xref: /freebsd-14-stable/sys/sys/kernel.h (revision 467fa302c3aeefd7ccd9fa6bd0ad0381d86cd75b)
1 /*-
2  * SPDX-License-Identifier: BSD-4-Clause
3  *
4  * Copyright (c) 1995 Terrence R. Lambert
5  * All rights reserved.
6  *
7  * Copyright (c) 1990, 1993
8  *	The Regents of the University of California.  All rights reserved.
9  * (c) UNIX System Laboratories, Inc.
10  * All or some portions of this file are derived from material licensed
11  * to the University of California by American Telephone and Telegraph
12  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
13  * the permission of UNIX System Laboratories, Inc.
14  *
15  * Redistribution and use in source and binary forms, with or without
16  * modification, are permitted provided that the following conditions
17  * are met:
18  * 1. Redistributions of source code must retain the above copyright
19  *    notice, this list of conditions and the following disclaimer.
20  * 2. Redistributions in binary form must reproduce the above copyright
21  *    notice, this list of conditions and the following disclaimer in the
22  *    documentation and/or other materials provided with the distribution.
23  * 3. All advertising materials mentioning features or use of this software
24  *    must display the following acknowledgement:
25  *	This product includes software developed by the University of
26  *	California, Berkeley and its contributors.
27  * 4. Neither the name of the University nor the names of its contributors
28  *    may be used to endorse or promote products derived from this software
29  *    without specific prior written permission.
30  *
31  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
32  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
33  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
34  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
35  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
36  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
37  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
38  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
39  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
40  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
41  * SUCH DAMAGE.
42  *
43  *	@(#)kernel.h	8.3 (Berkeley) 1/21/94
44  */
45 
46 #ifndef _SYS_KERNEL_H_
47 #define	_SYS_KERNEL_H_
48 
49 #include <sys/linker_set.h>
50 
51 #ifdef _KERNEL
52 
53 /* for intrhook and sysinit linked list below */
54 #include <sys/queue.h>
55 
56 /* for timestamping SYSINITs; other files may assume this is included here */
57 #include <sys/tslog.h>
58 
59 /* Global variables for the kernel. */
60 
61 /* 1.1 */
62 extern char kernelname[MAXPATHLEN];
63 
64 extern int tick;			/* usec per tick (1000000 / hz) */
65 extern int hz;				/* system clock's frequency */
66 extern int psratio;			/* ratio: prof / stat */
67 extern int stathz;			/* statistics clock's frequency */
68 extern int profhz;			/* profiling clock's frequency */
69 extern int profprocs;			/* number of process's profiling */
70 
71 /*
72  * The ticks and ticksl symbols overlap, giving a 64-bit tick counter on 64-bit
73  * platforms while still maintaining compatibility with the legacy 32-bit
74  * counter.  Either value can be used, but rollover must be handled; at 1000Hz,
75  * ticks (and ticksl on 32-bit platforms) roll over roughly every 25 days.  On
76  * 64-bit platforms, ticksl will not roll over in the foreseeable future.
77  */
78 extern volatile int ticks;
79 extern volatile long ticksl;
80 
81 #endif /* _KERNEL */
82 
83 /*
84  * Enumerated types for known system startup interfaces.
85  *
86  * Startup occurs in ascending numeric order; the list entries are
87  * sorted prior to attempting startup to guarantee order.  Items
88  * of the same level are arbitrated for order based on the 'order'
89  * element.
90  *
91  * These numbers are arbitrary and are chosen ONLY for ordering; the
92  * enumeration values are explicit rather than implicit to provide
93  * for binary compatibility with inserted elements.
94  *
95  * The SI_SUB_LAST value must have the highest lexical value.
96  */
97 enum sysinit_sub_id {
98 	SI_SUB_DUMMY		= 0x0000000,	/* not executed; for linker */
99 	SI_SUB_TUNABLES		= 0x0700000,	/* establish tunable values */
100 	SI_SUB_COPYRIGHT	= 0x0800001,	/* first use of console */
101 	SI_SUB_VM		= 0x1000000,	/* virtual memory system init */
102 	SI_SUB_COUNTER		= 0x1100000,	/* counter(9) is initialized */
103 	SI_SUB_KMEM		= 0x1800000,	/* kernel memory */
104 	SI_SUB_HYPERVISOR	= 0x1A40000,	/*
105 						 * Hypervisor detection and
106 						 * virtualization support
107 						 * setup.
108 						 */
109 	SI_SUB_WITNESS		= 0x1A80000,	/* witness initialization */
110 	SI_SUB_MTX_POOL_DYNAMIC	= 0x1AC0000,	/* dynamic mutex pool */
111 	SI_SUB_LOCK		= 0x1B00000,	/* various locks */
112 	SI_SUB_EVENTHANDLER	= 0x1C00000,	/* eventhandler init */
113 	SI_SUB_VNET_PRELINK	= 0x1E00000,	/* vnet init before modules */
114 	SI_SUB_KLD		= 0x2000000,	/* KLD and module setup */
115 	SI_SUB_KHELP		= 0x2080000,	/* khelp modules */
116 	SI_SUB_CPU		= 0x2100000,	/* CPU resource(s) */
117 	SI_SUB_RACCT		= 0x2110000,	/* resource accounting */
118 	SI_SUB_KDTRACE		= 0x2140000,	/* Kernel dtrace hooks */
119 	SI_SUB_RANDOM		= 0x2160000,	/* random number generator */
120 	SI_SUB_MAC		= 0x2180000,	/* TrustedBSD MAC subsystem */
121 	SI_SUB_MAC_POLICY	= 0x21C0000,	/* TrustedBSD MAC policies */
122 	SI_SUB_MAC_LATE		= 0x21D0000,	/* TrustedBSD MAC subsystem */
123 	SI_SUB_VNET		= 0x21E0000,	/* vnet 0 */
124 	SI_SUB_INTRINSIC	= 0x2200000,	/* proc 0 */
125 	SI_SUB_VM_CONF		= 0x2300000,	/* config VM, set limits */
126 	SI_SUB_DDB_SERVICES	= 0x2380000,	/* capture, scripting, etc. */
127 	SI_SUB_RUN_QUEUE	= 0x2400000,	/* set up run queue */
128 	SI_SUB_KTRACE		= 0x2480000,	/* ktrace */
129 	SI_SUB_OPENSOLARIS	= 0x2490000,	/* OpenSolaris compatibility */
130 	SI_SUB_AUDIT		= 0x24C0000,	/* audit */
131 	SI_SUB_CREATE_INIT	= 0x2500000,	/* create init process */
132 	SI_SUB_SCHED_IDLE	= 0x2600000,	/* required idle procs */
133 	SI_SUB_MBUF		= 0x2700000,	/* mbuf subsystem */
134 	SI_SUB_INTR		= 0x2800000,	/* interrupt threads */
135 	SI_SUB_TASKQ		= 0x2880000,	/* task queues */
136 	SI_SUB_EPOCH		= 0x2888000,	/* epoch subsystem */
137 #ifdef EARLY_AP_STARTUP
138 	SI_SUB_SMP		= 0x2900000,	/* start the APs */
139 #endif
140 	SI_SUB_SOFTINTR		= 0x2A00000,	/* start soft interrupt thread */
141 	SI_SUB_DEVFS		= 0x2F00000,	/* devfs ready for devices */
142 	SI_SUB_INIT_IF		= 0x3000000,	/* prep for net interfaces */
143 	SI_SUB_NETGRAPH		= 0x3010000,	/* Let Netgraph initialize */
144 	SI_SUB_DTRACE		= 0x3020000,	/* DTrace subsystem */
145 	SI_SUB_DTRACE_PROVIDER	= 0x3048000,	/* DTrace providers */
146 	SI_SUB_DTRACE_ANON	= 0x308C000,	/* DTrace anon enabling */
147 	SI_SUB_DRIVERS		= 0x3100000,	/* Let Drivers initialize */
148 	SI_SUB_CONFIGURE	= 0x3800000,	/* Configure devices */
149 	SI_SUB_VFS		= 0x4000000,	/* virtual filesystem */
150 	SI_SUB_CLOCKS		= 0x4800000,	/* real time and stat clocks */
151 	SI_SUB_SYSV_SHM		= 0x6400000,	/* System V shared memory */
152 	SI_SUB_SYSV_SEM		= 0x6800000,	/* System V semaphores */
153 	SI_SUB_SYSV_MSG		= 0x6C00000,	/* System V message queues */
154 	SI_SUB_P1003_1B		= 0x6E00000,	/* P1003.1B realtime */
155 	SI_SUB_PSEUDO		= 0x7000000,	/* pseudo devices */
156 	SI_SUB_EXEC		= 0x7400000,	/* execve() handlers */
157 	SI_SUB_PROTO_BEGIN	= 0x8000000,	/* VNET initialization */
158 	SI_SUB_PROTO_PFIL	= 0x8100000,	/* Initialize pfil before FWs */
159 	SI_SUB_PROTO_IF		= 0x8400000,	/* interfaces */
160 	SI_SUB_PROTO_DOMAININIT	= 0x8600000,	/* domain registration system */
161 	SI_SUB_PROTO_MC		= 0x8700000,	/* Multicast */
162 	SI_SUB_PROTO_DOMAIN	= 0x8800000,	/* domains (address families?) */
163 	SI_SUB_PROTO_FIREWALL	= 0x8806000,	/* Firewalls */
164 	SI_SUB_PROTO_IFATTACHDOMAIN = 0x8808000,/* domain dependent data init */
165 	SI_SUB_PROTO_END	= 0x8ffffff,	/* VNET helper functions */
166 	SI_SUB_KPROF		= 0x9000000,	/* kernel profiling */
167 	SI_SUB_KICK_SCHEDULER	= 0xa000000,	/* start the timeout events */
168 	SI_SUB_INT_CONFIG_HOOKS	= 0xa800000,	/* Interrupts enabled config */
169 	SI_SUB_ROOT_CONF	= 0xb000000,	/* Find root devices */
170 	SI_SUB_INTRINSIC_POST	= 0xd000000,	/* proc 0 cleanup */
171 	SI_SUB_SYSCALLS		= 0xd800000,	/* register system calls */
172 	SI_SUB_VNET_DONE	= 0xdc00000,	/* vnet registration complete */
173 	SI_SUB_KTHREAD_INIT	= 0xe000000,	/* init process */
174 	SI_SUB_KTHREAD_PAGE	= 0xe400000,	/* pageout daemon */
175 	SI_SUB_KTHREAD_VM	= 0xe800000,	/* vm daemon */
176 	SI_SUB_KTHREAD_BUF	= 0xea00000,	/* buffer daemon */
177 	SI_SUB_KTHREAD_UPDATE	= 0xec00000,	/* update daemon */
178 	SI_SUB_KTHREAD_IDLE	= 0xee00000,	/* idle procs */
179 #ifndef EARLY_AP_STARTUP
180 	SI_SUB_SMP		= 0xf000000,	/* start the APs */
181 #endif
182 	SI_SUB_RACCTD		= 0xf100000,	/* start racctd */
183 	SI_SUB_LAST		= 0xfffffff	/* final initialization */
184 };
185 
186 /*
187  * Some enumerated orders; "ANY" sorts last.
188  */
189 enum sysinit_elem_order {
190 	SI_ORDER_FIRST		= 0x0000000,	/* first */
191 	SI_ORDER_SECOND		= 0x0000001,	/* second */
192 	SI_ORDER_THIRD		= 0x0000002,	/* third */
193 	SI_ORDER_FOURTH		= 0x0000003,	/* fourth */
194 	SI_ORDER_FIFTH		= 0x0000004,	/* fifth */
195 	SI_ORDER_SIXTH		= 0x0000005,	/* sixth */
196 	SI_ORDER_SEVENTH	= 0x0000006,	/* seventh */
197 	SI_ORDER_EIGHTH		= 0x0000007,	/* eighth */
198 	SI_ORDER_MIDDLE		= 0x1000000,	/* somewhere in the middle */
199 	SI_ORDER_ANY		= 0xfffffff	/* last */
200 };
201 
202 /*
203  * A system initialization call instance
204  *
205  * At the moment there is one instance of sysinit.  We probably do not
206  * want two which is why this code is if'd out, but we definitely want
207  * to discern SYSINIT's which take non-constant data pointers and
208  * SYSINIT's which take constant data pointers,
209  *
210  * The C_* macros take functions expecting const void * arguments
211  * while the non-C_* macros take functions expecting just void * arguments.
212  *
213  * With -Wcast-qual on, the compiler issues warnings:
214  *	- if we pass non-const data or functions taking non-const data
215  *	  to a C_* macro.
216  *
217  *	- if we pass const data to the normal macros
218  *
219  * However, no warning is issued if we pass a function taking const data
220  * through a normal non-const macro.  This is ok because the function is
221  * saying it won't modify the data so we don't care whether the data is
222  * modifiable or not.
223  */
224 
225 typedef void (*sysinit_nfunc_t)(void *);
226 typedef void (*sysinit_cfunc_t)(const void *);
227 
228 struct sysinit {
229 	enum sysinit_sub_id	subsystem;	/* subsystem identifier */
230 	enum sysinit_elem_order	order;		/* init order within subsystem */
231 	STAILQ_ENTRY(sysinit)	next;		/* singly-linked list */
232 	sysinit_cfunc_t func;			/* function */
233 	const void	*udata;			/* multiplexer/argument */
234 };
235 
236 /*
237  * Default: no special processing
238  *
239  * The C_ version of SYSINIT is for data pointers to const
240  * data ( and functions taking data pointers to const data ).
241  * At the moment it is no different from SYSINIT and thus
242  * still results in warnings.
243  *
244  * The casts are necessary to have the compiler produce the
245  * correct warnings when -Wcast-qual is used.
246  *
247  */
248 #ifdef TSLOG
249 struct sysinit_tslog {
250 	sysinit_cfunc_t func;
251 	const void *data;
252 	const char *name;
253 };
254 static inline void
sysinit_tslog_shim(const void * data)255 sysinit_tslog_shim(const void *data)
256 {
257 	const struct sysinit_tslog *x = data;
258 
259 	TSRAW(curthread, TS_ENTER, "SYSINIT", x->name);
260 	(x->func)(x->data);
261 	TSRAW(curthread, TS_EXIT, "SYSINIT", x->name);
262 }
263 #define	C_SYSINIT(uniquifier, subsystem, order, func, ident)	\
264 	static struct sysinit_tslog uniquifier ## _sys_init_tslog = {	\
265 		func,						\
266 		(ident),					\
267 		#uniquifier					\
268 	};							\
269 	static struct sysinit uniquifier ## _sys_init = {	\
270 		subsystem,					\
271 		order,						\
272 		{ NULL },					\
273 		sysinit_tslog_shim,				\
274 		&uniquifier ## _sys_init_tslog			\
275 	};							\
276 	DATA_WSET(sysinit_set, uniquifier ## _sys_init)
277 #else
278 #define	C_SYSINIT(uniquifier, subsystem, order, func, ident)	\
279 	static struct sysinit uniquifier ## _sys_init = {	\
280 		subsystem,					\
281 		order,						\
282 		{ NULL },					\
283 		func,						\
284 		(ident)						\
285 	};							\
286 	DATA_WSET(sysinit_set, uniquifier ## _sys_init)
287 #endif
288 
289 #define	SYSINIT(uniquifier, subsystem, order, func, ident)	\
290 	C_SYSINIT(uniquifier, subsystem, order,			\
291 	(sysinit_cfunc_t)(sysinit_nfunc_t)func, (void *)(ident))
292 
293 /*
294  * Called on module unload: no special processing
295  */
296 #define	C_SYSUNINIT(uniquifier, subsystem, order, func, ident)	\
297 	static struct sysinit uniquifier ## _sys_uninit = {	\
298 		subsystem,					\
299 		order,						\
300 		{ NULL },					\
301 		func,						\
302 		(ident)						\
303 	};							\
304 	DATA_WSET(sysuninit_set, uniquifier ## _sys_uninit)
305 
306 #define	SYSUNINIT(uniquifier, subsystem, order, func, ident)	\
307 	C_SYSUNINIT(uniquifier, subsystem, order,		\
308 	(sysinit_cfunc_t)(sysinit_nfunc_t)func, (void *)(ident))
309 
310 void	sysinit_add(struct sysinit **set, struct sysinit **set_end);
311 
312 #ifdef _KERNEL
313 
314 /*
315  * Infrastructure for tunable 'constants'.  Value may be specified at compile
316  * time or kernel load time.  Rules relating tunables together can be placed
317  * in a SYSINIT function at SI_SUB_TUNABLES with SI_ORDER_ANY.
318  *
319  * WARNING: developers should never use the reserved suffixes specified in
320  * loader.conf(5) for any tunables or conflicts will result.
321  */
322 
323 /*
324  * int
325  * please avoid using for new tunables!
326  */
327 extern void tunable_int_init(void *);
328 struct tunable_int {
329 	const char *path;
330 	int *var;
331 };
332 #define	TUNABLE_INT(path, var)					\
333 	static struct tunable_int __CONCAT(__tunable_int_, __LINE__) = { \
334 		(path),						\
335 		(var),						\
336 	};							\
337 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
338 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_int_init,	\
339 	    &__CONCAT(__tunable_int_, __LINE__))
340 
341 #define	TUNABLE_INT_FETCH(path, var)	getenv_int((path), (var))
342 
343 /*
344  * long
345  */
346 extern void tunable_long_init(void *);
347 struct tunable_long {
348 	const char *path;
349 	long *var;
350 };
351 #define	TUNABLE_LONG(path, var)					\
352 	static struct tunable_long __CONCAT(__tunable_long_, __LINE__) = { \
353 		(path),						\
354 		(var),						\
355 	};							\
356 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
357 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_long_init,\
358 	    &__CONCAT(__tunable_long_, __LINE__))
359 
360 #define	TUNABLE_LONG_FETCH(path, var)	getenv_long((path), (var))
361 
362 /*
363  * unsigned long
364  */
365 extern void tunable_ulong_init(void *);
366 struct tunable_ulong {
367 	const char *path;
368 	unsigned long *var;
369 };
370 #define	TUNABLE_ULONG(path, var)				\
371 	static struct tunable_ulong __CONCAT(__tunable_ulong_, __LINE__) = { \
372 		(path),						\
373 		(var),						\
374 	};							\
375 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
376 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_ulong_init, \
377 	    &__CONCAT(__tunable_ulong_, __LINE__))
378 
379 #define	TUNABLE_ULONG_FETCH(path, var)	getenv_ulong((path), (var))
380 
381 /*
382  * int64_t
383  */
384 extern void tunable_int64_init(void *);
385 struct tunable_int64 {
386 	const char *path;
387 	int64_t *var;
388 };
389 #define	TUNABLE_INT64(path, var)				\
390 	static struct tunable_int64 __CONCAT(__tunable_int64_, __LINE__) = { \
391 		(path),						\
392 		(var),						\
393 	};							\
394 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
395 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_int64_init, \
396 	    &__CONCAT(__tunable_int64_, __LINE__))
397 
398 #define	TUNABLE_INT64_FETCH(path, var)	getenv_int64((path), (var))
399 
400 /*
401  * uint64_t
402  */
403 extern void tunable_uint64_init(void *);
404 struct tunable_uint64 {
405 	const char *path;
406 	uint64_t *var;
407 };
408 #define	TUNABLE_UINT64(path, var)				\
409 	static struct tunable_uint64 __CONCAT(__tunable_uint64_, __LINE__) = { \
410 		(path),						\
411 		(var),						\
412 	};							\
413 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
414 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_uint64_init, \
415 	    &__CONCAT(__tunable_uint64_, __LINE__))
416 
417 #define	TUNABLE_UINT64_FETCH(path, var)	getenv_uint64((path), (var))
418 
419 /*
420  * quad
421  */
422 extern void tunable_quad_init(void *);
423 struct tunable_quad {
424 	const char *path;
425 	quad_t *var;
426 };
427 #define	TUNABLE_QUAD(path, var)					\
428 	static struct tunable_quad __CONCAT(__tunable_quad_, __LINE__) = { \
429 		(path),						\
430 		(var),						\
431 	};							\
432 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
433 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_quad_init, \
434 	    &__CONCAT(__tunable_quad_, __LINE__))
435 
436 #define	TUNABLE_QUAD_FETCH(path, var)	getenv_quad((path), (var))
437 
438 /*
439  * bool
440  */
441 extern void tunable_bool_init(void *);
442 struct tunable_bool {
443 	const char *path;
444 	bool *var;
445 };
446 #define	TUNABLE_BOOL(path, var) \
447 	static struct tunable_bool __CONCAT(__tunable_bool_, __LINE__) = { \
448 		(path),						\
449 		(var),						\
450 	};							\
451 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
452 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_bool_init, \
453 	    &__CONCAT(__tunable_bool_, __LINE__))
454 
455 #define	TUNABLE_BOOL_FETCH(path, var)	getenv_bool((path), (var))
456 
457 extern void tunable_str_init(void *);
458 struct tunable_str {
459 	const char *path;
460 	char *var;
461 	int size;
462 };
463 #define	TUNABLE_STR(path, var, size)				\
464 	static struct tunable_str __CONCAT(__tunable_str_, __LINE__) = { \
465 		(path),						\
466 		(var),						\
467 		(size),						\
468 	};							\
469 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
470 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_str_init,	\
471 	    &__CONCAT(__tunable_str_, __LINE__))
472 
473 #define	TUNABLE_STR_FETCH(path, var, size)			\
474 	getenv_string((path), (var), (size))
475 
476 #endif /* _KERNEL */
477 
478 typedef void (*ich_func_t)(void *_arg);
479 
480 struct intr_config_hook {
481 	STAILQ_ENTRY(intr_config_hook) ich_links;
482 	uintptr_t	ich_state;
483 #define ICHS_QUEUED	0x1
484 #define ICHS_RUNNING	0x2
485 #define	ICHS_DONE	0x3
486 	ich_func_t	ich_func;
487 	void		*ich_arg;
488 };
489 
490 int	config_intrhook_establish(struct intr_config_hook *hook);
491 void	config_intrhook_disestablish(struct intr_config_hook *hook);
492 int	config_intrhook_drain(struct intr_config_hook *hook);
493 void	config_intrhook_oneshot(ich_func_t _func, void *_arg);
494 
495 #endif /* !_SYS_KERNEL_H_*/
496