;
204
205extern int num_to_str(char *buf, int size,
206 unsigned long long num, unsigned int width);
207
208/* lib/printf utilities */
209
210extern __printf(2, 3) int sprintf(char *buf, const char * fmt, ...);
211extern __printf(2, 0) int vsprintf(char *buf, const char *, va_list);
212extern __printf(3, 4)
213int snprintf(char *buf, size_t size, const char *fmt, ...);
214extern __printf(3, 0)
215int vsnprintf(char *buf, size_t size, const char *fmt, va_list args);
216extern __printf(3, 4)
217int scnprintf(char *buf, size_t size, const char *fmt, ...);
218extern __printf(3, 0)
219int vscnprintf(char *buf, size_t size, const char *fmt, va_list args);
220extern __printf(2, 3) __malloc
221char *kasprintf(gfp_t gfp, const char *fmt, ...);
222extern __printf(2, 0) __malloc
223char *kvasprintf(gfp_t gfp, const char *fmt, va_list args);
224extern __printf(2, 0)
225const char *kvasprintf_const(gfp_t gfp, const char *fmt, va_list args);
226
227extern __scanf(2, 3)
228int sscanf(const char *, const char *, ...);
229extern __scanf(2, 0)
230int vsscanf(const char *, const char *, va_list);
231
232extern int no_hash_pointers_enable(char *str);
233
234extern int get_option(char **str, int *pint);
235extern char *get_options(const char *str, int nints, int *ints);
236extern unsigned long long memparse(const char *ptr, char **retptr);
237extern bool parse_option_str(const char *str, const char *option);
238extern char *next_arg(char *args, char **param, char **val);
239
240extern int core_kernel_text(unsigned long addr);
241extern int __kernel_text_address(unsigned long addr);
242extern int kernel_text_address(unsigned long addr);
243extern int func_ptr_is_kernel_text(void *ptr);
244
245extern void bust_spinlocks(int yes);
246
247extern int root_mountflags;
248
249extern bool early_boot_irqs_disabled;
250
251/*
252 * Values used for system_state. Ordering of the states must not be changed
253 * as code checks for <, <=, >, >= STATE.
254 */
255extern enum system_states {
256 SYSTEM_BOOTING,
257 SYSTEM_SCHEDULING,
258 SYSTEM_FREEING_INITMEM,
259 SYSTEM_RUNNING,
260 SYSTEM_HALT,
261 SYSTEM_POWER_OFF,
262 SYSTEM_RESTART,
263 SYSTEM_SUSPEND,
264} system_state;
265
266extern const char hex_asc[];
267#define hex_asc_lo(x) hex_asc[((x) & 0x0f)]
268#define hex_asc_hi(x) hex_asc[((x) & 0xf0) >> 4]
269
270static inline char *hex_byte_pack(char *buf, u8 byte)
271{
272 *buf++ = hex_asc_hi(byte);
273 *buf++ = hex_asc_lo(byte);
274 return buf;
275}
276
277extern const char hex_asc_upper[];
278#define hex_asc_upper_lo(x) hex_asc_upper[((x) & 0x0f)]
279#define hex_asc_upper_hi(x) hex_asc_upper[((x) & 0xf0) >> 4]
280
281static inline char *hex_byte_pack_upper(char *buf, u8 byte)
282{
283 *buf++ = hex_asc_upper_hi(byte);
284 *buf++ = hex_asc_upper_lo(byte);
285 return buf;
286}
287
288extern int hex_to_bin(unsigned char ch);
289extern int __must_check hex2bin(u8 *dst, const char *src, size_t count);
290extern char *bin2hex(char *dst, const void *src, size_t count);
291
292bool mac_pton(const char *s, u8 *mac);
293
294/*
295 * General tracing related utility functions - trace_printk(),
296 * tracing_on/tracing_off and tracing_start()/tracing_stop
297 *
298 * Use tracing_on/tracing_off when you want to quickly turn on or off
299 * tracing. It simply enables or disables the recording of the trace events.
300 * This also corresponds to the user space /sys/kernel/debug/tracing/tracing_on
301 * file, which gives a means for the kernel and userspace to interact.
302 * Place a tracing_off() in the kernel where you want tracing to end.
303 * From user space, examine the trace, and then echo 1 > tracing_on
304 * to continue tracing.
305 *
306 * tracing_stop/tracing_start has slightly more overhead. It is used
307 * by things like suspend to ram where disabling the recording of the
308 * trace is not enough, but tracing must actually stop because things
309 * like calling smp_processor_id() may crash the system.
310 *
311 * Most likely, you want to use tracing_on/tracing_off.
312 */
313
314enum ftrace_dump_mode {
315 DUMP_NONE,
316 DUMP_ALL,
317 DUMP_ORIG,
318};
319
320#ifdef CONFIG_TRACING
321void tracing_on(void);
322void tracing_off(void);
323int tracing_is_on(void);
324void tracing_snapshot(void);
325void tracing_snapshot_alloc(void);
326
327extern void tracing_start(void);
328extern void tracing_stop(void);
329
330static inline __printf(1, 2)
331void ____trace_printk_check_format(const char *fmt, ...)
332{
333}
334#define __trace_printk_check_format(fmt, args...) \
335do { \
336 if (0) \
337 ____trace_printk_check_format(fmt, ##args); \
338} while (0)
339
340/**
341 * trace_printk - printf formatting in the ftrace buffer
342 * @fmt: the printf format for printing
343 *
344 * Note: __trace_printk is an internal function for trace_printk() and
345 * the @ip is passed in via the trace_printk() macro.
346 *
347 * This function allows a kernel developer to debug fast path sections
348 * that printk is not appropriate for. By scattering in various
349 * printk like tracing in the code, a developer can quickly see
350 * where problems are occurring.
351 *
352 * This is intended as a debugging tool for the developer only.
353 * Please refrain from leaving trace_printks scattered around in
354 * your code. (Extra memory is used for special buffers that are
355 * allocated when trace_printk() is used.)
356 *
357 * A little optimization trick is done here. If there's only one
358 * argument, there's no need to scan the string for printf formats.
359 * The trace_puts() will suffice. But how can we take advantage of
360 * using trace_puts() when trace_printk() has only one argument?
361 * By stringifying the args and checking the size we can tell
362 * whether or not there are args. __stringify((__VA_ARGS__)) will
363 * turn into "()\0" with a size of 3 when there are no args, anything
364 * else will be bigger. All we need to do is define a string to this,
365 * and then take its size and compare to 3. If it's bigger, use
366 * do_trace_printk() otherwise, optimize it to trace_puts(). Then just
367 * let gcc optimize the rest.
368 */
369
370#define trace_printk(fmt, ...) \
371do { \
372 char _______STR[] = __stringify((__VA_ARGS__)); \
373 if (sizeof(_______STR) > 3) \
374 do_trace_printk(fmt, ##__VA_ARGS__); \
375 else \
376 trace_puts(fmt); \
377} while (0)
378
379#define do_trace_printk(fmt, args...) \
380do { \
381 static const char *trace_printk_fmt __used \
382 __section("__trace_printk_fmt") = \
383 __builtin_constant_p(fmt) ? fmt : NULL; \
384 \
385 __trace_printk_check_format(fmt, ##args); \
386 \
387 if (__builtin_constant_p(fmt)) \
388 __trace_bprintk(_THIS_IP_, trace_printk_fmt, ##args); \
389 else \
390 __trace_printk(_THIS_IP_, fmt, ##args); \
391} while (0)
392
393extern __printf(2, 3)
394int __trace_bprintk(unsigned long ip, const char *fmt, ...);
395
396extern __printf(2, 3)
397int __trace_printk(unsigned long ip, const char *fmt, ...);
398
399/**
400 * trace_puts - write a string into the ftrace buffer
401 * @str: the string to record
402 *
403 * Note: __trace_bputs is an internal function for trace_puts and
404 * the @ip is passed in via the trace_puts macro.
405 *
406 * This is similar to trace_printk() but is made for those really fast
407 * paths that a developer wants the least amount of "Heisenbug" effects,
408 * where the processing of the print format is still too much.
409 *
410 * This function allows a kernel developer to debug fast path sections
411 * that printk is not appropriate for. By scattering in various
412 * printk like tracing in the code, a developer can quickly see
413 * where problems are occurring.
414 *
415 * This is intended as a debugging tool for the developer only.
416 * Please refrain from leaving trace_puts scattered around in
417 * your code. (Extra memory is used for special buffers that are
418 * allocated when trace_puts() is used.)
419 *
420 * Returns: 0 if nothing was written, positive # if string was.
421 * (1 when __trace_bputs is used, strlen(str) when __trace_puts is used)
422 */
423
424#define trace_puts(str) ({ \
425 static const char *trace_printk_fmt __used \
426 __section("__trace_printk_fmt") = \
427 __builtin_constant_p(str) ? str : NULL; \
428 \
429 if (__builtin_constant_p(str)) \
430 __trace_bputs(_THIS_IP_, trace_printk_fmt); \
431 else \
432 __trace_puts(_THIS_IP_, str, strlen(str)); \
433})
434extern int __trace_bputs(unsigned long ip, const char *str);
435extern int __trace_puts(unsigned long ip, const char *str, int size);
436
437extern void trace_dump_stack(int skip);
438
439/*
440 * The double __builtin_constant_p is because gcc will give us an error
441 * if we try to allocate the static variable to fmt if it is not a
442 * constant. Even with the outer if statement.
443 */
444#define ftrace_vprintk(fmt, vargs) \
445do { \
446 if (__builtin_constant_p(fmt)) { \
447 static const char *trace_printk_fmt __used \
448 __section("__trace_printk_fmt") = \
449 __builtin_constant_p(fmt) ? fmt : NULL; \
450 \
451 __ftrace_vbprintk(_THIS_IP_, trace_printk_fmt, vargs); \
452 } else \
453 __ftrace_vprintk(_THIS_IP_, fmt, vargs); \
454} while (0)
455
456extern __printf(2, 0) int
457__ftrace_vbprintk(unsigned long ip, const char *fmt, va_list ap);
458
459extern __printf(2, 0) int
460__ftrace_vprintk(unsigned long ip, const char *fmt, va_list ap);
461
462extern void ftrace_dump(enum ftrace_dump_mode oops_dump_mode);
463#else
464static inline void tracing_start(void) { }
465static inline void tracing_stop(void) { }
466static inline void trace_dump_stack(int skip) { }
467
468static inline void tracing_on(void) { }
469static inline void tracing_off(void) { }
470static inline int tracing_is_on(void) { return 0; }
471static inline void tracing_snapshot(void) { }
472static inline void tracing_snapshot_alloc(void) { }
473
474static inline __printf(1, 2)
475int trace_printk(const char *fmt, ...)
476{
477 return 0;
478}
479static __printf(1, 0) inline int
480ftrace_vprintk(const char *fmt, va_list ap)
481{
482 return 0;
483}
484static inline void ftrace_dump(enum ftrace_dump_mode oops_dump_mode) { }
485#endif /* CONFIG_TRACING */
486
487/* This counts to 12. Any more, it will return 13th argument. */
488#define __COUNT_ARGS(_0, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, _11, _12, _n, X...) _n
489#define COUNT_ARGS(X...) __COUNT_ARGS(, ##X, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0)
490
491#define __CONCAT(a, b) a ## b
492#define CONCATENATE(a, b) __CONCAT(a, b)
493
494/* Rebuild everything on CONFIG_FTRACE_MCOUNT_RECORD */
495#ifdef CONFIG_FTRACE_MCOUNT_RECORD
496# define REBUILD_DUE_TO_FTRACE_MCOUNT_RECORD
497#endif
498
499/* Permissions on a sysfs file: you didn't miss the 0 prefix did you? */
500#define VERIFY_OCTAL_PERMISSIONS(perms) \
501 (BUILD_BUG_ON_ZERO((perms) < 0) + \
502 BUILD_BUG_ON_ZERO((perms) > 0777) + \
503 /* USER_READABLE >= GROUP_READABLE >= OTHER_READABLE */ \
504 BUILD_BUG_ON_ZERO((((perms) >> 6) & 4) < (((perms) >> 3) & 4)) + \
505 BUILD_BUG_ON_ZERO((((perms) >> 3) & 4) < ((perms) & 4)) + \
506 /* USER_WRITABLE >= GROUP_WRITABLE */ \
507 BUILD_BUG_ON_ZERO((((perms) >> 6) & 2) < (((perms) >> 3) & 2)) + \
508 /* OTHER_WRITABLE? Generally considered a bad idea. */ \
509 BUILD_BUG_ON_ZERO((perms) & 2) + \
510 (perms))
511#endif
512

source code of linux/include/linux/kernel.h