xref: /freebsd/sys/sys/kernel.h (revision 6b82130e6c9add4a8892ca897df5a0ec04663ea2)
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_TUNABLES		= 0x0700000,	/* establish tunable values */
98 	SI_SUB_COPYRIGHT	= 0x0800001,	/* first use of console */
99 	SI_SUB_VM		= 0x1000000,	/* virtual memory system init */
100 	SI_SUB_COUNTER		= 0x1100000,	/* counter(9) is initialized */
101 	SI_SUB_KMEM		= 0x1800000,	/* kernel memory */
102 	SI_SUB_HYPERVISOR	= 0x1A40000,	/*
103 						 * Hypervisor detection and
104 						 * virtualization support
105 						 * setup.
106 						 */
107 	SI_SUB_WITNESS		= 0x1A80000,	/* witness initialization */
108 	SI_SUB_MTX_POOL_DYNAMIC	= 0x1AC0000,	/* dynamic mutex pool */
109 	SI_SUB_LOCK		= 0x1B00000,	/* various locks */
110 	SI_SUB_EVENTHANDLER	= 0x1C00000,	/* eventhandler init */
111 	SI_SUB_VNET_PRELINK	= 0x1E00000,	/* vnet init before modules */
112 	SI_SUB_KLD		= 0x2000000,	/* KLD and module setup */
113 	SI_SUB_KHELP		= 0x2080000,	/* khelp modules */
114 	SI_SUB_CPU		= 0x2100000,	/* CPU resource(s) */
115 	SI_SUB_RACCT		= 0x2110000,	/* resource accounting */
116 	SI_SUB_KDTRACE		= 0x2140000,	/* Kernel dtrace hooks */
117 	SI_SUB_RANDOM		= 0x2160000,	/* random number generator */
118 	SI_SUB_MAC		= 0x2180000,	/* TrustedBSD MAC subsystem */
119 	SI_SUB_MAC_POLICY	= 0x21C0000,	/* TrustedBSD MAC policies */
120 	SI_SUB_MAC_LATE		= 0x21D0000,	/* TrustedBSD MAC subsystem */
121 	SI_SUB_VNET		= 0x21E0000,	/* vnet 0 */
122 	SI_SUB_INTRINSIC	= 0x2200000,	/* proc 0 */
123 	SI_SUB_VM_CONF		= 0x2300000,	/* config VM, set limits */
124 	SI_SUB_DDB_SERVICES	= 0x2380000,	/* capture, scripting, etc. */
125 	SI_SUB_RUN_QUEUE	= 0x2400000,	/* set up run queue */
126 	SI_SUB_KTRACE		= 0x2480000,	/* ktrace */
127 	SI_SUB_OPENSOLARIS	= 0x2490000,	/* OpenSolaris compatibility */
128 	SI_SUB_AUDIT		= 0x24C0000,	/* audit */
129 	SI_SUB_CREATE_INIT	= 0x2500000,	/* create init process */
130 	SI_SUB_SCHED_IDLE	= 0x2600000,	/* required idle procs */
131 	SI_SUB_MBUF		= 0x2700000,	/* mbuf subsystem */
132 	SI_SUB_INTR		= 0x2800000,	/* interrupt threads */
133 	SI_SUB_TASKQ		= 0x2880000,	/* task queues */
134 	SI_SUB_EPOCH		= 0x2888000,	/* epoch subsystem */
135 #ifdef EARLY_AP_STARTUP
136 	SI_SUB_SMP		= 0x2900000,	/* start the APs */
137 #endif
138 	SI_SUB_SOFTINTR		= 0x2A00000,	/* start soft interrupt thread */
139 	SI_SUB_DEVFS		= 0x2F00000,	/* devfs ready for devices */
140 	SI_SUB_INIT_IF		= 0x3000000,	/* prep for net interfaces */
141 	SI_SUB_NETGRAPH		= 0x3010000,	/* Let Netgraph initialize */
142 	SI_SUB_DTRACE		= 0x3020000,	/* DTrace subsystem */
143 	SI_SUB_DTRACE_PROVIDER	= 0x3048000,	/* DTrace providers */
144 	SI_SUB_DTRACE_ANON	= 0x308C000,	/* DTrace anon enabling */
145 	SI_SUB_DRIVERS		= 0x3100000,	/* Let Drivers initialize */
146 	SI_SUB_CONFIGURE	= 0x3800000,	/* Configure devices */
147 	SI_SUB_VFS		= 0x4000000,	/* virtual filesystem */
148 	SI_SUB_CLOCKS		= 0x4800000,	/* real time and stat clocks */
149 	SI_SUB_SYSV_SHM		= 0x6400000,	/* System V shared memory */
150 	SI_SUB_SYSV_SEM		= 0x6800000,	/* System V semaphores */
151 	SI_SUB_SYSV_MSG		= 0x6C00000,	/* System V message queues */
152 	SI_SUB_P1003_1B		= 0x6E00000,	/* P1003.1B realtime */
153 	SI_SUB_PSEUDO		= 0x7000000,	/* pseudo devices */
154 	SI_SUB_EXEC		= 0x7400000,	/* execve() handlers */
155 	SI_SUB_PROTO_BEGIN	= 0x8000000,	/* VNET initialization */
156 	SI_SUB_PROTO_PFIL	= 0x8100000,	/* Initialize pfil before FWs */
157 	SI_SUB_PROTO_IF		= 0x8400000,	/* interfaces */
158 	SI_SUB_PROTO_DOMAININIT	= 0x8600000,	/* domain registration system */
159 	SI_SUB_PROTO_MC		= 0x8700000,	/* Multicast */
160 	SI_SUB_PROTO_DOMAIN	= 0x8800000,	/* domains (address families?) */
161 	SI_SUB_PROTO_FIREWALL	= 0x8806000,	/* Firewalls */
162 	SI_SUB_PROTO_IFATTACHDOMAIN = 0x8808000,/* domain dependent data init */
163 	SI_SUB_PROTO_END	= 0x8ffffff,	/* VNET helper functions */
164 	SI_SUB_KPROF		= 0x9000000,	/* kernel profiling */
165 	SI_SUB_KICK_SCHEDULER	= 0xa000000,	/* start the timeout events */
166 	SI_SUB_INT_CONFIG_HOOKS	= 0xa800000,	/* Interrupts enabled config */
167 	SI_SUB_ROOT_CONF	= 0xb000000,	/* Find root devices */
168 	SI_SUB_INTRINSIC_POST	= 0xd000000,	/* proc 0 cleanup */
169 	SI_SUB_SYSCALLS		= 0xd800000,	/* register system calls */
170 	SI_SUB_VNET_DONE	= 0xdc00000,	/* vnet registration complete */
171 	SI_SUB_KTHREAD_INIT	= 0xe000000,	/* init process */
172 	SI_SUB_KTHREAD_PAGE	= 0xe400000,	/* pageout daemon */
173 	SI_SUB_KTHREAD_VM	= 0xe800000,	/* vm daemon */
174 	SI_SUB_KTHREAD_BUF	= 0xea00000,	/* buffer daemon */
175 	SI_SUB_KTHREAD_UPDATE	= 0xec00000,	/* update daemon */
176 	SI_SUB_KTHREAD_IDLE	= 0xee00000,	/* idle procs */
177 #ifndef EARLY_AP_STARTUP
178 	SI_SUB_SMP		= 0xf000000,	/* start the APs */
179 #endif
180 	SI_SUB_RACCTD		= 0xf100000,	/* start racctd */
181 	SI_SUB_LAST		= 0xfffffff	/* final initialization */
182 };
183 
184 /*
185  * Some enumerated orders; "ANY" sorts last.
186  */
187 enum sysinit_elem_order {
188 	SI_ORDER_FIRST		= 0x0000000,	/* first */
189 	SI_ORDER_SECOND		= 0x0000001,	/* second */
190 	SI_ORDER_THIRD		= 0x0000002,	/* third */
191 	SI_ORDER_FOURTH		= 0x0000003,	/* fourth */
192 	SI_ORDER_FIFTH		= 0x0000004,	/* fifth */
193 	SI_ORDER_SIXTH		= 0x0000005,	/* sixth */
194 	SI_ORDER_SEVENTH	= 0x0000006,	/* seventh */
195 	SI_ORDER_EIGHTH		= 0x0000007,	/* eighth */
196 	SI_ORDER_MIDDLE		= 0x1000000,	/* somewhere in the middle */
197 	SI_ORDER_ANY		= 0xfffffff	/* last */
198 };
199 
200 /*
201  * A system initialization call instance
202  *
203  * At the moment there is one instance of sysinit.  We probably do not
204  * want two which is why this code is if'd out, but we definitely want
205  * to discern SYSINIT's which take non-constant data pointers and
206  * SYSINIT's which take constant data pointers,
207  *
208  * The C_* macros take functions expecting const void * arguments
209  * while the non-C_* macros take functions expecting just void * arguments.
210  *
211  * With -Wcast-qual on, the compiler issues warnings:
212  *	- if we pass non-const data or functions taking non-const data
213  *	  to a C_* macro.
214  *
215  *	- if we pass const data to the normal macros
216  *
217  * However, no warning is issued if we pass a function taking const data
218  * through a normal non-const macro.  This is ok because the function is
219  * saying it won't modify the data so we don't care whether the data is
220  * modifiable or not.
221  */
222 
223 typedef void (*sysinit_nfunc_t)(void *);
224 typedef void (*sysinit_cfunc_t)(const void *);
225 
226 struct sysinit {
227 	enum sysinit_sub_id	subsystem;	/* subsystem identifier */
228 	enum sysinit_elem_order	order;		/* init order within subsystem */
229 	STAILQ_ENTRY(sysinit)	next;		/* singly-linked list */
230 	sysinit_cfunc_t func;			/* function */
231 	const void	*udata;			/* multiplexer/argument */
232 };
233 
234 /*
235  * Default: no special processing
236  *
237  * The C_ version of SYSINIT is for data pointers to const
238  * data ( and functions taking data pointers to const data ).
239  * At the moment it is no different from SYSINIT and thus
240  * still results in warnings.
241  *
242  * The casts are necessary to have the compiler produce the
243  * correct warnings when -Wcast-qual is used.
244  *
245  */
246 #ifdef TSLOG
247 struct sysinit_tslog {
248 	sysinit_cfunc_t func;
249 	const void *data;
250 	const char *name;
251 };
252 static inline void
sysinit_tslog_shim(const void * data)253 sysinit_tslog_shim(const void *data)
254 {
255 	const struct sysinit_tslog *x = data;
256 
257 	TSRAW(curthread, TS_ENTER, "SYSINIT", x->name);
258 	(x->func)(x->data);
259 	TSRAW(curthread, TS_EXIT, "SYSINIT", x->name);
260 }
261 #define	C_SYSINIT(uniquifier, subsystem, order, func, ident)	\
262 	static struct sysinit_tslog uniquifier ## _sys_init_tslog = {	\
263 		func,						\
264 		(ident),					\
265 		#uniquifier					\
266 	};							\
267 	static struct sysinit uniquifier ## _sys_init = {	\
268 		subsystem,					\
269 		order,						\
270 		{ NULL },					\
271 		sysinit_tslog_shim,				\
272 		&uniquifier ## _sys_init_tslog			\
273 	};							\
274 	DATA_WSET(sysinit_set, uniquifier ## _sys_init)
275 #else
276 #define	C_SYSINIT(uniquifier, subsystem, order, func, ident)	\
277 	static struct sysinit uniquifier ## _sys_init = {	\
278 		subsystem,					\
279 		order,						\
280 		{ NULL },					\
281 		func,						\
282 		(ident)						\
283 	};							\
284 	DATA_WSET(sysinit_set, uniquifier ## _sys_init)
285 #endif
286 
287 #define	SYSINIT(uniquifier, subsystem, order, func, ident)	\
288 	C_SYSINIT(uniquifier, subsystem, order,			\
289 	(sysinit_cfunc_t)(sysinit_nfunc_t)func, (void *)(ident))
290 
291 /*
292  * Called on module unload: no special processing
293  */
294 #define	C_SYSUNINIT(uniquifier, subsystem, order, func, ident)	\
295 	static struct sysinit uniquifier ## _sys_uninit = {	\
296 		subsystem,					\
297 		order,						\
298 		{ NULL },					\
299 		func,						\
300 		(ident)						\
301 	};							\
302 	DATA_WSET(sysuninit_set, uniquifier ## _sys_uninit)
303 
304 #define	SYSUNINIT(uniquifier, subsystem, order, func, ident)	\
305 	C_SYSUNINIT(uniquifier, subsystem, order,		\
306 	(sysinit_cfunc_t)(sysinit_nfunc_t)func, (void *)(ident))
307 
308 void	sysinit_add(struct sysinit **set, struct sysinit **set_end);
309 
310 #ifdef _KERNEL
311 
312 /*
313  * Infrastructure for tunable 'constants'.  Value may be specified at compile
314  * time or kernel load time.  Rules relating tunables together can be placed
315  * in a SYSINIT function at SI_SUB_TUNABLES with SI_ORDER_ANY.
316  *
317  * WARNING: developers should never use the reserved suffixes specified in
318  * loader.conf(5) for any tunables or conflicts will result.
319  */
320 
321 /*
322  * int
323  * please avoid using for new tunables!
324  */
325 extern void tunable_int_init(void *);
326 struct tunable_int {
327 	const char *path;
328 	int *var;
329 };
330 #define	TUNABLE_INT(path, var)					\
331 	static struct tunable_int __CONCAT(__tunable_int_, __LINE__) = { \
332 		(path),						\
333 		(var),						\
334 	};							\
335 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
336 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_int_init,	\
337 	    &__CONCAT(__tunable_int_, __LINE__))
338 
339 #define	TUNABLE_INT_FETCH(path, var)	getenv_int((path), (var))
340 
341 /*
342  * long
343  */
344 extern void tunable_long_init(void *);
345 struct tunable_long {
346 	const char *path;
347 	long *var;
348 };
349 #define	TUNABLE_LONG(path, var)					\
350 	static struct tunable_long __CONCAT(__tunable_long_, __LINE__) = { \
351 		(path),						\
352 		(var),						\
353 	};							\
354 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
355 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_long_init,\
356 	    &__CONCAT(__tunable_long_, __LINE__))
357 
358 #define	TUNABLE_LONG_FETCH(path, var)	getenv_long((path), (var))
359 
360 /*
361  * unsigned long
362  */
363 extern void tunable_ulong_init(void *);
364 struct tunable_ulong {
365 	const char *path;
366 	unsigned long *var;
367 };
368 #define	TUNABLE_ULONG(path, var)				\
369 	static struct tunable_ulong __CONCAT(__tunable_ulong_, __LINE__) = { \
370 		(path),						\
371 		(var),						\
372 	};							\
373 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
374 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_ulong_init, \
375 	    &__CONCAT(__tunable_ulong_, __LINE__))
376 
377 #define	TUNABLE_ULONG_FETCH(path, var)	getenv_ulong((path), (var))
378 
379 /*
380  * int64_t
381  */
382 extern void tunable_int64_init(void *);
383 struct tunable_int64 {
384 	const char *path;
385 	int64_t *var;
386 };
387 #define	TUNABLE_INT64(path, var)				\
388 	static struct tunable_int64 __CONCAT(__tunable_int64_, __LINE__) = { \
389 		(path),						\
390 		(var),						\
391 	};							\
392 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
393 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_int64_init, \
394 	    &__CONCAT(__tunable_int64_, __LINE__))
395 
396 #define	TUNABLE_INT64_FETCH(path, var)	getenv_int64((path), (var))
397 
398 /*
399  * uint64_t
400  */
401 extern void tunable_uint64_init(void *);
402 struct tunable_uint64 {
403 	const char *path;
404 	uint64_t *var;
405 };
406 #define	TUNABLE_UINT64(path, var)				\
407 	static struct tunable_uint64 __CONCAT(__tunable_uint64_, __LINE__) = { \
408 		(path),						\
409 		(var),						\
410 	};							\
411 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
412 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_uint64_init, \
413 	    &__CONCAT(__tunable_uint64_, __LINE__))
414 
415 #define	TUNABLE_UINT64_FETCH(path, var)	getenv_uint64((path), (var))
416 
417 /*
418  * quad
419  */
420 extern void tunable_quad_init(void *);
421 struct tunable_quad {
422 	const char *path;
423 	quad_t *var;
424 };
425 #define	TUNABLE_QUAD(path, var)					\
426 	static struct tunable_quad __CONCAT(__tunable_quad_, __LINE__) = { \
427 		(path),						\
428 		(var),						\
429 	};							\
430 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
431 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_quad_init, \
432 	    &__CONCAT(__tunable_quad_, __LINE__))
433 
434 #define	TUNABLE_QUAD_FETCH(path, var)	getenv_quad((path), (var))
435 
436 /*
437  * bool
438  */
439 extern void tunable_bool_init(void *);
440 struct tunable_bool {
441 	const char *path;
442 	bool *var;
443 };
444 #define	TUNABLE_BOOL(path, var) \
445 	static struct tunable_bool __CONCAT(__tunable_bool_, __LINE__) = { \
446 		(path),						\
447 		(var),						\
448 	};							\
449 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
450 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_bool_init, \
451 	    &__CONCAT(__tunable_bool_, __LINE__))
452 
453 #define	TUNABLE_BOOL_FETCH(path, var)	getenv_bool((path), (var))
454 
455 extern void tunable_str_init(void *);
456 struct tunable_str {
457 	const char *path;
458 	char *var;
459 	int size;
460 };
461 #define	TUNABLE_STR(path, var, size)				\
462 	static struct tunable_str __CONCAT(__tunable_str_, __LINE__) = { \
463 		(path),						\
464 		(var),						\
465 		(size),						\
466 	};							\
467 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
468 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_str_init,	\
469 	    &__CONCAT(__tunable_str_, __LINE__))
470 
471 #define	TUNABLE_STR_FETCH(path, var, size)			\
472 	getenv_string((path), (var), (size))
473 
474 #endif /* _KERNEL */
475 
476 typedef void (*ich_func_t)(void *_arg);
477 
478 struct intr_config_hook {
479 	STAILQ_ENTRY(intr_config_hook) ich_links;
480 	uintptr_t	ich_state;
481 #define ICHS_QUEUED	0x1
482 #define ICHS_RUNNING	0x2
483 #define	ICHS_DONE	0x3
484 	ich_func_t	ich_func;
485 	void		*ich_arg;
486 };
487 
488 int	config_intrhook_establish(struct intr_config_hook *hook);
489 void	config_intrhook_disestablish(struct intr_config_hook *hook);
490 int	config_intrhook_drain(struct intr_config_hook *hook);
491 void	config_intrhook_oneshot(ich_func_t _func, void *_arg);
492 
493 #endif /* !_SYS_KERNEL_H_*/
494