xref: /freebsd/sys/sys/kernel.h (revision a77edeb0e54ad3e96e2a6c685d5c26eac4ba190a)
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 
44 #ifndef _SYS_KERNEL_H_
45 #define	_SYS_KERNEL_H_
46 
47 #include <sys/linker_set.h>
48 
49 #ifdef _KERNEL
50 
51 /* for intrhook and sysinit linked list below */
52 #include <sys/queue.h>
53 
54 /* for timestamping SYSINITs; other files may assume this is included here */
55 #include <sys/tslog.h>
56 
57 /* Global variables for the kernel. */
58 
59 /* 1.1 */
60 extern char kernelname[MAXPATHLEN];
61 
62 extern int tick;			/* usec per tick (1000000 / hz) */
63 extern int hz;				/* system clock's frequency */
64 extern int psratio;			/* ratio: prof / stat */
65 extern int stathz;			/* statistics clock's frequency */
66 extern int profhz;			/* profiling clock's frequency */
67 extern int profprocs;			/* number of process's profiling */
68 
69 /*
70  * The ticks and ticksl symbols overlap, giving a 64-bit tick counter on 64-bit
71  * platforms while still maintaining compatibility with the legacy 32-bit
72  * counter.  Either value can be used, but rollover must be handled; at 1000Hz,
73  * ticks (and ticksl on 32-bit platforms) roll over roughly every 25 days.  On
74  * 64-bit platforms, ticksl will not roll over in the foreseeable future.
75  */
76 extern volatile int ticks;
77 extern volatile long ticksl;
78 
79 #endif /* _KERNEL */
80 
81 /*
82  * Enumerated types for known system startup interfaces.
83  *
84  * Startup occurs in ascending numeric order; the list entries are
85  * sorted prior to attempting startup to guarantee order.  Items
86  * of the same level are arbitrated for order based on the 'order'
87  * element.
88  *
89  * These numbers are arbitrary and are chosen ONLY for ordering; the
90  * enumeration values are explicit rather than implicit to provide
91  * for binary compatibility with inserted elements.
92  *
93  * The SI_SUB_LAST value must have the highest lexical value.
94  */
95 enum sysinit_sub_id {
96 	SI_SUB_DUMMY		= 0x0000000,	/* not executed; for linker */
97 	SI_SUB_FIRST		= 0x0600000,	/* mp_maxid, etc... */
98 	SI_SUB_TUNABLES		= 0x0700000,	/* establish tunable values */
99 	SI_SUB_COPYRIGHT	= 0x0800001,	/* first use of console */
100 	SI_SUB_NUMA		= 0x0810000,	/* initialize NUMA topology */
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_KENV		= 0x1900000,	/* kernel environment */
105 	SI_SUB_HYPERVISOR	= 0x1A40000,	/*
106 						 * Hypervisor detection and
107 						 * virtualization support
108 						 * setup.
109 						 */
110 	SI_SUB_WITNESS		= 0x1A80000,	/* witness initialization */
111 	SI_SUB_MTX_POOL_DYNAMIC	= 0x1AC0000,	/* dynamic mutex pool */
112 	SI_SUB_LOCK		= 0x1B00000,	/* various locks */
113 	SI_SUB_EVENTHANDLER	= 0x1C00000,	/* eventhandler init */
114 	SI_SUB_VNET_PRELINK	= 0x1E00000,	/* vnet init before modules */
115 	SI_SUB_KLD		= 0x2000000,	/* KLD and module setup */
116 	SI_SUB_KHELP		= 0x2080000,	/* khelp modules */
117 	SI_SUB_CPU		= 0x2100000,	/* CPU resource(s) */
118 	SI_SUB_RACCT		= 0x2110000,	/* resource accounting */
119 	SI_SUB_KDTRACE		= 0x2140000,	/* Kernel dtrace hooks */
120 	SI_SUB_RANDOM		= 0x2160000,	/* random number generator */
121 	SI_SUB_MAC		= 0x2180000,	/* TrustedBSD MAC subsystem */
122 	SI_SUB_MAC_POLICY	= 0x21C0000,	/* TrustedBSD MAC policies */
123 	SI_SUB_MAC_LATE		= 0x21D0000,	/* TrustedBSD MAC subsystem */
124 	SI_SUB_VNET		= 0x21E0000,	/* vnet 0 */
125 	SI_SUB_INTRINSIC	= 0x2200000,	/* proc 0 */
126 	SI_SUB_VM_CONF		= 0x2300000,	/* config VM, set limits */
127 	SI_SUB_DDB_SERVICES	= 0x2380000,	/* capture, scripting, etc. */
128 	SI_SUB_RUN_QUEUE	= 0x2400000,	/* set up run queue */
129 	SI_SUB_KTRACE		= 0x2480000,	/* ktrace */
130 	SI_SUB_OPENSOLARIS	= 0x2490000,	/* OpenSolaris compatibility */
131 	SI_SUB_AUDIT		= 0x24C0000,	/* audit */
132 	SI_SUB_CREATE_INIT	= 0x2500000,	/* create init process */
133 	SI_SUB_SCHED_IDLE	= 0x2600000,	/* required idle procs */
134 	SI_SUB_MBUF		= 0x2700000,	/* mbuf subsystem */
135 	SI_SUB_INTR		= 0x2800000,	/* interrupt threads */
136 	SI_SUB_TASKQ		= 0x2880000,	/* task queues */
137 	SI_SUB_EPOCH		= 0x2888000,	/* epoch subsystem */
138 #ifdef EARLY_AP_STARTUP
139 	SI_SUB_SMP		= 0x2900000,	/* start the APs */
140 #endif
141 	SI_SUB_SOFTINTR		= 0x2A00000,	/* start soft interrupt thread */
142 	SI_SUB_DEVFS		= 0x2F00000,	/* devfs ready for devices */
143 	SI_SUB_INIT_IF		= 0x3000000,	/* prep for net interfaces */
144 	SI_SUB_NETGRAPH		= 0x3010000,	/* Let Netgraph initialize */
145 	SI_SUB_DTRACE		= 0x3020000,	/* DTrace subsystem */
146 	SI_SUB_DTRACE_PROVIDER	= 0x3048000,	/* DTrace providers */
147 	SI_SUB_DTRACE_ANON	= 0x308C000,	/* DTrace anon enabling */
148 	SI_SUB_DRIVERS		= 0x3100000,	/* Let Drivers initialize */
149 	SI_SUB_CONFIGURE	= 0x3800000,	/* Configure devices */
150 	SI_SUB_VFS		= 0x4000000,	/* virtual filesystem */
151 	SI_SUB_CLOCKS		= 0x4800000,	/* real time and stat clocks */
152 	SI_SUB_SYSV_SHM		= 0x6400000,	/* System V shared memory */
153 	SI_SUB_SYSV_SEM		= 0x6800000,	/* System V semaphores */
154 	SI_SUB_SYSV_MSG		= 0x6C00000,	/* System V message queues */
155 	SI_SUB_P1003_1B		= 0x6E00000,	/* P1003.1B realtime */
156 	SI_SUB_PSEUDO		= 0x7000000,	/* pseudo devices */
157 	SI_SUB_EXEC		= 0x7400000,	/* execve() handlers */
158 	SI_SUB_PROTO_BEGIN	= 0x8000000,	/* VNET initialization */
159 	SI_SUB_PROTO_PFIL	= 0x8100000,	/* Initialize pfil before FWs */
160 	SI_SUB_PROTO_MC		= 0x8300000,	/* Multicast */
161 	SI_SUB_PROTO_IF		= 0x8400000,	/* interfaces */
162 	SI_SUB_PROTO_DOMAININIT	= 0x8600000,	/* domain registration system */
163 	SI_SUB_PROTO_DOMAIN	= 0x8800000,	/* domains (address families?) */
164 	SI_SUB_PROTO_FIREWALL	= 0x8806000,	/* Firewalls */
165 	SI_SUB_PROTO_IFATTACHDOMAIN = 0x8808000,/* domain dependent data init */
166 	SI_SUB_PROTO_END	= 0x8ffffff,	/* VNET helper functions */
167 	SI_SUB_KPROF		= 0x9000000,	/* kernel profiling */
168 	SI_SUB_KICK_SCHEDULER	= 0xa000000,	/* start the timeout events */
169 	SI_SUB_INT_CONFIG_HOOKS	= 0xa800000,	/* Interrupts enabled config */
170 	SI_SUB_ROOT_CONF	= 0xb000000,	/* Find root devices */
171 	SI_SUB_INTRINSIC_POST	= 0xd000000,	/* proc 0 cleanup */
172 	SI_SUB_SYSCALLS		= 0xd800000,	/* register system calls */
173 	SI_SUB_VNET_DONE	= 0xdc00000,	/* vnet registration complete */
174 	SI_SUB_KTHREAD_INIT	= 0xe000000,	/* init process */
175 	SI_SUB_KTHREAD_PAGE	= 0xe400000,	/* pageout daemon */
176 	SI_SUB_KTHREAD_VM	= 0xe800000,	/* vm daemon */
177 	SI_SUB_KTHREAD_BUF	= 0xea00000,	/* buffer daemon */
178 	SI_SUB_KTHREAD_UPDATE	= 0xec00000,	/* update daemon */
179 	SI_SUB_KTHREAD_IDLE	= 0xee00000,	/* idle procs */
180 #ifndef EARLY_AP_STARTUP
181 	SI_SUB_SMP		= 0xf000000,	/* start the APs */
182 #endif
183 	SI_SUB_RACCTD		= 0xf100000,	/* start racctd */
184 	SI_SUB_LAST		= 0xfffffff	/* final initialization */
185 };
186 
187 /*
188  * Some enumerated orders; "ANY" sorts last.
189  */
190 enum sysinit_elem_order {
191 	SI_ORDER_FIRST		= 0x0000000,	/* first */
192 	SI_ORDER_SECOND		= 0x0000001,	/* second */
193 	SI_ORDER_THIRD		= 0x0000002,	/* third */
194 	SI_ORDER_FOURTH		= 0x0000003,	/* fourth */
195 	SI_ORDER_FIFTH		= 0x0000004,	/* fifth */
196 	SI_ORDER_SIXTH		= 0x0000005,	/* sixth */
197 	SI_ORDER_SEVENTH	= 0x0000006,	/* seventh */
198 	SI_ORDER_EIGHTH		= 0x0000007,	/* eighth */
199 	SI_ORDER_MIDDLE		= 0x1000000,	/* somewhere in the middle */
200 	SI_ORDER_ANY		= 0xf000000,	/* default */
201 	SI_ORDER_LAST		= 0xfffffff	/* last */
202 };
203 
204 /*
205  * A system initialization call instance
206  *
207  * At the moment there is one instance of sysinit.  We probably do not
208  * want two which is why this code is if'd out, but we definitely want
209  * to discern SYSINIT's which take non-constant data pointers and
210  * SYSINIT's which take constant data pointers,
211  *
212  * The C_* macros take functions expecting const void * arguments
213  * while the non-C_* macros take functions expecting just void * arguments.
214  *
215  * With -Wcast-qual on, the compiler issues warnings:
216  *	- if we pass non-const data or functions taking non-const data
217  *	  to a C_* macro.
218  *
219  *	- if we pass const data to the normal macros
220  *
221  * However, no warning is issued if we pass a function taking const data
222  * through a normal non-const macro.  This is ok because the function is
223  * saying it won't modify the data so we don't care whether the data is
224  * modifiable or not.
225  */
226 
227 typedef void (*sysinit_nfunc_t)(void *);
228 typedef void (*sysinit_cfunc_t)(const void *);
229 
230 struct sysinit {
231 	enum sysinit_sub_id	subsystem;	/* subsystem identifier */
232 	enum sysinit_elem_order	order;		/* init order within subsystem */
233 	STAILQ_ENTRY(sysinit)	next;		/* singly-linked list */
234 	sysinit_cfunc_t func;			/* function */
235 	const void	*udata;			/* multiplexer/argument */
236 };
237 
238 /*
239  * Default: no special processing
240  *
241  * The C_ version of SYSINIT is for data pointers to const
242  * data ( and functions taking data pointers to const data ).
243  * At the moment it is no different from SYSINIT and thus
244  * still results in warnings.
245  *
246  * The casts are necessary to have the compiler produce the
247  * correct warnings when -Wcast-qual is used.
248  *
249  */
250 #ifdef TSLOG
251 struct sysinit_tslog {
252 	sysinit_cfunc_t func;
253 	const void *data;
254 	const char *name;
255 };
256 void sysinit_tslog_shim(const void *);
257 
258 #define	C_SYSINIT(uniquifier, subsystem, order, func, ident)	\
259 	static struct sysinit_tslog uniquifier ## _sys_init_tslog = {	\
260 		func,						\
261 		(ident),					\
262 		#uniquifier					\
263 	};							\
264 	static struct sysinit uniquifier ## _sys_init = {	\
265 		subsystem,					\
266 		order,						\
267 		{ NULL },					\
268 		sysinit_tslog_shim,				\
269 		&uniquifier ## _sys_init_tslog			\
270 	};							\
271 	DATA_WSET(sysinit_set, uniquifier ## _sys_init)
272 #else
273 #define	C_SYSINIT(uniquifier, subsystem, order, func, ident)	\
274 	static struct sysinit uniquifier ## _sys_init = {	\
275 		subsystem,					\
276 		order,						\
277 		{ NULL },					\
278 		func,						\
279 		(ident)						\
280 	};							\
281 	DATA_WSET(sysinit_set, uniquifier ## _sys_init)
282 #endif
283 
284 #define	SYSINIT(uniquifier, subsystem, order, func, ident)	\
285 	C_SYSINIT(uniquifier, subsystem, order,			\
286 	(sysinit_cfunc_t)(sysinit_nfunc_t)func, (void *)(ident))
287 
288 /*
289  * Called on module unload: no special processing
290  */
291 #define	C_SYSUNINIT(uniquifier, subsystem, order, func, ident)	\
292 	static struct sysinit uniquifier ## _sys_uninit = {	\
293 		subsystem,					\
294 		order,						\
295 		{ NULL },					\
296 		func,						\
297 		(ident)						\
298 	};							\
299 	DATA_WSET(sysuninit_set, uniquifier ## _sys_uninit)
300 
301 #define	SYSUNINIT(uniquifier, subsystem, order, func, ident)	\
302 	C_SYSUNINIT(uniquifier, subsystem, order,		\
303 	(sysinit_cfunc_t)(sysinit_nfunc_t)func, (void *)(ident))
304 
305 void	sysinit_add(struct sysinit **set, struct sysinit **set_end);
306 
307 #ifdef _KERNEL
308 
309 /*
310  * Infrastructure for tunable 'constants'.  Value may be specified at compile
311  * time or kernel load time.  Rules relating tunables together can be placed
312  * in a SYSINIT function at SI_SUB_TUNABLES with SI_ORDER_ANY.
313  *
314  * WARNING: developers should never use the reserved suffixes specified in
315  * loader.conf(5) for any tunables or conflicts will result.
316  */
317 
318 /*
319  * int
320  * please avoid using for new tunables!
321  */
322 extern void tunable_int_init(const void *);
323 struct tunable_int {
324 	const char *path;
325 	int *var;
326 };
327 #define	TUNABLE_INT(path, var)					\
328 	static struct tunable_int __CONCAT(__tunable_int_, __LINE__) = { \
329 		(path),						\
330 		(var),						\
331 	};							\
332 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
333 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_int_init,	\
334 	    &__CONCAT(__tunable_int_, __LINE__))
335 
336 #define	TUNABLE_INT_FETCH(path, var)	getenv_int((path), (var))
337 
338 /*
339  * long
340  */
341 extern void tunable_long_init(const void *);
342 struct tunable_long {
343 	const char *path;
344 	long *var;
345 };
346 #define	TUNABLE_LONG(path, var)					\
347 	static struct tunable_long __CONCAT(__tunable_long_, __LINE__) = { \
348 		(path),						\
349 		(var),						\
350 	};							\
351 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
352 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_long_init,\
353 	    &__CONCAT(__tunable_long_, __LINE__))
354 
355 #define	TUNABLE_LONG_FETCH(path, var)	getenv_long((path), (var))
356 
357 /*
358  * unsigned long
359  */
360 extern void tunable_ulong_init(const void *);
361 struct tunable_ulong {
362 	const char *path;
363 	unsigned long *var;
364 };
365 #define	TUNABLE_ULONG(path, var)				\
366 	static struct tunable_ulong __CONCAT(__tunable_ulong_, __LINE__) = { \
367 		(path),						\
368 		(var),						\
369 	};							\
370 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
371 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_ulong_init, \
372 	    &__CONCAT(__tunable_ulong_, __LINE__))
373 
374 #define	TUNABLE_ULONG_FETCH(path, var)	getenv_ulong((path), (var))
375 
376 /*
377  * int64_t
378  */
379 extern void tunable_int64_init(const void *);
380 struct tunable_int64 {
381 	const char *path;
382 	int64_t *var;
383 };
384 #define	TUNABLE_INT64(path, var)				\
385 	static struct tunable_int64 __CONCAT(__tunable_int64_, __LINE__) = { \
386 		(path),						\
387 		(var),						\
388 	};							\
389 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
390 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_int64_init, \
391 	    &__CONCAT(__tunable_int64_, __LINE__))
392 
393 #define	TUNABLE_INT64_FETCH(path, var)	getenv_int64((path), (var))
394 
395 /*
396  * uint64_t
397  */
398 extern void tunable_uint64_init(const void *);
399 struct tunable_uint64 {
400 	const char *path;
401 	uint64_t *var;
402 };
403 #define	TUNABLE_UINT64(path, var)				\
404 	static struct tunable_uint64 __CONCAT(__tunable_uint64_, __LINE__) = { \
405 		(path),						\
406 		(var),						\
407 	};							\
408 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
409 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_uint64_init, \
410 	    &__CONCAT(__tunable_uint64_, __LINE__))
411 
412 #define	TUNABLE_UINT64_FETCH(path, var)	getenv_uint64((path), (var))
413 
414 /*
415  * quad
416  */
417 extern void tunable_quad_init(const void *);
418 struct tunable_quad {
419 	const char *path;
420 	quad_t *var;
421 };
422 #define	TUNABLE_QUAD(path, var)					\
423 	static struct tunable_quad __CONCAT(__tunable_quad_, __LINE__) = { \
424 		(path),						\
425 		(var),						\
426 	};							\
427 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
428 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_quad_init, \
429 	    &__CONCAT(__tunable_quad_, __LINE__))
430 
431 #define	TUNABLE_QUAD_FETCH(path, var)	getenv_quad((path), (var))
432 
433 /*
434  * bool
435  */
436 extern void tunable_bool_init(const void *);
437 struct tunable_bool {
438 	const char *path;
439 	bool *var;
440 };
441 #define	TUNABLE_BOOL(path, var) \
442 	static struct tunable_bool __CONCAT(__tunable_bool_, __LINE__) = { \
443 		(path),						\
444 		(var),						\
445 	};							\
446 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
447 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_bool_init, \
448 	    &__CONCAT(__tunable_bool_, __LINE__))
449 
450 #define	TUNABLE_BOOL_FETCH(path, var)	getenv_bool((path), (var))
451 
452 extern void tunable_str_init(const void *);
453 struct tunable_str {
454 	const char *path;
455 	char *var;
456 	int size;
457 };
458 #define	TUNABLE_STR(path, var, size)				\
459 	static struct tunable_str __CONCAT(__tunable_str_, __LINE__) = { \
460 		(path),						\
461 		(var),						\
462 		(size),						\
463 	};							\
464 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
465 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_str_init,	\
466 	    &__CONCAT(__tunable_str_, __LINE__))
467 
468 #define	TUNABLE_STR_FETCH(path, var, size)			\
469 	getenv_string((path), (var), (size))
470 
471 #endif /* _KERNEL */
472 
473 typedef void (*ich_func_t)(void *_arg);
474 
475 struct intr_config_hook {
476 	STAILQ_ENTRY(intr_config_hook) ich_links;
477 	uintptr_t	ich_state;
478 #define ICHS_QUEUED	0x1
479 #define ICHS_RUNNING	0x2
480 #define	ICHS_DONE	0x3
481 	ich_func_t	ich_func;
482 	void		*ich_arg;
483 };
484 
485 int	config_intrhook_establish(struct intr_config_hook *hook);
486 void	config_intrhook_disestablish(struct intr_config_hook *hook);
487 int	config_intrhook_drain(struct intr_config_hook *hook);
488 void	config_intrhook_oneshot(ich_func_t _func, void *_arg);
489 
490 #endif /* !_SYS_KERNEL_H_*/
491