1 /* $NetBSD: sysv_shm.c,v 1.23 1994/07/04 23:25:12 glass Exp $ */ 2 /*- 3 * Copyright (c) 1994 Adam Glass and Charles Hannum. 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 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 3. All advertising materials mentioning features or use of this software 14 * must display the following acknowledgement: 15 * This product includes software developed by Adam Glass and Charles 16 * Hannum. 17 * 4. The names of the authors may not be used to endorse or promote products 18 * derived from this software without specific prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE AUTHORS ``AS IS'' AND ANY EXPRESS OR 21 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 22 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 23 * IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY DIRECT, INDIRECT, 24 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 25 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 26 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 27 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 28 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 29 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 30 */ 31 /*- 32 * Copyright (c) 2003-2005 McAfee, Inc. 33 * Copyright (c) 2016-2017 Robert N. M. Watson 34 * All rights reserved. 35 * 36 * This software was developed for the FreeBSD Project in part by McAfee 37 * Research, the Security Research Division of McAfee, Inc under DARPA/SPAWAR 38 * contract N66001-01-C-8035 ("CBOSS"), as part of the DARPA CHATS research 39 * program. 40 * 41 * Portions of this software were developed by BAE Systems, the University of 42 * Cambridge Computer Laboratory, and Memorial University under DARPA/AFRL 43 * contract FA8650-15-C-7558 ("CADETS"), as part of the DARPA Transparent 44 * Computing (TC) research program. 45 * 46 * Redistribution and use in source and binary forms, with or without 47 * modification, are permitted provided that the following conditions 48 * are met: 49 * 1. Redistributions of source code must retain the above copyright 50 * notice, this list of conditions and the following disclaimer. 51 * 2. Redistributions in binary form must reproduce the above copyright 52 * notice, this list of conditions and the following disclaimer in the 53 * documentation and/or other materials provided with the distribution. 54 * 55 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 56 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 57 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 58 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 59 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 60 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 61 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 62 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 63 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 64 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 65 * SUCH DAMAGE. 66 */ 67 68 #include <sys/cdefs.h> 69 __FBSDID("$FreeBSD$"); 70 71 #include "opt_compat.h" 72 #include "opt_sysvipc.h" 73 74 #include <sys/param.h> 75 #include <sys/systm.h> 76 #include <sys/kernel.h> 77 #include <sys/limits.h> 78 #include <sys/lock.h> 79 #include <sys/sysctl.h> 80 #include <sys/shm.h> 81 #include <sys/proc.h> 82 #include <sys/malloc.h> 83 #include <sys/mman.h> 84 #include <sys/module.h> 85 #include <sys/mutex.h> 86 #include <sys/racct.h> 87 #include <sys/resourcevar.h> 88 #include <sys/rwlock.h> 89 #include <sys/stat.h> 90 #include <sys/syscall.h> 91 #include <sys/syscallsubr.h> 92 #include <sys/sysent.h> 93 #include <sys/sysproto.h> 94 #include <sys/jail.h> 95 96 #include <security/audit/audit.h> 97 #include <security/mac/mac_framework.h> 98 99 #include <vm/vm.h> 100 #include <vm/vm_param.h> 101 #include <vm/pmap.h> 102 #include <vm/vm_object.h> 103 #include <vm/vm_map.h> 104 #include <vm/vm_page.h> 105 #include <vm/vm_pager.h> 106 107 FEATURE(sysv_shm, "System V shared memory segments support"); 108 109 static MALLOC_DEFINE(M_SHM, "shm", "SVID compatible shared memory segments"); 110 111 static int shmget_allocate_segment(struct thread *td, 112 struct shmget_args *uap, int mode); 113 static int shmget_existing(struct thread *td, struct shmget_args *uap, 114 int mode, int segnum); 115 116 #define SHMSEG_FREE 0x0200 117 #define SHMSEG_REMOVED 0x0400 118 #define SHMSEG_ALLOCATED 0x0800 119 120 static int shm_last_free, shm_nused, shmalloced; 121 vm_size_t shm_committed; 122 static struct shmid_kernel *shmsegs; 123 static unsigned shm_prison_slot; 124 125 struct shmmap_state { 126 vm_offset_t va; 127 int shmid; 128 }; 129 130 static void shm_deallocate_segment(struct shmid_kernel *); 131 static int shm_find_segment_by_key(struct prison *, key_t); 132 static struct shmid_kernel *shm_find_segment(struct prison *, int, bool); 133 static int shm_delete_mapping(struct vmspace *vm, struct shmmap_state *); 134 static void shmrealloc(void); 135 static int shminit(void); 136 static int sysvshm_modload(struct module *, int, void *); 137 static int shmunload(void); 138 static void shmexit_myhook(struct vmspace *vm); 139 static void shmfork_myhook(struct proc *p1, struct proc *p2); 140 static int sysctl_shmsegs(SYSCTL_HANDLER_ARGS); 141 static void shm_remove(struct shmid_kernel *, int); 142 static struct prison *shm_find_prison(struct ucred *); 143 static int shm_prison_cansee(struct prison *, struct shmid_kernel *); 144 static int shm_prison_check(void *, void *); 145 static int shm_prison_set(void *, void *); 146 static int shm_prison_get(void *, void *); 147 static int shm_prison_remove(void *, void *); 148 static void shm_prison_cleanup(struct prison *); 149 150 /* 151 * Tuneable values. 152 */ 153 #ifndef SHMMAXPGS 154 #define SHMMAXPGS 131072 /* Note: sysv shared memory is swap backed. */ 155 #endif 156 #ifndef SHMMAX 157 #define SHMMAX (SHMMAXPGS*PAGE_SIZE) 158 #endif 159 #ifndef SHMMIN 160 #define SHMMIN 1 161 #endif 162 #ifndef SHMMNI 163 #define SHMMNI 192 164 #endif 165 #ifndef SHMSEG 166 #define SHMSEG 128 167 #endif 168 #ifndef SHMALL 169 #define SHMALL (SHMMAXPGS) 170 #endif 171 172 struct shminfo shminfo = { 173 .shmmax = SHMMAX, 174 .shmmin = SHMMIN, 175 .shmmni = SHMMNI, 176 .shmseg = SHMSEG, 177 .shmall = SHMALL 178 }; 179 180 static int shm_use_phys; 181 static int shm_allow_removed = 1; 182 183 SYSCTL_ULONG(_kern_ipc, OID_AUTO, shmmax, CTLFLAG_RWTUN, &shminfo.shmmax, 0, 184 "Maximum shared memory segment size"); 185 SYSCTL_ULONG(_kern_ipc, OID_AUTO, shmmin, CTLFLAG_RWTUN, &shminfo.shmmin, 0, 186 "Minimum shared memory segment size"); 187 SYSCTL_ULONG(_kern_ipc, OID_AUTO, shmmni, CTLFLAG_RDTUN, &shminfo.shmmni, 0, 188 "Number of shared memory identifiers"); 189 SYSCTL_ULONG(_kern_ipc, OID_AUTO, shmseg, CTLFLAG_RDTUN, &shminfo.shmseg, 0, 190 "Number of segments per process"); 191 SYSCTL_ULONG(_kern_ipc, OID_AUTO, shmall, CTLFLAG_RWTUN, &shminfo.shmall, 0, 192 "Maximum number of pages available for shared memory"); 193 SYSCTL_INT(_kern_ipc, OID_AUTO, shm_use_phys, CTLFLAG_RWTUN, 194 &shm_use_phys, 0, "Enable/Disable locking of shared memory pages in core"); 195 SYSCTL_INT(_kern_ipc, OID_AUTO, shm_allow_removed, CTLFLAG_RWTUN, 196 &shm_allow_removed, 0, 197 "Enable/Disable attachment to attached segments marked for removal"); 198 SYSCTL_PROC(_kern_ipc, OID_AUTO, shmsegs, CTLTYPE_OPAQUE | CTLFLAG_RD | 199 CTLFLAG_MPSAFE, NULL, 0, sysctl_shmsegs, "", 200 "Current number of shared memory segments allocated"); 201 202 static struct sx sysvshmsx; 203 #define SYSVSHM_LOCK() sx_xlock(&sysvshmsx) 204 #define SYSVSHM_UNLOCK() sx_xunlock(&sysvshmsx) 205 #define SYSVSHM_ASSERT_LOCKED() sx_assert(&sysvshmsx, SA_XLOCKED) 206 207 static int 208 shm_find_segment_by_key(struct prison *pr, key_t key) 209 { 210 int i; 211 212 for (i = 0; i < shmalloced; i++) 213 if ((shmsegs[i].u.shm_perm.mode & SHMSEG_ALLOCATED) && 214 shmsegs[i].cred != NULL && 215 shmsegs[i].cred->cr_prison == pr && 216 shmsegs[i].u.shm_perm.key == key) 217 return (i); 218 return (-1); 219 } 220 221 /* 222 * Finds segment either by shmid if is_shmid is true, or by segnum if 223 * is_shmid is false. 224 */ 225 static struct shmid_kernel * 226 shm_find_segment(struct prison *rpr, int arg, bool is_shmid) 227 { 228 struct shmid_kernel *shmseg; 229 int segnum; 230 231 segnum = is_shmid ? IPCID_TO_IX(arg) : arg; 232 if (segnum < 0 || segnum >= shmalloced) 233 return (NULL); 234 shmseg = &shmsegs[segnum]; 235 if ((shmseg->u.shm_perm.mode & SHMSEG_ALLOCATED) == 0 || 236 (!shm_allow_removed && 237 (shmseg->u.shm_perm.mode & SHMSEG_REMOVED) != 0) || 238 (is_shmid && shmseg->u.shm_perm.seq != IPCID_TO_SEQ(arg)) || 239 shm_prison_cansee(rpr, shmseg) != 0) 240 return (NULL); 241 return (shmseg); 242 } 243 244 static void 245 shm_deallocate_segment(struct shmid_kernel *shmseg) 246 { 247 vm_size_t size; 248 249 SYSVSHM_ASSERT_LOCKED(); 250 251 vm_object_deallocate(shmseg->object); 252 shmseg->object = NULL; 253 size = round_page(shmseg->u.shm_segsz); 254 shm_committed -= btoc(size); 255 shm_nused--; 256 shmseg->u.shm_perm.mode = SHMSEG_FREE; 257 #ifdef MAC 258 mac_sysvshm_cleanup(shmseg); 259 #endif 260 racct_sub_cred(shmseg->cred, RACCT_NSHM, 1); 261 racct_sub_cred(shmseg->cred, RACCT_SHMSIZE, size); 262 crfree(shmseg->cred); 263 shmseg->cred = NULL; 264 } 265 266 static int 267 shm_delete_mapping(struct vmspace *vm, struct shmmap_state *shmmap_s) 268 { 269 struct shmid_kernel *shmseg; 270 int segnum, result; 271 vm_size_t size; 272 273 SYSVSHM_ASSERT_LOCKED(); 274 segnum = IPCID_TO_IX(shmmap_s->shmid); 275 KASSERT(segnum >= 0 && segnum < shmalloced, 276 ("segnum %d shmalloced %d", segnum, shmalloced)); 277 278 shmseg = &shmsegs[segnum]; 279 size = round_page(shmseg->u.shm_segsz); 280 result = vm_map_remove(&vm->vm_map, shmmap_s->va, shmmap_s->va + size); 281 if (result != KERN_SUCCESS) 282 return (EINVAL); 283 shmmap_s->shmid = -1; 284 shmseg->u.shm_dtime = time_second; 285 if (--shmseg->u.shm_nattch == 0 && 286 (shmseg->u.shm_perm.mode & SHMSEG_REMOVED)) { 287 shm_deallocate_segment(shmseg); 288 shm_last_free = segnum; 289 } 290 return (0); 291 } 292 293 static void 294 shm_remove(struct shmid_kernel *shmseg, int segnum) 295 { 296 297 shmseg->u.shm_perm.key = IPC_PRIVATE; 298 shmseg->u.shm_perm.mode |= SHMSEG_REMOVED; 299 if (shmseg->u.shm_nattch == 0) { 300 shm_deallocate_segment(shmseg); 301 shm_last_free = segnum; 302 } 303 } 304 305 static struct prison * 306 shm_find_prison(struct ucred *cred) 307 { 308 struct prison *pr, *rpr; 309 310 pr = cred->cr_prison; 311 prison_lock(pr); 312 rpr = osd_jail_get(pr, shm_prison_slot); 313 prison_unlock(pr); 314 return rpr; 315 } 316 317 static int 318 shm_prison_cansee(struct prison *rpr, struct shmid_kernel *shmseg) 319 { 320 321 if (shmseg->cred == NULL || 322 !(rpr == shmseg->cred->cr_prison || 323 prison_ischild(rpr, shmseg->cred->cr_prison))) 324 return (EINVAL); 325 return (0); 326 } 327 328 static int 329 kern_shmdt_locked(struct thread *td, const void *shmaddr) 330 { 331 struct proc *p = td->td_proc; 332 struct shmmap_state *shmmap_s; 333 #if defined(AUDIT) || defined(MAC) 334 struct shmid_kernel *shmsegptr; 335 #endif 336 #ifdef MAC 337 int error; 338 #endif 339 int i; 340 341 SYSVSHM_ASSERT_LOCKED(); 342 if (shm_find_prison(td->td_ucred) == NULL) 343 return (ENOSYS); 344 shmmap_s = p->p_vmspace->vm_shm; 345 if (shmmap_s == NULL) 346 return (EINVAL); 347 AUDIT_ARG_SVIPC_ID(shmmap_s->shmid); 348 for (i = 0; i < shminfo.shmseg; i++, shmmap_s++) { 349 if (shmmap_s->shmid != -1 && 350 shmmap_s->va == (vm_offset_t)shmaddr) { 351 break; 352 } 353 } 354 if (i == shminfo.shmseg) 355 return (EINVAL); 356 #if (defined(AUDIT) && defined(KDTRACE_HOOKS)) || defined(MAC) 357 shmsegptr = &shmsegs[IPCID_TO_IX(shmmap_s->shmid)]; 358 #endif 359 #ifdef MAC 360 error = mac_sysvshm_check_shmdt(td->td_ucred, shmsegptr); 361 if (error != 0) 362 return (error); 363 #endif 364 return (shm_delete_mapping(p->p_vmspace, shmmap_s)); 365 } 366 367 #ifndef _SYS_SYSPROTO_H_ 368 struct shmdt_args { 369 const void *shmaddr; 370 }; 371 #endif 372 int 373 sys_shmdt(struct thread *td, struct shmdt_args *uap) 374 { 375 int error; 376 377 SYSVSHM_LOCK(); 378 error = kern_shmdt_locked(td, uap->shmaddr); 379 SYSVSHM_UNLOCK(); 380 return (error); 381 } 382 383 static int 384 kern_shmat_locked(struct thread *td, int shmid, const void *shmaddr, 385 int shmflg) 386 { 387 struct prison *rpr; 388 struct proc *p = td->td_proc; 389 struct shmid_kernel *shmseg; 390 struct shmmap_state *shmmap_s; 391 vm_offset_t attach_va; 392 vm_prot_t prot; 393 vm_size_t size; 394 int error, i, rv; 395 396 AUDIT_ARG_SVIPC_ID(shmid); 397 AUDIT_ARG_VALUE(shmflg); 398 399 SYSVSHM_ASSERT_LOCKED(); 400 rpr = shm_find_prison(td->td_ucred); 401 if (rpr == NULL) 402 return (ENOSYS); 403 shmmap_s = p->p_vmspace->vm_shm; 404 if (shmmap_s == NULL) { 405 shmmap_s = malloc(shminfo.shmseg * sizeof(struct shmmap_state), 406 M_SHM, M_WAITOK); 407 for (i = 0; i < shminfo.shmseg; i++) 408 shmmap_s[i].shmid = -1; 409 KASSERT(p->p_vmspace->vm_shm == NULL, ("raced")); 410 p->p_vmspace->vm_shm = shmmap_s; 411 } 412 shmseg = shm_find_segment(rpr, shmid, true); 413 if (shmseg == NULL) 414 return (EINVAL); 415 error = ipcperm(td, &shmseg->u.shm_perm, 416 (shmflg & SHM_RDONLY) ? IPC_R : IPC_R|IPC_W); 417 if (error != 0) 418 return (error); 419 #ifdef MAC 420 error = mac_sysvshm_check_shmat(td->td_ucred, shmseg, shmflg); 421 if (error != 0) 422 return (error); 423 #endif 424 for (i = 0; i < shminfo.shmseg; i++) { 425 if (shmmap_s->shmid == -1) 426 break; 427 shmmap_s++; 428 } 429 if (i >= shminfo.shmseg) 430 return (EMFILE); 431 size = round_page(shmseg->u.shm_segsz); 432 prot = VM_PROT_READ; 433 if ((shmflg & SHM_RDONLY) == 0) 434 prot |= VM_PROT_WRITE; 435 if (shmaddr != NULL) { 436 if ((shmflg & SHM_RND) != 0) 437 attach_va = rounddown2((vm_offset_t)shmaddr, SHMLBA); 438 else if (((vm_offset_t)shmaddr & (SHMLBA-1)) == 0) 439 attach_va = (vm_offset_t)shmaddr; 440 else 441 return (EINVAL); 442 } else { 443 /* 444 * This is just a hint to vm_map_find() about where to 445 * put it. 446 */ 447 attach_va = round_page((vm_offset_t)p->p_vmspace->vm_daddr + 448 lim_max(td, RLIMIT_DATA)); 449 } 450 451 vm_object_reference(shmseg->object); 452 rv = vm_map_find(&p->p_vmspace->vm_map, shmseg->object, 0, &attach_va, 453 size, 0, shmaddr != NULL ? VMFS_NO_SPACE : VMFS_OPTIMAL_SPACE, 454 prot, prot, MAP_INHERIT_SHARE | MAP_PREFAULT_PARTIAL); 455 if (rv != KERN_SUCCESS) { 456 vm_object_deallocate(shmseg->object); 457 return (ENOMEM); 458 } 459 460 shmmap_s->va = attach_va; 461 shmmap_s->shmid = shmid; 462 shmseg->u.shm_lpid = p->p_pid; 463 shmseg->u.shm_atime = time_second; 464 shmseg->u.shm_nattch++; 465 td->td_retval[0] = attach_va; 466 return (error); 467 } 468 469 int 470 kern_shmat(struct thread *td, int shmid, const void *shmaddr, int shmflg) 471 { 472 int error; 473 474 SYSVSHM_LOCK(); 475 error = kern_shmat_locked(td, shmid, shmaddr, shmflg); 476 SYSVSHM_UNLOCK(); 477 return (error); 478 } 479 480 #ifndef _SYS_SYSPROTO_H_ 481 struct shmat_args { 482 int shmid; 483 const void *shmaddr; 484 int shmflg; 485 }; 486 #endif 487 int 488 sys_shmat(struct thread *td, struct shmat_args *uap) 489 { 490 491 return (kern_shmat(td, uap->shmid, uap->shmaddr, uap->shmflg)); 492 } 493 494 static int 495 kern_shmctl_locked(struct thread *td, int shmid, int cmd, void *buf, 496 size_t *bufsz) 497 { 498 struct prison *rpr; 499 struct shmid_kernel *shmseg; 500 struct shmid_ds *shmidp; 501 struct shm_info shm_info; 502 int error; 503 504 SYSVSHM_ASSERT_LOCKED(); 505 506 rpr = shm_find_prison(td->td_ucred); 507 if (rpr == NULL) 508 return (ENOSYS); 509 510 AUDIT_ARG_SVIPC_ID(shmid); 511 AUDIT_ARG_SVIPC_CMD(cmd); 512 513 switch (cmd) { 514 /* 515 * It is possible that kern_shmctl is being called from the Linux ABI 516 * layer, in which case, we will need to implement IPC_INFO. It should 517 * be noted that other shmctl calls will be funneled through here for 518 * Linix binaries as well. 519 * 520 * NB: The Linux ABI layer will convert this data to structure(s) more 521 * consistent with the Linux ABI. 522 */ 523 case IPC_INFO: 524 memcpy(buf, &shminfo, sizeof(shminfo)); 525 if (bufsz) 526 *bufsz = sizeof(shminfo); 527 td->td_retval[0] = shmalloced; 528 return (0); 529 case SHM_INFO: { 530 shm_info.used_ids = shm_nused; 531 shm_info.shm_rss = 0; /*XXX where to get from ? */ 532 shm_info.shm_tot = 0; /*XXX where to get from ? */ 533 shm_info.shm_swp = 0; /*XXX where to get from ? */ 534 shm_info.swap_attempts = 0; /*XXX where to get from ? */ 535 shm_info.swap_successes = 0; /*XXX where to get from ? */ 536 memcpy(buf, &shm_info, sizeof(shm_info)); 537 if (bufsz != NULL) 538 *bufsz = sizeof(shm_info); 539 td->td_retval[0] = shmalloced; 540 return (0); 541 } 542 } 543 shmseg = shm_find_segment(rpr, shmid, cmd != SHM_STAT); 544 if (shmseg == NULL) 545 return (EINVAL); 546 #ifdef MAC 547 error = mac_sysvshm_check_shmctl(td->td_ucred, shmseg, cmd); 548 if (error != 0) 549 return (error); 550 #endif 551 switch (cmd) { 552 case SHM_STAT: 553 case IPC_STAT: 554 shmidp = (struct shmid_ds *)buf; 555 error = ipcperm(td, &shmseg->u.shm_perm, IPC_R); 556 if (error != 0) 557 return (error); 558 memcpy(shmidp, &shmseg->u, sizeof(struct shmid_ds)); 559 if (td->td_ucred->cr_prison != shmseg->cred->cr_prison) 560 shmidp->shm_perm.key = IPC_PRIVATE; 561 if (bufsz != NULL) 562 *bufsz = sizeof(struct shmid_ds); 563 if (cmd == SHM_STAT) { 564 td->td_retval[0] = IXSEQ_TO_IPCID(shmid, 565 shmseg->u.shm_perm); 566 } 567 break; 568 case IPC_SET: 569 shmidp = (struct shmid_ds *)buf; 570 AUDIT_ARG_SVIPC_PERM(&shmidp->shm_perm); 571 error = ipcperm(td, &shmseg->u.shm_perm, IPC_M); 572 if (error != 0) 573 return (error); 574 shmseg->u.shm_perm.uid = shmidp->shm_perm.uid; 575 shmseg->u.shm_perm.gid = shmidp->shm_perm.gid; 576 shmseg->u.shm_perm.mode = 577 (shmseg->u.shm_perm.mode & ~ACCESSPERMS) | 578 (shmidp->shm_perm.mode & ACCESSPERMS); 579 shmseg->u.shm_ctime = time_second; 580 break; 581 case IPC_RMID: 582 error = ipcperm(td, &shmseg->u.shm_perm, IPC_M); 583 if (error != 0) 584 return (error); 585 shm_remove(shmseg, IPCID_TO_IX(shmid)); 586 break; 587 #if 0 588 case SHM_LOCK: 589 case SHM_UNLOCK: 590 #endif 591 default: 592 error = EINVAL; 593 break; 594 } 595 return (error); 596 } 597 598 int 599 kern_shmctl(struct thread *td, int shmid, int cmd, void *buf, size_t *bufsz) 600 { 601 int error; 602 603 SYSVSHM_LOCK(); 604 error = kern_shmctl_locked(td, shmid, cmd, buf, bufsz); 605 SYSVSHM_UNLOCK(); 606 return (error); 607 } 608 609 610 #ifndef _SYS_SYSPROTO_H_ 611 struct shmctl_args { 612 int shmid; 613 int cmd; 614 struct shmid_ds *buf; 615 }; 616 #endif 617 int 618 sys_shmctl(struct thread *td, struct shmctl_args *uap) 619 { 620 int error; 621 struct shmid_ds buf; 622 size_t bufsz; 623 624 /* 625 * The only reason IPC_INFO, SHM_INFO, SHM_STAT exists is to support 626 * Linux binaries. If we see the call come through the FreeBSD ABI, 627 * return an error back to the user since we do not to support this. 628 */ 629 if (uap->cmd == IPC_INFO || uap->cmd == SHM_INFO || 630 uap->cmd == SHM_STAT) 631 return (EINVAL); 632 633 /* IPC_SET needs to copyin the buffer before calling kern_shmctl */ 634 if (uap->cmd == IPC_SET) { 635 if ((error = copyin(uap->buf, &buf, sizeof(struct shmid_ds)))) 636 goto done; 637 } 638 639 error = kern_shmctl(td, uap->shmid, uap->cmd, (void *)&buf, &bufsz); 640 if (error) 641 goto done; 642 643 /* Cases in which we need to copyout */ 644 switch (uap->cmd) { 645 case IPC_STAT: 646 error = copyout(&buf, uap->buf, bufsz); 647 break; 648 } 649 650 done: 651 if (error) { 652 /* Invalidate the return value */ 653 td->td_retval[0] = -1; 654 } 655 return (error); 656 } 657 658 659 static int 660 shmget_existing(struct thread *td, struct shmget_args *uap, int mode, 661 int segnum) 662 { 663 struct shmid_kernel *shmseg; 664 #ifdef MAC 665 int error; 666 #endif 667 668 SYSVSHM_ASSERT_LOCKED(); 669 KASSERT(segnum >= 0 && segnum < shmalloced, 670 ("segnum %d shmalloced %d", segnum, shmalloced)); 671 shmseg = &shmsegs[segnum]; 672 if ((uap->shmflg & (IPC_CREAT | IPC_EXCL)) == (IPC_CREAT | IPC_EXCL)) 673 return (EEXIST); 674 #ifdef MAC 675 error = mac_sysvshm_check_shmget(td->td_ucred, shmseg, uap->shmflg); 676 if (error != 0) 677 return (error); 678 #endif 679 if (uap->size != 0 && uap->size > shmseg->u.shm_segsz) 680 return (EINVAL); 681 td->td_retval[0] = IXSEQ_TO_IPCID(segnum, shmseg->u.shm_perm); 682 return (0); 683 } 684 685 static int 686 shmget_allocate_segment(struct thread *td, struct shmget_args *uap, int mode) 687 { 688 struct ucred *cred = td->td_ucred; 689 struct shmid_kernel *shmseg; 690 vm_object_t shm_object; 691 int i, segnum; 692 size_t size; 693 694 SYSVSHM_ASSERT_LOCKED(); 695 696 if (uap->size < shminfo.shmmin || uap->size > shminfo.shmmax) 697 return (EINVAL); 698 if (shm_nused >= shminfo.shmmni) /* Any shmids left? */ 699 return (ENOSPC); 700 size = round_page(uap->size); 701 if (shm_committed + btoc(size) > shminfo.shmall) 702 return (ENOMEM); 703 if (shm_last_free < 0) { 704 shmrealloc(); /* Maybe expand the shmsegs[] array. */ 705 for (i = 0; i < shmalloced; i++) 706 if (shmsegs[i].u.shm_perm.mode & SHMSEG_FREE) 707 break; 708 if (i == shmalloced) 709 return (ENOSPC); 710 segnum = i; 711 } else { 712 segnum = shm_last_free; 713 shm_last_free = -1; 714 } 715 KASSERT(segnum >= 0 && segnum < shmalloced, 716 ("segnum %d shmalloced %d", segnum, shmalloced)); 717 shmseg = &shmsegs[segnum]; 718 #ifdef RACCT 719 if (racct_enable) { 720 PROC_LOCK(td->td_proc); 721 if (racct_add(td->td_proc, RACCT_NSHM, 1)) { 722 PROC_UNLOCK(td->td_proc); 723 return (ENOSPC); 724 } 725 if (racct_add(td->td_proc, RACCT_SHMSIZE, size)) { 726 racct_sub(td->td_proc, RACCT_NSHM, 1); 727 PROC_UNLOCK(td->td_proc); 728 return (ENOMEM); 729 } 730 PROC_UNLOCK(td->td_proc); 731 } 732 #endif 733 734 /* 735 * We make sure that we have allocated a pager before we need 736 * to. 737 */ 738 shm_object = vm_pager_allocate(shm_use_phys ? OBJT_PHYS : OBJT_SWAP, 739 0, size, VM_PROT_DEFAULT, 0, cred); 740 if (shm_object == NULL) { 741 #ifdef RACCT 742 if (racct_enable) { 743 PROC_LOCK(td->td_proc); 744 racct_sub(td->td_proc, RACCT_NSHM, 1); 745 racct_sub(td->td_proc, RACCT_SHMSIZE, size); 746 PROC_UNLOCK(td->td_proc); 747 } 748 #endif 749 return (ENOMEM); 750 } 751 shm_object->pg_color = 0; 752 VM_OBJECT_WLOCK(shm_object); 753 vm_object_clear_flag(shm_object, OBJ_ONEMAPPING); 754 vm_object_set_flag(shm_object, OBJ_COLORED | OBJ_NOSPLIT); 755 VM_OBJECT_WUNLOCK(shm_object); 756 757 shmseg->object = shm_object; 758 shmseg->u.shm_perm.cuid = shmseg->u.shm_perm.uid = cred->cr_uid; 759 shmseg->u.shm_perm.cgid = shmseg->u.shm_perm.gid = cred->cr_gid; 760 shmseg->u.shm_perm.mode = (mode & ACCESSPERMS) | SHMSEG_ALLOCATED; 761 shmseg->u.shm_perm.key = uap->key; 762 shmseg->u.shm_perm.seq = (shmseg->u.shm_perm.seq + 1) & 0x7fff; 763 shmseg->cred = crhold(cred); 764 shmseg->u.shm_segsz = uap->size; 765 shmseg->u.shm_cpid = td->td_proc->p_pid; 766 shmseg->u.shm_lpid = shmseg->u.shm_nattch = 0; 767 shmseg->u.shm_atime = shmseg->u.shm_dtime = 0; 768 #ifdef MAC 769 mac_sysvshm_create(cred, shmseg); 770 #endif 771 shmseg->u.shm_ctime = time_second; 772 shm_committed += btoc(size); 773 shm_nused++; 774 td->td_retval[0] = IXSEQ_TO_IPCID(segnum, shmseg->u.shm_perm); 775 776 return (0); 777 } 778 779 #ifndef _SYS_SYSPROTO_H_ 780 struct shmget_args { 781 key_t key; 782 size_t size; 783 int shmflg; 784 }; 785 #endif 786 int 787 sys_shmget(struct thread *td, struct shmget_args *uap) 788 { 789 int segnum, mode; 790 int error; 791 792 if (shm_find_prison(td->td_ucred) == NULL) 793 return (ENOSYS); 794 mode = uap->shmflg & ACCESSPERMS; 795 SYSVSHM_LOCK(); 796 if (uap->key == IPC_PRIVATE) { 797 error = shmget_allocate_segment(td, uap, mode); 798 } else { 799 segnum = shm_find_segment_by_key(td->td_ucred->cr_prison, 800 uap->key); 801 if (segnum >= 0) 802 error = shmget_existing(td, uap, mode, segnum); 803 else if ((uap->shmflg & IPC_CREAT) == 0) 804 error = ENOENT; 805 else 806 error = shmget_allocate_segment(td, uap, mode); 807 } 808 SYSVSHM_UNLOCK(); 809 return (error); 810 } 811 812 static void 813 shmfork_myhook(struct proc *p1, struct proc *p2) 814 { 815 struct shmmap_state *shmmap_s; 816 size_t size; 817 int i; 818 819 SYSVSHM_LOCK(); 820 size = shminfo.shmseg * sizeof(struct shmmap_state); 821 shmmap_s = malloc(size, M_SHM, M_WAITOK); 822 bcopy(p1->p_vmspace->vm_shm, shmmap_s, size); 823 p2->p_vmspace->vm_shm = shmmap_s; 824 for (i = 0; i < shminfo.shmseg; i++, shmmap_s++) { 825 if (shmmap_s->shmid != -1) { 826 KASSERT(IPCID_TO_IX(shmmap_s->shmid) >= 0 && 827 IPCID_TO_IX(shmmap_s->shmid) < shmalloced, 828 ("segnum %d shmalloced %d", 829 IPCID_TO_IX(shmmap_s->shmid), shmalloced)); 830 shmsegs[IPCID_TO_IX(shmmap_s->shmid)].u.shm_nattch++; 831 } 832 } 833 SYSVSHM_UNLOCK(); 834 } 835 836 static void 837 shmexit_myhook(struct vmspace *vm) 838 { 839 struct shmmap_state *base, *shm; 840 int i; 841 842 base = vm->vm_shm; 843 if (base != NULL) { 844 vm->vm_shm = NULL; 845 SYSVSHM_LOCK(); 846 for (i = 0, shm = base; i < shminfo.shmseg; i++, shm++) { 847 if (shm->shmid != -1) 848 shm_delete_mapping(vm, shm); 849 } 850 SYSVSHM_UNLOCK(); 851 free(base, M_SHM); 852 } 853 } 854 855 static void 856 shmrealloc(void) 857 { 858 struct shmid_kernel *newsegs; 859 int i; 860 861 SYSVSHM_ASSERT_LOCKED(); 862 863 if (shmalloced >= shminfo.shmmni) 864 return; 865 866 newsegs = malloc(shminfo.shmmni * sizeof(*newsegs), M_SHM, M_WAITOK); 867 for (i = 0; i < shmalloced; i++) 868 bcopy(&shmsegs[i], &newsegs[i], sizeof(newsegs[0])); 869 for (; i < shminfo.shmmni; i++) { 870 newsegs[i].u.shm_perm.mode = SHMSEG_FREE; 871 newsegs[i].u.shm_perm.seq = 0; 872 #ifdef MAC 873 mac_sysvshm_init(&newsegs[i]); 874 #endif 875 } 876 free(shmsegs, M_SHM); 877 shmsegs = newsegs; 878 shmalloced = shminfo.shmmni; 879 } 880 881 static struct syscall_helper_data shm_syscalls[] = { 882 SYSCALL_INIT_HELPER(shmat), 883 SYSCALL_INIT_HELPER(shmctl), 884 SYSCALL_INIT_HELPER(shmdt), 885 SYSCALL_INIT_HELPER(shmget), 886 #if defined(COMPAT_FREEBSD4) || defined(COMPAT_FREEBSD5) || \ 887 defined(COMPAT_FREEBSD6) || defined(COMPAT_FREEBSD7) 888 SYSCALL_INIT_HELPER_COMPAT(freebsd7_shmctl), 889 #endif 890 #if defined(__i386__) && (defined(COMPAT_FREEBSD4) || defined(COMPAT_43)) 891 SYSCALL_INIT_HELPER(shmsys), 892 #endif 893 SYSCALL_INIT_LAST 894 }; 895 896 #ifdef COMPAT_FREEBSD32 897 #include <compat/freebsd32/freebsd32.h> 898 #include <compat/freebsd32/freebsd32_ipc.h> 899 #include <compat/freebsd32/freebsd32_proto.h> 900 #include <compat/freebsd32/freebsd32_signal.h> 901 #include <compat/freebsd32/freebsd32_syscall.h> 902 #include <compat/freebsd32/freebsd32_util.h> 903 904 static struct syscall_helper_data shm32_syscalls[] = { 905 SYSCALL32_INIT_HELPER_COMPAT(shmat), 906 SYSCALL32_INIT_HELPER_COMPAT(shmdt), 907 SYSCALL32_INIT_HELPER_COMPAT(shmget), 908 SYSCALL32_INIT_HELPER(freebsd32_shmsys), 909 SYSCALL32_INIT_HELPER(freebsd32_shmctl), 910 #if defined(COMPAT_FREEBSD4) || defined(COMPAT_FREEBSD5) || \ 911 defined(COMPAT_FREEBSD6) || defined(COMPAT_FREEBSD7) 912 SYSCALL32_INIT_HELPER(freebsd7_freebsd32_shmctl), 913 #endif 914 SYSCALL_INIT_LAST 915 }; 916 #endif 917 918 static int 919 shminit(void) 920 { 921 struct prison *pr; 922 void **rsv; 923 int i, error; 924 osd_method_t methods[PR_MAXMETHOD] = { 925 [PR_METHOD_CHECK] = shm_prison_check, 926 [PR_METHOD_SET] = shm_prison_set, 927 [PR_METHOD_GET] = shm_prison_get, 928 [PR_METHOD_REMOVE] = shm_prison_remove, 929 }; 930 931 #ifndef BURN_BRIDGES 932 if (TUNABLE_ULONG_FETCH("kern.ipc.shmmaxpgs", &shminfo.shmall) != 0) 933 printf("kern.ipc.shmmaxpgs is now called kern.ipc.shmall!\n"); 934 #endif 935 if (shminfo.shmmax == SHMMAX) { 936 /* Initialize shmmax dealing with possible overflow. */ 937 for (i = PAGE_SIZE; i != 0; i--) { 938 shminfo.shmmax = shminfo.shmall * i; 939 if ((shminfo.shmmax / shminfo.shmall) == (u_long)i) 940 break; 941 } 942 } 943 shmalloced = shminfo.shmmni; 944 shmsegs = malloc(shmalloced * sizeof(shmsegs[0]), M_SHM, M_WAITOK); 945 for (i = 0; i < shmalloced; i++) { 946 shmsegs[i].u.shm_perm.mode = SHMSEG_FREE; 947 shmsegs[i].u.shm_perm.seq = 0; 948 #ifdef MAC 949 mac_sysvshm_init(&shmsegs[i]); 950 #endif 951 } 952 shm_last_free = 0; 953 shm_nused = 0; 954 shm_committed = 0; 955 sx_init(&sysvshmsx, "sysvshmsx"); 956 shmexit_hook = &shmexit_myhook; 957 shmfork_hook = &shmfork_myhook; 958 959 /* Set current prisons according to their allow.sysvipc. */ 960 shm_prison_slot = osd_jail_register(NULL, methods); 961 rsv = osd_reserve(shm_prison_slot); 962 prison_lock(&prison0); 963 (void)osd_jail_set_reserved(&prison0, shm_prison_slot, rsv, &prison0); 964 prison_unlock(&prison0); 965 rsv = NULL; 966 sx_slock(&allprison_lock); 967 TAILQ_FOREACH(pr, &allprison, pr_list) { 968 if (rsv == NULL) 969 rsv = osd_reserve(shm_prison_slot); 970 prison_lock(pr); 971 if ((pr->pr_allow & PR_ALLOW_SYSVIPC) && pr->pr_ref > 0) { 972 (void)osd_jail_set_reserved(pr, shm_prison_slot, rsv, 973 &prison0); 974 rsv = NULL; 975 } 976 prison_unlock(pr); 977 } 978 if (rsv != NULL) 979 osd_free_reserved(rsv); 980 sx_sunlock(&allprison_lock); 981 982 error = syscall_helper_register(shm_syscalls, SY_THR_STATIC_KLD); 983 if (error != 0) 984 return (error); 985 #ifdef COMPAT_FREEBSD32 986 error = syscall32_helper_register(shm32_syscalls, SY_THR_STATIC_KLD); 987 if (error != 0) 988 return (error); 989 #endif 990 return (0); 991 } 992 993 static int 994 shmunload(void) 995 { 996 int i; 997 998 if (shm_nused > 0) 999 return (EBUSY); 1000 1001 #ifdef COMPAT_FREEBSD32 1002 syscall32_helper_unregister(shm32_syscalls); 1003 #endif 1004 syscall_helper_unregister(shm_syscalls); 1005 if (shm_prison_slot != 0) 1006 osd_jail_deregister(shm_prison_slot); 1007 1008 for (i = 0; i < shmalloced; i++) { 1009 #ifdef MAC 1010 mac_sysvshm_destroy(&shmsegs[i]); 1011 #endif 1012 /* 1013 * Objects might be still mapped into the processes 1014 * address spaces. Actual free would happen on the 1015 * last mapping destruction. 1016 */ 1017 if (shmsegs[i].u.shm_perm.mode != SHMSEG_FREE) 1018 vm_object_deallocate(shmsegs[i].object); 1019 } 1020 free(shmsegs, M_SHM); 1021 shmexit_hook = NULL; 1022 shmfork_hook = NULL; 1023 sx_destroy(&sysvshmsx); 1024 return (0); 1025 } 1026 1027 static int 1028 sysctl_shmsegs(SYSCTL_HANDLER_ARGS) 1029 { 1030 struct shmid_kernel tshmseg; 1031 struct prison *pr, *rpr; 1032 int error, i; 1033 1034 SYSVSHM_LOCK(); 1035 pr = req->td->td_ucred->cr_prison; 1036 rpr = shm_find_prison(req->td->td_ucred); 1037 error = 0; 1038 for (i = 0; i < shmalloced; i++) { 1039 if ((shmsegs[i].u.shm_perm.mode & SHMSEG_ALLOCATED) == 0 || 1040 rpr == NULL || shm_prison_cansee(rpr, &shmsegs[i]) != 0) { 1041 bzero(&tshmseg, sizeof(tshmseg)); 1042 tshmseg.u.shm_perm.mode = SHMSEG_FREE; 1043 } else { 1044 tshmseg = shmsegs[i]; 1045 if (tshmseg.cred->cr_prison != pr) 1046 tshmseg.u.shm_perm.key = IPC_PRIVATE; 1047 } 1048 error = SYSCTL_OUT(req, &tshmseg, sizeof(tshmseg)); 1049 if (error != 0) 1050 break; 1051 } 1052 SYSVSHM_UNLOCK(); 1053 return (error); 1054 } 1055 1056 static int 1057 shm_prison_check(void *obj, void *data) 1058 { 1059 struct prison *pr = obj; 1060 struct prison *prpr; 1061 struct vfsoptlist *opts = data; 1062 int error, jsys; 1063 1064 /* 1065 * sysvshm is a jailsys integer. 1066 * It must be "disable" if the parent jail is disabled. 1067 */ 1068 error = vfs_copyopt(opts, "sysvshm", &jsys, sizeof(jsys)); 1069 if (error != ENOENT) { 1070 if (error != 0) 1071 return (error); 1072 switch (jsys) { 1073 case JAIL_SYS_DISABLE: 1074 break; 1075 case JAIL_SYS_NEW: 1076 case JAIL_SYS_INHERIT: 1077 prison_lock(pr->pr_parent); 1078 prpr = osd_jail_get(pr->pr_parent, shm_prison_slot); 1079 prison_unlock(pr->pr_parent); 1080 if (prpr == NULL) 1081 return (EPERM); 1082 break; 1083 default: 1084 return (EINVAL); 1085 } 1086 } 1087 1088 return (0); 1089 } 1090 1091 static int 1092 shm_prison_set(void *obj, void *data) 1093 { 1094 struct prison *pr = obj; 1095 struct prison *tpr, *orpr, *nrpr, *trpr; 1096 struct vfsoptlist *opts = data; 1097 void *rsv; 1098 int jsys, descend; 1099 1100 /* 1101 * sysvshm controls which jail is the root of the associated segments 1102 * (this jail or same as the parent), or if the feature is available 1103 * at all. 1104 */ 1105 if (vfs_copyopt(opts, "sysvshm", &jsys, sizeof(jsys)) == ENOENT) 1106 jsys = vfs_flagopt(opts, "allow.sysvipc", NULL, 0) 1107 ? JAIL_SYS_INHERIT 1108 : vfs_flagopt(opts, "allow.nosysvipc", NULL, 0) 1109 ? JAIL_SYS_DISABLE 1110 : -1; 1111 if (jsys == JAIL_SYS_DISABLE) { 1112 prison_lock(pr); 1113 orpr = osd_jail_get(pr, shm_prison_slot); 1114 if (orpr != NULL) 1115 osd_jail_del(pr, shm_prison_slot); 1116 prison_unlock(pr); 1117 if (orpr != NULL) { 1118 if (orpr == pr) 1119 shm_prison_cleanup(pr); 1120 /* Disable all child jails as well. */ 1121 FOREACH_PRISON_DESCENDANT(pr, tpr, descend) { 1122 prison_lock(tpr); 1123 trpr = osd_jail_get(tpr, shm_prison_slot); 1124 if (trpr != NULL) { 1125 osd_jail_del(tpr, shm_prison_slot); 1126 prison_unlock(tpr); 1127 if (trpr == tpr) 1128 shm_prison_cleanup(tpr); 1129 } else { 1130 prison_unlock(tpr); 1131 descend = 0; 1132 } 1133 } 1134 } 1135 } else if (jsys != -1) { 1136 if (jsys == JAIL_SYS_NEW) 1137 nrpr = pr; 1138 else { 1139 prison_lock(pr->pr_parent); 1140 nrpr = osd_jail_get(pr->pr_parent, shm_prison_slot); 1141 prison_unlock(pr->pr_parent); 1142 } 1143 rsv = osd_reserve(shm_prison_slot); 1144 prison_lock(pr); 1145 orpr = osd_jail_get(pr, shm_prison_slot); 1146 if (orpr != nrpr) 1147 (void)osd_jail_set_reserved(pr, shm_prison_slot, rsv, 1148 nrpr); 1149 else 1150 osd_free_reserved(rsv); 1151 prison_unlock(pr); 1152 if (orpr != nrpr) { 1153 if (orpr == pr) 1154 shm_prison_cleanup(pr); 1155 if (orpr != NULL) { 1156 /* Change child jails matching the old root, */ 1157 FOREACH_PRISON_DESCENDANT(pr, tpr, descend) { 1158 prison_lock(tpr); 1159 trpr = osd_jail_get(tpr, 1160 shm_prison_slot); 1161 if (trpr == orpr) { 1162 (void)osd_jail_set(tpr, 1163 shm_prison_slot, nrpr); 1164 prison_unlock(tpr); 1165 if (trpr == tpr) 1166 shm_prison_cleanup(tpr); 1167 } else { 1168 prison_unlock(tpr); 1169 descend = 0; 1170 } 1171 } 1172 } 1173 } 1174 } 1175 1176 return (0); 1177 } 1178 1179 static int 1180 shm_prison_get(void *obj, void *data) 1181 { 1182 struct prison *pr = obj; 1183 struct prison *rpr; 1184 struct vfsoptlist *opts = data; 1185 int error, jsys; 1186 1187 /* Set sysvshm based on the jail's root prison. */ 1188 prison_lock(pr); 1189 rpr = osd_jail_get(pr, shm_prison_slot); 1190 prison_unlock(pr); 1191 jsys = rpr == NULL ? JAIL_SYS_DISABLE 1192 : rpr == pr ? JAIL_SYS_NEW : JAIL_SYS_INHERIT; 1193 error = vfs_setopt(opts, "sysvshm", &jsys, sizeof(jsys)); 1194 if (error == ENOENT) 1195 error = 0; 1196 return (error); 1197 } 1198 1199 static int 1200 shm_prison_remove(void *obj, void *data __unused) 1201 { 1202 struct prison *pr = obj; 1203 struct prison *rpr; 1204 1205 SYSVSHM_LOCK(); 1206 prison_lock(pr); 1207 rpr = osd_jail_get(pr, shm_prison_slot); 1208 prison_unlock(pr); 1209 if (rpr == pr) 1210 shm_prison_cleanup(pr); 1211 SYSVSHM_UNLOCK(); 1212 return (0); 1213 } 1214 1215 static void 1216 shm_prison_cleanup(struct prison *pr) 1217 { 1218 struct shmid_kernel *shmseg; 1219 int i; 1220 1221 /* Remove any segments that belong to this jail. */ 1222 for (i = 0; i < shmalloced; i++) { 1223 shmseg = &shmsegs[i]; 1224 if ((shmseg->u.shm_perm.mode & SHMSEG_ALLOCATED) && 1225 shmseg->cred != NULL && shmseg->cred->cr_prison == pr) { 1226 shm_remove(shmseg, i); 1227 } 1228 } 1229 } 1230 1231 SYSCTL_JAIL_PARAM_SYS_NODE(sysvshm, CTLFLAG_RW, "SYSV shared memory"); 1232 1233 #if defined(__i386__) && (defined(COMPAT_FREEBSD4) || defined(COMPAT_43)) 1234 struct oshmid_ds { 1235 struct ipc_perm_old shm_perm; /* operation perms */ 1236 int shm_segsz; /* size of segment (bytes) */ 1237 u_short shm_cpid; /* pid, creator */ 1238 u_short shm_lpid; /* pid, last operation */ 1239 short shm_nattch; /* no. of current attaches */ 1240 time_t shm_atime; /* last attach time */ 1241 time_t shm_dtime; /* last detach time */ 1242 time_t shm_ctime; /* last change time */ 1243 void *shm_handle; /* internal handle for shm segment */ 1244 }; 1245 1246 struct oshmctl_args { 1247 int shmid; 1248 int cmd; 1249 struct oshmid_ds *ubuf; 1250 }; 1251 1252 static int 1253 oshmctl(struct thread *td, struct oshmctl_args *uap) 1254 { 1255 #ifdef COMPAT_43 1256 int error = 0; 1257 struct prison *rpr; 1258 struct shmid_kernel *shmseg; 1259 struct oshmid_ds outbuf; 1260 1261 rpr = shm_find_prison(td->td_ucred); 1262 if (rpr == NULL) 1263 return (ENOSYS); 1264 if (uap->cmd != IPC_STAT) { 1265 return (freebsd7_shmctl(td, 1266 (struct freebsd7_shmctl_args *)uap)); 1267 } 1268 SYSVSHM_LOCK(); 1269 shmseg = shm_find_segment(rpr, uap->shmid, true); 1270 if (shmseg == NULL) { 1271 SYSVSHM_UNLOCK(); 1272 return (EINVAL); 1273 } 1274 error = ipcperm(td, &shmseg->u.shm_perm, IPC_R); 1275 if (error != 0) { 1276 SYSVSHM_UNLOCK(); 1277 return (error); 1278 } 1279 #ifdef MAC 1280 error = mac_sysvshm_check_shmctl(td->td_ucred, shmseg, uap->cmd); 1281 if (error != 0) { 1282 SYSVSHM_UNLOCK(); 1283 return (error); 1284 } 1285 #endif 1286 ipcperm_new2old(&shmseg->u.shm_perm, &outbuf.shm_perm); 1287 outbuf.shm_segsz = shmseg->u.shm_segsz; 1288 outbuf.shm_cpid = shmseg->u.shm_cpid; 1289 outbuf.shm_lpid = shmseg->u.shm_lpid; 1290 outbuf.shm_nattch = shmseg->u.shm_nattch; 1291 outbuf.shm_atime = shmseg->u.shm_atime; 1292 outbuf.shm_dtime = shmseg->u.shm_dtime; 1293 outbuf.shm_ctime = shmseg->u.shm_ctime; 1294 outbuf.shm_handle = shmseg->object; 1295 SYSVSHM_UNLOCK(); 1296 return (copyout(&outbuf, uap->ubuf, sizeof(outbuf))); 1297 #else 1298 return (EINVAL); 1299 #endif 1300 } 1301 1302 /* XXX casting to (sy_call_t *) is bogus, as usual. */ 1303 static sy_call_t *shmcalls[] = { 1304 (sy_call_t *)sys_shmat, (sy_call_t *)oshmctl, 1305 (sy_call_t *)sys_shmdt, (sy_call_t *)sys_shmget, 1306 (sy_call_t *)freebsd7_shmctl 1307 }; 1308 1309 #ifndef _SYS_SYSPROTO_H_ 1310 /* XXX actually varargs. */ 1311 struct shmsys_args { 1312 int which; 1313 int a2; 1314 int a3; 1315 int a4; 1316 }; 1317 #endif 1318 int 1319 sys_shmsys(struct thread *td, struct shmsys_args *uap) 1320 { 1321 1322 AUDIT_ARG_SVIPC_WHICH(uap->which); 1323 if (uap->which < 0 || uap->which >= nitems(shmcalls)) 1324 return (EINVAL); 1325 return ((*shmcalls[uap->which])(td, &uap->a2)); 1326 } 1327 1328 #endif /* i386 && (COMPAT_FREEBSD4 || COMPAT_43) */ 1329 1330 #ifdef COMPAT_FREEBSD32 1331 1332 int 1333 freebsd32_shmsys(struct thread *td, struct freebsd32_shmsys_args *uap) 1334 { 1335 1336 #if defined(COMPAT_FREEBSD4) || defined(COMPAT_FREEBSD5) || \ 1337 defined(COMPAT_FREEBSD6) || defined(COMPAT_FREEBSD7) 1338 AUDIT_ARG_SVIPC_WHICH(uap->which); 1339 switch (uap->which) { 1340 case 0: { /* shmat */ 1341 struct shmat_args ap; 1342 1343 ap.shmid = uap->a2; 1344 ap.shmaddr = PTRIN(uap->a3); 1345 ap.shmflg = uap->a4; 1346 return (sysent[SYS_shmat].sy_call(td, &ap)); 1347 } 1348 case 2: { /* shmdt */ 1349 struct shmdt_args ap; 1350 1351 ap.shmaddr = PTRIN(uap->a2); 1352 return (sysent[SYS_shmdt].sy_call(td, &ap)); 1353 } 1354 case 3: { /* shmget */ 1355 struct shmget_args ap; 1356 1357 ap.key = uap->a2; 1358 ap.size = uap->a3; 1359 ap.shmflg = uap->a4; 1360 return (sysent[SYS_shmget].sy_call(td, &ap)); 1361 } 1362 case 4: { /* shmctl */ 1363 struct freebsd7_freebsd32_shmctl_args ap; 1364 1365 ap.shmid = uap->a2; 1366 ap.cmd = uap->a3; 1367 ap.buf = PTRIN(uap->a4); 1368 return (freebsd7_freebsd32_shmctl(td, &ap)); 1369 } 1370 case 1: /* oshmctl */ 1371 default: 1372 return (EINVAL); 1373 } 1374 #else 1375 return (nosys(td, NULL)); 1376 #endif 1377 } 1378 1379 #if defined(COMPAT_FREEBSD4) || defined(COMPAT_FREEBSD5) || \ 1380 defined(COMPAT_FREEBSD6) || defined(COMPAT_FREEBSD7) 1381 int 1382 freebsd7_freebsd32_shmctl(struct thread *td, 1383 struct freebsd7_freebsd32_shmctl_args *uap) 1384 { 1385 int error; 1386 union { 1387 struct shmid_ds shmid_ds; 1388 struct shm_info shm_info; 1389 struct shminfo shminfo; 1390 } u; 1391 union { 1392 struct shmid_ds32_old shmid_ds32; 1393 struct shm_info32 shm_info32; 1394 struct shminfo32 shminfo32; 1395 } u32; 1396 size_t sz; 1397 1398 if (uap->cmd == IPC_SET) { 1399 if ((error = copyin(uap->buf, &u32.shmid_ds32, 1400 sizeof(u32.shmid_ds32)))) 1401 goto done; 1402 freebsd32_ipcperm_old_in(&u32.shmid_ds32.shm_perm, 1403 &u.shmid_ds.shm_perm); 1404 CP(u32.shmid_ds32, u.shmid_ds, shm_segsz); 1405 CP(u32.shmid_ds32, u.shmid_ds, shm_lpid); 1406 CP(u32.shmid_ds32, u.shmid_ds, shm_cpid); 1407 CP(u32.shmid_ds32, u.shmid_ds, shm_nattch); 1408 CP(u32.shmid_ds32, u.shmid_ds, shm_atime); 1409 CP(u32.shmid_ds32, u.shmid_ds, shm_dtime); 1410 CP(u32.shmid_ds32, u.shmid_ds, shm_ctime); 1411 } 1412 1413 error = kern_shmctl(td, uap->shmid, uap->cmd, (void *)&u, &sz); 1414 if (error) 1415 goto done; 1416 1417 /* Cases in which we need to copyout */ 1418 switch (uap->cmd) { 1419 case IPC_INFO: 1420 CP(u.shminfo, u32.shminfo32, shmmax); 1421 CP(u.shminfo, u32.shminfo32, shmmin); 1422 CP(u.shminfo, u32.shminfo32, shmmni); 1423 CP(u.shminfo, u32.shminfo32, shmseg); 1424 CP(u.shminfo, u32.shminfo32, shmall); 1425 error = copyout(&u32.shminfo32, uap->buf, 1426 sizeof(u32.shminfo32)); 1427 break; 1428 case SHM_INFO: 1429 CP(u.shm_info, u32.shm_info32, used_ids); 1430 CP(u.shm_info, u32.shm_info32, shm_rss); 1431 CP(u.shm_info, u32.shm_info32, shm_tot); 1432 CP(u.shm_info, u32.shm_info32, shm_swp); 1433 CP(u.shm_info, u32.shm_info32, swap_attempts); 1434 CP(u.shm_info, u32.shm_info32, swap_successes); 1435 error = copyout(&u32.shm_info32, uap->buf, 1436 sizeof(u32.shm_info32)); 1437 break; 1438 case SHM_STAT: 1439 case IPC_STAT: 1440 freebsd32_ipcperm_old_out(&u.shmid_ds.shm_perm, 1441 &u32.shmid_ds32.shm_perm); 1442 if (u.shmid_ds.shm_segsz > INT32_MAX) 1443 u32.shmid_ds32.shm_segsz = INT32_MAX; 1444 else 1445 CP(u.shmid_ds, u32.shmid_ds32, shm_segsz); 1446 CP(u.shmid_ds, u32.shmid_ds32, shm_lpid); 1447 CP(u.shmid_ds, u32.shmid_ds32, shm_cpid); 1448 CP(u.shmid_ds, u32.shmid_ds32, shm_nattch); 1449 CP(u.shmid_ds, u32.shmid_ds32, shm_atime); 1450 CP(u.shmid_ds, u32.shmid_ds32, shm_dtime); 1451 CP(u.shmid_ds, u32.shmid_ds32, shm_ctime); 1452 u32.shmid_ds32.shm_internal = 0; 1453 error = copyout(&u32.shmid_ds32, uap->buf, 1454 sizeof(u32.shmid_ds32)); 1455 break; 1456 } 1457 1458 done: 1459 if (error) { 1460 /* Invalidate the return value */ 1461 td->td_retval[0] = -1; 1462 } 1463 return (error); 1464 } 1465 #endif 1466 1467 int 1468 freebsd32_shmctl(struct thread *td, struct freebsd32_shmctl_args *uap) 1469 { 1470 int error; 1471 union { 1472 struct shmid_ds shmid_ds; 1473 struct shm_info shm_info; 1474 struct shminfo shminfo; 1475 } u; 1476 union { 1477 struct shmid_ds32 shmid_ds32; 1478 struct shm_info32 shm_info32; 1479 struct shminfo32 shminfo32; 1480 } u32; 1481 size_t sz; 1482 1483 if (uap->cmd == IPC_SET) { 1484 if ((error = copyin(uap->buf, &u32.shmid_ds32, 1485 sizeof(u32.shmid_ds32)))) 1486 goto done; 1487 freebsd32_ipcperm_in(&u32.shmid_ds32.shm_perm, 1488 &u.shmid_ds.shm_perm); 1489 CP(u32.shmid_ds32, u.shmid_ds, shm_segsz); 1490 CP(u32.shmid_ds32, u.shmid_ds, shm_lpid); 1491 CP(u32.shmid_ds32, u.shmid_ds, shm_cpid); 1492 CP(u32.shmid_ds32, u.shmid_ds, shm_nattch); 1493 CP(u32.shmid_ds32, u.shmid_ds, shm_atime); 1494 CP(u32.shmid_ds32, u.shmid_ds, shm_dtime); 1495 CP(u32.shmid_ds32, u.shmid_ds, shm_ctime); 1496 } 1497 1498 error = kern_shmctl(td, uap->shmid, uap->cmd, (void *)&u, &sz); 1499 if (error) 1500 goto done; 1501 1502 /* Cases in which we need to copyout */ 1503 switch (uap->cmd) { 1504 case IPC_INFO: 1505 CP(u.shminfo, u32.shminfo32, shmmax); 1506 CP(u.shminfo, u32.shminfo32, shmmin); 1507 CP(u.shminfo, u32.shminfo32, shmmni); 1508 CP(u.shminfo, u32.shminfo32, shmseg); 1509 CP(u.shminfo, u32.shminfo32, shmall); 1510 error = copyout(&u32.shminfo32, uap->buf, 1511 sizeof(u32.shminfo32)); 1512 break; 1513 case SHM_INFO: 1514 CP(u.shm_info, u32.shm_info32, used_ids); 1515 CP(u.shm_info, u32.shm_info32, shm_rss); 1516 CP(u.shm_info, u32.shm_info32, shm_tot); 1517 CP(u.shm_info, u32.shm_info32, shm_swp); 1518 CP(u.shm_info, u32.shm_info32, swap_attempts); 1519 CP(u.shm_info, u32.shm_info32, swap_successes); 1520 error = copyout(&u32.shm_info32, uap->buf, 1521 sizeof(u32.shm_info32)); 1522 break; 1523 case SHM_STAT: 1524 case IPC_STAT: 1525 freebsd32_ipcperm_out(&u.shmid_ds.shm_perm, 1526 &u32.shmid_ds32.shm_perm); 1527 if (u.shmid_ds.shm_segsz > INT32_MAX) 1528 u32.shmid_ds32.shm_segsz = INT32_MAX; 1529 else 1530 CP(u.shmid_ds, u32.shmid_ds32, shm_segsz); 1531 CP(u.shmid_ds, u32.shmid_ds32, shm_lpid); 1532 CP(u.shmid_ds, u32.shmid_ds32, shm_cpid); 1533 CP(u.shmid_ds, u32.shmid_ds32, shm_nattch); 1534 CP(u.shmid_ds, u32.shmid_ds32, shm_atime); 1535 CP(u.shmid_ds, u32.shmid_ds32, shm_dtime); 1536 CP(u.shmid_ds, u32.shmid_ds32, shm_ctime); 1537 error = copyout(&u32.shmid_ds32, uap->buf, 1538 sizeof(u32.shmid_ds32)); 1539 break; 1540 } 1541 1542 done: 1543 if (error) { 1544 /* Invalidate the return value */ 1545 td->td_retval[0] = -1; 1546 } 1547 return (error); 1548 } 1549 #endif 1550 1551 #if defined(COMPAT_FREEBSD4) || defined(COMPAT_FREEBSD5) || \ 1552 defined(COMPAT_FREEBSD6) || defined(COMPAT_FREEBSD7) 1553 1554 #ifndef CP 1555 #define CP(src, dst, fld) do { (dst).fld = (src).fld; } while (0) 1556 #endif 1557 1558 #ifndef _SYS_SYSPROTO_H_ 1559 struct freebsd7_shmctl_args { 1560 int shmid; 1561 int cmd; 1562 struct shmid_ds_old *buf; 1563 }; 1564 #endif 1565 int 1566 freebsd7_shmctl(struct thread *td, struct freebsd7_shmctl_args *uap) 1567 { 1568 int error; 1569 struct shmid_ds_old old; 1570 struct shmid_ds buf; 1571 size_t bufsz; 1572 1573 /* 1574 * The only reason IPC_INFO, SHM_INFO, SHM_STAT exists is to support 1575 * Linux binaries. If we see the call come through the FreeBSD ABI, 1576 * return an error back to the user since we do not to support this. 1577 */ 1578 if (uap->cmd == IPC_INFO || uap->cmd == SHM_INFO || 1579 uap->cmd == SHM_STAT) 1580 return (EINVAL); 1581 1582 /* IPC_SET needs to copyin the buffer before calling kern_shmctl */ 1583 if (uap->cmd == IPC_SET) { 1584 if ((error = copyin(uap->buf, &old, sizeof(old)))) 1585 goto done; 1586 ipcperm_old2new(&old.shm_perm, &buf.shm_perm); 1587 CP(old, buf, shm_segsz); 1588 CP(old, buf, shm_lpid); 1589 CP(old, buf, shm_cpid); 1590 CP(old, buf, shm_nattch); 1591 CP(old, buf, shm_atime); 1592 CP(old, buf, shm_dtime); 1593 CP(old, buf, shm_ctime); 1594 } 1595 1596 error = kern_shmctl(td, uap->shmid, uap->cmd, (void *)&buf, &bufsz); 1597 if (error) 1598 goto done; 1599 1600 /* Cases in which we need to copyout */ 1601 switch (uap->cmd) { 1602 case IPC_STAT: 1603 ipcperm_new2old(&buf.shm_perm, &old.shm_perm); 1604 if (buf.shm_segsz > INT_MAX) 1605 old.shm_segsz = INT_MAX; 1606 else 1607 CP(buf, old, shm_segsz); 1608 CP(buf, old, shm_lpid); 1609 CP(buf, old, shm_cpid); 1610 if (buf.shm_nattch > SHRT_MAX) 1611 old.shm_nattch = SHRT_MAX; 1612 else 1613 CP(buf, old, shm_nattch); 1614 CP(buf, old, shm_atime); 1615 CP(buf, old, shm_dtime); 1616 CP(buf, old, shm_ctime); 1617 old.shm_internal = NULL; 1618 error = copyout(&old, uap->buf, sizeof(old)); 1619 break; 1620 } 1621 1622 done: 1623 if (error) { 1624 /* Invalidate the return value */ 1625 td->td_retval[0] = -1; 1626 } 1627 return (error); 1628 } 1629 1630 #endif /* COMPAT_FREEBSD4 || COMPAT_FREEBSD5 || COMPAT_FREEBSD6 || 1631 COMPAT_FREEBSD7 */ 1632 1633 static int 1634 sysvshm_modload(struct module *module, int cmd, void *arg) 1635 { 1636 int error = 0; 1637 1638 switch (cmd) { 1639 case MOD_LOAD: 1640 error = shminit(); 1641 if (error != 0) 1642 shmunload(); 1643 break; 1644 case MOD_UNLOAD: 1645 error = shmunload(); 1646 break; 1647 case MOD_SHUTDOWN: 1648 break; 1649 default: 1650 error = EINVAL; 1651 break; 1652 } 1653 return (error); 1654 } 1655 1656 static moduledata_t sysvshm_mod = { 1657 "sysvshm", 1658 &sysvshm_modload, 1659 NULL 1660 }; 1661 1662 DECLARE_MODULE(sysvshm, sysvshm_mod, SI_SUB_SYSV_SHM, SI_ORDER_FIRST); 1663 MODULE_VERSION(sysvshm, 1); 1664