vfs_vnops.c revision 24206
1/* 2 * Copyright (c) 1982, 1986, 1989, 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 * @(#)vfs_vnops.c 8.2 (Berkeley) 1/21/94 39 * $Id: vfs_vnops.c,v 1.33 1997/03/23 03:36:38 bde Exp $ 40 */ 41 42#include <sys/param.h> 43#include <sys/systm.h> 44#include <sys/kernel.h> 45#include <sys/fcntl.h> 46#include <sys/file.h> 47#include <sys/stat.h> 48#include <sys/buf.h> 49#include <sys/proc.h> 50#include <sys/mount.h> 51#include <sys/namei.h> 52#include <sys/vnode.h> 53#include <sys/filio.h> 54#include <sys/ttycom.h> 55 56#include <vm/vm.h> 57#include <vm/vm_param.h> 58#include <vm/vm_object.h> 59#include <vm/vnode_pager.h> 60 61static int vn_closefile __P((struct file *fp, struct proc *p)); 62static int vn_ioctl __P((struct file *fp, int com, caddr_t data, 63 struct proc *p)); 64static int vn_read __P((struct file *fp, struct uio *uio, 65 struct ucred *cred)); 66static int vn_select __P((struct file *fp, int which, struct proc *p)); 67static int vn_write __P((struct file *fp, struct uio *uio, 68 struct ucred *cred)); 69 70struct fileops vnops = 71 { vn_read, vn_write, vn_ioctl, vn_select, vn_closefile }; 72 73/* 74 * Common code for vnode open operations. 75 * Check permissions, and call the VOP_OPEN or VOP_CREATE routine. 76 */ 77int 78vn_open(ndp, fmode, cmode) 79 register struct nameidata *ndp; 80 int fmode, cmode; 81{ 82 register struct vnode *vp; 83 register struct proc *p = ndp->ni_cnd.cn_proc; 84 register struct ucred *cred = p->p_ucred; 85 struct vattr vat; 86 struct vattr *vap = &vat; 87 int error; 88 89 if (fmode & O_CREAT) { 90 ndp->ni_cnd.cn_nameiop = CREATE; 91 ndp->ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF; 92 if ((fmode & O_EXCL) == 0) 93 ndp->ni_cnd.cn_flags |= FOLLOW; 94 error = namei(ndp); 95 if (error) 96 return (error); 97 if (ndp->ni_vp == NULL) { 98 VATTR_NULL(vap); 99 vap->va_type = VREG; 100 vap->va_mode = cmode; 101 if (fmode & O_EXCL) 102 vap->va_vaflags |= VA_EXCLUSIVE; 103 VOP_LEASE(ndp->ni_dvp, p, cred, LEASE_WRITE); 104 if (error = VOP_CREATE(ndp->ni_dvp, &ndp->ni_vp, 105 &ndp->ni_cnd, vap)) 106 return (error); 107 fmode &= ~O_TRUNC; 108 vp = ndp->ni_vp; 109 } else { 110 VOP_ABORTOP(ndp->ni_dvp, &ndp->ni_cnd); 111 if (ndp->ni_dvp == ndp->ni_vp) 112 vrele(ndp->ni_dvp); 113 else 114 vput(ndp->ni_dvp); 115 ndp->ni_dvp = NULL; 116 vp = ndp->ni_vp; 117 if (fmode & O_EXCL) { 118 error = EEXIST; 119 goto bad; 120 } 121 fmode &= ~O_CREAT; 122 } 123 } else { 124 ndp->ni_cnd.cn_nameiop = LOOKUP; 125 ndp->ni_cnd.cn_flags = FOLLOW | LOCKLEAF; 126 error = namei(ndp); 127 if (error) 128 return (error); 129 vp = ndp->ni_vp; 130 } 131 if (vp->v_type == VSOCK) { 132 error = EOPNOTSUPP; 133 goto bad; 134 } 135 if ((fmode & O_CREAT) == 0) { 136 if (fmode & FREAD) { 137 error = VOP_ACCESS(vp, VREAD, cred, p); 138 if (error) 139 goto bad; 140 } 141 if (fmode & (FWRITE | O_TRUNC)) { 142 if (vp->v_type == VDIR) { 143 error = EISDIR; 144 goto bad; 145 } 146 error = vn_writechk(vp); 147 if (error) 148 goto bad; 149 error = VOP_ACCESS(vp, VWRITE, cred, p); 150 if (error) 151 goto bad; 152 } 153 } 154 if (fmode & O_TRUNC) { 155 VOP_UNLOCK(vp, 0, p); /* XXX */ 156 VOP_LEASE(vp, p, cred, LEASE_WRITE); 157 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p); /* XXX */ 158 VATTR_NULL(vap); 159 vap->va_size = 0; 160 error = VOP_SETATTR(vp, vap, cred, p); 161 if (error) 162 goto bad; 163 } 164 error = VOP_OPEN(vp, fmode, cred, p); 165 if (error) 166 goto bad; 167 /* 168 * Make sure that a VM object is created for VMIO support. 169 */ 170 if (vp->v_type == VREG) { 171 if ((error = vfs_object_create(vp, p, cred, 1)) != 0) 172 goto bad; 173 } 174 175 if (fmode & FWRITE) 176 vp->v_writecount++; 177 return (0); 178bad: 179 vput(vp); 180 return (error); 181} 182 183/* 184 * Check for write permissions on the specified vnode. 185 * Prototype text segments cannot be written. 186 */ 187int 188vn_writechk(vp) 189 register struct vnode *vp; 190{ 191 192 /* 193 * If there's shared text associated with 194 * the vnode, try to free it up once. If 195 * we fail, we can't allow writing. 196 */ 197 if (vp->v_flag & VTEXT) 198 return (ETXTBSY); 199 return (0); 200} 201 202/* 203 * Vnode close call 204 */ 205int 206vn_close(vp, flags, cred, p) 207 register struct vnode *vp; 208 int flags; 209 struct ucred *cred; 210 struct proc *p; 211{ 212 int error; 213 214 if (flags & FWRITE) 215 vp->v_writecount--; 216 error = VOP_CLOSE(vp, flags, cred, p); 217 vrele(vp); 218 return (error); 219} 220 221/* 222 * Package up an I/O request on a vnode into a uio and do it. 223 */ 224int 225vn_rdwr(rw, vp, base, len, offset, segflg, ioflg, cred, aresid, p) 226 enum uio_rw rw; 227 struct vnode *vp; 228 caddr_t base; 229 int len; 230 off_t offset; 231 enum uio_seg segflg; 232 int ioflg; 233 struct ucred *cred; 234 int *aresid; 235 struct proc *p; 236{ 237 struct uio auio; 238 struct iovec aiov; 239 int error; 240 241 if ((ioflg & IO_NODELOCKED) == 0) 242 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p); 243 auio.uio_iov = &aiov; 244 auio.uio_iovcnt = 1; 245 aiov.iov_base = base; 246 aiov.iov_len = len; 247 auio.uio_resid = len; 248 auio.uio_offset = offset; 249 auio.uio_segflg = segflg; 250 auio.uio_rw = rw; 251 auio.uio_procp = p; 252 if (rw == UIO_READ) { 253 error = VOP_READ(vp, &auio, ioflg, cred); 254 } else { 255 error = VOP_WRITE(vp, &auio, ioflg, cred); 256 } 257 if (aresid) 258 *aresid = auio.uio_resid; 259 else 260 if (auio.uio_resid && error == 0) 261 error = EIO; 262 if ((ioflg & IO_NODELOCKED) == 0) 263 VOP_UNLOCK(vp, 0, p); 264 return (error); 265} 266 267/* 268 * File table vnode read routine. 269 */ 270static int 271vn_read(fp, uio, cred) 272 struct file *fp; 273 struct uio *uio; 274 struct ucred *cred; 275{ 276 struct vnode *vp = (struct vnode *)fp->f_data; 277 struct proc *p = uio->uio_procp; 278 int count, error; 279 int flag, seq; 280 281 VOP_LEASE(vp, p, cred, LEASE_READ); 282 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p); 283 uio->uio_offset = fp->f_offset; 284 count = uio->uio_resid; 285 flag = 0; 286 if (fp->f_flag & FNONBLOCK) 287 flag |= IO_NDELAY; 288 289 /* 290 * Sequential read heuristic. 291 * If we have been doing sequential input, 292 * a rewind operation doesn't turn off 293 * sequential input mode. 294 */ 295 if (((fp->f_offset == 0) && (fp->f_seqcount > 0)) || 296 (fp->f_offset == fp->f_nextread)) { 297 int tmpseq = fp->f_seqcount; 298 /* 299 * XXX we assume that the filesystem block size is 300 * the default. Not true, but still gives us a pretty 301 * good indicator of how sequential the read operations 302 * are. 303 */ 304 tmpseq += ((count + BKVASIZE - 1) / BKVASIZE); 305 if (tmpseq >= CHAR_MAX) 306 tmpseq = CHAR_MAX; 307 fp->f_seqcount = tmpseq; 308 flag |= (fp->f_seqcount << 16); 309 } else { 310 if (fp->f_seqcount > 1) 311 fp->f_seqcount = 1; 312 else 313 fp->f_seqcount = 0; 314 } 315 316 error = VOP_READ(vp, uio, flag, cred); 317 fp->f_offset += count - uio->uio_resid; 318 fp->f_nextread = fp->f_offset; 319 VOP_UNLOCK(vp, 0, p); 320 return (error); 321} 322 323/* 324 * File table vnode write routine. 325 */ 326static int 327vn_write(fp, uio, cred) 328 struct file *fp; 329 struct uio *uio; 330 struct ucred *cred; 331{ 332 struct vnode *vp = (struct vnode *)fp->f_data; 333 struct proc *p = uio->uio_procp; 334 int count, error, ioflag = IO_UNIT; 335 336 if (vp->v_type == VREG && (fp->f_flag & O_APPEND)) 337 ioflag |= IO_APPEND; 338 if (fp->f_flag & FNONBLOCK) 339 ioflag |= IO_NDELAY; 340 if ((fp->f_flag & O_FSYNC) || 341 (vp->v_mount && (vp->v_mount->mnt_flag & MNT_SYNCHRONOUS))) 342 ioflag |= IO_SYNC; 343 VOP_LEASE(vp, p, cred, LEASE_WRITE); 344 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p); 345 uio->uio_offset = fp->f_offset; 346 count = uio->uio_resid; 347 error = VOP_WRITE(vp, uio, ioflag, cred); 348 if (ioflag & IO_APPEND) 349 fp->f_offset = uio->uio_offset; 350 else 351 fp->f_offset += count - uio->uio_resid; 352 VOP_UNLOCK(vp, 0, p); 353 return (error); 354} 355 356/* 357 * File table vnode stat routine. 358 */ 359int 360vn_stat(vp, sb, p) 361 struct vnode *vp; 362 register struct stat *sb; 363 struct proc *p; 364{ 365 struct vattr vattr; 366 register struct vattr *vap; 367 int error; 368 u_short mode; 369 370 vap = &vattr; 371 error = VOP_GETATTR(vp, vap, p->p_ucred, p); 372 if (error) 373 return (error); 374 /* 375 * Copy from vattr table 376 */ 377 sb->st_dev = vap->va_fsid; 378 sb->st_ino = vap->va_fileid; 379 mode = vap->va_mode; 380 switch (vp->v_type) { 381 case VREG: 382 mode |= S_IFREG; 383 break; 384 case VDIR: 385 mode |= S_IFDIR; 386 break; 387 case VBLK: 388 mode |= S_IFBLK; 389 break; 390 case VCHR: 391 mode |= S_IFCHR; 392 break; 393 case VLNK: 394 mode |= S_IFLNK; 395 break; 396 case VSOCK: 397 mode |= S_IFSOCK; 398 break; 399 case VFIFO: 400 mode |= S_IFIFO; 401 break; 402 default: 403 return (EBADF); 404 }; 405 sb->st_mode = mode; 406 sb->st_nlink = vap->va_nlink; 407 sb->st_uid = vap->va_uid; 408 sb->st_gid = vap->va_gid; 409 sb->st_rdev = vap->va_rdev; 410 sb->st_size = vap->va_size; 411 sb->st_atimespec = vap->va_atime; 412 sb->st_mtimespec = vap->va_mtime; 413 sb->st_ctimespec = vap->va_ctime; 414 sb->st_blksize = vap->va_blocksize; 415 sb->st_flags = vap->va_flags; 416 if (p->p_ucred->cr_uid != 0) 417 sb->st_gen = 0; 418 else 419 sb->st_gen = vap->va_gen; 420 421#if (S_BLKSIZE == 512) 422 /* Optimize this case */ 423 sb->st_blocks = vap->va_bytes >> 9; 424#else 425 sb->st_blocks = vap->va_bytes / S_BLKSIZE; 426#endif 427 return (0); 428} 429 430/* 431 * File table vnode ioctl routine. 432 */ 433static int 434vn_ioctl(fp, com, data, p) 435 struct file *fp; 436 int com; 437 caddr_t data; 438 struct proc *p; 439{ 440 register struct vnode *vp = ((struct vnode *)fp->f_data); 441 struct vattr vattr; 442 int error; 443 444 switch (vp->v_type) { 445 446 case VREG: 447 case VDIR: 448 if (com == FIONREAD) { 449 error = VOP_GETATTR(vp, &vattr, p->p_ucred, p); 450 if (error) 451 return (error); 452 *(int *)data = vattr.va_size - fp->f_offset; 453 return (0); 454 } 455 if (com == FIONBIO || com == FIOASYNC) /* XXX */ 456 return (0); /* XXX */ 457 /* fall into ... */ 458 459 default: 460 return (ENOTTY); 461 462 case VFIFO: 463 case VCHR: 464 case VBLK: 465 error = VOP_IOCTL(vp, com, data, fp->f_flag, p->p_ucred, p); 466 if (error == 0 && com == TIOCSCTTY) { 467 468 /* Do nothing if reassigning same control tty */ 469 if (p->p_session->s_ttyvp == vp) 470 return (0); 471 472 /* Get rid of reference to old control tty */ 473 if (p->p_session->s_ttyvp) 474 vrele(p->p_session->s_ttyvp); 475 476 p->p_session->s_ttyvp = vp; 477 VREF(vp); 478 } 479 return (error); 480 } 481} 482 483/* 484 * File table vnode select routine. 485 */ 486static int 487vn_select(fp, which, p) 488 struct file *fp; 489 int which; 490 struct proc *p; 491{ 492 493 return (VOP_SELECT(((struct vnode *)fp->f_data), which, fp->f_flag, 494 fp->f_cred, p)); 495} 496 497/* 498 * File table vnode close routine. 499 */ 500static int 501vn_closefile(fp, p) 502 struct file *fp; 503 struct proc *p; 504{ 505 506 return (vn_close(((struct vnode *)fp->f_data), fp->f_flag, 507 fp->f_cred, p)); 508} 509 510/* 511 * Check that the vnode is still valid, and if so 512 * acquire requested lock. 513 */ 514int 515vn_lock(vp, flags, p) 516 struct vnode *vp; 517 int flags; 518 struct proc *p; 519{ 520 int error; 521 522 do { 523 if ((flags & LK_INTERLOCK) == 0) { 524 simple_lock(&vp->v_interlock); 525 } 526 if (vp->v_flag & VXLOCK) { 527 vp->v_flag |= VXWANT; 528 simple_unlock(&vp->v_interlock); 529 tsleep((caddr_t)vp, PINOD, "vn_lock", 0); 530 error = ENOENT; 531 } else { 532 error = VOP_LOCK(vp, flags | LK_INTERLOCK, p); 533 if (error == 0) 534 return (error); 535 } 536 flags &= ~LK_INTERLOCK; 537 } while (flags & LK_RETRY); 538 return (error); 539} 540