linux_misc.c revision 1.135.2.2
1 /* $NetBSD: linux_misc.c,v 1.135.2.2 2005/11/01 22:31:17 tron 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 1.135.2.2 2005/11/01 22:31:17 tron 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 99 #include <sys/ptrace.h> 100 #include <machine/ptrace.h> 101 102 #include <sys/sa.h> 103 #include <sys/syscallargs.h> 104 105 #include <compat/linux/common/linux_types.h> 106 #include <compat/linux/common/linux_signal.h> 107 108 #include <compat/linux/linux_syscallargs.h> 109 110 #include <compat/linux/common/linux_fcntl.h> 111 #include <compat/linux/common/linux_mmap.h> 112 #include <compat/linux/common/linux_dirent.h> 113 #include <compat/linux/common/linux_util.h> 114 #include <compat/linux/common/linux_misc.h> 115 #include <compat/linux/common/linux_ptrace.h> 116 #include <compat/linux/common/linux_reboot.h> 117 #include <compat/linux/common/linux_emuldata.h> 118 119 const int linux_ptrace_request_map[] = { 120 LINUX_PTRACE_TRACEME, PT_TRACE_ME, 121 LINUX_PTRACE_PEEKTEXT, PT_READ_I, 122 LINUX_PTRACE_PEEKDATA, PT_READ_D, 123 LINUX_PTRACE_POKETEXT, PT_WRITE_I, 124 LINUX_PTRACE_POKEDATA, PT_WRITE_D, 125 LINUX_PTRACE_CONT, PT_CONTINUE, 126 LINUX_PTRACE_KILL, PT_KILL, 127 LINUX_PTRACE_ATTACH, PT_ATTACH, 128 LINUX_PTRACE_DETACH, PT_DETACH, 129 #ifdef PT_STEP 130 LINUX_PTRACE_SINGLESTEP, PT_STEP, 131 #endif 132 -1 133 }; 134 135 const struct linux_mnttypes linux_fstypes[] = { 136 { MOUNT_FFS, LINUX_DEFAULT_SUPER_MAGIC }, 137 { MOUNT_NFS, LINUX_NFS_SUPER_MAGIC }, 138 { MOUNT_MFS, LINUX_DEFAULT_SUPER_MAGIC }, 139 { MOUNT_MSDOS, LINUX_MSDOS_SUPER_MAGIC }, 140 { MOUNT_LFS, LINUX_DEFAULT_SUPER_MAGIC }, 141 { MOUNT_FDESC, LINUX_DEFAULT_SUPER_MAGIC }, 142 { MOUNT_PORTAL, LINUX_DEFAULT_SUPER_MAGIC }, 143 { MOUNT_NULL, LINUX_DEFAULT_SUPER_MAGIC }, 144 { MOUNT_OVERLAY, LINUX_DEFAULT_SUPER_MAGIC }, 145 { MOUNT_UMAP, LINUX_DEFAULT_SUPER_MAGIC }, 146 { MOUNT_KERNFS, LINUX_DEFAULT_SUPER_MAGIC }, 147 { MOUNT_PROCFS, LINUX_PROC_SUPER_MAGIC }, 148 { MOUNT_AFS, LINUX_DEFAULT_SUPER_MAGIC }, 149 { MOUNT_CD9660, LINUX_ISOFS_SUPER_MAGIC }, 150 { MOUNT_UNION, LINUX_DEFAULT_SUPER_MAGIC }, 151 { MOUNT_ADOSFS, LINUX_ADFS_SUPER_MAGIC }, 152 { MOUNT_EXT2FS, LINUX_EXT2_SUPER_MAGIC }, 153 { MOUNT_CFS, LINUX_DEFAULT_SUPER_MAGIC }, 154 { MOUNT_CODA, LINUX_CODA_SUPER_MAGIC }, 155 { MOUNT_FILECORE, LINUX_DEFAULT_SUPER_MAGIC }, 156 { MOUNT_NTFS, LINUX_DEFAULT_SUPER_MAGIC }, 157 { MOUNT_SMBFS, LINUX_SMB_SUPER_MAGIC }, 158 { MOUNT_PTYFS, LINUX_DEVPTS_SUPER_MAGIC } 159 }; 160 const int linux_fstypes_cnt = sizeof(linux_fstypes) / sizeof(linux_fstypes[0]); 161 162 #ifdef DEBUG_LINUX 163 #define DPRINTF(a) uprintf a 164 #else 165 #define DPRINTF(a) 166 #endif 167 168 /* Local linux_misc.c functions: */ 169 static void bsd_to_linux_statfs __P((const struct statvfs *, 170 struct linux_statfs *)); 171 static int linux_to_bsd_limit __P((int)); 172 static void linux_to_bsd_mmap_args __P((struct sys_mmap_args *, 173 const struct linux_sys_mmap_args *)); 174 static int linux_mmap __P((struct lwp *, struct linux_sys_mmap_args *, 175 register_t *, off_t)); 176 177 178 /* 179 * The information on a terminated (or stopped) process needs 180 * to be converted in order for Linux binaries to get a valid signal 181 * number out of it. 182 */ 183 void 184 bsd_to_linux_wstat(st) 185 int *st; 186 { 187 188 int sig; 189 190 if (WIFSIGNALED(*st)) { 191 sig = WTERMSIG(*st); 192 if (sig >= 0 && sig < NSIG) 193 *st= (*st& ~0177) | native_to_linux_signo[sig]; 194 } else if (WIFSTOPPED(*st)) { 195 sig = WSTOPSIG(*st); 196 if (sig >= 0 && sig < NSIG) 197 *st = (*st & ~0xff00) | 198 (native_to_linux_signo[sig] << 8); 199 } 200 } 201 202 /* 203 * wait4(2). Passed on to the NetBSD call, surrounded by code to 204 * reserve some space for a NetBSD-style wait status, and converting 205 * it to what Linux wants. 206 */ 207 int 208 linux_sys_wait4(l, v, retval) 209 struct lwp *l; 210 void *v; 211 register_t *retval; 212 { 213 struct linux_sys_wait4_args /* { 214 syscallarg(int) pid; 215 syscallarg(int *) status; 216 syscallarg(int) options; 217 syscallarg(struct rusage *) rusage; 218 } */ *uap = v; 219 struct proc *p = l->l_proc; 220 struct sys_wait4_args w4a; 221 int error, *status, tstat, options, linux_options; 222 caddr_t sg; 223 224 if (SCARG(uap, status) != NULL) { 225 sg = stackgap_init(p, 0); 226 status = (int *) stackgap_alloc(p, &sg, sizeof *status); 227 } else 228 status = NULL; 229 230 linux_options = SCARG(uap, options); 231 options = 0; 232 if (linux_options & ~(LINUX_WAIT4_KNOWNFLAGS)) 233 return (EINVAL); 234 235 if (linux_options & LINUX_WAIT4_WNOHANG) 236 options |= WNOHANG; 237 if (linux_options & LINUX_WAIT4_WUNTRACED) 238 options |= WUNTRACED; 239 if (linux_options & LINUX_WAIT4_WALL) 240 options |= WALLSIG; 241 if (linux_options & LINUX_WAIT4_WCLONE) 242 options |= WALTSIG; 243 #ifdef DIAGNOSTIC 244 if (linux_options & LINUX_WAIT4_WNOTHREAD) 245 printf("WARNING: %s: linux process %d.%d called " 246 "waitpid with __WNOTHREAD set!", 247 __FILE__, p->p_pid, l->l_lid); 248 249 #endif 250 251 SCARG(&w4a, pid) = SCARG(uap, pid); 252 SCARG(&w4a, status) = status; 253 SCARG(&w4a, options) = options; 254 SCARG(&w4a, rusage) = SCARG(uap, rusage); 255 256 if ((error = sys_wait4(l, &w4a, retval))) 257 return error; 258 259 sigdelset(&p->p_sigctx.ps_siglist, SIGCHLD); 260 261 if (status != NULL) { 262 if ((error = copyin(status, &tstat, sizeof tstat))) 263 return error; 264 265 bsd_to_linux_wstat(&tstat); 266 return copyout(&tstat, SCARG(uap, status), sizeof tstat); 267 } 268 269 return 0; 270 } 271 272 /* 273 * Linux brk(2). The check if the new address is >= the old one is 274 * done in the kernel in Linux. NetBSD does it in the library. 275 */ 276 int 277 linux_sys_brk(l, v, retval) 278 struct lwp *l; 279 void *v; 280 register_t *retval; 281 { 282 struct linux_sys_brk_args /* { 283 syscallarg(char *) nsize; 284 } */ *uap = v; 285 struct proc *p = l->l_proc; 286 char *nbrk = SCARG(uap, nsize); 287 struct sys_obreak_args oba; 288 struct vmspace *vm = p->p_vmspace; 289 struct linux_emuldata *ed = (struct linux_emuldata*)p->p_emuldata; 290 291 SCARG(&oba, nsize) = nbrk; 292 293 if ((caddr_t) nbrk > vm->vm_daddr && sys_obreak(l, &oba, retval) == 0) 294 ed->s->p_break = (char*)nbrk; 295 else 296 nbrk = ed->s->p_break; 297 298 retval[0] = (register_t)nbrk; 299 300 return 0; 301 } 302 303 /* 304 * Convert NetBSD statvfs structure to Linux statfs structure. 305 * Linux doesn't have f_flag, and we can't set f_frsize due 306 * to glibc statvfs() bug (see below). 307 */ 308 static void 309 bsd_to_linux_statfs(bsp, lsp) 310 const struct statvfs *bsp; 311 struct linux_statfs *lsp; 312 { 313 int i; 314 315 for (i = 0; i < linux_fstypes_cnt; i++) { 316 if (strcmp(bsp->f_fstypename, linux_fstypes[i].bsd) == 0) { 317 lsp->l_ftype = linux_fstypes[i].linux; 318 break; 319 } 320 } 321 322 if (i == linux_fstypes_cnt) { 323 DPRINTF(("unhandled fstype in linux emulation: %s\n", 324 bsp->f_fstypename)); 325 lsp->l_ftype = LINUX_DEFAULT_SUPER_MAGIC; 326 } 327 328 /* 329 * The sizes are expressed in number of blocks. The block 330 * size used for the size is f_frsize for POSIX-compliant 331 * statvfs. Linux statfs uses f_bsize as the block size 332 * (f_frsize used to not be available in Linux struct statfs). 333 * However, glibc 2.3.3 statvfs() wrapper fails to adjust the block 334 * counts for different f_frsize if f_frsize is provided by the kernel. 335 * POSIX conforming apps thus get wrong size if f_frsize 336 * is different to f_bsize. Thus, we just pretend we don't 337 * support f_frsize. 338 */ 339 340 lsp->l_fbsize = bsp->f_frsize; 341 lsp->l_ffrsize = 0; /* compat */ 342 lsp->l_fblocks = bsp->f_blocks; 343 lsp->l_fbfree = bsp->f_bfree; 344 lsp->l_fbavail = bsp->f_bavail; 345 lsp->l_ffiles = bsp->f_files; 346 lsp->l_fffree = bsp->f_ffree; 347 /* Linux sets the fsid to 0..., we don't */ 348 lsp->l_ffsid.val[0] = bsp->f_fsidx.__fsid_val[0]; 349 lsp->l_ffsid.val[1] = bsp->f_fsidx.__fsid_val[1]; 350 lsp->l_fnamelen = bsp->f_namemax; 351 (void)memset(lsp->l_fspare, 0, sizeof(lsp->l_fspare)); 352 } 353 354 /* 355 * Implement the fs stat functions. Straightforward. 356 */ 357 int 358 linux_sys_statfs(l, v, retval) 359 struct lwp *l; 360 void *v; 361 register_t *retval; 362 { 363 struct linux_sys_statfs_args /* { 364 syscallarg(const char *) path; 365 syscallarg(struct linux_statfs *) sp; 366 } */ *uap = v; 367 struct proc *p = l->l_proc; 368 struct statvfs btmp, *bsp; 369 struct linux_statfs ltmp; 370 struct sys_statvfs1_args bsa; 371 caddr_t sg; 372 int error; 373 374 sg = stackgap_init(p, 0); 375 bsp = (struct statvfs *) stackgap_alloc(p, &sg, sizeof (struct statvfs)); 376 377 CHECK_ALT_EXIST(p, &sg, SCARG(uap, path)); 378 379 SCARG(&bsa, path) = SCARG(uap, path); 380 SCARG(&bsa, buf) = bsp; 381 SCARG(&bsa, flags) = ST_WAIT; 382 383 if ((error = sys_statvfs1(l, &bsa, retval))) 384 return error; 385 386 if ((error = copyin((caddr_t) bsp, (caddr_t) &btmp, sizeof btmp))) 387 return error; 388 389 bsd_to_linux_statfs(&btmp, <mp); 390 391 return copyout((caddr_t) <mp, (caddr_t) SCARG(uap, sp), sizeof ltmp); 392 } 393 394 int 395 linux_sys_fstatfs(l, v, retval) 396 struct lwp *l; 397 void *v; 398 register_t *retval; 399 { 400 struct linux_sys_fstatfs_args /* { 401 syscallarg(int) fd; 402 syscallarg(struct linux_statfs *) sp; 403 } */ *uap = v; 404 struct proc *p = l->l_proc; 405 struct statvfs btmp, *bsp; 406 struct linux_statfs ltmp; 407 struct sys_fstatvfs1_args bsa; 408 caddr_t sg; 409 int error; 410 411 sg = stackgap_init(p, 0); 412 bsp = (struct statvfs *) stackgap_alloc(p, &sg, sizeof (struct statvfs)); 413 414 SCARG(&bsa, fd) = SCARG(uap, fd); 415 SCARG(&bsa, buf) = bsp; 416 SCARG(&bsa, flags) = ST_WAIT; 417 418 if ((error = sys_fstatvfs1(l, &bsa, retval))) 419 return error; 420 421 if ((error = copyin((caddr_t) bsp, (caddr_t) &btmp, sizeof btmp))) 422 return error; 423 424 bsd_to_linux_statfs(&btmp, <mp); 425 426 return copyout((caddr_t) <mp, (caddr_t) SCARG(uap, sp), sizeof ltmp); 427 } 428 429 /* 430 * uname(). Just copy the info from the various strings stored in the 431 * kernel, and put it in the Linux utsname structure. That structure 432 * is almost the same as the NetBSD one, only it has fields 65 characters 433 * long, and an extra domainname field. 434 */ 435 int 436 linux_sys_uname(l, v, retval) 437 struct lwp *l; 438 void *v; 439 register_t *retval; 440 { 441 struct linux_sys_uname_args /* { 442 syscallarg(struct linux_utsname *) up; 443 } */ *uap = v; 444 struct linux_utsname luts; 445 446 strncpy(luts.l_sysname, linux_sysname, sizeof(luts.l_sysname)); 447 strncpy(luts.l_nodename, hostname, sizeof(luts.l_nodename)); 448 strncpy(luts.l_release, linux_release, sizeof(luts.l_release)); 449 strncpy(luts.l_version, linux_version, sizeof(luts.l_version)); 450 strncpy(luts.l_machine, machine, sizeof(luts.l_machine)); 451 strncpy(luts.l_domainname, domainname, sizeof(luts.l_domainname)); 452 453 return copyout(&luts, SCARG(uap, up), sizeof(luts)); 454 } 455 456 /* Used directly on: alpha, mips, ppc, sparc, sparc64 */ 457 /* Used indirectly on: arm, i386, m68k */ 458 459 /* 460 * New type Linux mmap call. 461 * Only called directly on machines with >= 6 free regs. 462 */ 463 int 464 linux_sys_mmap(l, v, retval) 465 struct lwp *l; 466 void *v; 467 register_t *retval; 468 { 469 struct linux_sys_mmap_args /* { 470 syscallarg(unsigned long) addr; 471 syscallarg(size_t) len; 472 syscallarg(int) prot; 473 syscallarg(int) flags; 474 syscallarg(int) fd; 475 syscallarg(linux_off_t) offset; 476 } */ *uap = v; 477 478 if (SCARG(uap, offset) & PAGE_MASK) 479 return EINVAL; 480 481 return linux_mmap(l, uap, retval, SCARG(uap, offset)); 482 } 483 484 /* 485 * Guts of most architectures' mmap64() implementations. This shares 486 * its list of arguments with linux_sys_mmap(). 487 * 488 * The difference in linux_sys_mmap2() is that "offset" is actually 489 * (offset / pagesize), not an absolute byte count. This translation 490 * to pagesize offsets is done inside glibc between the mmap64() call 491 * point, and the actual syscall. 492 */ 493 int 494 linux_sys_mmap2(l, v, retval) 495 struct lwp *l; 496 void *v; 497 register_t *retval; 498 { 499 struct linux_sys_mmap2_args /* { 500 syscallarg(unsigned long) addr; 501 syscallarg(size_t) len; 502 syscallarg(int) prot; 503 syscallarg(int) flags; 504 syscallarg(int) fd; 505 syscallarg(linux_off_t) offset; 506 } */ *uap = v; 507 508 return linux_mmap(l, uap, retval, 509 ((off_t)SCARG(uap, offset)) << PAGE_SHIFT); 510 } 511 512 /* 513 * Massage arguments and call system mmap(2). 514 */ 515 static int 516 linux_mmap(l, uap, retval, offset) 517 struct lwp *l; 518 struct linux_sys_mmap_args *uap; 519 register_t *retval; 520 off_t offset; 521 { 522 struct sys_mmap_args cma; 523 int error; 524 size_t mmoff=0; 525 526 if (SCARG(uap, flags) & LINUX_MAP_GROWSDOWN) { 527 /* 528 * Request for stack-like memory segment. On linux, this 529 * works by mmap()ping (small) segment, which is automatically 530 * extended when page fault happens below the currently 531 * allocated area. We emulate this by allocating (typically 532 * bigger) segment sized at current stack size limit, and 533 * offsetting the requested and returned address accordingly. 534 * Since physical pages are only allocated on-demand, this 535 * is effectively identical. 536 */ 537 rlim_t ssl = l->l_proc->p_rlimit[RLIMIT_STACK].rlim_cur; 538 539 if (SCARG(uap, len) < ssl) { 540 /* Compute the address offset */ 541 mmoff = round_page(ssl) - SCARG(uap, len); 542 543 if (SCARG(uap, addr)) 544 SCARG(uap, addr) -= mmoff; 545 546 SCARG(uap, len) = (size_t) ssl; 547 } 548 } 549 550 linux_to_bsd_mmap_args(&cma, uap); 551 SCARG(&cma, pos) = offset; 552 553 error = sys_mmap(l, &cma, retval); 554 if (error) 555 return (error); 556 557 /* Shift the returned address for stack-like segment if necessary */ 558 if (SCARG(uap, flags) & LINUX_MAP_GROWSDOWN && mmoff) 559 retval[0] += mmoff; 560 561 return (0); 562 } 563 564 static void 565 linux_to_bsd_mmap_args(cma, uap) 566 struct sys_mmap_args *cma; 567 const struct linux_sys_mmap_args *uap; 568 { 569 int flags = MAP_TRYFIXED, fl = SCARG(uap, flags); 570 571 flags |= cvtto_bsd_mask(fl, LINUX_MAP_SHARED, MAP_SHARED); 572 flags |= cvtto_bsd_mask(fl, LINUX_MAP_PRIVATE, MAP_PRIVATE); 573 flags |= cvtto_bsd_mask(fl, LINUX_MAP_FIXED, MAP_FIXED); 574 flags |= cvtto_bsd_mask(fl, LINUX_MAP_ANON, MAP_ANON); 575 /* XXX XAX ERH: Any other flags here? There are more defined... */ 576 577 SCARG(cma, addr) = (void *)SCARG(uap, addr); 578 SCARG(cma, len) = SCARG(uap, len); 579 SCARG(cma, prot) = SCARG(uap, prot); 580 if (SCARG(cma, prot) & VM_PROT_WRITE) /* XXX */ 581 SCARG(cma, prot) |= VM_PROT_READ; 582 SCARG(cma, flags) = flags; 583 SCARG(cma, fd) = flags & MAP_ANON ? -1 : SCARG(uap, fd); 584 SCARG(cma, pad) = 0; 585 } 586 587 int 588 linux_sys_mremap(l, v, retval) 589 struct lwp *l; 590 void *v; 591 register_t *retval; 592 { 593 struct linux_sys_mremap_args /* { 594 syscallarg(void *) old_address; 595 syscallarg(size_t) old_size; 596 syscallarg(size_t) new_size; 597 syscallarg(u_long) flags; 598 } */ *uap = v; 599 struct sys_munmap_args mua; 600 size_t old_size, new_size; 601 int error; 602 603 old_size = round_page(SCARG(uap, old_size)); 604 new_size = round_page(SCARG(uap, new_size)); 605 606 /* 607 * Growing mapped region. 608 */ 609 if (new_size > old_size) { 610 /* 611 * XXX Implement me. What we probably want to do is 612 * XXX dig out the guts of the old mapping, mmap that 613 * XXX object again with the new size, then munmap 614 * XXX the old mapping. 615 */ 616 *retval = 0; 617 return (ENOMEM); 618 } 619 620 /* 621 * Shrinking mapped region. 622 */ 623 if (new_size < old_size) { 624 SCARG(&mua, addr) = (caddr_t)SCARG(uap, old_address) + 625 new_size; 626 SCARG(&mua, len) = old_size - new_size; 627 error = sys_munmap(l, &mua, retval); 628 *retval = error ? 0 : (register_t)SCARG(uap, old_address); 629 return (error); 630 } 631 632 /* 633 * No change. 634 */ 635 *retval = (register_t)SCARG(uap, old_address); 636 return (0); 637 } 638 639 int 640 linux_sys_msync(l, v, retval) 641 struct lwp *l; 642 void *v; 643 register_t *retval; 644 { 645 struct linux_sys_msync_args /* { 646 syscallarg(caddr_t) addr; 647 syscallarg(int) len; 648 syscallarg(int) fl; 649 } */ *uap = v; 650 651 struct sys___msync13_args bma; 652 653 /* flags are ignored */ 654 SCARG(&bma, addr) = SCARG(uap, addr); 655 SCARG(&bma, len) = SCARG(uap, len); 656 SCARG(&bma, flags) = SCARG(uap, fl); 657 658 return sys___msync13(l, &bma, retval); 659 } 660 661 int 662 linux_sys_mprotect(l, v, retval) 663 struct lwp *l; 664 void *v; 665 register_t *retval; 666 { 667 struct linux_sys_mprotect_args /* { 668 syscallarg(const void *) start; 669 syscallarg(unsigned long) len; 670 syscallarg(int) prot; 671 } */ *uap = v; 672 struct vm_map_entry *entry; 673 struct vm_map *map; 674 struct proc *p; 675 vaddr_t end, start, len, stacklim; 676 int prot, grows; 677 678 start = (vaddr_t)SCARG(uap, start); 679 len = round_page(SCARG(uap, len)); 680 prot = SCARG(uap, prot); 681 grows = prot & (LINUX_PROT_GROWSDOWN | LINUX_PROT_GROWSUP); 682 prot &= ~grows; 683 end = start + len; 684 685 if (start & PAGE_MASK) 686 return EINVAL; 687 if (end < start) 688 return EINVAL; 689 if (end == start) 690 return 0; 691 692 if (prot & ~(PROT_READ | PROT_WRITE | PROT_EXEC)) 693 return EINVAL; 694 if (grows == (LINUX_PROT_GROWSDOWN | LINUX_PROT_GROWSUP)) 695 return EINVAL; 696 697 p = l->l_proc; 698 map = &p->p_vmspace->vm_map; 699 vm_map_lock(map); 700 #ifdef notdef 701 VM_MAP_RANGE_CHECK(map, start, end); 702 #endif 703 if (!uvm_map_lookup_entry(map, start, &entry) || entry->start > start) { 704 vm_map_unlock(map); 705 return ENOMEM; 706 } 707 708 /* 709 * Approximate the behaviour of PROT_GROWS{DOWN,UP}. 710 */ 711 712 stacklim = (vaddr_t)p->p_limit->pl_rlimit[RLIMIT_STACK].rlim_cur; 713 if (grows & LINUX_PROT_GROWSDOWN) { 714 if (USRSTACK - stacklim <= start && start < USRSTACK) { 715 start = USRSTACK - stacklim; 716 } else { 717 start = entry->start; 718 } 719 } else if (grows & LINUX_PROT_GROWSUP) { 720 if (USRSTACK <= end && end < USRSTACK + stacklim) { 721 end = USRSTACK + stacklim; 722 } else { 723 end = entry->end; 724 } 725 } 726 vm_map_unlock(map); 727 return uvm_map_protect(map, start, end, prot, FALSE); 728 } 729 730 /* 731 * This code is partly stolen from src/lib/libc/compat-43/times.c 732 */ 733 734 #define CONVTCK(r) (r.tv_sec * hz + r.tv_usec / (1000000 / hz)) 735 736 int 737 linux_sys_times(l, v, retval) 738 struct lwp *l; 739 void *v; 740 register_t *retval; 741 { 742 struct linux_sys_times_args /* { 743 syscallarg(struct times *) tms; 744 } */ *uap = v; 745 struct proc *p = l->l_proc; 746 struct timeval t; 747 int error, s; 748 749 if (SCARG(uap, tms)) { 750 struct linux_tms ltms; 751 struct rusage ru; 752 753 calcru(p, &ru.ru_utime, &ru.ru_stime, NULL); 754 ltms.ltms_utime = CONVTCK(ru.ru_utime); 755 ltms.ltms_stime = CONVTCK(ru.ru_stime); 756 757 ltms.ltms_cutime = CONVTCK(p->p_stats->p_cru.ru_utime); 758 ltms.ltms_cstime = CONVTCK(p->p_stats->p_cru.ru_stime); 759 760 if ((error = copyout(<ms, SCARG(uap, tms), sizeof ltms))) 761 return error; 762 } 763 764 s = splclock(); 765 timersub(&time, &boottime, &t); 766 splx(s); 767 768 retval[0] = ((linux_clock_t)(CONVTCK(t))); 769 return 0; 770 } 771 772 #undef CONVTCK 773 774 /* 775 * Linux 'readdir' call. This code is mostly taken from the 776 * SunOS getdents call (see compat/sunos/sunos_misc.c), though 777 * an attempt has been made to keep it a little cleaner (failing 778 * miserably, because of the cruft needed if count 1 is passed). 779 * 780 * The d_off field should contain the offset of the next valid entry, 781 * but in Linux it has the offset of the entry itself. We emulate 782 * that bug here. 783 * 784 * Read in BSD-style entries, convert them, and copy them out. 785 * 786 * Note that this doesn't handle union-mounted filesystems. 787 */ 788 int 789 linux_sys_getdents(l, v, retval) 790 struct lwp *l; 791 void *v; 792 register_t *retval; 793 { 794 struct linux_sys_getdents_args /* { 795 syscallarg(int) fd; 796 syscallarg(struct linux_dirent *) dent; 797 syscallarg(unsigned int) count; 798 } */ *uap = v; 799 struct proc *p = l->l_proc; 800 struct dirent *bdp; 801 struct vnode *vp; 802 caddr_t inp, buf; /* BSD-format */ 803 int len, reclen; /* BSD-format */ 804 caddr_t outp; /* Linux-format */ 805 int resid, linux_reclen = 0; /* Linux-format */ 806 struct file *fp; 807 struct uio auio; 808 struct iovec aiov; 809 struct linux_dirent idb; 810 off_t off; /* true file offset */ 811 int buflen, error, eofflag, nbytes, oldcall; 812 struct vattr va; 813 off_t *cookiebuf = NULL, *cookie; 814 int ncookies; 815 816 /* getvnode() will use the descriptor for us */ 817 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0) 818 return (error); 819 820 if ((fp->f_flag & FREAD) == 0) { 821 error = EBADF; 822 goto out1; 823 } 824 825 vp = (struct vnode *)fp->f_data; 826 if (vp->v_type != VDIR) { 827 error = EINVAL; 828 goto out1; 829 } 830 831 if ((error = VOP_GETATTR(vp, &va, p->p_ucred, p))) 832 goto out1; 833 834 nbytes = SCARG(uap, count); 835 if (nbytes == 1) { /* emulating old, broken behaviour */ 836 nbytes = sizeof (idb); 837 buflen = max(va.va_blocksize, nbytes); 838 oldcall = 1; 839 } else { 840 buflen = min(MAXBSIZE, nbytes); 841 if (buflen < va.va_blocksize) 842 buflen = va.va_blocksize; 843 oldcall = 0; 844 } 845 buf = malloc(buflen, M_TEMP, M_WAITOK); 846 847 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); 848 off = fp->f_offset; 849 again: 850 aiov.iov_base = buf; 851 aiov.iov_len = buflen; 852 auio.uio_iov = &aiov; 853 auio.uio_iovcnt = 1; 854 auio.uio_rw = UIO_READ; 855 auio.uio_segflg = UIO_SYSSPACE; 856 auio.uio_procp = NULL; 857 auio.uio_resid = buflen; 858 auio.uio_offset = off; 859 /* 860 * First we read into the malloc'ed buffer, then 861 * we massage it into user space, one record at a time. 862 */ 863 error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, &cookiebuf, 864 &ncookies); 865 if (error) 866 goto out; 867 868 inp = buf; 869 outp = (caddr_t)SCARG(uap, dent); 870 resid = nbytes; 871 if ((len = buflen - auio.uio_resid) == 0) 872 goto eof; 873 874 for (cookie = cookiebuf; len > 0; len -= reclen) { 875 bdp = (struct dirent *)inp; 876 reclen = bdp->d_reclen; 877 if (reclen & 3) 878 panic("linux_readdir"); 879 if (bdp->d_fileno == 0) { 880 inp += reclen; /* it is a hole; squish it out */ 881 if (cookie) 882 off = *cookie++; 883 else 884 off += reclen; 885 continue; 886 } 887 linux_reclen = LINUX_RECLEN(&idb, bdp->d_namlen); 888 if (reclen > len || resid < linux_reclen) { 889 /* entry too big for buffer, so just stop */ 890 outp++; 891 break; 892 } 893 /* 894 * Massage in place to make a Linux-shaped dirent (otherwise 895 * we have to worry about touching user memory outside of 896 * the copyout() call). 897 */ 898 idb.d_ino = bdp->d_fileno; 899 /* 900 * The old readdir() call misuses the offset and reclen fields. 901 */ 902 if (oldcall) { 903 idb.d_off = (linux_off_t)linux_reclen; 904 idb.d_reclen = (u_short)bdp->d_namlen; 905 } else { 906 if (sizeof (idb.d_off) <= 4 && (off >> 32) != 0) { 907 compat_offseterr(vp, "linux_getdents"); 908 error = EINVAL; 909 goto out; 910 } 911 idb.d_off = (linux_off_t)off; 912 idb.d_reclen = (u_short)linux_reclen; 913 } 914 strcpy(idb.d_name, bdp->d_name); 915 if ((error = copyout((caddr_t)&idb, outp, linux_reclen))) 916 goto out; 917 /* advance past this real entry */ 918 inp += reclen; 919 if (cookie) 920 off = *cookie++; /* each entry points to itself */ 921 else 922 off += reclen; 923 /* advance output past Linux-shaped entry */ 924 outp += linux_reclen; 925 resid -= linux_reclen; 926 if (oldcall) 927 break; 928 } 929 930 /* if we squished out the whole block, try again */ 931 if (outp == (caddr_t)SCARG(uap, dent)) 932 goto again; 933 fp->f_offset = off; /* update the vnode offset */ 934 935 if (oldcall) 936 nbytes = resid + linux_reclen; 937 938 eof: 939 *retval = nbytes - resid; 940 out: 941 VOP_UNLOCK(vp, 0); 942 if (cookiebuf) 943 free(cookiebuf, M_TEMP); 944 free(buf, M_TEMP); 945 out1: 946 FILE_UNUSE(fp, p); 947 return error; 948 } 949 950 /* 951 * Even when just using registers to pass arguments to syscalls you can 952 * have 5 of them on the i386. So this newer version of select() does 953 * this. 954 */ 955 int 956 linux_sys_select(l, v, retval) 957 struct lwp *l; 958 void *v; 959 register_t *retval; 960 { 961 struct linux_sys_select_args /* { 962 syscallarg(int) nfds; 963 syscallarg(fd_set *) readfds; 964 syscallarg(fd_set *) writefds; 965 syscallarg(fd_set *) exceptfds; 966 syscallarg(struct timeval *) timeout; 967 } */ *uap = v; 968 969 return linux_select1(l, retval, SCARG(uap, nfds), SCARG(uap, readfds), 970 SCARG(uap, writefds), SCARG(uap, exceptfds), SCARG(uap, timeout)); 971 } 972 973 /* 974 * Common code for the old and new versions of select(). A couple of 975 * things are important: 976 * 1) return the amount of time left in the 'timeout' parameter 977 * 2) select never returns ERESTART on Linux, always return EINTR 978 */ 979 int 980 linux_select1(l, retval, nfds, readfds, writefds, exceptfds, timeout) 981 struct lwp *l; 982 register_t *retval; 983 int nfds; 984 fd_set *readfds, *writefds, *exceptfds; 985 struct timeval *timeout; 986 { 987 struct sys_select_args bsa; 988 struct proc *p = l->l_proc; 989 struct timeval tv0, tv1, utv, *tvp; 990 caddr_t sg; 991 int error; 992 993 SCARG(&bsa, nd) = nfds; 994 SCARG(&bsa, in) = readfds; 995 SCARG(&bsa, ou) = writefds; 996 SCARG(&bsa, ex) = exceptfds; 997 SCARG(&bsa, tv) = timeout; 998 999 /* 1000 * Store current time for computation of the amount of 1001 * time left. 1002 */ 1003 if (timeout) { 1004 if ((error = copyin(timeout, &utv, sizeof(utv)))) 1005 return error; 1006 if (itimerfix(&utv)) { 1007 /* 1008 * The timeval was invalid. Convert it to something 1009 * valid that will act as it does under Linux. 1010 */ 1011 sg = stackgap_init(p, 0); 1012 tvp = stackgap_alloc(p, &sg, sizeof(utv)); 1013 utv.tv_sec += utv.tv_usec / 1000000; 1014 utv.tv_usec %= 1000000; 1015 if (utv.tv_usec < 0) { 1016 utv.tv_sec -= 1; 1017 utv.tv_usec += 1000000; 1018 } 1019 if (utv.tv_sec < 0) 1020 timerclear(&utv); 1021 if ((error = copyout(&utv, tvp, sizeof(utv)))) 1022 return error; 1023 SCARG(&bsa, tv) = tvp; 1024 } 1025 microtime(&tv0); 1026 } 1027 1028 error = sys_select(l, &bsa, retval); 1029 if (error) { 1030 /* 1031 * See fs/select.c in the Linux kernel. Without this, 1032 * Maelstrom doesn't work. 1033 */ 1034 if (error == ERESTART) 1035 error = EINTR; 1036 return error; 1037 } 1038 1039 if (timeout) { 1040 if (*retval) { 1041 /* 1042 * Compute how much time was left of the timeout, 1043 * by subtracting the current time and the time 1044 * before we started the call, and subtracting 1045 * that result from the user-supplied value. 1046 */ 1047 microtime(&tv1); 1048 timersub(&tv1, &tv0, &tv1); 1049 timersub(&utv, &tv1, &utv); 1050 if (utv.tv_sec < 0) 1051 timerclear(&utv); 1052 } else 1053 timerclear(&utv); 1054 if ((error = copyout(&utv, timeout, sizeof(utv)))) 1055 return error; 1056 } 1057 1058 return 0; 1059 } 1060 1061 /* 1062 * Get the process group of a certain process. Look it up 1063 * and return the value. 1064 */ 1065 int 1066 linux_sys_getpgid(l, v, retval) 1067 struct lwp *l; 1068 void *v; 1069 register_t *retval; 1070 { 1071 struct linux_sys_getpgid_args /* { 1072 syscallarg(int) pid; 1073 } */ *uap = v; 1074 struct proc *p = l->l_proc; 1075 struct proc *targp; 1076 1077 if (SCARG(uap, pid) != 0 && SCARG(uap, pid) != p->p_pid) { 1078 if ((targp = pfind(SCARG(uap, pid))) == 0) 1079 return ESRCH; 1080 } 1081 else 1082 targp = p; 1083 1084 retval[0] = targp->p_pgid; 1085 return 0; 1086 } 1087 1088 /* 1089 * Set the 'personality' (emulation mode) for the current process. Only 1090 * accept the Linux personality here (0). This call is needed because 1091 * the Linux ELF crt0 issues it in an ugly kludge to make sure that 1092 * ELF binaries run in Linux mode, not SVR4 mode. 1093 */ 1094 int 1095 linux_sys_personality(l, v, retval) 1096 struct lwp *l; 1097 void *v; 1098 register_t *retval; 1099 { 1100 struct linux_sys_personality_args /* { 1101 syscallarg(int) per; 1102 } */ *uap = v; 1103 1104 if (SCARG(uap, per) != 0) 1105 return EINVAL; 1106 retval[0] = 0; 1107 return 0; 1108 } 1109 1110 #if defined(__i386__) || defined(__m68k__) 1111 /* 1112 * The calls are here because of type conversions. 1113 */ 1114 int 1115 linux_sys_setreuid16(l, v, retval) 1116 struct lwp *l; 1117 void *v; 1118 register_t *retval; 1119 { 1120 struct linux_sys_setreuid16_args /* { 1121 syscallarg(int) ruid; 1122 syscallarg(int) euid; 1123 } */ *uap = v; 1124 struct sys_setreuid_args bsa; 1125 1126 SCARG(&bsa, ruid) = ((linux_uid_t)SCARG(uap, ruid) == (linux_uid_t)-1) ? 1127 (uid_t)-1 : SCARG(uap, ruid); 1128 SCARG(&bsa, euid) = ((linux_uid_t)SCARG(uap, euid) == (linux_uid_t)-1) ? 1129 (uid_t)-1 : SCARG(uap, euid); 1130 1131 return sys_setreuid(l, &bsa, retval); 1132 } 1133 1134 int 1135 linux_sys_setregid16(l, v, retval) 1136 struct lwp *l; 1137 void *v; 1138 register_t *retval; 1139 { 1140 struct linux_sys_setregid16_args /* { 1141 syscallarg(int) rgid; 1142 syscallarg(int) egid; 1143 } */ *uap = v; 1144 struct sys_setregid_args bsa; 1145 1146 SCARG(&bsa, rgid) = ((linux_gid_t)SCARG(uap, rgid) == (linux_gid_t)-1) ? 1147 (uid_t)-1 : SCARG(uap, rgid); 1148 SCARG(&bsa, egid) = ((linux_gid_t)SCARG(uap, egid) == (linux_gid_t)-1) ? 1149 (uid_t)-1 : SCARG(uap, egid); 1150 1151 return sys_setregid(l, &bsa, retval); 1152 } 1153 1154 int 1155 linux_sys_setresuid16(l, v, retval) 1156 struct lwp *l; 1157 void *v; 1158 register_t *retval; 1159 { 1160 struct linux_sys_setresuid16_args /* { 1161 syscallarg(uid_t) ruid; 1162 syscallarg(uid_t) euid; 1163 syscallarg(uid_t) suid; 1164 } */ *uap = v; 1165 struct linux_sys_setresuid16_args lsa; 1166 1167 SCARG(&lsa, ruid) = ((linux_uid_t)SCARG(uap, ruid) == (linux_uid_t)-1) ? 1168 (uid_t)-1 : SCARG(uap, ruid); 1169 SCARG(&lsa, euid) = ((linux_uid_t)SCARG(uap, euid) == (linux_uid_t)-1) ? 1170 (uid_t)-1 : SCARG(uap, euid); 1171 SCARG(&lsa, suid) = ((linux_uid_t)SCARG(uap, suid) == (linux_uid_t)-1) ? 1172 (uid_t)-1 : SCARG(uap, suid); 1173 1174 return linux_sys_setresuid(l, &lsa, retval); 1175 } 1176 1177 int 1178 linux_sys_setresgid16(l, v, retval) 1179 struct lwp *l; 1180 void *v; 1181 register_t *retval; 1182 { 1183 struct linux_sys_setresgid16_args /* { 1184 syscallarg(gid_t) rgid; 1185 syscallarg(gid_t) egid; 1186 syscallarg(gid_t) sgid; 1187 } */ *uap = v; 1188 struct linux_sys_setresgid16_args lsa; 1189 1190 SCARG(&lsa, rgid) = ((linux_gid_t)SCARG(uap, rgid) == (linux_gid_t)-1) ? 1191 (gid_t)-1 : SCARG(uap, rgid); 1192 SCARG(&lsa, egid) = ((linux_gid_t)SCARG(uap, egid) == (linux_gid_t)-1) ? 1193 (gid_t)-1 : SCARG(uap, egid); 1194 SCARG(&lsa, sgid) = ((linux_gid_t)SCARG(uap, sgid) == (linux_gid_t)-1) ? 1195 (gid_t)-1 : SCARG(uap, sgid); 1196 1197 return linux_sys_setresgid(l, &lsa, retval); 1198 } 1199 1200 int 1201 linux_sys_getgroups16(l, v, retval) 1202 struct lwp *l; 1203 void *v; 1204 register_t *retval; 1205 { 1206 struct linux_sys_getgroups16_args /* { 1207 syscallarg(int) gidsetsize; 1208 syscallarg(linux_gid_t *) gidset; 1209 } */ *uap = v; 1210 struct proc *p = l->l_proc; 1211 caddr_t sg; 1212 int n, error, i; 1213 struct sys_getgroups_args bsa; 1214 gid_t *bset, *kbset; 1215 linux_gid_t *lset; 1216 struct pcred *pc = p->p_cred; 1217 1218 n = SCARG(uap, gidsetsize); 1219 if (n < 0) 1220 return EINVAL; 1221 error = 0; 1222 bset = kbset = NULL; 1223 lset = NULL; 1224 if (n > 0) { 1225 n = min(pc->pc_ucred->cr_ngroups, n); 1226 sg = stackgap_init(p, 0); 1227 bset = stackgap_alloc(p, &sg, n * sizeof (gid_t)); 1228 kbset = malloc(n * sizeof (gid_t), M_TEMP, M_WAITOK); 1229 lset = malloc(n * sizeof (linux_gid_t), M_TEMP, M_WAITOK); 1230 if (bset == NULL || kbset == NULL || lset == NULL) 1231 return ENOMEM; 1232 SCARG(&bsa, gidsetsize) = n; 1233 SCARG(&bsa, gidset) = bset; 1234 error = sys_getgroups(l, &bsa, retval); 1235 if (error != 0) 1236 goto out; 1237 error = copyin(bset, kbset, n * sizeof (gid_t)); 1238 if (error != 0) 1239 goto out; 1240 for (i = 0; i < n; i++) 1241 lset[i] = (linux_gid_t)kbset[i]; 1242 error = copyout(lset, SCARG(uap, gidset), 1243 n * sizeof (linux_gid_t)); 1244 } else 1245 *retval = pc->pc_ucred->cr_ngroups; 1246 out: 1247 if (kbset != NULL) 1248 free(kbset, M_TEMP); 1249 if (lset != NULL) 1250 free(lset, M_TEMP); 1251 return error; 1252 } 1253 1254 int 1255 linux_sys_setgroups16(l, v, retval) 1256 struct lwp *l; 1257 void *v; 1258 register_t *retval; 1259 { 1260 struct linux_sys_setgroups16_args /* { 1261 syscallarg(int) gidsetsize; 1262 syscallarg(linux_gid_t *) gidset; 1263 } */ *uap = v; 1264 struct proc *p = l->l_proc; 1265 caddr_t sg; 1266 int n; 1267 int error, i; 1268 struct sys_setgroups_args bsa; 1269 gid_t *bset, *kbset; 1270 linux_gid_t *lset; 1271 1272 n = SCARG(uap, gidsetsize); 1273 if (n < 0 || n > NGROUPS) 1274 return EINVAL; 1275 sg = stackgap_init(p, 0); 1276 bset = stackgap_alloc(p, &sg, n * sizeof (gid_t)); 1277 lset = malloc(n * sizeof (linux_gid_t), M_TEMP, M_WAITOK); 1278 kbset = malloc(n * sizeof (gid_t), M_TEMP, M_WAITOK); 1279 if (lset == NULL || bset == NULL) 1280 return ENOMEM; 1281 error = copyin(SCARG(uap, gidset), lset, n * sizeof (linux_gid_t)); 1282 if (error != 0) 1283 goto out; 1284 for (i = 0; i < n; i++) 1285 kbset[i] = (gid_t)lset[i]; 1286 error = copyout(kbset, bset, n * sizeof (gid_t)); 1287 if (error != 0) 1288 goto out; 1289 SCARG(&bsa, gidsetsize) = n; 1290 SCARG(&bsa, gidset) = bset; 1291 error = sys_setgroups(l, &bsa, retval); 1292 1293 out: 1294 if (lset != NULL) 1295 free(lset, M_TEMP); 1296 if (kbset != NULL) 1297 free(kbset, M_TEMP); 1298 1299 return error; 1300 } 1301 1302 #endif /* __i386__ || __m68k__ */ 1303 1304 /* 1305 * We have nonexistent fsuid equal to uid. 1306 * If modification is requested, refuse. 1307 */ 1308 int 1309 linux_sys_setfsuid(l, v, retval) 1310 struct lwp *l; 1311 void *v; 1312 register_t *retval; 1313 { 1314 struct linux_sys_setfsuid_args /* { 1315 syscallarg(uid_t) uid; 1316 } */ *uap = v; 1317 struct proc *p = l->l_proc; 1318 uid_t uid; 1319 1320 uid = SCARG(uap, uid); 1321 if (p->p_cred->p_ruid != uid) 1322 return sys_nosys(l, v, retval); 1323 else 1324 return (0); 1325 } 1326 1327 /* XXX XXX XXX */ 1328 #ifndef alpha 1329 int 1330 linux_sys_getfsuid(l, v, retval) 1331 struct lwp *l; 1332 void *v; 1333 register_t *retval; 1334 { 1335 return sys_getuid(l, v, retval); 1336 } 1337 #endif 1338 1339 int 1340 linux_sys_setresuid(l, v, retval) 1341 struct lwp *l; 1342 void *v; 1343 register_t *retval; 1344 { 1345 struct linux_sys_setresuid_args /* { 1346 syscallarg(uid_t) ruid; 1347 syscallarg(uid_t) euid; 1348 syscallarg(uid_t) suid; 1349 } */ *uap = v; 1350 1351 /* 1352 * Note: These checks are a little different than the NetBSD 1353 * setreuid(2) call performs. This precisely follows the 1354 * behavior of the Linux kernel. 1355 */ 1356 1357 return do_setresuid(l, SCARG(uap, ruid), SCARG(uap, euid), 1358 SCARG(uap, suid), 1359 ID_R_EQ_R | ID_R_EQ_E | ID_R_EQ_S | 1360 ID_E_EQ_R | ID_E_EQ_E | ID_E_EQ_S | 1361 ID_S_EQ_R | ID_S_EQ_E | ID_S_EQ_S ); 1362 } 1363 1364 int 1365 linux_sys_getresuid(l, v, retval) 1366 struct lwp *l; 1367 void *v; 1368 register_t *retval; 1369 { 1370 struct linux_sys_getresuid_args /* { 1371 syscallarg(uid_t *) ruid; 1372 syscallarg(uid_t *) euid; 1373 syscallarg(uid_t *) suid; 1374 } */ *uap = v; 1375 struct proc *p = l->l_proc; 1376 struct pcred *pc = p->p_cred; 1377 int error; 1378 1379 /* 1380 * Linux copies these values out to userspace like so: 1381 * 1382 * 1. Copy out ruid. 1383 * 2. If that succeeds, copy out euid. 1384 * 3. If both of those succeed, copy out suid. 1385 */ 1386 if ((error = copyout(&pc->p_ruid, SCARG(uap, ruid), 1387 sizeof(uid_t))) != 0) 1388 return (error); 1389 1390 if ((error = copyout(&pc->pc_ucred->cr_uid, SCARG(uap, euid), 1391 sizeof(uid_t))) != 0) 1392 return (error); 1393 1394 return (copyout(&pc->p_svuid, SCARG(uap, suid), sizeof(uid_t))); 1395 } 1396 1397 int 1398 linux_sys_ptrace(l, v, retval) 1399 struct lwp *l; 1400 void *v; 1401 register_t *retval; 1402 { 1403 struct linux_sys_ptrace_args /* { 1404 i386, m68k, powerpc: T=int 1405 alpha: T=long 1406 syscallarg(T) request; 1407 syscallarg(T) pid; 1408 syscallarg(T) addr; 1409 syscallarg(T) data; 1410 } */ *uap = v; 1411 const int *ptr; 1412 int request; 1413 int error; 1414 1415 ptr = linux_ptrace_request_map; 1416 request = SCARG(uap, request); 1417 while (*ptr != -1) 1418 if (*ptr++ == request) { 1419 struct sys_ptrace_args pta; 1420 1421 SCARG(&pta, req) = *ptr; 1422 SCARG(&pta, pid) = SCARG(uap, pid); 1423 SCARG(&pta, addr) = (caddr_t)SCARG(uap, addr); 1424 SCARG(&pta, data) = SCARG(uap, data); 1425 1426 /* 1427 * Linux ptrace(PTRACE_CONT, pid, 0, 0) means actually 1428 * to continue where the process left off previously. 1429 * The same thing is achieved by addr == (caddr_t) 1 1430 * on NetBSD, so rewrite 'addr' appropriately. 1431 */ 1432 if (request == LINUX_PTRACE_CONT && SCARG(uap, addr)==0) 1433 SCARG(&pta, addr) = (caddr_t) 1; 1434 1435 error = sys_ptrace(l, &pta, retval); 1436 if (error) 1437 return error; 1438 switch (request) { 1439 case LINUX_PTRACE_PEEKTEXT: 1440 case LINUX_PTRACE_PEEKDATA: 1441 error = copyout (retval, 1442 (caddr_t)SCARG(uap, data), sizeof *retval); 1443 *retval = SCARG(uap, data); 1444 break; 1445 default: 1446 break; 1447 } 1448 return error; 1449 } 1450 else 1451 ptr++; 1452 1453 return LINUX_SYS_PTRACE_ARCH(l, uap, retval); 1454 } 1455 1456 int 1457 linux_sys_reboot(struct lwp *l, void *v, register_t *retval) 1458 { 1459 struct linux_sys_reboot_args /* { 1460 syscallarg(int) magic1; 1461 syscallarg(int) magic2; 1462 syscallarg(int) cmd; 1463 syscallarg(void *) arg; 1464 } */ *uap = v; 1465 struct sys_reboot_args /* { 1466 syscallarg(int) opt; 1467 syscallarg(char *) bootstr; 1468 } */ sra; 1469 struct proc *p = l->l_proc; 1470 int error; 1471 1472 if ((error = suser(p->p_ucred, &p->p_acflag)) != 0) 1473 return(error); 1474 1475 if (SCARG(uap, magic1) != LINUX_REBOOT_MAGIC1) 1476 return(EINVAL); 1477 if (SCARG(uap, magic2) != LINUX_REBOOT_MAGIC2 && 1478 SCARG(uap, magic2) != LINUX_REBOOT_MAGIC2A && 1479 SCARG(uap, magic2) != LINUX_REBOOT_MAGIC2B) 1480 return(EINVAL); 1481 1482 switch (SCARG(uap, cmd)) { 1483 case LINUX_REBOOT_CMD_RESTART: 1484 SCARG(&sra, opt) = RB_AUTOBOOT; 1485 break; 1486 case LINUX_REBOOT_CMD_HALT: 1487 SCARG(&sra, opt) = RB_HALT; 1488 break; 1489 case LINUX_REBOOT_CMD_POWER_OFF: 1490 SCARG(&sra, opt) = RB_HALT|RB_POWERDOWN; 1491 break; 1492 case LINUX_REBOOT_CMD_RESTART2: 1493 /* Reboot with an argument. */ 1494 SCARG(&sra, opt) = RB_AUTOBOOT|RB_STRING; 1495 SCARG(&sra, bootstr) = SCARG(uap, arg); 1496 break; 1497 case LINUX_REBOOT_CMD_CAD_ON: 1498 return(EINVAL); /* We don't implement ctrl-alt-delete */ 1499 case LINUX_REBOOT_CMD_CAD_OFF: 1500 return(0); 1501 default: 1502 return(EINVAL); 1503 } 1504 1505 return(sys_reboot(l, &sra, retval)); 1506 } 1507 1508 /* 1509 * Copy of compat_12_sys_swapon(). 1510 */ 1511 int 1512 linux_sys_swapon(l, v, retval) 1513 struct lwp *l; 1514 void *v; 1515 register_t *retval; 1516 { 1517 struct sys_swapctl_args ua; 1518 struct linux_sys_swapon_args /* { 1519 syscallarg(const char *) name; 1520 } */ *uap = v; 1521 1522 SCARG(&ua, cmd) = SWAP_ON; 1523 SCARG(&ua, arg) = (void *)SCARG(uap, name); 1524 SCARG(&ua, misc) = 0; /* priority */ 1525 return (sys_swapctl(l, &ua, retval)); 1526 } 1527 1528 /* 1529 * Stop swapping to the file or block device specified by path. 1530 */ 1531 int 1532 linux_sys_swapoff(l, v, retval) 1533 struct lwp *l; 1534 void *v; 1535 register_t *retval; 1536 { 1537 struct sys_swapctl_args ua; 1538 struct linux_sys_swapoff_args /* { 1539 syscallarg(const char *) path; 1540 } */ *uap = v; 1541 1542 SCARG(&ua, cmd) = SWAP_OFF; 1543 SCARG(&ua, arg) = (void *)SCARG(uap, path); 1544 return (sys_swapctl(l, &ua, retval)); 1545 } 1546 1547 /* 1548 * Copy of compat_09_sys_setdomainname() 1549 */ 1550 /* ARGSUSED */ 1551 int 1552 linux_sys_setdomainname(l, v, retval) 1553 struct lwp *l; 1554 void *v; 1555 register_t *retval; 1556 { 1557 struct linux_sys_setdomainname_args /* { 1558 syscallarg(char *) domainname; 1559 syscallarg(int) len; 1560 } */ *uap = v; 1561 int name[2]; 1562 1563 name[0] = CTL_KERN; 1564 name[1] = KERN_DOMAINNAME; 1565 return (old_sysctl(&name[0], 2, 0, 0, SCARG(uap, domainname), 1566 SCARG(uap, len), l)); 1567 } 1568 1569 /* 1570 * sysinfo() 1571 */ 1572 /* ARGSUSED */ 1573 int 1574 linux_sys_sysinfo(l, v, retval) 1575 struct lwp *l; 1576 void *v; 1577 register_t *retval; 1578 { 1579 struct linux_sys_sysinfo_args /* { 1580 syscallarg(struct linux_sysinfo *) arg; 1581 } */ *uap = v; 1582 struct linux_sysinfo si; 1583 struct loadavg *la; 1584 1585 si.uptime = time.tv_sec - boottime.tv_sec; 1586 la = &averunnable; 1587 si.loads[0] = la->ldavg[0] * LINUX_SYSINFO_LOADS_SCALE / la->fscale; 1588 si.loads[1] = la->ldavg[1] * LINUX_SYSINFO_LOADS_SCALE / la->fscale; 1589 si.loads[2] = la->ldavg[2] * LINUX_SYSINFO_LOADS_SCALE / la->fscale; 1590 si.totalram = ctob(physmem); 1591 si.freeram = uvmexp.free * uvmexp.pagesize; 1592 si.sharedram = 0; /* XXX */ 1593 si.bufferram = uvmexp.filepages * uvmexp.pagesize; 1594 si.totalswap = uvmexp.swpages * uvmexp.pagesize; 1595 si.freeswap = (uvmexp.swpages - uvmexp.swpginuse) * uvmexp.pagesize; 1596 si.procs = nprocs; 1597 1598 /* The following are only present in newer Linux kernels. */ 1599 si.totalbig = 0; 1600 si.freebig = 0; 1601 si.mem_unit = 1; 1602 1603 return (copyout(&si, SCARG(uap, arg), sizeof si)); 1604 } 1605 1606 #define bsd_to_linux_rlimit1(l, b, f) \ 1607 (l)->f = ((b)->f == RLIM_INFINITY || \ 1608 ((b)->f & 0xffffffff00000000ULL) != 0) ? \ 1609 LINUX_RLIM_INFINITY : (int32_t)(b)->f 1610 #define bsd_to_linux_rlimit(l, b) \ 1611 bsd_to_linux_rlimit1(l, b, rlim_cur); \ 1612 bsd_to_linux_rlimit1(l, b, rlim_max) 1613 1614 #define linux_to_bsd_rlimit1(b, l, f) \ 1615 (b)->f = (l)->f == LINUX_RLIM_INFINITY ? RLIM_INFINITY : (l)->f 1616 #define linux_to_bsd_rlimit(b, l) \ 1617 linux_to_bsd_rlimit1(b, l, rlim_cur); \ 1618 linux_to_bsd_rlimit1(b, l, rlim_max) 1619 1620 static int 1621 linux_to_bsd_limit(lim) 1622 int lim; 1623 { 1624 switch (lim) { 1625 case LINUX_RLIMIT_CPU: 1626 return RLIMIT_CPU; 1627 case LINUX_RLIMIT_FSIZE: 1628 return RLIMIT_FSIZE; 1629 case LINUX_RLIMIT_DATA: 1630 return RLIMIT_DATA; 1631 case LINUX_RLIMIT_STACK: 1632 return RLIMIT_STACK; 1633 case LINUX_RLIMIT_CORE: 1634 return RLIMIT_CORE; 1635 case LINUX_RLIMIT_RSS: 1636 return RLIMIT_RSS; 1637 case LINUX_RLIMIT_NPROC: 1638 return RLIMIT_NPROC; 1639 case LINUX_RLIMIT_NOFILE: 1640 return RLIMIT_NOFILE; 1641 case LINUX_RLIMIT_MEMLOCK: 1642 return RLIMIT_MEMLOCK; 1643 case LINUX_RLIMIT_AS: 1644 case LINUX_RLIMIT_LOCKS: 1645 return -EOPNOTSUPP; 1646 default: 1647 return -EINVAL; 1648 } 1649 } 1650 1651 1652 int 1653 linux_sys_getrlimit(l, v, retval) 1654 struct lwp *l; 1655 void *v; 1656 register_t *retval; 1657 { 1658 struct linux_sys_getrlimit_args /* { 1659 syscallarg(int) which; 1660 syscallarg(struct orlimit *) rlp; 1661 } */ *uap = v; 1662 struct proc *p = l->l_proc; 1663 caddr_t sg = stackgap_init(p, 0); 1664 struct sys_getrlimit_args ap; 1665 struct rlimit rl; 1666 struct orlimit orl; 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 return copyout(&orl, SCARG(uap, rlp), sizeof(orl)); 1679 } 1680 1681 int 1682 linux_sys_setrlimit(l, v, retval) 1683 struct lwp *l; 1684 void *v; 1685 register_t *retval; 1686 { 1687 struct linux_sys_setrlimit_args /* { 1688 syscallarg(int) which; 1689 syscallarg(struct orlimit *) rlp; 1690 } */ *uap = v; 1691 struct proc *p = l->l_proc; 1692 caddr_t sg = stackgap_init(p, 0); 1693 struct sys_setrlimit_args ap; 1694 struct rlimit rl; 1695 struct orlimit orl; 1696 int error; 1697 1698 SCARG(&ap, which) = linux_to_bsd_limit(SCARG(uap, which)); 1699 SCARG(&ap, rlp) = stackgap_alloc(p, &sg, sizeof rl); 1700 if ((error = SCARG(&ap, which)) < 0) 1701 return -error; 1702 if ((error = copyin(SCARG(uap, rlp), &orl, sizeof(orl))) != 0) 1703 return error; 1704 linux_to_bsd_rlimit(&rl, &orl); 1705 /* XXX: alpha complains about this */ 1706 if ((error = copyout(&rl, (void *)SCARG(&ap, rlp), sizeof(rl))) != 0) 1707 return error; 1708 return sys_setrlimit(l, &ap, retval); 1709 } 1710 1711 #ifndef __mips__ 1712 /* XXX: this doesn't look 100% common, at least mips doesn't have it */ 1713 int 1714 linux_sys_ugetrlimit(l, v, retval) 1715 struct lwp *l; 1716 void *v; 1717 register_t *retval; 1718 { 1719 return linux_sys_getrlimit(l, v, retval); 1720 } 1721 #endif 1722 1723 /* 1724 * This gets called for unsupported syscalls. The difference to sys_nosys() 1725 * is that process does not get SIGSYS, the call just returns with ENOSYS. 1726 * This is the way Linux does it and glibc depends on this behaviour. 1727 */ 1728 int 1729 linux_sys_nosys(l, v, retval) 1730 struct lwp *l; 1731 void *v; 1732 register_t *retval; 1733 { 1734 return (ENOSYS); 1735 } 1736
Indexes created Thu Sep 25 06:09:39 GMT 2025