1 /*- 2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD 3 * 4 * Copyright (c) 2004 Tim J. Robbins 5 * Copyright (c) 2002 Doug Rabson 6 * Copyright (c) 2000 Marcel Moolenaar 7 * All rights reserved. 8 * 9 * Redistribution and use in source and binary forms, with or without 10 * modification, are permitted provided that the following conditions 11 * are met: 12 * 1. Redistributions of source code must retain the above copyright 13 * notice, this list of conditions and the following disclaimer 14 * in this position and unchanged. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 20 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 21 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 22 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 23 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 24 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 25 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 26 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 27 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 28 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 29 */ 30 31 #include <sys/cdefs.h> 32 __FBSDID("$FreeBSD$"); 33 34 #include "opt_compat.h" 35 36 #include <sys/param.h> 37 #include <sys/systm.h> 38 #include <sys/imgact.h> 39 #include <sys/ktr.h> 40 #include <sys/lock.h> 41 #include <sys/mutex.h> 42 #include <sys/proc.h> 43 #include <sys/ptrace.h> 44 #include <sys/racct.h> 45 #include <sys/sched.h> 46 #include <sys/syscallsubr.h> 47 #include <sys/sx.h> 48 #include <sys/umtx.h> 49 #include <sys/unistd.h> 50 #include <sys/wait.h> 51 52 #include <vm/vm.h> 53 #include <vm/pmap.h> 54 #include <vm/vm_map.h> 55 56 #ifdef COMPAT_LINUX32 57 #include <machine/../linux32/linux.h> 58 #include <machine/../linux32/linux32_proto.h> 59 #else 60 #include <machine/../linux/linux.h> 61 #include <machine/../linux/linux_proto.h> 62 #endif 63 #include <compat/linux/linux_emul.h> 64 #include <compat/linux/linux_futex.h> 65 #include <compat/linux/linux_misc.h> 66 #include <compat/linux/linux_util.h> 67 68 #ifdef LINUX_LEGACY_SYSCALLS 69 int 70 linux_fork(struct thread *td, struct linux_fork_args *args) 71 { 72 struct fork_req fr; 73 int error; 74 struct proc *p2; 75 struct thread *td2; 76 77 bzero(&fr, sizeof(fr)); 78 fr.fr_flags = RFFDG | RFPROC | RFSTOPPED; 79 fr.fr_procp = &p2; 80 if ((error = fork1(td, &fr)) != 0) 81 return (error); 82 83 td2 = FIRST_THREAD_IN_PROC(p2); 84 85 linux_proc_init(td, td2, 0); 86 87 td->td_retval[0] = p2->p_pid; 88 89 /* 90 * Make this runnable after we are finished with it. 91 */ 92 thread_lock(td2); 93 TD_SET_CAN_RUN(td2); 94 sched_add(td2, SRQ_BORING); 95 96 return (0); 97 } 98 99 int 100 linux_vfork(struct thread *td, struct linux_vfork_args *args) 101 { 102 struct fork_req fr; 103 int error; 104 struct proc *p2; 105 struct thread *td2; 106 107 bzero(&fr, sizeof(fr)); 108 fr.fr_flags = RFFDG | RFPROC | RFMEM | RFPPWAIT | RFSTOPPED; 109 fr.fr_procp = &p2; 110 if ((error = fork1(td, &fr)) != 0) 111 return (error); 112 113 td2 = FIRST_THREAD_IN_PROC(p2); 114 115 linux_proc_init(td, td2, 0); 116 117 td->td_retval[0] = p2->p_pid; 118 119 /* 120 * Make this runnable after we are finished with it. 121 */ 122 thread_lock(td2); 123 TD_SET_CAN_RUN(td2); 124 sched_add(td2, SRQ_BORING); 125 126 return (0); 127 } 128 #endif 129 130 static int 131 linux_clone_proc(struct thread *td, struct linux_clone_args *args) 132 { 133 struct fork_req fr; 134 int error, ff = RFPROC | RFSTOPPED; 135 struct proc *p2; 136 struct thread *td2; 137 int exit_signal; 138 struct linux_emuldata *em; 139 140 exit_signal = args->flags & 0x000000ff; 141 if (LINUX_SIG_VALID(exit_signal)) { 142 exit_signal = linux_to_bsd_signal(exit_signal); 143 } else if (exit_signal != 0) 144 return (EINVAL); 145 146 if (args->flags & LINUX_CLONE_VM) 147 ff |= RFMEM; 148 if (args->flags & LINUX_CLONE_SIGHAND) 149 ff |= RFSIGSHARE; 150 /* 151 * XXX: In Linux, sharing of fs info (chroot/cwd/umask) 152 * and open files is independent. In FreeBSD, its in one 153 * structure but in reality it does not cause any problems 154 * because both of these flags are usually set together. 155 */ 156 if (!(args->flags & (LINUX_CLONE_FILES | LINUX_CLONE_FS))) 157 ff |= RFFDG; 158 159 if (args->flags & LINUX_CLONE_PARENT_SETTID) 160 if (args->parent_tidptr == NULL) 161 return (EINVAL); 162 163 if (args->flags & LINUX_CLONE_VFORK) 164 ff |= RFPPWAIT; 165 166 bzero(&fr, sizeof(fr)); 167 fr.fr_flags = ff; 168 fr.fr_procp = &p2; 169 error = fork1(td, &fr); 170 if (error) 171 return (error); 172 173 td2 = FIRST_THREAD_IN_PROC(p2); 174 175 /* create the emuldata */ 176 linux_proc_init(td, td2, args->flags); 177 178 em = em_find(td2); 179 KASSERT(em != NULL, ("clone_proc: emuldata not found.\n")); 180 181 if (args->flags & LINUX_CLONE_CHILD_SETTID) 182 em->child_set_tid = args->child_tidptr; 183 else 184 em->child_set_tid = NULL; 185 186 if (args->flags & LINUX_CLONE_CHILD_CLEARTID) 187 em->child_clear_tid = args->child_tidptr; 188 else 189 em->child_clear_tid = NULL; 190 191 if (args->flags & LINUX_CLONE_PARENT_SETTID) { 192 error = copyout(&p2->p_pid, args->parent_tidptr, 193 sizeof(p2->p_pid)); 194 if (error) 195 linux_msg(td, "copyout p_pid failed!"); 196 } 197 198 PROC_LOCK(p2); 199 p2->p_sigparent = exit_signal; 200 PROC_UNLOCK(p2); 201 /* 202 * In a case of stack = NULL, we are supposed to COW calling process 203 * stack. This is what normal fork() does, so we just keep tf_rsp arg 204 * intact. 205 */ 206 linux_set_upcall_kse(td2, PTROUT(args->stack)); 207 208 if (args->flags & LINUX_CLONE_SETTLS) 209 linux_set_cloned_tls(td2, args->tls); 210 211 /* 212 * If CLONE_PARENT is set, then the parent of the new process will be 213 * the same as that of the calling process. 214 */ 215 if (args->flags & LINUX_CLONE_PARENT) { 216 sx_xlock(&proctree_lock); 217 PROC_LOCK(p2); 218 proc_reparent(p2, td->td_proc->p_pptr, true); 219 PROC_UNLOCK(p2); 220 sx_xunlock(&proctree_lock); 221 } 222 223 /* 224 * Make this runnable after we are finished with it. 225 */ 226 thread_lock(td2); 227 TD_SET_CAN_RUN(td2); 228 sched_add(td2, SRQ_BORING); 229 230 td->td_retval[0] = p2->p_pid; 231 232 return (0); 233 } 234 235 static int 236 linux_clone_thread(struct thread *td, struct linux_clone_args *args) 237 { 238 struct linux_emuldata *em; 239 struct thread *newtd; 240 struct proc *p; 241 int error; 242 243 LINUX_CTR4(clone_thread, "thread(%d) flags %x ptid %p ctid %p", 244 td->td_tid, (unsigned)args->flags, 245 args->parent_tidptr, args->child_tidptr); 246 247 if (args->flags & LINUX_CLONE_PARENT_SETTID) 248 if (args->parent_tidptr == NULL) 249 return (EINVAL); 250 251 /* Threads should be created with own stack */ 252 if (args->stack == NULL) 253 return (EINVAL); 254 255 p = td->td_proc; 256 257 #ifdef RACCT 258 if (racct_enable) { 259 PROC_LOCK(p); 260 error = racct_add(p, RACCT_NTHR, 1); 261 PROC_UNLOCK(p); 262 if (error != 0) 263 return (EPROCLIM); 264 } 265 #endif 266 267 /* Initialize our td */ 268 error = kern_thr_alloc(p, 0, &newtd); 269 if (error) 270 goto fail; 271 272 cpu_copy_thread(newtd, td); 273 274 bzero(&newtd->td_startzero, 275 __rangeof(struct thread, td_startzero, td_endzero)); 276 bcopy(&td->td_startcopy, &newtd->td_startcopy, 277 __rangeof(struct thread, td_startcopy, td_endcopy)); 278 279 newtd->td_proc = p; 280 thread_cow_get(newtd, td); 281 282 /* create the emuldata */ 283 linux_proc_init(td, newtd, args->flags); 284 285 em = em_find(newtd); 286 KASSERT(em != NULL, ("clone_thread: emuldata not found.\n")); 287 288 if (args->flags & LINUX_CLONE_SETTLS) 289 linux_set_cloned_tls(newtd, args->tls); 290 291 if (args->flags & LINUX_CLONE_CHILD_SETTID) 292 em->child_set_tid = args->child_tidptr; 293 else 294 em->child_set_tid = NULL; 295 296 if (args->flags & LINUX_CLONE_CHILD_CLEARTID) 297 em->child_clear_tid = args->child_tidptr; 298 else 299 em->child_clear_tid = NULL; 300 301 cpu_thread_clean(newtd); 302 303 linux_set_upcall_kse(newtd, PTROUT(args->stack)); 304 305 PROC_LOCK(p); 306 p->p_flag |= P_HADTHREADS; 307 bcopy(p->p_comm, newtd->td_name, sizeof(newtd->td_name)); 308 309 if (args->flags & LINUX_CLONE_PARENT) 310 thread_link(newtd, p->p_pptr); 311 else 312 thread_link(newtd, p); 313 314 thread_lock(td); 315 /* let the scheduler know about these things. */ 316 sched_fork_thread(td, newtd); 317 thread_unlock(td); 318 if (P_SHOULDSTOP(p)) 319 newtd->td_flags |= TDF_ASTPENDING | TDF_NEEDSUSPCHK; 320 321 if (p->p_ptevents & PTRACE_LWP) 322 newtd->td_dbgflags |= TDB_BORN; 323 PROC_UNLOCK(p); 324 325 tidhash_add(newtd); 326 327 LINUX_CTR2(clone_thread, "thread(%d) successful clone to %d", 328 td->td_tid, newtd->td_tid); 329 330 if (args->flags & LINUX_CLONE_PARENT_SETTID) { 331 error = copyout(&newtd->td_tid, args->parent_tidptr, 332 sizeof(newtd->td_tid)); 333 if (error) 334 linux_msg(td, "clone_thread: copyout td_tid failed!"); 335 } 336 337 /* 338 * Make this runnable after we are finished with it. 339 */ 340 thread_lock(newtd); 341 TD_SET_CAN_RUN(newtd); 342 sched_add(newtd, SRQ_BORING); 343 344 td->td_retval[0] = newtd->td_tid; 345 346 return (0); 347 348 fail: 349 #ifdef RACCT 350 if (racct_enable) { 351 PROC_LOCK(p); 352 racct_sub(p, RACCT_NTHR, 1); 353 PROC_UNLOCK(p); 354 } 355 #endif 356 return (error); 357 } 358 359 int 360 linux_clone(struct thread *td, struct linux_clone_args *args) 361 { 362 363 if (args->flags & LINUX_CLONE_THREAD) 364 return (linux_clone_thread(td, args)); 365 else 366 return (linux_clone_proc(td, args)); 367 } 368 369 int 370 linux_exit(struct thread *td, struct linux_exit_args *args) 371 { 372 struct linux_emuldata *em; 373 374 em = em_find(td); 375 KASSERT(em != NULL, ("exit: emuldata not found.\n")); 376 377 LINUX_CTR2(exit, "thread(%d) (%d)", em->em_tid, args->rval); 378 379 umtx_thread_exit(td); 380 381 linux_thread_detach(td); 382 383 /* 384 * XXX. When the last two threads of a process 385 * exit via pthread_exit() try thr_exit() first. 386 */ 387 kern_thr_exit(td); 388 exit1(td, args->rval, 0); 389 /* NOTREACHED */ 390 } 391 392 int 393 linux_set_tid_address(struct thread *td, struct linux_set_tid_address_args *args) 394 { 395 struct linux_emuldata *em; 396 397 em = em_find(td); 398 KASSERT(em != NULL, ("set_tid_address: emuldata not found.\n")); 399 400 em->child_clear_tid = args->tidptr; 401 402 td->td_retval[0] = em->em_tid; 403 404 LINUX_CTR3(set_tid_address, "tidptr(%d) %p, returns %d", 405 em->em_tid, args->tidptr, td->td_retval[0]); 406 407 return (0); 408 } 409 410 void 411 linux_thread_detach(struct thread *td) 412 { 413 struct linux_sys_futex_args cup; 414 struct linux_emuldata *em; 415 int *child_clear_tid; 416 int error; 417 418 em = em_find(td); 419 KASSERT(em != NULL, ("thread_detach: emuldata not found.\n")); 420 421 LINUX_CTR1(thread_detach, "thread(%d)", em->em_tid); 422 423 release_futexes(td, em); 424 425 child_clear_tid = em->child_clear_tid; 426 427 if (child_clear_tid != NULL) { 428 429 LINUX_CTR2(thread_detach, "thread(%d) %p", 430 em->em_tid, child_clear_tid); 431 432 error = suword32(child_clear_tid, 0); 433 if (error != 0) 434 return; 435 436 cup.uaddr = child_clear_tid; 437 cup.op = LINUX_FUTEX_WAKE; 438 cup.val = 1; /* wake one */ 439 cup.timeout = NULL; 440 cup.uaddr2 = NULL; 441 cup.val3 = 0; 442 error = linux_sys_futex(td, &cup); 443 /* 444 * this cannot happen at the moment and if this happens it 445 * probably means there is a user space bug 446 */ 447 if (error != 0) 448 linux_msg(td, "futex stuff in thread_detach failed."); 449 } 450 } 451