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 1.59 1998/06/27 06:43:09 phk Exp $ 40 */ 41 42 #include <sys/param.h> 43 #include <sys/systm.h> 44 #include <sys/fcntl.h> 45 #include <sys/file.h> 46 #include <sys/stat.h> 47 #include <sys/proc.h> 48 #include <sys/mount.h> 49 #include <sys/namei.h> 50 #include <sys/vnode.h> 51 #include <sys/filio.h> 52 #include <sys/ttycom.h> 53 54 static int vn_closefile __P((struct file *fp, struct proc *p)); 55 static int vn_ioctl __P((struct file *fp, u_long com, caddr_t data, 56 struct proc *p)); 57 static int vn_read __P((struct file *fp, struct uio *uio, 58 struct ucred *cred)); 59 static int vn_poll __P((struct file *fp, int events, struct ucred *cred, 60 struct proc *p)); 61 static int vn_write __P((struct file *fp, struct uio *uio, 62 struct ucred *cred)); 63 64 struct fileops vnops = 65 { vn_read, vn_write, vn_ioctl, vn_poll, vn_closefile }; 66 67 /* 68 * Common code for vnode open operations. 69 * Check permissions, and call the VOP_OPEN or VOP_CREATE routine. 70 */ 71 int 72 vn_open(ndp, fmode, cmode) 73 register struct nameidata *ndp; 74 int fmode, cmode; 75 { 76 register struct vnode *vp; 77 register struct proc *p = ndp->ni_cnd.cn_proc; 78 register struct ucred *cred = p->p_ucred; 79 struct vattr vat; 80 struct vattr *vap = &vat; 81 int mode, error; 82 83 if (fmode & O_CREAT) { 84 ndp->ni_cnd.cn_nameiop = CREATE; 85 ndp->ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF; 86 if ((fmode & O_EXCL) == 0 && (fmode & O_NOFOLLOW) == 0) 87 ndp->ni_cnd.cn_flags |= FOLLOW; 88 error = namei(ndp); 89 if (error) 90 return (error); 91 if (ndp->ni_vp == NULL) { 92 VATTR_NULL(vap); 93 vap->va_type = VREG; 94 vap->va_mode = cmode; 95 if (fmode & O_EXCL) 96 vap->va_vaflags |= VA_EXCLUSIVE; 97 VOP_LEASE(ndp->ni_dvp, p, cred, LEASE_WRITE); 98 error = VOP_CREATE(ndp->ni_dvp, &ndp->ni_vp, 99 &ndp->ni_cnd, vap); 100 vput(ndp->ni_dvp); 101 if (error) 102 return (error); 103 ASSERT_VOP_UNLOCKED(ndp->ni_dvp, "create"); 104 ASSERT_VOP_LOCKED(ndp->ni_vp, "create"); 105 fmode &= ~O_TRUNC; 106 vp = ndp->ni_vp; 107 } else { 108 VOP_ABORTOP(ndp->ni_dvp, &ndp->ni_cnd); 109 if (ndp->ni_dvp == ndp->ni_vp) 110 vrele(ndp->ni_dvp); 111 else 112 vput(ndp->ni_dvp); 113 ndp->ni_dvp = NULL; 114 vp = ndp->ni_vp; 115 if (fmode & O_EXCL) { 116 error = EEXIST; 117 goto bad; 118 } 119 fmode &= ~O_CREAT; 120 } 121 } else { 122 ndp->ni_cnd.cn_nameiop = LOOKUP; 123 ndp->ni_cnd.cn_flags = 124 ((fmode & O_NOFOLLOW) ? NOFOLLOW : FOLLOW) | LOCKLEAF; 125 error = namei(ndp); 126 if (error) 127 return (error); 128 vp = ndp->ni_vp; 129 } 130 if (vp->v_type == VLNK) { 131 error = EMLINK; 132 goto bad; 133 } 134 if (vp->v_type == VSOCK) { 135 error = EOPNOTSUPP; 136 goto bad; 137 } 138 if ((fmode & O_CREAT) == 0) { 139 mode = 0; 140 if (fmode & (FWRITE | O_TRUNC)) { 141 if (vp->v_type == VDIR) { 142 error = EISDIR; 143 goto bad; 144 } 145 error = vn_writechk(vp); 146 if (error) 147 goto bad; 148 mode |= VWRITE; 149 } 150 if (fmode & FREAD) 151 mode |= VREAD; 152 if (mode) { 153 error = VOP_ACCESS(vp, mode, cred, p); 154 if (error) 155 goto bad; 156 } 157 } 158 if (fmode & O_TRUNC) { 159 VOP_UNLOCK(vp, 0, p); /* XXX */ 160 VOP_LEASE(vp, p, cred, LEASE_WRITE); 161 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p); /* XXX */ 162 VATTR_NULL(vap); 163 vap->va_size = 0; 164 error = VOP_SETATTR(vp, vap, cred, p); 165 if (error) 166 goto bad; 167 } 168 error = VOP_OPEN(vp, fmode, cred, p); 169 if (error) 170 goto bad; 171 /* 172 * Make sure that a VM object is created for VMIO support. 173 */ 174 if (vp->v_type == VREG) { 175 if ((error = vfs_object_create(vp, p, cred, 1)) != 0) 176 goto bad; 177 } 178 179 if (fmode & FWRITE) 180 vp->v_writecount++; 181 return (0); 182 bad: 183 vput(vp); 184 return (error); 185 } 186 187 /* 188 * Check for write permissions on the specified vnode. 189 * Prototype text segments cannot be written. 190 */ 191 int 192 vn_writechk(vp) 193 register struct vnode *vp; 194 { 195 196 /* 197 * If there's shared text associated with 198 * the vnode, try to free it up once. If 199 * we fail, we can't allow writing. 200 */ 201 if (vp->v_flag & VTEXT) 202 return (ETXTBSY); 203 return (0); 204 } 205 206 /* 207 * Vnode close call 208 */ 209 int 210 vn_close(vp, flags, cred, p) 211 register struct vnode *vp; 212 int flags; 213 struct ucred *cred; 214 struct proc *p; 215 { 216 int error; 217 218 if (flags & FWRITE) 219 vp->v_writecount--; 220 error = VOP_CLOSE(vp, flags, cred, p); 221 vrele(vp); 222 return (error); 223 } 224 225 /* 226 * Package up an I/O request on a vnode into a uio and do it. 227 */ 228 int 229 vn_rdwr(rw, vp, base, len, offset, segflg, ioflg, cred, aresid, p) 230 enum uio_rw rw; 231 struct vnode *vp; 232 caddr_t base; 233 int len; 234 off_t offset; 235 enum uio_seg segflg; 236 int ioflg; 237 struct ucred *cred; 238 int *aresid; 239 struct proc *p; 240 { 241 struct uio auio; 242 struct iovec aiov; 243 int error; 244 245 if ((ioflg & IO_NODELOCKED) == 0) 246 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p); 247 auio.uio_iov = &aiov; 248 auio.uio_iovcnt = 1; 249 aiov.iov_base = base; 250 aiov.iov_len = len; 251 auio.uio_resid = len; 252 auio.uio_offset = offset; 253 auio.uio_segflg = segflg; 254 auio.uio_rw = rw; 255 auio.uio_procp = p; 256 if (rw == UIO_READ) { 257 error = VOP_READ(vp, &auio, ioflg, cred); 258 } else { 259 error = VOP_WRITE(vp, &auio, ioflg, cred); 260 } 261 if (aresid) 262 *aresid = auio.uio_resid; 263 else 264 if (auio.uio_resid && error == 0) 265 error = EIO; 266 if ((ioflg & IO_NODELOCKED) == 0) 267 VOP_UNLOCK(vp, 0, p); 268 return (error); 269 } 270 271 /* 272 * File table vnode read routine. 273 */ 274 static int 275 vn_read(fp, uio, cred) 276 struct file *fp; 277 struct uio *uio; 278 struct ucred *cred; 279 { 280 struct vnode *vp = (struct vnode *)fp->f_data; 281 struct proc *p = uio->uio_procp; 282 int count, error; 283 int flag; 284 285 VOP_LEASE(vp, p, cred, LEASE_READ); 286 vn_lock(vp, LK_SHARED | LK_NOPAUSE | LK_RETRY, p); 287 if (uio->uio_offset == -1) 288 uio->uio_offset = fp->f_offset; 289 count = uio->uio_resid; 290 flag = 0; 291 if (fp->f_flag & FNONBLOCK) 292 flag |= IO_NDELAY; 293 294 /* 295 * Sequential read heuristic. 296 * If we have been doing sequential input, 297 * a rewind operation doesn't turn off 298 * sequential input mode. 299 */ 300 if (((fp->f_offset == 0) && (fp->f_seqcount > 0)) || 301 (fp->f_offset == fp->f_nextread)) { 302 int tmpseq = fp->f_seqcount; 303 /* 304 * XXX we assume that the filesystem block size is 305 * the default. Not true, but still gives us a pretty 306 * good indicator of how sequential the read operations 307 * are. 308 */ 309 tmpseq += ((count + BKVASIZE - 1) / BKVASIZE); 310 if (tmpseq >= 127) 311 tmpseq = 127; 312 fp->f_seqcount = tmpseq; 313 flag |= (fp->f_seqcount << 16); 314 } else { 315 if (fp->f_seqcount > 1) 316 fp->f_seqcount = 1; 317 else 318 fp->f_seqcount = 0; 319 } 320 321 error = VOP_READ(vp, uio, flag, cred); 322 fp->f_offset += count - uio->uio_resid; 323 fp->f_nextread = fp->f_offset; 324 VOP_UNLOCK(vp, 0, p); 325 return (error); 326 } 327 328 /* 329 * File table vnode write routine. 330 */ 331 static int 332 vn_write(fp, uio, cred) 333 struct file *fp; 334 struct uio *uio; 335 struct ucred *cred; 336 { 337 struct vnode *vp = (struct vnode *)fp->f_data; 338 struct proc *p = uio->uio_procp; 339 int count, error, ioflag = IO_UNIT; 340 341 if (uio->uio_offset == -1 && vp->v_type == VREG && (fp->f_flag & O_APPEND)) 342 ioflag |= IO_APPEND; 343 if (fp->f_flag & FNONBLOCK) 344 ioflag |= IO_NDELAY; 345 if ((fp->f_flag & O_FSYNC) || 346 (vp->v_mount && (vp->v_mount->mnt_flag & MNT_SYNCHRONOUS))) 347 ioflag |= IO_SYNC; 348 VOP_LEASE(vp, p, cred, LEASE_WRITE); 349 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p); 350 uio->uio_offset = fp->f_offset; 351 count = uio->uio_resid; 352 error = VOP_WRITE(vp, uio, ioflag, cred); 353 if (ioflag & IO_APPEND) 354 fp->f_offset = uio->uio_offset; 355 else 356 fp->f_offset += count - uio->uio_resid; 357 VOP_UNLOCK(vp, 0, p); 358 return (error); 359 } 360 361 /* 362 * File table vnode stat routine. 363 */ 364 int 365 vn_stat(vp, sb, p) 366 struct vnode *vp; 367 register struct stat *sb; 368 struct proc *p; 369 { 370 struct vattr vattr; 371 register struct vattr *vap; 372 int error; 373 u_short mode; 374 375 vap = &vattr; 376 error = VOP_GETATTR(vp, vap, p->p_ucred, p); 377 if (error) 378 return (error); 379 /* 380 * Copy from vattr table 381 */ 382 sb->st_dev = vap->va_fsid; 383 sb->st_ino = vap->va_fileid; 384 mode = vap->va_mode; 385 switch (vap->va_type) { 386 case VREG: 387 mode |= S_IFREG; 388 break; 389 case VDIR: 390 mode |= S_IFDIR; 391 break; 392 case VBLK: 393 mode |= S_IFBLK; 394 break; 395 case VCHR: 396 mode |= S_IFCHR; 397 break; 398 case VLNK: 399 mode |= S_IFLNK; 400 /* This is a cosmetic change, symlinks do not have a mode. */ 401 if (vp->v_mount->mnt_flag & MNT_NOSYMFOLLOW) 402 sb->st_mode &= ~ACCESSPERMS; /* 0000 */ 403 else 404 sb->st_mode |= ACCESSPERMS; /* 0777 */ 405 break; 406 case VSOCK: 407 mode |= S_IFSOCK; 408 break; 409 case VFIFO: 410 mode |= S_IFIFO; 411 break; 412 default: 413 return (EBADF); 414 }; 415 sb->st_mode = mode; 416 sb->st_nlink = vap->va_nlink; 417 sb->st_uid = vap->va_uid; 418 sb->st_gid = vap->va_gid; 419 sb->st_rdev = vap->va_rdev; 420 sb->st_size = vap->va_size; 421 sb->st_atimespec = vap->va_atime; 422 sb->st_mtimespec = vap->va_mtime; 423 sb->st_ctimespec = vap->va_ctime; 424 sb->st_blksize = vap->va_blocksize; 425 sb->st_flags = vap->va_flags; 426 if (p->p_ucred->cr_uid != 0) 427 sb->st_gen = 0; 428 else 429 sb->st_gen = vap->va_gen; 430 431 #if (S_BLKSIZE == 512) 432 /* Optimize this case */ 433 sb->st_blocks = vap->va_bytes >> 9; 434 #else 435 sb->st_blocks = vap->va_bytes / S_BLKSIZE; 436 #endif 437 return (0); 438 } 439 440 /* 441 * File table vnode ioctl routine. 442 */ 443 static int 444 vn_ioctl(fp, com, data, p) 445 struct file *fp; 446 u_long com; 447 caddr_t data; 448 struct proc *p; 449 { 450 register struct vnode *vp = ((struct vnode *)fp->f_data); 451 struct vattr vattr; 452 int error; 453 454 switch (vp->v_type) { 455 456 case VREG: 457 case VDIR: 458 if (com == FIONREAD) { 459 error = VOP_GETATTR(vp, &vattr, p->p_ucred, p); 460 if (error) 461 return (error); 462 *(int *)data = vattr.va_size - fp->f_offset; 463 return (0); 464 } 465 if (com == FIONBIO || com == FIOASYNC) /* XXX */ 466 return (0); /* XXX */ 467 /* fall into ... */ 468 469 default: 470 #if 0 471 return (ENOTTY); 472 #endif 473 case VFIFO: 474 case VCHR: 475 case VBLK: 476 error = VOP_IOCTL(vp, com, data, fp->f_flag, p->p_ucred, p); 477 if (error == 0 && com == TIOCSCTTY) { 478 479 /* Do nothing if reassigning same control tty */ 480 if (p->p_session->s_ttyvp == vp) 481 return (0); 482 483 /* Get rid of reference to old control tty */ 484 if (p->p_session->s_ttyvp) 485 vrele(p->p_session->s_ttyvp); 486 487 p->p_session->s_ttyvp = vp; 488 VREF(vp); 489 } 490 return (error); 491 } 492 } 493 494 /* 495 * File table vnode poll routine. 496 */ 497 static int 498 vn_poll(fp, events, cred, p) 499 struct file *fp; 500 int events; 501 struct ucred *cred; 502 struct proc *p; 503 { 504 505 return (VOP_POLL(((struct vnode *)fp->f_data), events, cred, p)); 506 } 507 508 /* 509 * Check that the vnode is still valid, and if so 510 * acquire requested lock. 511 */ 512 int 513 vn_lock(vp, flags, p) 514 struct vnode *vp; 515 int flags; 516 struct proc *p; 517 { 518 int error; 519 520 do { 521 if ((flags & LK_INTERLOCK) == 0) 522 simple_lock(&vp->v_interlock); 523 if (vp->v_flag & VXLOCK) { 524 vp->v_flag |= VXWANT; 525 simple_unlock(&vp->v_interlock); 526 tsleep((caddr_t)vp, PINOD, "vn_lock", 0); 527 error = ENOENT; 528 } else { 529 error = VOP_LOCK(vp, flags | LK_NOPAUSE | LK_INTERLOCK, p); 530 if (error == 0) 531 return (error); 532 } 533 flags &= ~LK_INTERLOCK; 534 } while (flags & LK_RETRY); 535 return (error); 536 } 537 538 /* 539 * File table vnode close routine. 540 */ 541 static int 542 vn_closefile(fp, p) 543 struct file *fp; 544 struct proc *p; 545 { 546 547 return (vn_close(((struct vnode *)fp->f_data), fp->f_flag, 548 fp->f_cred, p)); 549 } 550