1 /* $NetBSD: linux_misc.c,v 1.158 2006/07/23 22:06:09 ad Exp $ */ 2 3 /*- 4 * Copyright (c) 1995, 1998, 1999 The NetBSD Foundation, Inc. 5 * All rights reserved. 6 * 7 * This code is derived from software contributed to The NetBSD Foundation 8 * by Frank van der Linden and Eric Haszlakiewicz; by Jason R. Thorpe 9 * of the Numerical Aerospace Simulation Facility, NASA Ames Research Center. 10 * 11 * Redistribution and use in source and binary forms, with or without 12 * modification, are permitted provided that the following conditions 13 * are met: 14 * 1. Redistributions of source code must retain the above copyright 15 * notice, this list of conditions and the following disclaimer. 16 * 2. Redistributions in binary form must reproduce the above copyright 17 * notice, this list of conditions and the following disclaimer in the 18 * documentation and/or other materials provided with the distribution. 19 * 3. All advertising materials mentioning features or use of this software 20 * must display the following acknowledgement: 21 * This product includes software developed by the NetBSD 22 * Foundation, Inc. and its contributors. 23 * 4. Neither the name of The NetBSD Foundation nor the names of its 24 * contributors may be used to endorse or promote products derived 25 * from this software without specific prior written permission. 26 * 27 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 28 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 29 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 30 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 31 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 32 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 33 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 34 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 35 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 36 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 37 * POSSIBILITY OF SUCH DAMAGE. 38 */ 39 40 /* 41 * Linux compatibility module. Try to deal with various Linux system calls. 42 */ 43 44 /* 45 * These functions have been moved to multiarch to allow 46 * selection of which machines include them to be 47 * determined by the individual files.linux_<arch> files. 48 * 49 * Function in multiarch: 50 * linux_sys_break : linux_break.c 51 * linux_sys_alarm : linux_misc_notalpha.c 52 * linux_sys_getresgid : linux_misc_notalpha.c 53 * linux_sys_nice : linux_misc_notalpha.c 54 * linux_sys_readdir : linux_misc_notalpha.c 55 * linux_sys_setresgid : linux_misc_notalpha.c 56 * linux_sys_time : linux_misc_notalpha.c 57 * linux_sys_utime : linux_misc_notalpha.c 58 * linux_sys_waitpid : linux_misc_notalpha.c 59 * linux_sys_old_mmap : linux_oldmmap.c 60 * linux_sys_oldolduname : linux_oldolduname.c 61 * linux_sys_oldselect : linux_oldselect.c 62 * linux_sys_olduname : linux_olduname.c 63 * linux_sys_pipe : linux_pipe.c 64 */ 65 66 #include <sys/cdefs.h> 67 __KERNEL_RCSID(0, "$NetBSD: linux_misc.c,v 1.158 2006/07/23 22:06:09 ad Exp $"); 68 69 #include <sys/param.h> 70 #include <sys/systm.h> 71 #include <sys/namei.h> 72 #include <sys/proc.h> 73 #include <sys/dirent.h> 74 #include <sys/file.h> 75 #include <sys/stat.h> 76 #include <sys/filedesc.h> 77 #include <sys/ioctl.h> 78 #include <sys/kernel.h> 79 #include <sys/malloc.h> 80 #include <sys/mbuf.h> 81 #include <sys/mman.h> 82 #include <sys/mount.h> 83 #include <sys/reboot.h> 84 #include <sys/resource.h> 85 #include <sys/resourcevar.h> 86 #include <sys/signal.h> 87 #include <sys/signalvar.h> 88 #include <sys/socket.h> 89 #include <sys/time.h> 90 #include <sys/times.h> 91 #include <sys/vnode.h> 92 #include <sys/uio.h> 93 #include <sys/wait.h> 94 #include <sys/utsname.h> 95 #include <sys/unistd.h> 96 #include <sys/swap.h> /* for SWAP_ON */ 97 #include <sys/sysctl.h> /* for KERN_DOMAINNAME */ 98 #include <sys/kauth.h> 99 100 #include <sys/ptrace.h> 101 #include <machine/ptrace.h> 102 103 #include <sys/sa.h> 104 #include <sys/syscallargs.h> 105 106 #include <compat/linux/common/linux_machdep.h> 107 #include <compat/linux/common/linux_types.h> 108 #include <compat/linux/common/linux_signal.h> 109 110 #include <compat/linux/linux_syscallargs.h> 111 112 #include <compat/linux/common/linux_fcntl.h> 113 #include <compat/linux/common/linux_mmap.h> 114 #include <compat/linux/common/linux_dirent.h> 115 #include <compat/linux/common/linux_util.h> 116 #include <compat/linux/common/linux_misc.h> 117 #ifndef COMPAT_LINUX32 118 #include <compat/linux/common/linux_limit.h> 119 #endif 120 #include <compat/linux/common/linux_ptrace.h> 121 #include <compat/linux/common/linux_reboot.h> 122 #include <compat/linux/common/linux_emuldata.h> 123 124 #ifndef COMPAT_LINUX32 125 const int linux_ptrace_request_map[] = { 126 LINUX_PTRACE_TRACEME, PT_TRACE_ME, 127 LINUX_PTRACE_PEEKTEXT, PT_READ_I, 128 LINUX_PTRACE_PEEKDATA, PT_READ_D, 129 LINUX_PTRACE_POKETEXT, PT_WRITE_I, 130 LINUX_PTRACE_POKEDATA, PT_WRITE_D, 131 LINUX_PTRACE_CONT, PT_CONTINUE, 132 LINUX_PTRACE_KILL, PT_KILL, 133 LINUX_PTRACE_ATTACH, PT_ATTACH, 134 LINUX_PTRACE_DETACH, PT_DETACH, 135 # ifdef PT_STEP 136 LINUX_PTRACE_SINGLESTEP, PT_STEP, 137 # endif 138 -1 139 }; 140 141 const struct linux_mnttypes linux_fstypes[] = { 142 { MOUNT_FFS, LINUX_DEFAULT_SUPER_MAGIC }, 143 { MOUNT_NFS, LINUX_NFS_SUPER_MAGIC }, 144 { MOUNT_MFS, LINUX_DEFAULT_SUPER_MAGIC }, 145 { MOUNT_MSDOS, LINUX_MSDOS_SUPER_MAGIC }, 146 { MOUNT_LFS, LINUX_DEFAULT_SUPER_MAGIC }, 147 { MOUNT_FDESC, LINUX_DEFAULT_SUPER_MAGIC }, 148 { MOUNT_PORTAL, LINUX_DEFAULT_SUPER_MAGIC }, 149 { MOUNT_NULL, LINUX_DEFAULT_SUPER_MAGIC }, 150 { MOUNT_OVERLAY, LINUX_DEFAULT_SUPER_MAGIC }, 151 { MOUNT_UMAP, LINUX_DEFAULT_SUPER_MAGIC }, 152 { MOUNT_KERNFS, LINUX_DEFAULT_SUPER_MAGIC }, 153 { MOUNT_PROCFS, LINUX_PROC_SUPER_MAGIC }, 154 { MOUNT_AFS, LINUX_DEFAULT_SUPER_MAGIC }, 155 { MOUNT_CD9660, LINUX_ISOFS_SUPER_MAGIC }, 156 { MOUNT_UNION, LINUX_DEFAULT_SUPER_MAGIC }, 157 { MOUNT_ADOSFS, LINUX_ADFS_SUPER_MAGIC }, 158 { MOUNT_EXT2FS, LINUX_EXT2_SUPER_MAGIC }, 159 { MOUNT_CFS, LINUX_DEFAULT_SUPER_MAGIC }, 160 { MOUNT_CODA, LINUX_CODA_SUPER_MAGIC }, 161 { MOUNT_FILECORE, LINUX_DEFAULT_SUPER_MAGIC }, 162 { MOUNT_NTFS, LINUX_DEFAULT_SUPER_MAGIC }, 163 { MOUNT_SMBFS, LINUX_SMB_SUPER_MAGIC }, 164 { MOUNT_PTYFS, LINUX_DEVPTS_SUPER_MAGIC }, 165 { MOUNT_TMPFS, LINUX_DEFAULT_SUPER_MAGIC } 166 }; 167 const int linux_fstypes_cnt = sizeof(linux_fstypes) / sizeof(linux_fstypes[0]); 168 169 # ifdef DEBUG_LINUX 170 #define DPRINTF(a) uprintf a 171 # else 172 #define DPRINTF(a) 173 # endif 174 175 /* Local linux_misc.c functions: */ 176 # ifndef __amd64__ 177 static void bsd_to_linux_statfs __P((const struct statvfs *, 178 struct linux_statfs *)); 179 # endif 180 static void linux_to_bsd_mmap_args __P((struct sys_mmap_args *, 181 const struct linux_sys_mmap_args *)); 182 static int linux_mmap __P((struct lwp *, struct linux_sys_mmap_args *, 183 register_t *, off_t)); 184 185 186 /* 187 * The information on a terminated (or stopped) process needs 188 * to be converted in order for Linux binaries to get a valid signal 189 * number out of it. 190 */ 191 void 192 bsd_to_linux_wstat(st) 193 int *st; 194 { 195 196 int sig; 197 198 if (WIFSIGNALED(*st)) { 199 sig = WTERMSIG(*st); 200 if (sig >= 0 && sig < NSIG) 201 *st= (*st& ~0177) | native_to_linux_signo[sig]; 202 } else if (WIFSTOPPED(*st)) { 203 sig = WSTOPSIG(*st); 204 if (sig >= 0 && sig < NSIG) 205 *st = (*st & ~0xff00) | 206 (native_to_linux_signo[sig] << 8); 207 } 208 } 209 210 /* 211 * wait4(2). Passed on to the NetBSD call, surrounded by code to 212 * reserve some space for a NetBSD-style wait status, and converting 213 * it to what Linux wants. 214 */ 215 int 216 linux_sys_wait4(l, v, retval) 217 struct lwp *l; 218 void *v; 219 register_t *retval; 220 { 221 struct linux_sys_wait4_args /* { 222 syscallarg(int) pid; 223 syscallarg(int *) status; 224 syscallarg(int) options; 225 syscallarg(struct rusage *) rusage; 226 } */ *uap = v; 227 struct proc *p = l->l_proc; 228 struct sys_wait4_args w4a; 229 int error, *status, tstat, options, linux_options; 230 caddr_t sg; 231 232 if (SCARG(uap, status) != NULL) { 233 sg = stackgap_init(p, 0); 234 status = (int *) stackgap_alloc(p, &sg, sizeof *status); 235 } else 236 status = NULL; 237 238 linux_options = SCARG(uap, options); 239 options = 0; 240 if (linux_options & ~(LINUX_WAIT4_KNOWNFLAGS)) 241 return (EINVAL); 242 243 if (linux_options & LINUX_WAIT4_WNOHANG) 244 options |= WNOHANG; 245 if (linux_options & LINUX_WAIT4_WUNTRACED) 246 options |= WUNTRACED; 247 if (linux_options & LINUX_WAIT4_WALL) 248 options |= WALLSIG; 249 if (linux_options & LINUX_WAIT4_WCLONE) 250 options |= WALTSIG; 251 # ifdef DIAGNOSTIC 252 if (linux_options & LINUX_WAIT4_WNOTHREAD) 253 printf("WARNING: %s: linux process %d.%d called " 254 "waitpid with __WNOTHREAD set!", 255 __FILE__, p->p_pid, l->l_lid); 256 257 # endif 258 259 SCARG(&w4a, pid) = SCARG(uap, pid); 260 SCARG(&w4a, status) = status; 261 SCARG(&w4a, options) = options; 262 SCARG(&w4a, rusage) = SCARG(uap, rusage); 263 264 if ((error = sys_wait4(l, &w4a, retval))) 265 return error; 266 267 sigdelset(&p->p_sigctx.ps_siglist, SIGCHLD); 268 269 if (status != NULL) { 270 if ((error = copyin(status, &tstat, sizeof tstat))) 271 return error; 272 273 bsd_to_linux_wstat(&tstat); 274 return copyout(&tstat, SCARG(uap, status), sizeof tstat); 275 } 276 277 return 0; 278 } 279 280 /* 281 * Linux brk(2). The check if the new address is >= the old one is 282 * done in the kernel in Linux. NetBSD does it in the library. 283 */ 284 int 285 linux_sys_brk(l, v, retval) 286 struct lwp *l; 287 void *v; 288 register_t *retval; 289 { 290 struct linux_sys_brk_args /* { 291 syscallarg(char *) nsize; 292 } */ *uap = v; 293 struct proc *p = l->l_proc; 294 char *nbrk = SCARG(uap, nsize); 295 struct sys_obreak_args oba; 296 struct vmspace *vm = p->p_vmspace; 297 struct linux_emuldata *ed = (struct linux_emuldata*)p->p_emuldata; 298 299 SCARG(&oba, nsize) = nbrk; 300 301 if ((caddr_t) nbrk > vm->vm_daddr && sys_obreak(l, &oba, retval) == 0) 302 ed->s->p_break = (char*)nbrk; 303 else 304 nbrk = ed->s->p_break; 305 306 retval[0] = (register_t)nbrk; 307 308 return 0; 309 } 310 311 # ifndef __amd64__ 312 /* 313 * Convert NetBSD statvfs structure to Linux statfs structure. 314 * Linux doesn't have f_flag, and we can't set f_frsize due 315 * to glibc statvfs() bug (see below). 316 */ 317 static void 318 bsd_to_linux_statfs(bsp, lsp) 319 const struct statvfs *bsp; 320 struct linux_statfs *lsp; 321 { 322 int i; 323 324 for (i = 0; i < linux_fstypes_cnt; i++) { 325 if (strcmp(bsp->f_fstypename, linux_fstypes[i].bsd) == 0) { 326 lsp->l_ftype = linux_fstypes[i].linux; 327 break; 328 } 329 } 330 331 if (i == linux_fstypes_cnt) { 332 DPRINTF(("unhandled fstype in linux emulation: %s\n", 333 bsp->f_fstypename)); 334 lsp->l_ftype = LINUX_DEFAULT_SUPER_MAGIC; 335 } 336 337 /* 338 * The sizes are expressed in number of blocks. The block 339 * size used for the size is f_frsize for POSIX-compliant 340 * statvfs. Linux statfs uses f_bsize as the block size 341 * (f_frsize used to not be available in Linux struct statfs). 342 * However, glibc 2.3.3 statvfs() wrapper fails to adjust the block 343 * counts for different f_frsize if f_frsize is provided by the kernel. 344 * POSIX conforming apps thus get wrong size if f_frsize 345 * is different to f_bsize. Thus, we just pretend we don't 346 * support f_frsize. 347 */ 348 349 lsp->l_fbsize = bsp->f_frsize; 350 lsp->l_ffrsize = 0; /* compat */ 351 lsp->l_fblocks = bsp->f_blocks; 352 lsp->l_fbfree = bsp->f_bfree; 353 lsp->l_fbavail = bsp->f_bavail; 354 lsp->l_ffiles = bsp->f_files; 355 lsp->l_fffree = bsp->f_ffree; 356 /* Linux sets the fsid to 0..., we don't */ 357 lsp->l_ffsid.val[0] = bsp->f_fsidx.__fsid_val[0]; 358 lsp->l_ffsid.val[1] = bsp->f_fsidx.__fsid_val[1]; 359 lsp->l_fnamelen = bsp->f_namemax; 360 (void)memset(lsp->l_fspare, 0, sizeof(lsp->l_fspare)); 361 } 362 363 /* 364 * Implement the fs stat functions. Straightforward. 365 */ 366 int 367 linux_sys_statfs(l, v, retval) 368 struct lwp *l; 369 void *v; 370 register_t *retval; 371 { 372 struct linux_sys_statfs_args /* { 373 syscallarg(const char *) path; 374 syscallarg(struct linux_statfs *) sp; 375 } */ *uap = v; 376 struct proc *p = l->l_proc; 377 struct statvfs *btmp, *bsp; 378 struct linux_statfs ltmp; 379 struct sys_statvfs1_args bsa; 380 caddr_t sg; 381 int error; 382 383 sg = stackgap_init(p, 0); 384 bsp = stackgap_alloc(p, &sg, sizeof (struct statvfs)); 385 386 CHECK_ALT_EXIST(l, &sg, SCARG(uap, path)); 387 388 SCARG(&bsa, path) = SCARG(uap, path); 389 SCARG(&bsa, buf) = bsp; 390 SCARG(&bsa, flags) = ST_WAIT; 391 392 if ((error = sys_statvfs1(l, &bsa, retval))) 393 return error; 394 395 btmp = STATVFSBUF_GET(); 396 error = copyin(bsp, btmp, sizeof(*btmp)); 397 if (error) { 398 goto out; 399 } 400 bsd_to_linux_statfs(btmp, <mp); 401 error = copyout(<mp, SCARG(uap, sp), sizeof ltmp); 402 out: 403 STATVFSBUF_PUT(btmp); 404 return error; 405 } 406 407 int 408 linux_sys_fstatfs(l, v, retval) 409 struct lwp *l; 410 void *v; 411 register_t *retval; 412 { 413 struct linux_sys_fstatfs_args /* { 414 syscallarg(int) fd; 415 syscallarg(struct linux_statfs *) sp; 416 } */ *uap = v; 417 struct proc *p = l->l_proc; 418 struct statvfs *btmp, *bsp; 419 struct linux_statfs ltmp; 420 struct sys_fstatvfs1_args bsa; 421 caddr_t sg; 422 int error; 423 424 sg = stackgap_init(p, 0); 425 bsp = stackgap_alloc(p, &sg, sizeof (struct statvfs)); 426 427 SCARG(&bsa, fd) = SCARG(uap, fd); 428 SCARG(&bsa, buf) = bsp; 429 SCARG(&bsa, flags) = ST_WAIT; 430 431 if ((error = sys_fstatvfs1(l, &bsa, retval))) 432 return error; 433 434 btmp = STATVFSBUF_GET(); 435 error = copyin(bsp, btmp, sizeof(*btmp)); 436 if (error) { 437 goto out; 438 } 439 bsd_to_linux_statfs(btmp, <mp); 440 error = copyout(<mp, SCARG(uap, sp), sizeof ltmp); 441 out: 442 STATVFSBUF_PUT(btmp); 443 return error; 444 } 445 # endif /* !__amd64__ */ 446 447 /* 448 * uname(). Just copy the info from the various strings stored in the 449 * kernel, and put it in the Linux utsname structure. That structure 450 * is almost the same as the NetBSD one, only it has fields 65 characters 451 * long, and an extra domainname field. 452 */ 453 int 454 linux_sys_uname(l, v, retval) 455 struct lwp *l; 456 void *v; 457 register_t *retval; 458 { 459 struct linux_sys_uname_args /* { 460 syscallarg(struct linux_utsname *) up; 461 } */ *uap = v; 462 struct linux_utsname luts; 463 464 strncpy(luts.l_sysname, linux_sysname, sizeof(luts.l_sysname)); 465 strncpy(luts.l_nodename, hostname, sizeof(luts.l_nodename)); 466 strncpy(luts.l_release, linux_release, sizeof(luts.l_release)); 467 strncpy(luts.l_version, linux_version, sizeof(luts.l_version)); 468 # ifdef LINUX_UNAME_ARCH 469 strncpy(luts.l_machine, LINUX_UNAME_ARCH, sizeof(luts.l_machine)); 470 # else 471 strncpy(luts.l_machine, machine, sizeof(luts.l_machine)); 472 # endif 473 strncpy(luts.l_domainname, domainname, sizeof(luts.l_domainname)); 474 475 return copyout(&luts, SCARG(uap, up), sizeof(luts)); 476 } 477 478 /* Used directly on: alpha, mips, ppc, sparc, sparc64 */ 479 /* Used indirectly on: arm, i386, m68k */ 480 481 /* 482 * New type Linux mmap call. 483 * Only called directly on machines with >= 6 free regs. 484 */ 485 int 486 linux_sys_mmap(l, v, retval) 487 struct lwp *l; 488 void *v; 489 register_t *retval; 490 { 491 struct linux_sys_mmap_args /* { 492 syscallarg(unsigned long) addr; 493 syscallarg(size_t) len; 494 syscallarg(int) prot; 495 syscallarg(int) flags; 496 syscallarg(int) fd; 497 syscallarg(linux_off_t) offset; 498 } */ *uap = v; 499 500 if (SCARG(uap, offset) & PAGE_MASK) 501 return EINVAL; 502 503 return linux_mmap(l, uap, retval, SCARG(uap, offset)); 504 } 505 506 /* 507 * Guts of most architectures' mmap64() implementations. This shares 508 * its list of arguments with linux_sys_mmap(). 509 * 510 * The difference in linux_sys_mmap2() is that "offset" is actually 511 * (offset / pagesize), not an absolute byte count. This translation 512 * to pagesize offsets is done inside glibc between the mmap64() call 513 * point, and the actual syscall. 514 */ 515 int 516 linux_sys_mmap2(l, v, retval) 517 struct lwp *l; 518 void *v; 519 register_t *retval; 520 { 521 struct linux_sys_mmap2_args /* { 522 syscallarg(unsigned long) addr; 523 syscallarg(size_t) len; 524 syscallarg(int) prot; 525 syscallarg(int) flags; 526 syscallarg(int) fd; 527 syscallarg(linux_off_t) offset; 528 } */ *uap = v; 529 530 return linux_mmap(l, uap, retval, 531 ((off_t)SCARG(uap, offset)) << PAGE_SHIFT); 532 } 533 534 /* 535 * Massage arguments and call system mmap(2). 536 */ 537 static int 538 linux_mmap(l, uap, retval, offset) 539 struct lwp *l; 540 struct linux_sys_mmap_args *uap; 541 register_t *retval; 542 off_t offset; 543 { 544 struct sys_mmap_args cma; 545 int error; 546 size_t mmoff=0; 547 548 if (SCARG(uap, flags) & LINUX_MAP_GROWSDOWN) { 549 /* 550 * Request for stack-like memory segment. On linux, this 551 * works by mmap()ping (small) segment, which is automatically 552 * extended when page fault happens below the currently 553 * allocated area. We emulate this by allocating (typically 554 * bigger) segment sized at current stack size limit, and 555 * offsetting the requested and returned address accordingly. 556 * Since physical pages are only allocated on-demand, this 557 * is effectively identical. 558 */ 559 rlim_t ssl = l->l_proc->p_rlimit[RLIMIT_STACK].rlim_cur; 560 561 if (SCARG(uap, len) < ssl) { 562 /* Compute the address offset */ 563 mmoff = round_page(ssl) - SCARG(uap, len); 564 565 if (SCARG(uap, addr)) 566 SCARG(uap, addr) -= mmoff; 567 568 SCARG(uap, len) = (size_t) ssl; 569 } 570 } 571 572 linux_to_bsd_mmap_args(&cma, uap); 573 SCARG(&cma, pos) = offset; 574 575 error = sys_mmap(l, &cma, retval); 576 if (error) 577 return (error); 578 579 /* Shift the returned address for stack-like segment if necessary */ 580 if (SCARG(uap, flags) & LINUX_MAP_GROWSDOWN && mmoff) 581 retval[0] += mmoff; 582 583 return (0); 584 } 585 586 static void 587 linux_to_bsd_mmap_args(cma, uap) 588 struct sys_mmap_args *cma; 589 const struct linux_sys_mmap_args *uap; 590 { 591 int flags = MAP_TRYFIXED, fl = SCARG(uap, flags); 592 593 flags |= cvtto_bsd_mask(fl, LINUX_MAP_SHARED, MAP_SHARED); 594 flags |= cvtto_bsd_mask(fl, LINUX_MAP_PRIVATE, MAP_PRIVATE); 595 flags |= cvtto_bsd_mask(fl, LINUX_MAP_FIXED, MAP_FIXED); 596 flags |= cvtto_bsd_mask(fl, LINUX_MAP_ANON, MAP_ANON); 597 /* XXX XAX ERH: Any other flags here? There are more defined... */ 598 599 SCARG(cma, addr) = (void *)SCARG(uap, addr); 600 SCARG(cma, len) = SCARG(uap, len); 601 SCARG(cma, prot) = SCARG(uap, prot); 602 if (SCARG(cma, prot) & VM_PROT_WRITE) /* XXX */ 603 SCARG(cma, prot) |= VM_PROT_READ; 604 SCARG(cma, flags) = flags; 605 SCARG(cma, fd) = flags & MAP_ANON ? -1 : SCARG(uap, fd); 606 SCARG(cma, pad) = 0; 607 } 608 609 #define LINUX_MREMAP_MAYMOVE 1 610 #define LINUX_MREMAP_FIXED 2 611 612 int 613 linux_sys_mremap(l, v, retval) 614 struct lwp *l; 615 void *v; 616 register_t *retval; 617 { 618 struct linux_sys_mremap_args /* { 619 syscallarg(void *) old_address; 620 syscallarg(size_t) old_size; 621 syscallarg(size_t) new_size; 622 syscallarg(u_long) flags; 623 } */ *uap = v; 624 625 struct proc *p; 626 struct vm_map *map; 627 vaddr_t oldva; 628 vaddr_t newva; 629 size_t oldsize; 630 size_t newsize; 631 int flags; 632 int uvmflags; 633 int error; 634 635 flags = SCARG(uap, flags); 636 oldva = (vaddr_t)SCARG(uap, old_address); 637 oldsize = round_page(SCARG(uap, old_size)); 638 newsize = round_page(SCARG(uap, new_size)); 639 if ((flags & ~(LINUX_MREMAP_FIXED|LINUX_MREMAP_MAYMOVE)) != 0) { 640 error = EINVAL; 641 goto done; 642 } 643 if ((flags & LINUX_MREMAP_FIXED) != 0) { 644 if ((flags & LINUX_MREMAP_MAYMOVE) == 0) { 645 error = EINVAL; 646 goto done; 647 } 648 #if 0 /* notyet */ 649 newva = SCARG(uap, new_address); 650 uvmflags = UVM_MREMAP_FIXED; 651 #else /* notyet */ 652 error = EOPNOTSUPP; 653 goto done; 654 #endif /* notyet */ 655 } else if ((flags & LINUX_MREMAP_MAYMOVE) != 0) { 656 uvmflags = 0; 657 } else { 658 newva = oldva; 659 uvmflags = UVM_MREMAP_FIXED; 660 } 661 p = l->l_proc; 662 map = &p->p_vmspace->vm_map; 663 error = uvm_mremap(map, oldva, oldsize, map, &newva, newsize, p, 664 uvmflags); 665 666 done: 667 *retval = (error != 0) ? 0 : (register_t)newva; 668 return error; 669 } 670 671 int 672 linux_sys_msync(l, v, retval) 673 struct lwp *l; 674 void *v; 675 register_t *retval; 676 { 677 struct linux_sys_msync_args /* { 678 syscallarg(caddr_t) addr; 679 syscallarg(int) len; 680 syscallarg(int) fl; 681 } */ *uap = v; 682 683 struct sys___msync13_args bma; 684 685 /* flags are ignored */ 686 SCARG(&bma, addr) = SCARG(uap, addr); 687 SCARG(&bma, len) = SCARG(uap, len); 688 SCARG(&bma, flags) = SCARG(uap, fl); 689 690 return sys___msync13(l, &bma, retval); 691 } 692 693 int 694 linux_sys_mprotect(l, v, retval) 695 struct lwp *l; 696 void *v; 697 register_t *retval; 698 { 699 struct linux_sys_mprotect_args /* { 700 syscallarg(const void *) start; 701 syscallarg(unsigned long) len; 702 syscallarg(int) prot; 703 } */ *uap = v; 704 struct vm_map_entry *entry; 705 struct vm_map *map; 706 struct proc *p; 707 vaddr_t end, start, len, stacklim; 708 int prot, grows; 709 710 start = (vaddr_t)SCARG(uap, start); 711 len = round_page(SCARG(uap, len)); 712 prot = SCARG(uap, prot); 713 grows = prot & (LINUX_PROT_GROWSDOWN | LINUX_PROT_GROWSUP); 714 prot &= ~grows; 715 end = start + len; 716 717 if (start & PAGE_MASK) 718 return EINVAL; 719 if (end < start) 720 return EINVAL; 721 if (end == start) 722 return 0; 723 724 if (prot & ~(PROT_READ | PROT_WRITE | PROT_EXEC)) 725 return EINVAL; 726 if (grows == (LINUX_PROT_GROWSDOWN | LINUX_PROT_GROWSUP)) 727 return EINVAL; 728 729 p = l->l_proc; 730 map = &p->p_vmspace->vm_map; 731 vm_map_lock(map); 732 # ifdef notdef 733 VM_MAP_RANGE_CHECK(map, start, end); 734 # endif 735 if (!uvm_map_lookup_entry(map, start, &entry) || entry->start > start) { 736 vm_map_unlock(map); 737 return ENOMEM; 738 } 739 740 /* 741 * Approximate the behaviour of PROT_GROWS{DOWN,UP}. 742 */ 743 744 stacklim = (vaddr_t)p->p_limit->pl_rlimit[RLIMIT_STACK].rlim_cur; 745 if (grows & LINUX_PROT_GROWSDOWN) { 746 if (USRSTACK - stacklim <= start && start < USRSTACK) { 747 start = USRSTACK - stacklim; 748 } else { 749 start = entry->start; 750 } 751 } else if (grows & LINUX_PROT_GROWSUP) { 752 if (USRSTACK <= end && end < USRSTACK + stacklim) { 753 end = USRSTACK + stacklim; 754 } else { 755 end = entry->end; 756 } 757 } 758 vm_map_unlock(map); 759 return uvm_map_protect(map, start, end, prot, FALSE); 760 } 761 762 /* 763 * This code is partly stolen from src/lib/libc/compat-43/times.c 764 */ 765 766 #define CONVTCK(r) (r.tv_sec * hz + r.tv_usec / (1000000 / hz)) 767 768 int 769 linux_sys_times(l, v, retval) 770 struct lwp *l; 771 void *v; 772 register_t *retval; 773 { 774 struct linux_sys_times_args /* { 775 syscallarg(struct times *) tms; 776 } */ *uap = v; 777 struct proc *p = l->l_proc; 778 struct timeval t; 779 int error; 780 781 if (SCARG(uap, tms)) { 782 struct linux_tms ltms; 783 struct rusage ru; 784 785 calcru(p, &ru.ru_utime, &ru.ru_stime, NULL); 786 ltms.ltms_utime = CONVTCK(ru.ru_utime); 787 ltms.ltms_stime = CONVTCK(ru.ru_stime); 788 789 ltms.ltms_cutime = CONVTCK(p->p_stats->p_cru.ru_utime); 790 ltms.ltms_cstime = CONVTCK(p->p_stats->p_cru.ru_stime); 791 792 if ((error = copyout(<ms, SCARG(uap, tms), sizeof ltms))) 793 return error; 794 } 795 796 getmicrouptime(&t); 797 798 retval[0] = ((linux_clock_t)(CONVTCK(t))); 799 return 0; 800 } 801 802 #undef CONVTCK 803 804 /* 805 * Linux 'readdir' call. This code is mostly taken from the 806 * SunOS getdents call (see compat/sunos/sunos_misc.c), though 807 * an attempt has been made to keep it a little cleaner (failing 808 * miserably, because of the cruft needed if count 1 is passed). 809 * 810 * The d_off field should contain the offset of the next valid entry, 811 * but in Linux it has the offset of the entry itself. We emulate 812 * that bug here. 813 * 814 * Read in BSD-style entries, convert them, and copy them out. 815 * 816 * Note that this doesn't handle union-mounted filesystems. 817 */ 818 int 819 linux_sys_getdents(l, v, retval) 820 struct lwp *l; 821 void *v; 822 register_t *retval; 823 { 824 struct linux_sys_getdents_args /* { 825 syscallarg(int) fd; 826 syscallarg(struct linux_dirent *) dent; 827 syscallarg(unsigned int) count; 828 } */ *uap = v; 829 struct dirent *bdp; 830 struct vnode *vp; 831 caddr_t inp, tbuf; /* BSD-format */ 832 int len, reclen; /* BSD-format */ 833 caddr_t outp; /* Linux-format */ 834 int resid, linux_reclen = 0; /* Linux-format */ 835 struct file *fp; 836 struct uio auio; 837 struct iovec aiov; 838 struct linux_dirent idb; 839 off_t off; /* true file offset */ 840 int buflen, error, eofflag, nbytes, oldcall; 841 struct vattr va; 842 off_t *cookiebuf = NULL, *cookie; 843 int ncookies; 844 845 /* getvnode() will use the descriptor for us */ 846 if ((error = getvnode(l->l_proc->p_fd, SCARG(uap, fd), &fp)) != 0) 847 return (error); 848 849 if ((fp->f_flag & FREAD) == 0) { 850 error = EBADF; 851 goto out1; 852 } 853 854 vp = (struct vnode *)fp->f_data; 855 if (vp->v_type != VDIR) { 856 error = EINVAL; 857 goto out1; 858 } 859 860 if ((error = VOP_GETATTR(vp, &va, l->l_cred, l))) 861 goto out1; 862 863 nbytes = SCARG(uap, count); 864 if (nbytes == 1) { /* emulating old, broken behaviour */ 865 nbytes = sizeof (idb); 866 buflen = max(va.va_blocksize, nbytes); 867 oldcall = 1; 868 } else { 869 buflen = min(MAXBSIZE, nbytes); 870 if (buflen < va.va_blocksize) 871 buflen = va.va_blocksize; 872 oldcall = 0; 873 } 874 tbuf = malloc(buflen, M_TEMP, M_WAITOK); 875 876 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 877 off = fp->f_offset; 878 again: 879 aiov.iov_base = tbuf; 880 aiov.iov_len = buflen; 881 auio.uio_iov = &aiov; 882 auio.uio_iovcnt = 1; 883 auio.uio_rw = UIO_READ; 884 auio.uio_resid = buflen; 885 auio.uio_offset = off; 886 UIO_SETUP_SYSSPACE(&auio); 887 /* 888 * First we read into the malloc'ed buffer, then 889 * we massage it into user space, one record at a time. 890 */ 891 error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, &cookiebuf, 892 &ncookies); 893 if (error) 894 goto out; 895 896 inp = tbuf; 897 outp = (caddr_t)SCARG(uap, dent); 898 resid = nbytes; 899 if ((len = buflen - auio.uio_resid) == 0) 900 goto eof; 901 902 for (cookie = cookiebuf; len > 0; len -= reclen) { 903 bdp = (struct dirent *)inp; 904 reclen = bdp->d_reclen; 905 if (reclen & 3) 906 panic("linux_readdir"); 907 if (bdp->d_fileno == 0) { 908 inp += reclen; /* it is a hole; squish it out */ 909 if (cookie) 910 off = *cookie++; 911 else 912 off += reclen; 913 continue; 914 } 915 linux_reclen = LINUX_RECLEN(&idb, bdp->d_namlen); 916 if (reclen > len || resid < linux_reclen) { 917 /* entry too big for buffer, so just stop */ 918 outp++; 919 break; 920 } 921 /* 922 * Massage in place to make a Linux-shaped dirent (otherwise 923 * we have to worry about touching user memory outside of 924 * the copyout() call). 925 */ 926 idb.d_ino = bdp->d_fileno; 927 /* 928 * The old readdir() call misuses the offset and reclen fields. 929 */ 930 if (oldcall) { 931 idb.d_off = (linux_off_t)linux_reclen; 932 idb.d_reclen = (u_short)bdp->d_namlen; 933 } else { 934 if (sizeof (idb.d_off) <= 4 && (off >> 32) != 0) { 935 compat_offseterr(vp, "linux_getdents"); 936 error = EINVAL; 937 goto out; 938 } 939 idb.d_off = (linux_off_t)off; 940 idb.d_reclen = (u_short)linux_reclen; 941 } 942 strcpy(idb.d_name, bdp->d_name); 943 if ((error = copyout((caddr_t)&idb, outp, linux_reclen))) 944 goto out; 945 /* advance past this real entry */ 946 inp += reclen; 947 if (cookie) 948 off = *cookie++; /* each entry points to itself */ 949 else 950 off += reclen; 951 /* advance output past Linux-shaped entry */ 952 outp += linux_reclen; 953 resid -= linux_reclen; 954 if (oldcall) 955 break; 956 } 957 958 /* if we squished out the whole block, try again */ 959 if (outp == (caddr_t)SCARG(uap, dent)) 960 goto again; 961 fp->f_offset = off; /* update the vnode offset */ 962 963 if (oldcall) 964 nbytes = resid + linux_reclen; 965 966 eof: 967 *retval = nbytes - resid; 968 out: 969 VOP_UNLOCK(vp, 0); 970 if (cookiebuf) 971 free(cookiebuf, M_TEMP); 972 free(tbuf, M_TEMP); 973 out1: 974 FILE_UNUSE(fp, l); 975 return error; 976 } 977 978 /* 979 * Even when just using registers to pass arguments to syscalls you can 980 * have 5 of them on the i386. So this newer version of select() does 981 * this. 982 */ 983 int 984 linux_sys_select(l, v, retval) 985 struct lwp *l; 986 void *v; 987 register_t *retval; 988 { 989 struct linux_sys_select_args /* { 990 syscallarg(int) nfds; 991 syscallarg(fd_set *) readfds; 992 syscallarg(fd_set *) writefds; 993 syscallarg(fd_set *) exceptfds; 994 syscallarg(struct timeval *) timeout; 995 } */ *uap = v; 996 997 return linux_select1(l, retval, SCARG(uap, nfds), SCARG(uap, readfds), 998 SCARG(uap, writefds), SCARG(uap, exceptfds), SCARG(uap, timeout)); 999 } 1000 1001 /* 1002 * Common code for the old and new versions of select(). A couple of 1003 * things are important: 1004 * 1) return the amount of time left in the 'timeout' parameter 1005 * 2) select never returns ERESTART on Linux, always return EINTR 1006 */ 1007 int 1008 linux_select1(l, retval, nfds, readfds, writefds, exceptfds, timeout) 1009 struct lwp *l; 1010 register_t *retval; 1011 int nfds; 1012 fd_set *readfds, *writefds, *exceptfds; 1013 struct timeval *timeout; 1014 { 1015 struct sys_select_args bsa; 1016 struct proc *p = l->l_proc; 1017 struct timeval tv0, tv1, utv, *tvp; 1018 caddr_t sg; 1019 int error; 1020 1021 SCARG(&bsa, nd) = nfds; 1022 SCARG(&bsa, in) = readfds; 1023 SCARG(&bsa, ou) = writefds; 1024 SCARG(&bsa, ex) = exceptfds; 1025 SCARG(&bsa, tv) = timeout; 1026 1027 /* 1028 * Store current time for computation of the amount of 1029 * time left. 1030 */ 1031 if (timeout) { 1032 if ((error = copyin(timeout, &utv, sizeof(utv)))) 1033 return error; 1034 if (itimerfix(&utv)) { 1035 /* 1036 * The timeval was invalid. Convert it to something 1037 * valid that will act as it does under Linux. 1038 */ 1039 sg = stackgap_init(p, 0); 1040 tvp = stackgap_alloc(p, &sg, sizeof(utv)); 1041 utv.tv_sec += utv.tv_usec / 1000000; 1042 utv.tv_usec %= 1000000; 1043 if (utv.tv_usec < 0) { 1044 utv.tv_sec -= 1; 1045 utv.tv_usec += 1000000; 1046 } 1047 if (utv.tv_sec < 0) 1048 timerclear(&utv); 1049 if ((error = copyout(&utv, tvp, sizeof(utv)))) 1050 return error; 1051 SCARG(&bsa, tv) = tvp; 1052 } 1053 microtime(&tv0); 1054 } 1055 1056 error = sys_select(l, &bsa, retval); 1057 if (error) { 1058 /* 1059 * See fs/select.c in the Linux kernel. Without this, 1060 * Maelstrom doesn't work. 1061 */ 1062 if (error == ERESTART) 1063 error = EINTR; 1064 return error; 1065 } 1066 1067 if (timeout) { 1068 if (*retval) { 1069 /* 1070 * Compute how much time was left of the timeout, 1071 * by subtracting the current time and the time 1072 * before we started the call, and subtracting 1073 * that result from the user-supplied value. 1074 */ 1075 microtime(&tv1); 1076 timersub(&tv1, &tv0, &tv1); 1077 timersub(&utv, &tv1, &utv); 1078 if (utv.tv_sec < 0) 1079 timerclear(&utv); 1080 } else 1081 timerclear(&utv); 1082 if ((error = copyout(&utv, timeout, sizeof(utv)))) 1083 return error; 1084 } 1085 1086 return 0; 1087 } 1088 1089 /* 1090 * Get the process group of a certain process. Look it up 1091 * and return the value. 1092 */ 1093 int 1094 linux_sys_getpgid(l, v, retval) 1095 struct lwp *l; 1096 void *v; 1097 register_t *retval; 1098 { 1099 struct linux_sys_getpgid_args /* { 1100 syscallarg(int) pid; 1101 } */ *uap = v; 1102 struct proc *p = l->l_proc; 1103 struct proc *targp; 1104 1105 if (SCARG(uap, pid) != 0 && SCARG(uap, pid) != p->p_pid) { 1106 if ((targp = pfind(SCARG(uap, pid))) == 0) 1107 return ESRCH; 1108 } 1109 else 1110 targp = p; 1111 1112 retval[0] = targp->p_pgid; 1113 return 0; 1114 } 1115 1116 /* 1117 * Set the 'personality' (emulation mode) for the current process. Only 1118 * accept the Linux personality here (0). This call is needed because 1119 * the Linux ELF crt0 issues it in an ugly kludge to make sure that 1120 * ELF binaries run in Linux mode, not SVR4 mode. 1121 */ 1122 int 1123 linux_sys_personality(l, v, retval) 1124 struct lwp *l; 1125 void *v; 1126 register_t *retval; 1127 { 1128 struct linux_sys_personality_args /* { 1129 syscallarg(int) per; 1130 } */ *uap = v; 1131 1132 if (SCARG(uap, per) != 0) 1133 return EINVAL; 1134 retval[0] = 0; 1135 return 0; 1136 } 1137 #endif /* !COMPAT_LINUX32 */ 1138 1139 #if defined(__i386__) || defined(__m68k__) || defined(COMPAT_LINUX32) 1140 /* 1141 * The calls are here because of type conversions. 1142 */ 1143 int 1144 linux_sys_setreuid16(l, v, retval) 1145 struct lwp *l; 1146 void *v; 1147 register_t *retval; 1148 { 1149 struct linux_sys_setreuid16_args /* { 1150 syscallarg(int) ruid; 1151 syscallarg(int) euid; 1152 } */ *uap = v; 1153 struct sys_setreuid_args bsa; 1154 1155 SCARG(&bsa, ruid) = ((linux_uid_t)SCARG(uap, ruid) == (linux_uid_t)-1) ? 1156 (uid_t)-1 : SCARG(uap, ruid); 1157 SCARG(&bsa, euid) = ((linux_uid_t)SCARG(uap, euid) == (linux_uid_t)-1) ? 1158 (uid_t)-1 : SCARG(uap, euid); 1159 1160 return sys_setreuid(l, &bsa, retval); 1161 } 1162 1163 int 1164 linux_sys_setregid16(l, v, retval) 1165 struct lwp *l; 1166 void *v; 1167 register_t *retval; 1168 { 1169 struct linux_sys_setregid16_args /* { 1170 syscallarg(int) rgid; 1171 syscallarg(int) egid; 1172 } */ *uap = v; 1173 struct sys_setregid_args bsa; 1174 1175 SCARG(&bsa, rgid) = ((linux_gid_t)SCARG(uap, rgid) == (linux_gid_t)-1) ? 1176 (uid_t)-1 : SCARG(uap, rgid); 1177 SCARG(&bsa, egid) = ((linux_gid_t)SCARG(uap, egid) == (linux_gid_t)-1) ? 1178 (uid_t)-1 : SCARG(uap, egid); 1179 1180 return sys_setregid(l, &bsa, retval); 1181 } 1182 1183 int 1184 linux_sys_setresuid16(l, v, retval) 1185 struct lwp *l; 1186 void *v; 1187 register_t *retval; 1188 { 1189 struct linux_sys_setresuid16_args /* { 1190 syscallarg(uid_t) ruid; 1191 syscallarg(uid_t) euid; 1192 syscallarg(uid_t) suid; 1193 } */ *uap = v; 1194 struct linux_sys_setresuid16_args lsa; 1195 1196 SCARG(&lsa, ruid) = ((linux_uid_t)SCARG(uap, ruid) == (linux_uid_t)-1) ? 1197 (uid_t)-1 : SCARG(uap, ruid); 1198 SCARG(&lsa, euid) = ((linux_uid_t)SCARG(uap, euid) == (linux_uid_t)-1) ? 1199 (uid_t)-1 : SCARG(uap, euid); 1200 SCARG(&lsa, suid) = ((linux_uid_t)SCARG(uap, suid) == (linux_uid_t)-1) ? 1201 (uid_t)-1 : SCARG(uap, suid); 1202 1203 return linux_sys_setresuid(l, &lsa, retval); 1204 } 1205 1206 int 1207 linux_sys_setresgid16(l, v, retval) 1208 struct lwp *l; 1209 void *v; 1210 register_t *retval; 1211 { 1212 struct linux_sys_setresgid16_args /* { 1213 syscallarg(gid_t) rgid; 1214 syscallarg(gid_t) egid; 1215 syscallarg(gid_t) sgid; 1216 } */ *uap = v; 1217 struct linux_sys_setresgid16_args lsa; 1218 1219 SCARG(&lsa, rgid) = ((linux_gid_t)SCARG(uap, rgid) == (linux_gid_t)-1) ? 1220 (gid_t)-1 : SCARG(uap, rgid); 1221 SCARG(&lsa, egid) = ((linux_gid_t)SCARG(uap, egid) == (linux_gid_t)-1) ? 1222 (gid_t)-1 : SCARG(uap, egid); 1223 SCARG(&lsa, sgid) = ((linux_gid_t)SCARG(uap, sgid) == (linux_gid_t)-1) ? 1224 (gid_t)-1 : SCARG(uap, sgid); 1225 1226 return linux_sys_setresgid(l, &lsa, retval); 1227 } 1228 1229 int 1230 linux_sys_getgroups16(l, v, retval) 1231 struct lwp *l; 1232 void *v; 1233 register_t *retval; 1234 { 1235 struct linux_sys_getgroups16_args /* { 1236 syscallarg(int) gidsetsize; 1237 syscallarg(linux_gid_t *) gidset; 1238 } */ *uap = v; 1239 struct proc *p = l->l_proc; 1240 caddr_t sg; 1241 int n, error, i; 1242 struct sys_getgroups_args bsa; 1243 gid_t *bset, *kbset; 1244 linux_gid_t *lset; 1245 kauth_cred_t pc = l->l_cred; 1246 1247 n = SCARG(uap, gidsetsize); 1248 if (n < 0) 1249 return EINVAL; 1250 error = 0; 1251 bset = kbset = NULL; 1252 lset = NULL; 1253 if (n > 0) { 1254 n = min(kauth_cred_ngroups(pc), n); 1255 sg = stackgap_init(p, 0); 1256 bset = stackgap_alloc(p, &sg, n * sizeof (gid_t)); 1257 kbset = malloc(n * sizeof (gid_t), M_TEMP, M_WAITOK); 1258 lset = malloc(n * sizeof (linux_gid_t), M_TEMP, M_WAITOK); 1259 if (bset == NULL || kbset == NULL || lset == NULL) 1260 { 1261 error = ENOMEM; 1262 goto out; 1263 } 1264 SCARG(&bsa, gidsetsize) = n; 1265 SCARG(&bsa, gidset) = bset; 1266 error = sys_getgroups(l, &bsa, retval); 1267 if (error != 0) 1268 goto out; 1269 error = copyin(bset, kbset, n * sizeof (gid_t)); 1270 if (error != 0) 1271 goto out; 1272 for (i = 0; i < n; i++) 1273 lset[i] = (linux_gid_t)kbset[i]; 1274 error = copyout(lset, SCARG(uap, gidset), 1275 n * sizeof (linux_gid_t)); 1276 } else 1277 *retval = kauth_cred_ngroups(pc); 1278 out: 1279 if (kbset != NULL) 1280 free(kbset, M_TEMP); 1281 if (lset != NULL) 1282 free(lset, M_TEMP); 1283 return error; 1284 } 1285 1286 int 1287 linux_sys_setgroups16(l, v, retval) 1288 struct lwp *l; 1289 void *v; 1290 register_t *retval; 1291 { 1292 struct linux_sys_setgroups16_args /* { 1293 syscallarg(int) gidsetsize; 1294 syscallarg(linux_gid_t *) gidset; 1295 } */ *uap = v; 1296 struct proc *p = l->l_proc; 1297 caddr_t sg; 1298 int n; 1299 int error, i; 1300 struct sys_setgroups_args bsa; 1301 gid_t *bset, *kbset; 1302 linux_gid_t *lset; 1303 1304 n = SCARG(uap, gidsetsize); 1305 if (n < 0 || n > NGROUPS) 1306 return EINVAL; 1307 sg = stackgap_init(p, 0); 1308 bset = stackgap_alloc(p, &sg, n * sizeof (gid_t)); 1309 lset = malloc(n * sizeof (linux_gid_t), M_TEMP, M_WAITOK); 1310 kbset = malloc(n * sizeof (gid_t), M_TEMP, M_WAITOK); 1311 if (bset == NULL || kbset == NULL || lset == NULL) 1312 { 1313 error = ENOMEM; 1314 goto out; 1315 } 1316 error = copyin(SCARG(uap, gidset), lset, n * sizeof (linux_gid_t)); 1317 if (error != 0) 1318 goto out; 1319 for (i = 0; i < n; i++) 1320 kbset[i] = (gid_t)lset[i]; 1321 error = copyout(kbset, bset, n * sizeof (gid_t)); 1322 if (error != 0) 1323 goto out; 1324 SCARG(&bsa, gidsetsize) = n; 1325 SCARG(&bsa, gidset) = bset; 1326 error = sys_setgroups(l, &bsa, retval); 1327 1328 out: 1329 if (lset != NULL) 1330 free(lset, M_TEMP); 1331 if (kbset != NULL) 1332 free(kbset, M_TEMP); 1333 1334 return error; 1335 } 1336 1337 #endif /* __i386__ || __m68k__ || COMPAT_LINUX32 */ 1338 1339 #ifndef COMPAT_LINUX32 1340 /* 1341 * We have nonexistent fsuid equal to uid. 1342 * If modification is requested, refuse. 1343 */ 1344 int 1345 linux_sys_setfsuid(l, v, retval) 1346 struct lwp *l; 1347 void *v; 1348 register_t *retval; 1349 { 1350 struct linux_sys_setfsuid_args /* { 1351 syscallarg(uid_t) uid; 1352 } */ *uap = v; 1353 uid_t uid; 1354 1355 uid = SCARG(uap, uid); 1356 if (kauth_cred_getuid(l->l_cred) != uid) 1357 return sys_nosys(l, v, retval); 1358 else 1359 return (0); 1360 } 1361 1362 /* XXX XXX XXX */ 1363 # ifndef alpha 1364 int 1365 linux_sys_getfsuid(l, v, retval) 1366 struct lwp *l; 1367 void *v; 1368 register_t *retval; 1369 { 1370 return sys_getuid(l, v, retval); 1371 } 1372 # endif 1373 1374 int 1375 linux_sys_setresuid(l, v, retval) 1376 struct lwp *l; 1377 void *v; 1378 register_t *retval; 1379 { 1380 struct linux_sys_setresuid_args /* { 1381 syscallarg(uid_t) ruid; 1382 syscallarg(uid_t) euid; 1383 syscallarg(uid_t) suid; 1384 } */ *uap = v; 1385 1386 /* 1387 * Note: These checks are a little different than the NetBSD 1388 * setreuid(2) call performs. This precisely follows the 1389 * behavior of the Linux kernel. 1390 */ 1391 1392 return do_setresuid(l, SCARG(uap, ruid), SCARG(uap, euid), 1393 SCARG(uap, suid), 1394 ID_R_EQ_R | ID_R_EQ_E | ID_R_EQ_S | 1395 ID_E_EQ_R | ID_E_EQ_E | ID_E_EQ_S | 1396 ID_S_EQ_R | ID_S_EQ_E | ID_S_EQ_S ); 1397 } 1398 1399 int 1400 linux_sys_getresuid(l, v, retval) 1401 struct lwp *l; 1402 void *v; 1403 register_t *retval; 1404 { 1405 struct linux_sys_getresuid_args /* { 1406 syscallarg(uid_t *) ruid; 1407 syscallarg(uid_t *) euid; 1408 syscallarg(uid_t *) suid; 1409 } */ *uap = v; 1410 kauth_cred_t pc = l->l_cred; 1411 int error; 1412 uid_t uid; 1413 1414 /* 1415 * Linux copies these values out to userspace like so: 1416 * 1417 * 1. Copy out ruid. 1418 * 2. If that succeeds, copy out euid. 1419 * 3. If both of those succeed, copy out suid. 1420 */ 1421 uid = kauth_cred_getuid(pc); 1422 if ((error = copyout(&uid, SCARG(uap, ruid), sizeof(uid_t))) != 0) 1423 return (error); 1424 1425 uid = kauth_cred_geteuid(pc); 1426 if ((error = copyout(&uid, SCARG(uap, euid), sizeof(uid_t))) != 0) 1427 return (error); 1428 1429 uid = kauth_cred_getsvuid(pc); 1430 1431 return (copyout(&uid, SCARG(uap, suid), sizeof(uid_t))); 1432 } 1433 1434 int 1435 linux_sys_ptrace(l, v, retval) 1436 struct lwp *l; 1437 void *v; 1438 register_t *retval; 1439 { 1440 struct linux_sys_ptrace_args /* { 1441 i386, m68k, powerpc: T=int 1442 alpha, amd64: T=long 1443 syscallarg(T) request; 1444 syscallarg(T) pid; 1445 syscallarg(T) addr; 1446 syscallarg(T) data; 1447 } */ *uap = v; 1448 const int *ptr; 1449 int request; 1450 int error; 1451 1452 ptr = linux_ptrace_request_map; 1453 request = SCARG(uap, request); 1454 while (*ptr != -1) 1455 if (*ptr++ == request) { 1456 struct sys_ptrace_args pta; 1457 1458 SCARG(&pta, req) = *ptr; 1459 SCARG(&pta, pid) = SCARG(uap, pid); 1460 SCARG(&pta, addr) = (caddr_t)SCARG(uap, addr); 1461 SCARG(&pta, data) = SCARG(uap, data); 1462 1463 /* 1464 * Linux ptrace(PTRACE_CONT, pid, 0, 0) means actually 1465 * to continue where the process left off previously. 1466 * The same thing is achieved by addr == (caddr_t) 1 1467 * on NetBSD, so rewrite 'addr' appropriately. 1468 */ 1469 if (request == LINUX_PTRACE_CONT && SCARG(uap, addr)==0) 1470 SCARG(&pta, addr) = (caddr_t) 1; 1471 1472 error = sys_ptrace(l, &pta, retval); 1473 if (error) 1474 return error; 1475 switch (request) { 1476 case LINUX_PTRACE_PEEKTEXT: 1477 case LINUX_PTRACE_PEEKDATA: 1478 error = copyout (retval, 1479 (caddr_t)SCARG(uap, data), 1480 sizeof *retval); 1481 *retval = SCARG(uap, data); 1482 break; 1483 default: 1484 break; 1485 } 1486 return error; 1487 } 1488 else 1489 ptr++; 1490 1491 return LINUX_SYS_PTRACE_ARCH(l, uap, retval); 1492 } 1493 1494 int 1495 linux_sys_reboot(struct lwp *l, void *v, register_t *retval) 1496 { 1497 struct linux_sys_reboot_args /* { 1498 syscallarg(int) magic1; 1499 syscallarg(int) magic2; 1500 syscallarg(int) cmd; 1501 syscallarg(void *) arg; 1502 } */ *uap = v; 1503 struct sys_reboot_args /* { 1504 syscallarg(int) opt; 1505 syscallarg(char *) bootstr; 1506 } */ sra; 1507 int error; 1508 1509 if ((error = kauth_authorize_generic(l->l_cred, 1510 KAUTH_GENERIC_ISSUSER, &l->l_acflag)) != 0) 1511 return(error); 1512 1513 if (SCARG(uap, magic1) != LINUX_REBOOT_MAGIC1) 1514 return(EINVAL); 1515 if (SCARG(uap, magic2) != LINUX_REBOOT_MAGIC2 && 1516 SCARG(uap, magic2) != LINUX_REBOOT_MAGIC2A && 1517 SCARG(uap, magic2) != LINUX_REBOOT_MAGIC2B) 1518 return(EINVAL); 1519 1520 switch (SCARG(uap, cmd)) { 1521 case LINUX_REBOOT_CMD_RESTART: 1522 SCARG(&sra, opt) = RB_AUTOBOOT; 1523 break; 1524 case LINUX_REBOOT_CMD_HALT: 1525 SCARG(&sra, opt) = RB_HALT; 1526 break; 1527 case LINUX_REBOOT_CMD_POWER_OFF: 1528 SCARG(&sra, opt) = RB_HALT|RB_POWERDOWN; 1529 break; 1530 case LINUX_REBOOT_CMD_RESTART2: 1531 /* Reboot with an argument. */ 1532 SCARG(&sra, opt) = RB_AUTOBOOT|RB_STRING; 1533 SCARG(&sra, bootstr) = SCARG(uap, arg); 1534 break; 1535 case LINUX_REBOOT_CMD_CAD_ON: 1536 return(EINVAL); /* We don't implement ctrl-alt-delete */ 1537 case LINUX_REBOOT_CMD_CAD_OFF: 1538 return(0); 1539 default: 1540 return(EINVAL); 1541 } 1542 1543 return(sys_reboot(l, &sra, retval)); 1544 } 1545 1546 /* 1547 * Copy of compat_12_sys_swapon(). 1548 */ 1549 int 1550 linux_sys_swapon(l, v, retval) 1551 struct lwp *l; 1552 void *v; 1553 register_t *retval; 1554 { 1555 struct sys_swapctl_args ua; 1556 struct linux_sys_swapon_args /* { 1557 syscallarg(const char *) name; 1558 } */ *uap = v; 1559 1560 SCARG(&ua, cmd) = SWAP_ON; 1561 SCARG(&ua, arg) = (void *)__UNCONST(SCARG(uap, name)); 1562 SCARG(&ua, misc) = 0; /* priority */ 1563 return (sys_swapctl(l, &ua, retval)); 1564 } 1565 1566 /* 1567 * Stop swapping to the file or block device specified by path. 1568 */ 1569 int 1570 linux_sys_swapoff(l, v, retval) 1571 struct lwp *l; 1572 void *v; 1573 register_t *retval; 1574 { 1575 struct sys_swapctl_args ua; 1576 struct linux_sys_swapoff_args /* { 1577 syscallarg(const char *) path; 1578 } */ *uap = v; 1579 1580 SCARG(&ua, cmd) = SWAP_OFF; 1581 SCARG(&ua, arg) = __UNCONST(SCARG(uap, path)); /*XXXUNCONST*/ 1582 return (sys_swapctl(l, &ua, retval)); 1583 } 1584 1585 /* 1586 * Copy of compat_09_sys_setdomainname() 1587 */ 1588 /* ARGSUSED */ 1589 int 1590 linux_sys_setdomainname(l, v, retval) 1591 struct lwp *l; 1592 void *v; 1593 register_t *retval; 1594 { 1595 struct linux_sys_setdomainname_args /* { 1596 syscallarg(char *) domainname; 1597 syscallarg(int) len; 1598 } */ *uap = v; 1599 int name[2]; 1600 1601 name[0] = CTL_KERN; 1602 name[1] = KERN_DOMAINNAME; 1603 return (old_sysctl(&name[0], 2, 0, 0, SCARG(uap, domainname), 1604 SCARG(uap, len), l)); 1605 } 1606 1607 /* 1608 * sysinfo() 1609 */ 1610 /* ARGSUSED */ 1611 int 1612 linux_sys_sysinfo(l, v, retval) 1613 struct lwp *l; 1614 void *v; 1615 register_t *retval; 1616 { 1617 struct linux_sys_sysinfo_args /* { 1618 syscallarg(struct linux_sysinfo *) arg; 1619 } */ *uap = v; 1620 struct linux_sysinfo si; 1621 struct loadavg *la; 1622 1623 si.uptime = time_uptime; 1624 la = &averunnable; 1625 si.loads[0] = la->ldavg[0] * LINUX_SYSINFO_LOADS_SCALE / la->fscale; 1626 si.loads[1] = la->ldavg[1] * LINUX_SYSINFO_LOADS_SCALE / la->fscale; 1627 si.loads[2] = la->ldavg[2] * LINUX_SYSINFO_LOADS_SCALE / la->fscale; 1628 si.totalram = ctob(physmem); 1629 si.freeram = uvmexp.free * uvmexp.pagesize; 1630 si.sharedram = 0; /* XXX */ 1631 si.bufferram = uvmexp.filepages * uvmexp.pagesize; 1632 si.totalswap = uvmexp.swpages * uvmexp.pagesize; 1633 si.freeswap = (uvmexp.swpages - uvmexp.swpginuse) * uvmexp.pagesize; 1634 si.procs = nprocs; 1635 1636 /* The following are only present in newer Linux kernels. */ 1637 si.totalbig = 0; 1638 si.freebig = 0; 1639 si.mem_unit = 1; 1640 1641 return (copyout(&si, SCARG(uap, arg), sizeof si)); 1642 } 1643 1644 int 1645 linux_sys_getrlimit(l, v, retval) 1646 struct lwp *l; 1647 void *v; 1648 register_t *retval; 1649 { 1650 struct linux_sys_getrlimit_args /* { 1651 syscallarg(int) which; 1652 # ifdef LINUX_LARGEFILE64 1653 syscallarg(struct rlimit *) rlp; 1654 # else 1655 syscallarg(struct orlimit *) rlp; 1656 # endif 1657 } */ *uap = v; 1658 struct proc *p = l->l_proc; 1659 caddr_t sg = stackgap_init(p, 0); 1660 struct sys_getrlimit_args ap; 1661 struct rlimit rl; 1662 # ifdef LINUX_LARGEFILE64 1663 struct rlimit orl; 1664 # else 1665 struct orlimit orl; 1666 # endif 1667 int error; 1668 1669 SCARG(&ap, which) = linux_to_bsd_limit(SCARG(uap, which)); 1670 if ((error = SCARG(&ap, which)) < 0) 1671 return -error; 1672 SCARG(&ap, rlp) = stackgap_alloc(p, &sg, sizeof rl); 1673 if ((error = sys_getrlimit(l, &ap, retval)) != 0) 1674 return error; 1675 if ((error = copyin(SCARG(&ap, rlp), &rl, sizeof(rl))) != 0) 1676 return error; 1677 bsd_to_linux_rlimit(&orl, &rl); 1678 1679 return copyout(&orl, SCARG(uap, rlp), sizeof(orl)); 1680 } 1681 1682 int 1683 linux_sys_setrlimit(l, v, retval) 1684 struct lwp *l; 1685 void *v; 1686 register_t *retval; 1687 { 1688 struct linux_sys_setrlimit_args /* { 1689 syscallarg(int) which; 1690 # ifdef LINUX_LARGEFILE64 1691 syscallarg(struct rlimit *) rlp; 1692 # else 1693 syscallarg(struct orlimit *) rlp; 1694 # endif 1695 } */ *uap = v; 1696 struct proc *p = l->l_proc; 1697 caddr_t sg = stackgap_init(p, 0); 1698 struct sys_getrlimit_args ap; 1699 struct rlimit rl; 1700 # ifdef LINUX_LARGEFILE64 1701 struct rlimit orl; 1702 # else 1703 struct orlimit orl; 1704 # endif 1705 int error; 1706 1707 SCARG(&ap, which) = linux_to_bsd_limit(SCARG(uap, which)); 1708 SCARG(&ap, rlp) = stackgap_alloc(p, &sg, sizeof rl); 1709 if ((error = SCARG(&ap, which)) < 0) 1710 return -error; 1711 if ((error = copyin(SCARG(uap, rlp), &orl, sizeof(orl))) != 0) 1712 return error; 1713 linux_to_bsd_rlimit(&rl, &orl); 1714 if ((error = copyout(&rl, SCARG(&ap, rlp), sizeof(rl))) != 0) 1715 return error; 1716 return sys_setrlimit(l, &ap, retval); 1717 } 1718 1719 # if !defined(__mips__) && !defined(__amd64__) 1720 /* XXX: this doesn't look 100% common, at least mips doesn't have it */ 1721 int 1722 linux_sys_ugetrlimit(l, v, retval) 1723 struct lwp *l; 1724 void *v; 1725 register_t *retval; 1726 { 1727 return linux_sys_getrlimit(l, v, retval); 1728 } 1729 # endif 1730 1731 /* 1732 * This gets called for unsupported syscalls. The difference to sys_nosys() 1733 * is that process does not get SIGSYS, the call just returns with ENOSYS. 1734 * This is the way Linux does it and glibc depends on this behaviour. 1735 */ 1736 int 1737 linux_sys_nosys(l, v, retval) 1738 struct lwp *l; 1739 void *v; 1740 register_t *retval; 1741 { 1742 return (ENOSYS); 1743 } 1744 1745 int 1746 linux_sys_getpriority(l, v, retval) 1747 struct lwp *l; 1748 void *v; 1749 register_t *retval; 1750 { 1751 struct linux_sys_getpriority_args /* { 1752 syscallarg(int) which; 1753 syscallarg(int) who; 1754 } */ *uap = v; 1755 struct sys_getpriority_args bsa; 1756 int error; 1757 1758 SCARG(&bsa, which) = SCARG(uap, which); 1759 SCARG(&bsa, who) = SCARG(uap, who); 1760 1761 if ((error = sys_getpriority(l, &bsa, retval))) 1762 return error; 1763 1764 *retval = NZERO - *retval; 1765 1766 return 0; 1767 } 1768 1769 #endif /* !COMPAT_LINUX32 */ 1770