1 /* 2 * Copyright (c) 1982, 1986, 1991, 1993 3 * The Regents of the University of California. All rights reserved. 4 * (c) UNIX System Laboratories, Inc. 5 * All or some portions of this file are derived from material licensed 6 * to the University of California by American Telephone and Telegraph 7 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 8 * the permission of UNIX System Laboratories, Inc. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 3. All advertising materials mentioning features or use of this software 19 * must display the following acknowledgement: 20 * This product includes software developed by the University of 21 * California, Berkeley and its contributors. 22 * 4. Neither the name of the University nor the names of its contributors 23 * may be used to endorse or promote products derived from this software 24 * without specific prior written permission. 25 * 26 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 27 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 28 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 29 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 30 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 31 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 32 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 33 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 34 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 35 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 36 * SUCH DAMAGE. 37 * 38 * @(#)kern_subr.c 8.3 (Berkeley) 1/21/94 39 * $FreeBSD: src/sys/kern/kern_subr.c,v 1.31.2.2 2002/04/21 08:09:37 bde Exp $ 40 * $DragonFly: src/sys/kern/kern_subr.c,v 1.27 2007/01/29 20:44:02 tgen Exp $ 41 */ 42 43 #include "opt_ddb.h" 44 45 #include <sys/param.h> 46 #include <sys/systm.h> 47 #include <sys/kernel.h> 48 #include <sys/proc.h> 49 #include <sys/malloc.h> 50 #include <sys/lock.h> 51 #include <sys/resourcevar.h> 52 #include <sys/sysctl.h> 53 #include <sys/uio.h> 54 #include <sys/vnode.h> 55 #include <sys/sfbuf.h> 56 #include <sys/thread2.h> 57 #include <machine/limits.h> 58 59 #include <vm/vm.h> 60 #include <vm/vm_page.h> 61 #include <vm/vm_map.h> 62 63 SYSCTL_INT(_kern, KERN_IOV_MAX, iov_max, CTLFLAG_RD, NULL, UIO_MAXIOV, 64 "Maximum number of elements in an I/O vector; sysconf(_SC_IOV_MAX)"); 65 66 /* 67 * UIO_READ: copy the kernelspace cp to the user or kernelspace UIO 68 * UIO_WRITE: copy the user or kernelspace UIO to the kernelspace cp 69 * 70 * For userspace UIO's, uio_td must be the current thread. 71 */ 72 int 73 uiomove(caddr_t cp, int n, struct uio *uio) 74 { 75 struct iovec *iov; 76 u_int cnt; 77 int error = 0; 78 int save = 0; 79 int baseticks = ticks; 80 81 KASSERT(uio->uio_rw == UIO_READ || uio->uio_rw == UIO_WRITE, 82 ("uiomove: mode")); 83 KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == curthread, 84 ("uiomove proc")); 85 86 if (curproc) { 87 save = curproc->p_flag & P_DEADLKTREAT; 88 curproc->p_flag |= P_DEADLKTREAT; 89 } 90 91 while (n > 0 && uio->uio_resid) { 92 iov = uio->uio_iov; 93 cnt = iov->iov_len; 94 if (cnt == 0) { 95 uio->uio_iov++; 96 uio->uio_iovcnt--; 97 continue; 98 } 99 if (cnt > n) 100 cnt = n; 101 102 switch (uio->uio_segflg) { 103 104 case UIO_USERSPACE: 105 if (ticks - baseticks >= hogticks) { 106 uio_yield(); 107 baseticks = ticks; 108 } 109 if (uio->uio_rw == UIO_READ) 110 error = copyout(cp, iov->iov_base, cnt); 111 else 112 error = copyin(iov->iov_base, cp, cnt); 113 if (error) 114 break; 115 break; 116 117 case UIO_SYSSPACE: 118 if (uio->uio_rw == UIO_READ) 119 bcopy((caddr_t)cp, iov->iov_base, cnt); 120 else 121 bcopy(iov->iov_base, (caddr_t)cp, cnt); 122 break; 123 case UIO_NOCOPY: 124 break; 125 } 126 iov->iov_base += cnt; 127 iov->iov_len -= cnt; 128 uio->uio_resid -= cnt; 129 uio->uio_offset += cnt; 130 cp += cnt; 131 n -= cnt; 132 } 133 if (curproc) 134 curproc->p_flag = (curproc->p_flag & ~P_DEADLKTREAT) | save; 135 return (error); 136 } 137 /* 138 * Wrapper for uiomove() that validates the arguments against a known-good 139 * kernel buffer. Currently, uiomove accepts a signed (n) argument, which 140 * is almost definitely a bad thing, so we catch that here as well. We 141 * return a runtime failure, but it might be desirable to generate a runtime 142 * assertion failure instead. 143 */ 144 int 145 uiomove_frombuf(void *buf, int buflen, struct uio *uio) 146 { 147 unsigned int offset, n; 148 149 if (uio->uio_offset < 0 || uio->uio_resid < 0 || 150 (offset = uio->uio_offset) != uio->uio_offset) 151 return (EINVAL); 152 if (buflen <= 0 || offset >= buflen) 153 return (0); 154 if ((n = buflen - offset) > INT_MAX) 155 return (EINVAL); 156 return (uiomove((char *)buf + offset, n, uio)); 157 } 158 159 160 int 161 uiomoveco(caddr_t cp, int n, struct uio *uio, struct vm_object *obj) 162 { 163 struct iovec *iov; 164 u_int cnt; 165 int error; 166 int baseticks = ticks; 167 168 KASSERT(uio->uio_rw == UIO_READ || uio->uio_rw == UIO_WRITE, 169 ("uiomoveco: mode")); 170 KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == curthread, 171 ("uiomoveco proc")); 172 173 while (n > 0 && uio->uio_resid) { 174 iov = uio->uio_iov; 175 cnt = iov->iov_len; 176 if (cnt == 0) { 177 uio->uio_iov++; 178 uio->uio_iovcnt--; 179 continue; 180 } 181 if (cnt > n) 182 cnt = n; 183 184 switch (uio->uio_segflg) { 185 186 case UIO_USERSPACE: 187 if (ticks - baseticks >= hogticks) { 188 uio_yield(); 189 baseticks = ticks; 190 } 191 if (uio->uio_rw == UIO_READ) { 192 error = copyout(cp, iov->iov_base, cnt); 193 } else { 194 error = copyin(iov->iov_base, cp, cnt); 195 } 196 if (error) 197 return (error); 198 break; 199 200 case UIO_SYSSPACE: 201 if (uio->uio_rw == UIO_READ) 202 bcopy((caddr_t)cp, iov->iov_base, cnt); 203 else 204 bcopy(iov->iov_base, (caddr_t)cp, cnt); 205 break; 206 case UIO_NOCOPY: 207 break; 208 } 209 iov->iov_base += cnt; 210 iov->iov_len -= cnt; 211 uio->uio_resid -= cnt; 212 uio->uio_offset += cnt; 213 cp += cnt; 214 n -= cnt; 215 } 216 return (0); 217 } 218 219 /* 220 * Give next character to user as result of read. 221 */ 222 int 223 ureadc(int c, struct uio *uio) 224 { 225 struct iovec *iov; 226 227 again: 228 if (uio->uio_iovcnt == 0 || uio->uio_resid == 0) 229 panic("ureadc"); 230 iov = uio->uio_iov; 231 if (iov->iov_len == 0) { 232 uio->uio_iovcnt--; 233 uio->uio_iov++; 234 goto again; 235 } 236 switch (uio->uio_segflg) { 237 238 case UIO_USERSPACE: 239 if (subyte(iov->iov_base, c) < 0) 240 return (EFAULT); 241 break; 242 243 case UIO_SYSSPACE: 244 *iov->iov_base = c; 245 break; 246 247 case UIO_NOCOPY: 248 break; 249 } 250 iov->iov_base++; 251 iov->iov_len--; 252 uio->uio_resid--; 253 uio->uio_offset++; 254 return (0); 255 } 256 257 /* 258 * General routine to allocate a hash table. Make the hash table size a 259 * power of 2 greater or equal to the number of elements requested, and 260 * store the masking value in *hashmask. 261 */ 262 void * 263 hashinit(int elements, struct malloc_type *type, u_long *hashmask) 264 { 265 long hashsize; 266 LIST_HEAD(generic, generic) *hashtbl; 267 int i; 268 269 if (elements <= 0) 270 panic("hashinit: bad elements"); 271 for (hashsize = 2; hashsize < elements; hashsize <<= 1) 272 continue; 273 hashtbl = kmalloc((u_long)hashsize * sizeof(*hashtbl), type, M_WAITOK); 274 for (i = 0; i < hashsize; i++) 275 LIST_INIT(&hashtbl[i]); 276 *hashmask = hashsize - 1; 277 return (hashtbl); 278 } 279 280 static int primes[] = { 1, 13, 31, 61, 127, 251, 509, 761, 1021, 1531, 2039, 281 2557, 3067, 3583, 4093, 4603, 5119, 5623, 6143, 6653, 282 7159, 7673, 8191, 12281, 16381, 24571, 32749 }; 283 #define NPRIMES (sizeof(primes) / sizeof(primes[0])) 284 285 /* 286 * General routine to allocate a prime number sized hash table. 287 */ 288 void * 289 phashinit(int elements, struct malloc_type *type, u_long *nentries) 290 { 291 long hashsize; 292 LIST_HEAD(generic, generic) *hashtbl; 293 int i; 294 295 if (elements <= 0) 296 panic("phashinit: bad elements"); 297 for (i = 1, hashsize = primes[1]; hashsize <= elements;) { 298 i++; 299 if (i == NPRIMES) 300 break; 301 hashsize = primes[i]; 302 } 303 hashsize = primes[i - 1]; 304 hashtbl = kmalloc((u_long)hashsize * sizeof(*hashtbl), type, M_WAITOK); 305 for (i = 0; i < hashsize; i++) 306 LIST_INIT(&hashtbl[i]); 307 *nentries = hashsize; 308 return (hashtbl); 309 } 310 311 /* 312 * Copyin an iovec. If the iovec array fits, use the preallocated small 313 * iovec structure. If it is too big, dynamically allocate an iovec array 314 * of sufficient size. 315 * 316 * MPSAFE 317 */ 318 int 319 iovec_copyin(struct iovec *uiov, struct iovec **kiov, struct iovec *siov, 320 size_t iov_cnt, int *iov_len) 321 { 322 struct iovec *iovp; 323 int error, i; 324 325 if (iov_cnt > UIO_MAXIOV) 326 return EMSGSIZE; 327 if (iov_cnt > UIO_SMALLIOV) { 328 MALLOC(*kiov, struct iovec *, sizeof(struct iovec) * iov_cnt, 329 M_IOV, M_WAITOK); 330 } else { 331 *kiov = siov; 332 } 333 error = copyin(uiov, *kiov, iov_cnt * sizeof(struct iovec)); 334 if (error == 0) { 335 *iov_len = 0; 336 for (i = 0, iovp = *kiov; i < iov_cnt; i++, iovp++) { 337 /* 338 * Check for both *iov_len overflows and out of 339 * range iovp->iov_len's. We limit to the 340 * capabilities of signed integers. 341 */ 342 if (*iov_len + (int)iovp->iov_len < *iov_len) 343 error = EINVAL; 344 *iov_len += (int)iovp->iov_len; 345 } 346 } 347 if (error) 348 iovec_free(kiov, siov); 349 return (error); 350 } 351 352 353 /* 354 * Copyright (c) 2004 Alan L. Cox <alc@cs.rice.edu> 355 * Copyright (c) 1982, 1986, 1991, 1993 356 * The Regents of the University of California. All rights reserved. 357 * (c) UNIX System Laboratories, Inc. 358 * All or some portions of this file are derived from material licensed 359 * to the University of California by American Telephone and Telegraph 360 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 361 * the permission of UNIX System Laboratories, Inc. 362 * 363 * Redistribution and use in source and binary forms, with or without 364 * modification, are permitted provided that the following conditions 365 * are met: 366 * 1. Redistributions of source code must retain the above copyright 367 * notice, this list of conditions and the following disclaimer. 368 * 2. Redistributions in binary form must reproduce the above copyright 369 * notice, this list of conditions and the following disclaimer in the 370 * documentation and/or other materials provided with the distribution. 371 * 4. Neither the name of the University nor the names of its contributors 372 * may be used to endorse or promote products derived from this software 373 * without specific prior written permission. 374 * 375 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 376 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 377 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 378 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 379 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 380 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 381 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 382 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 383 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 384 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 385 * SUCH DAMAGE. 386 * 387 * @(#)kern_subr.c 8.3 (Berkeley) 1/21/94 388 * $FreeBSD: src/sys/i386/i386/uio_machdep.c,v 1.1 2004/03/21 20:28:36 alc Exp $ 389 */ 390 391 /* 392 * Implement uiomove(9) from physical memory using sf_bufs to reduce 393 * the creation and destruction of ephemeral mappings. 394 */ 395 int 396 uiomove_fromphys(vm_page_t *ma, vm_offset_t offset, int n, struct uio *uio) 397 { 398 struct sf_buf *sf; 399 struct thread *td = curthread; 400 struct iovec *iov; 401 void *cp; 402 vm_offset_t page_offset; 403 vm_page_t m; 404 size_t cnt; 405 int error = 0; 406 int save = 0; 407 408 KASSERT(uio->uio_rw == UIO_READ || uio->uio_rw == UIO_WRITE, 409 ("uiomove_fromphys: mode")); 410 KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == curthread, 411 ("uiomove_fromphys proc")); 412 413 crit_enter(); 414 save = td->td_flags & TDF_DEADLKTREAT; 415 td->td_flags |= TDF_DEADLKTREAT; 416 crit_exit(); 417 418 while (n > 0 && uio->uio_resid) { 419 iov = uio->uio_iov; 420 cnt = iov->iov_len; 421 if (cnt == 0) { 422 uio->uio_iov++; 423 uio->uio_iovcnt--; 424 continue; 425 } 426 if (cnt > n) 427 cnt = n; 428 page_offset = offset & PAGE_MASK; 429 cnt = min(cnt, PAGE_SIZE - page_offset); 430 m = ma[offset >> PAGE_SHIFT]; 431 sf = sf_buf_alloc(m, SFB_CPUPRIVATE); 432 cp = (char *)sf_buf_kva(sf) + page_offset; 433 switch (uio->uio_segflg) { 434 case UIO_USERSPACE: 435 /* 436 * note: removed uioyield (it was the wrong place to 437 * put it). 438 */ 439 if (uio->uio_rw == UIO_READ) 440 error = copyout(cp, iov->iov_base, cnt); 441 else 442 error = copyin(iov->iov_base, cp, cnt); 443 if (error) { 444 sf_buf_free(sf); 445 goto out; 446 } 447 break; 448 case UIO_SYSSPACE: 449 if (uio->uio_rw == UIO_READ) 450 bcopy(cp, iov->iov_base, cnt); 451 else 452 bcopy(iov->iov_base, cp, cnt); 453 break; 454 case UIO_NOCOPY: 455 break; 456 } 457 sf_buf_free(sf); 458 iov->iov_base = (char *)iov->iov_base + cnt; 459 iov->iov_len -= cnt; 460 uio->uio_resid -= cnt; 461 uio->uio_offset += cnt; 462 offset += cnt; 463 n -= cnt; 464 } 465 out: 466 if (save == 0) { 467 crit_enter(); 468 td->td_flags &= ~TDF_DEADLKTREAT; 469 crit_exit(); 470 } 471 return (error); 472 } 473 474