1 /*- 2 * SPDX-License-Identifier: BSD-4-Clause 3 * 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 8 * 9 * This code is derived from software contributed to Berkeley by 10 * the University of Utah, and William Jolitz. 11 * 12 * Portions of this software were developed by A. Joseph Koshy under 13 * sponsorship from the FreeBSD Foundation and Google, 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 * from: @(#)trap.c 7.4 (Berkeley) 5/13/91 44 */ 45 46 #include <sys/cdefs.h> 47 __FBSDID("$FreeBSD$"); 48 49 #include "opt_hwpmc_hooks.h" 50 #include "opt_ktrace.h" 51 #include "opt_sched.h" 52 53 #include <sys/param.h> 54 #include <sys/bus.h> 55 #include <sys/capsicum.h> 56 #include <sys/kernel.h> 57 #include <sys/lock.h> 58 #include <sys/mutex.h> 59 #include <sys/pmckern.h> 60 #include <sys/proc.h> 61 #include <sys/ktr.h> 62 #include <sys/ptrace.h> 63 #include <sys/racct.h> 64 #include <sys/resourcevar.h> 65 #include <sys/sched.h> 66 #include <sys/signalvar.h> 67 #include <sys/syscall.h> 68 #include <sys/syscallsubr.h> 69 #include <sys/sysent.h> 70 #include <sys/systm.h> 71 #include <sys/vmmeter.h> 72 #ifdef KTRACE 73 #include <sys/uio.h> 74 #include <sys/ktrace.h> 75 #endif 76 #include <security/audit/audit.h> 77 78 #include <machine/cpu.h> 79 80 #ifdef VIMAGE 81 #include <net/vnet.h> 82 #endif 83 84 #ifdef HWPMC_HOOKS 85 #include <sys/pmckern.h> 86 #endif 87 88 #include <security/mac/mac_framework.h> 89 90 void (*softdep_ast_cleanup)(struct thread *); 91 92 /* 93 * Define the code needed before returning to user mode, for trap and 94 * syscall. 95 */ 96 void 97 userret(struct thread *td, struct trapframe *frame) 98 { 99 struct proc *p = td->td_proc; 100 101 CTR3(KTR_SYSC, "userret: thread %p (pid %d, %s)", td, p->p_pid, 102 td->td_name); 103 KASSERT((p->p_flag & P_WEXIT) == 0, 104 ("Exiting process returns to usermode")); 105 #ifdef DIAGNOSTIC 106 /* 107 * Check that we called signotify() enough. For 108 * multi-threaded processes, where signal distribution might 109 * change due to other threads changing sigmask, the check is 110 * racy and cannot be performed reliably. 111 * If current process is vfork child, indicated by P_PPWAIT, then 112 * issignal() ignores stops, so we block the check to avoid 113 * classifying pending signals. 114 */ 115 if (p->p_numthreads == 1) { 116 PROC_LOCK(p); 117 thread_lock(td); 118 if ((p->p_flag & P_PPWAIT) == 0 && 119 (td->td_pflags & TDP_SIGFASTBLOCK) == 0) { 120 if (SIGPENDING(td) && (td->td_flags & 121 (TDF_NEEDSIGCHK | TDF_ASTPENDING)) != 122 (TDF_NEEDSIGCHK | TDF_ASTPENDING)) { 123 thread_unlock(td); 124 panic( 125 "failed to set signal flags for ast p %p td %p fl %x", 126 p, td, td->td_flags); 127 } 128 } 129 thread_unlock(td); 130 PROC_UNLOCK(p); 131 } 132 #endif 133 #ifdef KTRACE 134 KTRUSERRET(td); 135 #endif 136 137 /* 138 * Charge system time if profiling. 139 */ 140 if (__predict_false(p->p_flag & P_PROFIL)) 141 addupc_task(td, TRAPF_PC(frame), td->td_pticks * psratio); 142 143 #ifdef HWPMC_HOOKS 144 if (PMC_THREAD_HAS_SAMPLES(td)) 145 PMC_CALL_HOOK(td, PMC_FN_THR_USERRET, NULL); 146 #endif 147 /* 148 * Let the scheduler adjust our priority etc. 149 */ 150 sched_userret(td); 151 152 /* 153 * Check for misbehavior. 154 * 155 * In case there is a callchain tracing ongoing because of 156 * hwpmc(4), skip the scheduler pinning check. 157 * hwpmc(4) subsystem, infact, will collect callchain informations 158 * at ast() checkpoint, which is past userret(). 159 */ 160 WITNESS_WARN(WARN_PANIC, NULL, "userret: returning"); 161 KASSERT(td->td_critnest == 0, 162 ("userret: Returning in a critical section")); 163 KASSERT(td->td_locks == 0, 164 ("userret: Returning with %d locks held", td->td_locks)); 165 KASSERT(td->td_rw_rlocks == 0, 166 ("userret: Returning with %d rwlocks held in read mode", 167 td->td_rw_rlocks)); 168 KASSERT(td->td_sx_slocks == 0, 169 ("userret: Returning with %d sx locks held in shared mode", 170 td->td_sx_slocks)); 171 KASSERT(td->td_lk_slocks == 0, 172 ("userret: Returning with %d lockmanager locks held in shared mode", 173 td->td_lk_slocks)); 174 KASSERT((td->td_pflags & TDP_NOFAULTING) == 0, 175 ("userret: Returning with pagefaults disabled")); 176 if (__predict_false(!THREAD_CAN_SLEEP())) { 177 #ifdef EPOCH_TRACE 178 epoch_trace_list(curthread); 179 #endif 180 KASSERT(0, ("userret: Returning with sleep disabled")); 181 } 182 KASSERT(td->td_pinned == 0 || (td->td_pflags & TDP_CALLCHAIN) != 0, 183 ("userret: Returning with with pinned thread")); 184 KASSERT(td->td_vp_reserved == NULL, 185 ("userret: Returning with preallocated vnode")); 186 KASSERT((td->td_flags & (TDF_SBDRY | TDF_SEINTR | TDF_SERESTART)) == 0, 187 ("userret: Returning with stop signals deferred")); 188 KASSERT(td->td_vslock_sz == 0, 189 ("userret: Returning with vslock-wired space")); 190 #ifdef VIMAGE 191 /* Unfortunately td_vnet_lpush needs VNET_DEBUG. */ 192 VNET_ASSERT(curvnet == NULL, 193 ("%s: Returning on td %p (pid %d, %s) with vnet %p set in %s", 194 __func__, td, p->p_pid, td->td_name, curvnet, 195 (td->td_vnet_lpush != NULL) ? td->td_vnet_lpush : "N/A")); 196 #endif 197 } 198 199 /* 200 * Process an asynchronous software trap. 201 * This is relatively easy. 202 * This function will return with preemption disabled. 203 */ 204 void 205 ast(struct trapframe *framep) 206 { 207 struct thread *td; 208 struct proc *p; 209 int flags, sig; 210 211 td = curthread; 212 p = td->td_proc; 213 214 CTR3(KTR_SYSC, "ast: thread %p (pid %d, %s)", td, p->p_pid, 215 p->p_comm); 216 KASSERT(TRAPF_USERMODE(framep), ("ast in kernel mode")); 217 WITNESS_WARN(WARN_PANIC, NULL, "Returning to user mode"); 218 mtx_assert(&Giant, MA_NOTOWNED); 219 THREAD_LOCK_ASSERT(td, MA_NOTOWNED); 220 td->td_frame = framep; 221 td->td_pticks = 0; 222 223 /* 224 * This updates the td_flag's for the checks below in one 225 * "atomic" operation with turning off the astpending flag. 226 * If another AST is triggered while we are handling the 227 * AST's saved in flags, the astpending flag will be set and 228 * ast() will be called again. 229 */ 230 thread_lock(td); 231 flags = td->td_flags; 232 td->td_flags &= ~(TDF_ASTPENDING | TDF_NEEDSIGCHK | TDF_NEEDSUSPCHK | 233 TDF_NEEDRESCHED | TDF_ALRMPEND | TDF_PROFPEND | TDF_MACPEND); 234 thread_unlock(td); 235 VM_CNT_INC(v_trap); 236 237 if (td->td_cowgen != p->p_cowgen) 238 thread_cow_update(td); 239 if (td->td_pflags & TDP_OWEUPC && p->p_flag & P_PROFIL) { 240 addupc_task(td, td->td_profil_addr, td->td_profil_ticks); 241 td->td_profil_ticks = 0; 242 td->td_pflags &= ~TDP_OWEUPC; 243 } 244 #ifdef HWPMC_HOOKS 245 /* Handle Software PMC callchain capture. */ 246 if (PMC_IS_PENDING_CALLCHAIN(td)) 247 PMC_CALL_HOOK_UNLOCKED(td, PMC_FN_USER_CALLCHAIN_SOFT, (void *) framep); 248 #endif 249 if (flags & TDF_ALRMPEND) { 250 PROC_LOCK(p); 251 kern_psignal(p, SIGVTALRM); 252 PROC_UNLOCK(p); 253 } 254 if (flags & TDF_PROFPEND) { 255 PROC_LOCK(p); 256 kern_psignal(p, SIGPROF); 257 PROC_UNLOCK(p); 258 } 259 #ifdef MAC 260 if (flags & TDF_MACPEND) 261 mac_thread_userret(td); 262 #endif 263 if (flags & TDF_NEEDRESCHED) { 264 #ifdef KTRACE 265 if (KTRPOINT(td, KTR_CSW)) 266 ktrcsw(1, 1, __func__); 267 #endif 268 thread_lock(td); 269 sched_prio(td, td->td_user_pri); 270 mi_switch(SW_INVOL | SWT_NEEDRESCHED); 271 #ifdef KTRACE 272 if (KTRPOINT(td, KTR_CSW)) 273 ktrcsw(0, 1, __func__); 274 #endif 275 } 276 277 td_softdep_cleanup(td); 278 MPASS(td->td_su == NULL); 279 280 /* 281 * If this thread tickled GEOM, we need to wait for the giggling to 282 * stop before we return to userland 283 */ 284 if (__predict_false(td->td_pflags & TDP_GEOM)) 285 g_waitidle(); 286 287 #ifdef DIAGNOSTIC 288 if (p->p_numthreads == 1 && (flags & TDF_NEEDSIGCHK) == 0) { 289 PROC_LOCK(p); 290 thread_lock(td); 291 /* 292 * Note that TDF_NEEDSIGCHK should be re-read from 293 * td_flags, since signal might have been delivered 294 * after we cleared td_flags above. This is one of 295 * the reason for looping check for AST condition. 296 * See comment in userret() about P_PPWAIT. 297 */ 298 if ((p->p_flag & P_PPWAIT) == 0 && 299 (td->td_pflags & TDP_SIGFASTBLOCK) == 0) { 300 if (SIGPENDING(td) && (td->td_flags & 301 (TDF_NEEDSIGCHK | TDF_ASTPENDING)) != 302 (TDF_NEEDSIGCHK | TDF_ASTPENDING)) { 303 thread_unlock(td); /* fix dumps */ 304 panic( 305 "failed2 to set signal flags for ast p %p td %p fl %x %x", 306 p, td, flags, td->td_flags); 307 } 308 } 309 thread_unlock(td); 310 PROC_UNLOCK(p); 311 } 312 #endif 313 314 /* 315 * Check for signals. Unlocked reads of p_pendingcnt or 316 * p_siglist might cause process-directed signal to be handled 317 * later. 318 */ 319 if (flags & TDF_NEEDSIGCHK || p->p_pendingcnt > 0 || 320 !SIGISEMPTY(p->p_siglist)) { 321 sigfastblock_fetch(td); 322 if ((td->td_pflags & TDP_SIGFASTBLOCK) != 0 && 323 td->td_sigblock_val != 0) { 324 sigfastblock_setpend(td, true); 325 } else { 326 PROC_LOCK(p); 327 mtx_lock(&p->p_sigacts->ps_mtx); 328 while ((sig = cursig(td)) != 0) { 329 KASSERT(sig >= 0, ("sig %d", sig)); 330 postsig(sig); 331 } 332 mtx_unlock(&p->p_sigacts->ps_mtx); 333 PROC_UNLOCK(p); 334 } 335 } 336 337 /* 338 * Handle deferred update of the fast sigblock value, after 339 * the postsig() loop was performed. 340 */ 341 if (td->td_pflags & TDP_SIGFASTPENDING) 342 sigfastblock_setpend(td, false); 343 344 /* 345 * We need to check to see if we have to exit or wait due to a 346 * single threading requirement or some other STOP condition. 347 */ 348 if (flags & TDF_NEEDSUSPCHK) { 349 PROC_LOCK(p); 350 thread_suspend_check(0); 351 PROC_UNLOCK(p); 352 } 353 354 if (td->td_pflags & TDP_OLDMASK) { 355 td->td_pflags &= ~TDP_OLDMASK; 356 kern_sigprocmask(td, SIG_SETMASK, &td->td_oldsigmask, NULL, 0); 357 } 358 359 #ifdef RACCT 360 if (__predict_false(racct_enable && p->p_throttled != 0)) 361 racct_proc_throttled(p); 362 #endif 363 364 userret(td, framep); 365 } 366 367 const char * 368 syscallname(struct proc *p, u_int code) 369 { 370 static const char unknown[] = "unknown"; 371 struct sysentvec *sv; 372 373 sv = p->p_sysent; 374 if (sv->sv_syscallnames == NULL || code >= sv->sv_size) 375 return (unknown); 376 return (sv->sv_syscallnames[code]); 377 } 378