FreeBSD/Linux Kernel Cross Reference
sys/kern/subr_trap.c
1 /*-
2 * Copyright (C) 1994, David Greenman
3 * Copyright (c) 1990, 1993
4 * The Regents of the University of California. All rights reserved.
5 * Copyright (c) 2007 The FreeBSD Foundation
6 *
7 * This code is derived from software contributed to Berkeley by
8 * the University of Utah, and William Jolitz.
9 *
10 * Portions of this software were developed by A. Joseph Koshy under
11 * sponsorship from the FreeBSD Foundation and Google, Inc.
12 *
13 * Redistribution and use in source and binary forms, with or without
14 * modification, are permitted provided that the following conditions
15 * are met:
16 * 1. Redistributions of source code must retain the above copyright
17 * notice, this list of conditions and the following disclaimer.
18 * 2. Redistributions in binary form must reproduce the above copyright
19 * notice, this list of conditions and the following disclaimer in the
20 * documentation and/or other materials provided with the distribution.
21 * 3. All advertising materials mentioning features or use of this software
22 * must display the following acknowledgement:
23 * This product includes software developed by the University of
24 * California, Berkeley and its contributors.
25 * 4. Neither the name of the University nor the names of its contributors
26 * may be used to endorse or promote products derived from this software
27 * without specific prior written permission.
28 *
29 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
30 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
31 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
32 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
33 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
34 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
35 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
36 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
37 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
38 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
39 * SUCH DAMAGE.
40 *
41 * from: @(#)trap.c 7.4 (Berkeley) 5/13/91
42 */
43
44 #include <sys/cdefs.h>
45 __FBSDID("$FreeBSD: releng/8.2/sys/kern/subr_trap.c 215513 2010-11-19 09:49:14Z kib $");
46
47 #include "opt_ktrace.h"
48 #include "opt_kdtrace.h"
49 #include "opt_sched.h"
50
51 #include <sys/param.h>
52 #include <sys/bus.h>
53 #include <sys/kernel.h>
54 #include <sys/lock.h>
55 #include <sys/mutex.h>
56 #include <sys/pmckern.h>
57 #include <sys/proc.h>
58 #include <sys/ktr.h>
59 #include <sys/pioctl.h>
60 #include <sys/ptrace.h>
61 #include <sys/resourcevar.h>
62 #include <sys/sched.h>
63 #include <sys/signalvar.h>
64 #include <sys/syscall.h>
65 #include <sys/sysent.h>
66 #include <sys/systm.h>
67 #include <sys/vmmeter.h>
68 #ifdef KTRACE
69 #include <sys/uio.h>
70 #include <sys/ktrace.h>
71 #endif
72 #include <security/audit/audit.h>
73
74 #include <machine/cpu.h>
75
76 #ifdef XEN
77 #include <vm/vm.h>
78 #include <vm/vm_param.h>
79 #include <vm/pmap.h>
80 #endif
81
82 #include <security/mac/mac_framework.h>
83
84 /*
85 * Define the code needed before returning to user mode, for trap and
86 * syscall.
87 */
88 void
89 userret(struct thread *td, struct trapframe *frame)
90 {
91 struct proc *p = td->td_proc;
92
93 CTR3(KTR_SYSC, "userret: thread %p (pid %d, %s)", td, p->p_pid,
94 td->td_name);
95 #if 0
96 #ifdef DIAGNOSTIC
97 /* Check that we called signotify() enough. */
98 PROC_LOCK(p);
99 thread_lock(td);
100 if (SIGPENDING(td) && ((td->td_flags & TDF_NEEDSIGCHK) == 0 ||
101 (td->td_flags & TDF_ASTPENDING) == 0))
102 printf("failed to set signal flags properly for ast()\n");
103 thread_unlock(td);
104 PROC_UNLOCK(p);
105 #endif
106 #endif
107 #ifdef KTRACE
108 KTRUSERRET(td);
109 #endif
110 /*
111 * If this thread tickled GEOM, we need to wait for the giggling to
112 * stop before we return to userland
113 */
114 if (td->td_pflags & TDP_GEOM)
115 g_waitidle();
116
117 /*
118 * Charge system time if profiling.
119 */
120 if (p->p_flag & P_PROFIL)
121 addupc_task(td, TRAPF_PC(frame), td->td_pticks * psratio);
122 /*
123 * Let the scheduler adjust our priority etc.
124 */
125 sched_userret(td);
126 KASSERT(td->td_locks == 0,
127 ("userret: Returning with %d locks held.", td->td_locks));
128 #ifdef XEN
129 PT_UPDATES_FLUSH();
130 #endif
131 }
132
133 /*
134 * Process an asynchronous software trap.
135 * This is relatively easy.
136 * This function will return with preemption disabled.
137 */
138 void
139 ast(struct trapframe *framep)
140 {
141 struct thread *td;
142 struct proc *p;
143 int flags;
144 int sig;
145
146 td = curthread;
147 p = td->td_proc;
148
149 CTR3(KTR_SYSC, "ast: thread %p (pid %d, %s)", td, p->p_pid,
150 p->p_comm);
151 KASSERT(TRAPF_USERMODE(framep), ("ast in kernel mode"));
152 WITNESS_WARN(WARN_PANIC, NULL, "Returning to user mode");
153 mtx_assert(&Giant, MA_NOTOWNED);
154 THREAD_LOCK_ASSERT(td, MA_NOTOWNED);
155 td->td_frame = framep;
156 td->td_pticks = 0;
157
158 /*
159 * This updates the td_flag's for the checks below in one
160 * "atomic" operation with turning off the astpending flag.
161 * If another AST is triggered while we are handling the
162 * AST's saved in flags, the astpending flag will be set and
163 * ast() will be called again.
164 */
165 thread_lock(td);
166 flags = td->td_flags;
167 td->td_flags &= ~(TDF_ASTPENDING | TDF_NEEDSIGCHK | TDF_NEEDSUSPCHK |
168 TDF_NEEDRESCHED | TDF_ALRMPEND | TDF_PROFPEND | TDF_MACPEND);
169 thread_unlock(td);
170 PCPU_INC(cnt.v_trap);
171
172 if (td->td_ucred != p->p_ucred)
173 cred_update_thread(td);
174 if (td->td_pflags & TDP_OWEUPC && p->p_flag & P_PROFIL) {
175 addupc_task(td, td->td_profil_addr, td->td_profil_ticks);
176 td->td_profil_ticks = 0;
177 td->td_pflags &= ~TDP_OWEUPC;
178 }
179 if (flags & TDF_ALRMPEND) {
180 PROC_LOCK(p);
181 psignal(p, SIGVTALRM);
182 PROC_UNLOCK(p);
183 }
184 if (flags & TDF_PROFPEND) {
185 PROC_LOCK(p);
186 psignal(p, SIGPROF);
187 PROC_UNLOCK(p);
188 }
189 #ifdef MAC
190 if (flags & TDF_MACPEND)
191 mac_thread_userret(td);
192 #endif
193 if (flags & TDF_NEEDRESCHED) {
194 #ifdef KTRACE
195 if (KTRPOINT(td, KTR_CSW))
196 ktrcsw(1, 1);
197 #endif
198 thread_lock(td);
199 sched_prio(td, td->td_user_pri);
200 mi_switch(SW_INVOL | SWT_NEEDRESCHED, NULL);
201 thread_unlock(td);
202 #ifdef KTRACE
203 if (KTRPOINT(td, KTR_CSW))
204 ktrcsw(0, 1);
205 #endif
206 }
207
208 /*
209 * Check for signals. Unlocked reads of p_pendingcnt or
210 * p_siglist might cause process-directed signal to be handled
211 * later.
212 */
213 if (flags & TDF_NEEDSIGCHK || p->p_pendingcnt > 0 ||
214 !SIGISEMPTY(p->p_siglist)) {
215 PROC_LOCK(p);
216 mtx_lock(&p->p_sigacts->ps_mtx);
217 while ((sig = cursig(td, SIG_STOP_ALLOWED)) != 0)
218 postsig(sig);
219 mtx_unlock(&p->p_sigacts->ps_mtx);
220 PROC_UNLOCK(p);
221 }
222 /*
223 * We need to check to see if we have to exit or wait due to a
224 * single threading requirement or some other STOP condition.
225 */
226 if (flags & TDF_NEEDSUSPCHK) {
227 PROC_LOCK(p);
228 thread_suspend_check(0);
229 PROC_UNLOCK(p);
230 }
231
232 if (td->td_pflags & TDP_OLDMASK) {
233 td->td_pflags &= ~TDP_OLDMASK;
234 kern_sigprocmask(td, SIG_SETMASK, &td->td_oldsigmask, NULL, 0);
235 }
236
237 userret(td, framep);
238 mtx_assert(&Giant, MA_NOTOWNED);
239 }
240
241 #ifdef HAVE_SYSCALL_ARGS_DEF
242 const char *
243 syscallname(struct proc *p, u_int code)
244 {
245 static const char unknown[] = "unknown";
246 struct sysentvec *sv;
247
248 sv = p->p_sysent;
249 if (sv->sv_syscallnames == NULL || code >= sv->sv_size)
250 return (unknown);
251 return (sv->sv_syscallnames[code]);
252 }
253
254 int
255 syscallenter(struct thread *td, struct syscall_args *sa)
256 {
257 struct proc *p;
258 int error, traced;
259
260 PCPU_INC(cnt.v_syscall);
261 p = td->td_proc;
262 td->td_syscalls++;
263
264 td->td_pticks = 0;
265 if (td->td_ucred != p->p_ucred)
266 cred_update_thread(td);
267 if (p->p_flag & P_TRACED) {
268 traced = 1;
269 PROC_LOCK(p);
270 td->td_dbgflags &= ~TDB_USERWR;
271 td->td_dbgflags |= TDB_SCE;
272 PROC_UNLOCK(p);
273 } else
274 traced = 0;
275 error = (p->p_sysent->sv_fetch_syscall_args)(td, sa);
276 #ifdef KTRACE
277 if (KTRPOINT(td, KTR_SYSCALL))
278 ktrsyscall(sa->code, sa->narg, sa->args);
279 #endif
280
281 CTR6(KTR_SYSC,
282 "syscall: td=%p pid %d %s (%#lx, %#lx, %#lx)",
283 td, td->td_proc->p_pid, syscallname(p, sa->code),
284 sa->args[0], sa->args[1], sa->args[2]);
285
286 if (error == 0) {
287 STOPEVENT(p, S_SCE, sa->narg);
288 PTRACESTOP_SC(p, td, S_PT_SCE);
289 if (td->td_dbgflags & TDB_USERWR) {
290 /*
291 * Reread syscall number and arguments if
292 * debugger modified registers or memory.
293 */
294 error = (p->p_sysent->sv_fetch_syscall_args)(td, sa);
295 #ifdef KTRACE
296 if (KTRPOINT(td, KTR_SYSCALL))
297 ktrsyscall(sa->code, sa->narg, sa->args);
298 #endif
299 if (error != 0)
300 goto retval;
301 }
302
303 #ifdef KDTRACE_HOOKS
304 /*
305 * If the systrace module has registered it's probe
306 * callback and if there is a probe active for the
307 * syscall 'entry', process the probe.
308 */
309 if (systrace_probe_func != NULL && sa->callp->sy_entry != 0)
310 (*systrace_probe_func)(sa->callp->sy_entry, sa->code,
311 sa->callp, sa->args);
312 #endif
313
314 AUDIT_SYSCALL_ENTER(sa->code, td);
315 error = (sa->callp->sy_call)(td, sa->args);
316 AUDIT_SYSCALL_EXIT(error, td);
317
318 /* Save the latest error return value. */
319 td->td_errno = error;
320
321 #ifdef KDTRACE_HOOKS
322 /*
323 * If the systrace module has registered it's probe
324 * callback and if there is a probe active for the
325 * syscall 'return', process the probe.
326 */
327 if (systrace_probe_func != NULL && sa->callp->sy_return != 0)
328 (*systrace_probe_func)(sa->callp->sy_return, sa->code,
329 sa->callp, sa->args);
330 #endif
331 CTR4(KTR_SYSC, "syscall: p=%p error=%d return %#lx %#lx",
332 p, error, td->td_retval[0], td->td_retval[1]);
333 }
334 retval:
335 if (traced) {
336 PROC_LOCK(p);
337 td->td_dbgflags &= ~TDB_SCE;
338 PROC_UNLOCK(p);
339 }
340 (p->p_sysent->sv_set_syscall_retval)(td, error);
341 return (error);
342 }
343
344 void
345 syscallret(struct thread *td, int error, struct syscall_args *sa __unused)
346 {
347 struct proc *p;
348 int traced;
349
350 p = td->td_proc;
351
352 /*
353 * Check for misbehavior.
354 */
355 WITNESS_WARN(WARN_PANIC, NULL, "System call %s returning",
356 syscallname(p, sa->code));
357 KASSERT(td->td_critnest == 0,
358 ("System call %s returning in a critical section",
359 syscallname(p, sa->code)));
360 KASSERT(td->td_locks == 0,
361 ("System call %s returning with %d locks held",
362 syscallname(p, sa->code), td->td_locks));
363
364 /*
365 * Handle reschedule and other end-of-syscall issues
366 */
367 userret(td, td->td_frame);
368
369 CTR4(KTR_SYSC, "syscall %s exit thread %p pid %d proc %s",
370 syscallname(p, sa->code), td, td->td_proc->p_pid, td->td_name);
371
372 #ifdef KTRACE
373 if (KTRPOINT(td, KTR_SYSRET))
374 ktrsysret(sa->code, error, td->td_retval[0]);
375 #endif
376
377 if (p->p_flag & P_TRACED) {
378 traced = 1;
379 PROC_LOCK(p);
380 td->td_dbgflags |= TDB_SCX;
381 PROC_UNLOCK(p);
382 } else
383 traced = 0;
384 /*
385 * This works because errno is findable through the
386 * register set. If we ever support an emulation where this
387 * is not the case, this code will need to be revisited.
388 */
389 STOPEVENT(p, S_SCX, sa->code);
390 PTRACESTOP_SC(p, td, S_PT_SCX);
391 if (traced || (td->td_dbgflags & TDB_EXEC) != 0) {
392 PROC_LOCK(p);
393 td->td_dbgflags &= ~(TDB_SCX | TDB_EXEC);
394 PROC_UNLOCK(p);
395 }
396 }
397 #endif /* HAVE_SYSCALL_ARGS_DEF */
Cache object: 4e434fc68269652d51dd22495550c237
|