1 /*- 2 * Copyright (c) 2002 Doug Rabson 3 * Copyright (c) 1994-1995 S�ren Schmidt 4 * 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 * in this position and unchanged. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. The name of the author may not be used to endorse or promote products 16 * derived from this software without specific prior written permission 17 * 18 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 19 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 20 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 21 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 22 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 23 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 24 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 25 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 26 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 27 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 28 */ 29 30 #include <sys/cdefs.h> 31 __FBSDID("$FreeBSD$"); 32 33 #include "opt_compat.h" 34 #include "opt_mac.h" 35 36 #include <sys/param.h> 37 #include <sys/blist.h> 38 #include <sys/fcntl.h> 39 #if defined(__i386__) 40 #include <sys/imgact_aout.h> 41 #endif 42 #include <sys/jail.h> 43 #include <sys/kernel.h> 44 #include <sys/limits.h> 45 #include <sys/lock.h> 46 #include <sys/malloc.h> 47 #include <sys/mman.h> 48 #include <sys/mount.h> 49 #include <sys/mutex.h> 50 #include <sys/namei.h> 51 #include <sys/priv.h> 52 #include <sys/proc.h> 53 #include <sys/reboot.h> 54 #include <sys/resourcevar.h> 55 #include <sys/signalvar.h> 56 #include <sys/stat.h> 57 #include <sys/syscallsubr.h> 58 #include <sys/sysctl.h> 59 #include <sys/sysproto.h> 60 #include <sys/systm.h> 61 #include <sys/time.h> 62 #include <sys/vmmeter.h> 63 #include <sys/vnode.h> 64 #include <sys/wait.h> 65 66 #include <security/mac/mac_framework.h> 67 68 #include <vm/vm.h> 69 #include <vm/pmap.h> 70 #include <vm/vm_kern.h> 71 #include <vm/vm_map.h> 72 #include <vm/vm_extern.h> 73 #include <vm/vm_object.h> 74 #include <vm/swap_pager.h> 75 76 #include <posix4/sched.h> 77 78 #include <compat/linux/linux_sysproto.h> 79 #include <compat/linux/linux_emul.h> 80 #include <compat/linux/linux_misc.h> 81 82 #ifdef COMPAT_LINUX32 83 #include <machine/../linux32/linux.h> 84 #include <machine/../linux32/linux32_proto.h> 85 #else 86 #include <machine/../linux/linux.h> 87 #include <machine/../linux/linux_proto.h> 88 #endif 89 90 #include <compat/linux/linux_mib.h> 91 #include <compat/linux/linux_signal.h> 92 #include <compat/linux/linux_util.h> 93 94 #ifdef __i386__ 95 #include <machine/cputypes.h> 96 #endif 97 98 #define BSD_TO_LINUX_SIGNAL(sig) \ 99 (((sig) <= LINUX_SIGTBLSZ) ? bsd_to_linux_signal[_SIG_IDX(sig)] : sig) 100 101 static unsigned int linux_to_bsd_resource[LINUX_RLIM_NLIMITS] = { 102 RLIMIT_CPU, RLIMIT_FSIZE, RLIMIT_DATA, RLIMIT_STACK, 103 RLIMIT_CORE, RLIMIT_RSS, RLIMIT_NPROC, RLIMIT_NOFILE, 104 RLIMIT_MEMLOCK, -1 105 }; 106 107 struct l_sysinfo { 108 l_long uptime; /* Seconds since boot */ 109 l_ulong loads[3]; /* 1, 5, and 15 minute load averages */ 110 #define LINUX_SYSINFO_LOADS_SCALE 65536 111 l_ulong totalram; /* Total usable main memory size */ 112 l_ulong freeram; /* Available memory size */ 113 l_ulong sharedram; /* Amount of shared memory */ 114 l_ulong bufferram; /* Memory used by buffers */ 115 l_ulong totalswap; /* Total swap space size */ 116 l_ulong freeswap; /* swap space still available */ 117 l_ushort procs; /* Number of current processes */ 118 l_ulong totalbig; 119 l_ulong freebig; 120 l_uint mem_unit; 121 char _f[6]; /* Pads structure to 64 bytes */ 122 }; 123 int 124 linux_sysinfo(struct thread *td, struct linux_sysinfo_args *args) 125 { 126 struct l_sysinfo sysinfo; 127 vm_object_t object; 128 int i, j; 129 struct timespec ts; 130 131 getnanouptime(&ts); 132 if (ts.tv_nsec != 0) 133 ts.tv_sec++; 134 sysinfo.uptime = ts.tv_sec; 135 136 /* Use the information from the mib to get our load averages */ 137 for (i = 0; i < 3; i++) 138 sysinfo.loads[i] = averunnable.ldavg[i] * 139 LINUX_SYSINFO_LOADS_SCALE / averunnable.fscale; 140 141 sysinfo.totalram = physmem * PAGE_SIZE; 142 sysinfo.freeram = sysinfo.totalram - cnt.v_wire_count * PAGE_SIZE; 143 144 sysinfo.sharedram = 0; 145 mtx_lock(&vm_object_list_mtx); 146 TAILQ_FOREACH(object, &vm_object_list, object_list) 147 if (object->shadow_count > 1) 148 sysinfo.sharedram += object->resident_page_count; 149 mtx_unlock(&vm_object_list_mtx); 150 151 sysinfo.sharedram *= PAGE_SIZE; 152 sysinfo.bufferram = 0; 153 154 swap_pager_status(&i, &j); 155 sysinfo.totalswap= i * PAGE_SIZE; 156 sysinfo.freeswap = (i - j) * PAGE_SIZE; 157 158 sysinfo.procs = nprocs; 159 160 /* The following are only present in newer Linux kernels. */ 161 sysinfo.totalbig = 0; 162 sysinfo.freebig = 0; 163 sysinfo.mem_unit = 1; 164 165 return copyout(&sysinfo, args->info, sizeof(sysinfo)); 166 } 167 168 int 169 linux_alarm(struct thread *td, struct linux_alarm_args *args) 170 { 171 struct itimerval it, old_it; 172 int error; 173 174 #ifdef DEBUG 175 if (ldebug(alarm)) 176 printf(ARGS(alarm, "%u"), args->secs); 177 #endif 178 179 if (args->secs > 100000000) 180 return (EINVAL); 181 182 it.it_value.tv_sec = (long)args->secs; 183 it.it_value.tv_usec = 0; 184 it.it_interval.tv_sec = 0; 185 it.it_interval.tv_usec = 0; 186 error = kern_setitimer(td, ITIMER_REAL, &it, &old_it); 187 if (error) 188 return (error); 189 if (timevalisset(&old_it.it_value)) { 190 if (old_it.it_value.tv_usec != 0) 191 old_it.it_value.tv_sec++; 192 td->td_retval[0] = old_it.it_value.tv_sec; 193 } 194 return (0); 195 } 196 197 int 198 linux_brk(struct thread *td, struct linux_brk_args *args) 199 { 200 struct vmspace *vm = td->td_proc->p_vmspace; 201 vm_offset_t new, old; 202 struct obreak_args /* { 203 char * nsize; 204 } */ tmp; 205 206 #ifdef DEBUG 207 if (ldebug(brk)) 208 printf(ARGS(brk, "%p"), (void *)(uintptr_t)args->dsend); 209 #endif 210 old = (vm_offset_t)vm->vm_daddr + ctob(vm->vm_dsize); 211 new = (vm_offset_t)args->dsend; 212 tmp.nsize = (char *) new; 213 if (((caddr_t)new > vm->vm_daddr) && !obreak(td, &tmp)) 214 td->td_retval[0] = (long)new; 215 else 216 td->td_retval[0] = (long)old; 217 218 return 0; 219 } 220 221 #if defined(__i386__) 222 /* XXX: what about amd64/linux32? */ 223 224 int 225 linux_uselib(struct thread *td, struct linux_uselib_args *args) 226 { 227 struct nameidata ni; 228 struct vnode *vp; 229 struct exec *a_out; 230 struct vattr attr; 231 vm_offset_t vmaddr; 232 unsigned long file_offset; 233 vm_offset_t buffer; 234 unsigned long bss_size; 235 char *library; 236 int error; 237 int locked, vfslocked; 238 239 LCONVPATHEXIST(td, args->library, &library); 240 241 #ifdef DEBUG 242 if (ldebug(uselib)) 243 printf(ARGS(uselib, "%s"), library); 244 #endif 245 246 a_out = NULL; 247 vfslocked = 0; 248 locked = 0; 249 vp = NULL; 250 251 NDINIT(&ni, LOOKUP, ISOPEN | FOLLOW | LOCKLEAF | MPSAFE | AUDITVNODE1, 252 UIO_SYSSPACE, library, td); 253 error = namei(&ni); 254 LFREEPATH(library); 255 if (error) 256 goto cleanup; 257 258 vp = ni.ni_vp; 259 vfslocked = NDHASGIANT(&ni); 260 NDFREE(&ni, NDF_ONLY_PNBUF); 261 262 /* 263 * From here on down, we have a locked vnode that must be unlocked. 264 * XXX: The code below largely duplicates exec_check_permissions(). 265 */ 266 locked = 1; 267 268 /* Writable? */ 269 if (vp->v_writecount) { 270 error = ETXTBSY; 271 goto cleanup; 272 } 273 274 /* Executable? */ 275 error = VOP_GETATTR(vp, &attr, td->td_ucred, td); 276 if (error) 277 goto cleanup; 278 279 if ((vp->v_mount->mnt_flag & MNT_NOEXEC) || 280 ((attr.va_mode & 0111) == 0) || (attr.va_type != VREG)) { 281 /* EACCESS is what exec(2) returns. */ 282 error = ENOEXEC; 283 goto cleanup; 284 } 285 286 /* Sensible size? */ 287 if (attr.va_size == 0) { 288 error = ENOEXEC; 289 goto cleanup; 290 } 291 292 /* Can we access it? */ 293 error = VOP_ACCESS(vp, VEXEC, td->td_ucred, td); 294 if (error) 295 goto cleanup; 296 297 /* 298 * XXX: This should use vn_open() so that it is properly authorized, 299 * and to reduce code redundancy all over the place here. 300 * XXX: Not really, it duplicates far more of exec_check_permissions() 301 * than vn_open(). 302 */ 303 #ifdef MAC 304 error = mac_check_vnode_open(td->td_ucred, vp, FREAD); 305 if (error) 306 goto cleanup; 307 #endif 308 error = VOP_OPEN(vp, FREAD, td->td_ucred, td, -1); 309 if (error) 310 goto cleanup; 311 312 /* Pull in executable header into kernel_map */ 313 error = vm_mmap(kernel_map, (vm_offset_t *)&a_out, PAGE_SIZE, 314 VM_PROT_READ, VM_PROT_READ, 0, OBJT_VNODE, vp, 0); 315 if (error) 316 goto cleanup; 317 318 /* Is it a Linux binary ? */ 319 if (((a_out->a_magic >> 16) & 0xff) != 0x64) { 320 error = ENOEXEC; 321 goto cleanup; 322 } 323 324 /* 325 * While we are here, we should REALLY do some more checks 326 */ 327 328 /* Set file/virtual offset based on a.out variant. */ 329 switch ((int)(a_out->a_magic & 0xffff)) { 330 case 0413: /* ZMAGIC */ 331 file_offset = 1024; 332 break; 333 case 0314: /* QMAGIC */ 334 file_offset = 0; 335 break; 336 default: 337 error = ENOEXEC; 338 goto cleanup; 339 } 340 341 bss_size = round_page(a_out->a_bss); 342 343 /* Check various fields in header for validity/bounds. */ 344 if (a_out->a_text & PAGE_MASK || a_out->a_data & PAGE_MASK) { 345 error = ENOEXEC; 346 goto cleanup; 347 } 348 349 /* text + data can't exceed file size */ 350 if (a_out->a_data + a_out->a_text > attr.va_size) { 351 error = EFAULT; 352 goto cleanup; 353 } 354 355 /* 356 * text/data/bss must not exceed limits 357 * XXX - this is not complete. it should check current usage PLUS 358 * the resources needed by this library. 359 */ 360 PROC_LOCK(td->td_proc); 361 if (a_out->a_text > maxtsiz || 362 a_out->a_data + bss_size > lim_cur(td->td_proc, RLIMIT_DATA)) { 363 PROC_UNLOCK(td->td_proc); 364 error = ENOMEM; 365 goto cleanup; 366 } 367 PROC_UNLOCK(td->td_proc); 368 369 /* 370 * Prevent more writers. 371 * XXX: Note that if any of the VM operations fail below we don't 372 * clear this flag. 373 */ 374 vp->v_vflag |= VV_TEXT; 375 376 /* 377 * Lock no longer needed 378 */ 379 locked = 0; 380 VOP_UNLOCK(vp, 0, td); 381 VFS_UNLOCK_GIANT(vfslocked); 382 383 /* 384 * Check if file_offset page aligned. Currently we cannot handle 385 * misalinged file offsets, and so we read in the entire image 386 * (what a waste). 387 */ 388 if (file_offset & PAGE_MASK) { 389 #ifdef DEBUG 390 printf("uselib: Non page aligned binary %lu\n", file_offset); 391 #endif 392 /* Map text+data read/write/execute */ 393 394 /* a_entry is the load address and is page aligned */ 395 vmaddr = trunc_page(a_out->a_entry); 396 397 /* get anon user mapping, read+write+execute */ 398 error = vm_map_find(&td->td_proc->p_vmspace->vm_map, NULL, 0, 399 &vmaddr, a_out->a_text + a_out->a_data, FALSE, VM_PROT_ALL, 400 VM_PROT_ALL, 0); 401 if (error) 402 goto cleanup; 403 404 /* map file into kernel_map */ 405 error = vm_mmap(kernel_map, &buffer, 406 round_page(a_out->a_text + a_out->a_data + file_offset), 407 VM_PROT_READ, VM_PROT_READ, 0, OBJT_VNODE, vp, 408 trunc_page(file_offset)); 409 if (error) 410 goto cleanup; 411 412 /* copy from kernel VM space to user space */ 413 error = copyout(PTRIN(buffer + file_offset), 414 (void *)vmaddr, a_out->a_text + a_out->a_data); 415 416 /* release temporary kernel space */ 417 vm_map_remove(kernel_map, buffer, buffer + 418 round_page(a_out->a_text + a_out->a_data + file_offset)); 419 420 if (error) 421 goto cleanup; 422 } else { 423 #ifdef DEBUG 424 printf("uselib: Page aligned binary %lu\n", file_offset); 425 #endif 426 /* 427 * for QMAGIC, a_entry is 20 bytes beyond the load address 428 * to skip the executable header 429 */ 430 vmaddr = trunc_page(a_out->a_entry); 431 432 /* 433 * Map it all into the process's space as a single 434 * copy-on-write "data" segment. 435 */ 436 error = vm_mmap(&td->td_proc->p_vmspace->vm_map, &vmaddr, 437 a_out->a_text + a_out->a_data, VM_PROT_ALL, VM_PROT_ALL, 438 MAP_PRIVATE | MAP_FIXED, OBJT_VNODE, vp, file_offset); 439 if (error) 440 goto cleanup; 441 } 442 #ifdef DEBUG 443 printf("mem=%08lx = %08lx %08lx\n", (long)vmaddr, ((long*)vmaddr)[0], 444 ((long*)vmaddr)[1]); 445 #endif 446 if (bss_size != 0) { 447 /* Calculate BSS start address */ 448 vmaddr = trunc_page(a_out->a_entry) + a_out->a_text + 449 a_out->a_data; 450 451 /* allocate some 'anon' space */ 452 error = vm_map_find(&td->td_proc->p_vmspace->vm_map, NULL, 0, 453 &vmaddr, bss_size, FALSE, VM_PROT_ALL, VM_PROT_ALL, 0); 454 if (error) 455 goto cleanup; 456 } 457 458 cleanup: 459 /* Unlock vnode if needed */ 460 if (locked) { 461 VOP_UNLOCK(vp, 0, td); 462 VFS_UNLOCK_GIANT(vfslocked); 463 } 464 465 /* Release the kernel mapping. */ 466 if (a_out) 467 vm_map_remove(kernel_map, (vm_offset_t)a_out, 468 (vm_offset_t)a_out + PAGE_SIZE); 469 470 return error; 471 } 472 473 #endif /* __i386__ */ 474 475 int 476 linux_select(struct thread *td, struct linux_select_args *args) 477 { 478 l_timeval ltv; 479 struct timeval tv0, tv1, utv, *tvp; 480 int error; 481 482 #ifdef DEBUG 483 if (ldebug(select)) 484 printf(ARGS(select, "%d, %p, %p, %p, %p"), args->nfds, 485 (void *)args->readfds, (void *)args->writefds, 486 (void *)args->exceptfds, (void *)args->timeout); 487 #endif 488 489 /* 490 * Store current time for computation of the amount of 491 * time left. 492 */ 493 if (args->timeout) { 494 if ((error = copyin(args->timeout, <v, sizeof(ltv)))) 495 goto select_out; 496 utv.tv_sec = ltv.tv_sec; 497 utv.tv_usec = ltv.tv_usec; 498 #ifdef DEBUG 499 if (ldebug(select)) 500 printf(LMSG("incoming timeout (%jd/%ld)"), 501 (intmax_t)utv.tv_sec, utv.tv_usec); 502 #endif 503 504 if (itimerfix(&utv)) { 505 /* 506 * The timeval was invalid. Convert it to something 507 * valid that will act as it does under Linux. 508 */ 509 utv.tv_sec += utv.tv_usec / 1000000; 510 utv.tv_usec %= 1000000; 511 if (utv.tv_usec < 0) { 512 utv.tv_sec -= 1; 513 utv.tv_usec += 1000000; 514 } 515 if (utv.tv_sec < 0) 516 timevalclear(&utv); 517 } 518 microtime(&tv0); 519 tvp = &utv; 520 } else 521 tvp = NULL; 522 523 error = kern_select(td, args->nfds, args->readfds, args->writefds, 524 args->exceptfds, tvp); 525 526 #ifdef DEBUG 527 if (ldebug(select)) 528 printf(LMSG("real select returns %d"), error); 529 #endif 530 if (error) { 531 /* 532 * See fs/select.c in the Linux kernel. Without this, 533 * Maelstrom doesn't work. 534 */ 535 if (error == ERESTART) 536 error = EINTR; 537 goto select_out; 538 } 539 540 if (args->timeout) { 541 if (td->td_retval[0]) { 542 /* 543 * Compute how much time was left of the timeout, 544 * by subtracting the current time and the time 545 * before we started the call, and subtracting 546 * that result from the user-supplied value. 547 */ 548 microtime(&tv1); 549 timevalsub(&tv1, &tv0); 550 timevalsub(&utv, &tv1); 551 if (utv.tv_sec < 0) 552 timevalclear(&utv); 553 } else 554 timevalclear(&utv); 555 #ifdef DEBUG 556 if (ldebug(select)) 557 printf(LMSG("outgoing timeout (%jd/%ld)"), 558 (intmax_t)utv.tv_sec, utv.tv_usec); 559 #endif 560 ltv.tv_sec = utv.tv_sec; 561 ltv.tv_usec = utv.tv_usec; 562 if ((error = copyout(<v, args->timeout, sizeof(ltv)))) 563 goto select_out; 564 } 565 566 select_out: 567 #ifdef DEBUG 568 if (ldebug(select)) 569 printf(LMSG("select_out -> %d"), error); 570 #endif 571 return error; 572 } 573 574 int 575 linux_mremap(struct thread *td, struct linux_mremap_args *args) 576 { 577 struct munmap_args /* { 578 void *addr; 579 size_t len; 580 } */ bsd_args; 581 int error = 0; 582 583 #ifdef DEBUG 584 if (ldebug(mremap)) 585 printf(ARGS(mremap, "%p, %08lx, %08lx, %08lx"), 586 (void *)(uintptr_t)args->addr, 587 (unsigned long)args->old_len, 588 (unsigned long)args->new_len, 589 (unsigned long)args->flags); 590 #endif 591 args->new_len = round_page(args->new_len); 592 args->old_len = round_page(args->old_len); 593 594 if (args->new_len > args->old_len) { 595 td->td_retval[0] = 0; 596 return ENOMEM; 597 } 598 599 if (args->new_len < args->old_len) { 600 bsd_args.addr = 601 (caddr_t)((uintptr_t)args->addr + args->new_len); 602 bsd_args.len = args->old_len - args->new_len; 603 error = munmap(td, &bsd_args); 604 } 605 606 td->td_retval[0] = error ? 0 : (uintptr_t)args->addr; 607 return error; 608 } 609 610 #define LINUX_MS_ASYNC 0x0001 611 #define LINUX_MS_INVALIDATE 0x0002 612 #define LINUX_MS_SYNC 0x0004 613 614 int 615 linux_msync(struct thread *td, struct linux_msync_args *args) 616 { 617 struct msync_args bsd_args; 618 619 bsd_args.addr = (caddr_t)(uintptr_t)args->addr; 620 bsd_args.len = (uintptr_t)args->len; 621 bsd_args.flags = args->fl & ~LINUX_MS_SYNC; 622 623 return msync(td, &bsd_args); 624 } 625 626 int 627 linux_time(struct thread *td, struct linux_time_args *args) 628 { 629 struct timeval tv; 630 l_time_t tm; 631 int error; 632 633 #ifdef DEBUG 634 if (ldebug(time)) 635 printf(ARGS(time, "*")); 636 #endif 637 638 microtime(&tv); 639 tm = tv.tv_sec; 640 if (args->tm && (error = copyout(&tm, args->tm, sizeof(tm)))) 641 return error; 642 td->td_retval[0] = tm; 643 return 0; 644 } 645 646 struct l_times_argv { 647 l_long tms_utime; 648 l_long tms_stime; 649 l_long tms_cutime; 650 l_long tms_cstime; 651 }; 652 653 #define CLK_TCK 100 /* Linux uses 100 */ 654 655 #define CONVTCK(r) (r.tv_sec * CLK_TCK + r.tv_usec / (1000000 / CLK_TCK)) 656 657 int 658 linux_times(struct thread *td, struct linux_times_args *args) 659 { 660 struct timeval tv, utime, stime, cutime, cstime; 661 struct l_times_argv tms; 662 struct proc *p; 663 int error; 664 665 #ifdef DEBUG 666 if (ldebug(times)) 667 printf(ARGS(times, "*")); 668 #endif 669 670 if (args->buf != NULL) { 671 p = td->td_proc; 672 PROC_LOCK(p); 673 calcru(p, &utime, &stime); 674 calccru(p, &cutime, &cstime); 675 PROC_UNLOCK(p); 676 677 tms.tms_utime = CONVTCK(utime); 678 tms.tms_stime = CONVTCK(stime); 679 680 tms.tms_cutime = CONVTCK(cutime); 681 tms.tms_cstime = CONVTCK(cstime); 682 683 if ((error = copyout(&tms, args->buf, sizeof(tms)))) 684 return error; 685 } 686 687 microuptime(&tv); 688 td->td_retval[0] = (int)CONVTCK(tv); 689 return 0; 690 } 691 692 int 693 linux_newuname(struct thread *td, struct linux_newuname_args *args) 694 { 695 struct l_new_utsname utsname; 696 char osname[LINUX_MAX_UTSNAME]; 697 char osrelease[LINUX_MAX_UTSNAME]; 698 char *p; 699 700 #ifdef DEBUG 701 if (ldebug(newuname)) 702 printf(ARGS(newuname, "*")); 703 #endif 704 705 linux_get_osname(td, osname); 706 linux_get_osrelease(td, osrelease); 707 708 bzero(&utsname, sizeof(utsname)); 709 strlcpy(utsname.sysname, osname, LINUX_MAX_UTSNAME); 710 getcredhostname(td->td_ucred, utsname.nodename, LINUX_MAX_UTSNAME); 711 strlcpy(utsname.release, osrelease, LINUX_MAX_UTSNAME); 712 strlcpy(utsname.version, version, LINUX_MAX_UTSNAME); 713 for (p = utsname.version; *p != '\0'; ++p) 714 if (*p == '\n') { 715 *p = '\0'; 716 break; 717 } 718 #ifdef __i386__ 719 { 720 const char *class; 721 switch (cpu_class) { 722 case CPUCLASS_686: 723 class = "i686"; 724 break; 725 case CPUCLASS_586: 726 class = "i586"; 727 break; 728 case CPUCLASS_486: 729 class = "i486"; 730 break; 731 default: 732 class = "i386"; 733 } 734 strlcpy(utsname.machine, class, LINUX_MAX_UTSNAME); 735 } 736 #elif defined(__amd64__) /* XXX: Linux can change 'personality'. */ 737 #ifdef COMPAT_LINUX32 738 strlcpy(utsname.machine, "i686", LINUX_MAX_UTSNAME); 739 #else 740 strlcpy(utsname.machine, "x86_64", LINUX_MAX_UTSNAME); 741 #endif /* COMPAT_LINUX32 */ 742 #else /* something other than i386 or amd64 - assume we and Linux agree */ 743 strlcpy(utsname.machine, machine, LINUX_MAX_UTSNAME); 744 #endif /* __i386__ */ 745 strlcpy(utsname.domainname, domainname, LINUX_MAX_UTSNAME); 746 747 return (copyout(&utsname, args->buf, sizeof(utsname))); 748 } 749 750 #if defined(__i386__) || (defined(__amd64__) && defined(COMPAT_LINUX32)) 751 struct l_utimbuf { 752 l_time_t l_actime; 753 l_time_t l_modtime; 754 }; 755 756 int 757 linux_utime(struct thread *td, struct linux_utime_args *args) 758 { 759 struct timeval tv[2], *tvp; 760 struct l_utimbuf lut; 761 char *fname; 762 int error; 763 764 LCONVPATHEXIST(td, args->fname, &fname); 765 766 #ifdef DEBUG 767 if (ldebug(utime)) 768 printf(ARGS(utime, "%s, *"), fname); 769 #endif 770 771 if (args->times) { 772 if ((error = copyin(args->times, &lut, sizeof lut))) { 773 LFREEPATH(fname); 774 return error; 775 } 776 tv[0].tv_sec = lut.l_actime; 777 tv[0].tv_usec = 0; 778 tv[1].tv_sec = lut.l_modtime; 779 tv[1].tv_usec = 0; 780 tvp = tv; 781 } else 782 tvp = NULL; 783 784 error = kern_utimes(td, fname, UIO_SYSSPACE, tvp, UIO_SYSSPACE); 785 LFREEPATH(fname); 786 return (error); 787 } 788 #endif /* __i386__ || (__amd64__ && COMPAT_LINUX32) */ 789 790 #define __WCLONE 0x80000000 791 792 int 793 linux_waitpid(struct thread *td, struct linux_waitpid_args *args) 794 { 795 int error, options, tmpstat; 796 797 #ifdef DEBUG 798 if (ldebug(waitpid)) 799 printf(ARGS(waitpid, "%d, %p, %d"), 800 args->pid, (void *)args->status, args->options); 801 #endif 802 /* 803 * this is necessary because the test in kern_wait doesnt 804 * work because we mess with the options here 805 */ 806 if (args->options &~ (WUNTRACED|WNOHANG|WCONTINUED)) 807 return (EINVAL); 808 809 options = (args->options & (WNOHANG | WUNTRACED)); 810 /* WLINUXCLONE should be equal to __WCLONE, but we make sure */ 811 if (args->options & __WCLONE) 812 options |= WLINUXCLONE; 813 814 error = kern_wait(td, args->pid, &tmpstat, options, NULL); 815 if (error) 816 return error; 817 818 if (args->status) { 819 tmpstat &= 0xffff; 820 if (WIFSIGNALED(tmpstat)) 821 tmpstat = (tmpstat & 0xffffff80) | 822 BSD_TO_LINUX_SIGNAL(WTERMSIG(tmpstat)); 823 else if (WIFSTOPPED(tmpstat)) 824 tmpstat = (tmpstat & 0xffff00ff) | 825 (BSD_TO_LINUX_SIGNAL(WSTOPSIG(tmpstat)) << 8); 826 return copyout(&tmpstat, args->status, sizeof(int)); 827 } 828 829 return 0; 830 } 831 832 int 833 linux_wait4(struct thread *td, struct linux_wait4_args *args) 834 { 835 int error, options, tmpstat; 836 struct rusage ru, *rup; 837 struct proc *p; 838 839 #ifdef DEBUG 840 if (ldebug(wait4)) 841 printf(ARGS(wait4, "%d, %p, %d, %p"), 842 args->pid, (void *)args->status, args->options, 843 (void *)args->rusage); 844 #endif 845 846 options = (args->options & (WNOHANG | WUNTRACED)); 847 /* WLINUXCLONE should be equal to __WCLONE, but we make sure */ 848 if (args->options & __WCLONE) 849 options |= WLINUXCLONE; 850 851 if (args->rusage != NULL) 852 rup = &ru; 853 else 854 rup = NULL; 855 error = kern_wait(td, args->pid, &tmpstat, options, rup); 856 if (error) 857 return error; 858 859 p = td->td_proc; 860 PROC_LOCK(p); 861 sigqueue_delete(&p->p_sigqueue, SIGCHLD); 862 PROC_UNLOCK(p); 863 864 if (args->status) { 865 tmpstat &= 0xffff; 866 if (WIFSIGNALED(tmpstat)) 867 tmpstat = (tmpstat & 0xffffff80) | 868 BSD_TO_LINUX_SIGNAL(WTERMSIG(tmpstat)); 869 else if (WIFSTOPPED(tmpstat)) 870 tmpstat = (tmpstat & 0xffff00ff) | 871 (BSD_TO_LINUX_SIGNAL(WSTOPSIG(tmpstat)) << 8); 872 error = copyout(&tmpstat, args->status, sizeof(int)); 873 } 874 if (args->rusage != NULL && error == 0) 875 error = copyout(&ru, args->rusage, sizeof(ru)); 876 877 return (error); 878 } 879 880 int 881 linux_mknod(struct thread *td, struct linux_mknod_args *args) 882 { 883 char *path; 884 int error; 885 886 LCONVPATHCREAT(td, args->path, &path); 887 888 #ifdef DEBUG 889 if (ldebug(mknod)) 890 printf(ARGS(mknod, "%s, %d, %d"), path, args->mode, args->dev); 891 #endif 892 893 if (S_ISFIFO(args->mode)) 894 error = kern_mkfifo(td, path, UIO_SYSSPACE, args->mode); 895 else 896 error = kern_mknod(td, path, UIO_SYSSPACE, args->mode, 897 args->dev); 898 LFREEPATH(path); 899 return (error); 900 } 901 902 /* 903 * UGH! This is just about the dumbest idea I've ever heard!! 904 */ 905 int 906 linux_personality(struct thread *td, struct linux_personality_args *args) 907 { 908 #ifdef DEBUG 909 if (ldebug(personality)) 910 printf(ARGS(personality, "%lu"), (unsigned long)args->per); 911 #endif 912 if (args->per != 0) 913 return EINVAL; 914 915 /* Yes Jim, it's still a Linux... */ 916 td->td_retval[0] = 0; 917 return 0; 918 } 919 920 struct l_itimerval { 921 l_timeval it_interval; 922 l_timeval it_value; 923 }; 924 925 #define B2L_ITIMERVAL(bip, lip) \ 926 (bip)->it_interval.tv_sec = (lip)->it_interval.tv_sec; \ 927 (bip)->it_interval.tv_usec = (lip)->it_interval.tv_usec; \ 928 (bip)->it_value.tv_sec = (lip)->it_value.tv_sec; \ 929 (bip)->it_value.tv_usec = (lip)->it_value.tv_usec; 930 931 int 932 linux_setitimer(struct thread *td, struct linux_setitimer_args *uap) 933 { 934 int error; 935 struct l_itimerval ls; 936 struct itimerval aitv, oitv; 937 938 #ifdef DEBUG 939 if (ldebug(setitimer)) 940 printf(ARGS(setitimer, "%p, %p"), 941 (void *)uap->itv, (void *)uap->oitv); 942 #endif 943 944 if (uap->itv == NULL) { 945 uap->itv = uap->oitv; 946 return (linux_getitimer(td, (struct linux_getitimer_args *)uap)); 947 } 948 949 error = copyin(uap->itv, &ls, sizeof(ls)); 950 if (error != 0) 951 return (error); 952 B2L_ITIMERVAL(&aitv, &ls); 953 #ifdef DEBUG 954 if (ldebug(setitimer)) { 955 printf("setitimer: value: sec: %jd, usec: %ld\n", 956 (intmax_t)aitv.it_value.tv_sec, aitv.it_value.tv_usec); 957 printf("setitimer: interval: sec: %jd, usec: %ld\n", 958 (intmax_t)aitv.it_interval.tv_sec, aitv.it_interval.tv_usec); 959 } 960 #endif 961 error = kern_setitimer(td, uap->which, &aitv, &oitv); 962 if (error != 0 || uap->oitv == NULL) 963 return (error); 964 B2L_ITIMERVAL(&ls, &oitv); 965 966 return (copyout(&ls, uap->oitv, sizeof(ls))); 967 } 968 969 int 970 linux_getitimer(struct thread *td, struct linux_getitimer_args *uap) 971 { 972 int error; 973 struct l_itimerval ls; 974 struct itimerval aitv; 975 976 #ifdef DEBUG 977 if (ldebug(getitimer)) 978 printf(ARGS(getitimer, "%p"), (void *)uap->itv); 979 #endif 980 error = kern_getitimer(td, uap->which, &aitv); 981 if (error != 0) 982 return (error); 983 B2L_ITIMERVAL(&ls, &aitv); 984 return (copyout(&ls, uap->itv, sizeof(ls))); 985 } 986 987 int 988 linux_nice(struct thread *td, struct linux_nice_args *args) 989 { 990 struct setpriority_args bsd_args; 991 992 bsd_args.which = PRIO_PROCESS; 993 bsd_args.who = 0; /* current process */ 994 bsd_args.prio = args->inc; 995 return setpriority(td, &bsd_args); 996 } 997 998 int 999 linux_setgroups(struct thread *td, struct linux_setgroups_args *args) 1000 { 1001 struct ucred *newcred, *oldcred; 1002 l_gid_t linux_gidset[NGROUPS]; 1003 gid_t *bsd_gidset; 1004 int ngrp, error; 1005 struct proc *p; 1006 1007 ngrp = args->gidsetsize; 1008 if (ngrp < 0 || ngrp >= NGROUPS) 1009 return (EINVAL); 1010 error = copyin(args->grouplist, linux_gidset, ngrp * sizeof(l_gid_t)); 1011 if (error) 1012 return (error); 1013 newcred = crget(); 1014 p = td->td_proc; 1015 PROC_LOCK(p); 1016 oldcred = p->p_ucred; 1017 1018 /* 1019 * cr_groups[0] holds egid. Setting the whole set from 1020 * the supplied set will cause egid to be changed too. 1021 * Keep cr_groups[0] unchanged to prevent that. 1022 */ 1023 1024 if ((error = priv_check_cred(oldcred, PRIV_CRED_SETGROUPS, 1025 SUSER_ALLOWJAIL)) != 0) { 1026 PROC_UNLOCK(p); 1027 crfree(newcred); 1028 return (error); 1029 } 1030 1031 crcopy(newcred, oldcred); 1032 if (ngrp > 0) { 1033 newcred->cr_ngroups = ngrp + 1; 1034 1035 bsd_gidset = newcred->cr_groups; 1036 ngrp--; 1037 while (ngrp >= 0) { 1038 bsd_gidset[ngrp + 1] = linux_gidset[ngrp]; 1039 ngrp--; 1040 } 1041 } 1042 else 1043 newcred->cr_ngroups = 1; 1044 1045 setsugid(p); 1046 p->p_ucred = newcred; 1047 PROC_UNLOCK(p); 1048 crfree(oldcred); 1049 return (0); 1050 } 1051 1052 int 1053 linux_getgroups(struct thread *td, struct linux_getgroups_args *args) 1054 { 1055 struct ucred *cred; 1056 l_gid_t linux_gidset[NGROUPS]; 1057 gid_t *bsd_gidset; 1058 int bsd_gidsetsz, ngrp, error; 1059 1060 cred = td->td_ucred; 1061 bsd_gidset = cred->cr_groups; 1062 bsd_gidsetsz = cred->cr_ngroups - 1; 1063 1064 /* 1065 * cr_groups[0] holds egid. Returning the whole set 1066 * here will cause a duplicate. Exclude cr_groups[0] 1067 * to prevent that. 1068 */ 1069 1070 if ((ngrp = args->gidsetsize) == 0) { 1071 td->td_retval[0] = bsd_gidsetsz; 1072 return (0); 1073 } 1074 1075 if (ngrp < bsd_gidsetsz) 1076 return (EINVAL); 1077 1078 ngrp = 0; 1079 while (ngrp < bsd_gidsetsz) { 1080 linux_gidset[ngrp] = bsd_gidset[ngrp + 1]; 1081 ngrp++; 1082 } 1083 1084 if ((error = copyout(linux_gidset, args->grouplist, 1085 ngrp * sizeof(l_gid_t)))) 1086 return (error); 1087 1088 td->td_retval[0] = ngrp; 1089 return (0); 1090 } 1091 1092 int 1093 linux_setrlimit(struct thread *td, struct linux_setrlimit_args *args) 1094 { 1095 struct rlimit bsd_rlim; 1096 struct l_rlimit rlim; 1097 u_int which; 1098 int error; 1099 1100 #ifdef DEBUG 1101 if (ldebug(setrlimit)) 1102 printf(ARGS(setrlimit, "%d, %p"), 1103 args->resource, (void *)args->rlim); 1104 #endif 1105 1106 if (args->resource >= LINUX_RLIM_NLIMITS) 1107 return (EINVAL); 1108 1109 which = linux_to_bsd_resource[args->resource]; 1110 if (which == -1) 1111 return (EINVAL); 1112 1113 error = copyin(args->rlim, &rlim, sizeof(rlim)); 1114 if (error) 1115 return (error); 1116 1117 bsd_rlim.rlim_cur = (rlim_t)rlim.rlim_cur; 1118 bsd_rlim.rlim_max = (rlim_t)rlim.rlim_max; 1119 return (kern_setrlimit(td, which, &bsd_rlim)); 1120 } 1121 1122 int 1123 linux_old_getrlimit(struct thread *td, struct linux_old_getrlimit_args *args) 1124 { 1125 struct l_rlimit rlim; 1126 struct proc *p = td->td_proc; 1127 struct rlimit bsd_rlim; 1128 u_int which; 1129 1130 #ifdef DEBUG 1131 if (ldebug(old_getrlimit)) 1132 printf(ARGS(old_getrlimit, "%d, %p"), 1133 args->resource, (void *)args->rlim); 1134 #endif 1135 1136 if (args->resource >= LINUX_RLIM_NLIMITS) 1137 return (EINVAL); 1138 1139 which = linux_to_bsd_resource[args->resource]; 1140 if (which == -1) 1141 return (EINVAL); 1142 1143 PROC_LOCK(p); 1144 lim_rlimit(p, which, &bsd_rlim); 1145 PROC_UNLOCK(p); 1146 1147 #ifdef COMPAT_LINUX32 1148 rlim.rlim_cur = (unsigned int)bsd_rlim.rlim_cur; 1149 if (rlim.rlim_cur == UINT_MAX) 1150 rlim.rlim_cur = INT_MAX; 1151 rlim.rlim_max = (unsigned int)bsd_rlim.rlim_max; 1152 if (rlim.rlim_max == UINT_MAX) 1153 rlim.rlim_max = INT_MAX; 1154 #else 1155 rlim.rlim_cur = (unsigned long)bsd_rlim.rlim_cur; 1156 if (rlim.rlim_cur == ULONG_MAX) 1157 rlim.rlim_cur = LONG_MAX; 1158 rlim.rlim_max = (unsigned long)bsd_rlim.rlim_max; 1159 if (rlim.rlim_max == ULONG_MAX) 1160 rlim.rlim_max = LONG_MAX; 1161 #endif 1162 return (copyout(&rlim, args->rlim, sizeof(rlim))); 1163 } 1164 1165 int 1166 linux_getrlimit(struct thread *td, struct linux_getrlimit_args *args) 1167 { 1168 struct l_rlimit rlim; 1169 struct proc *p = td->td_proc; 1170 struct rlimit bsd_rlim; 1171 u_int which; 1172 1173 #ifdef DEBUG 1174 if (ldebug(getrlimit)) 1175 printf(ARGS(getrlimit, "%d, %p"), 1176 args->resource, (void *)args->rlim); 1177 #endif 1178 1179 if (args->resource >= LINUX_RLIM_NLIMITS) 1180 return (EINVAL); 1181 1182 which = linux_to_bsd_resource[args->resource]; 1183 if (which == -1) 1184 return (EINVAL); 1185 1186 PROC_LOCK(p); 1187 lim_rlimit(p, which, &bsd_rlim); 1188 PROC_UNLOCK(p); 1189 1190 rlim.rlim_cur = (l_ulong)bsd_rlim.rlim_cur; 1191 rlim.rlim_max = (l_ulong)bsd_rlim.rlim_max; 1192 return (copyout(&rlim, args->rlim, sizeof(rlim))); 1193 } 1194 1195 int 1196 linux_sched_setscheduler(struct thread *td, 1197 struct linux_sched_setscheduler_args *args) 1198 { 1199 struct sched_setscheduler_args bsd; 1200 1201 #ifdef DEBUG 1202 if (ldebug(sched_setscheduler)) 1203 printf(ARGS(sched_setscheduler, "%d, %d, %p"), 1204 args->pid, args->policy, (const void *)args->param); 1205 #endif 1206 1207 switch (args->policy) { 1208 case LINUX_SCHED_OTHER: 1209 bsd.policy = SCHED_OTHER; 1210 break; 1211 case LINUX_SCHED_FIFO: 1212 bsd.policy = SCHED_FIFO; 1213 break; 1214 case LINUX_SCHED_RR: 1215 bsd.policy = SCHED_RR; 1216 break; 1217 default: 1218 return EINVAL; 1219 } 1220 1221 bsd.pid = args->pid; 1222 bsd.param = (struct sched_param *)args->param; 1223 return sched_setscheduler(td, &bsd); 1224 } 1225 1226 int 1227 linux_sched_getscheduler(struct thread *td, 1228 struct linux_sched_getscheduler_args *args) 1229 { 1230 struct sched_getscheduler_args bsd; 1231 int error; 1232 1233 #ifdef DEBUG 1234 if (ldebug(sched_getscheduler)) 1235 printf(ARGS(sched_getscheduler, "%d"), args->pid); 1236 #endif 1237 1238 bsd.pid = args->pid; 1239 error = sched_getscheduler(td, &bsd); 1240 1241 switch (td->td_retval[0]) { 1242 case SCHED_OTHER: 1243 td->td_retval[0] = LINUX_SCHED_OTHER; 1244 break; 1245 case SCHED_FIFO: 1246 td->td_retval[0] = LINUX_SCHED_FIFO; 1247 break; 1248 case SCHED_RR: 1249 td->td_retval[0] = LINUX_SCHED_RR; 1250 break; 1251 } 1252 1253 return error; 1254 } 1255 1256 int 1257 linux_sched_get_priority_max(struct thread *td, 1258 struct linux_sched_get_priority_max_args *args) 1259 { 1260 struct sched_get_priority_max_args bsd; 1261 1262 #ifdef DEBUG 1263 if (ldebug(sched_get_priority_max)) 1264 printf(ARGS(sched_get_priority_max, "%d"), args->policy); 1265 #endif 1266 1267 switch (args->policy) { 1268 case LINUX_SCHED_OTHER: 1269 bsd.policy = SCHED_OTHER; 1270 break; 1271 case LINUX_SCHED_FIFO: 1272 bsd.policy = SCHED_FIFO; 1273 break; 1274 case LINUX_SCHED_RR: 1275 bsd.policy = SCHED_RR; 1276 break; 1277 default: 1278 return EINVAL; 1279 } 1280 return sched_get_priority_max(td, &bsd); 1281 } 1282 1283 int 1284 linux_sched_get_priority_min(struct thread *td, 1285 struct linux_sched_get_priority_min_args *args) 1286 { 1287 struct sched_get_priority_min_args bsd; 1288 1289 #ifdef DEBUG 1290 if (ldebug(sched_get_priority_min)) 1291 printf(ARGS(sched_get_priority_min, "%d"), args->policy); 1292 #endif 1293 1294 switch (args->policy) { 1295 case LINUX_SCHED_OTHER: 1296 bsd.policy = SCHED_OTHER; 1297 break; 1298 case LINUX_SCHED_FIFO: 1299 bsd.policy = SCHED_FIFO; 1300 break; 1301 case LINUX_SCHED_RR: 1302 bsd.policy = SCHED_RR; 1303 break; 1304 default: 1305 return EINVAL; 1306 } 1307 return sched_get_priority_min(td, &bsd); 1308 } 1309 1310 #define REBOOT_CAD_ON 0x89abcdef 1311 #define REBOOT_CAD_OFF 0 1312 #define REBOOT_HALT 0xcdef0123 1313 #define REBOOT_RESTART 0x01234567 1314 #define REBOOT_RESTART2 0xA1B2C3D4 1315 #define REBOOT_POWEROFF 0x4321FEDC 1316 #define REBOOT_MAGIC1 0xfee1dead 1317 #define REBOOT_MAGIC2 0x28121969 1318 #define REBOOT_MAGIC2A 0x05121996 1319 #define REBOOT_MAGIC2B 0x16041998 1320 1321 int 1322 linux_reboot(struct thread *td, struct linux_reboot_args *args) 1323 { 1324 struct reboot_args bsd_args; 1325 1326 #ifdef DEBUG 1327 if (ldebug(reboot)) 1328 printf(ARGS(reboot, "0x%x"), args->cmd); 1329 #endif 1330 1331 if (args->magic1 != REBOOT_MAGIC1) 1332 return EINVAL; 1333 1334 switch (args->magic2) { 1335 case REBOOT_MAGIC2: 1336 case REBOOT_MAGIC2A: 1337 case REBOOT_MAGIC2B: 1338 break; 1339 default: 1340 return EINVAL; 1341 } 1342 1343 switch (args->cmd) { 1344 case REBOOT_CAD_ON: 1345 case REBOOT_CAD_OFF: 1346 return (priv_check(td, PRIV_REBOOT)); 1347 case REBOOT_HALT: 1348 bsd_args.opt = RB_HALT; 1349 break; 1350 case REBOOT_RESTART: 1351 case REBOOT_RESTART2: 1352 bsd_args.opt = 0; 1353 break; 1354 case REBOOT_POWEROFF: 1355 bsd_args.opt = RB_POWEROFF; 1356 break; 1357 default: 1358 return EINVAL; 1359 } 1360 return reboot(td, &bsd_args); 1361 } 1362 1363 1364 /* 1365 * The FreeBSD native getpid(2), getgid(2) and getuid(2) also modify 1366 * td->td_retval[1] when COMPAT_43 is defined. This 1367 * globbers registers that are assumed to be preserved. The following 1368 * lightweight syscalls fixes this. See also linux_getgid16() and 1369 * linux_getuid16() in linux_uid16.c. 1370 * 1371 * linux_getpid() - MP SAFE 1372 * linux_getgid() - MP SAFE 1373 * linux_getuid() - MP SAFE 1374 */ 1375 1376 int 1377 linux_getpid(struct thread *td, struct linux_getpid_args *args) 1378 { 1379 struct linux_emuldata *em; 1380 char osrel[LINUX_MAX_UTSNAME]; 1381 1382 linux_get_osrelease(td, osrel); 1383 if (strlen(osrel) >= 3 && osrel[2] == '6') { 1384 em = em_find(td->td_proc, EMUL_UNLOCKED); 1385 KASSERT(em != NULL, ("getpid: emuldata not found.\n")); 1386 td->td_retval[0] = em->shared->group_pid; 1387 EMUL_UNLOCK(&emul_lock); 1388 } else { 1389 PROC_LOCK(td->td_proc); 1390 td->td_retval[0] = td->td_proc->p_pid; 1391 PROC_UNLOCK(td->td_proc); 1392 } 1393 1394 return (0); 1395 } 1396 1397 int 1398 linux_gettid(struct thread *td, struct linux_gettid_args *args) 1399 { 1400 #ifdef DEBUG 1401 if (ldebug(gettid)) 1402 printf(ARGS(gettid, "")); 1403 #endif 1404 1405 td->td_retval[0] = td->td_proc->p_pid; 1406 return (0); 1407 } 1408 1409 1410 int 1411 linux_getppid(struct thread *td, struct linux_getppid_args *args) 1412 { 1413 struct linux_emuldata *em; 1414 struct proc *p, *pp; 1415 char osrel[LINUX_MAX_UTSNAME]; 1416 1417 linux_get_osrelease(td, osrel); 1418 if (strlen(osrel) >= 3 && osrel[2] != '6') { 1419 PROC_LOCK(td->td_proc); 1420 td->td_retval[0] = td->td_proc->p_pptr->p_pid; 1421 PROC_UNLOCK(td->td_proc); 1422 return (0); 1423 } 1424 1425 em = em_find(td->td_proc, EMUL_UNLOCKED); 1426 1427 KASSERT(em != NULL, ("getppid: process emuldata not found.\n")); 1428 1429 /* find the group leader */ 1430 p = pfind(em->shared->group_pid); 1431 1432 if (p == NULL) { 1433 #ifdef DEBUG 1434 printf(LMSG("parent process not found.\n")); 1435 #endif 1436 return (0); 1437 } 1438 1439 pp = p->p_pptr; /* switch to parent */ 1440 PROC_LOCK(pp); 1441 PROC_UNLOCK(p); 1442 1443 /* if its also linux process */ 1444 if (pp->p_sysent == &elf_linux_sysvec) { 1445 em = em_find(pp, EMUL_LOCKED); 1446 KASSERT(em != NULL, ("getppid: parent emuldata not found.\n")); 1447 1448 td->td_retval[0] = em->shared->group_pid; 1449 } else 1450 td->td_retval[0] = pp->p_pid; 1451 1452 EMUL_UNLOCK(&emul_lock); 1453 PROC_UNLOCK(pp); 1454 1455 return (0); 1456 } 1457 1458 int 1459 linux_getgid(struct thread *td, struct linux_getgid_args *args) 1460 { 1461 1462 td->td_retval[0] = td->td_ucred->cr_rgid; 1463 return (0); 1464 } 1465 1466 int 1467 linux_getuid(struct thread *td, struct linux_getuid_args *args) 1468 { 1469 1470 td->td_retval[0] = td->td_ucred->cr_ruid; 1471 return (0); 1472 } 1473 1474 1475 int 1476 linux_getsid(struct thread *td, struct linux_getsid_args *args) 1477 { 1478 struct getsid_args bsd; 1479 bsd.pid = args->pid; 1480 return getsid(td, &bsd); 1481 } 1482 1483 int 1484 linux_nosys(struct thread *td, struct nosys_args *ignore) 1485 { 1486 1487 return (ENOSYS); 1488 } 1489 1490 int 1491 linux_getpriority(struct thread *td, struct linux_getpriority_args *args) 1492 { 1493 struct getpriority_args bsd_args; 1494 int error; 1495 1496 bsd_args.which = args->which; 1497 bsd_args.who = args->who; 1498 error = getpriority(td, &bsd_args); 1499 td->td_retval[0] = 20 - td->td_retval[0]; 1500 return error; 1501 } 1502 1503 int 1504 linux_sethostname(struct thread *td, struct linux_sethostname_args *args) 1505 { 1506 int name[2]; 1507 1508 name[0] = CTL_KERN; 1509 name[1] = KERN_HOSTNAME; 1510 return (userland_sysctl(td, name, 2, 0, 0, 0, args->hostname, 1511 args->len, 0, 0)); 1512 } 1513 1514 int 1515 linux_exit_group(struct thread *td, struct linux_exit_group_args *args) 1516 { 1517 struct linux_emuldata *em, *td_em, *tmp_em; 1518 struct proc *sp; 1519 char osrel[LINUX_MAX_UTSNAME]; 1520 1521 #ifdef DEBUG 1522 if (ldebug(exit_group)) 1523 printf(ARGS(exit_group, "%i"), args->error_code); 1524 #endif 1525 1526 linux_get_osrelease(td, osrel); 1527 if (strlen(osrel) >= 3 && osrel[2] == '6') { 1528 td_em = em_find(td->td_proc, EMUL_UNLOCKED); 1529 1530 KASSERT(td_em != NULL, ("exit_group: emuldata not found.\n")); 1531 1532 EMUL_SHARED_RLOCK(&emul_shared_lock); 1533 LIST_FOREACH_SAFE(em, &td_em->shared->threads, threads, tmp_em) { 1534 if (em->pid == td_em->pid) 1535 continue; 1536 1537 sp = pfind(em->pid); 1538 psignal(sp, SIGKILL); 1539 PROC_UNLOCK(sp); 1540 #ifdef DEBUG 1541 printf(LMSG("linux_sys_exit_group: kill PID %d\n"), em->pid); 1542 #endif 1543 } 1544 1545 EMUL_SHARED_RUNLOCK(&emul_shared_lock); 1546 EMUL_UNLOCK(&emul_lock); 1547 } 1548 1549 exit1(td, W_EXITCODE(args->error_code,0)); 1550 1551 return (0); 1552 } 1553 1554 int 1555 linux_prctl(struct thread *td, struct linux_prctl_args *args) 1556 { 1557 int error = 0; 1558 struct proc *p = td->td_proc; 1559 char comm[LINUX_MAX_COMM_LEN]; 1560 struct linux_emuldata *em; 1561 1562 #ifdef DEBUG 1563 if (ldebug(prctl)) 1564 printf(ARGS(prctl, "%d, %d, %d, %d, %d"), args->option, args->arg2, 1565 args->arg3, args->arg4, args->arg5); 1566 #endif 1567 1568 switch (args->option) { 1569 case LINUX_PR_SET_PDEATHSIG: 1570 if (!LINUX_SIG_VALID(args->arg2)) 1571 return (EINVAL); 1572 em = em_find(p, EMUL_UNLOCKED); 1573 KASSERT(em != NULL, ("prctl: emuldata not found.\n")); 1574 em->pdeath_signal = args->arg2; 1575 EMUL_UNLOCK(&emul_lock); 1576 break; 1577 case LINUX_PR_GET_PDEATHSIG: 1578 em = em_find(p, EMUL_UNLOCKED); 1579 KASSERT(em != NULL, ("prctl: emuldata not found.\n")); 1580 error = copyout(&em->pdeath_signal, (void *)(register_t) args->arg2, sizeof(em->pdeath_signal)); 1581 EMUL_UNLOCK(&emul_lock); 1582 break; 1583 case LINUX_PR_SET_NAME: 1584 comm[LINUX_MAX_COMM_LEN-1] = 0; 1585 error = copyin(comm, (void *)(register_t) args->arg2, LINUX_MAX_COMM_LEN-1); 1586 if (error) 1587 return (error); 1588 PROC_LOCK(p); 1589 strcpy(p->p_comm, comm); 1590 PROC_UNLOCK(p); 1591 break; 1592 case LINUX_PR_GET_NAME: 1593 error = copyout(&p->p_comm, (void *)(register_t) args->arg2, MAXCOMLEN+1); 1594 break; 1595 default: 1596 error = EINVAL; 1597 break; 1598 } 1599 1600 return (error); 1601 } 1602