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("$FreeBSD$"); 46 47 #include "opt_hwpmc_hooks.h" 48 #include "opt_ktrace.h" 49 #include "opt_sched.h" 50 51 #include <sys/param.h> 52 #include <sys/bus.h> 53 #include <sys/capability.h> 54 #include <sys/kernel.h> 55 #include <sys/lock.h> 56 #include <sys/mutex.h> 57 #include <sys/pmckern.h> 58 #include <sys/proc.h> 59 #include <sys/ktr.h> 60 #include <sys/pioctl.h> 61 #include <sys/ptrace.h> 62 #include <sys/resourcevar.h> 63 #include <sys/sched.h> 64 #include <sys/signalvar.h> 65 #include <sys/syscall.h> 66 #include <sys/syscallsubr.h> 67 #include <sys/sysent.h> 68 #include <sys/systm.h> 69 #include <sys/vmmeter.h> 70 #ifdef KTRACE 71 #include <sys/uio.h> 72 #include <sys/ktrace.h> 73 #endif 74 #include <security/audit/audit.h> 75 76 #include <machine/cpu.h> 77 78 #ifdef VIMAGE 79 #include <net/vnet.h> 80 #endif 81 82 #ifdef XEN 83 #include <vm/vm.h> 84 #include <vm/vm_param.h> 85 #include <vm/pmap.h> 86 #endif 87 88 #ifdef HWPMC_HOOKS 89 #include <sys/pmckern.h> 90 #endif 91 92 #include <security/mac/mac_framework.h> 93 94 /* 95 * Define the code needed before returning to user mode, for trap and 96 * syscall. 97 */ 98 void 99 userret(struct thread *td, struct trapframe *frame) 100 { 101 struct proc *p = td->td_proc; 102 103 CTR3(KTR_SYSC, "userret: thread %p (pid %d, %s)", td, p->p_pid, 104 td->td_name); 105 KASSERT((p->p_flag & P_WEXIT) == 0, 106 ("Exiting process returns to usermode")); 107 #if 0 108 #ifdef DIAGNOSTIC 109 /* Check that we called signotify() enough. */ 110 PROC_LOCK(p); 111 thread_lock(td); 112 if (SIGPENDING(td) && ((td->td_flags & TDF_NEEDSIGCHK) == 0 || 113 (td->td_flags & TDF_ASTPENDING) == 0)) 114 printf("failed to set signal flags properly for ast()\n"); 115 thread_unlock(td); 116 PROC_UNLOCK(p); 117 #endif 118 #endif 119 #ifdef KTRACE 120 KTRUSERRET(td); 121 #endif 122 /* 123 * If this thread tickled GEOM, we need to wait for the giggling to 124 * stop before we return to userland 125 */ 126 if (td->td_pflags & TDP_GEOM) 127 g_waitidle(); 128 129 /* 130 * Charge system time if profiling. 131 */ 132 if (p->p_flag & P_PROFIL) 133 addupc_task(td, TRAPF_PC(frame), td->td_pticks * psratio); 134 /* 135 * Let the scheduler adjust our priority etc. 136 */ 137 sched_userret(td); 138 #ifdef XEN 139 PT_UPDATES_FLUSH(); 140 #endif 141 142 /* 143 * Check for misbehavior. 144 * 145 * In case there is a callchain tracing ongoing because of 146 * hwpmc(4), skip the scheduler pinning check. 147 * hwpmc(4) subsystem, infact, will collect callchain informations 148 * at ast() checkpoint, which is past userret(). 149 */ 150 WITNESS_WARN(WARN_PANIC, NULL, "userret: returning"); 151 KASSERT(td->td_critnest == 0, 152 ("userret: Returning in a critical section")); 153 KASSERT(td->td_locks == 0, 154 ("userret: Returning with %d locks held", td->td_locks)); 155 KASSERT((td->td_pflags & TDP_NOFAULTING) == 0, 156 ("userret: Returning with pagefaults disabled")); 157 KASSERT(td->td_no_sleeping == 0, 158 ("userret: Returning with sleep disabled")); 159 KASSERT(td->td_pinned == 0 || (td->td_pflags & TDP_CALLCHAIN) != 0, 160 ("userret: Returning with with pinned thread")); 161 KASSERT(td->td_vp_reserv == 0, 162 ("userret: Returning while holding vnode reservation")); 163 KASSERT((td->td_flags & TDF_SBDRY) == 0, 164 ("userret: Returning with stop signals deferred")); 165 #ifdef VIMAGE 166 /* Unfortunately td_vnet_lpush needs VNET_DEBUG. */ 167 VNET_ASSERT(curvnet == NULL, 168 ("%s: Returning on td %p (pid %d, %s) with vnet %p set in %s", 169 __func__, td, p->p_pid, td->td_name, curvnet, 170 (td->td_vnet_lpush != NULL) ? td->td_vnet_lpush : "N/A")); 171 #endif 172 #ifdef RACCT 173 PROC_LOCK(p); 174 while (p->p_throttled == 1) 175 msleep(p->p_racct, &p->p_mtx, 0, "racct", 0); 176 PROC_UNLOCK(p); 177 #endif 178 } 179 180 /* 181 * Process an asynchronous software trap. 182 * This is relatively easy. 183 * This function will return with preemption disabled. 184 */ 185 void 186 ast(struct trapframe *framep) 187 { 188 struct thread *td; 189 struct proc *p; 190 int flags; 191 int sig; 192 193 td = curthread; 194 p = td->td_proc; 195 196 CTR3(KTR_SYSC, "ast: thread %p (pid %d, %s)", td, p->p_pid, 197 p->p_comm); 198 KASSERT(TRAPF_USERMODE(framep), ("ast in kernel mode")); 199 WITNESS_WARN(WARN_PANIC, NULL, "Returning to user mode"); 200 mtx_assert(&Giant, MA_NOTOWNED); 201 THREAD_LOCK_ASSERT(td, MA_NOTOWNED); 202 td->td_frame = framep; 203 td->td_pticks = 0; 204 205 /* 206 * This updates the td_flag's for the checks below in one 207 * "atomic" operation with turning off the astpending flag. 208 * If another AST is triggered while we are handling the 209 * AST's saved in flags, the astpending flag will be set and 210 * ast() will be called again. 211 */ 212 thread_lock(td); 213 flags = td->td_flags; 214 td->td_flags &= ~(TDF_ASTPENDING | TDF_NEEDSIGCHK | TDF_NEEDSUSPCHK | 215 TDF_NEEDRESCHED | TDF_ALRMPEND | TDF_PROFPEND | TDF_MACPEND); 216 thread_unlock(td); 217 PCPU_INC(cnt.v_trap); 218 219 if (td->td_ucred != p->p_ucred) 220 cred_update_thread(td); 221 if (td->td_pflags & TDP_OWEUPC && p->p_flag & P_PROFIL) { 222 addupc_task(td, td->td_profil_addr, td->td_profil_ticks); 223 td->td_profil_ticks = 0; 224 td->td_pflags &= ~TDP_OWEUPC; 225 } 226 #ifdef HWPMC_HOOKS 227 /* Handle Software PMC callchain capture. */ 228 if (PMC_IS_PENDING_CALLCHAIN(td)) 229 PMC_CALL_HOOK_UNLOCKED(td, PMC_FN_USER_CALLCHAIN_SOFT, (void *) framep); 230 #endif 231 if (flags & TDF_ALRMPEND) { 232 PROC_LOCK(p); 233 kern_psignal(p, SIGVTALRM); 234 PROC_UNLOCK(p); 235 } 236 if (flags & TDF_PROFPEND) { 237 PROC_LOCK(p); 238 kern_psignal(p, SIGPROF); 239 PROC_UNLOCK(p); 240 } 241 #ifdef MAC 242 if (flags & TDF_MACPEND) 243 mac_thread_userret(td); 244 #endif 245 if (flags & TDF_NEEDRESCHED) { 246 #ifdef KTRACE 247 if (KTRPOINT(td, KTR_CSW)) 248 ktrcsw(1, 1, __func__); 249 #endif 250 thread_lock(td); 251 sched_prio(td, td->td_user_pri); 252 mi_switch(SW_INVOL | SWT_NEEDRESCHED, NULL); 253 thread_unlock(td); 254 #ifdef KTRACE 255 if (KTRPOINT(td, KTR_CSW)) 256 ktrcsw(0, 1, __func__); 257 #endif 258 } 259 260 /* 261 * Check for signals. Unlocked reads of p_pendingcnt or 262 * p_siglist might cause process-directed signal to be handled 263 * later. 264 */ 265 if (flags & TDF_NEEDSIGCHK || p->p_pendingcnt > 0 || 266 !SIGISEMPTY(p->p_siglist)) { 267 PROC_LOCK(p); 268 mtx_lock(&p->p_sigacts->ps_mtx); 269 while ((sig = cursig(td)) != 0) 270 postsig(sig); 271 mtx_unlock(&p->p_sigacts->ps_mtx); 272 PROC_UNLOCK(p); 273 } 274 /* 275 * We need to check to see if we have to exit or wait due to a 276 * single threading requirement or some other STOP condition. 277 */ 278 if (flags & TDF_NEEDSUSPCHK) { 279 PROC_LOCK(p); 280 thread_suspend_check(0); 281 PROC_UNLOCK(p); 282 } 283 284 if (td->td_pflags & TDP_OLDMASK) { 285 td->td_pflags &= ~TDP_OLDMASK; 286 kern_sigprocmask(td, SIG_SETMASK, &td->td_oldsigmask, NULL, 0); 287 } 288 289 userret(td, framep); 290 } 291 292 const char * 293 syscallname(struct proc *p, u_int code) 294 { 295 static const char unknown[] = "unknown"; 296 struct sysentvec *sv; 297 298 sv = p->p_sysent; 299 if (sv->sv_syscallnames == NULL || code >= sv->sv_size) 300 return (unknown); 301 return (sv->sv_syscallnames[code]); 302 } 303