| 1 | /*- |
| 2 | * Copyright (c) 2000 Marcel Moolenaar |
| 3 | * All rights reserved. |
| 4 | * |
| 5 | * Redistribution and use in source and binary forms, with or without |
| 6 | * modification, are permitted provided that the following conditions |
| 7 | * are met: |
| 8 | * 1. Redistributions of source code must retain the above copyright |
| 9 | * notice, this list of conditions and the following disclaimer |
| 10 | * in this position and unchanged. |
| 11 | * 2. Redistributions in binary form must reproduce the above copyright |
| 12 | * notice, this list of conditions and the following disclaimer in the |
| 13 | * documentation and/or other materials provided with the distribution. |
| 14 | * 3. The name of the author may not be used to endorse or promote products |
| 15 | * derived from this software without specific prior written permission. |
| 16 | * |
| 17 | * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR |
| 18 | * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES |
| 19 | * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. |
| 20 | * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, |
| 21 | * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT |
| 22 | * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, |
| 23 | * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY |
| 24 | * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
| 25 | * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF |
| 26 | * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
| 27 | * |
| 28 | * $FreeBSD: src/sys/i386/linux/linux_machdep.c,v 1.6.2.4 2001/11/05 19:08:23 marcel Exp $ |
| 29 | * $DragonFly: src/sys/emulation/linux/i386/linux_machdep.c,v 1.23 2007/07/30 17:41:23 pavalos Exp $ |
| 30 | */ |
| 31 | |
| 32 | #include <sys/param.h> |
| 33 | #include <sys/systm.h> |
| 34 | #include <sys/imgact.h> |
| 35 | #include <sys/kern_syscall.h> |
| 36 | #include <sys/lock.h> |
| 37 | #include <sys/mman.h> |
| 38 | #include <sys/nlookup.h> |
| 39 | #include <sys/proc.h> |
| 40 | #include <sys/priv.h> |
| 41 | #include <sys/resource.h> |
| 42 | #include <sys/resourcevar.h> |
| 43 | #include <sys/sysproto.h> |
| 44 | #include <sys/unistd.h> |
| 45 | #include <sys/wait.h> |
| 46 | |
| 47 | #include <machine/frame.h> |
| 48 | #include <machine/psl.h> |
| 49 | #include <machine/segments.h> |
| 50 | #include <machine/sysarch.h> |
| 51 | |
| 52 | #include <vm/vm.h> |
| 53 | #include <vm/pmap.h> |
| 54 | #include <vm/vm_map.h> |
| 55 | |
| 56 | #include "linux.h" |
| 57 | #include "linux_proto.h" |
| 58 | #include "../linux_ipc.h" |
| 59 | #include "../linux_signal.h" |
| 60 | #include "../linux_util.h" |
| 61 | |
| 62 | struct l_descriptor { |
| 63 | l_uint entry_number; |
| 64 | l_ulong base_addr; |
| 65 | l_uint limit; |
| 66 | l_uint seg_32bit:1; |
| 67 | l_uint contents:2; |
| 68 | l_uint read_exec_only:1; |
| 69 | l_uint limit_in_pages:1; |
| 70 | l_uint seg_not_present:1; |
| 71 | l_uint useable:1; |
| 72 | }; |
| 73 | |
| 74 | struct l_old_select_argv { |
| 75 | l_int nfds; |
| 76 | l_fd_set *readfds; |
| 77 | l_fd_set *writefds; |
| 78 | l_fd_set *exceptfds; |
| 79 | struct l_timeval *timeout; |
| 80 | }; |
| 81 | |
| 82 | int |
| 83 | linux_to_bsd_sigaltstack(int lsa) |
| 84 | { |
| 85 | int bsa = 0; |
| 86 | |
| 87 | if (lsa & LINUX_SS_DISABLE) |
| 88 | bsa |= SS_DISABLE; |
| 89 | if (lsa & LINUX_SS_ONSTACK) |
| 90 | bsa |= SS_ONSTACK; |
| 91 | return (bsa); |
| 92 | } |
| 93 | |
| 94 | int |
| 95 | bsd_to_linux_sigaltstack(int bsa) |
| 96 | { |
| 97 | int lsa = 0; |
| 98 | |
| 99 | if (bsa & SS_DISABLE) |
| 100 | lsa |= LINUX_SS_DISABLE; |
| 101 | if (bsa & SS_ONSTACK) |
| 102 | lsa |= LINUX_SS_ONSTACK; |
| 103 | return (lsa); |
| 104 | } |
| 105 | |
| 106 | int |
| 107 | sys_linux_execve(struct linux_execve_args *args) |
| 108 | { |
| 109 | struct nlookupdata nd; |
| 110 | struct image_args exec_args; |
| 111 | char *path; |
| 112 | int error; |
| 113 | |
| 114 | error = linux_copyin_path(args->path, &path, LINUX_PATH_EXISTS); |
| 115 | if (error) |
| 116 | return (error); |
| 117 | #ifdef DEBUG |
| 118 | if (ldebug(execve)) |
| 119 | kprintf(ARGS(execve, "%s"), path); |
| 120 | #endif |
| 121 | error = nlookup_init(&nd, path, UIO_SYSSPACE, NLC_FOLLOW); |
| 122 | bzero(&exec_args, sizeof(exec_args)); |
| 123 | if (error == 0) { |
| 124 | error = exec_copyin_args(&exec_args, path, PATH_SYSSPACE, |
| 125 | args->argp, args->envp); |
| 126 | } |
| 127 | if (error == 0) |
| 128 | error = kern_execve(&nd, &exec_args); |
| 129 | nlookup_done(&nd); |
| 130 | |
| 131 | /* |
| 132 | * The syscall result is returned in registers to the new program. |
| 133 | * Linux will register %edx as an atexit function and we must be |
| 134 | * sure to set it to 0. XXX |
| 135 | */ |
| 136 | if (error == 0) |
| 137 | args->sysmsg_result64 = 0; |
| 138 | |
| 139 | exec_free_args(&exec_args); |
| 140 | linux_free_path(&path); |
| 141 | |
| 142 | if (error < 0) { |
| 143 | /* We hit a lethal error condition. Let's die now. */ |
| 144 | exit1(W_EXITCODE(0, SIGABRT)); |
| 145 | /* NOTREACHED */ |
| 146 | } |
| 147 | |
| 148 | return(error); |
| 149 | } |
| 150 | |
| 151 | struct l_ipc_kludge { |
| 152 | struct l_msgbuf *msgp; |
| 153 | l_long msgtyp; |
| 154 | }; |
| 155 | |
| 156 | int |
| 157 | sys_linux_ipc(struct linux_ipc_args *args) |
| 158 | { |
| 159 | int error = 0; |
| 160 | |
| 161 | switch (args->what & 0xFFFF) { |
| 162 | case LINUX_SEMOP: { |
| 163 | struct linux_semop_args a; |
| 164 | |
| 165 | a.semid = args->arg1; |
| 166 | a.tsops = args->ptr; |
| 167 | a.nsops = args->arg2; |
| 168 | a.sysmsg_lresult = 0; |
| 169 | error = linux_semop(&a); |
| 170 | args->sysmsg_lresult = a.sysmsg_lresult; |
| 171 | break; |
| 172 | } |
| 173 | case LINUX_SEMGET: { |
| 174 | struct linux_semget_args a; |
| 175 | |
| 176 | a.key = args->arg1; |
| 177 | a.nsems = args->arg2; |
| 178 | a.semflg = args->arg3; |
| 179 | a.sysmsg_lresult = 0; |
| 180 | error = linux_semget(&a); |
| 181 | args->sysmsg_lresult = a.sysmsg_lresult; |
| 182 | break; |
| 183 | } |
| 184 | case LINUX_SEMCTL: { |
| 185 | struct linux_semctl_args a; |
| 186 | int error; |
| 187 | |
| 188 | a.semid = args->arg1; |
| 189 | a.semnum = args->arg2; |
| 190 | a.cmd = args->arg3; |
| 191 | a.sysmsg_lresult = 0; |
| 192 | error = copyin((caddr_t)args->ptr, &a.arg, sizeof(a.arg)); |
| 193 | if (error) |
| 194 | return (error); |
| 195 | error = linux_semctl(&a); |
| 196 | args->sysmsg_lresult = a.sysmsg_lresult; |
| 197 | break; |
| 198 | } |
| 199 | case LINUX_MSGSND: { |
| 200 | struct linux_msgsnd_args a; |
| 201 | |
| 202 | a.msqid = args->arg1; |
| 203 | a.msgp = args->ptr; |
| 204 | a.msgsz = args->arg2; |
| 205 | a.msgflg = args->arg3; |
| 206 | a.sysmsg_lresult = 0; |
| 207 | error = linux_msgsnd(&a); |
| 208 | args->sysmsg_lresult = a.sysmsg_lresult; |
| 209 | break; |
| 210 | } |
| 211 | case LINUX_MSGRCV: { |
| 212 | struct linux_msgrcv_args a; |
| 213 | |
| 214 | a.msqid = args->arg1; |
| 215 | a.msgsz = args->arg2; |
| 216 | a.msgflg = args->arg3; |
| 217 | a.sysmsg_lresult = 0; |
| 218 | if ((args->what >> 16) == 0) { |
| 219 | struct l_ipc_kludge tmp; |
| 220 | int error; |
| 221 | |
| 222 | if (args->ptr == NULL) |
| 223 | return (EINVAL); |
| 224 | error = copyin((caddr_t)args->ptr, &tmp, sizeof(tmp)); |
| 225 | if (error) |
| 226 | return (error); |
| 227 | a.msgp = tmp.msgp; |
| 228 | a.msgtyp = tmp.msgtyp; |
| 229 | } else { |
| 230 | a.msgp = args->ptr; |
| 231 | a.msgtyp = args->arg5; |
| 232 | } |
| 233 | error = linux_msgrcv(&a); |
| 234 | args->sysmsg_lresult = a.sysmsg_lresult; |
| 235 | break; |
| 236 | } |
| 237 | case LINUX_MSGGET: { |
| 238 | struct linux_msgget_args a; |
| 239 | |
| 240 | a.key = args->arg1; |
| 241 | a.msgflg = args->arg2; |
| 242 | a.sysmsg_lresult = 0; |
| 243 | error = linux_msgget(&a); |
| 244 | args->sysmsg_lresult = a.sysmsg_lresult; |
| 245 | break; |
| 246 | } |
| 247 | case LINUX_MSGCTL: { |
| 248 | struct linux_msgctl_args a; |
| 249 | |
| 250 | a.msqid = args->arg1; |
| 251 | a.cmd = args->arg2; |
| 252 | a.buf = args->ptr; |
| 253 | a.sysmsg_lresult = 0; |
| 254 | error = linux_msgctl(&a); |
| 255 | args->sysmsg_lresult = a.sysmsg_lresult; |
| 256 | break; |
| 257 | } |
| 258 | case LINUX_SHMAT: { |
| 259 | struct linux_shmat_args a; |
| 260 | |
| 261 | a.shmid = args->arg1; |
| 262 | a.shmaddr = args->ptr; |
| 263 | a.shmflg = args->arg2; |
| 264 | a.raddr = (l_ulong *)args->arg3; |
| 265 | a.sysmsg_lresult = 0; |
| 266 | error = linux_shmat(&a); |
| 267 | args->sysmsg_lresult = a.sysmsg_lresult; |
| 268 | break; |
| 269 | } |
| 270 | case LINUX_SHMDT: { |
| 271 | struct linux_shmdt_args a; |
| 272 | |
| 273 | a.shmaddr = args->ptr; |
| 274 | a.sysmsg_lresult = 0; |
| 275 | error = linux_shmdt(&a); |
| 276 | args->sysmsg_lresult = a.sysmsg_lresult; |
| 277 | break; |
| 278 | } |
| 279 | case LINUX_SHMGET: { |
| 280 | struct linux_shmget_args a; |
| 281 | |
| 282 | a.key = args->arg1; |
| 283 | a.size = args->arg2; |
| 284 | a.shmflg = args->arg3; |
| 285 | a.sysmsg_lresult = 0; |
| 286 | error = linux_shmget(&a); |
| 287 | args->sysmsg_lresult = a.sysmsg_lresult; |
| 288 | break; |
| 289 | } |
| 290 | case LINUX_SHMCTL: { |
| 291 | struct linux_shmctl_args a; |
| 292 | |
| 293 | a.shmid = args->arg1; |
| 294 | a.cmd = args->arg2; |
| 295 | a.buf = args->ptr; |
| 296 | a.sysmsg_lresult = 0; |
| 297 | error = linux_shmctl(&a); |
| 298 | args->sysmsg_lresult = a.sysmsg_lresult; |
| 299 | break; |
| 300 | } |
| 301 | default: |
| 302 | error = EINVAL; |
| 303 | break; |
| 304 | } |
| 305 | return(error); |
| 306 | } |
| 307 | |
| 308 | int |
| 309 | sys_linux_old_select(struct linux_old_select_args *args) |
| 310 | { |
| 311 | struct l_old_select_argv linux_args; |
| 312 | struct linux_select_args newsel; |
| 313 | int error; |
| 314 | |
| 315 | #ifdef DEBUG |
| 316 | if (ldebug(old_select)) |
| 317 | kprintf(ARGS(old_select, "%p"), args->ptr); |
| 318 | #endif |
| 319 | |
| 320 | error = copyin((caddr_t)args->ptr, &linux_args, sizeof(linux_args)); |
| 321 | if (error) |
| 322 | return (error); |
| 323 | |
| 324 | newsel.sysmsg_iresult = 0; |
| 325 | newsel.nfds = linux_args.nfds; |
| 326 | newsel.readfds = linux_args.readfds; |
| 327 | newsel.writefds = linux_args.writefds; |
| 328 | newsel.exceptfds = linux_args.exceptfds; |
| 329 | newsel.timeout = linux_args.timeout; |
| 330 | error = sys_linux_select(&newsel); |
| 331 | args->sysmsg_iresult = newsel.sysmsg_iresult; |
| 332 | return(error); |
| 333 | } |
| 334 | |
| 335 | int |
| 336 | sys_linux_fork(struct linux_fork_args *args) |
| 337 | { |
| 338 | int error; |
| 339 | |
| 340 | #ifdef DEBUG |
| 341 | if (ldebug(fork)) |
| 342 | kprintf(ARGS(fork, "")); |
| 343 | #endif |
| 344 | |
| 345 | if ((error = sys_fork((struct fork_args *)args)) != 0) |
| 346 | return (error); |
| 347 | |
| 348 | if (args->sysmsg_iresult == 1) |
| 349 | args->sysmsg_iresult = 0; |
| 350 | return (0); |
| 351 | } |
| 352 | |
| 353 | int |
| 354 | sys_linux_exit_group(struct linux_exit_group_args *args) |
| 355 | { |
| 356 | struct exit_args newargs; |
| 357 | int error; |
| 358 | |
| 359 | newargs.sysmsg_iresult = 0; |
| 360 | newargs.rval = args->rval; |
| 361 | error = sys_exit(&newargs); |
| 362 | args->sysmsg_iresult = newargs.sysmsg_iresult; |
| 363 | return (error); |
| 364 | } |
| 365 | |
| 366 | int |
| 367 | sys_linux_vfork(struct linux_vfork_args *args) |
| 368 | { |
| 369 | int error; |
| 370 | |
| 371 | #ifdef DEBUG |
| 372 | if (ldebug(vfork)) |
| 373 | kprintf(ARGS(vfork, "")); |
| 374 | #endif |
| 375 | |
| 376 | if ((error = sys_vfork((struct vfork_args *)args)) != 0) |
| 377 | return (error); |
| 378 | /* Are we the child? */ |
| 379 | if (args->sysmsg_iresult == 1) |
| 380 | args->sysmsg_iresult = 0; |
| 381 | return (0); |
| 382 | } |
| 383 | |
| 384 | #define CLONE_VM 0x100 |
| 385 | #define CLONE_FS 0x200 |
| 386 | #define CLONE_FILES 0x400 |
| 387 | #define CLONE_SIGHAND 0x800 |
| 388 | #define CLONE_PID 0x1000 |
| 389 | |
| 390 | int |
| 391 | sys_linux_clone(struct linux_clone_args *args) |
| 392 | { |
| 393 | int error, ff = RFPROC; |
| 394 | struct proc *p2; |
| 395 | int exit_signal; |
| 396 | vm_offset_t start; |
| 397 | struct rfork_args rf_args; |
| 398 | |
| 399 | #ifdef DEBUG |
| 400 | if (ldebug(clone)) { |
| 401 | kprintf(ARGS(clone, "flags %x, stack %x"), |
| 402 | (unsigned int)args->flags, (unsigned int)args->stack); |
| 403 | if (args->flags & CLONE_PID) |
| 404 | kprintf(LMSG("CLONE_PID not yet supported")); |
| 405 | } |
| 406 | #endif |
| 407 | |
| 408 | if (!args->stack) |
| 409 | return (EINVAL); |
| 410 | |
| 411 | exit_signal = args->flags & 0x000000ff; |
| 412 | if (exit_signal >= LINUX_NSIG) |
| 413 | return (EINVAL); |
| 414 | |
| 415 | if (exit_signal <= LINUX_SIGTBLSZ) |
| 416 | exit_signal = linux_to_bsd_signal[_SIG_IDX(exit_signal)]; |
| 417 | |
| 418 | /* RFTHREAD probably not necessary here, but it shouldn't hurt */ |
| 419 | ff |= RFTHREAD; |
| 420 | |
| 421 | if (args->flags & CLONE_VM) |
| 422 | ff |= RFMEM; |
| 423 | if (args->flags & CLONE_SIGHAND) |
| 424 | ff |= RFSIGSHARE; |
| 425 | if (!(args->flags & CLONE_FILES)) |
| 426 | ff |= RFFDG; |
| 427 | |
| 428 | error = 0; |
| 429 | start = 0; |
| 430 | |
| 431 | rf_args.flags = ff; |
| 432 | rf_args.sysmsg_iresult = 0; |
| 433 | if ((error = sys_rfork(&rf_args)) != 0) |
| 434 | return (error); |
| 435 | args->sysmsg_iresult = rf_args.sysmsg_iresult; |
| 436 | |
| 437 | p2 = pfind(rf_args.sysmsg_iresult); |
| 438 | if (p2 == NULL) |
| 439 | return (ESRCH); |
| 440 | |
| 441 | p2->p_sigparent = exit_signal; |
| 442 | ONLY_LWP_IN_PROC(p2)->lwp_md.md_regs->tf_esp = |
| 443 | (unsigned int)args->stack; |
| 444 | |
| 445 | #ifdef DEBUG |
| 446 | if (ldebug(clone)) |
| 447 | kprintf(LMSG("clone: successful rfork to %ld"), |
| 448 | (long)p2->p_pid); |
| 449 | #endif |
| 450 | |
| 451 | return (0); |
| 452 | } |
| 453 | |
| 454 | /* XXX move */ |
| 455 | struct l_mmap_argv { |
| 456 | l_caddr_t addr; |
| 457 | l_int len; |
| 458 | l_int prot; |
| 459 | l_int flags; |
| 460 | l_int fd; |
| 461 | l_int pos; |
| 462 | }; |
| 463 | |
| 464 | #define STACK_SIZE (2 * 1024 * 1024) |
| 465 | #define GUARD_SIZE (4 * PAGE_SIZE) |
| 466 | |
| 467 | static int |
| 468 | linux_mmap_common(caddr_t linux_addr, size_t linux_len, int linux_prot, |
| 469 | int linux_flags, int linux_fd, off_t pos, void **res) |
| 470 | { |
| 471 | struct thread *td = curthread; |
| 472 | struct proc *p = td->td_proc; |
| 473 | caddr_t addr; |
| 474 | void *new; |
| 475 | int error, flags, len, prot, fd; |
| 476 | |
| 477 | flags = 0; |
| 478 | if (linux_flags & LINUX_MAP_SHARED) |
| 479 | flags |= MAP_SHARED; |
| 480 | if (linux_flags & LINUX_MAP_PRIVATE) |
| 481 | flags |= MAP_PRIVATE; |
| 482 | if (linux_flags & LINUX_MAP_FIXED) |
| 483 | flags |= MAP_FIXED; |
| 484 | if (linux_flags & LINUX_MAP_ANON) { |
| 485 | flags |= MAP_ANON; |
| 486 | } else { |
| 487 | flags |= MAP_NOSYNC; |
| 488 | } |
| 489 | if (linux_flags & LINUX_MAP_GROWSDOWN) { |
| 490 | flags |= MAP_STACK; |
| 491 | /* The linux MAP_GROWSDOWN option does not limit auto |
| 492 | * growth of the region. Linux mmap with this option |
| 493 | * takes as addr the inital BOS, and as len, the initial |
| 494 | * region size. It can then grow down from addr without |
| 495 | * limit. However, linux threads has an implicit internal |
| 496 | * limit to stack size of STACK_SIZE. Its just not |
| 497 | * enforced explicitly in linux. But, here we impose |
| 498 | * a limit of (STACK_SIZE - GUARD_SIZE) on the stack |
| 499 | * region, since we can do this with our mmap. |
| 500 | * |
| 501 | * Our mmap with MAP_STACK takes addr as the maximum |
| 502 | * downsize limit on BOS, and as len the max size of |
| 503 | * the region. It them maps the top SGROWSIZ bytes, |
| 504 | * and autgrows the region down, up to the limit |
| 505 | * in addr. |
| 506 | * |
| 507 | * If we don't use the MAP_STACK option, the effect |
| 508 | * of this code is to allocate a stack region of a |
| 509 | * fixed size of (STACK_SIZE - GUARD_SIZE). |
| 510 | */ |
| 511 | |
| 512 | /* This gives us TOS */ |
| 513 | addr = linux_addr + linux_len; |
| 514 | |
| 515 | if (addr > p->p_vmspace->vm_maxsaddr) { |
| 516 | /* Some linux apps will attempt to mmap |
| 517 | * thread stacks near the top of their |
| 518 | * address space. If their TOS is greater |
| 519 | * than vm_maxsaddr, vm_map_growstack() |
| 520 | * will confuse the thread stack with the |
| 521 | * process stack and deliver a SEGV if they |
| 522 | * attempt to grow the thread stack past their |
| 523 | * current stacksize rlimit. To avoid this, |
| 524 | * adjust vm_maxsaddr upwards to reflect |
| 525 | * the current stacksize rlimit rather |
| 526 | * than the maximum possible stacksize. |
| 527 | * It would be better to adjust the |
| 528 | * mmap'ed region, but some apps do not check |
| 529 | * mmap's return value. |
| 530 | */ |
| 531 | p->p_vmspace->vm_maxsaddr = (char *)USRSTACK - |
| 532 | p->p_rlimit[RLIMIT_STACK].rlim_cur; |
| 533 | } |
| 534 | |
| 535 | /* This gives us our maximum stack size */ |
| 536 | if (linux_len > STACK_SIZE - GUARD_SIZE) { |
| 537 | len = linux_len; |
| 538 | } else { |
| 539 | len = STACK_SIZE - GUARD_SIZE; |
| 540 | } |
| 541 | /* This gives us a new BOS. If we're using VM_STACK, then |
| 542 | * mmap will just map the top SGROWSIZ bytes, and let |
| 543 | * the stack grow down to the limit at BOS. If we're |
| 544 | * not using VM_STACK we map the full stack, since we |
| 545 | * don't have a way to autogrow it. |
| 546 | */ |
| 547 | addr -= len; |
| 548 | } else { |
| 549 | addr = linux_addr; |
| 550 | len = linux_len; |
| 551 | } |
| 552 | |
| 553 | prot = linux_prot | PROT_READ; |
| 554 | if (linux_flags & LINUX_MAP_ANON) { |
| 555 | fd = -1; |
| 556 | } else { |
| 557 | fd = linux_fd; |
| 558 | } |
| 559 | |
| 560 | #ifdef DEBUG |
| 561 | if (ldebug(mmap) || ldebug(mmap2)) |
| 562 | kprintf("-> (%p, %d, %d, 0x%08x, %d, %lld)\n", |
| 563 | addr, len, prot, flags, fd, pos); |
| 564 | #endif |
| 565 | error = kern_mmap(curproc->p_vmspace, addr, len, |
| 566 | prot, flags, fd, pos, &new); |
| 567 | |
| 568 | if (error == 0) |
| 569 | *res = new; |
| 570 | return (error); |
| 571 | } |
| 572 | |
| 573 | int |
| 574 | sys_linux_mmap(struct linux_mmap_args *args) |
| 575 | { |
| 576 | struct l_mmap_argv linux_args; |
| 577 | int error; |
| 578 | |
| 579 | error = copyin((caddr_t)args->ptr, &linux_args, sizeof(linux_args)); |
| 580 | if (error) |
| 581 | return (error); |
| 582 | |
| 583 | #ifdef DEBUG |
| 584 | if (ldebug(mmap)) |
| 585 | kprintf(ARGS(mmap, "%p, %d, %d, 0x%08x, %d, %d"), |
| 586 | (void *)linux_args.addr, linux_args.len, linux_args.prot, |
| 587 | linux_args.flags, linux_args.fd, linux_args.pos); |
| 588 | #endif |
| 589 | error = linux_mmap_common(linux_args.addr, linux_args.len, |
| 590 | linux_args.prot, linux_args.flags, linux_args.fd, |
| 591 | linux_args.pos, &args->sysmsg_resultp); |
| 592 | #ifdef DEBUG |
| 593 | if (ldebug(mmap)) |
| 594 | kprintf("-> %p\n", args->sysmsg_resultp); |
| 595 | #endif |
| 596 | return(error); |
| 597 | } |
| 598 | |
| 599 | int |
| 600 | sys_linux_mmap2(struct linux_mmap2_args *args) |
| 601 | { |
| 602 | int error; |
| 603 | |
| 604 | #ifdef DEBUG |
| 605 | if (ldebug(mmap2)) |
| 606 | kprintf(ARGS(mmap2, "%p, %d, %d, 0x%08x, %d, %d"), |
| 607 | (void *)args->addr, args->len, args->prot, args->flags, |
| 608 | args->fd, args->pgoff); |
| 609 | #endif |
| 610 | error = linux_mmap_common((void *)args->addr, args->len, args->prot, |
| 611 | args->flags, args->fd, args->pgoff * PAGE_SIZE, |
| 612 | &args->sysmsg_resultp); |
| 613 | #ifdef DEBUG |
| 614 | if (ldebug(mmap2)) |
| 615 | kprintf("-> %p\n", args->sysmsg_resultp); |
| 616 | #endif |
| 617 | return (error); |
| 618 | } |
| 619 | |
| 620 | int |
| 621 | sys_linux_pipe(struct linux_pipe_args *args) |
| 622 | { |
| 623 | int error; |
| 624 | int reg_edx; |
| 625 | struct pipe_args bsd_args; |
| 626 | |
| 627 | #ifdef DEBUG |
| 628 | if (ldebug(pipe)) |
| 629 | kprintf(ARGS(pipe, "*")); |
| 630 | #endif |
| 631 | |
| 632 | reg_edx = args->sysmsg_fds[1]; |
| 633 | error = sys_pipe(&bsd_args); |
| 634 | if (error) { |
| 635 | args->sysmsg_fds[1] = reg_edx; |
| 636 | return (error); |
| 637 | } |
| 638 | |
| 639 | error = copyout(bsd_args.sysmsg_fds, args->pipefds, 2*sizeof(int)); |
| 640 | if (error) { |
| 641 | args->sysmsg_fds[1] = reg_edx; |
| 642 | return (error); |
| 643 | } |
| 644 | |
| 645 | args->sysmsg_fds[1] = reg_edx; |
| 646 | args->sysmsg_fds[0] = 0; |
| 647 | return (0); |
| 648 | } |
| 649 | |
| 650 | int |
| 651 | sys_linux_ioperm(struct linux_ioperm_args *args) |
| 652 | { |
| 653 | struct sysarch_args sa; |
| 654 | struct i386_ioperm_args *iia; |
| 655 | caddr_t sg; |
| 656 | int error; |
| 657 | |
| 658 | sg = stackgap_init(); |
| 659 | iia = stackgap_alloc(&sg, sizeof(struct i386_ioperm_args)); |
| 660 | iia->start = args->start; |
| 661 | iia->length = args->length; |
| 662 | iia->enable = args->enable; |
| 663 | sa.sysmsg_resultp = NULL; |
| 664 | sa.op = I386_SET_IOPERM; |
| 665 | sa.parms = (char *)iia; |
| 666 | error = sys_sysarch(&sa); |
| 667 | args->sysmsg_resultp = sa.sysmsg_resultp; |
| 668 | return(error); |
| 669 | } |
| 670 | |
| 671 | int |
| 672 | sys_linux_iopl(struct linux_iopl_args *args) |
| 673 | { |
| 674 | struct thread *td = curthread; |
| 675 | struct lwp *lp = td->td_lwp; |
| 676 | int error; |
| 677 | |
| 678 | KKASSERT(lp); |
| 679 | |
| 680 | if (args->level < 0 || args->level > 3) |
| 681 | return (EINVAL); |
| 682 | if ((error = priv_check(td, PRIV_ROOT)) != 0) |
| 683 | return (error); |
| 684 | if (securelevel > 0) |
| 685 | return (EPERM); |
| 686 | lp->lwp_md.md_regs->tf_eflags = |
| 687 | (lp->lwp_md.md_regs->tf_eflags & ~PSL_IOPL) | |
| 688 | (args->level * (PSL_IOPL / 3)); |
| 689 | return (0); |
| 690 | } |
| 691 | |
| 692 | int |
| 693 | sys_linux_modify_ldt(struct linux_modify_ldt_args *uap) |
| 694 | { |
| 695 | int error; |
| 696 | caddr_t sg; |
| 697 | struct sysarch_args args; |
| 698 | struct i386_ldt_args *ldt; |
| 699 | struct l_descriptor ld; |
| 700 | union descriptor *desc; |
| 701 | |
| 702 | sg = stackgap_init(); |
| 703 | |
| 704 | if (uap->ptr == NULL) |
| 705 | return (EINVAL); |
| 706 | |
| 707 | switch (uap->func) { |
| 708 | case 0x00: /* read_ldt */ |
| 709 | ldt = stackgap_alloc(&sg, sizeof(*ldt)); |
| 710 | ldt->start = 0; |
| 711 | ldt->descs = uap->ptr; |
| 712 | ldt->num = uap->bytecount / sizeof(union descriptor); |
| 713 | args.op = I386_GET_LDT; |
| 714 | args.parms = (char*)ldt; |
| 715 | args.sysmsg_iresult = 0; |
| 716 | error = sys_sysarch(&args); |
| 717 | uap->sysmsg_iresult = args.sysmsg_iresult * |
| 718 | sizeof(union descriptor); |
| 719 | break; |
| 720 | case 0x01: /* write_ldt */ |
| 721 | case 0x11: /* write_ldt */ |
| 722 | if (uap->bytecount != sizeof(ld)) |
| 723 | return (EINVAL); |
| 724 | |
| 725 | error = copyin(uap->ptr, &ld, sizeof(ld)); |
| 726 | if (error) |
| 727 | return (error); |
| 728 | |
| 729 | ldt = stackgap_alloc(&sg, sizeof(*ldt)); |
| 730 | desc = stackgap_alloc(&sg, sizeof(*desc)); |
| 731 | ldt->start = ld.entry_number; |
| 732 | ldt->descs = desc; |
| 733 | ldt->num = 1; |
| 734 | desc->sd.sd_lolimit = (ld.limit & 0x0000ffff); |
| 735 | desc->sd.sd_hilimit = (ld.limit & 0x000f0000) >> 16; |
| 736 | desc->sd.sd_lobase = (ld.base_addr & 0x00ffffff); |
| 737 | desc->sd.sd_hibase = (ld.base_addr & 0xff000000) >> 24; |
| 738 | desc->sd.sd_type = SDT_MEMRO | ((ld.read_exec_only ^ 1) << 1) | |
| 739 | (ld.contents << 2); |
| 740 | desc->sd.sd_dpl = 3; |
| 741 | desc->sd.sd_p = (ld.seg_not_present ^ 1); |
| 742 | desc->sd.sd_xx = 0; |
| 743 | desc->sd.sd_def32 = ld.seg_32bit; |
| 744 | desc->sd.sd_gran = ld.limit_in_pages; |
| 745 | args.op = I386_SET_LDT; |
| 746 | args.parms = (char*)ldt; |
| 747 | args.sysmsg_iresult = 0; |
| 748 | error = sys_sysarch(&args); |
| 749 | uap->sysmsg_iresult = args.sysmsg_iresult; |
| 750 | break; |
| 751 | default: |
| 752 | error = EINVAL; |
| 753 | break; |
| 754 | } |
| 755 | |
| 756 | return (error); |
| 757 | } |
| 758 | |
| 759 | int |
| 760 | sys_linux_sigaction(struct linux_sigaction_args *args) |
| 761 | { |
| 762 | l_osigaction_t osa; |
| 763 | l_sigaction_t linux_act, linux_oact; |
| 764 | struct sigaction act, oact; |
| 765 | int error; |
| 766 | |
| 767 | #ifdef DEBUG |
| 768 | if (ldebug(sigaction)) |
| 769 | kprintf(ARGS(sigaction, "%d, %p, %p"), |
| 770 | args->sig, (void *)args->nsa, (void *)args->osa); |
| 771 | #endif |
| 772 | |
| 773 | if (args->nsa) { |
| 774 | error = copyin(args->nsa, &osa, sizeof(l_osigaction_t)); |
| 775 | if (error) |
| 776 | return (error); |
| 777 | linux_act.lsa_handler = osa.lsa_handler; |
| 778 | linux_act.lsa_flags = osa.lsa_flags; |
| 779 | linux_act.lsa_restorer = osa.lsa_restorer; |
| 780 | LINUX_SIGEMPTYSET(linux_act.lsa_mask); |
| 781 | linux_act.lsa_mask.__bits[0] = osa.lsa_mask; |
| 782 | linux_to_bsd_sigaction(&linux_act, &act); |
| 783 | } |
| 784 | |
| 785 | error = kern_sigaction(args->sig, args->nsa ? &act : NULL, |
| 786 | args->osa ? &oact : NULL); |
| 787 | |
| 788 | if (args->osa != NULL && !error) { |
| 789 | bsd_to_linux_sigaction(&oact, &linux_oact); |
| 790 | osa.lsa_handler = linux_oact.lsa_handler; |
| 791 | osa.lsa_flags = linux_oact.lsa_flags; |
| 792 | osa.lsa_restorer = linux_oact.lsa_restorer; |
| 793 | osa.lsa_mask = linux_oact.lsa_mask.__bits[0]; |
| 794 | error = copyout(&osa, args->osa, sizeof(l_osigaction_t)); |
| 795 | } |
| 796 | return (error); |
| 797 | } |
| 798 | |
| 799 | /* |
| 800 | * Linux has two extra args, restart and oldmask. We dont use these, |
| 801 | * but it seems that "restart" is actually a context pointer that |
| 802 | * enables the signal to happen with a different register set. |
| 803 | */ |
| 804 | int |
| 805 | sys_linux_sigsuspend(struct linux_sigsuspend_args *args) |
| 806 | { |
| 807 | l_sigset_t linux_mask; |
| 808 | sigset_t mask; |
| 809 | int error; |
| 810 | |
| 811 | #ifdef DEBUG |
| 812 | if (ldebug(sigsuspend)) |
| 813 | kprintf(ARGS(sigsuspend, "%08lx"), (unsigned long)args->mask); |
| 814 | #endif |
| 815 | |
| 816 | LINUX_SIGEMPTYSET(mask); |
| 817 | mask.__bits[0] = args->mask; |
| 818 | linux_to_bsd_sigset(&linux_mask, &mask); |
| 819 | |
| 820 | error = kern_sigsuspend(&mask); |
| 821 | |
| 822 | return(error); |
| 823 | } |
| 824 | |
| 825 | int |
| 826 | sys_linux_rt_sigsuspend(struct linux_rt_sigsuspend_args *uap) |
| 827 | { |
| 828 | l_sigset_t linux_mask; |
| 829 | sigset_t mask; |
| 830 | int error; |
| 831 | |
| 832 | #ifdef DEBUG |
| 833 | if (ldebug(rt_sigsuspend)) |
| 834 | kprintf(ARGS(rt_sigsuspend, "%p, %d"), |
| 835 | (void *)uap->newset, uap->sigsetsize); |
| 836 | #endif |
| 837 | |
| 838 | if (uap->sigsetsize != sizeof(l_sigset_t)) |
| 839 | return (EINVAL); |
| 840 | |
| 841 | error = copyin(uap->newset, &linux_mask, sizeof(l_sigset_t)); |
| 842 | if (error) |
| 843 | return (error); |
| 844 | |
| 845 | linux_to_bsd_sigset(&linux_mask, &mask); |
| 846 | |
| 847 | error = kern_sigsuspend(&mask); |
| 848 | |
| 849 | return(error); |
| 850 | } |
| 851 | |
| 852 | int |
| 853 | sys_linux_pause(struct linux_pause_args *args) |
| 854 | { |
| 855 | struct thread *td = curthread; |
| 856 | struct lwp *lp = td->td_lwp; |
| 857 | sigset_t mask; |
| 858 | int error; |
| 859 | |
| 860 | #ifdef DEBUG |
| 861 | if (ldebug(pause)) |
| 862 | kprintf(ARGS(pause, "")); |
| 863 | #endif |
| 864 | |
| 865 | mask = lp->lwp_sigmask; |
| 866 | |
| 867 | error = kern_sigsuspend(&mask); |
| 868 | |
| 869 | return(error); |
| 870 | } |
| 871 | |
| 872 | int |
| 873 | sys_linux_sigaltstack(struct linux_sigaltstack_args *uap) |
| 874 | { |
| 875 | stack_t ss, oss; |
| 876 | l_stack_t linux_ss; |
| 877 | int error; |
| 878 | |
| 879 | #ifdef DEBUG |
| 880 | if (ldebug(sigaltstack)) |
| 881 | kprintf(ARGS(sigaltstack, "%p, %p"), uap->uss, uap->uoss); |
| 882 | #endif |
| 883 | |
| 884 | if (uap->uss) { |
| 885 | error = copyin(uap->uss, &linux_ss, sizeof(l_stack_t)); |
| 886 | if (error) |
| 887 | return (error); |
| 888 | |
| 889 | ss.ss_sp = linux_ss.ss_sp; |
| 890 | ss.ss_size = linux_ss.ss_size; |
| 891 | ss.ss_flags = linux_to_bsd_sigaltstack(linux_ss.ss_flags); |
| 892 | } |
| 893 | |
| 894 | error = kern_sigaltstack(uap->uss ? &ss : NULL, |
| 895 | uap->uoss ? &oss : NULL); |
| 896 | |
| 897 | if (error == 0 && uap->uoss) { |
| 898 | linux_ss.ss_sp = oss.ss_sp; |
| 899 | linux_ss.ss_size = oss.ss_size; |
| 900 | linux_ss.ss_flags = bsd_to_linux_sigaltstack(oss.ss_flags); |
| 901 | error = copyout(&linux_ss, uap->uoss, sizeof(l_stack_t)); |
| 902 | } |
| 903 | |
| 904 | return (error); |
| 905 | } |