| 1 | /*- |
| 2 | * Copyright (c) 1988 University of Utah. |
| 3 | * Copyright (c) 1982, 1986, 1990 The Regents of the University of California. |
| 4 | * All rights reserved. |
| 5 | * |
| 6 | * This code is derived from software contributed to Berkeley by |
| 7 | * the Systems Programming Group of the University of Utah Computer |
| 8 | * Science Department, and code derived from software contributed to |
| 9 | * Berkeley by William Jolitz. |
| 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. Neither the name of the University nor the names of its contributors |
| 20 | * may be used to endorse or promote products derived from this software |
| 21 | * without specific prior written permission. |
| 22 | * |
| 23 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND |
| 24 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 25 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
| 26 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE |
| 27 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
| 28 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
| 29 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
| 30 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
| 31 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
| 32 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
| 33 | * SUCH DAMAGE. |
| 34 | * |
| 35 | * from: Utah $Hdr: mem.c 1.13 89/10/08$ |
| 36 | * from: @(#)mem.c 7.2 (Berkeley) 5/9/91 |
| 37 | * $FreeBSD: src/sys/i386/i386/mem.c,v 1.79.2.9 2003/01/04 22:58:01 njl Exp $ |
| 38 | */ |
| 39 | |
| 40 | /* |
| 41 | * Memory special file |
| 42 | */ |
| 43 | |
| 44 | #include <sys/param.h> |
| 45 | #include <sys/systm.h> |
| 46 | #include <sys/buf.h> |
| 47 | #include <sys/conf.h> |
| 48 | #include <sys/fcntl.h> |
| 49 | #include <sys/filio.h> |
| 50 | #include <sys/kernel.h> |
| 51 | #include <sys/malloc.h> |
| 52 | #include <sys/memrange.h> |
| 53 | #include <sys/proc.h> |
| 54 | #include <sys/priv.h> |
| 55 | #include <sys/random.h> |
| 56 | #include <sys/signalvar.h> |
| 57 | #include <sys/uio.h> |
| 58 | #include <sys/vnode.h> |
| 59 | #include <sys/sysctl.h> |
| 60 | |
| 61 | #include <sys/signal2.h> |
| 62 | #include <sys/mplock2.h> |
| 63 | |
| 64 | #include <vm/vm.h> |
| 65 | #include <vm/pmap.h> |
| 66 | #include <vm/vm_extern.h> |
| 67 | |
| 68 | |
| 69 | static d_open_t mmopen; |
| 70 | static d_close_t mmclose; |
| 71 | static d_read_t mmread; |
| 72 | static d_write_t mmwrite; |
| 73 | static d_ioctl_t mmioctl; |
| 74 | #if 0 |
| 75 | static d_mmap_t memmmap; |
| 76 | #endif |
| 77 | static d_kqfilter_t mmkqfilter; |
| 78 | static int memuksmap(cdev_t dev, vm_page_t fake); |
| 79 | |
| 80 | #define CDEV_MAJOR 2 |
| 81 | static struct dev_ops mem_ops = { |
| 82 | { "mem", 0, D_MPSAFE }, |
| 83 | .d_open = mmopen, |
| 84 | .d_close = mmclose, |
| 85 | .d_read = mmread, |
| 86 | .d_write = mmwrite, |
| 87 | .d_ioctl = mmioctl, |
| 88 | .d_kqfilter = mmkqfilter, |
| 89 | #if 0 |
| 90 | .d_mmap = memmmap, |
| 91 | #endif |
| 92 | .d_uksmap = memuksmap |
| 93 | }; |
| 94 | |
| 95 | static int rand_bolt; |
| 96 | static caddr_t zbuf; |
| 97 | static cdev_t zerodev = NULL; |
| 98 | |
| 99 | MALLOC_DEFINE(M_MEMDESC, "memdesc", "memory range descriptors"); |
| 100 | static int mem_ioctl (cdev_t, u_long, caddr_t, int, struct ucred *); |
| 101 | static int random_ioctl (cdev_t, u_long, caddr_t, int, struct ucred *); |
| 102 | |
| 103 | struct mem_range_softc mem_range_softc; |
| 104 | |
| 105 | static int seedenable; |
| 106 | SYSCTL_INT(_kern, OID_AUTO, seedenable, CTLFLAG_RW, &seedenable, 0, ""); |
| 107 | |
| 108 | static int |
| 109 | mmopen(struct dev_open_args *ap) |
| 110 | { |
| 111 | cdev_t dev = ap->a_head.a_dev; |
| 112 | int error; |
| 113 | |
| 114 | switch (minor(dev)) { |
| 115 | case 0: |
| 116 | case 1: |
| 117 | /* |
| 118 | * /dev/mem and /dev/kmem |
| 119 | */ |
| 120 | if (ap->a_oflags & FWRITE) { |
| 121 | if (securelevel > 0 || kernel_mem_readonly) |
| 122 | return (EPERM); |
| 123 | } |
| 124 | error = 0; |
| 125 | break; |
| 126 | case 6: |
| 127 | /* |
| 128 | * /dev/kpmap can only be opened for reading. |
| 129 | */ |
| 130 | if (ap->a_oflags & FWRITE) |
| 131 | return (EPERM); |
| 132 | error = 0; |
| 133 | break; |
| 134 | case 14: |
| 135 | error = priv_check_cred(ap->a_cred, PRIV_ROOT, 0); |
| 136 | if (error != 0) |
| 137 | break; |
| 138 | if (securelevel > 0 || kernel_mem_readonly) { |
| 139 | error = EPERM; |
| 140 | break; |
| 141 | } |
| 142 | error = cpu_set_iopl(); |
| 143 | break; |
| 144 | default: |
| 145 | error = 0; |
| 146 | break; |
| 147 | } |
| 148 | return (error); |
| 149 | } |
| 150 | |
| 151 | static int |
| 152 | mmclose(struct dev_close_args *ap) |
| 153 | { |
| 154 | cdev_t dev = ap->a_head.a_dev; |
| 155 | int error; |
| 156 | |
| 157 | switch (minor(dev)) { |
| 158 | case 14: |
| 159 | error = cpu_clr_iopl(); |
| 160 | break; |
| 161 | default: |
| 162 | error = 0; |
| 163 | break; |
| 164 | } |
| 165 | return (error); |
| 166 | } |
| 167 | |
| 168 | |
| 169 | static int |
| 170 | mmrw(cdev_t dev, struct uio *uio, int flags) |
| 171 | { |
| 172 | int o; |
| 173 | u_int c; |
| 174 | u_int poolsize; |
| 175 | u_long v; |
| 176 | struct iovec *iov; |
| 177 | int error = 0; |
| 178 | caddr_t buf = NULL; |
| 179 | |
| 180 | while (uio->uio_resid > 0 && error == 0) { |
| 181 | iov = uio->uio_iov; |
| 182 | if (iov->iov_len == 0) { |
| 183 | uio->uio_iov++; |
| 184 | uio->uio_iovcnt--; |
| 185 | if (uio->uio_iovcnt < 0) |
| 186 | panic("mmrw"); |
| 187 | continue; |
| 188 | } |
| 189 | switch (minor(dev)) { |
| 190 | case 0: |
| 191 | /* |
| 192 | * minor device 0 is physical memory, /dev/mem |
| 193 | */ |
| 194 | v = uio->uio_offset; |
| 195 | v &= ~(long)PAGE_MASK; |
| 196 | pmap_kenter((vm_offset_t)ptvmmap, v); |
| 197 | o = (int)uio->uio_offset & PAGE_MASK; |
| 198 | c = (u_int)(PAGE_SIZE - ((uintptr_t)iov->iov_base & PAGE_MASK)); |
| 199 | c = min(c, (u_int)(PAGE_SIZE - o)); |
| 200 | c = min(c, (u_int)iov->iov_len); |
| 201 | error = uiomove((caddr_t)&ptvmmap[o], (int)c, uio); |
| 202 | pmap_kremove((vm_offset_t)ptvmmap); |
| 203 | continue; |
| 204 | |
| 205 | case 1: { |
| 206 | /* |
| 207 | * minor device 1 is kernel memory, /dev/kmem |
| 208 | */ |
| 209 | vm_offset_t saddr, eaddr; |
| 210 | int prot; |
| 211 | |
| 212 | c = iov->iov_len; |
| 213 | |
| 214 | /* |
| 215 | * Make sure that all of the pages are currently |
| 216 | * resident so that we don't create any zero-fill |
| 217 | * pages. |
| 218 | */ |
| 219 | saddr = trunc_page(uio->uio_offset); |
| 220 | eaddr = round_page(uio->uio_offset + c); |
| 221 | if (saddr > eaddr) |
| 222 | return EFAULT; |
| 223 | |
| 224 | /* |
| 225 | * Make sure the kernel addresses are mapped. |
| 226 | * platform_direct_mapped() can be used to bypass |
| 227 | * default mapping via the page table (virtual kernels |
| 228 | * contain a lot of out-of-band data). |
| 229 | */ |
| 230 | prot = VM_PROT_READ; |
| 231 | if (uio->uio_rw != UIO_READ) |
| 232 | prot |= VM_PROT_WRITE; |
| 233 | error = kvm_access_check(saddr, eaddr, prot); |
| 234 | if (error) |
| 235 | return (error); |
| 236 | error = uiomove((caddr_t)(vm_offset_t)uio->uio_offset, |
| 237 | (int)c, uio); |
| 238 | continue; |
| 239 | } |
| 240 | case 2: |
| 241 | /* |
| 242 | * minor device 2 (/dev/null) is EOF/RATHOLE |
| 243 | */ |
| 244 | if (uio->uio_rw == UIO_READ) |
| 245 | return (0); |
| 246 | c = iov->iov_len; |
| 247 | break; |
| 248 | case 3: |
| 249 | /* |
| 250 | * minor device 3 (/dev/random) is source of filth |
| 251 | * on read, seeder on write |
| 252 | */ |
| 253 | if (buf == NULL) |
| 254 | buf = kmalloc(PAGE_SIZE, M_TEMP, M_WAITOK); |
| 255 | c = min(iov->iov_len, PAGE_SIZE); |
| 256 | if (uio->uio_rw == UIO_WRITE) { |
| 257 | error = uiomove(buf, (int)c, uio); |
| 258 | if (error == 0 && |
| 259 | seedenable && |
| 260 | securelevel <= 0) { |
| 261 | error = add_buffer_randomness_src(buf, c, RAND_SRC_SEEDING); |
| 262 | } else if (error == 0) { |
| 263 | error = EPERM; |
| 264 | } |
| 265 | } else { |
| 266 | poolsize = read_random(buf, c); |
| 267 | if (poolsize == 0) { |
| 268 | if (buf) |
| 269 | kfree(buf, M_TEMP); |
| 270 | if ((flags & IO_NDELAY) != 0) |
| 271 | return (EWOULDBLOCK); |
| 272 | return (0); |
| 273 | } |
| 274 | c = min(c, poolsize); |
| 275 | error = uiomove(buf, (int)c, uio); |
| 276 | } |
| 277 | continue; |
| 278 | case 4: |
| 279 | /* |
| 280 | * minor device 4 (/dev/urandom) is source of muck |
| 281 | * on read, writes are disallowed. |
| 282 | */ |
| 283 | c = min(iov->iov_len, PAGE_SIZE); |
| 284 | if (uio->uio_rw == UIO_WRITE) { |
| 285 | error = EPERM; |
| 286 | break; |
| 287 | } |
| 288 | if (CURSIG(curthread->td_lwp) != 0) { |
| 289 | /* |
| 290 | * Use tsleep() to get the error code right. |
| 291 | * It should return immediately. |
| 292 | */ |
| 293 | error = tsleep(&rand_bolt, PCATCH, "urand", 1); |
| 294 | if (error != 0 && error != EWOULDBLOCK) |
| 295 | continue; |
| 296 | } |
| 297 | if (buf == NULL) |
| 298 | buf = kmalloc(PAGE_SIZE, M_TEMP, M_WAITOK); |
| 299 | poolsize = read_random_unlimited(buf, c); |
| 300 | c = min(c, poolsize); |
| 301 | error = uiomove(buf, (int)c, uio); |
| 302 | continue; |
| 303 | /* case 5: read/write not supported, mmap only */ |
| 304 | /* case 6: read/write not supported, mmap only */ |
| 305 | case 12: |
| 306 | /* |
| 307 | * minor device 12 (/dev/zero) is source of nulls |
| 308 | * on read, write are disallowed. |
| 309 | */ |
| 310 | if (uio->uio_rw == UIO_WRITE) { |
| 311 | c = iov->iov_len; |
| 312 | break; |
| 313 | } |
| 314 | if (zbuf == NULL) { |
| 315 | zbuf = (caddr_t)kmalloc(PAGE_SIZE, M_TEMP, |
| 316 | M_WAITOK | M_ZERO); |
| 317 | } |
| 318 | c = min(iov->iov_len, PAGE_SIZE); |
| 319 | error = uiomove(zbuf, (int)c, uio); |
| 320 | continue; |
| 321 | default: |
| 322 | return (ENODEV); |
| 323 | } |
| 324 | if (error) |
| 325 | break; |
| 326 | iov->iov_base = (char *)iov->iov_base + c; |
| 327 | iov->iov_len -= c; |
| 328 | uio->uio_offset += c; |
| 329 | uio->uio_resid -= c; |
| 330 | } |
| 331 | if (buf) |
| 332 | kfree(buf, M_TEMP); |
| 333 | return (error); |
| 334 | } |
| 335 | |
| 336 | static int |
| 337 | mmread(struct dev_read_args *ap) |
| 338 | { |
| 339 | return(mmrw(ap->a_head.a_dev, ap->a_uio, ap->a_ioflag)); |
| 340 | } |
| 341 | |
| 342 | static int |
| 343 | mmwrite(struct dev_write_args *ap) |
| 344 | { |
| 345 | return(mmrw(ap->a_head.a_dev, ap->a_uio, ap->a_ioflag)); |
| 346 | } |
| 347 | |
| 348 | /*******************************************************\ |
| 349 | * allow user processes to MMAP some memory sections * |
| 350 | * instead of going through read/write * |
| 351 | \*******************************************************/ |
| 352 | |
| 353 | static int user_kernel_mapping(int num, vm_ooffset_t offset, |
| 354 | vm_ooffset_t *resultp); |
| 355 | |
| 356 | #if 0 |
| 357 | |
| 358 | static int |
| 359 | memmmap(struct dev_mmap_args *ap) |
| 360 | { |
| 361 | cdev_t dev = ap->a_head.a_dev; |
| 362 | vm_ooffset_t result; |
| 363 | int error; |
| 364 | |
| 365 | switch (minor(dev)) { |
| 366 | case 0: |
| 367 | /* |
| 368 | * minor device 0 is physical memory |
| 369 | */ |
| 370 | ap->a_result = atop(ap->a_offset); |
| 371 | error = 0; |
| 372 | break; |
| 373 | case 1: |
| 374 | /* |
| 375 | * minor device 1 is kernel memory |
| 376 | */ |
| 377 | ap->a_result = atop(vtophys(ap->a_offset)); |
| 378 | error = 0; |
| 379 | break; |
| 380 | case 5: |
| 381 | case 6: |
| 382 | /* |
| 383 | * minor device 5 is /dev/upmap (see sys/upmap.h) |
| 384 | * minor device 6 is /dev/kpmap (see sys/upmap.h) |
| 385 | */ |
| 386 | result = 0; |
| 387 | error = user_kernel_mapping(minor(dev), ap->a_offset, &result); |
| 388 | ap->a_result = atop(result); |
| 389 | break; |
| 390 | default: |
| 391 | error = EINVAL; |
| 392 | break; |
| 393 | } |
| 394 | return error; |
| 395 | } |
| 396 | |
| 397 | #endif |
| 398 | |
| 399 | static int |
| 400 | memuksmap(cdev_t dev, vm_page_t fake) |
| 401 | { |
| 402 | vm_ooffset_t result; |
| 403 | int error; |
| 404 | |
| 405 | switch (minor(dev)) { |
| 406 | case 0: |
| 407 | /* |
| 408 | * minor device 0 is physical memory |
| 409 | */ |
| 410 | fake->phys_addr = ptoa(fake->pindex); |
| 411 | error = 0; |
| 412 | break; |
| 413 | case 1: |
| 414 | /* |
| 415 | * minor device 1 is kernel memory |
| 416 | */ |
| 417 | fake->phys_addr = vtophys(ptoa(fake->pindex)); |
| 418 | error = 0; |
| 419 | break; |
| 420 | case 5: |
| 421 | case 6: |
| 422 | /* |
| 423 | * minor device 5 is /dev/upmap (see sys/upmap.h) |
| 424 | * minor device 6 is /dev/kpmap (see sys/upmap.h) |
| 425 | */ |
| 426 | result = 0; |
| 427 | error = user_kernel_mapping(minor(dev), |
| 428 | ptoa(fake->pindex), &result); |
| 429 | fake->phys_addr = result; |
| 430 | break; |
| 431 | default: |
| 432 | error = EINVAL; |
| 433 | break; |
| 434 | } |
| 435 | return error; |
| 436 | } |
| 437 | |
| 438 | static int |
| 439 | mmioctl(struct dev_ioctl_args *ap) |
| 440 | { |
| 441 | cdev_t dev = ap->a_head.a_dev; |
| 442 | int error; |
| 443 | |
| 444 | get_mplock(); |
| 445 | |
| 446 | switch (minor(dev)) { |
| 447 | case 0: |
| 448 | error = mem_ioctl(dev, ap->a_cmd, ap->a_data, |
| 449 | ap->a_fflag, ap->a_cred); |
| 450 | break; |
| 451 | case 3: |
| 452 | case 4: |
| 453 | error = random_ioctl(dev, ap->a_cmd, ap->a_data, |
| 454 | ap->a_fflag, ap->a_cred); |
| 455 | break; |
| 456 | default: |
| 457 | error = ENODEV; |
| 458 | break; |
| 459 | } |
| 460 | |
| 461 | rel_mplock(); |
| 462 | return (error); |
| 463 | } |
| 464 | |
| 465 | /* |
| 466 | * Operations for changing memory attributes. |
| 467 | * |
| 468 | * This is basically just an ioctl shim for mem_range_attr_get |
| 469 | * and mem_range_attr_set. |
| 470 | */ |
| 471 | static int |
| 472 | mem_ioctl(cdev_t dev, u_long cmd, caddr_t data, int flags, struct ucred *cred) |
| 473 | { |
| 474 | int nd, error = 0; |
| 475 | struct mem_range_op *mo = (struct mem_range_op *)data; |
| 476 | struct mem_range_desc *md; |
| 477 | |
| 478 | /* is this for us? */ |
| 479 | if ((cmd != MEMRANGE_GET) && |
| 480 | (cmd != MEMRANGE_SET)) |
| 481 | return (ENOTTY); |
| 482 | |
| 483 | /* any chance we can handle this? */ |
| 484 | if (mem_range_softc.mr_op == NULL) |
| 485 | return (EOPNOTSUPP); |
| 486 | |
| 487 | /* do we have any descriptors? */ |
| 488 | if (mem_range_softc.mr_ndesc == 0) |
| 489 | return (ENXIO); |
| 490 | |
| 491 | switch (cmd) { |
| 492 | case MEMRANGE_GET: |
| 493 | nd = imin(mo->mo_arg[0], mem_range_softc.mr_ndesc); |
| 494 | if (nd > 0) { |
| 495 | md = (struct mem_range_desc *) |
| 496 | kmalloc(nd * sizeof(struct mem_range_desc), |
| 497 | M_MEMDESC, M_WAITOK); |
| 498 | error = mem_range_attr_get(md, &nd); |
| 499 | if (!error) |
| 500 | error = copyout(md, mo->mo_desc, |
| 501 | nd * sizeof(struct mem_range_desc)); |
| 502 | kfree(md, M_MEMDESC); |
| 503 | } else { |
| 504 | nd = mem_range_softc.mr_ndesc; |
| 505 | } |
| 506 | mo->mo_arg[0] = nd; |
| 507 | break; |
| 508 | |
| 509 | case MEMRANGE_SET: |
| 510 | md = (struct mem_range_desc *)kmalloc(sizeof(struct mem_range_desc), |
| 511 | M_MEMDESC, M_WAITOK); |
| 512 | error = copyin(mo->mo_desc, md, sizeof(struct mem_range_desc)); |
| 513 | /* clamp description string */ |
| 514 | md->mr_owner[sizeof(md->mr_owner) - 1] = 0; |
| 515 | if (error == 0) |
| 516 | error = mem_range_attr_set(md, &mo->mo_arg[0]); |
| 517 | kfree(md, M_MEMDESC); |
| 518 | break; |
| 519 | } |
| 520 | return (error); |
| 521 | } |
| 522 | |
| 523 | /* |
| 524 | * Implementation-neutral, kernel-callable functions for manipulating |
| 525 | * memory range attributes. |
| 526 | */ |
| 527 | int |
| 528 | mem_range_attr_get(struct mem_range_desc *mrd, int *arg) |
| 529 | { |
| 530 | /* can we handle this? */ |
| 531 | if (mem_range_softc.mr_op == NULL) |
| 532 | return (EOPNOTSUPP); |
| 533 | |
| 534 | if (*arg == 0) { |
| 535 | *arg = mem_range_softc.mr_ndesc; |
| 536 | } else { |
| 537 | bcopy(mem_range_softc.mr_desc, mrd, (*arg) * sizeof(struct mem_range_desc)); |
| 538 | } |
| 539 | return (0); |
| 540 | } |
| 541 | |
| 542 | int |
| 543 | mem_range_attr_set(struct mem_range_desc *mrd, int *arg) |
| 544 | { |
| 545 | /* can we handle this? */ |
| 546 | if (mem_range_softc.mr_op == NULL) |
| 547 | return (EOPNOTSUPP); |
| 548 | |
| 549 | return (mem_range_softc.mr_op->set(&mem_range_softc, mrd, arg)); |
| 550 | } |
| 551 | |
| 552 | void |
| 553 | mem_range_AP_init(void) |
| 554 | { |
| 555 | if (mem_range_softc.mr_op && mem_range_softc.mr_op->initAP) |
| 556 | mem_range_softc.mr_op->initAP(&mem_range_softc); |
| 557 | } |
| 558 | |
| 559 | static int |
| 560 | random_ioctl(cdev_t dev, u_long cmd, caddr_t data, int flags, struct ucred *cred) |
| 561 | { |
| 562 | int error; |
| 563 | int intr; |
| 564 | |
| 565 | /* |
| 566 | * Even inspecting the state is privileged, since it gives a hint |
| 567 | * about how easily the randomness might be guessed. |
| 568 | */ |
| 569 | error = 0; |
| 570 | |
| 571 | switch (cmd) { |
| 572 | /* Really handled in upper layer */ |
| 573 | case FIOASYNC: |
| 574 | break; |
| 575 | case MEM_SETIRQ: |
| 576 | intr = *(int16_t *)data; |
| 577 | if ((error = priv_check_cred(cred, PRIV_ROOT, 0)) != 0) |
| 578 | break; |
| 579 | if (intr < 0 || intr >= MAX_INTS) |
| 580 | return (EINVAL); |
| 581 | register_randintr(intr); |
| 582 | break; |
| 583 | case MEM_CLEARIRQ: |
| 584 | intr = *(int16_t *)data; |
| 585 | if ((error = priv_check_cred(cred, PRIV_ROOT, 0)) != 0) |
| 586 | break; |
| 587 | if (intr < 0 || intr >= MAX_INTS) |
| 588 | return (EINVAL); |
| 589 | unregister_randintr(intr); |
| 590 | break; |
| 591 | case MEM_RETURNIRQ: |
| 592 | error = ENOTSUP; |
| 593 | break; |
| 594 | case MEM_FINDIRQ: |
| 595 | intr = *(int16_t *)data; |
| 596 | if ((error = priv_check_cred(cred, PRIV_ROOT, 0)) != 0) |
| 597 | break; |
| 598 | if (intr < 0 || intr >= MAX_INTS) |
| 599 | return (EINVAL); |
| 600 | intr = next_registered_randintr(intr); |
| 601 | if (intr == MAX_INTS) |
| 602 | return (ENOENT); |
| 603 | *(u_int16_t *)data = intr; |
| 604 | break; |
| 605 | default: |
| 606 | error = ENOTSUP; |
| 607 | break; |
| 608 | } |
| 609 | return (error); |
| 610 | } |
| 611 | |
| 612 | static int |
| 613 | mm_filter_read(struct knote *kn, long hint) |
| 614 | { |
| 615 | return (1); |
| 616 | } |
| 617 | |
| 618 | static int |
| 619 | mm_filter_write(struct knote *kn, long hint) |
| 620 | { |
| 621 | return (1); |
| 622 | } |
| 623 | |
| 624 | static void |
| 625 | dummy_filter_detach(struct knote *kn) {} |
| 626 | |
| 627 | /* Implemented in kern_nrandom.c */ |
| 628 | static struct filterops random_read_filtops = |
| 629 | { FILTEROP_ISFD|FILTEROP_MPSAFE, NULL, dummy_filter_detach, random_filter_read }; |
| 630 | |
| 631 | static struct filterops mm_read_filtops = |
| 632 | { FILTEROP_ISFD|FILTEROP_MPSAFE, NULL, dummy_filter_detach, mm_filter_read }; |
| 633 | |
| 634 | static struct filterops mm_write_filtops = |
| 635 | { FILTEROP_ISFD|FILTEROP_MPSAFE, NULL, dummy_filter_detach, mm_filter_write }; |
| 636 | |
| 637 | static int |
| 638 | mmkqfilter(struct dev_kqfilter_args *ap) |
| 639 | { |
| 640 | struct knote *kn = ap->a_kn; |
| 641 | cdev_t dev = ap->a_head.a_dev; |
| 642 | |
| 643 | ap->a_result = 0; |
| 644 | switch (kn->kn_filter) { |
| 645 | case EVFILT_READ: |
| 646 | switch (minor(dev)) { |
| 647 | case 3: |
| 648 | kn->kn_fop = &random_read_filtops; |
| 649 | break; |
| 650 | default: |
| 651 | kn->kn_fop = &mm_read_filtops; |
| 652 | break; |
| 653 | } |
| 654 | break; |
| 655 | case EVFILT_WRITE: |
| 656 | kn->kn_fop = &mm_write_filtops; |
| 657 | break; |
| 658 | default: |
| 659 | ap->a_result = EOPNOTSUPP; |
| 660 | return (0); |
| 661 | } |
| 662 | |
| 663 | return (0); |
| 664 | } |
| 665 | |
| 666 | int |
| 667 | iszerodev(cdev_t dev) |
| 668 | { |
| 669 | return (zerodev == dev); |
| 670 | } |
| 671 | |
| 672 | /* |
| 673 | * /dev/upmap and /dev/kpmap. |
| 674 | */ |
| 675 | static int |
| 676 | user_kernel_mapping(int num, vm_ooffset_t offset, vm_ooffset_t *resultp) |
| 677 | { |
| 678 | struct proc *p; |
| 679 | int error; |
| 680 | int invfork; |
| 681 | |
| 682 | if ((p = curproc) == NULL) |
| 683 | return (EINVAL); |
| 684 | |
| 685 | /* |
| 686 | * If this is a child currently in vfork the pmap is shared with |
| 687 | * the parent! We need to actually set-up the parent's p_upmap, |
| 688 | * not the child's, and we need to set the invfork flag. Userland |
| 689 | * will probably adjust its static state so it must be consistent |
| 690 | * with the parent or userland will be really badly confused. |
| 691 | * |
| 692 | * (this situation can happen when user code in vfork() calls |
| 693 | * libc's getpid() or some other function which then decides |
| 694 | * it wants the upmap). |
| 695 | */ |
| 696 | if (p->p_flags & P_PPWAIT) { |
| 697 | p = p->p_pptr; |
| 698 | if (p == NULL) |
| 699 | return (EINVAL); |
| 700 | invfork = 1; |
| 701 | } else { |
| 702 | invfork = 0; |
| 703 | } |
| 704 | |
| 705 | error = EINVAL; |
| 706 | |
| 707 | switch(num) { |
| 708 | case 5: |
| 709 | /* |
| 710 | * /dev/upmap - maps RW per-process shared user-kernel area. |
| 711 | */ |
| 712 | if (p->p_upmap == NULL) |
| 713 | proc_usermap(p, invfork); |
| 714 | else if (invfork) |
| 715 | p->p_upmap->invfork = invfork; |
| 716 | |
| 717 | if (p->p_upmap && |
| 718 | offset < roundup2(sizeof(*p->p_upmap), PAGE_SIZE)) { |
| 719 | /* only good for current process */ |
| 720 | *resultp = pmap_kextract((vm_offset_t)p->p_upmap + |
| 721 | offset); |
| 722 | error = 0; |
| 723 | } |
| 724 | break; |
| 725 | case 6: |
| 726 | /* |
| 727 | * /dev/kpmap - maps RO shared kernel global page |
| 728 | */ |
| 729 | if (kpmap && |
| 730 | offset < roundup2(sizeof(*kpmap), PAGE_SIZE)) { |
| 731 | *resultp = pmap_kextract((vm_offset_t)kpmap + |
| 732 | offset); |
| 733 | error = 0; |
| 734 | } |
| 735 | break; |
| 736 | default: |
| 737 | break; |
| 738 | } |
| 739 | return error; |
| 740 | } |
| 741 | |
| 742 | static void |
| 743 | mem_drvinit(void *unused) |
| 744 | { |
| 745 | |
| 746 | /* Initialise memory range handling */ |
| 747 | if (mem_range_softc.mr_op != NULL) |
| 748 | mem_range_softc.mr_op->init(&mem_range_softc); |
| 749 | |
| 750 | make_dev(&mem_ops, 0, UID_ROOT, GID_KMEM, 0640, "mem"); |
| 751 | make_dev(&mem_ops, 1, UID_ROOT, GID_KMEM, 0640, "kmem"); |
| 752 | make_dev(&mem_ops, 2, UID_ROOT, GID_WHEEL, 0666, "null"); |
| 753 | make_dev(&mem_ops, 3, UID_ROOT, GID_WHEEL, 0644, "random"); |
| 754 | make_dev(&mem_ops, 4, UID_ROOT, GID_WHEEL, 0644, "urandom"); |
| 755 | make_dev(&mem_ops, 5, UID_ROOT, GID_WHEEL, 0666, "upmap"); |
| 756 | make_dev(&mem_ops, 6, UID_ROOT, GID_WHEEL, 0444, "kpmap"); |
| 757 | zerodev = make_dev(&mem_ops, 12, UID_ROOT, GID_WHEEL, 0666, "zero"); |
| 758 | make_dev(&mem_ops, 14, UID_ROOT, GID_WHEEL, 0600, "io"); |
| 759 | } |
| 760 | |
| 761 | SYSINIT(memdev,SI_SUB_DRIVERS,SI_ORDER_MIDDLE+CDEV_MAJOR,mem_drvinit,NULL) |
| 762 | |