xref: /freebsd/sys/sys/kernel.h (revision 5c623697d6011de63bc762560a648c09379e06d2)
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 extern volatile int ticks;
69 
70 #endif /* _KERNEL */
71 
72 /*
73  * Enumerated types for known system startup interfaces.
74  *
75  * Startup occurs in ascending numeric order; the list entries are
76  * sorted prior to attempting startup to guarantee order.  Items
77  * of the same level are arbitrated for order based on the 'order'
78  * element.
79  *
80  * These numbers are arbitrary and are chosen ONLY for ordering; the
81  * enumeration values are explicit rather than implicit to provide
82  * for binary compatibility with inserted elements.
83  *
84  * The SI_SUB_LAST value must have the highest lexical value.
85  */
86 enum sysinit_sub_id {
87 	SI_SUB_DUMMY		= 0x0000000,	/* not executed; for linker */
88 	SI_SUB_DONE		= 0x0000001,	/* processed */
89 	SI_SUB_TUNABLES		= 0x0700000,	/* establish tunable values */
90 	SI_SUB_COPYRIGHT	= 0x0800001,	/* first use of console */
91 	SI_SUB_VM		= 0x1000000,	/* virtual memory system init */
92 	SI_SUB_COUNTER		= 0x1100000,	/* counter(9) is initialized */
93 	SI_SUB_KMEM		= 0x1800000,	/* kernel memory */
94 	SI_SUB_HYPERVISOR	= 0x1A40000,	/*
95 						 * Hypervisor detection and
96 						 * virtualization support
97 						 * setup.
98 						 */
99 	SI_SUB_WITNESS		= 0x1A80000,	/* witness initialization */
100 	SI_SUB_MTX_POOL_DYNAMIC	= 0x1AC0000,	/* dynamic mutex pool */
101 	SI_SUB_LOCK		= 0x1B00000,	/* various locks */
102 	SI_SUB_EVENTHANDLER	= 0x1C00000,	/* eventhandler init */
103 	SI_SUB_VNET_PRELINK	= 0x1E00000,	/* vnet init before modules */
104 	SI_SUB_KLD		= 0x2000000,	/* KLD and module setup */
105 	SI_SUB_KHELP		= 0x2080000,	/* khelp modules */
106 	SI_SUB_CPU		= 0x2100000,	/* CPU resource(s) */
107 	SI_SUB_RACCT		= 0x2110000,	/* resource accounting */
108 	SI_SUB_KDTRACE		= 0x2140000,	/* Kernel dtrace hooks */
109 	SI_SUB_RANDOM		= 0x2160000,	/* random number generator */
110 	SI_SUB_MAC		= 0x2180000,	/* TrustedBSD MAC subsystem */
111 	SI_SUB_MAC_POLICY	= 0x21C0000,	/* TrustedBSD MAC policies */
112 	SI_SUB_MAC_LATE		= 0x21D0000,	/* TrustedBSD MAC subsystem */
113 	SI_SUB_VNET		= 0x21E0000,	/* vnet 0 */
114 	SI_SUB_INTRINSIC	= 0x2200000,	/* proc 0 */
115 	SI_SUB_VM_CONF		= 0x2300000,	/* config VM, set limits */
116 	SI_SUB_DDB_SERVICES	= 0x2380000,	/* capture, scripting, etc. */
117 	SI_SUB_RUN_QUEUE	= 0x2400000,	/* set up run queue */
118 	SI_SUB_KTRACE		= 0x2480000,	/* ktrace */
119 	SI_SUB_OPENSOLARIS	= 0x2490000,	/* OpenSolaris compatibility */
120 	SI_SUB_AUDIT		= 0x24C0000,	/* audit */
121 	SI_SUB_CREATE_INIT	= 0x2500000,	/* create init process */
122 	SI_SUB_SCHED_IDLE	= 0x2600000,	/* required idle procs */
123 	SI_SUB_MBUF		= 0x2700000,	/* mbuf subsystem */
124 	SI_SUB_INTR		= 0x2800000,	/* interrupt threads */
125 	SI_SUB_TASKQ		= 0x2880000,	/* task queues */
126 	SI_SUB_EPOCH		= 0x2888000,	/* epoch subsystem */
127 #ifdef EARLY_AP_STARTUP
128 	SI_SUB_SMP		= 0x2900000,	/* start the APs */
129 #endif
130 	SI_SUB_SOFTINTR		= 0x2A00000,	/* start soft interrupt thread */
131 	SI_SUB_DEVFS		= 0x2F00000,	/* devfs ready for devices */
132 	SI_SUB_INIT_IF		= 0x3000000,	/* prep for net interfaces */
133 	SI_SUB_NETGRAPH		= 0x3010000,	/* Let Netgraph initialize */
134 	SI_SUB_DTRACE		= 0x3020000,	/* DTrace subsystem */
135 	SI_SUB_DTRACE_PROVIDER	= 0x3048000,	/* DTrace providers */
136 	SI_SUB_DTRACE_ANON	= 0x308C000,	/* DTrace anon enabling */
137 	SI_SUB_DRIVERS		= 0x3100000,	/* Let Drivers initialize */
138 	SI_SUB_CONFIGURE	= 0x3800000,	/* Configure devices */
139 	SI_SUB_VFS		= 0x4000000,	/* virtual filesystem */
140 	SI_SUB_CLOCKS		= 0x4800000,	/* real time and stat clocks */
141 	SI_SUB_SYSV_SHM		= 0x6400000,	/* System V shared memory */
142 	SI_SUB_SYSV_SEM		= 0x6800000,	/* System V semaphores */
143 	SI_SUB_SYSV_MSG		= 0x6C00000,	/* System V message queues */
144 	SI_SUB_P1003_1B		= 0x6E00000,	/* P1003.1B realtime */
145 	SI_SUB_PSEUDO		= 0x7000000,	/* pseudo devices */
146 	SI_SUB_EXEC		= 0x7400000,	/* execve() handlers */
147 	SI_SUB_PROTO_BEGIN	= 0x8000000,	/* VNET initialization */
148 	SI_SUB_PROTO_PFIL	= 0x8100000,	/* Initialize pfil before FWs */
149 	SI_SUB_PROTO_IF		= 0x8400000,	/* interfaces */
150 	SI_SUB_PROTO_DOMAININIT	= 0x8600000,	/* domain registration system */
151 	SI_SUB_PROTO_MC		= 0x8700000,	/* Multicast */
152 	SI_SUB_PROTO_DOMAIN	= 0x8800000,	/* domains (address families?) */
153 	SI_SUB_PROTO_FIREWALL	= 0x8806000,	/* Firewalls */
154 	SI_SUB_PROTO_IFATTACHDOMAIN = 0x8808000,/* domain dependent data init */
155 	SI_SUB_PROTO_END	= 0x8ffffff,	/* VNET helper functions */
156 	SI_SUB_KPROF		= 0x9000000,	/* kernel profiling */
157 	SI_SUB_KICK_SCHEDULER	= 0xa000000,	/* start the timeout events */
158 	SI_SUB_INT_CONFIG_HOOKS	= 0xa800000,	/* Interrupts enabled config */
159 	SI_SUB_ROOT_CONF	= 0xb000000,	/* Find root devices */
160 	SI_SUB_INTRINSIC_POST	= 0xd000000,	/* proc 0 cleanup */
161 	SI_SUB_SYSCALLS		= 0xd800000,	/* register system calls */
162 	SI_SUB_VNET_DONE	= 0xdc00000,	/* vnet registration complete */
163 	SI_SUB_KTHREAD_INIT	= 0xe000000,	/* init process */
164 	SI_SUB_KTHREAD_PAGE	= 0xe400000,	/* pageout daemon */
165 	SI_SUB_KTHREAD_VM	= 0xe800000,	/* vm daemon */
166 	SI_SUB_KTHREAD_BUF	= 0xea00000,	/* buffer daemon */
167 	SI_SUB_KTHREAD_UPDATE	= 0xec00000,	/* update daemon */
168 	SI_SUB_KTHREAD_IDLE	= 0xee00000,	/* idle procs */
169 #ifndef EARLY_AP_STARTUP
170 	SI_SUB_SMP		= 0xf000000,	/* start the APs */
171 #endif
172 	SI_SUB_RACCTD		= 0xf100000,	/* start racctd */
173 	SI_SUB_LAST		= 0xfffffff	/* final initialization */
174 };
175 
176 /*
177  * Some enumerated orders; "ANY" sorts last.
178  */
179 enum sysinit_elem_order {
180 	SI_ORDER_FIRST		= 0x0000000,	/* first */
181 	SI_ORDER_SECOND		= 0x0000001,	/* second */
182 	SI_ORDER_THIRD		= 0x0000002,	/* third */
183 	SI_ORDER_FOURTH		= 0x0000003,	/* fourth */
184 	SI_ORDER_FIFTH		= 0x0000004,	/* fifth */
185 	SI_ORDER_SIXTH		= 0x0000005,	/* sixth */
186 	SI_ORDER_SEVENTH	= 0x0000006,	/* seventh */
187 	SI_ORDER_EIGHTH		= 0x0000007,	/* eighth */
188 	SI_ORDER_MIDDLE		= 0x1000000,	/* somewhere in the middle */
189 	SI_ORDER_ANY		= 0xfffffff	/* last */
190 };
191 
192 /*
193  * A system initialization call instance
194  *
195  * At the moment there is one instance of sysinit.  We probably do not
196  * want two which is why this code is if'd out, but we definitely want
197  * to discern SYSINIT's which take non-constant data pointers and
198  * SYSINIT's which take constant data pointers,
199  *
200  * The C_* macros take functions expecting const void * arguments
201  * while the non-C_* macros take functions expecting just void * arguments.
202  *
203  * With -Wcast-qual on, the compiler issues warnings:
204  *	- if we pass non-const data or functions taking non-const data
205  *	  to a C_* macro.
206  *
207  *	- if we pass const data to the normal macros
208  *
209  * However, no warning is issued if we pass a function taking const data
210  * through a normal non-const macro.  This is ok because the function is
211  * saying it won't modify the data so we don't care whether the data is
212  * modifiable or not.
213  */
214 
215 typedef void (*sysinit_nfunc_t)(void *);
216 typedef void (*sysinit_cfunc_t)(const void *);
217 
218 struct sysinit {
219 	enum sysinit_sub_id	subsystem;	/* subsystem identifier */
220 	enum sysinit_elem_order	order;		/* init order within subsystem */
221 	STAILQ_ENTRY(sysinit)	next;		/* singly-linked list */
222 	sysinit_cfunc_t func;			/* function */
223 	const void	*udata;			/* multiplexer/argument */
224 };
225 
226 /*
227  * Default: no special processing
228  *
229  * The C_ version of SYSINIT is for data pointers to const
230  * data ( and functions taking data pointers to const data ).
231  * At the moment it is no different from SYSINIT and thus
232  * still results in warnings.
233  *
234  * The casts are necessary to have the compiler produce the
235  * correct warnings when -Wcast-qual is used.
236  *
237  */
238 #ifdef TSLOG
239 struct sysinit_tslog {
240 	sysinit_cfunc_t func;
241 	const void *data;
242 	const char *name;
243 };
244 static inline void
sysinit_tslog_shim(const void * data)245 sysinit_tslog_shim(const void *data)
246 {
247 	const struct sysinit_tslog *x = data;
248 
249 	TSRAW(curthread, TS_ENTER, "SYSINIT", x->name);
250 	(x->func)(x->data);
251 	TSRAW(curthread, TS_EXIT, "SYSINIT", x->name);
252 }
253 #define	C_SYSINIT(uniquifier, subsystem, order, func, ident)	\
254 	static struct sysinit_tslog uniquifier ## _sys_init_tslog = {	\
255 		func,						\
256 		(ident),					\
257 		#uniquifier					\
258 	};							\
259 	static struct sysinit uniquifier ## _sys_init = {	\
260 		subsystem,					\
261 		order,						\
262 		{ NULL },					\
263 		sysinit_tslog_shim,				\
264 		&uniquifier ## _sys_init_tslog			\
265 	};							\
266 	DATA_WSET(sysinit_set, uniquifier ## _sys_init)
267 #else
268 #define	C_SYSINIT(uniquifier, subsystem, order, func, ident)	\
269 	static struct sysinit uniquifier ## _sys_init = {	\
270 		subsystem,					\
271 		order,						\
272 		{ NULL },					\
273 		func,						\
274 		(ident)						\
275 	};							\
276 	DATA_WSET(sysinit_set, uniquifier ## _sys_init)
277 #endif
278 
279 #define	SYSINIT(uniquifier, subsystem, order, func, ident)	\
280 	C_SYSINIT(uniquifier, subsystem, order,			\
281 	(sysinit_cfunc_t)(sysinit_nfunc_t)func, (void *)(ident))
282 
283 /*
284  * Called on module unload: no special processing
285  */
286 #define	C_SYSUNINIT(uniquifier, subsystem, order, func, ident)	\
287 	static struct sysinit uniquifier ## _sys_uninit = {	\
288 		subsystem,					\
289 		order,						\
290 		{ NULL },					\
291 		func,						\
292 		(ident)						\
293 	};							\
294 	DATA_WSET(sysuninit_set, uniquifier ## _sys_uninit)
295 
296 #define	SYSUNINIT(uniquifier, subsystem, order, func, ident)	\
297 	C_SYSUNINIT(uniquifier, subsystem, order,		\
298 	(sysinit_cfunc_t)(sysinit_nfunc_t)func, (void *)(ident))
299 
300 void	sysinit_add(struct sysinit **set, struct sysinit **set_end);
301 
302 #ifdef _KERNEL
303 
304 /*
305  * Infrastructure for tunable 'constants'.  Value may be specified at compile
306  * time or kernel load time.  Rules relating tunables together can be placed
307  * in a SYSINIT function at SI_SUB_TUNABLES with SI_ORDER_ANY.
308  *
309  * WARNING: developers should never use the reserved suffixes specified in
310  * loader.conf(5) for any tunables or conflicts will result.
311  */
312 
313 /*
314  * int
315  * please avoid using for new tunables!
316  */
317 extern void tunable_int_init(void *);
318 struct tunable_int {
319 	const char *path;
320 	int *var;
321 };
322 #define	TUNABLE_INT(path, var)					\
323 	static struct tunable_int __CONCAT(__tunable_int_, __LINE__) = { \
324 		(path),						\
325 		(var),						\
326 	};							\
327 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
328 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_int_init,	\
329 	    &__CONCAT(__tunable_int_, __LINE__))
330 
331 #define	TUNABLE_INT_FETCH(path, var)	getenv_int((path), (var))
332 
333 /*
334  * long
335  */
336 extern void tunable_long_init(void *);
337 struct tunable_long {
338 	const char *path;
339 	long *var;
340 };
341 #define	TUNABLE_LONG(path, var)					\
342 	static struct tunable_long __CONCAT(__tunable_long_, __LINE__) = { \
343 		(path),						\
344 		(var),						\
345 	};							\
346 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
347 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_long_init,\
348 	    &__CONCAT(__tunable_long_, __LINE__))
349 
350 #define	TUNABLE_LONG_FETCH(path, var)	getenv_long((path), (var))
351 
352 /*
353  * unsigned long
354  */
355 extern void tunable_ulong_init(void *);
356 struct tunable_ulong {
357 	const char *path;
358 	unsigned long *var;
359 };
360 #define	TUNABLE_ULONG(path, var)				\
361 	static struct tunable_ulong __CONCAT(__tunable_ulong_, __LINE__) = { \
362 		(path),						\
363 		(var),						\
364 	};							\
365 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
366 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_ulong_init, \
367 	    &__CONCAT(__tunable_ulong_, __LINE__))
368 
369 #define	TUNABLE_ULONG_FETCH(path, var)	getenv_ulong((path), (var))
370 
371 /*
372  * int64_t
373  */
374 extern void tunable_int64_init(void *);
375 struct tunable_int64 {
376 	const char *path;
377 	int64_t *var;
378 };
379 #define	TUNABLE_INT64(path, var)				\
380 	static struct tunable_int64 __CONCAT(__tunable_int64_, __LINE__) = { \
381 		(path),						\
382 		(var),						\
383 	};							\
384 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
385 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_int64_init, \
386 	    &__CONCAT(__tunable_int64_, __LINE__))
387 
388 #define	TUNABLE_INT64_FETCH(path, var)	getenv_int64((path), (var))
389 
390 /*
391  * uint64_t
392  */
393 extern void tunable_uint64_init(void *);
394 struct tunable_uint64 {
395 	const char *path;
396 	uint64_t *var;
397 };
398 #define	TUNABLE_UINT64(path, var)				\
399 	static struct tunable_uint64 __CONCAT(__tunable_uint64_, __LINE__) = { \
400 		(path),						\
401 		(var),						\
402 	};							\
403 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
404 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_uint64_init, \
405 	    &__CONCAT(__tunable_uint64_, __LINE__))
406 
407 #define	TUNABLE_UINT64_FETCH(path, var)	getenv_uint64((path), (var))
408 
409 /*
410  * quad
411  */
412 extern void tunable_quad_init(void *);
413 struct tunable_quad {
414 	const char *path;
415 	quad_t *var;
416 };
417 #define	TUNABLE_QUAD(path, var)					\
418 	static struct tunable_quad __CONCAT(__tunable_quad_, __LINE__) = { \
419 		(path),						\
420 		(var),						\
421 	};							\
422 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
423 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_quad_init, \
424 	    &__CONCAT(__tunable_quad_, __LINE__))
425 
426 #define	TUNABLE_QUAD_FETCH(path, var)	getenv_quad((path), (var))
427 
428 /*
429  * bool
430  */
431 extern void tunable_bool_init(void *);
432 struct tunable_bool {
433 	const char *path;
434 	bool *var;
435 };
436 #define	TUNABLE_BOOL(path, var) \
437 	static struct tunable_bool __CONCAT(__tunable_bool_, __LINE__) = { \
438 		(path),						\
439 		(var),						\
440 	};							\
441 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
442 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_bool_init, \
443 	    &__CONCAT(__tunable_bool_, __LINE__))
444 
445 #define	TUNABLE_BOOL_FETCH(path, var)	getenv_bool((path), (var))
446 
447 extern void tunable_str_init(void *);
448 struct tunable_str {
449 	const char *path;
450 	char *var;
451 	int size;
452 };
453 #define	TUNABLE_STR(path, var, size)				\
454 	static struct tunable_str __CONCAT(__tunable_str_, __LINE__) = { \
455 		(path),						\
456 		(var),						\
457 		(size),						\
458 	};							\
459 	SYSINIT(__CONCAT(__Tunable_init_, __LINE__),		\
460 	    SI_SUB_TUNABLES, SI_ORDER_MIDDLE, tunable_str_init,	\
461 	    &__CONCAT(__tunable_str_, __LINE__))
462 
463 #define	TUNABLE_STR_FETCH(path, var, size)			\
464 	getenv_string((path), (var), (size))
465 
466 #endif /* _KERNEL */
467 
468 typedef void (*ich_func_t)(void *_arg);
469 
470 struct intr_config_hook {
471 	STAILQ_ENTRY(intr_config_hook) ich_links;
472 	uintptr_t	ich_state;
473 #define ICHS_QUEUED	0x1
474 #define ICHS_RUNNING	0x2
475 #define	ICHS_DONE	0x3
476 	ich_func_t	ich_func;
477 	void		*ich_arg;
478 };
479 
480 int	config_intrhook_establish(struct intr_config_hook *hook);
481 void	config_intrhook_disestablish(struct intr_config_hook *hook);
482 int	config_intrhook_drain(struct intr_config_hook *hook);
483 void	config_intrhook_oneshot(ich_func_t _func, void *_arg);
484 
485 #endif /* !_SYS_KERNEL_H_*/
486