2 * SPDX-License-Identifier: BSD-4-Clause
4 * Copyright (C) 1994, David Greenman
5 * Copyright (c) 1990, 1993
6 * The Regents of the University of California. All rights reserved.
7 * Copyright (c) 2007 The FreeBSD Foundation
9 * This code is derived from software contributed to Berkeley by
10 * the University of Utah, and William Jolitz.
12 * Portions of this software were developed by A. Joseph Koshy under
13 * sponsorship from the FreeBSD Foundation and Google, Inc.
15 * Redistribution and use in source and binary forms, with or without
16 * modification, are permitted provided that the following conditions
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.
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
43 * from: @(#)trap.c 7.4 (Berkeley) 5/13/91
46 #include <sys/cdefs.h>
47 __FBSDID("$FreeBSD$");
49 #include "opt_hwpmc_hooks.h"
50 #include "opt_ktrace.h"
51 #include "opt_sched.h"
53 #include <sys/param.h>
55 #include <sys/capsicum.h>
56 #include <sys/event.h>
57 #include <sys/kernel.h>
60 #include <sys/mutex.h>
61 #include <sys/pmckern.h>
64 #include <sys/ptrace.h>
65 #include <sys/racct.h>
66 #include <sys/resourcevar.h>
67 #include <sys/sched.h>
68 #include <sys/signalvar.h>
69 #include <sys/syscall.h>
70 #include <sys/syscallsubr.h>
71 #include <sys/sysent.h>
72 #include <sys/systm.h>
73 #include <sys/vmmeter.h>
76 #include <sys/ktrace.h>
78 #include <security/audit/audit.h>
80 #include <machine/cpu.h>
87 #include <sys/pmckern.h>
90 #include <security/mac/mac_framework.h>
92 void (*softdep_ast_cleanup)(struct thread *);
95 * Define the code needed before returning to user mode, for trap and
99 userret(struct thread *td, struct trapframe *frame)
101 struct proc *p = td->td_proc;
103 CTR3(KTR_SYSC, "userret: thread %p (pid %d, %s)", td, p->p_pid,
105 KASSERT((p->p_flag & P_WEXIT) == 0,
106 ("Exiting process returns to usermode"));
109 * Check that we called signotify() enough. For
110 * multi-threaded processes, where signal distribution might
111 * change due to other threads changing sigmask, the check is
112 * racy and cannot be performed reliably.
113 * If current process is vfork child, indicated by P_PPWAIT, then
114 * issignal() ignores stops, so we block the check to avoid
115 * classifying pending signals.
117 if (p->p_numthreads == 1) {
120 if ((p->p_flag & P_PPWAIT) == 0 &&
121 (td->td_pflags & TDP_SIGFASTBLOCK) == 0) {
122 if (SIGPENDING(td) && (td->td_flags &
123 (TDF_NEEDSIGCHK | TDF_ASTPENDING)) !=
124 (TDF_NEEDSIGCHK | TDF_ASTPENDING)) {
127 "failed to set signal flags for ast p %p td %p fl %x",
128 p, td, td->td_flags);
137 * Charge system time if profiling.
139 if (__predict_false(p->p_flag & P_PROFIL))
140 addupc_task(td, TRAPF_PC(frame), td->td_pticks * psratio);
143 if (PMC_THREAD_HAS_SAMPLES(td))
144 PMC_CALL_HOOK(td, PMC_FN_THR_USERRET, NULL);
148 * @gallatin is adament that this needs to go here, I
149 * am not so sure. Running hpts is a lot like
150 * a lro_flush() that happens while a user process
151 * is running. But he may know best so I will go
152 * with his view of accounting. :-)
157 * Let the scheduler adjust our priority etc.
162 * Check for misbehavior.
164 * In case there is a callchain tracing ongoing because of
165 * hwpmc(4), skip the scheduler pinning check.
166 * hwpmc(4) subsystem, infact, will collect callchain informations
167 * at ast() checkpoint, which is past userret().
169 WITNESS_WARN(WARN_PANIC, NULL, "userret: returning");
170 KASSERT(td->td_critnest == 0,
171 ("userret: Returning in a critical section"));
172 KASSERT(td->td_locks == 0,
173 ("userret: Returning with %d locks held", td->td_locks));
174 KASSERT(td->td_rw_rlocks == 0,
175 ("userret: Returning with %d rwlocks held in read mode",
177 KASSERT(td->td_sx_slocks == 0,
178 ("userret: Returning with %d sx locks held in shared mode",
180 KASSERT(td->td_lk_slocks == 0,
181 ("userret: Returning with %d lockmanager locks held in shared mode",
183 KASSERT((td->td_pflags & TDP_NOFAULTING) == 0,
184 ("userret: Returning with pagefaults disabled"));
185 if (__predict_false(!THREAD_CAN_SLEEP())) {
187 epoch_trace_list(curthread);
189 KASSERT(0, ("userret: Returning with sleep disabled"));
191 KASSERT(td->td_pinned == 0 || (td->td_pflags & TDP_CALLCHAIN) != 0,
192 ("userret: Returning with with pinned thread"));
193 KASSERT(td->td_vp_reserved == NULL,
194 ("userret: Returning with preallocated vnode"));
195 KASSERT((td->td_flags & (TDF_SBDRY | TDF_SEINTR | TDF_SERESTART)) == 0,
196 ("userret: Returning with stop signals deferred"));
197 KASSERT(td->td_vslock_sz == 0,
198 ("userret: Returning with vslock-wired space"));
200 /* Unfortunately td_vnet_lpush needs VNET_DEBUG. */
201 VNET_ASSERT(curvnet == NULL,
202 ("%s: Returning on td %p (pid %d, %s) with vnet %p set in %s",
203 __func__, td, p->p_pid, td->td_name, curvnet,
204 (td->td_vnet_lpush != NULL) ? td->td_vnet_lpush : "N/A"));
209 * Process an asynchronous software trap.
210 * This is relatively easy.
211 * This function will return with preemption disabled.
214 ast(struct trapframe *framep)
221 kmsan_mark(framep, sizeof(*framep), KMSAN_STATE_INITED);
226 CTR3(KTR_SYSC, "ast: thread %p (pid %d, %s)", td, p->p_pid,
228 KASSERT(TRAPF_USERMODE(framep), ("ast in kernel mode"));
229 WITNESS_WARN(WARN_PANIC, NULL, "Returning to user mode");
230 mtx_assert(&Giant, MA_NOTOWNED);
231 THREAD_LOCK_ASSERT(td, MA_NOTOWNED);
232 td->td_frame = framep;
236 * This updates the td_flag's for the checks below in one
237 * "atomic" operation with turning off the astpending flag.
238 * If another AST is triggered while we are handling the
239 * AST's saved in flags, the astpending flag will be set and
240 * ast() will be called again.
243 flags = td->td_flags;
244 td->td_flags &= ~(TDF_ASTPENDING | TDF_NEEDSIGCHK | TDF_NEEDSUSPCHK |
245 TDF_NEEDRESCHED | TDF_ALRMPEND | TDF_PROFPEND | TDF_MACPEND |
250 if (td->td_cowgen != p->p_cowgen)
251 thread_cow_update(td);
252 if (td->td_pflags & TDP_OWEUPC && p->p_flag & P_PROFIL) {
253 addupc_task(td, td->td_profil_addr, td->td_profil_ticks);
254 td->td_profil_ticks = 0;
255 td->td_pflags &= ~TDP_OWEUPC;
258 /* Handle Software PMC callchain capture. */
259 if (PMC_IS_PENDING_CALLCHAIN(td))
260 PMC_CALL_HOOK_UNLOCKED(td, PMC_FN_USER_CALLCHAIN_SOFT, (void *) framep);
262 if (flags & TDF_ALRMPEND) {
264 kern_psignal(p, SIGVTALRM);
267 if (flags & TDF_PROFPEND) {
269 kern_psignal(p, SIGPROF);
273 if (flags & TDF_MACPEND)
274 mac_thread_userret(td);
276 if (flags & TDF_NEEDRESCHED) {
278 if (KTRPOINT(td, KTR_CSW))
279 ktrcsw(1, 1, __func__);
282 sched_prio(td, td->td_user_pri);
283 mi_switch(SW_INVOL | SWT_NEEDRESCHED);
285 if (KTRPOINT(td, KTR_CSW))
286 ktrcsw(0, 1, __func__);
290 td_softdep_cleanup(td);
291 MPASS(td->td_su == NULL);
294 * If this thread tickled GEOM, we need to wait for the giggling to
295 * stop before we return to userland
297 if (__predict_false(td->td_pflags & TDP_GEOM))
301 if (p->p_numthreads == 1 && (flags & TDF_NEEDSIGCHK) == 0) {
305 * Note that TDF_NEEDSIGCHK should be re-read from
306 * td_flags, since signal might have been delivered
307 * after we cleared td_flags above. This is one of
308 * the reason for looping check for AST condition.
309 * See comment in userret() about P_PPWAIT.
311 if ((p->p_flag & P_PPWAIT) == 0 &&
312 (td->td_pflags & TDP_SIGFASTBLOCK) == 0) {
313 if (SIGPENDING(td) && (td->td_flags &
314 (TDF_NEEDSIGCHK | TDF_ASTPENDING)) !=
315 (TDF_NEEDSIGCHK | TDF_ASTPENDING)) {
316 thread_unlock(td); /* fix dumps */
318 "failed2 to set signal flags for ast p %p td %p fl %x %x",
319 p, td, flags, td->td_flags);
328 * Check for signals. Unlocked reads of p_pendingcnt or
329 * p_siglist might cause process-directed signal to be handled
332 if (flags & TDF_NEEDSIGCHK || p->p_pendingcnt > 0 ||
333 !SIGISEMPTY(p->p_siglist)) {
334 sigfastblock_fetch(td);
336 mtx_lock(&p->p_sigacts->ps_mtx);
337 while ((sig = cursig(td)) != 0) {
338 KASSERT(sig >= 0, ("sig %d", sig));
341 mtx_unlock(&p->p_sigacts->ps_mtx);
345 resched_sigs = false;
348 if ((flags & TDF_KQTICKLED) != 0)
349 kqueue_drain_schedtask();
352 * Handle deferred update of the fast sigblock value, after
353 * the postsig() loop was performed.
355 sigfastblock_setpend(td, resched_sigs);
362 * We need to check to see if we have to exit or wait due to a
363 * single threading requirement or some other STOP condition.
365 if (flags & TDF_NEEDSUSPCHK) {
367 thread_suspend_check(0);
371 if (td->td_pflags & TDP_OLDMASK) {
372 td->td_pflags &= ~TDP_OLDMASK;
373 kern_sigprocmask(td, SIG_SETMASK, &td->td_oldsigmask, NULL, 0);
377 if (__predict_false(racct_enable && p->p_throttled != 0))
378 racct_proc_throttled(p);
385 syscallname(struct proc *p, u_int code)
387 static const char unknown[] = "unknown";
388 struct sysentvec *sv;
391 if (sv->sv_syscallnames == NULL || code >= sv->sv_size)
393 return (sv->sv_syscallnames[code]);