1 /*- 2 * SPDX-License-Identifier: BSD-4-Clause AND BSD-2-Clause-FreeBSD 3 * 4 * Copyright (c) 1994 Adam Glass and Charles Hannum. All rights reserved. 5 * 6 * Redistribution and use in source and binary forms, with or without 7 * modification, are permitted provided that the following conditions 8 * are met: 9 * 1. Redistributions of source code must retain the above copyright 10 * notice, this list of conditions and the following disclaimer. 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. All advertising materials mentioning features or use of this software 15 * must display the following acknowledgement: 16 * This product includes software developed by Adam Glass and Charles 17 * Hannum. 18 * 4. The names of the authors may not be used to endorse or promote products 19 * derived from this software without specific prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE AUTHORS ``AS IS'' AND ANY EXPRESS OR 22 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 23 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 24 * IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY DIRECT, INDIRECT, 25 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 26 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 27 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 28 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 29 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 30 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 31 * 32 * $NetBSD: sysv_shm.c,v 1.39 1997/10/07 10:02:03 drochner Exp $ 33 */ 34 /*- 35 * Copyright (c) 2003-2005 McAfee, Inc. 36 * Copyright (c) 2016-2017 Robert N. M. Watson 37 * All rights reserved. 38 * 39 * This software was developed for the FreeBSD Project in part by McAfee 40 * Research, the Security Research Division of McAfee, Inc under DARPA/SPAWAR 41 * contract N66001-01-C-8035 ("CBOSS"), as part of the DARPA CHATS research 42 * program. 43 * 44 * Portions of this software were developed by BAE Systems, the University of 45 * Cambridge Computer Laboratory, and Memorial University under DARPA/AFRL 46 * contract FA8650-15-C-7558 ("CADETS"), as part of the DARPA Transparent 47 * Computing (TC) research program. 48 * 49 * Redistribution and use in source and binary forms, with or without 50 * modification, are permitted provided that the following conditions 51 * are met: 52 * 1. Redistributions of source code must retain the above copyright 53 * notice, this list of conditions and the following disclaimer. 54 * 2. Redistributions in binary form must reproduce the above copyright 55 * notice, this list of conditions and the following disclaimer in the 56 * documentation and/or other materials provided with the distribution. 57 * 58 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 59 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 60 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 61 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 62 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 63 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 64 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 65 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 66 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 67 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 68 * SUCH DAMAGE. 69 */ 70 71 #include <sys/cdefs.h> 72 __FBSDID("$FreeBSD$"); 73 74 #include "opt_compat.h" 75 #include "opt_sysvipc.h" 76 77 #include <sys/param.h> 78 #include <sys/systm.h> 79 #include <sys/kernel.h> 80 #include <sys/limits.h> 81 #include <sys/lock.h> 82 #include <sys/sysctl.h> 83 #include <sys/shm.h> 84 #include <sys/proc.h> 85 #include <sys/malloc.h> 86 #include <sys/mman.h> 87 #include <sys/module.h> 88 #include <sys/mutex.h> 89 #include <sys/racct.h> 90 #include <sys/resourcevar.h> 91 #include <sys/rwlock.h> 92 #include <sys/stat.h> 93 #include <sys/syscall.h> 94 #include <sys/syscallsubr.h> 95 #include <sys/sysent.h> 96 #include <sys/sysproto.h> 97 #include <sys/jail.h> 98 99 #include <security/audit/audit.h> 100 #include <security/mac/mac_framework.h> 101 102 #include <vm/vm.h> 103 #include <vm/vm_param.h> 104 #include <vm/pmap.h> 105 #include <vm/vm_object.h> 106 #include <vm/vm_map.h> 107 #include <vm/vm_page.h> 108 #include <vm/vm_pager.h> 109 110 FEATURE(sysv_shm, "System V shared memory segments support"); 111 112 static MALLOC_DEFINE(M_SHM, "shm", "SVID compatible shared memory segments"); 113 114 static int shmget_allocate_segment(struct thread *td, 115 struct shmget_args *uap, int mode); 116 static int shmget_existing(struct thread *td, struct shmget_args *uap, 117 int mode, int segnum); 118 119 #define SHMSEG_FREE 0x0200 120 #define SHMSEG_REMOVED 0x0400 121 #define SHMSEG_ALLOCATED 0x0800 122 123 static int shm_last_free, shm_nused, shmalloced; 124 vm_size_t shm_committed; 125 static struct shmid_kernel *shmsegs; 126 static unsigned shm_prison_slot; 127 128 struct shmmap_state { 129 vm_offset_t va; 130 int shmid; 131 }; 132 133 static void shm_deallocate_segment(struct shmid_kernel *); 134 static int shm_find_segment_by_key(struct prison *, key_t); 135 static struct shmid_kernel *shm_find_segment(struct prison *, int, bool); 136 static int shm_delete_mapping(struct vmspace *vm, struct shmmap_state *); 137 static void shmrealloc(void); 138 static int shminit(void); 139 static int sysvshm_modload(struct module *, int, void *); 140 static int shmunload(void); 141 static void shmexit_myhook(struct vmspace *vm); 142 static void shmfork_myhook(struct proc *p1, struct proc *p2); 143 static int sysctl_shmsegs(SYSCTL_HANDLER_ARGS); 144 static void shm_remove(struct shmid_kernel *, int); 145 static struct prison *shm_find_prison(struct ucred *); 146 static int shm_prison_cansee(struct prison *, struct shmid_kernel *); 147 static int shm_prison_check(void *, void *); 148 static int shm_prison_set(void *, void *); 149 static int shm_prison_get(void *, void *); 150 static int shm_prison_remove(void *, void *); 151 static void shm_prison_cleanup(struct prison *); 152 153 /* 154 * Tuneable values. 155 */ 156 #ifndef SHMMAXPGS 157 #define SHMMAXPGS 131072 /* Note: sysv shared memory is swap backed. */ 158 #endif 159 #ifndef SHMMAX 160 #define SHMMAX (SHMMAXPGS*PAGE_SIZE) 161 #endif 162 #ifndef SHMMIN 163 #define SHMMIN 1 164 #endif 165 #ifndef SHMMNI 166 #define SHMMNI 192 167 #endif 168 #ifndef SHMSEG 169 #define SHMSEG 128 170 #endif 171 #ifndef SHMALL 172 #define SHMALL (SHMMAXPGS) 173 #endif 174 175 struct shminfo shminfo = { 176 .shmmax = SHMMAX, 177 .shmmin = SHMMIN, 178 .shmmni = SHMMNI, 179 .shmseg = SHMSEG, 180 .shmall = SHMALL 181 }; 182 183 static int shm_use_phys; 184 static int shm_allow_removed = 1; 185 186 SYSCTL_ULONG(_kern_ipc, OID_AUTO, shmmax, CTLFLAG_RWTUN, &shminfo.shmmax, 0, 187 "Maximum shared memory segment size"); 188 SYSCTL_ULONG(_kern_ipc, OID_AUTO, shmmin, CTLFLAG_RWTUN, &shminfo.shmmin, 0, 189 "Minimum shared memory segment size"); 190 SYSCTL_ULONG(_kern_ipc, OID_AUTO, shmmni, CTLFLAG_RDTUN, &shminfo.shmmni, 0, 191 "Number of shared memory identifiers"); 192 SYSCTL_ULONG(_kern_ipc, OID_AUTO, shmseg, CTLFLAG_RDTUN, &shminfo.shmseg, 0, 193 "Number of segments per process"); 194 SYSCTL_ULONG(_kern_ipc, OID_AUTO, shmall, CTLFLAG_RWTUN, &shminfo.shmall, 0, 195 "Maximum number of pages available for shared memory"); 196 SYSCTL_INT(_kern_ipc, OID_AUTO, shm_use_phys, CTLFLAG_RWTUN, 197 &shm_use_phys, 0, "Enable/Disable locking of shared memory pages in core"); 198 SYSCTL_INT(_kern_ipc, OID_AUTO, shm_allow_removed, CTLFLAG_RWTUN, 199 &shm_allow_removed, 0, 200 "Enable/Disable attachment to attached segments marked for removal"); 201 SYSCTL_PROC(_kern_ipc, OID_AUTO, shmsegs, CTLTYPE_OPAQUE | CTLFLAG_RD | 202 CTLFLAG_MPSAFE, NULL, 0, sysctl_shmsegs, "", 203 "Current number of shared memory segments allocated"); 204 205 static struct sx sysvshmsx; 206 #define SYSVSHM_LOCK() sx_xlock(&sysvshmsx) 207 #define SYSVSHM_UNLOCK() sx_xunlock(&sysvshmsx) 208 #define SYSVSHM_ASSERT_LOCKED() sx_assert(&sysvshmsx, SA_XLOCKED) 209 210 static int 211 shm_find_segment_by_key(struct prison *pr, key_t key) 212 { 213 int i; 214 215 for (i = 0; i < shmalloced; i++) 216 if ((shmsegs[i].u.shm_perm.mode & SHMSEG_ALLOCATED) && 217 shmsegs[i].cred != NULL && 218 shmsegs[i].cred->cr_prison == pr && 219 shmsegs[i].u.shm_perm.key == key) 220 return (i); 221 return (-1); 222 } 223 224 /* 225 * Finds segment either by shmid if is_shmid is true, or by segnum if 226 * is_shmid is false. 227 */ 228 static struct shmid_kernel * 229 shm_find_segment(struct prison *rpr, int arg, bool is_shmid) 230 { 231 struct shmid_kernel *shmseg; 232 int segnum; 233 234 segnum = is_shmid ? IPCID_TO_IX(arg) : arg; 235 if (segnum < 0 || segnum >= shmalloced) 236 return (NULL); 237 shmseg = &shmsegs[segnum]; 238 if ((shmseg->u.shm_perm.mode & SHMSEG_ALLOCATED) == 0 || 239 (!shm_allow_removed && 240 (shmseg->u.shm_perm.mode & SHMSEG_REMOVED) != 0) || 241 (is_shmid && shmseg->u.shm_perm.seq != IPCID_TO_SEQ(arg)) || 242 shm_prison_cansee(rpr, shmseg) != 0) 243 return (NULL); 244 return (shmseg); 245 } 246 247 static void 248 shm_deallocate_segment(struct shmid_kernel *shmseg) 249 { 250 vm_size_t size; 251 252 SYSVSHM_ASSERT_LOCKED(); 253 254 vm_object_deallocate(shmseg->object); 255 shmseg->object = NULL; 256 size = round_page(shmseg->u.shm_segsz); 257 shm_committed -= btoc(size); 258 shm_nused--; 259 shmseg->u.shm_perm.mode = SHMSEG_FREE; 260 #ifdef MAC 261 mac_sysvshm_cleanup(shmseg); 262 #endif 263 racct_sub_cred(shmseg->cred, RACCT_NSHM, 1); 264 racct_sub_cred(shmseg->cred, RACCT_SHMSIZE, size); 265 crfree(shmseg->cred); 266 shmseg->cred = NULL; 267 } 268 269 static int 270 shm_delete_mapping(struct vmspace *vm, struct shmmap_state *shmmap_s) 271 { 272 struct shmid_kernel *shmseg; 273 int segnum, result; 274 vm_size_t size; 275 276 SYSVSHM_ASSERT_LOCKED(); 277 segnum = IPCID_TO_IX(shmmap_s->shmid); 278 KASSERT(segnum >= 0 && segnum < shmalloced, 279 ("segnum %d shmalloced %d", segnum, shmalloced)); 280 281 shmseg = &shmsegs[segnum]; 282 size = round_page(shmseg->u.shm_segsz); 283 result = vm_map_remove(&vm->vm_map, shmmap_s->va, shmmap_s->va + size); 284 if (result != KERN_SUCCESS) 285 return (EINVAL); 286 shmmap_s->shmid = -1; 287 shmseg->u.shm_dtime = time_second; 288 if (--shmseg->u.shm_nattch == 0 && 289 (shmseg->u.shm_perm.mode & SHMSEG_REMOVED)) { 290 shm_deallocate_segment(shmseg); 291 shm_last_free = segnum; 292 } 293 return (0); 294 } 295 296 static void 297 shm_remove(struct shmid_kernel *shmseg, int segnum) 298 { 299 300 shmseg->u.shm_perm.key = IPC_PRIVATE; 301 shmseg->u.shm_perm.mode |= SHMSEG_REMOVED; 302 if (shmseg->u.shm_nattch == 0) { 303 shm_deallocate_segment(shmseg); 304 shm_last_free = segnum; 305 } 306 } 307 308 static struct prison * 309 shm_find_prison(struct ucred *cred) 310 { 311 struct prison *pr, *rpr; 312 313 pr = cred->cr_prison; 314 prison_lock(pr); 315 rpr = osd_jail_get(pr, shm_prison_slot); 316 prison_unlock(pr); 317 return rpr; 318 } 319 320 static int 321 shm_prison_cansee(struct prison *rpr, struct shmid_kernel *shmseg) 322 { 323 324 if (shmseg->cred == NULL || 325 !(rpr == shmseg->cred->cr_prison || 326 prison_ischild(rpr, shmseg->cred->cr_prison))) 327 return (EINVAL); 328 return (0); 329 } 330 331 static int 332 kern_shmdt_locked(struct thread *td, const void *shmaddr) 333 { 334 struct proc *p = td->td_proc; 335 struct shmmap_state *shmmap_s; 336 #if defined(AUDIT) || defined(MAC) 337 struct shmid_kernel *shmsegptr; 338 #endif 339 #ifdef MAC 340 int error; 341 #endif 342 int i; 343 344 SYSVSHM_ASSERT_LOCKED(); 345 if (shm_find_prison(td->td_ucred) == NULL) 346 return (ENOSYS); 347 shmmap_s = p->p_vmspace->vm_shm; 348 if (shmmap_s == NULL) 349 return (EINVAL); 350 AUDIT_ARG_SVIPC_ID(shmmap_s->shmid); 351 for (i = 0; i < shminfo.shmseg; i++, shmmap_s++) { 352 if (shmmap_s->shmid != -1 && 353 shmmap_s->va == (vm_offset_t)shmaddr) { 354 break; 355 } 356 } 357 if (i == shminfo.shmseg) 358 return (EINVAL); 359 #if (defined(AUDIT) && defined(KDTRACE_HOOKS)) || defined(MAC) 360 shmsegptr = &shmsegs[IPCID_TO_IX(shmmap_s->shmid)]; 361 #endif 362 #ifdef MAC 363 error = mac_sysvshm_check_shmdt(td->td_ucred, shmsegptr); 364 if (error != 0) 365 return (error); 366 #endif 367 return (shm_delete_mapping(p->p_vmspace, shmmap_s)); 368 } 369 370 #ifndef _SYS_SYSPROTO_H_ 371 struct shmdt_args { 372 const void *shmaddr; 373 }; 374 #endif 375 int 376 sys_shmdt(struct thread *td, struct shmdt_args *uap) 377 { 378 int error; 379 380 SYSVSHM_LOCK(); 381 error = kern_shmdt_locked(td, uap->shmaddr); 382 SYSVSHM_UNLOCK(); 383 return (error); 384 } 385 386 static int 387 kern_shmat_locked(struct thread *td, int shmid, const void *shmaddr, 388 int shmflg) 389 { 390 struct prison *rpr; 391 struct proc *p = td->td_proc; 392 struct shmid_kernel *shmseg; 393 struct shmmap_state *shmmap_s; 394 vm_offset_t attach_va; 395 vm_prot_t prot; 396 vm_size_t size; 397 int error, i, rv; 398 399 AUDIT_ARG_SVIPC_ID(shmid); 400 AUDIT_ARG_VALUE(shmflg); 401 402 SYSVSHM_ASSERT_LOCKED(); 403 rpr = shm_find_prison(td->td_ucred); 404 if (rpr == NULL) 405 return (ENOSYS); 406 shmmap_s = p->p_vmspace->vm_shm; 407 if (shmmap_s == NULL) { 408 shmmap_s = malloc(shminfo.shmseg * sizeof(struct shmmap_state), 409 M_SHM, M_WAITOK); 410 for (i = 0; i < shminfo.shmseg; i++) 411 shmmap_s[i].shmid = -1; 412 KASSERT(p->p_vmspace->vm_shm == NULL, ("raced")); 413 p->p_vmspace->vm_shm = shmmap_s; 414 } 415 shmseg = shm_find_segment(rpr, shmid, true); 416 if (shmseg == NULL) 417 return (EINVAL); 418 error = ipcperm(td, &shmseg->u.shm_perm, 419 (shmflg & SHM_RDONLY) ? IPC_R : IPC_R|IPC_W); 420 if (error != 0) 421 return (error); 422 #ifdef MAC 423 error = mac_sysvshm_check_shmat(td->td_ucred, shmseg, shmflg); 424 if (error != 0) 425 return (error); 426 #endif 427 for (i = 0; i < shminfo.shmseg; i++) { 428 if (shmmap_s->shmid == -1) 429 break; 430 shmmap_s++; 431 } 432 if (i >= shminfo.shmseg) 433 return (EMFILE); 434 size = round_page(shmseg->u.shm_segsz); 435 prot = VM_PROT_READ; 436 if ((shmflg & SHM_RDONLY) == 0) 437 prot |= VM_PROT_WRITE; 438 if (shmaddr != NULL) { 439 if ((shmflg & SHM_RND) != 0) 440 attach_va = rounddown2((vm_offset_t)shmaddr, SHMLBA); 441 else if (((vm_offset_t)shmaddr & (SHMLBA-1)) == 0) 442 attach_va = (vm_offset_t)shmaddr; 443 else 444 return (EINVAL); 445 } else { 446 /* 447 * This is just a hint to vm_map_find() about where to 448 * put it. 449 */ 450 attach_va = round_page((vm_offset_t)p->p_vmspace->vm_daddr + 451 lim_max(td, RLIMIT_DATA)); 452 } 453 454 vm_object_reference(shmseg->object); 455 rv = vm_map_find(&p->p_vmspace->vm_map, shmseg->object, 0, &attach_va, 456 size, 0, shmaddr != NULL ? VMFS_NO_SPACE : VMFS_OPTIMAL_SPACE, 457 prot, prot, MAP_INHERIT_SHARE | MAP_PREFAULT_PARTIAL); 458 if (rv != KERN_SUCCESS) { 459 vm_object_deallocate(shmseg->object); 460 return (ENOMEM); 461 } 462 463 shmmap_s->va = attach_va; 464 shmmap_s->shmid = shmid; 465 shmseg->u.shm_lpid = p->p_pid; 466 shmseg->u.shm_atime = time_second; 467 shmseg->u.shm_nattch++; 468 td->td_retval[0] = attach_va; 469 return (error); 470 } 471 472 int 473 kern_shmat(struct thread *td, int shmid, const void *shmaddr, int shmflg) 474 { 475 int error; 476 477 SYSVSHM_LOCK(); 478 error = kern_shmat_locked(td, shmid, shmaddr, shmflg); 479 SYSVSHM_UNLOCK(); 480 return (error); 481 } 482 483 #ifndef _SYS_SYSPROTO_H_ 484 struct shmat_args { 485 int shmid; 486 const void *shmaddr; 487 int shmflg; 488 }; 489 #endif 490 int 491 sys_shmat(struct thread *td, struct shmat_args *uap) 492 { 493 494 return (kern_shmat(td, uap->shmid, uap->shmaddr, uap->shmflg)); 495 } 496 497 static int 498 kern_shmctl_locked(struct thread *td, int shmid, int cmd, void *buf, 499 size_t *bufsz) 500 { 501 struct prison *rpr; 502 struct shmid_kernel *shmseg; 503 struct shmid_ds *shmidp; 504 struct shm_info shm_info; 505 int error; 506 507 SYSVSHM_ASSERT_LOCKED(); 508 509 rpr = shm_find_prison(td->td_ucred); 510 if (rpr == NULL) 511 return (ENOSYS); 512 513 AUDIT_ARG_SVIPC_ID(shmid); 514 AUDIT_ARG_SVIPC_CMD(cmd); 515 516 switch (cmd) { 517 /* 518 * It is possible that kern_shmctl is being called from the Linux ABI 519 * layer, in which case, we will need to implement IPC_INFO. It should 520 * be noted that other shmctl calls will be funneled through here for 521 * Linix binaries as well. 522 * 523 * NB: The Linux ABI layer will convert this data to structure(s) more 524 * consistent with the Linux ABI. 525 */ 526 case IPC_INFO: 527 memcpy(buf, &shminfo, sizeof(shminfo)); 528 if (bufsz) 529 *bufsz = sizeof(shminfo); 530 td->td_retval[0] = shmalloced; 531 return (0); 532 case SHM_INFO: { 533 shm_info.used_ids = shm_nused; 534 shm_info.shm_rss = 0; /*XXX where to get from ? */ 535 shm_info.shm_tot = 0; /*XXX where to get from ? */ 536 shm_info.shm_swp = 0; /*XXX where to get from ? */ 537 shm_info.swap_attempts = 0; /*XXX where to get from ? */ 538 shm_info.swap_successes = 0; /*XXX where to get from ? */ 539 memcpy(buf, &shm_info, sizeof(shm_info)); 540 if (bufsz != NULL) 541 *bufsz = sizeof(shm_info); 542 td->td_retval[0] = shmalloced; 543 return (0); 544 } 545 } 546 shmseg = shm_find_segment(rpr, shmid, cmd != SHM_STAT); 547 if (shmseg == NULL) 548 return (EINVAL); 549 #ifdef MAC 550 error = mac_sysvshm_check_shmctl(td->td_ucred, shmseg, cmd); 551 if (error != 0) 552 return (error); 553 #endif 554 switch (cmd) { 555 case SHM_STAT: 556 case IPC_STAT: 557 shmidp = (struct shmid_ds *)buf; 558 error = ipcperm(td, &shmseg->u.shm_perm, IPC_R); 559 if (error != 0) 560 return (error); 561 memcpy(shmidp, &shmseg->u, sizeof(struct shmid_ds)); 562 if (td->td_ucred->cr_prison != shmseg->cred->cr_prison) 563 shmidp->shm_perm.key = IPC_PRIVATE; 564 if (bufsz != NULL) 565 *bufsz = sizeof(struct shmid_ds); 566 if (cmd == SHM_STAT) { 567 td->td_retval[0] = IXSEQ_TO_IPCID(shmid, 568 shmseg->u.shm_perm); 569 } 570 break; 571 case IPC_SET: 572 shmidp = (struct shmid_ds *)buf; 573 AUDIT_ARG_SVIPC_PERM(&shmidp->shm_perm); 574 error = ipcperm(td, &shmseg->u.shm_perm, IPC_M); 575 if (error != 0) 576 return (error); 577 shmseg->u.shm_perm.uid = shmidp->shm_perm.uid; 578 shmseg->u.shm_perm.gid = shmidp->shm_perm.gid; 579 shmseg->u.shm_perm.mode = 580 (shmseg->u.shm_perm.mode & ~ACCESSPERMS) | 581 (shmidp->shm_perm.mode & ACCESSPERMS); 582 shmseg->u.shm_ctime = time_second; 583 break; 584 case IPC_RMID: 585 error = ipcperm(td, &shmseg->u.shm_perm, IPC_M); 586 if (error != 0) 587 return (error); 588 shm_remove(shmseg, IPCID_TO_IX(shmid)); 589 break; 590 #if 0 591 case SHM_LOCK: 592 case SHM_UNLOCK: 593 #endif 594 default: 595 error = EINVAL; 596 break; 597 } 598 return (error); 599 } 600 601 int 602 kern_shmctl(struct thread *td, int shmid, int cmd, void *buf, size_t *bufsz) 603 { 604 int error; 605 606 SYSVSHM_LOCK(); 607 error = kern_shmctl_locked(td, shmid, cmd, buf, bufsz); 608 SYSVSHM_UNLOCK(); 609 return (error); 610 } 611 612 613 #ifndef _SYS_SYSPROTO_H_ 614 struct shmctl_args { 615 int shmid; 616 int cmd; 617 struct shmid_ds *buf; 618 }; 619 #endif 620 int 621 sys_shmctl(struct thread *td, struct shmctl_args *uap) 622 { 623 int error; 624 struct shmid_ds buf; 625 size_t bufsz; 626 627 /* 628 * The only reason IPC_INFO, SHM_INFO, SHM_STAT exists is to support 629 * Linux binaries. If we see the call come through the FreeBSD ABI, 630 * return an error back to the user since we do not to support this. 631 */ 632 if (uap->cmd == IPC_INFO || uap->cmd == SHM_INFO || 633 uap->cmd == SHM_STAT) 634 return (EINVAL); 635 636 /* IPC_SET needs to copyin the buffer before calling kern_shmctl */ 637 if (uap->cmd == IPC_SET) { 638 if ((error = copyin(uap->buf, &buf, sizeof(struct shmid_ds)))) 639 goto done; 640 } 641 642 error = kern_shmctl(td, uap->shmid, uap->cmd, (void *)&buf, &bufsz); 643 if (error) 644 goto done; 645 646 /* Cases in which we need to copyout */ 647 switch (uap->cmd) { 648 case IPC_STAT: 649 error = copyout(&buf, uap->buf, bufsz); 650 break; 651 } 652 653 done: 654 if (error) { 655 /* Invalidate the return value */ 656 td->td_retval[0] = -1; 657 } 658 return (error); 659 } 660 661 662 static int 663 shmget_existing(struct thread *td, struct shmget_args *uap, int mode, 664 int segnum) 665 { 666 struct shmid_kernel *shmseg; 667 #ifdef MAC 668 int error; 669 #endif 670 671 SYSVSHM_ASSERT_LOCKED(); 672 KASSERT(segnum >= 0 && segnum < shmalloced, 673 ("segnum %d shmalloced %d", segnum, shmalloced)); 674 shmseg = &shmsegs[segnum]; 675 if ((uap->shmflg & (IPC_CREAT | IPC_EXCL)) == (IPC_CREAT | IPC_EXCL)) 676 return (EEXIST); 677 #ifdef MAC 678 error = mac_sysvshm_check_shmget(td->td_ucred, shmseg, uap->shmflg); 679 if (error != 0) 680 return (error); 681 #endif 682 if (uap->size != 0 && uap->size > shmseg->u.shm_segsz) 683 return (EINVAL); 684 td->td_retval[0] = IXSEQ_TO_IPCID(segnum, shmseg->u.shm_perm); 685 return (0); 686 } 687 688 static int 689 shmget_allocate_segment(struct thread *td, struct shmget_args *uap, int mode) 690 { 691 struct ucred *cred = td->td_ucred; 692 struct shmid_kernel *shmseg; 693 vm_object_t shm_object; 694 int i, segnum; 695 size_t size; 696 697 SYSVSHM_ASSERT_LOCKED(); 698 699 if (uap->size < shminfo.shmmin || uap->size > shminfo.shmmax) 700 return (EINVAL); 701 if (shm_nused >= shminfo.shmmni) /* Any shmids left? */ 702 return (ENOSPC); 703 size = round_page(uap->size); 704 if (shm_committed + btoc(size) > shminfo.shmall) 705 return (ENOMEM); 706 if (shm_last_free < 0) { 707 shmrealloc(); /* Maybe expand the shmsegs[] array. */ 708 for (i = 0; i < shmalloced; i++) 709 if (shmsegs[i].u.shm_perm.mode & SHMSEG_FREE) 710 break; 711 if (i == shmalloced) 712 return (ENOSPC); 713 segnum = i; 714 } else { 715 segnum = shm_last_free; 716 shm_last_free = -1; 717 } 718 KASSERT(segnum >= 0 && segnum < shmalloced, 719 ("segnum %d shmalloced %d", segnum, shmalloced)); 720 shmseg = &shmsegs[segnum]; 721 #ifdef RACCT 722 if (racct_enable) { 723 PROC_LOCK(td->td_proc); 724 if (racct_add(td->td_proc, RACCT_NSHM, 1)) { 725 PROC_UNLOCK(td->td_proc); 726 return (ENOSPC); 727 } 728 if (racct_add(td->td_proc, RACCT_SHMSIZE, size)) { 729 racct_sub(td->td_proc, RACCT_NSHM, 1); 730 PROC_UNLOCK(td->td_proc); 731 return (ENOMEM); 732 } 733 PROC_UNLOCK(td->td_proc); 734 } 735 #endif 736 737 /* 738 * We make sure that we have allocated a pager before we need 739 * to. 740 */ 741 shm_object = vm_pager_allocate(shm_use_phys ? OBJT_PHYS : OBJT_SWAP, 742 0, size, VM_PROT_DEFAULT, 0, cred); 743 if (shm_object == NULL) { 744 #ifdef RACCT 745 if (racct_enable) { 746 PROC_LOCK(td->td_proc); 747 racct_sub(td->td_proc, RACCT_NSHM, 1); 748 racct_sub(td->td_proc, RACCT_SHMSIZE, size); 749 PROC_UNLOCK(td->td_proc); 750 } 751 #endif 752 return (ENOMEM); 753 } 754 shm_object->pg_color = 0; 755 VM_OBJECT_WLOCK(shm_object); 756 vm_object_clear_flag(shm_object, OBJ_ONEMAPPING); 757 vm_object_set_flag(shm_object, OBJ_COLORED | OBJ_NOSPLIT); 758 VM_OBJECT_WUNLOCK(shm_object); 759 760 shmseg->object = shm_object; 761 shmseg->u.shm_perm.cuid = shmseg->u.shm_perm.uid = cred->cr_uid; 762 shmseg->u.shm_perm.cgid = shmseg->u.shm_perm.gid = cred->cr_gid; 763 shmseg->u.shm_perm.mode = (mode & ACCESSPERMS) | SHMSEG_ALLOCATED; 764 shmseg->u.shm_perm.key = uap->key; 765 shmseg->u.shm_perm.seq = (shmseg->u.shm_perm.seq + 1) & 0x7fff; 766 shmseg->cred = crhold(cred); 767 shmseg->u.shm_segsz = uap->size; 768 shmseg->u.shm_cpid = td->td_proc->p_pid; 769 shmseg->u.shm_lpid = shmseg->u.shm_nattch = 0; 770 shmseg->u.shm_atime = shmseg->u.shm_dtime = 0; 771 #ifdef MAC 772 mac_sysvshm_create(cred, shmseg); 773 #endif 774 shmseg->u.shm_ctime = time_second; 775 shm_committed += btoc(size); 776 shm_nused++; 777 td->td_retval[0] = IXSEQ_TO_IPCID(segnum, shmseg->u.shm_perm); 778 779 return (0); 780 } 781 782 #ifndef _SYS_SYSPROTO_H_ 783 struct shmget_args { 784 key_t key; 785 size_t size; 786 int shmflg; 787 }; 788 #endif 789 int 790 sys_shmget(struct thread *td, struct shmget_args *uap) 791 { 792 int segnum, mode; 793 int error; 794 795 if (shm_find_prison(td->td_ucred) == NULL) 796 return (ENOSYS); 797 mode = uap->shmflg & ACCESSPERMS; 798 SYSVSHM_LOCK(); 799 if (uap->key == IPC_PRIVATE) { 800 error = shmget_allocate_segment(td, uap, mode); 801 } else { 802 segnum = shm_find_segment_by_key(td->td_ucred->cr_prison, 803 uap->key); 804 if (segnum >= 0) 805 error = shmget_existing(td, uap, mode, segnum); 806 else if ((uap->shmflg & IPC_CREAT) == 0) 807 error = ENOENT; 808 else 809 error = shmget_allocate_segment(td, uap, mode); 810 } 811 SYSVSHM_UNLOCK(); 812 return (error); 813 } 814 815 static void 816 shmfork_myhook(struct proc *p1, struct proc *p2) 817 { 818 struct shmmap_state *shmmap_s; 819 size_t size; 820 int i; 821 822 SYSVSHM_LOCK(); 823 size = shminfo.shmseg * sizeof(struct shmmap_state); 824 shmmap_s = malloc(size, M_SHM, M_WAITOK); 825 bcopy(p1->p_vmspace->vm_shm, shmmap_s, size); 826 p2->p_vmspace->vm_shm = shmmap_s; 827 for (i = 0; i < shminfo.shmseg; i++, shmmap_s++) { 828 if (shmmap_s->shmid != -1) { 829 KASSERT(IPCID_TO_IX(shmmap_s->shmid) >= 0 && 830 IPCID_TO_IX(shmmap_s->shmid) < shmalloced, 831 ("segnum %d shmalloced %d", 832 IPCID_TO_IX(shmmap_s->shmid), shmalloced)); 833 shmsegs[IPCID_TO_IX(shmmap_s->shmid)].u.shm_nattch++; 834 } 835 } 836 SYSVSHM_UNLOCK(); 837 } 838 839 static void 840 shmexit_myhook(struct vmspace *vm) 841 { 842 struct shmmap_state *base, *shm; 843 int i; 844 845 base = vm->vm_shm; 846 if (base != NULL) { 847 vm->vm_shm = NULL; 848 SYSVSHM_LOCK(); 849 for (i = 0, shm = base; i < shminfo.shmseg; i++, shm++) { 850 if (shm->shmid != -1) 851 shm_delete_mapping(vm, shm); 852 } 853 SYSVSHM_UNLOCK(); 854 free(base, M_SHM); 855 } 856 } 857 858 static void 859 shmrealloc(void) 860 { 861 struct shmid_kernel *newsegs; 862 int i; 863 864 SYSVSHM_ASSERT_LOCKED(); 865 866 if (shmalloced >= shminfo.shmmni) 867 return; 868 869 newsegs = malloc(shminfo.shmmni * sizeof(*newsegs), M_SHM, M_WAITOK); 870 for (i = 0; i < shmalloced; i++) 871 bcopy(&shmsegs[i], &newsegs[i], sizeof(newsegs[0])); 872 for (; i < shminfo.shmmni; i++) { 873 newsegs[i].u.shm_perm.mode = SHMSEG_FREE; 874 newsegs[i].u.shm_perm.seq = 0; 875 #ifdef MAC 876 mac_sysvshm_init(&newsegs[i]); 877 #endif 878 } 879 free(shmsegs, M_SHM); 880 shmsegs = newsegs; 881 shmalloced = shminfo.shmmni; 882 } 883 884 static struct syscall_helper_data shm_syscalls[] = { 885 SYSCALL_INIT_HELPER(shmat), 886 SYSCALL_INIT_HELPER(shmctl), 887 SYSCALL_INIT_HELPER(shmdt), 888 SYSCALL_INIT_HELPER(shmget), 889 #if defined(COMPAT_FREEBSD4) || defined(COMPAT_FREEBSD5) || \ 890 defined(COMPAT_FREEBSD6) || defined(COMPAT_FREEBSD7) 891 SYSCALL_INIT_HELPER_COMPAT(freebsd7_shmctl), 892 #endif 893 #if defined(__i386__) && (defined(COMPAT_FREEBSD4) || defined(COMPAT_43)) 894 SYSCALL_INIT_HELPER(shmsys), 895 #endif 896 SYSCALL_INIT_LAST 897 }; 898 899 #ifdef COMPAT_FREEBSD32 900 #include <compat/freebsd32/freebsd32.h> 901 #include <compat/freebsd32/freebsd32_ipc.h> 902 #include <compat/freebsd32/freebsd32_proto.h> 903 #include <compat/freebsd32/freebsd32_signal.h> 904 #include <compat/freebsd32/freebsd32_syscall.h> 905 #include <compat/freebsd32/freebsd32_util.h> 906 907 static struct syscall_helper_data shm32_syscalls[] = { 908 SYSCALL32_INIT_HELPER_COMPAT(shmat), 909 SYSCALL32_INIT_HELPER_COMPAT(shmdt), 910 SYSCALL32_INIT_HELPER_COMPAT(shmget), 911 SYSCALL32_INIT_HELPER(freebsd32_shmsys), 912 SYSCALL32_INIT_HELPER(freebsd32_shmctl), 913 #if defined(COMPAT_FREEBSD4) || defined(COMPAT_FREEBSD5) || \ 914 defined(COMPAT_FREEBSD6) || defined(COMPAT_FREEBSD7) 915 SYSCALL32_INIT_HELPER(freebsd7_freebsd32_shmctl), 916 #endif 917 SYSCALL_INIT_LAST 918 }; 919 #endif 920 921 static int 922 shminit(void) 923 { 924 struct prison *pr; 925 void **rsv; 926 int i, error; 927 osd_method_t methods[PR_MAXMETHOD] = { 928 [PR_METHOD_CHECK] = shm_prison_check, 929 [PR_METHOD_SET] = shm_prison_set, 930 [PR_METHOD_GET] = shm_prison_get, 931 [PR_METHOD_REMOVE] = shm_prison_remove, 932 }; 933 934 #ifndef BURN_BRIDGES 935 if (TUNABLE_ULONG_FETCH("kern.ipc.shmmaxpgs", &shminfo.shmall) != 0) 936 printf("kern.ipc.shmmaxpgs is now called kern.ipc.shmall!\n"); 937 #endif 938 if (shminfo.shmmax == SHMMAX) { 939 /* Initialize shmmax dealing with possible overflow. */ 940 for (i = PAGE_SIZE; i != 0; i--) { 941 shminfo.shmmax = shminfo.shmall * i; 942 if ((shminfo.shmmax / shminfo.shmall) == (u_long)i) 943 break; 944 } 945 } 946 shmalloced = shminfo.shmmni; 947 shmsegs = malloc(shmalloced * sizeof(shmsegs[0]), M_SHM, M_WAITOK); 948 for (i = 0; i < shmalloced; i++) { 949 shmsegs[i].u.shm_perm.mode = SHMSEG_FREE; 950 shmsegs[i].u.shm_perm.seq = 0; 951 #ifdef MAC 952 mac_sysvshm_init(&shmsegs[i]); 953 #endif 954 } 955 shm_last_free = 0; 956 shm_nused = 0; 957 shm_committed = 0; 958 sx_init(&sysvshmsx, "sysvshmsx"); 959 shmexit_hook = &shmexit_myhook; 960 shmfork_hook = &shmfork_myhook; 961 962 /* Set current prisons according to their allow.sysvipc. */ 963 shm_prison_slot = osd_jail_register(NULL, methods); 964 rsv = osd_reserve(shm_prison_slot); 965 prison_lock(&prison0); 966 (void)osd_jail_set_reserved(&prison0, shm_prison_slot, rsv, &prison0); 967 prison_unlock(&prison0); 968 rsv = NULL; 969 sx_slock(&allprison_lock); 970 TAILQ_FOREACH(pr, &allprison, pr_list) { 971 if (rsv == NULL) 972 rsv = osd_reserve(shm_prison_slot); 973 prison_lock(pr); 974 if ((pr->pr_allow & PR_ALLOW_SYSVIPC) && pr->pr_ref > 0) { 975 (void)osd_jail_set_reserved(pr, shm_prison_slot, rsv, 976 &prison0); 977 rsv = NULL; 978 } 979 prison_unlock(pr); 980 } 981 if (rsv != NULL) 982 osd_free_reserved(rsv); 983 sx_sunlock(&allprison_lock); 984 985 error = syscall_helper_register(shm_syscalls, SY_THR_STATIC_KLD); 986 if (error != 0) 987 return (error); 988 #ifdef COMPAT_FREEBSD32 989 error = syscall32_helper_register(shm32_syscalls, SY_THR_STATIC_KLD); 990 if (error != 0) 991 return (error); 992 #endif 993 return (0); 994 } 995 996 static int 997 shmunload(void) 998 { 999 int i; 1000 1001 if (shm_nused > 0) 1002 return (EBUSY); 1003 1004 #ifdef COMPAT_FREEBSD32 1005 syscall32_helper_unregister(shm32_syscalls); 1006 #endif 1007 syscall_helper_unregister(shm_syscalls); 1008 if (shm_prison_slot != 0) 1009 osd_jail_deregister(shm_prison_slot); 1010 1011 for (i = 0; i < shmalloced; i++) { 1012 #ifdef MAC 1013 mac_sysvshm_destroy(&shmsegs[i]); 1014 #endif 1015 /* 1016 * Objects might be still mapped into the processes 1017 * address spaces. Actual free would happen on the 1018 * last mapping destruction. 1019 */ 1020 if (shmsegs[i].u.shm_perm.mode != SHMSEG_FREE) 1021 vm_object_deallocate(shmsegs[i].object); 1022 } 1023 free(shmsegs, M_SHM); 1024 shmexit_hook = NULL; 1025 shmfork_hook = NULL; 1026 sx_destroy(&sysvshmsx); 1027 return (0); 1028 } 1029 1030 static int 1031 sysctl_shmsegs(SYSCTL_HANDLER_ARGS) 1032 { 1033 struct shmid_kernel tshmseg; 1034 struct prison *pr, *rpr; 1035 int error, i; 1036 1037 SYSVSHM_LOCK(); 1038 pr = req->td->td_ucred->cr_prison; 1039 rpr = shm_find_prison(req->td->td_ucred); 1040 error = 0; 1041 for (i = 0; i < shmalloced; i++) { 1042 if ((shmsegs[i].u.shm_perm.mode & SHMSEG_ALLOCATED) == 0 || 1043 rpr == NULL || shm_prison_cansee(rpr, &shmsegs[i]) != 0) { 1044 bzero(&tshmseg, sizeof(tshmseg)); 1045 tshmseg.u.shm_perm.mode = SHMSEG_FREE; 1046 } else { 1047 tshmseg = shmsegs[i]; 1048 if (tshmseg.cred->cr_prison != pr) 1049 tshmseg.u.shm_perm.key = IPC_PRIVATE; 1050 } 1051 error = SYSCTL_OUT(req, &tshmseg, sizeof(tshmseg)); 1052 if (error != 0) 1053 break; 1054 } 1055 SYSVSHM_UNLOCK(); 1056 return (error); 1057 } 1058 1059 static int 1060 shm_prison_check(void *obj, void *data) 1061 { 1062 struct prison *pr = obj; 1063 struct prison *prpr; 1064 struct vfsoptlist *opts = data; 1065 int error, jsys; 1066 1067 /* 1068 * sysvshm is a jailsys integer. 1069 * It must be "disable" if the parent jail is disabled. 1070 */ 1071 error = vfs_copyopt(opts, "sysvshm", &jsys, sizeof(jsys)); 1072 if (error != ENOENT) { 1073 if (error != 0) 1074 return (error); 1075 switch (jsys) { 1076 case JAIL_SYS_DISABLE: 1077 break; 1078 case JAIL_SYS_NEW: 1079 case JAIL_SYS_INHERIT: 1080 prison_lock(pr->pr_parent); 1081 prpr = osd_jail_get(pr->pr_parent, shm_prison_slot); 1082 prison_unlock(pr->pr_parent); 1083 if (prpr == NULL) 1084 return (EPERM); 1085 break; 1086 default: 1087 return (EINVAL); 1088 } 1089 } 1090 1091 return (0); 1092 } 1093 1094 static int 1095 shm_prison_set(void *obj, void *data) 1096 { 1097 struct prison *pr = obj; 1098 struct prison *tpr, *orpr, *nrpr, *trpr; 1099 struct vfsoptlist *opts = data; 1100 void *rsv; 1101 int jsys, descend; 1102 1103 /* 1104 * sysvshm controls which jail is the root of the associated segments 1105 * (this jail or same as the parent), or if the feature is available 1106 * at all. 1107 */ 1108 if (vfs_copyopt(opts, "sysvshm", &jsys, sizeof(jsys)) == ENOENT) 1109 jsys = vfs_flagopt(opts, "allow.sysvipc", NULL, 0) 1110 ? JAIL_SYS_INHERIT 1111 : vfs_flagopt(opts, "allow.nosysvipc", NULL, 0) 1112 ? JAIL_SYS_DISABLE 1113 : -1; 1114 if (jsys == JAIL_SYS_DISABLE) { 1115 prison_lock(pr); 1116 orpr = osd_jail_get(pr, shm_prison_slot); 1117 if (orpr != NULL) 1118 osd_jail_del(pr, shm_prison_slot); 1119 prison_unlock(pr); 1120 if (orpr != NULL) { 1121 if (orpr == pr) 1122 shm_prison_cleanup(pr); 1123 /* Disable all child jails as well. */ 1124 FOREACH_PRISON_DESCENDANT(pr, tpr, descend) { 1125 prison_lock(tpr); 1126 trpr = osd_jail_get(tpr, shm_prison_slot); 1127 if (trpr != NULL) { 1128 osd_jail_del(tpr, shm_prison_slot); 1129 prison_unlock(tpr); 1130 if (trpr == tpr) 1131 shm_prison_cleanup(tpr); 1132 } else { 1133 prison_unlock(tpr); 1134 descend = 0; 1135 } 1136 } 1137 } 1138 } else if (jsys != -1) { 1139 if (jsys == JAIL_SYS_NEW) 1140 nrpr = pr; 1141 else { 1142 prison_lock(pr->pr_parent); 1143 nrpr = osd_jail_get(pr->pr_parent, shm_prison_slot); 1144 prison_unlock(pr->pr_parent); 1145 } 1146 rsv = osd_reserve(shm_prison_slot); 1147 prison_lock(pr); 1148 orpr = osd_jail_get(pr, shm_prison_slot); 1149 if (orpr != nrpr) 1150 (void)osd_jail_set_reserved(pr, shm_prison_slot, rsv, 1151 nrpr); 1152 else 1153 osd_free_reserved(rsv); 1154 prison_unlock(pr); 1155 if (orpr != nrpr) { 1156 if (orpr == pr) 1157 shm_prison_cleanup(pr); 1158 if (orpr != NULL) { 1159 /* Change child jails matching the old root, */ 1160 FOREACH_PRISON_DESCENDANT(pr, tpr, descend) { 1161 prison_lock(tpr); 1162 trpr = osd_jail_get(tpr, 1163 shm_prison_slot); 1164 if (trpr == orpr) { 1165 (void)osd_jail_set(tpr, 1166 shm_prison_slot, nrpr); 1167 prison_unlock(tpr); 1168 if (trpr == tpr) 1169 shm_prison_cleanup(tpr); 1170 } else { 1171 prison_unlock(tpr); 1172 descend = 0; 1173 } 1174 } 1175 } 1176 } 1177 } 1178 1179 return (0); 1180 } 1181 1182 static int 1183 shm_prison_get(void *obj, void *data) 1184 { 1185 struct prison *pr = obj; 1186 struct prison *rpr; 1187 struct vfsoptlist *opts = data; 1188 int error, jsys; 1189 1190 /* Set sysvshm based on the jail's root prison. */ 1191 prison_lock(pr); 1192 rpr = osd_jail_get(pr, shm_prison_slot); 1193 prison_unlock(pr); 1194 jsys = rpr == NULL ? JAIL_SYS_DISABLE 1195 : rpr == pr ? JAIL_SYS_NEW : JAIL_SYS_INHERIT; 1196 error = vfs_setopt(opts, "sysvshm", &jsys, sizeof(jsys)); 1197 if (error == ENOENT) 1198 error = 0; 1199 return (error); 1200 } 1201 1202 static int 1203 shm_prison_remove(void *obj, void *data __unused) 1204 { 1205 struct prison *pr = obj; 1206 struct prison *rpr; 1207 1208 SYSVSHM_LOCK(); 1209 prison_lock(pr); 1210 rpr = osd_jail_get(pr, shm_prison_slot); 1211 prison_unlock(pr); 1212 if (rpr == pr) 1213 shm_prison_cleanup(pr); 1214 SYSVSHM_UNLOCK(); 1215 return (0); 1216 } 1217 1218 static void 1219 shm_prison_cleanup(struct prison *pr) 1220 { 1221 struct shmid_kernel *shmseg; 1222 int i; 1223 1224 /* Remove any segments that belong to this jail. */ 1225 for (i = 0; i < shmalloced; i++) { 1226 shmseg = &shmsegs[i]; 1227 if ((shmseg->u.shm_perm.mode & SHMSEG_ALLOCATED) && 1228 shmseg->cred != NULL && shmseg->cred->cr_prison == pr) { 1229 shm_remove(shmseg, i); 1230 } 1231 } 1232 } 1233 1234 SYSCTL_JAIL_PARAM_SYS_NODE(sysvshm, CTLFLAG_RW, "SYSV shared memory"); 1235 1236 #if defined(__i386__) && (defined(COMPAT_FREEBSD4) || defined(COMPAT_43)) 1237 struct oshmid_ds { 1238 struct ipc_perm_old shm_perm; /* operation perms */ 1239 int shm_segsz; /* size of segment (bytes) */ 1240 u_short shm_cpid; /* pid, creator */ 1241 u_short shm_lpid; /* pid, last operation */ 1242 short shm_nattch; /* no. of current attaches */ 1243 time_t shm_atime; /* last attach time */ 1244 time_t shm_dtime; /* last detach time */ 1245 time_t shm_ctime; /* last change time */ 1246 void *shm_handle; /* internal handle for shm segment */ 1247 }; 1248 1249 struct oshmctl_args { 1250 int shmid; 1251 int cmd; 1252 struct oshmid_ds *ubuf; 1253 }; 1254 1255 static int 1256 oshmctl(struct thread *td, struct oshmctl_args *uap) 1257 { 1258 #ifdef COMPAT_43 1259 int error = 0; 1260 struct prison *rpr; 1261 struct shmid_kernel *shmseg; 1262 struct oshmid_ds outbuf; 1263 1264 rpr = shm_find_prison(td->td_ucred); 1265 if (rpr == NULL) 1266 return (ENOSYS); 1267 if (uap->cmd != IPC_STAT) { 1268 return (freebsd7_shmctl(td, 1269 (struct freebsd7_shmctl_args *)uap)); 1270 } 1271 SYSVSHM_LOCK(); 1272 shmseg = shm_find_segment(rpr, uap->shmid, true); 1273 if (shmseg == NULL) { 1274 SYSVSHM_UNLOCK(); 1275 return (EINVAL); 1276 } 1277 error = ipcperm(td, &shmseg->u.shm_perm, IPC_R); 1278 if (error != 0) { 1279 SYSVSHM_UNLOCK(); 1280 return (error); 1281 } 1282 #ifdef MAC 1283 error = mac_sysvshm_check_shmctl(td->td_ucred, shmseg, uap->cmd); 1284 if (error != 0) { 1285 SYSVSHM_UNLOCK(); 1286 return (error); 1287 } 1288 #endif 1289 ipcperm_new2old(&shmseg->u.shm_perm, &outbuf.shm_perm); 1290 outbuf.shm_segsz = shmseg->u.shm_segsz; 1291 outbuf.shm_cpid = shmseg->u.shm_cpid; 1292 outbuf.shm_lpid = shmseg->u.shm_lpid; 1293 outbuf.shm_nattch = shmseg->u.shm_nattch; 1294 outbuf.shm_atime = shmseg->u.shm_atime; 1295 outbuf.shm_dtime = shmseg->u.shm_dtime; 1296 outbuf.shm_ctime = shmseg->u.shm_ctime; 1297 outbuf.shm_handle = shmseg->object; 1298 SYSVSHM_UNLOCK(); 1299 return (copyout(&outbuf, uap->ubuf, sizeof(outbuf))); 1300 #else 1301 return (EINVAL); 1302 #endif 1303 } 1304 1305 /* XXX casting to (sy_call_t *) is bogus, as usual. */ 1306 static sy_call_t *shmcalls[] = { 1307 (sy_call_t *)sys_shmat, (sy_call_t *)oshmctl, 1308 (sy_call_t *)sys_shmdt, (sy_call_t *)sys_shmget, 1309 (sy_call_t *)freebsd7_shmctl 1310 }; 1311 1312 #ifndef _SYS_SYSPROTO_H_ 1313 /* XXX actually varargs. */ 1314 struct shmsys_args { 1315 int which; 1316 int a2; 1317 int a3; 1318 int a4; 1319 }; 1320 #endif 1321 int 1322 sys_shmsys(struct thread *td, struct shmsys_args *uap) 1323 { 1324 1325 AUDIT_ARG_SVIPC_WHICH(uap->which); 1326 if (uap->which < 0 || uap->which >= nitems(shmcalls)) 1327 return (EINVAL); 1328 return ((*shmcalls[uap->which])(td, &uap->a2)); 1329 } 1330 1331 #endif /* i386 && (COMPAT_FREEBSD4 || COMPAT_43) */ 1332 1333 #ifdef COMPAT_FREEBSD32 1334 1335 int 1336 freebsd32_shmsys(struct thread *td, struct freebsd32_shmsys_args *uap) 1337 { 1338 1339 #if defined(COMPAT_FREEBSD4) || defined(COMPAT_FREEBSD5) || \ 1340 defined(COMPAT_FREEBSD6) || defined(COMPAT_FREEBSD7) 1341 AUDIT_ARG_SVIPC_WHICH(uap->which); 1342 switch (uap->which) { 1343 case 0: { /* shmat */ 1344 struct shmat_args ap; 1345 1346 ap.shmid = uap->a2; 1347 ap.shmaddr = PTRIN(uap->a3); 1348 ap.shmflg = uap->a4; 1349 return (sysent[SYS_shmat].sy_call(td, &ap)); 1350 } 1351 case 2: { /* shmdt */ 1352 struct shmdt_args ap; 1353 1354 ap.shmaddr = PTRIN(uap->a2); 1355 return (sysent[SYS_shmdt].sy_call(td, &ap)); 1356 } 1357 case 3: { /* shmget */ 1358 struct shmget_args ap; 1359 1360 ap.key = uap->a2; 1361 ap.size = uap->a3; 1362 ap.shmflg = uap->a4; 1363 return (sysent[SYS_shmget].sy_call(td, &ap)); 1364 } 1365 case 4: { /* shmctl */ 1366 struct freebsd7_freebsd32_shmctl_args ap; 1367 1368 ap.shmid = uap->a2; 1369 ap.cmd = uap->a3; 1370 ap.buf = PTRIN(uap->a4); 1371 return (freebsd7_freebsd32_shmctl(td, &ap)); 1372 } 1373 case 1: /* oshmctl */ 1374 default: 1375 return (EINVAL); 1376 } 1377 #else 1378 return (nosys(td, NULL)); 1379 #endif 1380 } 1381 1382 #if defined(COMPAT_FREEBSD4) || defined(COMPAT_FREEBSD5) || \ 1383 defined(COMPAT_FREEBSD6) || defined(COMPAT_FREEBSD7) 1384 int 1385 freebsd7_freebsd32_shmctl(struct thread *td, 1386 struct freebsd7_freebsd32_shmctl_args *uap) 1387 { 1388 int error; 1389 union { 1390 struct shmid_ds shmid_ds; 1391 struct shm_info shm_info; 1392 struct shminfo shminfo; 1393 } u; 1394 union { 1395 struct shmid_ds32_old shmid_ds32; 1396 struct shm_info32 shm_info32; 1397 struct shminfo32 shminfo32; 1398 } u32; 1399 size_t sz; 1400 1401 if (uap->cmd == IPC_SET) { 1402 if ((error = copyin(uap->buf, &u32.shmid_ds32, 1403 sizeof(u32.shmid_ds32)))) 1404 goto done; 1405 freebsd32_ipcperm_old_in(&u32.shmid_ds32.shm_perm, 1406 &u.shmid_ds.shm_perm); 1407 CP(u32.shmid_ds32, u.shmid_ds, shm_segsz); 1408 CP(u32.shmid_ds32, u.shmid_ds, shm_lpid); 1409 CP(u32.shmid_ds32, u.shmid_ds, shm_cpid); 1410 CP(u32.shmid_ds32, u.shmid_ds, shm_nattch); 1411 CP(u32.shmid_ds32, u.shmid_ds, shm_atime); 1412 CP(u32.shmid_ds32, u.shmid_ds, shm_dtime); 1413 CP(u32.shmid_ds32, u.shmid_ds, shm_ctime); 1414 } 1415 1416 error = kern_shmctl(td, uap->shmid, uap->cmd, (void *)&u, &sz); 1417 if (error) 1418 goto done; 1419 1420 /* Cases in which we need to copyout */ 1421 switch (uap->cmd) { 1422 case IPC_INFO: 1423 CP(u.shminfo, u32.shminfo32, shmmax); 1424 CP(u.shminfo, u32.shminfo32, shmmin); 1425 CP(u.shminfo, u32.shminfo32, shmmni); 1426 CP(u.shminfo, u32.shminfo32, shmseg); 1427 CP(u.shminfo, u32.shminfo32, shmall); 1428 error = copyout(&u32.shminfo32, uap->buf, 1429 sizeof(u32.shminfo32)); 1430 break; 1431 case SHM_INFO: 1432 CP(u.shm_info, u32.shm_info32, used_ids); 1433 CP(u.shm_info, u32.shm_info32, shm_rss); 1434 CP(u.shm_info, u32.shm_info32, shm_tot); 1435 CP(u.shm_info, u32.shm_info32, shm_swp); 1436 CP(u.shm_info, u32.shm_info32, swap_attempts); 1437 CP(u.shm_info, u32.shm_info32, swap_successes); 1438 error = copyout(&u32.shm_info32, uap->buf, 1439 sizeof(u32.shm_info32)); 1440 break; 1441 case SHM_STAT: 1442 case IPC_STAT: 1443 freebsd32_ipcperm_old_out(&u.shmid_ds.shm_perm, 1444 &u32.shmid_ds32.shm_perm); 1445 if (u.shmid_ds.shm_segsz > INT32_MAX) 1446 u32.shmid_ds32.shm_segsz = INT32_MAX; 1447 else 1448 CP(u.shmid_ds, u32.shmid_ds32, shm_segsz); 1449 CP(u.shmid_ds, u32.shmid_ds32, shm_lpid); 1450 CP(u.shmid_ds, u32.shmid_ds32, shm_cpid); 1451 CP(u.shmid_ds, u32.shmid_ds32, shm_nattch); 1452 CP(u.shmid_ds, u32.shmid_ds32, shm_atime); 1453 CP(u.shmid_ds, u32.shmid_ds32, shm_dtime); 1454 CP(u.shmid_ds, u32.shmid_ds32, shm_ctime); 1455 u32.shmid_ds32.shm_internal = 0; 1456 error = copyout(&u32.shmid_ds32, uap->buf, 1457 sizeof(u32.shmid_ds32)); 1458 break; 1459 } 1460 1461 done: 1462 if (error) { 1463 /* Invalidate the return value */ 1464 td->td_retval[0] = -1; 1465 } 1466 return (error); 1467 } 1468 #endif 1469 1470 int 1471 freebsd32_shmctl(struct thread *td, struct freebsd32_shmctl_args *uap) 1472 { 1473 int error; 1474 union { 1475 struct shmid_ds shmid_ds; 1476 struct shm_info shm_info; 1477 struct shminfo shminfo; 1478 } u; 1479 union { 1480 struct shmid_ds32 shmid_ds32; 1481 struct shm_info32 shm_info32; 1482 struct shminfo32 shminfo32; 1483 } u32; 1484 size_t sz; 1485 1486 if (uap->cmd == IPC_SET) { 1487 if ((error = copyin(uap->buf, &u32.shmid_ds32, 1488 sizeof(u32.shmid_ds32)))) 1489 goto done; 1490 freebsd32_ipcperm_in(&u32.shmid_ds32.shm_perm, 1491 &u.shmid_ds.shm_perm); 1492 CP(u32.shmid_ds32, u.shmid_ds, shm_segsz); 1493 CP(u32.shmid_ds32, u.shmid_ds, shm_lpid); 1494 CP(u32.shmid_ds32, u.shmid_ds, shm_cpid); 1495 CP(u32.shmid_ds32, u.shmid_ds, shm_nattch); 1496 CP(u32.shmid_ds32, u.shmid_ds, shm_atime); 1497 CP(u32.shmid_ds32, u.shmid_ds, shm_dtime); 1498 CP(u32.shmid_ds32, u.shmid_ds, shm_ctime); 1499 } 1500 1501 error = kern_shmctl(td, uap->shmid, uap->cmd, (void *)&u, &sz); 1502 if (error) 1503 goto done; 1504 1505 /* Cases in which we need to copyout */ 1506 switch (uap->cmd) { 1507 case IPC_INFO: 1508 CP(u.shminfo, u32.shminfo32, shmmax); 1509 CP(u.shminfo, u32.shminfo32, shmmin); 1510 CP(u.shminfo, u32.shminfo32, shmmni); 1511 CP(u.shminfo, u32.shminfo32, shmseg); 1512 CP(u.shminfo, u32.shminfo32, shmall); 1513 error = copyout(&u32.shminfo32, uap->buf, 1514 sizeof(u32.shminfo32)); 1515 break; 1516 case SHM_INFO: 1517 CP(u.shm_info, u32.shm_info32, used_ids); 1518 CP(u.shm_info, u32.shm_info32, shm_rss); 1519 CP(u.shm_info, u32.shm_info32, shm_tot); 1520 CP(u.shm_info, u32.shm_info32, shm_swp); 1521 CP(u.shm_info, u32.shm_info32, swap_attempts); 1522 CP(u.shm_info, u32.shm_info32, swap_successes); 1523 error = copyout(&u32.shm_info32, uap->buf, 1524 sizeof(u32.shm_info32)); 1525 break; 1526 case SHM_STAT: 1527 case IPC_STAT: 1528 freebsd32_ipcperm_out(&u.shmid_ds.shm_perm, 1529 &u32.shmid_ds32.shm_perm); 1530 if (u.shmid_ds.shm_segsz > INT32_MAX) 1531 u32.shmid_ds32.shm_segsz = INT32_MAX; 1532 else 1533 CP(u.shmid_ds, u32.shmid_ds32, shm_segsz); 1534 CP(u.shmid_ds, u32.shmid_ds32, shm_lpid); 1535 CP(u.shmid_ds, u32.shmid_ds32, shm_cpid); 1536 CP(u.shmid_ds, u32.shmid_ds32, shm_nattch); 1537 CP(u.shmid_ds, u32.shmid_ds32, shm_atime); 1538 CP(u.shmid_ds, u32.shmid_ds32, shm_dtime); 1539 CP(u.shmid_ds, u32.shmid_ds32, shm_ctime); 1540 error = copyout(&u32.shmid_ds32, uap->buf, 1541 sizeof(u32.shmid_ds32)); 1542 break; 1543 } 1544 1545 done: 1546 if (error) { 1547 /* Invalidate the return value */ 1548 td->td_retval[0] = -1; 1549 } 1550 return (error); 1551 } 1552 #endif 1553 1554 #if defined(COMPAT_FREEBSD4) || defined(COMPAT_FREEBSD5) || \ 1555 defined(COMPAT_FREEBSD6) || defined(COMPAT_FREEBSD7) 1556 1557 #ifndef CP 1558 #define CP(src, dst, fld) do { (dst).fld = (src).fld; } while (0) 1559 #endif 1560 1561 #ifndef _SYS_SYSPROTO_H_ 1562 struct freebsd7_shmctl_args { 1563 int shmid; 1564 int cmd; 1565 struct shmid_ds_old *buf; 1566 }; 1567 #endif 1568 int 1569 freebsd7_shmctl(struct thread *td, struct freebsd7_shmctl_args *uap) 1570 { 1571 int error; 1572 struct shmid_ds_old old; 1573 struct shmid_ds buf; 1574 size_t bufsz; 1575 1576 /* 1577 * The only reason IPC_INFO, SHM_INFO, SHM_STAT exists is to support 1578 * Linux binaries. If we see the call come through the FreeBSD ABI, 1579 * return an error back to the user since we do not to support this. 1580 */ 1581 if (uap->cmd == IPC_INFO || uap->cmd == SHM_INFO || 1582 uap->cmd == SHM_STAT) 1583 return (EINVAL); 1584 1585 /* IPC_SET needs to copyin the buffer before calling kern_shmctl */ 1586 if (uap->cmd == IPC_SET) { 1587 if ((error = copyin(uap->buf, &old, sizeof(old)))) 1588 goto done; 1589 ipcperm_old2new(&old.shm_perm, &buf.shm_perm); 1590 CP(old, buf, shm_segsz); 1591 CP(old, buf, shm_lpid); 1592 CP(old, buf, shm_cpid); 1593 CP(old, buf, shm_nattch); 1594 CP(old, buf, shm_atime); 1595 CP(old, buf, shm_dtime); 1596 CP(old, buf, shm_ctime); 1597 } 1598 1599 error = kern_shmctl(td, uap->shmid, uap->cmd, (void *)&buf, &bufsz); 1600 if (error) 1601 goto done; 1602 1603 /* Cases in which we need to copyout */ 1604 switch (uap->cmd) { 1605 case IPC_STAT: 1606 ipcperm_new2old(&buf.shm_perm, &old.shm_perm); 1607 if (buf.shm_segsz > INT_MAX) 1608 old.shm_segsz = INT_MAX; 1609 else 1610 CP(buf, old, shm_segsz); 1611 CP(buf, old, shm_lpid); 1612 CP(buf, old, shm_cpid); 1613 if (buf.shm_nattch > SHRT_MAX) 1614 old.shm_nattch = SHRT_MAX; 1615 else 1616 CP(buf, old, shm_nattch); 1617 CP(buf, old, shm_atime); 1618 CP(buf, old, shm_dtime); 1619 CP(buf, old, shm_ctime); 1620 old.shm_internal = NULL; 1621 error = copyout(&old, uap->buf, sizeof(old)); 1622 break; 1623 } 1624 1625 done: 1626 if (error) { 1627 /* Invalidate the return value */ 1628 td->td_retval[0] = -1; 1629 } 1630 return (error); 1631 } 1632 1633 #endif /* COMPAT_FREEBSD4 || COMPAT_FREEBSD5 || COMPAT_FREEBSD6 || 1634 COMPAT_FREEBSD7 */ 1635 1636 static int 1637 sysvshm_modload(struct module *module, int cmd, void *arg) 1638 { 1639 int error = 0; 1640 1641 switch (cmd) { 1642 case MOD_LOAD: 1643 error = shminit(); 1644 if (error != 0) 1645 shmunload(); 1646 break; 1647 case MOD_UNLOAD: 1648 error = shmunload(); 1649 break; 1650 case MOD_SHUTDOWN: 1651 break; 1652 default: 1653 error = EINVAL; 1654 break; 1655 } 1656 return (error); 1657 } 1658 1659 static moduledata_t sysvshm_mod = { 1660 "sysvshm", 1661 &sysvshm_modload, 1662 NULL 1663 }; 1664 1665 DECLARE_MODULE(sysvshm, sysvshm_mod, SI_SUB_SYSV_SHM, SI_ORDER_FIRST); 1666 MODULE_VERSION(sysvshm, 1); 1667