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->robust_futexes = NULL; 90 if (flags & LINUX_CLONE_THREAD) { 91 /* handled later in the code */ 92 } else { 93 struct linux_emuldata_shared *s; 94 95 s = malloc(sizeof *s, M_LINUX, M_WAITOK | M_ZERO); 96 s->refs = 1; 97 s->group_pid = child; 98 99 LIST_INIT(&s->threads); 100 em->shared = s; 101 } 102 } else { 103 /* lookup the old one */ 104 em = em_find(td->td_proc, EMUL_DOLOCK); 105 KASSERT(em != NULL, ("proc_init: emuldata not found in exec case.\n")); 106 } 107 108 em->child_clear_tid = NULL; 109 em->child_set_tid = NULL; 110 111 /* 112 * allocate the shared struct only in clone()/fork cases in the case 113 * of clone() td = calling proc and child = pid of the newly created 114 * proc 115 */ 116 if (child != 0) { 117 if (flags & LINUX_CLONE_THREAD) { 118 /* lookup the parent */ 119 /* 120 * we dont have to lock the p_em because 121 * its waiting for us in linux_clone so 122 * there is no chance of it changing the 123 * p_em->shared address 124 */ 125 p_em = em_find(td->td_proc, EMUL_DONTLOCK); 126 KASSERT(p_em != NULL, ("proc_init: parent emuldata not found for CLONE_THREAD\n")); 127 em->shared = p_em->shared; 128 EMUL_SHARED_WLOCK(&emul_shared_lock); 129 em->shared->refs++; 130 EMUL_SHARED_WUNLOCK(&emul_shared_lock); 131 } else { 132 /* 133 * handled earlier to avoid malloc(M_WAITOK) with 134 * rwlock held 135 */ 136 } 137 } 138 if (child != 0) { 139 EMUL_SHARED_WLOCK(&emul_shared_lock); 140 LIST_INSERT_HEAD(&em->shared->threads, em, threads); 141 EMUL_SHARED_WUNLOCK(&emul_shared_lock); 142 143 p = pfind(child); 144 KASSERT(p != NULL, ("process not found in proc_init\n")); 145 p->p_emuldata = em; 146 PROC_UNLOCK(p); 147 } else 148 EMUL_UNLOCK(&emul_lock); 149 150 return (0); 151 } 152 153 void 154 linux_proc_exit(void *arg __unused, struct proc *p) 155 { 156 struct linux_emuldata *em; 157 int error; 158 struct thread *td = FIRST_THREAD_IN_PROC(p); 159 int *child_clear_tid; 160 struct proc *q, *nq; 161 162 if (__predict_true(p->p_sysent != &elf_linux_sysvec)) 163 return; 164 165 release_futexes(p); 166 167 /* find the emuldata */ 168 em = em_find(p, EMUL_DOLOCK); 169 170 KASSERT(em != NULL, ("proc_exit: emuldata not found.\n")); 171 172 /* reparent all procs that are not a thread leader to initproc */ 173 if (em->shared->group_pid != p->p_pid) { 174 child_clear_tid = em->child_clear_tid; 175 EMUL_UNLOCK(&emul_lock); 176 sx_xlock(&proctree_lock); 177 wakeup(initproc); 178 PROC_LOCK(p); 179 proc_reparent(p, initproc); 180 p->p_sigparent = SIGCHLD; 181 PROC_UNLOCK(p); 182 sx_xunlock(&proctree_lock); 183 } else { 184 child_clear_tid = em->child_clear_tid; 185 EMUL_UNLOCK(&emul_lock); 186 } 187 188 EMUL_SHARED_WLOCK(&emul_shared_lock); 189 LIST_REMOVE(em, threads); 190 191 em->shared->refs--; 192 if (em->shared->refs == 0) { 193 EMUL_SHARED_WUNLOCK(&emul_shared_lock); 194 free(em->shared, M_LINUX); 195 } else 196 EMUL_SHARED_WUNLOCK(&emul_shared_lock); 197 198 if (child_clear_tid != NULL) { 199 struct linux_sys_futex_args cup; 200 int null = 0; 201 202 error = copyout(&null, child_clear_tid, sizeof(null)); 203 if (error) { 204 free(em, M_LINUX); 205 return; 206 } 207 208 /* futexes stuff */ 209 cup.uaddr = child_clear_tid; 210 cup.op = LINUX_FUTEX_WAKE; 211 cup.val = 0x7fffffff; /* Awake everyone */ 212 cup.timeout = NULL; 213 cup.uaddr2 = NULL; 214 cup.val3 = 0; 215 error = linux_sys_futex(FIRST_THREAD_IN_PROC(p), &cup); 216 /* 217 * this cannot happen at the moment and if this happens it 218 * probably means there is a user space bug 219 */ 220 if (error) 221 printf(LMSG("futex stuff in proc_exit failed.\n")); 222 } 223 224 /* clean the stuff up */ 225 free(em, M_LINUX); 226 227 /* this is a little weird but rewritten from exit1() */ 228 sx_xlock(&proctree_lock); 229 q = LIST_FIRST(&p->p_children); 230 for (; q != NULL; q = nq) { 231 nq = LIST_NEXT(q, p_sibling); 232 if (q->p_flag & P_WEXIT) 233 continue; 234 if (__predict_false(q->p_sysent != &elf_linux_sysvec)) 235 continue; 236 em = em_find(q, EMUL_DOLOCK); 237 KASSERT(em != NULL, ("linux_reparent: emuldata not found: %i\n", q->p_pid)); 238 PROC_LOCK(q); 239 if ((q->p_flag & P_WEXIT) == 0 && em->pdeath_signal != 0) { 240 psignal(q, em->pdeath_signal); 241 } 242 PROC_UNLOCK(q); 243 EMUL_UNLOCK(&emul_lock); 244 } 245 sx_xunlock(&proctree_lock); 246 } 247 248 /* 249 * This is used in a case of transition from FreeBSD binary execing to linux binary 250 * in this case we create linux emuldata proc entry with the pid of the currently running 251 * process. 252 */ 253 void 254 linux_proc_exec(void *arg __unused, struct proc *p, struct image_params *imgp) 255 { 256 if (__predict_false(imgp->sysent == &elf_linux_sysvec 257 && p->p_sysent != &elf_linux_sysvec)) 258 linux_proc_init(FIRST_THREAD_IN_PROC(p), p->p_pid, 0); 259 if (__predict_false(imgp->sysent != &elf_linux_sysvec 260 && p->p_sysent == &elf_linux_sysvec)) { 261 struct linux_emuldata *em; 262 263 /* 264 * XXX:There's a race because here we assign p->p_emuldata NULL 265 * but the process is still counted as linux one for a short 266 * time so some other process might reference it and try to 267 * access its p->p_emuldata and panicing on a NULL reference. 268 */ 269 em = em_find(p, EMUL_DONTLOCK); 270 271 KASSERT(em != NULL, ("proc_exec: emuldata not found.\n")); 272 273 EMUL_SHARED_WLOCK(&emul_shared_lock); 274 LIST_REMOVE(em, threads); 275 276 PROC_LOCK(p); 277 p->p_emuldata = NULL; 278 PROC_UNLOCK(p); 279 280 em->shared->refs--; 281 if (em->shared->refs == 0) { 282 EMUL_SHARED_WUNLOCK(&emul_shared_lock); 283 free(em->shared, M_LINUX); 284 } else 285 EMUL_SHARED_WUNLOCK(&emul_shared_lock); 286 287 free(em, M_LINUX); 288 } 289 } 290 291 void 292 linux_schedtail(void *arg __unused, struct proc *p) 293 { 294 struct linux_emuldata *em; 295 int error = 0; 296 int *child_set_tid; 297 298 if (__predict_true(p->p_sysent != &elf_linux_sysvec)) 299 return; 300 301 /* find the emuldata */ 302 em = em_find(p, EMUL_DOLOCK); 303 304 KASSERT(em != NULL, ("linux_schedtail: emuldata not found.\n")); 305 child_set_tid = em->child_set_tid; 306 EMUL_UNLOCK(&emul_lock); 307 308 if (child_set_tid != NULL) 309 error = copyout(&p->p_pid, (int *)child_set_tid, 310 sizeof(p->p_pid)); 311 312 return; 313 } 314 315 int 316 linux_set_tid_address(struct thread *td, struct linux_set_tid_address_args *args) 317 { 318 struct linux_emuldata *em; 319 320 #ifdef DEBUG 321 if (ldebug(set_tid_address)) 322 printf(ARGS(set_tid_address, "%p"), args->tidptr); 323 #endif 324 325 /* find the emuldata */ 326 em = em_find(td->td_proc, EMUL_DOLOCK); 327 328 KASSERT(em != NULL, ("set_tid_address: emuldata not found.\n")); 329 330 em->child_clear_tid = args->tidptr; 331 td->td_retval[0] = td->td_proc->p_pid; 332 333 EMUL_UNLOCK(&emul_lock); 334 return 0; 335 } 336