2 * Copyright (c) 1986, 1989, 1991, 1993
3 * The Regents of the University of California. All rights reserved.
4 * (c) UNIX System Laboratories, Inc.
5 * All or some portions of this file are derived from material licensed
6 * to the University of California by American Telephone and Telegraph
7 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
8 * the permission of UNIX System Laboratories, Inc.
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 * 4. Neither the name of the University nor the names of its contributors
19 * may be used to endorse or promote products derived from this software
20 * without specific prior written permission.
22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 * @(#)proc.h 8.15 (Berkeley) 5/19/95
41 #include <sys/callout.h> /* For struct callout. */
42 #include <sys/event.h> /* For struct klist. */
44 #include <sys/filedesc.h>
46 #include <sys/queue.h>
47 #include <sys/_lock.h>
48 #include <sys/_mutex.h>
49 #include <sys/priority.h>
50 #include <sys/rtprio.h> /* XXX. */
52 #include <sys/sigio.h>
53 #include <sys/signal.h>
54 #include <sys/signalvar.h>
56 #include <sys/time.h> /* For structs itimerval, timeval. */
60 #include <sys/ucontext.h>
61 #include <sys/ucred.h>
62 #include <machine/proc.h> /* Machine-dependent proc substruct. */
65 * One structure allocated per session.
68 * (m) locked by s_mtx mtx
69 * (e) locked by proctree_lock sx
70 * (c) const until freeing
73 int s_count; /* (m) Ref cnt; pgrps in session. */
74 struct proc *s_leader; /* (m + e) Session leader. */
75 struct vnode *s_ttyvp; /* (m) Vnode of controlling tty. */
76 struct tty *s_ttyp; /* (m) Controlling tty. */
77 pid_t s_sid; /* (c) Session ID. */
78 /* (m) Setlogin() name: */
79 char s_login[roundup(MAXLOGNAME, sizeof(long))];
80 struct mtx s_mtx; /* Mutex to protect members. */
84 * One structure allocated per process group.
87 * (m) locked by pg_mtx mtx
88 * (e) locked by proctree_lock sx
89 * (c) const until freeing
92 LIST_ENTRY(pgrp) pg_hash; /* (e) Hash chain. */
93 LIST_HEAD(, proc) pg_members; /* (m + e) Pointer to pgrp members. */
94 struct session *pg_session; /* (c) Pointer to session. */
95 struct sigiolst pg_sigiolst; /* (m) List of sigio sources. */
96 pid_t pg_id; /* (c) Process group id. */
97 int pg_jobc; /* (m) Job control process count. */
98 struct mtx pg_mtx; /* Mutex to protect members */
102 * pargs, used to hold a copy of the command line, if it had a sane length.
105 u_int ar_ref; /* Reference count. */
106 u_int ar_length; /* Length. */
107 u_char ar_args[1]; /* Arguments. */
111 * Description of a process.
113 * This structure contains the information needed to manage a thread of
114 * control, known in UN*X as a process; it has references to substructures
115 * containing descriptions of things that the process uses, but may share
116 * with related processes. The process structure and the substructures
117 * are always addressable except for those marked "(CPU)" below,
118 * which might be addressable only on a processor on which the process
121 * Below is a key of locks used to protect each member of struct proc. The
122 * lock is indicated by a reference to a specific character in parens in the
123 * associated comment.
124 * * - not yet protected
125 * a - only touched by curproc or parent during fork/wait
126 * b - created at fork, never changes
127 * (exception aiods switch vmspaces, but they are also
128 * marked 'P_SYSTEM' so hopefully it will be left alone)
129 * c - locked by proc mtx
130 * d - locked by allproc_lock lock
131 * e - locked by proctree_lock lock
133 * g - process group mtx
134 * h - callout_lock mtx
135 * i - by curproc or the master session mtx
136 * j - locked by sched_lock mtx
137 * k - only accessed by curthread
138 * k*- only accessed by curthread and from an interrupt
139 * l - the attaching proc or attaching proc parent
141 * n - not locked, lazy
143 * p - select lock (sellock)
144 * q - td_contested lock
146 * x - created at fork, only changes during single threading in exec
147 * z - zombie threads/ksegroup lock
149 * If the locking key specifies two identifiers (for example, p_pptr) then
150 * either lock is sufficient for read access, but both locks must be held
154 struct kaudit_record;
163 struct mqueue_notifier;
166 * Here we define the three structures used for process information.
168 * The first is the thread. It might be thought of as a "Kernel
169 * Schedulable Entity Context".
170 * This structure contains all the information as to where a thread of
171 * execution is now, or was when it was suspended, why it was suspended,
172 * and anything else that will be needed to restart it when it is
173 * rescheduled. Always associated with a KSE when running, but can be
174 * reassigned to an equivalent KSE when being restarted for
175 * load balancing. Each of these is associated with a kernel stack
178 * It is important to remember that a particular thread structure may only
179 * exist as long as the system call or kernel entrance (e.g. by pagefault)
180 * which it is currently executing. It should therefore NEVER be referenced
181 * by pointers in long lived structures that live longer than a single
182 * request. If several threads complete their work at the same time,
183 * they will all rewind their stacks to the user boundary, report their
184 * completion state, and all but one will be freed. That last one will
185 * be kept to provide a kernel stack and pcb for the NEXT syscall or kernel
186 * entrance (basically to save freeing and then re-allocating it). The existing
187 * thread keeps a cached spare thread available to allow it to quickly
188 * get one when it needs a new one. There is also a system
189 * cache of free threads. Threads have priority and partake in priority
190 * inheritance schemes.
195 * The KSEGRP is allocated resources across a number of CPUs.
196 * (Including a number of CPUxQUANTA. It parcels these QUANTA up among
197 * its threads, each of which should be running in a different CPU.
198 * BASE priority and total available quanta are properties of a KSEGRP.
199 * Multiple KSEGRPs in a single process compete against each other
200 * for total quanta in the same way that a forked child competes against
201 * it's parent process.
206 * A process is the owner of all system resources allocated to a task
208 * All KSEGs under one process see, and have the same access to, these
209 * resources (e.g. files, memory, sockets, credential, kqueues).
210 * A process may compete for CPU cycles on the same basis as a
211 * forked process cluster by spawning several KSEGRPs.
217 With a single run queue used by all processors:
219 RUNQ: --->KSE---KSE--... SLEEPQ:[]---THREAD---THREAD---THREAD
221 KSEG---THREAD--THREAD--THREAD []
224 (processors run THREADs from the KSEG until they are exhausted or
225 the KSEG exhausts its quantum)
227 With PER-CPU run queues:
228 KSEs on the separate run queues directly
229 They would be given priorities calculated from the KSEG.
235 * Kernel runnable context (thread).
236 * This is what is put to sleep and reactivated.
237 * The first KSE available in the correct group will run this thread.
238 * If several are available, use the one on the same CPU as last time.
239 * When waiting to be run, threads are hung off the KSEGRP in priority order.
240 * With N runnable and queued KSEs in the KSEGRP, the first N threads
241 * are linked to them. Other threads are not yet assigned.
244 struct proc *td_proc; /* (*) Associated process. */
245 struct ksegrp *td_ksegrp; /* (*) Associated KSEG. */
246 TAILQ_ENTRY(thread) td_plist; /* (*) All threads in this proc. */
247 TAILQ_ENTRY(thread) td_kglist; /* (*) All threads in this ksegrp. */
249 /* The two queues below should someday be merged. */
250 TAILQ_ENTRY(thread) td_slpq; /* (j) Sleep queue. */
251 TAILQ_ENTRY(thread) td_lockq; /* (j) Lock queue. */
252 TAILQ_ENTRY(thread) td_runq; /* (j/z) Run queue(s). XXXKSE */
254 TAILQ_HEAD(, selinfo) td_selq; /* (p) List of selinfos. */
255 struct sleepqueue *td_sleepqueue; /* (k) Associated sleep queue. */
256 struct turnstile *td_turnstile; /* (k) Associated turnstile. */
257 struct umtx_q *td_umtxq; /* (c?) Link for when we're blocked. */
258 lwpid_t td_tid; /* (b) Thread ID. */
259 sigqueue_t td_sigqueue; /* (c) Sigs arrived, not delivered. */
260 #define td_siglist td_sigqueue.sq_signals
262 /* Cleared during fork1() or thread_schedule_upcall(). */
263 #define td_startzero td_flags
264 int td_flags; /* (j) TDF_* flags. */
265 int td_inhibitors; /* (j) Why can not run. */
266 int td_pflags; /* (k) Private thread (TDP_*) flags. */
267 int td_dupfd; /* (k) Ret value from fdopen. XXX */
268 void *td_wchan; /* (j) Sleep address. */
269 const char *td_wmesg; /* (j) Reason for sleep. */
270 u_char td_lastcpu; /* (j) Last cpu we were on. */
271 u_char td_oncpu; /* (j) Which cpu we are on. */
272 volatile u_char td_owepreempt; /* (k*) Preempt on last critical_exit */
273 short td_locks; /* (k) DEBUG: lockmgr count of locks. */
274 u_char td_tsqueue; /* (j) Turnstile queue blocked on. */
275 struct turnstile *td_blocked; /* (j) Lock thread is blocked on. */
276 const char *td_lockname; /* (j) Name of lock blocked on. */
277 LIST_HEAD(, turnstile) td_contested; /* (q) Contested locks. */
278 struct lock_list_entry *td_sleeplocks; /* (k) Held sleep locks. */
279 int td_intr_nesting_level; /* (k) Interrupt recursion. */
280 int td_pinned; /* (k) Temporary cpu pin count. */
281 struct kse_thr_mailbox *td_mailbox; /* (*) Userland mailbox address. */
282 struct ucred *td_ucred; /* (k) Reference to credentials. */
283 struct thread *td_standin; /* (k + a) Use this for an upcall. */
284 struct kse_upcall *td_upcall; /* (k + j) Upcall structure. */
285 u_int td_pticks; /* (k) Statclock hits for profiling */
286 u_int td_sticks; /* (k) Statclock hits in system mode. */
287 u_int td_iticks; /* (k) Statclock hits in intr mode. */
288 u_int td_uticks; /* (k) Statclock hits in user mode. */
289 u_int td_uuticks; /* (k) Statclock hits (usr), for UTS. */
290 u_int td_usticks; /* (k) Statclock hits (sys), for UTS. */
291 int td_intrval; /* (j) Return value of TDF_INTERRUPT. */
292 sigset_t td_oldsigmask; /* (k) Saved mask from pre sigpause. */
293 sigset_t td_sigmask; /* (c) Current signal mask. */
294 volatile u_int td_generation; /* (k) For detection of preemption */
295 stack_t td_sigstk; /* (k) Stack ptr and on-stack flag. */
296 int td_kflags; /* (c) Flags for KSE threading. */
297 int td_xsig; /* (c) Signal for ptrace */
298 u_long td_profil_addr; /* (k) Temporary addr until AST. */
299 u_int td_profil_ticks; /* (k) Temporary ticks until AST. */
300 char td_name[MAXCOMLEN + 1]; /* (*) Thread name. */
301 #define td_endzero td_base_pri
303 /* Copied during fork1() or thread_sched_upcall(). */
304 #define td_startcopy td_endzero
305 u_char td_base_pri; /* (j) Thread base kernel priority. */
306 u_char td_priority; /* (j) Thread active priority. */
307 #define td_endcopy td_pcb
310 * Fields that must be manually set in fork1() or thread_sched_upcall()
311 * or already have been set in the allocator, constructor, etc.
313 struct pcb *td_pcb; /* (k) Kernel VA of pcb and kstack. */
321 register_t td_retval[2]; /* (k) Syscall aux returns. */
322 struct callout td_slpcallout; /* (h) Callout for sleep. */
323 struct trapframe *td_frame; /* (k) */
324 struct vm_object *td_kstack_obj;/* (a) Kstack object. */
325 vm_offset_t td_kstack; /* (a) Kernel VA of kstack. */
326 int td_kstack_pages; /* (a) Size of the kstack. */
327 struct vm_object *td_altkstack_obj;/* (a) Alternate kstack object. */
328 vm_offset_t td_altkstack; /* (a) Kernel VA of alternate kstack. */
329 int td_altkstack_pages; /* (a) Size of alternate kstack. */
330 volatile u_int td_critnest; /* (k*) Critical section nest level. */
331 struct mdthread td_md; /* (k) Any machine-dependent fields. */
332 struct td_sched *td_sched; /* (*) Scheduler-specific data. */
333 struct kaudit_record *td_ar; /* (k) Active audit record, if any. */
337 * Flags kept in td_flags:
338 * To change these you MUST have the scheduler lock.
340 #define TDF_BORROWING 0x00000001 /* Thread is borrowing pri from another. */
341 #define TDF_INPANIC 0x00000002 /* Caused a panic, let it drive crashdump. */
342 #define TDF_SINTR 0x00000008 /* Sleep is interruptible. */
343 #define TDF_TIMEOUT 0x00000010 /* Timing out during sleep. */
344 #define TDF_IDLETD 0x00000020 /* This is a per-CPU idle thread. */
345 #define TDF_SELECT 0x00000040 /* Selecting; wakeup/waiting danger. */
346 #define TDF_SLEEPABORT 0x00000080 /* sleepq_abort was called. */
347 #define TDF_TSNOBLOCK 0x00000100 /* Don't block on a turnstile due to race. */
348 #define TDF_UNUSED9 0x00000200 /* --available -- */
349 #define TDF_BOUNDARY 0x00000400 /* Thread suspended at user boundary */
350 #define TDF_ASTPENDING 0x00000800 /* Thread has some asynchronous events. */
351 #define TDF_TIMOFAIL 0x00001000 /* Timeout from sleep after we were awake. */
352 #define TDF_INTERRUPT 0x00002000 /* Thread is marked as interrupted. */
353 #define TDF_UNUSED14 0x00004000 /* --available -- */
354 #define TDF_UNUSED15 0x00008000 /* --available -- */
355 #define TDF_NEEDRESCHED 0x00010000 /* Thread needs to yield. */
356 #define TDF_NEEDSIGCHK 0x00020000 /* Thread may need signal delivery. */
357 #define TDF_XSIG 0x00040000 /* Thread is exchanging signal under trace */
358 #define TDF_UMTXQ 0x00080000 /* Thread is sleeping on a umtx. */
359 #define TDF_THRWAKEUP 0x00100000 /* Libthr thread must not suspend itself. */
360 #define TDF_DBSUSPEND 0x00200000 /* Thread is suspended by debugger */
361 #define TDF_UNUSED22 0x00400000 /* --available -- */
362 #define TDF_UNUSED23 0x00800000 /* --available -- */
363 #define TDF_SCHED0 0x01000000 /* Reserved for scheduler private use */
364 #define TDF_SCHED1 0x02000000 /* Reserved for scheduler private use */
365 #define TDF_SCHED2 0x04000000 /* Reserved for scheduler private use */
366 #define TDF_SCHED3 0x08000000 /* Reserved for scheduler private use */
369 * "Private" flags kept in td_pflags:
370 * These are only accessed by curthread and thus need no locking.
372 #define TDP_OLDMASK 0x00000001 /* Need to restore mask after suspend. */
373 #define TDP_INKTR 0x00000002 /* Thread is currently in KTR code. */
374 #define TDP_INKTRACE 0x00000004 /* Thread is currently in KTRACE code. */
375 #define TDP_UPCALLING 0x00000008 /* This thread is doing an upcall. */
376 #define TDP_COWINPROGRESS 0x00000010 /* Snapshot copy-on-write in progress. */
377 #define TDP_ALTSTACK 0x00000020 /* Have alternate signal stack. */
378 #define TDP_DEADLKTREAT 0x00000040 /* Lock aquisition - deadlock treatment. */
379 #define TDP_SA 0x00000080 /* A scheduler activation based thread. */
380 #define TDP_NOSLEEPING 0x00000100 /* Thread is not allowed to sleep on a sq. */
381 #define TDP_OWEUPC 0x00000200 /* Call addupc() at next AST. */
382 #define TDP_ITHREAD 0x00000400 /* Thread is an interrupt thread. */
383 #define TDP_CAN_UNBIND 0x00000800 /* Only temporarily bound. */
384 #define TDP_SCHED1 0x00001000 /* Reserved for scheduler private use */
385 #define TDP_SCHED2 0x00002000 /* Reserved for scheduler private use */
386 #define TDP_SCHED3 0x00004000 /* Reserved for scheduler private use */
387 #define TDP_SCHED4 0x00008000 /* Reserved for scheduler private use */
388 #define TDP_GEOM 0x00010000 /* Settle GEOM before finishing syscall */
389 #define TDP_SOFTDEP 0x00020000 /* Stuck processing softdep worklist */
390 #define TDP_NORUNNINGBUF 0x00040000 /* Ignore runningbufspace check */
393 * Reasons that the current thread can not be run yet.
394 * More than one may apply.
396 #define TDI_SUSPENDED 0x0001 /* On suspension queue. */
397 #define TDI_SLEEPING 0x0002 /* Actually asleep! (tricky). */
398 #define TDI_SWAPPED 0x0004 /* Stack not in mem. Bad juju if run. */
399 #define TDI_LOCK 0x0008 /* Stopped on a lock. */
400 #define TDI_IWAIT 0x0010 /* Awaiting interrupt. */
403 * flags (in kflags) related to M:N threading.
405 #define TDK_KSEREL 0x0001 /* Blocked in msleep on kg->kg_completed. */
406 #define TDK_KSERELSIG 0x0002 /* Blocked in msleep on p->p_siglist. */
407 #define TDK_WAKEUP 0x0004 /* Thread has been woken by kse_wakeup. */
409 #define TD_CAN_UNBIND(td) \
410 (((td)->td_pflags & TDP_CAN_UNBIND) && \
411 ((td)->td_upcall != NULL))
413 #define TD_IS_SLEEPING(td) ((td)->td_inhibitors & TDI_SLEEPING)
414 #define TD_ON_SLEEPQ(td) ((td)->td_wchan != NULL)
415 #define TD_IS_SUSPENDED(td) ((td)->td_inhibitors & TDI_SUSPENDED)
416 #define TD_IS_SWAPPED(td) ((td)->td_inhibitors & TDI_SWAPPED)
417 #define TD_ON_LOCK(td) ((td)->td_inhibitors & TDI_LOCK)
418 #define TD_AWAITING_INTR(td) ((td)->td_inhibitors & TDI_IWAIT)
419 #define TD_IS_RUNNING(td) ((td)->td_state == TDS_RUNNING)
420 #define TD_ON_RUNQ(td) ((td)->td_state == TDS_RUNQ)
421 #define TD_CAN_RUN(td) ((td)->td_state == TDS_CAN_RUN)
422 #define TD_IS_INHIBITED(td) ((td)->td_state == TDS_INHIBITED)
424 #define TD_SET_INHIB(td, inhib) do { \
425 (td)->td_state = TDS_INHIBITED; \
426 (td)->td_inhibitors |= (inhib); \
429 #define TD_CLR_INHIB(td, inhib) do { \
430 if (((td)->td_inhibitors & (inhib)) && \
431 (((td)->td_inhibitors &= ~(inhib)) == 0)) \
432 (td)->td_state = TDS_CAN_RUN; \
435 #define TD_SET_SLEEPING(td) TD_SET_INHIB((td), TDI_SLEEPING)
436 #define TD_SET_SWAPPED(td) TD_SET_INHIB((td), TDI_SWAPPED)
437 #define TD_SET_LOCK(td) TD_SET_INHIB((td), TDI_LOCK)
438 #define TD_SET_SUSPENDED(td) TD_SET_INHIB((td), TDI_SUSPENDED)
439 #define TD_SET_IWAIT(td) TD_SET_INHIB((td), TDI_IWAIT)
440 #define TD_SET_EXITING(td) TD_SET_INHIB((td), TDI_EXITING)
442 #define TD_CLR_SLEEPING(td) TD_CLR_INHIB((td), TDI_SLEEPING)
443 #define TD_CLR_SWAPPED(td) TD_CLR_INHIB((td), TDI_SWAPPED)
444 #define TD_CLR_LOCK(td) TD_CLR_INHIB((td), TDI_LOCK)
445 #define TD_CLR_SUSPENDED(td) TD_CLR_INHIB((td), TDI_SUSPENDED)
446 #define TD_CLR_IWAIT(td) TD_CLR_INHIB((td), TDI_IWAIT)
448 #define TD_SET_RUNNING(td) (td)->td_state = TDS_RUNNING
449 #define TD_SET_RUNQ(td) (td)->td_state = TDS_RUNQ
450 #define TD_SET_CAN_RUN(td) (td)->td_state = TDS_CAN_RUN
453 * An upcall is used when returning to userland. If a thread does not have
454 * an upcall on return to userland the thread exports its context and exits.
457 TAILQ_ENTRY(kse_upcall) ku_link; /* List of upcalls in KSEG. */
458 struct ksegrp *ku_ksegrp; /* Associated KSEG. */
459 struct thread *ku_owner; /* Owning thread. */
460 int ku_flags; /* KUF_* flags. */
461 struct kse_mailbox *ku_mailbox; /* Userland mailbox address. */
462 stack_t ku_stack; /* Userland upcall stack. */
463 void *ku_func; /* Userland upcall function. */
464 unsigned int ku_mflags; /* Cached upcall mbox flags. */
467 #define KUF_DOUPCALL 0x00001 /* Do upcall now; don't wait. */
468 #define KUF_EXITING 0x00002 /* Upcall structure is exiting. */
471 * Kernel-scheduled entity group (KSEG). The scheduler considers each KSEG to
472 * be an indivisible unit from a time-sharing perspective, though each KSEG may
473 * contain multiple KSEs.
476 struct proc *kg_proc; /* (*) Proc that contains this KSEG. */
477 TAILQ_ENTRY(ksegrp) kg_ksegrp; /* (*) Queue of KSEGs in kg_proc. */
478 TAILQ_HEAD(, thread) kg_threads;/* (td_kglist) All threads. */
479 TAILQ_HEAD(, thread) kg_runq; /* (td_runq) waiting RUNNABLE threads */
480 TAILQ_HEAD(, kse_upcall) kg_upcalls; /* All upcalls in the group. */
482 #define kg_startzero kg_estcpu
483 u_int kg_estcpu; /* (j) Sum of the same field in KSEs. */
484 u_int kg_slptime; /* (j) How long completely blocked. */
485 int kg_numupcalls; /* (j) Num upcalls. */
486 int kg_upsleeps; /* (c) Num threads in kse_release(). */
487 struct kse_thr_mailbox *kg_completed; /* (c) Completed thread mboxes. */
488 int kg_nextupcall; /* (n) Next upcall time. */
489 int kg_upquantum; /* (n) Quantum to schedule an upcall. */
490 #define kg_endzero kg_pri_class
492 #define kg_startcopy kg_endzero
493 u_char kg_pri_class; /* (j) Scheduling class. */
494 u_char kg_user_pri; /* (j) User pri from estcpu and nice. */
495 #define kg_endcopy kg_numthreads
496 int kg_numthreads; /* (j) Num threads in total. */
497 struct kg_sched *kg_sched; /* (*) Scheduler-specific data. */
501 * XXX: Does this belong in resource.h or resourcevar.h instead?
502 * Resource usage extension. The times in rusage structs in the kernel are
503 * never up to date. The actual times are kept as runtimes and tick counts
504 * (with control info in the "previous" times), and are converted when
505 * userland asks for rusage info. Backwards compatibility prevents putting
506 * this directly in the user-visible rusage struct.
508 * Locking: (cj) means (j) for p_rux and (c) for p_crux.
511 u_int64_t rux_runtime; /* (cj) Real time. */
512 u_int64_t rux_uticks; /* (cj) Statclock hits in user mode. */
513 u_int64_t rux_sticks; /* (cj) Statclock hits in sys mode. */
514 u_int64_t rux_iticks; /* (cj) Statclock hits in intr mode. */
515 u_int64_t rux_uu; /* (c) Previous user time in usec. */
516 u_int64_t rux_su; /* (c) Previous sys time in usec. */
517 u_int64_t rux_iu; /* (c) Previous intr time in usec. */
521 * The old fashionned process. May have multiple threads, KSEGRPs
522 * and KSEs. Starts off with a single embedded KSEGRP and THREAD.
525 LIST_ENTRY(proc) p_list; /* (d) List of all processes. */
526 TAILQ_HEAD(, ksegrp) p_ksegrps; /* (c)(kg_ksegrp) All KSEGs. */
527 TAILQ_HEAD(, thread) p_threads; /* (j)(td_plist) Threads. (shortcut) */
528 TAILQ_HEAD(, thread) p_suspended; /* (td_runq) Suspended threads. */
529 struct ucred *p_ucred; /* (c) Process owner's identity. */
530 struct filedesc *p_fd; /* (b) Open files. */
531 struct filedesc_to_leader *p_fdtol; /* (b) Tracking node */
532 /* Accumulated stats for all threads? */
533 struct pstats *p_stats; /* (b) Accounting/statistics (CPU). */
534 struct plimit *p_limit; /* (c) Process limits. */
535 struct sigacts *p_sigacts; /* (x) Signal actions, state (CPU). */
538 * The following don't make too much sense.
539 * See the td_ or ke_ versions of the same flags.
541 int p_flag; /* (c) P_* flags. */
542 int p_sflag; /* (j) PS_* flags. */
544 PRS_NEW = 0, /* In creation */
545 PRS_NORMAL, /* threads can be run. */
547 } p_state; /* (j/c) S* process status. */
549 pid_t p_pid; /* (b) Process identifier. */
550 LIST_ENTRY(proc) p_hash; /* (d) Hash chain. */
551 LIST_ENTRY(proc) p_pglist; /* (g + e) List of processes in pgrp. */
552 struct proc *p_pptr; /* (c + e) Pointer to parent process. */
553 LIST_ENTRY(proc) p_sibling; /* (e) List of sibling processes. */
554 LIST_HEAD(, proc) p_children; /* (e) Pointer to list of children. */
555 struct mtx p_mtx; /* (n) Lock for this struct. */
556 struct ksiginfo *p_ksi; /* Locked by parent proc lock */
557 sigqueue_t p_sigqueue; /* (c) Sigs not delivered to a td. */
558 #define p_siglist p_sigqueue.sq_signals
560 /* The following fields are all zeroed upon creation in fork. */
561 #define p_startzero p_oppid
562 pid_t p_oppid; /* (c + e) Save ppid in ptrace. XXX */
563 struct vmspace *p_vmspace; /* (b) Address space. */
564 u_int p_swtime; /* (j) Time swapped in or out. */
565 struct itimerval p_realtimer; /* (c) Alarm timer. */
566 struct rusage_ext p_rux; /* (cj) Internal resource usage. */
567 struct rusage_ext p_crux; /* (c) Internal child resource usage. */
568 int p_profthreads; /* (c) Num threads in addupc_task. */
569 int p_maxthrwaits; /* (c) Max threads num waiters */
570 int p_traceflag; /* (o) Kernel trace points. */
571 struct vnode *p_tracevp; /* (c + o) Trace to vnode. */
572 struct ucred *p_tracecred; /* (o) Credentials to trace with. */
573 struct vnode *p_textvp; /* (b) Vnode of executable. */
574 char p_lock; /* (c) Proclock (prevent swap) count. */
575 struct sigiolst p_sigiolst; /* (c) List of sigio sources. */
576 int p_sigparent; /* (c) Signal to parent on exit. */
577 int p_sig; /* (n) For core dump/debugger XXX. */
578 u_long p_code; /* (n) For core dump/debugger XXX. */
579 u_int p_stops; /* (c) Stop event bitmask. */
580 u_int p_stype; /* (c) Stop event type. */
581 char p_step; /* (c) Process is stopped. */
582 u_char p_pfsflags; /* (c) Procfs flags. */
583 struct nlminfo *p_nlminfo; /* (?) Only used by/for lockd. */
584 struct kaioinfo *p_aioinfo; /* (c) ASYNC I/O info. */
585 struct thread *p_singlethread;/* (c + j) If single threading this is it */
586 int p_suspcount; /* (c) Num threads in suspended mode. */
587 struct thread *p_xthread; /* (c) Trap thread */
588 int p_boundary_count;/* (c) Num threads at user boundary */
589 struct ksegrp *p_procscopegrp;
590 int p_pendingcnt; /* how many signals are pending */
591 struct itimers *p_itimers; /* (c) POSIX interval timers. */
592 /* End area that is zeroed on creation. */
593 #define p_endzero p_magic
595 /* The following fields are all copied upon creation in fork. */
596 #define p_startcopy p_endzero
597 u_int p_magic; /* (b) Magic number. */
598 char p_comm[MAXCOMLEN + 1]; /* (b) Process name. */
599 struct pgrp *p_pgrp; /* (c + e) Pointer to process group. */
600 struct sysentvec *p_sysent; /* (b) Syscall dispatch info. */
601 struct pargs *p_args; /* (c) Process arguments. */
602 rlim_t p_cpulimit; /* (j) Current CPU limit in seconds. */
603 signed char p_nice; /* (c + j) Process "nice" value. */
604 /* End area that is copied on creation. */
605 #define p_endcopy p_xstat
607 u_short p_xstat; /* (c) Exit status; also stop sig. */
608 struct knlist p_klist; /* (c) Knotes attached to this proc. */
609 int p_numthreads; /* (j) Number of threads. */
610 int p_numksegrps; /* (c) Number of ksegrps. */
611 struct mdproc p_md; /* Any machine-dependent fields. */
612 struct callout p_itcallout; /* (h + c) Interval timer callout. */
613 u_short p_acflag; /* (c) Accounting flags. */
614 struct rusage *p_ru; /* (a) Exit information. XXX */
615 struct proc *p_peers; /* (r) */
616 struct proc *p_leader; /* (b) */
617 void *p_emuldata; /* (c) Emulator state data. */
618 struct label *p_label; /* (*) Proc (not subject) MAC label. */
619 struct p_sched *p_sched; /* (*) Scheduler-specific data. */
620 STAILQ_HEAD(, ktr_request) p_ktr; /* (o) KTR event queue. */
621 LIST_HEAD(, mqueue_notifier) p_mqnotifier; /* (c) mqueue notifiers.*/
622 struct auditinfo *p_au; /* (c) Process audit properties. */
625 #define p_session p_pgrp->pg_session
626 #define p_pgid p_pgrp->pg_id
628 #define NOCPU 0xff /* For when we aren't on a CPU. */
631 /* These flags are kept in p_flag. */
632 #define P_ADVLOCK 0x00001 /* Process may hold a POSIX advisory lock. */
633 #define P_CONTROLT 0x00002 /* Has a controlling terminal. */
634 #define P_KTHREAD 0x00004 /* Kernel thread (*). */
635 #define P_NOLOAD 0x00008 /* Ignore during load avg calculations. */
636 #define P_PPWAIT 0x00010 /* Parent is waiting for child to exec/exit. */
637 #define P_PROFIL 0x00020 /* Has started profiling. */
638 #define P_STOPPROF 0x00040 /* Has thread requesting to stop profiling. */
639 #define P_HADTHREADS 0x00080 /* Has had threads (no cleanup shortcuts) */
640 #define P_SUGID 0x00100 /* Had set id privileges since last exec. */
641 #define P_SYSTEM 0x00200 /* System proc: no sigs, stats or swapping. */
642 #define P_SINGLE_EXIT 0x00400 /* Threads suspending should exit, not wait. */
643 #define P_TRACED 0x00800 /* Debugged process being traced. */
644 #define P_WAITED 0x01000 /* Someone is waiting for us. */
645 #define P_WEXIT 0x02000 /* Working on exiting. */
646 #define P_EXEC 0x04000 /* Process called exec. */
647 #define P_SA 0x08000 /* Using scheduler activations. */
648 #define P_CONTINUED 0x10000 /* Proc has continued from a stopped state. */
649 #define P_STOPPED_SIG 0x20000 /* Stopped due to SIGSTOP/SIGTSTP. */
650 #define P_STOPPED_TRACE 0x40000 /* Stopped because of tracing. */
651 #define P_STOPPED_SINGLE 0x80000 /* Only 1 thread can continue (not to user). */
652 #define P_PROTECTED 0x100000 /* Do not kill on memory overcommit. */
653 #define P_SIGEVENT 0x200000 /* Process pending signals changed. */
654 #define P_SINGLE_BOUNDARY 0x400000 /* Threads should suspend at user boundary. */
655 #define P_HWPMC 0x800000 /* Process is using HWPMCs */
657 #define P_JAILED 0x1000000 /* Process is in jail. */
658 #define P_INEXEC 0x4000000 /* Process is in execve(). */
659 #define P_STATCHILD 0x8000000 /* Child process stopped or exited. */
661 #define P_STOPPED (P_STOPPED_SIG|P_STOPPED_SINGLE|P_STOPPED_TRACE)
662 #define P_SHOULDSTOP(p) ((p)->p_flag & P_STOPPED)
664 /* These flags are kept in p_sflag and are protected with sched_lock. */
665 #define PS_INMEM 0x00001 /* Loaded into memory. */
666 #define PS_XCPU 0x00002 /* Exceeded CPU limit. */
667 #define PS_ALRMPEND 0x00020 /* Pending SIGVTALRM needs to be posted. */
668 #define PS_PROFPEND 0x00040 /* Pending SIGPROF needs to be posted. */
669 #define PS_SWAPINREQ 0x00100 /* Swapin request due to wakeup. */
670 #define PS_SWAPPINGOUT 0x00200 /* Process is being swapped out. */
671 #define PS_SWAPPINGIN 0x04000 /* Process is being swapped in. */
672 #define PS_MACPEND 0x08000 /* AST-based MAC event pending. */
675 * These were process status values (p_stat), now they are only used in
676 * legacy conversion code.
678 #define SIDL 1 /* Process being created by fork. */
679 #define SRUN 2 /* Currently runnable. */
680 #define SSLEEP 3 /* Sleeping on an address. */
681 #define SSTOP 4 /* Process debugging or suspension. */
682 #define SZOMB 5 /* Awaiting collection by parent. */
683 #define SWAIT 6 /* Waiting for interrupt. */
684 #define SLOCK 7 /* Blocked on a lock. */
686 #define P_MAGIC 0xbeefface
690 /* Flags for mi_switch(). */
691 #define SW_VOL 0x0001 /* Voluntary switch. */
692 #define SW_INVOL 0x0002 /* Involuntary switch. */
693 #define SW_PREEMPT 0x0004 /* The invol switch is a preemption */
695 /* Flags for setrunqueue(). Why are we setting this thread on the run queue? */
696 #define SRQ_BORING 0x0000 /* No special circumstances. */
697 #define SRQ_YIELDING 0x0001 /* We are yielding (from mi_switch). */
698 #define SRQ_OURSELF 0x0002 /* It is ourself (from mi_switch). */
699 #define SRQ_INTR 0x0004 /* It is probably urgent. */
700 #define SRQ_PREEMPTED 0x0008 /* has been preempted.. be kind */
702 /* How values for thread_single(). */
703 #define SINGLE_NO_EXIT 0
704 #define SINGLE_EXIT 1
705 #define SINGLE_BOUNDARY 2
707 /* XXXKSE: Missing values for thread_suspsend_check(). */
709 #ifdef MALLOC_DECLARE
710 MALLOC_DECLARE(M_PARGS);
711 MALLOC_DECLARE(M_PGRP);
712 MALLOC_DECLARE(M_SESSION);
713 MALLOC_DECLARE(M_SUBPROC);
714 MALLOC_DECLARE(M_ZOMBIE);
717 #define FOREACH_PROC_IN_SYSTEM(p) \
718 LIST_FOREACH((p), &allproc, p_list)
719 #define FOREACH_KSEGRP_IN_PROC(p, kg) \
720 TAILQ_FOREACH((kg), &(p)->p_ksegrps, kg_ksegrp)
721 #define FOREACH_THREAD_IN_GROUP(kg, td) \
722 TAILQ_FOREACH((td), &(kg)->kg_threads, td_kglist)
723 #define FOREACH_UPCALL_IN_GROUP(kg, ku) \
724 TAILQ_FOREACH((ku), &(kg)->kg_upcalls, ku_link)
725 #define FOREACH_THREAD_IN_PROC(p, td) \
726 TAILQ_FOREACH((td), &(p)->p_threads, td_plist)
728 /* XXXKSE the following lines should probably only be used in 1:1 code: */
729 #define FIRST_THREAD_IN_PROC(p) TAILQ_FIRST(&(p)->p_threads)
730 #define FIRST_KSEGRP_IN_PROC(p) TAILQ_FIRST(&(p)->p_ksegrps)
733 * We use process IDs <= PID_MAX; PID_MAX + 1 must also fit in a pid_t,
734 * as it is used to represent "no process group".
736 #define PID_MAX 99999
737 #define NO_PID 100000
739 #define SESS_LEADER(p) ((p)->p_session->s_leader == (p))
740 #define SESSHOLD(s) ((s)->s_count++)
741 #define SESSRELE(s) sessrele(s)
744 #define STOPEVENT(p, e, v) do { \
745 if ((p)->p_stops & (e)) { \
747 stopevent((p), (e), (v)); \
751 #define _STOPEVENT(p, e, v) do { \
752 PROC_LOCK_ASSERT(p, MA_OWNED); \
753 WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, &p->p_mtx.mtx_object, \
754 "checking stopevent %d", (e)); \
755 if ((p)->p_stops & (e)) \
756 stopevent((p), (e), (v)); \
759 /* Lock and unlock a process. */
760 #define PROC_LOCK(p) mtx_lock(&(p)->p_mtx)
761 #define PROC_TRYLOCK(p) mtx_trylock(&(p)->p_mtx)
762 #define PROC_UNLOCK(p) mtx_unlock(&(p)->p_mtx)
763 #define PROC_LOCKED(p) mtx_owned(&(p)->p_mtx)
764 #define PROC_LOCK_ASSERT(p, type) mtx_assert(&(p)->p_mtx, (type))
766 /* Lock and unlock a process group. */
767 #define PGRP_LOCK(pg) mtx_lock(&(pg)->pg_mtx)
768 #define PGRP_UNLOCK(pg) mtx_unlock(&(pg)->pg_mtx)
769 #define PGRP_LOCKED(pg) mtx_owned(&(pg)->pg_mtx)
770 #define PGRP_LOCK_ASSERT(pg, type) mtx_assert(&(pg)->pg_mtx, (type))
772 #define PGRP_LOCK_PGSIGNAL(pg) do { \
776 #define PGRP_UNLOCK_PGSIGNAL(pg) do { \
781 /* Lock and unlock a session. */
782 #define SESS_LOCK(s) mtx_lock(&(s)->s_mtx)
783 #define SESS_UNLOCK(s) mtx_unlock(&(s)->s_mtx)
784 #define SESS_LOCKED(s) mtx_owned(&(s)->s_mtx)
785 #define SESS_LOCK_ASSERT(s, type) mtx_assert(&(s)->s_mtx, (type))
787 /* Hold process U-area in memory, normally for ptrace/procfs work. */
788 #define PHOLD(p) do { \
793 #define _PHOLD(p) do { \
794 PROC_LOCK_ASSERT((p), MA_OWNED); \
795 KASSERT(!((p)->p_flag & P_WEXIT) || (p) == curproc, \
796 ("PHOLD of exiting process")); \
798 if (((p)->p_sflag & PS_INMEM) == 0) \
802 #define PRELE(p) do { \
807 #define _PRELE(p) do { \
808 PROC_LOCK_ASSERT((p), MA_OWNED); \
810 if (((p)->p_flag & P_WEXIT) && (p)->p_lock == 0) \
811 wakeup(&(p)->p_lock); \
814 /* Check whether a thread is safe to be swapped out. */
815 #define thread_safetoswapout(td) (TD_IS_SLEEPING(td) || TD_IS_SUSPENDED(td))
817 /* Control whether or not it is safe for curthread to sleep. */
818 #define THREAD_NO_SLEEPING() do { \
819 KASSERT(!(curthread->td_pflags & TDP_NOSLEEPING), \
820 ("nested no sleeping")); \
821 curthread->td_pflags |= TDP_NOSLEEPING; \
824 #define THREAD_SLEEPING_OK() do { \
825 KASSERT((curthread->td_pflags & TDP_NOSLEEPING), \
826 ("nested sleeping ok")); \
827 curthread->td_pflags &= ~TDP_NOSLEEPING; \
830 #define PIDHASH(pid) (&pidhashtbl[(pid) & pidhash])
831 extern LIST_HEAD(pidhashhead, proc) *pidhashtbl;
832 extern u_long pidhash;
834 #define PGRPHASH(pgid) (&pgrphashtbl[(pgid) & pgrphash])
835 extern LIST_HEAD(pgrphashhead, pgrp) *pgrphashtbl;
836 extern u_long pgrphash;
838 extern struct sx allproc_lock;
839 extern struct sx proctree_lock;
840 extern struct mtx ppeers_lock;
841 extern struct ksegrp ksegrp0; /* Primary ksegrp in proc0. */
842 extern struct proc proc0; /* Process slot for swapper. */
843 extern struct thread thread0; /* Primary thread in proc0. */
844 extern struct vmspace vmspace0; /* VM space for proc0. */
845 extern int hogticks; /* Limit on kernel cpu hogs. */
847 extern int nprocs, maxproc; /* Current and max number of procs. */
848 extern int maxprocperuid; /* Max procs per uid. */
849 extern u_long ps_arg_cache_limit;
851 LIST_HEAD(proclist, proc);
852 TAILQ_HEAD(procqueue, proc);
853 TAILQ_HEAD(threadqueue, thread);
854 extern struct proclist allproc; /* List of all processes. */
855 extern struct proclist zombproc; /* List of zombie processes. */
856 extern struct proc *initproc, *pageproc; /* Process slots for init, pager. */
858 extern struct uma_zone *proc_zone;
860 struct proc *pfind(pid_t); /* Find process by id. */
861 struct pgrp *pgfind(pid_t); /* Find process group by id. */
862 struct proc *zpfind(pid_t); /* Find zombie process by id. */
864 void adjustrunqueue(struct thread *, int newpri);
865 void ast(struct trapframe *framep);
866 struct thread *choosethread(void);
867 int cr_cansignal(struct ucred *cred, struct proc *proc, int signum);
868 int enterpgrp(struct proc *p, pid_t pgid, struct pgrp *pgrp,
869 struct session *sess);
870 int enterthispgrp(struct proc *p, struct pgrp *pgrp);
871 void faultin(struct proc *p);
872 void fixjobc(struct proc *p, struct pgrp *pgrp, int entering);
873 int fork1(struct thread *, int, int, struct proc **);
874 void fork_exit(void (*)(void *, struct trapframe *), void *,
876 void fork_return(struct thread *, struct trapframe *);
877 int inferior(struct proc *p);
878 void kick_proc0(void);
879 int leavepgrp(struct proc *p);
880 int maybe_preempt(struct thread *td);
881 void mi_switch(int flags, struct thread *newtd);
882 int p_candebug(struct thread *td, struct proc *p);
883 int p_cansee(struct thread *td, struct proc *p);
884 int p_cansched(struct thread *td, struct proc *p);
885 int p_cansignal(struct thread *td, struct proc *p, int signum);
886 int p_canwait(struct thread *td, struct proc *p);
887 struct pargs *pargs_alloc(int len);
888 void pargs_drop(struct pargs *pa);
889 void pargs_free(struct pargs *pa);
890 void pargs_hold(struct pargs *pa);
892 void proc_linkup(struct proc *p, struct ksegrp *kg, struct thread *td);
893 void proc_reparent(struct proc *child, struct proc *newparent);
894 struct pstats *pstats_alloc(void);
895 void pstats_fork(struct pstats *src, struct pstats *dst);
896 void pstats_free(struct pstats *ps);
897 int securelevel_ge(struct ucred *cr, int level);
898 int securelevel_gt(struct ucred *cr, int level);
899 void sessrele(struct session *);
900 void setrunnable(struct thread *);
901 void setrunqueue(struct thread *, int flags);
902 void setsugid(struct proc *p);
903 int sigonstack(size_t sp);
904 void sleepinit(void);
905 void stopevent(struct proc *, u_int, u_int);
906 void threadinit(void);
908 extern void (*cpu_idle_hook)(void); /* Hook to machdep CPU idler. */
909 void cpu_switch(struct thread *old, struct thread *new);
910 void cpu_throw(struct thread *old, struct thread *new) __dead2;
911 void unsleep(struct thread *);
912 void userret(struct thread *, struct trapframe *);
914 void cpu_exit(struct thread *);
915 void exit1(struct thread *, int) __dead2;
916 void cpu_fork(struct thread *, struct proc *, struct thread *, int);
917 void cpu_set_fork_handler(struct thread *, void (*)(void *), void *);
920 struct ksegrp *ksegrp_alloc(void);
921 void ksegrp_free(struct ksegrp *kg);
922 void ksegrp_stash(struct ksegrp *kg);
925 void cpu_set_upcall(struct thread *td, struct thread *td0);
926 void cpu_set_upcall_kse(struct thread *, void (*)(void *), void *, stack_t *);
927 int cpu_set_user_tls(struct thread *, void *tls_base);
928 void cpu_thread_clean(struct thread *);
929 void cpu_thread_exit(struct thread *);
930 void cpu_thread_setup(struct thread *td);
931 void cpu_thread_swapin(struct thread *);
932 void cpu_thread_swapout(struct thread *);
933 void ksegrp_link(struct ksegrp *kg, struct proc *p);
934 void ksegrp_unlink(struct ksegrp *kg);
935 struct thread *thread_alloc(void);
936 void thread_continued(struct proc *p);
937 void thread_exit(void) __dead2;
938 int thread_export_context(struct thread *td, int willexit);
939 void thread_free(struct thread *td);
940 void thread_link(struct thread *td, struct ksegrp *kg);
941 void thread_reap(void);
942 struct thread *thread_schedule_upcall(struct thread *td, struct kse_upcall *ku);
943 void thread_signal_add(struct thread *td, ksiginfo_t *);
944 int thread_single(int how);
945 void thread_single_end(void);
946 int thread_sleep_check(struct thread *td);
947 void thread_stash(struct thread *td);
948 int thread_statclock(int user);
949 void thread_stopped(struct proc *p);
950 void childproc_stopped(struct proc *child, int reason);
951 void childproc_continued(struct proc *child);
952 void childproc_exited(struct proc *child);
953 int thread_suspend_check(int how);
954 void thread_suspend_one(struct thread *td);
955 struct thread *thread_switchout(struct thread *td, int flags,
956 struct thread *newtd);
957 void thread_unlink(struct thread *td);
958 void thread_unsuspend(struct proc *p);
959 void thread_unsuspend_one(struct thread *td);
960 void thread_unthread(struct thread *td);
961 int thread_userret(struct thread *td, struct trapframe *frame);
962 void thread_user_enter(struct thread *td);
963 void thread_wait(struct proc *p);
964 struct thread *thread_find(struct proc *p, lwpid_t tid);
965 void thr_exit1(void);
966 struct kse_upcall *upcall_alloc(void);
967 void upcall_free(struct kse_upcall *ku);
968 void upcall_link(struct kse_upcall *ku, struct ksegrp *kg);
969 void upcall_unlink(struct kse_upcall *ku);
970 void upcall_remove(struct thread *td);
971 void upcall_stash(struct kse_upcall *ke);
975 #endif /* !_SYS_PROC_H_ */