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_kdtrace.h" 50#include "opt_sched.h" 51 52#include <sys/param.h> 53#include <sys/bus.h> 54#include <sys/capability.h> 55#include <sys/kernel.h> 56#include <sys/lock.h> 57#include <sys/mutex.h> 58#include <sys/pmckern.h> 59#include <sys/proc.h> 60#include <sys/ktr.h> 61#include <sys/pioctl.h> 62#include <sys/ptrace.h> 63#include <sys/resourcevar.h> 64#include <sys/sched.h> 65#include <sys/signalvar.h> 66#include <sys/syscall.h> 67#include <sys/syscallsubr.h> 68#include <sys/sysent.h> 69#include <sys/systm.h> 70#include <sys/vmmeter.h> 71#ifdef KTRACE 72#include <sys/uio.h> 73#include <sys/ktrace.h> 74#endif 75#include <security/audit/audit.h> 76 77#include <machine/cpu.h> 78 79#ifdef VIMAGE 80#include <net/vnet.h> 81#endif 82 83#ifdef XEN 84#include <vm/vm.h> 85#include <vm/vm_param.h> 86#include <vm/pmap.h> 87#endif 88 89#ifdef HWPMC_HOOKS 90#include <sys/pmckern.h> 91#endif 92 93#include <security/mac/mac_framework.h> 94 95/* 96 * Define the code needed before returning to user mode, for trap and 97 * syscall. 98 */ 99void 100userret(struct thread *td, struct trapframe *frame) 101{ 102 struct proc *p = td->td_proc; 103 104 CTR3(KTR_SYSC, "userret: thread %p (pid %d, %s)", td, p->p_pid, 105 td->td_name); 106 KASSERT((p->p_flag & P_WEXIT) == 0, 107 ("Exiting process returns to usermode")); 108#if 0 109#ifdef DIAGNOSTIC 110 /* Check that we called signotify() enough. */ 111 PROC_LOCK(p); 112 thread_lock(td); 113 if (SIGPENDING(td) && ((td->td_flags & TDF_NEEDSIGCHK) == 0 || 114 (td->td_flags & TDF_ASTPENDING) == 0)) 115 printf("failed to set signal flags properly for ast()\n"); 116 thread_unlock(td); 117 PROC_UNLOCK(p); 118#endif 119#endif 120#ifdef KTRACE 121 KTRUSERRET(td); 122#endif 123 /* 124 * If this thread tickled GEOM, we need to wait for the giggling to 125 * stop before we return to userland 126 */ 127 if (td->td_pflags & TDP_GEOM) 128 g_waitidle(); 129 130 /* 131 * Charge system time if profiling. 132 */ 133 if (p->p_flag & P_PROFIL) 134 addupc_task(td, TRAPF_PC(frame), td->td_pticks * psratio); 135 /* 136 * Let the scheduler adjust our priority etc. 137 */ 138 sched_userret(td); 139#ifdef XEN 140 PT_UPDATES_FLUSH(); 141#endif 142 143 /* 144 * Check for misbehavior. 145 * 146 * In case there is a callchain tracing ongoing because of 147 * hwpmc(4), skip the scheduler pinning check. 148 * hwpmc(4) subsystem, infact, will collect callchain informations 149 * at ast() checkpoint, which is past userret(). 150 */ 151 WITNESS_WARN(WARN_PANIC, NULL, "userret: returning"); 152 KASSERT(td->td_critnest == 0, 153 ("userret: Returning in a critical section")); 154 KASSERT(td->td_locks == 0, 155 ("userret: Returning with %d locks held", td->td_locks)); 156 KASSERT((td->td_pflags & TDP_NOFAULTING) == 0, 157 ("userret: Returning with pagefaults disabled")); 158 KASSERT((td->td_pflags & TDP_DEVMEMIO) == 0, 159 ("userret: Returning with /dev/mem i/o leaked")); 160 KASSERT(td->td_no_sleeping == 0, 161 ("userret: Returning with sleep disabled")); 162 KASSERT(td->td_pinned == 0 || (td->td_pflags & TDP_CALLCHAIN) != 0, 163 ("userret: Returning with with pinned thread")); 164 KASSERT(td->td_vp_reserv == 0, 165 ("userret: Returning while holding vnode reservation")); 166 KASSERT((td->td_flags & TDF_SBDRY) == 0, 167 ("userret: Returning with stop signals deferred")); 168#ifdef VIMAGE 169 /* Unfortunately td_vnet_lpush needs VNET_DEBUG. */ 170 VNET_ASSERT(curvnet == NULL, 171 ("%s: Returning on td %p (pid %d, %s) with vnet %p set in %s", 172 __func__, td, p->p_pid, td->td_name, curvnet, 173 (td->td_vnet_lpush != NULL) ? td->td_vnet_lpush : "N/A")); 174#endif 175#ifdef RACCT 176 PROC_LOCK(p); 177 while (p->p_throttled == 1) 178 msleep(p->p_racct, &p->p_mtx, 0, "racct", 0); 179 PROC_UNLOCK(p); 180#endif 181} 182 183/* 184 * Process an asynchronous software trap. 185 * This is relatively easy. 186 * This function will return with preemption disabled. 187 */ 188void 189ast(struct trapframe *framep) 190{ 191 struct thread *td; 192 struct proc *p; 193 int flags; 194 int sig; 195 196 td = curthread; 197 p = td->td_proc; 198 199 CTR3(KTR_SYSC, "ast: thread %p (pid %d, %s)", td, p->p_pid, 200 p->p_comm); 201 KASSERT(TRAPF_USERMODE(framep), ("ast in kernel mode")); 202 WITNESS_WARN(WARN_PANIC, NULL, "Returning to user mode"); 203 mtx_assert(&Giant, MA_NOTOWNED); 204 THREAD_LOCK_ASSERT(td, MA_NOTOWNED); 205 td->td_frame = framep; 206 td->td_pticks = 0; 207 208 /* 209 * This updates the td_flag's for the checks below in one 210 * "atomic" operation with turning off the astpending flag. 211 * If another AST is triggered while we are handling the 212 * AST's saved in flags, the astpending flag will be set and 213 * ast() will be called again. 214 */ 215 thread_lock(td); 216 flags = td->td_flags; 217 td->td_flags &= ~(TDF_ASTPENDING | TDF_NEEDSIGCHK | TDF_NEEDSUSPCHK | 218 TDF_NEEDRESCHED | TDF_ALRMPEND | TDF_PROFPEND | TDF_MACPEND); 219 thread_unlock(td); 220 PCPU_INC(cnt.v_trap); 221 222 if (td->td_ucred != p->p_ucred) 223 cred_update_thread(td); 224 if (td->td_pflags & TDP_OWEUPC && p->p_flag & P_PROFIL) { 225 addupc_task(td, td->td_profil_addr, td->td_profil_ticks); 226 td->td_profil_ticks = 0; 227 td->td_pflags &= ~TDP_OWEUPC; 228 } 229#ifdef HWPMC_HOOKS 230 /* Handle Software PMC callchain capture. */ 231 if (PMC_IS_PENDING_CALLCHAIN(td)) 232 PMC_CALL_HOOK_UNLOCKED(td, PMC_FN_USER_CALLCHAIN_SOFT, (void *) framep); 233#endif 234 if (flags & TDF_ALRMPEND) { 235 PROC_LOCK(p); 236 kern_psignal(p, SIGVTALRM); 237 PROC_UNLOCK(p); 238 } 239 if (flags & TDF_PROFPEND) { 240 PROC_LOCK(p); 241 kern_psignal(p, SIGPROF); 242 PROC_UNLOCK(p); 243 } 244#ifdef MAC 245 if (flags & TDF_MACPEND) 246 mac_thread_userret(td); 247#endif 248 if (flags & TDF_NEEDRESCHED) { 249#ifdef KTRACE 250 if (KTRPOINT(td, KTR_CSW)) 251 ktrcsw(1, 1, __func__); 252#endif 253 thread_lock(td); 254 sched_prio(td, td->td_user_pri); 255 mi_switch(SW_INVOL | SWT_NEEDRESCHED, NULL); 256 thread_unlock(td); 257#ifdef KTRACE 258 if (KTRPOINT(td, KTR_CSW)) 259 ktrcsw(0, 1, __func__); 260#endif 261 } 262 263 /* 264 * Check for signals. Unlocked reads of p_pendingcnt or 265 * p_siglist might cause process-directed signal to be handled 266 * later. 267 */ 268 if (flags & TDF_NEEDSIGCHK || p->p_pendingcnt > 0 || 269 !SIGISEMPTY(p->p_siglist)) { 270 PROC_LOCK(p); 271 mtx_lock(&p->p_sigacts->ps_mtx); 272 while ((sig = cursig(td)) != 0) 273 postsig(sig); 274 mtx_unlock(&p->p_sigacts->ps_mtx); 275 PROC_UNLOCK(p); 276 } 277 /* 278 * We need to check to see if we have to exit or wait due to a 279 * single threading requirement or some other STOP condition. 280 */ 281 if (flags & TDF_NEEDSUSPCHK) { 282 PROC_LOCK(p); 283 thread_suspend_check(0); 284 PROC_UNLOCK(p); 285 } 286 287 if (td->td_pflags & TDP_OLDMASK) { 288 td->td_pflags &= ~TDP_OLDMASK; 289 kern_sigprocmask(td, SIG_SETMASK, &td->td_oldsigmask, NULL, 0); 290 } 291 292 userret(td, framep); 293} 294 295const char * 296syscallname(struct proc *p, u_int code) 297{ 298 static const char unknown[] = "unknown"; 299 struct sysentvec *sv; 300 301 sv = p->p_sysent; 302 if (sv->sv_syscallnames == NULL || code >= sv->sv_size) 303 return (unknown); 304 return (sv->sv_syscallnames[code]); 305} 306