1 /*- 2 * Copyright (c) 2006 Roman Divacky 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer 10 * in this position and unchanged. 11 * 2. Redistributions in binary form must reproduce the above copyright 12 * notice, this list of conditions and the following disclaimer in the 13 * documentation and/or other materials provided with the distribution. 14 * 3. The name of the author may not be used to endorse or promote products 15 * derived from this software without specific prior written permission 16 * 17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 18 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 19 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 20 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 21 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 22 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 23 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 24 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 25 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 26 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 27 */ 28 29 #include <sys/cdefs.h> 30 __FBSDID("$FreeBSD$"); 31 32 #include "opt_compat.h" 33 34 #include <sys/param.h> 35 #include <sys/systm.h> 36 #include <sys/imgact.h> 37 #include <sys/kernel.h> 38 #include <sys/lock.h> 39 #include <sys/malloc.h> 40 #include <sys/mutex.h> 41 #include <sys/sx.h> 42 #include <sys/proc.h> 43 #include <sys/syscallsubr.h> 44 #include <sys/sysproto.h> 45 #include <sys/unistd.h> 46 47 #ifdef COMPAT_LINUX32 48 #include <machine/../linux32/linux.h> 49 #include <machine/../linux32/linux32_proto.h> 50 #else 51 #include <machine/../linux/linux.h> 52 #include <machine/../linux/linux_proto.h> 53 #endif 54 55 #include <compat/linux/linux_emul.h> 56 #include <compat/linux/linux_futex.h> 57 58 struct sx emul_shared_lock; 59 struct mtx emul_lock; 60 61 /* this returns locked reference to the emuldata entry (if found) */ 62 struct linux_emuldata * 63 em_find(struct proc *p, int locked) 64 { 65 struct linux_emuldata *em; 66 67 if (locked == EMUL_DOLOCK) 68 EMUL_LOCK(&emul_lock); 69 70 em = p->p_emuldata; 71 72 if (em == NULL && locked == EMUL_DOLOCK) 73 EMUL_UNLOCK(&emul_lock); 74 75 return (em); 76 } 77 78 int 79 linux_proc_init(struct thread *td, pid_t child, int flags) 80 { 81 struct linux_emuldata *em, *p_em; 82 struct proc *p; 83 84 if (child != 0) { 85 /* non-exec call */ 86 em = malloc(sizeof *em, M_LINUX, M_WAITOK | M_ZERO); 87 em->pid = child; 88 em->pdeath_signal = 0; 89 em->used_requeue = 0; 90 em->robust_futexes = NULL; 91 if (flags & LINUX_CLONE_THREAD) { 92 /* handled later in the code */ 93 } else { 94 struct linux_emuldata_shared *s; 95 96 s = malloc(sizeof *s, M_LINUX, M_WAITOK | M_ZERO); 97 s->refs = 1; 98 s->group_pid = child; 99 100 LIST_INIT(&s->threads); 101 em->shared = s; 102 } 103 } else { 104 /* lookup the old one */ 105 em = em_find(td->td_proc, EMUL_DOLOCK); 106 KASSERT(em != NULL, ("proc_init: emuldata not found in exec case.\n")); 107 } 108 109 em->child_clear_tid = NULL; 110 em->child_set_tid = NULL; 111 112 /* 113 * allocate the shared struct only in clone()/fork cases in the case 114 * of clone() td = calling proc and child = pid of the newly created 115 * proc 116 */ 117 if (child != 0) { 118 if (flags & LINUX_CLONE_THREAD) { 119 /* lookup the parent */ 120 /* 121 * we dont have to lock the p_em because 122 * its waiting for us in linux_clone so 123 * there is no chance of it changing the 124 * p_em->shared address 125 */ 126 p_em = em_find(td->td_proc, EMUL_DONTLOCK); 127 KASSERT(p_em != NULL, ("proc_init: parent emuldata not found for CLONE_THREAD\n")); 128 em->shared = p_em->shared; 129 EMUL_SHARED_WLOCK(&emul_shared_lock); 130 em->shared->refs++; 131 EMUL_SHARED_WUNLOCK(&emul_shared_lock); 132 } else { 133 /* 134 * handled earlier to avoid malloc(M_WAITOK) with 135 * rwlock held 136 */ 137 } 138 } 139 if (child != 0) { 140 EMUL_SHARED_WLOCK(&emul_shared_lock); 141 LIST_INSERT_HEAD(&em->shared->threads, em, threads); 142 EMUL_SHARED_WUNLOCK(&emul_shared_lock); 143 144 p = pfind(child); 145 KASSERT(p != NULL, ("process not found in proc_init\n")); 146 p->p_emuldata = em; 147 PROC_UNLOCK(p); 148 } else 149 EMUL_UNLOCK(&emul_lock); 150 151 return (0); 152 } 153 154 void 155 linux_proc_exit(void *arg __unused, struct proc *p) 156 { 157 struct linux_emuldata *em; 158 int error; 159 struct thread *td = FIRST_THREAD_IN_PROC(p); 160 int *child_clear_tid; 161 struct proc *q, *nq; 162 163 if (__predict_true(p->p_sysent != &elf_linux_sysvec)) 164 return; 165 166 release_futexes(p); 167 168 /* find the emuldata */ 169 em = em_find(p, EMUL_DOLOCK); 170 171 KASSERT(em != NULL, ("proc_exit: emuldata not found.\n")); 172 173 /* reparent all procs that are not a thread leader to initproc */ 174 if (em->shared->group_pid != p->p_pid) { 175 child_clear_tid = em->child_clear_tid; 176 EMUL_UNLOCK(&emul_lock); 177 sx_xlock(&proctree_lock); 178 wakeup(initproc); 179 PROC_LOCK(p); 180 proc_reparent(p, initproc); 181 p->p_sigparent = SIGCHLD; 182 PROC_UNLOCK(p); 183 sx_xunlock(&proctree_lock); 184 } else { 185 child_clear_tid = em->child_clear_tid; 186 EMUL_UNLOCK(&emul_lock); 187 } 188 189 EMUL_SHARED_WLOCK(&emul_shared_lock); 190 LIST_REMOVE(em, threads); 191 192 em->shared->refs--; 193 if (em->shared->refs == 0) { 194 EMUL_SHARED_WUNLOCK(&emul_shared_lock); 195 free(em->shared, M_LINUX); 196 } else 197 EMUL_SHARED_WUNLOCK(&emul_shared_lock); 198 199 if (child_clear_tid != NULL) { 200 struct linux_sys_futex_args cup; 201 int null = 0; 202 203 error = copyout(&null, child_clear_tid, sizeof(null)); 204 if (error) { 205 free(em, M_LINUX); 206 return; 207 } 208 209 /* futexes stuff */ 210 cup.uaddr = child_clear_tid; 211 cup.op = LINUX_FUTEX_WAKE; 212 cup.val = 0x7fffffff; /* Awake everyone */ 213 cup.timeout = NULL; 214 cup.uaddr2 = NULL; 215 cup.val3 = 0; 216 error = linux_sys_futex(FIRST_THREAD_IN_PROC(p), &cup); 217 /* 218 * this cannot happen at the moment and if this happens it 219 * probably means there is a user space bug 220 */ 221 if (error) 222 printf(LMSG("futex stuff in proc_exit failed.\n")); 223 } 224 225 /* clean the stuff up */ 226 free(em, M_LINUX); 227 228 /* this is a little weird but rewritten from exit1() */ 229 sx_xlock(&proctree_lock); 230 q = LIST_FIRST(&p->p_children); 231 for (; q != NULL; q = nq) { 232 nq = LIST_NEXT(q, p_sibling); 233 if (q->p_flag & P_WEXIT) 234 continue; 235 if (__predict_false(q->p_sysent != &elf_linux_sysvec)) 236 continue; 237 em = em_find(q, EMUL_DOLOCK); 238 KASSERT(em != NULL, ("linux_reparent: emuldata not found: %i\n", q->p_pid)); 239 PROC_LOCK(q); 240 if ((q->p_flag & P_WEXIT) == 0 && em->pdeath_signal != 0) { 241 psignal(q, em->pdeath_signal); 242 } 243 PROC_UNLOCK(q); 244 EMUL_UNLOCK(&emul_lock); 245 } 246 sx_xunlock(&proctree_lock); 247 } 248 249 /* 250 * This is used in a case of transition from FreeBSD binary execing to linux binary 251 * in this case we create linux emuldata proc entry with the pid of the currently running 252 * process. 253 */ 254 void 255 linux_proc_exec(void *arg __unused, struct proc *p, struct image_params *imgp) 256 { 257 if (__predict_false(imgp->sysent == &elf_linux_sysvec 258 && p->p_sysent != &elf_linux_sysvec)) 259 linux_proc_init(FIRST_THREAD_IN_PROC(p), p->p_pid, 0); 260 if (__predict_false(imgp->sysent != &elf_linux_sysvec 261 && p->p_sysent == &elf_linux_sysvec)) { 262 struct linux_emuldata *em; 263 264 /* 265 * XXX:There's a race because here we assign p->p_emuldata NULL 266 * but the process is still counted as linux one for a short 267 * time so some other process might reference it and try to 268 * access its p->p_emuldata and panicing on a NULL reference. 269 */ 270 em = em_find(p, EMUL_DONTLOCK); 271 272 KASSERT(em != NULL, ("proc_exec: emuldata not found.\n")); 273 274 EMUL_SHARED_WLOCK(&emul_shared_lock); 275 LIST_REMOVE(em, threads); 276 277 PROC_LOCK(p); 278 p->p_emuldata = NULL; 279 PROC_UNLOCK(p); 280 281 em->shared->refs--; 282 if (em->shared->refs == 0) { 283 EMUL_SHARED_WUNLOCK(&emul_shared_lock); 284 free(em->shared, M_LINUX); 285 } else 286 EMUL_SHARED_WUNLOCK(&emul_shared_lock); 287 288 free(em, M_LINUX); 289 } 290 } 291 292 void 293 linux_schedtail(void *arg __unused, struct proc *p) 294 { 295 struct linux_emuldata *em; 296 int error = 0; 297 int *child_set_tid; 298 299 if (__predict_true(p->p_sysent != &elf_linux_sysvec)) 300 return; 301 302 /* find the emuldata */ 303 em = em_find(p, EMUL_DOLOCK); 304 305 KASSERT(em != NULL, ("linux_schedtail: emuldata not found.\n")); 306 child_set_tid = em->child_set_tid; 307 EMUL_UNLOCK(&emul_lock); 308 309 if (child_set_tid != NULL) 310 error = copyout(&p->p_pid, (int *)child_set_tid, 311 sizeof(p->p_pid)); 312 313 return; 314 } 315 316 int 317 linux_set_tid_address(struct thread *td, struct linux_set_tid_address_args *args) 318 { 319 struct linux_emuldata *em; 320 321 #ifdef DEBUG 322 if (ldebug(set_tid_address)) 323 printf(ARGS(set_tid_address, "%p"), args->tidptr); 324 #endif 325 326 /* find the emuldata */ 327 em = em_find(td->td_proc, EMUL_DOLOCK); 328 329 KASSERT(em != NULL, ("set_tid_address: emuldata not found.\n")); 330 331 em->child_clear_tid = args->tidptr; 332 td->td_retval[0] = td->td_proc->p_pid; 333 334 EMUL_UNLOCK(&emul_lock); 335 return 0; 336 } 337