| 1 | /* |
| 2 | * Copyright (c) 2000-2006 Apple Computer, Inc. All rights reserved. |
| 3 | * |
| 4 | * @APPLE_OSREFERENCE_LICENSE_HEADER_START@ |
| 5 | * |
| 6 | * This file contains Original Code and/or Modifications of Original Code |
| 7 | * as defined in and that are subject to the Apple Public Source License |
| 8 | * Version 2.0 (the 'License'). You may not use this file except in |
| 9 | * compliance with the License. The rights granted to you under the License |
| 10 | * may not be used to create, or enable the creation or redistribution of, |
| 11 | * unlawful or unlicensed copies of an Apple operating system, or to |
| 12 | * circumvent, violate, or enable the circumvention or violation of, any |
| 13 | * terms of an Apple operating system software license agreement. |
| 14 | * |
| 15 | * Please obtain a copy of the License at |
| 16 | * http://www.opensource.apple.com/apsl/ and read it before using this file. |
| 17 | * |
| 18 | * The Original Code and all software distributed under the License are |
| 19 | * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER |
| 20 | * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES, |
| 21 | * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY, |
| 22 | * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT. |
| 23 | * Please see the License for the specific language governing rights and |
| 24 | * limitations under the License. |
| 25 | * |
| 26 | * @APPLE_OSREFERENCE_LICENSE_HEADER_END@ |
| 27 | */ |
| 28 | /* Copyright (c) 1995 NeXT Computer, Inc. All Rights Reserved */ |
| 29 | /* |
| 30 | * Copyright (c) 1982, 1986, 1991, 1993 |
| 31 | * The Regents of the University of California. All rights reserved. |
| 32 | * (c) UNIX System Laboratories, Inc. |
| 33 | * All or some portions of this file are derived from material licensed |
| 34 | * to the University of California by American Telephone and Telegraph |
| 35 | * Co. or Unix System Laboratories, Inc. and are reproduced herein with |
| 36 | * the permission of UNIX System Laboratories, Inc. |
| 37 | * |
| 38 | * Redistribution and use in source and binary forms, with or without |
| 39 | * modification, are permitted provided that the following conditions |
| 40 | * are met: |
| 41 | * 1. Redistributions of source code must retain the above copyright |
| 42 | * notice, this list of conditions and the following disclaimer. |
| 43 | * 2. Redistributions in binary form must reproduce the above copyright |
| 44 | * notice, this list of conditions and the following disclaimer in the |
| 45 | * documentation and/or other materials provided with the distribution. |
| 46 | * 3. All advertising materials mentioning features or use of this software |
| 47 | * must display the following acknowledgement: |
| 48 | * This product includes software developed by the University of |
| 49 | * California, Berkeley and its contributors. |
| 50 | * 4. Neither the name of the University nor the names of its contributors |
| 51 | * may be used to endorse or promote products derived from this software |
| 52 | * without specific prior written permission. |
| 53 | * |
| 54 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND |
| 55 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 56 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
| 57 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE |
| 58 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
| 59 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
| 60 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
| 61 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
| 62 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
| 63 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
| 64 | * SUCH DAMAGE. |
| 65 | * |
| 66 | * @(#)kern_subr.c 8.3 (Berkeley) 1/21/94 |
| 67 | */ |
| 68 | |
| 69 | #include <machine/atomic.h> |
| 70 | |
| 71 | #include <sys/param.h> |
| 72 | #include <sys/systm.h> |
| 73 | #include <sys/proc_internal.h> |
| 74 | #include <sys/malloc.h> |
| 75 | #include <sys/queue.h> |
| 76 | #include <vm/pmap.h> |
| 77 | #include <sys/uio_internal.h> |
| 78 | #include <kern/kalloc.h> |
| 79 | |
| 80 | #include <kdebug.h> |
| 81 | |
| 82 | #include <sys/kdebug.h> |
| 83 | #define DBG_UIO_COPYOUT 16 |
| 84 | #define DBG_UIO_COPYIN 17 |
| 85 | |
| 86 | #if DEBUG |
| 87 | #include <kern/simple_lock.h> |
| 88 | |
| 89 | static uint32_t uio_t_count = 0; |
| 90 | #endif /* DEBUG */ |
| 91 | |
| 92 | #define IS_VALID_UIO_SEGFLG(segflg) \ |
| 93 | ( (1 << segflg) & (UIOF_USERSPACE | \ |
| 94 | UIOF_SYSSPACE | \ |
| 95 | UIOF_USERSPACE32 | \ |
| 96 | UIOF_USERSPACE64 | \ |
| 97 | UIOF_SYSSPACE32 | \ |
| 98 | UIOF_USERISPACE | \ |
| 99 | UIOF_PHYS_USERSPACE | \ |
| 100 | UIOF_PHYS_SYSSPACE | \ |
| 101 | UIOF_USERISPACE32 | \ |
| 102 | UIOF_PHYS_USERSPACE32 | \ |
| 103 | UIOF_USERISPACE64 | \ |
| 104 | UIOF_PHYS_USERSPACE64)) |
| 105 | |
| 106 | #define IS_SYS_OR_PHYS_SPACE_SEGFLG(segflg) \ |
| 107 | ( (1 << segflg) & (UIOF_SYSSPACE | \ |
| 108 | UIOF_PHYS_SYSSPACE | \ |
| 109 | UIOF_SYSSPACE32 | \ |
| 110 | UIOF_PHYS_USERSPACE | \ |
| 111 | UIOF_PHYS_SYSSPACE | \ |
| 112 | UIOF_PHYS_USERSPACE64 | \ |
| 113 | UIOF_PHYS_USERSPACE32)) |
| 114 | |
| 115 | #define IS_PURE_USER_SPACE_SEGFLG(segflg) \ |
| 116 | ( (1 << segflg) & (UIOF_USERSPACE | \ |
| 117 | UIOF_USERSPACE32 | \ |
| 118 | UIOF_USERSPACE64 | \ |
| 119 | UIOF_USERISPACE | \ |
| 120 | UIOF_USERISPACE32 | \ |
| 121 | UIOF_USERISPACE64)) |
| 122 | |
| 123 | #define IS_SYS_SPACE_SEGFLG(segflg) \ |
| 124 | ( (1 << segflg) & (UIOF_SYSSPACE | \ |
| 125 | UIOF_SYSSPACE32)) |
| 126 | |
| 127 | #define IS_PHYS_USER_SPACE_SEGFLG(segflg) \ |
| 128 | ( (1 << segflg) & (UIOF_PHYS_USERSPACE | \ |
| 129 | UIOF_PHYS_USERSPACE64 | \ |
| 130 | UIOF_PHYS_USERSPACE32)) |
| 131 | |
| 132 | #define IS_PHYS_SYS_SPACE_SEGFLG(segflg) \ |
| 133 | ( (1 << segflg) & (UIOF_PHYS_SYSSPACE)) |
| 134 | |
| 135 | static void uio_update_user(uio_t __attribute__((nonnull)) a_uio, user_size_t a_count); |
| 136 | static void uio_update_sys(uio_t __attribute__((nonnull)) a_uio, user_size_t a_count); |
| 137 | static user_size_t uio_curriovlen_user(const uio_t __attribute__((nonnull)) a_uio); |
| 138 | static user_size_t uio_curriovlen_sys(const uio_t __attribute__((nonnull)) a_uio); |
| 139 | |
| 140 | #if __has_feature(ptrauth_calls) |
| 141 | __attribute__((always_inline)) |
| 142 | static u_int64_t |
| 143 | blend_iov_components(const struct kern_iovec *kiovp) |
| 144 | { |
| 145 | return ptrauth_blend_discriminator( |
| 146 | (void *)((u_int64_t)&kiovp->iov_base ^ kiovp->iov_len), |
| 147 | ptrauth_string_discriminator("kiovp" )); |
| 148 | } |
| 149 | #endif |
| 150 | |
| 151 | __attribute__((always_inline)) |
| 152 | static u_int64_t |
| 153 | kiovp_get_base(const struct kern_iovec *kiovp) |
| 154 | { |
| 155 | #if __has_feature(ptrauth_calls) |
| 156 | if (kiovp->iov_base == 0) { |
| 157 | return 0; |
| 158 | } else { |
| 159 | return (u_int64_t)ptrauth_auth_data((void *)kiovp->iov_base, |
| 160 | ptrauth_key_process_independent_data, |
| 161 | blend_iov_components(kiovp)); |
| 162 | } |
| 163 | #else |
| 164 | return kiovp->iov_base; |
| 165 | #endif |
| 166 | } |
| 167 | |
| 168 | __attribute__((always_inline)) |
| 169 | static void |
| 170 | kiovp_set_base(struct kern_iovec *kiovp, u_int64_t addr) |
| 171 | { |
| 172 | #if __has_feature(ptrauth_calls) |
| 173 | if (addr == 0) { |
| 174 | kiovp->iov_base = 0; |
| 175 | } else { |
| 176 | kiovp->iov_base = (u_int64_t)ptrauth_sign_unauthenticated( |
| 177 | (void *)addr, ptrauth_key_process_independent_data, |
| 178 | blend_iov_components(kiovp)); |
| 179 | } |
| 180 | #else |
| 181 | kiovp->iov_base = addr; |
| 182 | #endif |
| 183 | } |
| 184 | |
| 185 | static struct kern_iovec * |
| 186 | uio_kiovp(uio_t uio) |
| 187 | { |
| 188 | #if DEBUG |
| 189 | if (__improbable(!UIO_IS_SYS_SPACE(uio))) { |
| 190 | panic("%s: uio is not sys space" , __func__); |
| 191 | } |
| 192 | #endif |
| 193 | |
| 194 | return (struct kern_iovec *)uio->uio_iovs; |
| 195 | } |
| 196 | |
| 197 | static struct user_iovec * |
| 198 | uio_uiovp(uio_t uio) |
| 199 | { |
| 200 | return (struct user_iovec *)uio->uio_iovs; |
| 201 | } |
| 202 | |
| 203 | static void * |
| 204 | uio_advance_user(uio_t uio) |
| 205 | { |
| 206 | uio->uio_iovs = (void *)((uintptr_t)uio->uio_iovs + sizeof(struct user_iovec)); |
| 207 | |
| 208 | return uio->uio_iovs; |
| 209 | } |
| 210 | |
| 211 | static void * |
| 212 | uio_advance_sys(uio_t uio) |
| 213 | { |
| 214 | uio->uio_iovs = (void *)((uintptr_t)uio->uio_iovs + sizeof(struct kern_iovec)); |
| 215 | |
| 216 | return uio->uio_iovs; |
| 217 | } |
| 218 | |
| 219 | /* |
| 220 | * Returns: 0 Success |
| 221 | * uiomove64:EFAULT |
| 222 | * |
| 223 | * Notes: The first argument should be a caddr_t, but const poisoning |
| 224 | * for typedef'ed types doesn't work in gcc. |
| 225 | */ |
| 226 | int |
| 227 | uiomove(const char * cp, int n, uio_t uio) |
| 228 | { |
| 229 | return uiomove64(cp: (const addr64_t)(uintptr_t)cp, n, uio); |
| 230 | } |
| 231 | |
| 232 | /* |
| 233 | * Returns: 0 Success |
| 234 | * EFAULT |
| 235 | * copyout:EFAULT |
| 236 | * copyin:EFAULT |
| 237 | * copywithin:EFAULT |
| 238 | * copypv:EFAULT |
| 239 | */ |
| 240 | int |
| 241 | uiomove64(const addr64_t c_cp, int n, struct uio *uio) |
| 242 | { |
| 243 | if (IS_PURE_USER_SPACE_SEGFLG(uio->uio_segflg)) { |
| 244 | if (uio->uio_rw == UIO_READ) { |
| 245 | return uio_copyout_user(c_cp: (const char *)c_cp, n, uio); |
| 246 | } else { |
| 247 | return uio_copyin_user(c_cp: (const char *)c_cp, n, uio); |
| 248 | } |
| 249 | } else if (IS_SYS_SPACE_SEGFLG(uio->uio_segflg)) { |
| 250 | if (uio->uio_rw == UIO_READ) { |
| 251 | return uio_copyout_sys(c_cp: (const char *)c_cp, n, uio); |
| 252 | } else { |
| 253 | return uio_copyin_sys(c_cp: (const char *)c_cp, n, uio); |
| 254 | } |
| 255 | } else if (IS_PHYS_USER_SPACE_SEGFLG(uio->uio_segflg)) { |
| 256 | if (uio->uio_rw == UIO_READ) { |
| 257 | return uio_copyout_phys_user(c_cp: (const char *)c_cp, n, uio); |
| 258 | } else { |
| 259 | return uio_copyin_phys_user(c_cp: (const char *)c_cp, n, uio); |
| 260 | } |
| 261 | } else if (IS_PHYS_SYS_SPACE_SEGFLG(uio->uio_segflg)) { |
| 262 | if (uio->uio_rw == UIO_READ) { |
| 263 | return uio_copyout_phys_sys(c_cp: (const char *)c_cp, n, uio); |
| 264 | } else { |
| 265 | return uio_copyin_phys_sys(c_cp: (const char *)c_cp, n, uio); |
| 266 | } |
| 267 | } else { |
| 268 | return EINVAL; |
| 269 | } |
| 270 | } |
| 271 | |
| 272 | int |
| 273 | uio_copyout_user(const char *c_cp, int n, uio_t uio) |
| 274 | { |
| 275 | addr64_t cp = (const addr64_t)(uintptr_t)c_cp; |
| 276 | |
| 277 | while (n > 0 && uio->uio_iovcnt > 0 && uio_resid(a_uio: uio)) { |
| 278 | struct user_iovec *uiovp; |
| 279 | uint64_t acnt; |
| 280 | int error; |
| 281 | |
| 282 | uio_update_user(a_uio: uio, a_count: 0); |
| 283 | acnt = uio_curriovlen_user(a_uio: uio); |
| 284 | if (acnt == 0) { |
| 285 | continue; |
| 286 | } |
| 287 | if (n > 0 && acnt > (uint64_t)n) { |
| 288 | acnt = n; |
| 289 | } |
| 290 | |
| 291 | uiovp = uio_uiovp(uio); |
| 292 | |
| 293 | KERNEL_DEBUG((FSDBG_CODE(DBG_FSRW, DBG_UIO_COPYOUT)) | DBG_FUNC_START, |
| 294 | (int)cp, (uintptr_t)uiovp->iov_base, acnt, 0, 0); |
| 295 | |
| 296 | error = copyout(CAST_DOWN(caddr_t, cp), uiovp->iov_base, (size_t)acnt); |
| 297 | |
| 298 | KERNEL_DEBUG((FSDBG_CODE(DBG_FSRW, DBG_UIO_COPYOUT)) | DBG_FUNC_END, |
| 299 | (int)cp, (uintptr_t)uiovp->iov_base, acnt, 0, 0); |
| 300 | |
| 301 | if (error) { |
| 302 | return error; |
| 303 | } |
| 304 | |
| 305 | uio_update_user(a_uio: uio, a_count: (user_size_t)acnt); |
| 306 | cp += acnt; |
| 307 | n -= acnt; |
| 308 | } |
| 309 | return 0; |
| 310 | } |
| 311 | |
| 312 | int |
| 313 | uio_copyin_user(const char *c_cp, int n, uio_t uio) |
| 314 | { |
| 315 | addr64_t cp = (const addr64_t)(uintptr_t)c_cp; |
| 316 | |
| 317 | while (n > 0 && uio->uio_iovcnt > 0 && uio_resid(a_uio: uio)) { |
| 318 | struct user_iovec *uiovp; |
| 319 | uint64_t acnt; |
| 320 | int error; |
| 321 | |
| 322 | uio_update_user(a_uio: uio, a_count: 0); |
| 323 | acnt = uio_curriovlen_user(a_uio: uio); |
| 324 | if (acnt == 0) { |
| 325 | continue; |
| 326 | } |
| 327 | if (n > 0 && acnt > (uint64_t)n) { |
| 328 | acnt = n; |
| 329 | } |
| 330 | |
| 331 | uiovp = uio_uiovp(uio); |
| 332 | |
| 333 | KERNEL_DEBUG((FSDBG_CODE(DBG_FSRW, DBG_UIO_COPYIN)) | DBG_FUNC_START, |
| 334 | (uintptr_t)uiovp->iov_base, (int)cp, acnt, 0, 0); |
| 335 | |
| 336 | error = copyin(uiovp->iov_base, CAST_DOWN(caddr_t, cp), (size_t)acnt); |
| 337 | |
| 338 | KERNEL_DEBUG((FSDBG_CODE(DBG_FSRW, DBG_UIO_COPYIN)) | DBG_FUNC_END, |
| 339 | (uintptr_t)uiovp->iov_base, (int)cp, acnt, 0, 0); |
| 340 | |
| 341 | if (error) { |
| 342 | return error; |
| 343 | } |
| 344 | |
| 345 | uio_update_user(a_uio: uio, a_count: (user_size_t)acnt); |
| 346 | cp += acnt; |
| 347 | n -= acnt; |
| 348 | } |
| 349 | return 0; |
| 350 | } |
| 351 | |
| 352 | int |
| 353 | uio_copyout_sys(const char *c_cp, int n, uio_t uio) |
| 354 | { |
| 355 | addr64_t cp = (const addr64_t)(uintptr_t)c_cp; |
| 356 | |
| 357 | while (n > 0 && uio->uio_iovcnt > 0 && uio_resid(a_uio: uio)) { |
| 358 | struct kern_iovec *kiovp; |
| 359 | uint64_t acnt; |
| 360 | |
| 361 | uio_update_sys(a_uio: uio, a_count: 0); |
| 362 | acnt = uio_curriovlen_sys(a_uio: uio); |
| 363 | if (acnt == 0) { |
| 364 | continue; |
| 365 | } |
| 366 | if (n > 0 && acnt > (uint64_t)n) { |
| 367 | acnt = n; |
| 368 | } |
| 369 | |
| 370 | kiovp = uio_kiovp(uio); |
| 371 | |
| 372 | copywithin(CAST_DOWN(caddr_t, cp), CAST_DOWN(caddr_t, kiovp_get_base(kiovp)), |
| 373 | len: (size_t)acnt); |
| 374 | |
| 375 | uio_update_sys(a_uio: uio, a_count: (user_size_t)acnt); |
| 376 | cp += acnt; |
| 377 | n -= acnt; |
| 378 | } |
| 379 | return 0; |
| 380 | } |
| 381 | |
| 382 | int |
| 383 | uio_copyin_sys(const char *c_cp, int n, uio_t uio) |
| 384 | { |
| 385 | addr64_t cp = (const addr64_t)(uintptr_t)c_cp; |
| 386 | |
| 387 | while (n > 0 && uio->uio_iovcnt > 0 && uio_resid(a_uio: uio)) { |
| 388 | struct kern_iovec *kiovp; |
| 389 | uint64_t acnt; |
| 390 | |
| 391 | uio_update_sys(a_uio: uio, a_count: 0); |
| 392 | acnt = uio_curriovlen_sys(a_uio: uio); |
| 393 | if (acnt == 0) { |
| 394 | continue; |
| 395 | } |
| 396 | if (n > 0 && acnt > (uint64_t)n) { |
| 397 | acnt = n; |
| 398 | } |
| 399 | |
| 400 | kiovp = uio_kiovp(uio); |
| 401 | |
| 402 | copywithin(CAST_DOWN(caddr_t, kiovp_get_base(kiovp)), CAST_DOWN(caddr_t, cp), |
| 403 | len: (size_t)acnt); |
| 404 | |
| 405 | uio_update_sys(a_uio: uio, a_count: (user_size_t)acnt); |
| 406 | cp += acnt; |
| 407 | n -= acnt; |
| 408 | } |
| 409 | return 0; |
| 410 | } |
| 411 | |
| 412 | int |
| 413 | uio_copyout_phys_user(const char *c_cp, int n, uio_t uio) |
| 414 | { |
| 415 | addr64_t cp = (const addr64_t)(uintptr_t)c_cp; |
| 416 | |
| 417 | while (n > 0 && uio->uio_iovcnt > 0 && uio_resid(a_uio: uio)) { |
| 418 | struct user_iovec *uiovp; |
| 419 | uint64_t acnt; |
| 420 | int error; |
| 421 | |
| 422 | uio_update_user(a_uio: uio, a_count: 0); |
| 423 | acnt = uio_curriovlen_user(a_uio: uio); |
| 424 | if (acnt == 0) { |
| 425 | continue; |
| 426 | } |
| 427 | if (n > 0 && acnt > (uint64_t)n) { |
| 428 | acnt = n; |
| 429 | } |
| 430 | |
| 431 | acnt = MIN(acnt, UINT_MAX); |
| 432 | uiovp = uio_uiovp(uio); |
| 433 | |
| 434 | KERNEL_DEBUG((FSDBG_CODE(DBG_FSRW, DBG_UIO_COPYOUT)) | DBG_FUNC_START, |
| 435 | (int)cp, (uintptr_t)uiovp->iov_base, acnt, 1, 0); |
| 436 | |
| 437 | error = copypv(source: (addr64_t)cp, sink: uiovp->iov_base, size: (unsigned int)acnt, cppvPsrc | cppvNoRefSrc); |
| 438 | |
| 439 | KERNEL_DEBUG((FSDBG_CODE(DBG_FSRW, DBG_UIO_COPYOUT)) | DBG_FUNC_END, |
| 440 | (int)cp, (uintptr_t)uiovp->iov_base, acnt, 1, 0); |
| 441 | |
| 442 | if (error) { /* Copy virtual to physical */ |
| 443 | return EFAULT; |
| 444 | } |
| 445 | |
| 446 | uio_update_user(a_uio: uio, a_count: (user_size_t)acnt); |
| 447 | cp += acnt; |
| 448 | n -= acnt; |
| 449 | } |
| 450 | return 0; |
| 451 | } |
| 452 | |
| 453 | int |
| 454 | uio_copyin_phys_user(const char *c_cp, int n, uio_t uio) |
| 455 | { |
| 456 | addr64_t cp = (const addr64_t)(uintptr_t)c_cp; |
| 457 | |
| 458 | while (n > 0 && uio->uio_iovcnt > 0 && uio_resid(a_uio: uio)) { |
| 459 | struct user_iovec *uiovp; |
| 460 | uint64_t acnt; |
| 461 | int error; |
| 462 | |
| 463 | uio_update_user(a_uio: uio, a_count: 0); |
| 464 | acnt = uio_curriovlen_user(a_uio: uio); |
| 465 | if (acnt == 0) { |
| 466 | continue; |
| 467 | } |
| 468 | if (n > 0 && acnt > (uint64_t)n) { |
| 469 | acnt = n; |
| 470 | } |
| 471 | |
| 472 | acnt = MIN(acnt, UINT_MAX); |
| 473 | uiovp = uio_uiovp(uio); |
| 474 | |
| 475 | KERNEL_DEBUG((FSDBG_CODE(DBG_FSRW, DBG_UIO_COPYIN)) | DBG_FUNC_START, |
| 476 | (uintptr_t)uiovp->iov_base, (int)cp, acnt, 1, 0); |
| 477 | |
| 478 | error = copypv(source: uiovp->iov_base, sink: (addr64_t)cp, size: (unsigned int)acnt, cppvPsnk | cppvNoRefSrc | cppvNoModSnk); |
| 479 | |
| 480 | KERNEL_DEBUG((FSDBG_CODE(DBG_FSRW, DBG_UIO_COPYIN)) | DBG_FUNC_END, |
| 481 | (uintptr_t)uiovp->iov_base, (int)cp, acnt, 1, 0); |
| 482 | |
| 483 | if (error) { /* Copy virtual to physical */ |
| 484 | return EFAULT; |
| 485 | } |
| 486 | |
| 487 | uio_update_user(a_uio: uio, a_count: (user_size_t)acnt); |
| 488 | cp += acnt; |
| 489 | n -= acnt; |
| 490 | } |
| 491 | return 0; |
| 492 | } |
| 493 | |
| 494 | int |
| 495 | uio_copyout_phys_sys(const char *c_cp, int n, uio_t uio) |
| 496 | { |
| 497 | addr64_t cp = (const addr64_t)(uintptr_t)c_cp; |
| 498 | |
| 499 | while (n > 0 && uio->uio_iovcnt > 0 && uio_resid(a_uio: uio)) { |
| 500 | struct kern_iovec *kiovp; |
| 501 | uint64_t acnt; |
| 502 | int error; |
| 503 | |
| 504 | uio_update_sys(a_uio: uio, a_count: 0); |
| 505 | acnt = uio_curriovlen_sys(a_uio: uio); |
| 506 | if (acnt == 0) { |
| 507 | continue; |
| 508 | } |
| 509 | if (n > 0 && acnt > (uint64_t)n) { |
| 510 | acnt = n; |
| 511 | } |
| 512 | |
| 513 | acnt = MIN(acnt, UINT_MAX); |
| 514 | kiovp = uio_kiovp(uio); |
| 515 | |
| 516 | KERNEL_DEBUG((FSDBG_CODE(DBG_FSRW, DBG_UIO_COPYOUT)) | DBG_FUNC_START, |
| 517 | (int)cp, (uintptr_t)kiovp_get_base(kiovp), acnt, 2, 0); |
| 518 | |
| 519 | error = copypv(source: (addr64_t)cp, sink: (addr64_t)kiovp_get_base(kiovp), size: (unsigned int)acnt, cppvKmap | cppvPsrc | cppvNoRefSrc); |
| 520 | |
| 521 | KERNEL_DEBUG((FSDBG_CODE(DBG_FSRW, DBG_UIO_COPYOUT)) | DBG_FUNC_END, |
| 522 | (int)cp, (uintptr_t)kiovp_get_base(kiovp), acnt, 2, 0); |
| 523 | |
| 524 | if (error) { /* Copy virtual to physical */ |
| 525 | return EFAULT; |
| 526 | } |
| 527 | |
| 528 | uio_update_sys(a_uio: uio, a_count: (user_size_t)acnt); |
| 529 | cp += acnt; |
| 530 | n -= acnt; |
| 531 | } |
| 532 | return 0; |
| 533 | } |
| 534 | |
| 535 | int |
| 536 | uio_copyin_phys_sys(const char *c_cp, int n, uio_t uio) |
| 537 | { |
| 538 | addr64_t cp = (const addr64_t)(uintptr_t)c_cp; |
| 539 | |
| 540 | while (n > 0 && uio->uio_iovcnt > 0 && uio_resid(a_uio: uio)) { |
| 541 | struct kern_iovec *kiovp; |
| 542 | uint64_t acnt; |
| 543 | int error; |
| 544 | |
| 545 | uio_update_sys(a_uio: uio, a_count: 0); |
| 546 | acnt = uio_curriovlen_sys(a_uio: uio); |
| 547 | if (acnt == 0) { |
| 548 | continue; |
| 549 | } |
| 550 | if (n > 0 && acnt > (uint64_t)n) { |
| 551 | acnt = n; |
| 552 | } |
| 553 | |
| 554 | acnt = MIN(acnt, UINT_MAX); |
| 555 | kiovp = uio_kiovp(uio); |
| 556 | |
| 557 | KERNEL_DEBUG((FSDBG_CODE(DBG_FSRW, DBG_UIO_COPYIN)) | DBG_FUNC_START, |
| 558 | (uintptr_t)kiovp_get_base(kiovp), (int)cp, acnt, 2, 0); |
| 559 | |
| 560 | error = copypv(source: (addr64_t)kiovp_get_base(kiovp), sink: (addr64_t)cp, size: (unsigned int)acnt, cppvKmap | cppvPsnk | cppvNoRefSrc | cppvNoModSnk); |
| 561 | |
| 562 | KERNEL_DEBUG((FSDBG_CODE(DBG_FSRW, DBG_UIO_COPYIN)) | DBG_FUNC_END, |
| 563 | (uintptr_t)kiovp_get_base(kiovp), (int)cp, acnt, 2, 0); |
| 564 | |
| 565 | if (error) { /* Copy virtual to physical */ |
| 566 | return EFAULT; |
| 567 | } |
| 568 | |
| 569 | uio_update_sys(a_uio: uio, a_count: (user_size_t)acnt); |
| 570 | cp += acnt; |
| 571 | n -= acnt; |
| 572 | } |
| 573 | return 0; |
| 574 | } |
| 575 | |
| 576 | /* |
| 577 | * Give next character to user as result of read. |
| 578 | */ |
| 579 | int |
| 580 | ureadc(int c, struct uio *uio) |
| 581 | { |
| 582 | struct kern_iovec *kiovp; |
| 583 | struct user_iovec *uiovp; |
| 584 | |
| 585 | if (__improbable(uio_resid(uio) <= 0)) { |
| 586 | panic("ureadc: non-positive resid" ); |
| 587 | } |
| 588 | |
| 589 | if (IS_PURE_USER_SPACE_SEGFLG(uio->uio_segflg)) { |
| 590 | uio_update_user(a_uio: uio, a_count: 0); |
| 591 | |
| 592 | uiovp = uio_uiovp(uio); |
| 593 | |
| 594 | if (subyte(addr: (user_addr_t)uiovp->iov_base, byte: c) < 0) { |
| 595 | return EFAULT; |
| 596 | } |
| 597 | |
| 598 | uio_update_user(a_uio: uio, a_count: 1); |
| 599 | } else if (IS_SYS_SPACE_SEGFLG(uio->uio_segflg)) { |
| 600 | uio_update_sys(a_uio: uio, a_count: 0); |
| 601 | |
| 602 | kiovp = uio_kiovp(uio); |
| 603 | *(CAST_DOWN(caddr_t, kiovp_get_base(kiovp))) = (char)c; |
| 604 | |
| 605 | uio_update_sys(a_uio: uio, a_count: 1); |
| 606 | } |
| 607 | return 0; |
| 608 | } |
| 609 | |
| 610 | LIST_HEAD(generic_hash_head, generic); |
| 611 | |
| 612 | /* |
| 613 | * General routine to allocate a hash table. |
| 614 | */ |
| 615 | void * |
| 616 | hashinit(int elements, int type __unused, u_long *hashmask) |
| 617 | { |
| 618 | struct generic_hash_head *hashtbl; |
| 619 | vm_size_t hashsize; |
| 620 | |
| 621 | if (__improbable(elements <= 0)) { |
| 622 | panic("hashinit: bad cnt" ); |
| 623 | } |
| 624 | |
| 625 | hashsize = 1UL << (fls(elements) - 1); |
| 626 | hashtbl = kalloc_type(struct generic_hash_head, hashsize, Z_WAITOK | Z_ZERO); |
| 627 | if (hashtbl != NULL) { |
| 628 | *hashmask = hashsize - 1; |
| 629 | } |
| 630 | return hashtbl; |
| 631 | } |
| 632 | |
| 633 | void |
| 634 | hashdestroy(void *hash, int type __unused, u_long hashmask) |
| 635 | { |
| 636 | assert(powerof2(hashmask + 1)); |
| 637 | kfree_type(struct generic_hash_head, hashmask + 1, hash); |
| 638 | } |
| 639 | |
| 640 | /* |
| 641 | * uio_resid - return the residual IO value for the given uio_t |
| 642 | */ |
| 643 | user_ssize_t |
| 644 | uio_resid( uio_t a_uio ) |
| 645 | { |
| 646 | #if DEBUG |
| 647 | if (a_uio == NULL) { |
| 648 | printf("%s :%d - invalid uio_t\n" , __FILE__, __LINE__); |
| 649 | } |
| 650 | #endif /* DEBUG */ |
| 651 | |
| 652 | /* return 0 if there are no active iovecs */ |
| 653 | if (a_uio == NULL) { |
| 654 | return 0; |
| 655 | } |
| 656 | |
| 657 | return a_uio->uio_resid_64; |
| 658 | } |
| 659 | |
| 660 | /* |
| 661 | * uio_setresid - set the residual IO value for the given uio_t |
| 662 | */ |
| 663 | void |
| 664 | uio_setresid( uio_t a_uio, user_ssize_t a_value ) |
| 665 | { |
| 666 | #if DEBUG |
| 667 | if (__improbable(a_uio == NULL)) { |
| 668 | panic("invalid uio_t" ); |
| 669 | } |
| 670 | #endif /* DEBUG */ |
| 671 | |
| 672 | if (a_uio == NULL) { |
| 673 | return; |
| 674 | } |
| 675 | |
| 676 | a_uio->uio_resid_64 = a_value; |
| 677 | return; |
| 678 | } |
| 679 | |
| 680 | /* |
| 681 | * uio_curriovbase - return the base address of the current iovec associated |
| 682 | * with the given uio_t. May return 0. |
| 683 | */ |
| 684 | user_addr_t |
| 685 | uio_curriovbase( uio_t a_uio ) |
| 686 | { |
| 687 | struct kern_iovec *kiovp; |
| 688 | struct user_iovec *uiovp; |
| 689 | |
| 690 | if (a_uio == NULL || a_uio->uio_iovcnt < 1) { |
| 691 | return 0; |
| 692 | } |
| 693 | |
| 694 | if (UIO_IS_USER_SPACE(a_uio)) { |
| 695 | uiovp = uio_uiovp(uio: a_uio); |
| 696 | return uiovp->iov_base; |
| 697 | } |
| 698 | |
| 699 | kiovp = uio_kiovp(uio: a_uio); |
| 700 | return (user_addr_t)kiovp_get_base(kiovp); |
| 701 | } |
| 702 | |
| 703 | /* |
| 704 | * uio_curriovlen_user - return the length value of the current iovec associated |
| 705 | * with the given uio_t. |
| 706 | */ |
| 707 | static user_size_t |
| 708 | uio_curriovlen_user(const uio_t __attribute__((nonnull)) a_uio) |
| 709 | { |
| 710 | return uio_uiovp(uio: a_uio)->iov_len; |
| 711 | } |
| 712 | |
| 713 | /* |
| 714 | * uio_curriovlen_sys - return the length value of the current iovec associated |
| 715 | * with the given uio_t. |
| 716 | */ |
| 717 | static user_size_t |
| 718 | uio_curriovlen_sys(const uio_t __attribute__((nonnull)) a_uio ) |
| 719 | { |
| 720 | return (user_size_t)uio_kiovp(uio: a_uio)->iov_len; |
| 721 | } |
| 722 | |
| 723 | /* |
| 724 | * uio_curriovlen - return the length value of the current iovec associated |
| 725 | * with the given uio_t. |
| 726 | */ |
| 727 | user_size_t |
| 728 | uio_curriovlen( uio_t a_uio ) |
| 729 | { |
| 730 | if (a_uio == NULL || a_uio->uio_iovcnt < 1) { |
| 731 | return 0; |
| 732 | } |
| 733 | |
| 734 | if (UIO_IS_USER_SPACE(a_uio)) { |
| 735 | return uio_curriovlen_user(a_uio); |
| 736 | } |
| 737 | |
| 738 | return uio_curriovlen_sys(a_uio); |
| 739 | } |
| 740 | |
| 741 | /* |
| 742 | * uio_iovcnt - return count of active iovecs for the given uio_t |
| 743 | */ |
| 744 | int |
| 745 | uio_iovcnt( uio_t a_uio ) |
| 746 | { |
| 747 | if (a_uio == NULL) { |
| 748 | return 0; |
| 749 | } |
| 750 | |
| 751 | return a_uio->uio_iovcnt; |
| 752 | } |
| 753 | |
| 754 | /* |
| 755 | * uio_offset - return the current offset value for the given uio_t |
| 756 | */ |
| 757 | off_t |
| 758 | uio_offset( uio_t a_uio ) |
| 759 | { |
| 760 | if (a_uio == NULL) { |
| 761 | return 0; |
| 762 | } |
| 763 | return a_uio->uio_offset; |
| 764 | } |
| 765 | |
| 766 | /* |
| 767 | * uio_setoffset - set the current offset value for the given uio_t |
| 768 | */ |
| 769 | void |
| 770 | uio_setoffset( uio_t a_uio, off_t a_offset ) |
| 771 | { |
| 772 | if (a_uio == NULL) { |
| 773 | return; |
| 774 | } |
| 775 | a_uio->uio_offset = a_offset; |
| 776 | return; |
| 777 | } |
| 778 | |
| 779 | /* |
| 780 | * uio_rw - return the read / write flag for the given uio_t |
| 781 | */ |
| 782 | int |
| 783 | uio_rw( uio_t a_uio ) |
| 784 | { |
| 785 | if (a_uio == NULL) { |
| 786 | return -1; |
| 787 | } |
| 788 | return a_uio->uio_rw; |
| 789 | } |
| 790 | |
| 791 | /* |
| 792 | * uio_setrw - set the read / write flag for the given uio_t |
| 793 | */ |
| 794 | void |
| 795 | uio_setrw( uio_t a_uio, int a_value ) |
| 796 | { |
| 797 | if (a_uio == NULL) { |
| 798 | return; |
| 799 | } |
| 800 | |
| 801 | if (a_value == UIO_READ || a_value == UIO_WRITE) { |
| 802 | a_uio->uio_rw = a_value; |
| 803 | } |
| 804 | return; |
| 805 | } |
| 806 | |
| 807 | /* |
| 808 | * uio_isuserspace - return non zero value if the address space |
| 809 | * flag is for a user address space (could be 32 or 64 bit). |
| 810 | */ |
| 811 | int |
| 812 | uio_isuserspace( uio_t a_uio ) |
| 813 | { |
| 814 | if (a_uio == NULL) { |
| 815 | return 0; |
| 816 | } |
| 817 | |
| 818 | if (UIO_SEG_IS_USER_SPACE(a_uio->uio_segflg)) { |
| 819 | return 1; |
| 820 | } |
| 821 | return 0; |
| 822 | } |
| 823 | |
| 824 | static void |
| 825 | uio_init(uio_t uio, |
| 826 | int a_iovcount, /* number of iovecs */ |
| 827 | off_t a_offset, /* current offset */ |
| 828 | int a_spacetype, /* type of address space */ |
| 829 | int a_iodirection, /* read or write flag */ |
| 830 | void *iovecs) /* pointer to iovec array */ |
| 831 | { |
| 832 | assert(a_iovcount >= 0 && a_iovcount <= UIO_MAXIOV); |
| 833 | assert(IS_VALID_UIO_SEGFLG(a_spacetype)); |
| 834 | assert(a_iodirection == UIO_READ || a_iodirection == UIO_WRITE); |
| 835 | |
| 836 | /* |
| 837 | * we use uio_segflg to indicate if the uio_t is the new format or |
| 838 | * old (pre LP64 support) legacy format |
| 839 | * This if-statement should canonicalize incoming space type |
| 840 | * to one of UIO_USERSPACE32/64, UIO_PHYS_USERSPACE32/64, or |
| 841 | * UIO_SYSSPACE/UIO_PHYS_SYSSPACE |
| 842 | */ |
| 843 | if (__improbable((1 << a_spacetype) & (UIOF_USERSPACE | UIOF_SYSSPACE32 | UIOF_PHYS_USERSPACE))) { |
| 844 | if (a_spacetype == UIO_USERSPACE) { |
| 845 | uio->uio_segflg = UIO_USERSPACE32; |
| 846 | } else if (a_spacetype == UIO_SYSSPACE32) { |
| 847 | uio->uio_segflg = UIO_SYSSPACE; |
| 848 | } else if (a_spacetype == UIO_PHYS_USERSPACE) { |
| 849 | uio->uio_segflg = UIO_PHYS_USERSPACE32; |
| 850 | } |
| 851 | } else { |
| 852 | uio->uio_segflg = a_spacetype; |
| 853 | } |
| 854 | |
| 855 | uio->uio_iovbase = iovecs; |
| 856 | uio->uio_iovs = iovecs; |
| 857 | uio->uio_max_iovs = a_iovcount; |
| 858 | uio->uio_offset = a_offset; |
| 859 | uio->uio_rw = a_iodirection; |
| 860 | uio->uio_flags = UIO_FLAGS_INITED; |
| 861 | } |
| 862 | |
| 863 | static void * |
| 864 | uio_alloc_iov_array(int a_spacetype, size_t a_iovcount) |
| 865 | { |
| 866 | if (IS_SYS_OR_PHYS_SPACE_SEGFLG(a_spacetype)) { |
| 867 | return kalloc_type(struct kern_iovec, a_iovcount, Z_WAITOK | Z_ZERO); |
| 868 | } |
| 869 | |
| 870 | size_t bytes = UIO_SIZEOF_IOVS(a_iovcount); |
| 871 | return kalloc_data(bytes, Z_WAITOK | Z_ZERO); |
| 872 | } |
| 873 | |
| 874 | static void |
| 875 | uio_free_iov_array(int a_spacetype, void *iovs, size_t a_iovcount) |
| 876 | { |
| 877 | if (IS_SYS_OR_PHYS_SPACE_SEGFLG(a_spacetype)) { |
| 878 | kfree_type(struct kern_iovec, a_iovcount, iovs); |
| 879 | } else { |
| 880 | size_t bytes = UIO_SIZEOF_IOVS(a_iovcount); |
| 881 | kfree_data(iovs, bytes); |
| 882 | } |
| 883 | } |
| 884 | |
| 885 | /* |
| 886 | * uio_create - create an uio_t. |
| 887 | * Space is allocated to hold up to a_iovcount number of iovecs. The uio_t |
| 888 | * is not fully initialized until all iovecs are added using uio_addiov calls. |
| 889 | * a_iovcount is the maximum number of iovecs you may add. |
| 890 | */ |
| 891 | uio_t |
| 892 | uio_create( int a_iovcount, /* number of iovecs */ |
| 893 | off_t a_offset, /* current offset */ |
| 894 | int a_spacetype, /* type of address space */ |
| 895 | int a_iodirection ) /* read or write flag */ |
| 896 | { |
| 897 | uio_t uio; |
| 898 | void *iovecs; |
| 899 | |
| 900 | if (a_iovcount < 0 || a_iovcount > UIO_MAXIOV) { |
| 901 | return NULL; |
| 902 | } |
| 903 | |
| 904 | uio = kalloc_type(struct uio, Z_WAITOK | Z_ZERO | Z_NOFAIL); |
| 905 | iovecs = uio_alloc_iov_array(a_spacetype, a_iovcount: (size_t)a_iovcount); |
| 906 | |
| 907 | uio_init(uio, a_iovcount, a_offset, a_spacetype, a_iodirection, iovecs); |
| 908 | |
| 909 | /* leave a note that we allocated this uio_t */ |
| 910 | uio->uio_flags |= UIO_FLAGS_WE_ALLOCED; |
| 911 | #if DEBUG |
| 912 | os_atomic_inc(&uio_t_count, relaxed); |
| 913 | #endif |
| 914 | |
| 915 | return uio; |
| 916 | } |
| 917 | |
| 918 | |
| 919 | /* |
| 920 | * uio_createwithbuffer - create an uio_t. |
| 921 | * Create a uio_t using the given buffer. The uio_t |
| 922 | * is not fully initialized until all iovecs are added using uio_addiov calls. |
| 923 | * a_iovcount is the maximum number of iovecs you may add. |
| 924 | * This call may fail if the given buffer is not large enough. |
| 925 | */ |
| 926 | __private_extern__ uio_t |
| 927 | uio_createwithbuffer( int a_iovcount, /* number of iovecs */ |
| 928 | off_t a_offset, /* current offset */ |
| 929 | int a_spacetype, /* type of address space */ |
| 930 | int a_iodirection, /* read or write flag */ |
| 931 | void *a_buf_p, /* pointer to a uio_t buffer */ |
| 932 | size_t a_buffer_size ) /* size of uio_t buffer */ |
| 933 | { |
| 934 | uio_t uio = (uio_t) a_buf_p; |
| 935 | void *iovecs = NULL; |
| 936 | |
| 937 | if (a_iovcount < 0 || a_iovcount > UIO_MAXIOV) { |
| 938 | return NULL; |
| 939 | } |
| 940 | |
| 941 | if (a_buffer_size < UIO_SIZEOF(a_iovcount)) { |
| 942 | return NULL; |
| 943 | } |
| 944 | |
| 945 | if (a_iovcount > 0) { |
| 946 | iovecs = (uint8_t *)uio + sizeof(struct uio); |
| 947 | } |
| 948 | |
| 949 | bzero(s: a_buf_p, n: a_buffer_size); |
| 950 | uio_init(uio, a_iovcount, a_offset, a_spacetype, a_iodirection, iovecs); |
| 951 | |
| 952 | return uio; |
| 953 | } |
| 954 | |
| 955 | /* |
| 956 | * uio_iovsaddr_user - get the address of the iovec array for the given uio_t. |
| 957 | * This returns the location of the iovecs within the uio. |
| 958 | * NOTE - for compatibility mode we just return the current value in uio_iovs |
| 959 | * which will increase as the IO is completed and is NOT embedded within the |
| 960 | * uio, it is a seperate array of one or more iovecs. |
| 961 | */ |
| 962 | __private_extern__ struct user_iovec * |
| 963 | uio_iovsaddr_user( uio_t a_uio ) |
| 964 | { |
| 965 | if (a_uio == NULL) { |
| 966 | return NULL; |
| 967 | } |
| 968 | |
| 969 | return uio_uiovp(uio: a_uio); |
| 970 | } |
| 971 | |
| 972 | static void |
| 973 | _uio_reset(uio_t a_uio, |
| 974 | off_t a_offset, /* current offset */ |
| 975 | int a_iodirection) /* read or write flag */ |
| 976 | { |
| 977 | void *my_iovs = a_uio->uio_iovbase; |
| 978 | int my_max_iovs = a_uio->uio_max_iovs; |
| 979 | |
| 980 | if (my_iovs != NULL) { |
| 981 | bzero(s: my_iovs, UIO_SIZEOF_IOVS(my_max_iovs)); |
| 982 | } |
| 983 | |
| 984 | a_uio->uio_iovs = my_iovs; |
| 985 | a_uio->uio_iovcnt = 0; |
| 986 | a_uio->uio_offset = a_offset; |
| 987 | a_uio->uio_segflg = 0; |
| 988 | a_uio->uio_rw = a_iodirection; |
| 989 | a_uio->uio_resid_64 = 0; |
| 990 | } |
| 991 | |
| 992 | void |
| 993 | uio_reset_fast( uio_t a_uio, |
| 994 | off_t a_offset, /* current offset */ |
| 995 | int a_spacetype, /* type of address space */ |
| 996 | int a_iodirection ) /* read or write flag */ |
| 997 | { |
| 998 | _uio_reset(a_uio, a_offset, a_iodirection); |
| 999 | |
| 1000 | a_uio->uio_segflg = a_spacetype; |
| 1001 | } |
| 1002 | |
| 1003 | /* |
| 1004 | * uio_reset - reset an uio_t. |
| 1005 | * Reset the given uio_t to initial values. The uio_t is not fully initialized |
| 1006 | * until all iovecs are added using uio_addiov calls. |
| 1007 | * The a_iovcount value passed in the uio_create is the maximum number of |
| 1008 | * iovecs you may add. |
| 1009 | */ |
| 1010 | void |
| 1011 | uio_reset( uio_t a_uio, |
| 1012 | off_t a_offset, /* current offset */ |
| 1013 | int a_spacetype, /* type of address space */ |
| 1014 | int a_iodirection ) /* read or write flag */ |
| 1015 | { |
| 1016 | if (a_uio == NULL) { |
| 1017 | return; |
| 1018 | } |
| 1019 | |
| 1020 | _uio_reset(a_uio, a_offset, a_iodirection); |
| 1021 | |
| 1022 | /* |
| 1023 | * we use uio_segflg to indicate if the uio_t is the new format or |
| 1024 | * old (pre LP64 support) legacy format |
| 1025 | * This switch statement should canonicalize incoming space type |
| 1026 | * to one of UIO_USERSPACE32/64, UIO_PHYS_USERSPACE32/64, or |
| 1027 | * UIO_SYSSPACE/UIO_PHYS_SYSSPACE |
| 1028 | */ |
| 1029 | switch (a_spacetype) { |
| 1030 | case UIO_USERSPACE: |
| 1031 | a_uio->uio_segflg = UIO_USERSPACE32; |
| 1032 | break; |
| 1033 | case UIO_SYSSPACE32: |
| 1034 | a_uio->uio_segflg = UIO_SYSSPACE; |
| 1035 | break; |
| 1036 | case UIO_PHYS_USERSPACE: |
| 1037 | a_uio->uio_segflg = UIO_PHYS_USERSPACE32; |
| 1038 | break; |
| 1039 | default: |
| 1040 | a_uio->uio_segflg = a_spacetype; |
| 1041 | break; |
| 1042 | } |
| 1043 | } |
| 1044 | |
| 1045 | /* |
| 1046 | * uio_free - free a uio_t allocated via uio_init. this also frees all |
| 1047 | * associated iovecs. |
| 1048 | */ |
| 1049 | void |
| 1050 | uio_free( uio_t a_uio ) |
| 1051 | { |
| 1052 | #if DEBUG |
| 1053 | if (__improbable(a_uio == NULL)) { |
| 1054 | panic("passing NULL uio_t" ); |
| 1055 | } |
| 1056 | #endif |
| 1057 | |
| 1058 | if (a_uio != NULL && (a_uio->uio_flags & UIO_FLAGS_WE_ALLOCED) != 0) { |
| 1059 | #if DEBUG |
| 1060 | if (__improbable(os_atomic_dec_orig(&uio_t_count, relaxed) == 0)) { |
| 1061 | panic("uio_t_count underflow" ); |
| 1062 | } |
| 1063 | #endif |
| 1064 | if (__improbable(a_uio->uio_max_iovs < 0 || a_uio->uio_max_iovs > UIO_MAXIOV)) { |
| 1065 | panic("%s: bad uio_max_iovs" , __func__); |
| 1066 | } |
| 1067 | |
| 1068 | uio_free_iov_array(a_spacetype: a_uio->uio_segflg, iovs: a_uio->uio_iovbase, |
| 1069 | a_iovcount: (size_t)a_uio->uio_max_iovs); |
| 1070 | |
| 1071 | kfree_type(struct uio, a_uio); |
| 1072 | } |
| 1073 | } |
| 1074 | |
| 1075 | /* |
| 1076 | * uio_addiov - add an iovec to the given uio_t. You may call this up to |
| 1077 | * the a_iovcount number that was passed to uio_create. This call will |
| 1078 | * increment the residual IO count as iovecs are added to the uio_t. |
| 1079 | * returns 0 if add was successful else non zero. |
| 1080 | */ |
| 1081 | int |
| 1082 | uio_addiov( uio_t a_uio, user_addr_t a_baseaddr, user_size_t a_length ) |
| 1083 | { |
| 1084 | int i; |
| 1085 | user_size_t resid; |
| 1086 | struct kern_iovec *kiovp; |
| 1087 | struct user_iovec *uiovp; |
| 1088 | |
| 1089 | if (__improbable(a_uio == NULL)) { |
| 1090 | #if DEBUG |
| 1091 | panic("invalid uio_t" ); |
| 1092 | #endif |
| 1093 | return -1; |
| 1094 | } |
| 1095 | |
| 1096 | if (__improbable(os_add_overflow(a_length, a_uio->uio_resid_64, &resid))) { |
| 1097 | #if DEBUG |
| 1098 | panic("invalid length %lu" , (unsigned long)a_length); |
| 1099 | #endif |
| 1100 | return -1; |
| 1101 | } |
| 1102 | |
| 1103 | if (UIO_IS_USER_SPACE(a_uio)) { |
| 1104 | uiovp = uio_uiovp(uio: a_uio); |
| 1105 | for (i = 0; i < a_uio->uio_max_iovs; i++) { |
| 1106 | if (uiovp[i].iov_len == 0 && |
| 1107 | uiovp[i].iov_base == 0) { |
| 1108 | uiovp[i].iov_len = a_length; |
| 1109 | uiovp[i].iov_base = a_baseaddr; |
| 1110 | a_uio->uio_iovcnt++; |
| 1111 | a_uio->uio_resid_64 = resid; |
| 1112 | return 0; |
| 1113 | } |
| 1114 | } |
| 1115 | } else { |
| 1116 | kiovp = uio_kiovp(uio: a_uio); |
| 1117 | for (i = 0; i < a_uio->uio_max_iovs; i++) { |
| 1118 | if (kiovp[i].iov_len == 0 && |
| 1119 | kiovp_get_base(kiovp: &kiovp[i]) == 0) { |
| 1120 | kiovp[i].iov_len = (u_int64_t)a_length; |
| 1121 | kiovp_set_base(kiovp: &kiovp[i], addr: (u_int64_t)a_baseaddr); |
| 1122 | a_uio->uio_iovcnt++; |
| 1123 | a_uio->uio_resid_64 = resid; |
| 1124 | return 0; |
| 1125 | } |
| 1126 | } |
| 1127 | } |
| 1128 | |
| 1129 | return -1; |
| 1130 | } |
| 1131 | |
| 1132 | /* |
| 1133 | * uio_getiov - get iovec data associated with the given uio_t. Use |
| 1134 | * a_index to iterate over each iovec (0 to (uio_iovcnt(uio_t) - 1)). |
| 1135 | * a_baseaddr_p and a_length_p may be NULL. |
| 1136 | * returns -1 when a_index is >= uio_t.uio_iovcnt or invalid uio_t. |
| 1137 | * returns 0 when data is returned. |
| 1138 | */ |
| 1139 | int |
| 1140 | uio_getiov( uio_t a_uio, |
| 1141 | int a_index, |
| 1142 | user_addr_t * a_baseaddr_p, |
| 1143 | user_size_t * a_length_p ) |
| 1144 | { |
| 1145 | struct kern_iovec *kiovp; |
| 1146 | struct user_iovec *uiovp; |
| 1147 | |
| 1148 | if (a_uio == NULL) { |
| 1149 | #if DEBUG |
| 1150 | panic("invalid uio_t" ); |
| 1151 | #endif /* DEBUG */ |
| 1152 | return -1; |
| 1153 | } |
| 1154 | if (a_index < 0 || a_index >= a_uio->uio_iovcnt) { |
| 1155 | return -1; |
| 1156 | } |
| 1157 | |
| 1158 | if (UIO_IS_USER_SPACE(a_uio)) { |
| 1159 | uiovp = uio_uiovp(uio: a_uio); |
| 1160 | |
| 1161 | if (a_baseaddr_p != NULL) { |
| 1162 | *a_baseaddr_p = uiovp[a_index].iov_base; |
| 1163 | } |
| 1164 | if (a_length_p != NULL) { |
| 1165 | *a_length_p = uiovp[a_index].iov_len; |
| 1166 | } |
| 1167 | } else { |
| 1168 | kiovp = uio_kiovp(uio: a_uio); |
| 1169 | |
| 1170 | if (a_baseaddr_p != NULL) { |
| 1171 | *a_baseaddr_p = (user_addr_t)kiovp_get_base(kiovp: &kiovp[a_index]); |
| 1172 | } |
| 1173 | if (a_length_p != NULL) { |
| 1174 | *a_length_p = (user_size_t)kiovp[a_index].iov_len; |
| 1175 | } |
| 1176 | } |
| 1177 | |
| 1178 | return 0; |
| 1179 | } |
| 1180 | |
| 1181 | /* |
| 1182 | * uio_calculateresid_user - runs through all iovecs associated with this |
| 1183 | * uio_t and calculates (and sets) the residual IO count. |
| 1184 | */ |
| 1185 | __private_extern__ int |
| 1186 | uio_calculateresid_user(uio_t __attribute((nonnull))a_uio) |
| 1187 | { |
| 1188 | int i; |
| 1189 | u_int64_t resid = 0; |
| 1190 | struct user_iovec *uiovp; |
| 1191 | |
| 1192 | a_uio->uio_iovcnt = a_uio->uio_max_iovs; |
| 1193 | uiovp = uio_uiovp(uio: a_uio); |
| 1194 | a_uio->uio_resid_64 = 0; |
| 1195 | for (i = 0; i < a_uio->uio_max_iovs; i++) { |
| 1196 | if (uiovp[i].iov_len != 0) { |
| 1197 | if (uiovp[i].iov_len > LONG_MAX) { |
| 1198 | return EINVAL; |
| 1199 | } |
| 1200 | resid += uiovp[i].iov_len; |
| 1201 | if (resid > LONG_MAX) { |
| 1202 | return EINVAL; |
| 1203 | } |
| 1204 | } |
| 1205 | } |
| 1206 | a_uio->uio_resid_64 = (user_size_t)resid; |
| 1207 | |
| 1208 | /* position to first non zero length iovec (4235922) */ |
| 1209 | while (a_uio->uio_iovcnt > 0 && uiovp->iov_len == 0) { |
| 1210 | a_uio->uio_iovcnt--; |
| 1211 | if (a_uio->uio_iovcnt > 0) { |
| 1212 | uiovp = uio_advance_user(uio: a_uio); |
| 1213 | } |
| 1214 | } |
| 1215 | |
| 1216 | return 0; |
| 1217 | } |
| 1218 | |
| 1219 | /* |
| 1220 | * uio_update_user - update the given uio_t for a_count of completed IO. |
| 1221 | * This call decrements the current iovec length and residual IO value |
| 1222 | * and increments the current iovec base address and offset value. |
| 1223 | * If the current iovec length is 0 then advance to the next |
| 1224 | * iovec (if any). |
| 1225 | * If the a_count passed in is 0, than only do the advancement |
| 1226 | * over any 0 length iovec's. |
| 1227 | */ |
| 1228 | static void |
| 1229 | uio_update_user(uio_t __attribute__((nonnull)) a_uio, user_size_t a_count) |
| 1230 | { |
| 1231 | struct user_iovec *uiovp; |
| 1232 | |
| 1233 | uiovp = uio_uiovp(uio: a_uio); |
| 1234 | |
| 1235 | /* |
| 1236 | * if a_count == 0, then we are asking to skip over |
| 1237 | * any empty iovs |
| 1238 | */ |
| 1239 | if (a_count) { |
| 1240 | if (a_count > uiovp->iov_len) { |
| 1241 | uiovp->iov_base += uiovp->iov_len; |
| 1242 | uiovp->iov_len = 0; |
| 1243 | } else { |
| 1244 | uiovp->iov_base += a_count; |
| 1245 | uiovp->iov_len -= a_count; |
| 1246 | } |
| 1247 | if (a_count > (user_size_t)a_uio->uio_resid_64) { |
| 1248 | a_uio->uio_offset += a_uio->uio_resid_64; |
| 1249 | a_uio->uio_resid_64 = 0; |
| 1250 | } else { |
| 1251 | a_uio->uio_offset += a_count; |
| 1252 | a_uio->uio_resid_64 -= a_count; |
| 1253 | } |
| 1254 | } |
| 1255 | /* |
| 1256 | * advance to next iovec if current one is totally consumed |
| 1257 | */ |
| 1258 | while (a_uio->uio_iovcnt > 0 && uiovp->iov_len == 0) { |
| 1259 | a_uio->uio_iovcnt--; |
| 1260 | if (a_uio->uio_iovcnt > 0) { |
| 1261 | uiovp = uio_advance_user(uio: a_uio); |
| 1262 | } |
| 1263 | } |
| 1264 | } |
| 1265 | |
| 1266 | /* |
| 1267 | * uio_update_sys - update the given uio_t for a_count of completed IO. |
| 1268 | * This call decrements the current iovec length and residual IO value |
| 1269 | * and increments the current iovec base address and offset value. |
| 1270 | * If the current iovec length is 0 then advance to the next |
| 1271 | * iovec (if any). |
| 1272 | * If the a_count passed in is 0, than only do the advancement |
| 1273 | * over any 0 length iovec's. |
| 1274 | */ |
| 1275 | static void |
| 1276 | uio_update_sys(uio_t __attribute__((nonnull)) a_uio, user_size_t a_count) |
| 1277 | { |
| 1278 | struct kern_iovec *kiovp; |
| 1279 | |
| 1280 | kiovp = uio_kiovp(uio: a_uio); |
| 1281 | |
| 1282 | /* |
| 1283 | * if a_count == 0, then we are asking to skip over |
| 1284 | * any empty iovs |
| 1285 | */ |
| 1286 | if (a_count) { |
| 1287 | u_int64_t prev_base = kiovp_get_base(kiovp); |
| 1288 | if (a_count > kiovp->iov_len) { |
| 1289 | u_int64_t len = kiovp->iov_len; |
| 1290 | kiovp->iov_len = 0; |
| 1291 | kiovp_set_base(kiovp, addr: prev_base + len); |
| 1292 | } else { |
| 1293 | kiovp->iov_len -= a_count; |
| 1294 | kiovp_set_base(kiovp, addr: prev_base + a_count); |
| 1295 | } |
| 1296 | if (a_count > (user_size_t)a_uio->uio_resid_64) { |
| 1297 | a_uio->uio_offset += a_uio->uio_resid_64; |
| 1298 | a_uio->uio_resid_64 = 0; |
| 1299 | } else { |
| 1300 | a_uio->uio_offset += a_count; |
| 1301 | a_uio->uio_resid_64 -= a_count; |
| 1302 | } |
| 1303 | } |
| 1304 | /* |
| 1305 | * advance to next iovec if current one is totally consumed |
| 1306 | */ |
| 1307 | while (a_uio->uio_iovcnt > 0 && kiovp->iov_len == 0) { |
| 1308 | a_uio->uio_iovcnt--; |
| 1309 | if (a_uio->uio_iovcnt > 0) { |
| 1310 | kiovp = uio_advance_sys(uio: a_uio); |
| 1311 | } |
| 1312 | } |
| 1313 | } |
| 1314 | |
| 1315 | /* |
| 1316 | * uio_update - update the given uio_t for a_count of completed IO. |
| 1317 | * This call decrements the current iovec length and residual IO value |
| 1318 | * and increments the current iovec base address and offset value. |
| 1319 | * If the current iovec length is 0 then advance to the next |
| 1320 | * iovec (if any). |
| 1321 | * If the a_count passed in is 0, than only do the advancement |
| 1322 | * over any 0 length iovec's. |
| 1323 | */ |
| 1324 | void |
| 1325 | uio_update(uio_t a_uio, user_size_t a_count) |
| 1326 | { |
| 1327 | if (a_uio == NULL || a_uio->uio_iovcnt < 1) { |
| 1328 | return; |
| 1329 | } |
| 1330 | |
| 1331 | if (UIO_IS_USER_SPACE(a_uio)) { |
| 1332 | uio_update_user(a_uio, a_count); |
| 1333 | } else { |
| 1334 | uio_update_sys(a_uio, a_count); |
| 1335 | } |
| 1336 | } |
| 1337 | |
| 1338 | /* |
| 1339 | * uio_duplicate - allocate a new uio and make a copy of the given uio_t. |
| 1340 | * may return NULL. |
| 1341 | */ |
| 1342 | uio_t |
| 1343 | uio_duplicate(uio_t uio) |
| 1344 | { |
| 1345 | uio_t new_uio; |
| 1346 | size_t n; |
| 1347 | struct kern_iovec *kiovp; |
| 1348 | struct user_iovec *uiovp; |
| 1349 | |
| 1350 | if (uio->uio_max_iovs < 0 || uio->uio_max_iovs > UIO_MAXIOV) { |
| 1351 | return NULL; |
| 1352 | } |
| 1353 | |
| 1354 | new_uio = kalloc_type(struct uio, Z_WAITOK | Z_ZERO | Z_NOFAIL); |
| 1355 | *new_uio = *uio; |
| 1356 | |
| 1357 | if (new_uio->uio_max_iovs > 0) { |
| 1358 | new_uio->uio_iovbase = uio_alloc_iov_array(a_spacetype: new_uio->uio_segflg, |
| 1359 | a_iovcount: (size_t)new_uio->uio_max_iovs); |
| 1360 | new_uio->uio_iovs = new_uio->uio_iovbase; |
| 1361 | |
| 1362 | n = UIO_SIZEOF_IOVS(new_uio->uio_iovcnt); |
| 1363 | bcopy(src: (const void *)uio->uio_iovs, dst: (void *)new_uio->uio_iovs, n); |
| 1364 | if (UIO_IS_SYS_SPACE(new_uio)) { |
| 1365 | struct kern_iovec *kiovp_old = uio_kiovp(uio); |
| 1366 | |
| 1367 | kiovp = uio_kiovp(uio: new_uio); |
| 1368 | |
| 1369 | for (n = 0; n < new_uio->uio_max_iovs; ++n) { |
| 1370 | kiovp_set_base(kiovp: &kiovp[n], |
| 1371 | addr: kiovp_get_base(kiovp: &kiovp_old[n])); |
| 1372 | } |
| 1373 | } else { |
| 1374 | uiovp = uio_uiovp(uio: new_uio); |
| 1375 | } |
| 1376 | |
| 1377 | /* advance to first nonzero iovec */ |
| 1378 | for (n = 0; n < new_uio->uio_max_iovs; ++n) { |
| 1379 | if (UIO_IS_USER_SPACE(new_uio)) { |
| 1380 | if (uiovp->iov_len != 0) { |
| 1381 | break; |
| 1382 | } |
| 1383 | |
| 1384 | uiovp = uio_advance_user(uio: new_uio); |
| 1385 | } else { |
| 1386 | if (kiovp->iov_len != 0) { |
| 1387 | break; |
| 1388 | } |
| 1389 | |
| 1390 | kiovp = uio_advance_sys(uio: new_uio); |
| 1391 | } |
| 1392 | } |
| 1393 | } else { |
| 1394 | new_uio->uio_iovs = NULL; |
| 1395 | } |
| 1396 | |
| 1397 | new_uio->uio_flags = UIO_FLAGS_WE_ALLOCED | UIO_FLAGS_INITED; |
| 1398 | #if DEBUG |
| 1399 | os_atomic_inc(&uio_t_count, relaxed); |
| 1400 | #endif |
| 1401 | |
| 1402 | return new_uio; |
| 1403 | } |
| 1404 | |
| 1405 | int |
| 1406 | copyin_user_iovec_array(user_addr_t uaddr, int spacetype, int count, struct user_iovec *dst) |
| 1407 | { |
| 1408 | size_t size_of_iovec = (spacetype == UIO_USERSPACE64 ? sizeof(struct user64_iovec) : sizeof(struct user32_iovec)); |
| 1409 | int error; |
| 1410 | int i; |
| 1411 | |
| 1412 | // copyin to the front of "dst", without regard for putting records in the right places |
| 1413 | error = copyin(uaddr, dst, count * size_of_iovec); |
| 1414 | if (error) { |
| 1415 | return error; |
| 1416 | } |
| 1417 | |
| 1418 | // now, unpack the entries in reverse order, so we don't overwrite anything |
| 1419 | for (i = count - 1; i >= 0; i--) { |
| 1420 | if (spacetype == UIO_USERSPACE64) { |
| 1421 | struct user64_iovec iovec = ((struct user64_iovec *)dst)[i]; |
| 1422 | dst[i].iov_base = (user_addr_t)iovec.iov_base; |
| 1423 | dst[i].iov_len = (user_size_t)iovec.iov_len; |
| 1424 | } else { |
| 1425 | struct user32_iovec iovec = ((struct user32_iovec *)dst)[i]; |
| 1426 | dst[i].iov_base = iovec.iov_base; |
| 1427 | dst[i].iov_len = iovec.iov_len; |
| 1428 | } |
| 1429 | } |
| 1430 | |
| 1431 | return 0; |
| 1432 | } |
| 1433 | |