bsd/vfs/vfs_vnops.c

   1 /*
   2  * Copyright (c) 2000-2001 Apple Computer, Inc. All rights reserved.
   3  *
   4  * @APPLE_LICENSE_HEADER_START@
   5  *
   6  * The contents of this file constitute Original Code as defined in and
   7  * are subject to the Apple Public Source License Version 1.1 (the
   8  * "License").  You may not use this file except in compliance with the
   9  * License.  Please obtain a copy of the License at
  10  * http://www.apple.com/publicsource and read it before using this file.
  11  *
  12  * This Original Code and all software distributed under the License are
  13  * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  14  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  15  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  16  * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT.  Please see the
  17  * License for the specific language governing rights and limitations
  18  * under the License.
  19  *
  20  * @APPLE_LICENSE_HEADER_END@
  21  */
  22 /* Copyright (c) 1995 NeXT Computer, Inc. All Rights Reserved */
  23 /*
  24  * Copyright (c) 1982, 1986, 1989, 1993
  25  *      The Regents of the University of California.  All rights reserved.
  26  * (c) UNIX System Laboratories, Inc.
  27  * All or some portions of this file are derived from material licensed
  28  * to the University of California by American Telephone and Telegraph
  29  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
  30  * the permission of UNIX System Laboratories, Inc.
  31  *
  32  * Redistribution and use in source and binary forms, with or without
  33  * modification, are permitted provided that the following conditions
  34  * are met:
  35  * 1. Redistributions of source code must retain the above copyright
  36  *    notice, this list of conditions and the following disclaimer.
  37  * 2. Redistributions in binary form must reproduce the above copyright
  38  *    notice, this list of conditions and the following disclaimer in the
  39  *    documentation and/or other materials provided with the distribution.
  40  * 3. All advertising materials mentioning features or use of this software
  41  *    must display the following acknowledgement:
  42  *      This product includes software developed by the University of
  43  *      California, Berkeley and its contributors.
  44  * 4. Neither the name of the University nor the names of its contributors
  45  *    may be used to endorse or promote products derived from this software
  46  *    without specific prior written permission.
  47  *
  48  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  49  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  50  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  51  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  52  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  53  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  54  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  55  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  56  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  57  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  58  * SUCH DAMAGE.
  59  *
  60  *      @(#)vfs_vnops.c 8.14 (Berkeley) 6/15/95
  61  *
  62  */
  63
  64 #include <sys/param.h>
  65 #include <sys/systm.h>
  66 #include <sys/kernel.h>
  67 #include <sys/file.h>
  68 #include <sys/stat.h>
  69 #include <sys/buf.h>
  70 #include <sys/proc.h>
  71 #include <sys/mount.h>
  72 #include <sys/namei.h>
  73 #include <sys/vnode.h>
  74 #include <sys/ioctl.h>
  75 #include <sys/tty.h>
  76 #include <sys/ubc.h>
  77
  78 struct  fileops vnops =
  79         { vn_read, vn_write, vn_ioctl, vn_select, vn_closefile };
  80
  81 /*
  82  * Common code for vnode open operations.
  83  * Check permissions, and call the VOP_OPEN or VOP_CREATE routine.
  84  */
  85 vn_open(ndp, fmode, cmode)
  86         register struct nameidata *ndp;
  87         int fmode, cmode;
  88 {
  89         register struct vnode *vp;
  90         register struct proc *p = ndp->ni_cnd.cn_proc;
  91         register struct ucred *cred = p->p_ucred;
  92         struct vattr vat;
  93         struct vattr *vap = &vat;
  94         int error;
  95
  96         if (fmode & O_CREAT) {
  97                 ndp->ni_cnd.cn_nameiop = CREATE;
  98                 ndp->ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF;
  99                 if ((fmode & O_EXCL) == 0)
 100                         ndp->ni_cnd.cn_flags |= FOLLOW;
 101                 if (error = namei(ndp))
 102                         return (error);
 103                 if (ndp->ni_vp == NULL) {
 104                         VATTR_NULL(vap);
 105                         vap->va_type = VREG;
 106                         vap->va_mode = cmode;
 107                         if (fmode & O_EXCL)
 108                                 vap->va_vaflags |= VA_EXCLUSIVE;
 109                         VOP_LEASE(ndp->ni_dvp, p, cred, LEASE_WRITE);
 110                         if (error = VOP_CREATE(ndp->ni_dvp, &ndp->ni_vp,
 111                             &ndp->ni_cnd, vap))
 112                                 return (error);
 113                         fmode &= ~O_TRUNC;
 114                         vp = ndp->ni_vp;
 115                 } else {
 116                         VOP_ABORTOP(ndp->ni_dvp, &ndp->ni_cnd);
 117                         if (ndp->ni_dvp == ndp->ni_vp)
 118                                 vrele(ndp->ni_dvp);
 119                         else
 120                                 vput(ndp->ni_dvp);
 121                         ndp->ni_dvp = NULL;
 122                         vp = ndp->ni_vp;
 123                         if (fmode & O_EXCL) {
 124                                 error = EEXIST;
 125                                 goto bad;
 126                         }
 127                         fmode &= ~O_CREAT;
 128                 }
 129         } else {
 130                 ndp->ni_cnd.cn_nameiop = LOOKUP;
 131                 ndp->ni_cnd.cn_flags = FOLLOW | LOCKLEAF;
 132                 if (error = namei(ndp))
 133                         return (error);
 134                 vp = ndp->ni_vp;
 135         }
 136         if (vp->v_type == VSOCK) {
 137                 error = EOPNOTSUPP;
 138                 goto bad;
 139         }
 140         if ((fmode & O_CREAT) == 0) {
 141                 if (fmode & FREAD && fmode & (FWRITE | O_TRUNC)) {
 142                         int err = 0;
 143                         if (vp->v_type == VDIR)
 144                                 err = EISDIR;
 145                         else
 146                                 err = vn_writechk(vp);
 147                         if (err && !(error = VOP_ACCESS(vp, VREAD, cred, p)))
 148                                 error = err;
 149                         if (error || (error = VOP_ACCESS(vp, VREAD|VWRITE,
 150                                                          cred, p)))
 151                                 goto bad;
 152                 } else if (fmode & FREAD) {
 153                         if ((error = VOP_ACCESS(vp, VREAD, cred, p)))
 154                                 goto bad;
 155                 } else if (fmode & (FWRITE | O_TRUNC)) {
 156                         if (vp->v_type == VDIR) {
 157                                 error = EISDIR;
 158                                 goto bad;
 159                         }
 160                         if ((error = vn_writechk(vp)) ||
 161                             (error = VOP_ACCESS(vp, VWRITE, cred, p)))
 162                                 goto bad;
 163                 }
 164         }
 165         if (fmode & O_TRUNC) {
 166                 VOP_UNLOCK(vp, 0, p);                           /* XXX */
 167                 VOP_LEASE(vp, p, cred, LEASE_WRITE);
 168                 (void)vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);  /* XXX */
 169                 VATTR_NULL(vap);
 170                 vap->va_size = 0;
 171                 if (error = VOP_SETATTR(vp, vap, cred, p))
 172                         goto bad;
 173         }
 174
 175 #if DIAGNOSTIC
 176         if (UBCINFOMISSING(vp))
 177                 panic("vn_open: ubc_info_init");
 178 #endif /* DIAGNOSTIC */
 179
 180         if (UBCINFOEXISTS(vp) && !ubc_hold(vp)) {
 181                 error = ENOENT;
 182                 goto bad;
 183         }
 184
 185         if (error = VOP_OPEN(vp, fmode, cred, p)) {
 186                 ubc_rele(vp);
 187                 goto bad;
 188         }
 189
 190         if (fmode & FWRITE)
 191                 if (++vp->v_writecount <= 0)
 192                         panic("vn_open: v_writecount");
 193         return (0);
 194 bad:
 195         vput(vp);
 196         return (error);
 197 }
 198
 199 /*
 200  * Check for write permissions on the specified vnode.
 201  * Prototype text segments cannot be written.
 202  */
 203 vn_writechk(vp)
 204         register struct vnode *vp;
 205 {
 206
 207         /*
 208          * If there's shared text associated with
 209          * the vnode, try to free it up once.  If
 210          * we fail, we can't allow writing.
 211          */
 212 #if 0
 213         /* XXXXX Not sure we need this */
 214         if (vp->v_flag & VTEXT)
 215                 return (ETXTBSY);
 216 #endif /* 0 */
 217         return (0);
 218 }
 219
 220 /*
 221  * Vnode close call
 222  */
 223 vn_close(vp, flags, cred, p)
 224         register struct vnode *vp;
 225         int flags;
 226         struct ucred *cred;
 227         struct proc *p;
 228 {
 229         int error;
 230
 231         if (flags & FWRITE)
 232                 vp->v_writecount--;
 233         error = VOP_CLOSE(vp, flags, cred, p);
 234         ubc_rele(vp);
 235         vrele(vp);
 236         return (error);
 237 }
 238
 239 /*
 240  * Package up an I/O request on a vnode into a uio and do it.
 241  */
 242 vn_rdwr(rw, vp, base, len, offset, segflg, ioflg, cred, aresid, p)
 243         enum uio_rw rw;
 244         struct vnode *vp;
 245         caddr_t base;
 246         int len;
 247         off_t offset;
 248         enum uio_seg segflg;
 249         int ioflg;
 250         struct ucred *cred;
 251         int *aresid;
 252         struct proc *p;
 253 {
 254         struct uio auio;
 255         struct iovec aiov;
 256         int error=0;
 257
 258          /* FIXME XXX */
 259         if ((ioflg & IO_NODELOCKED) == 0)
 260                 (void)vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
 261         auio.uio_iov = &aiov;
 262         auio.uio_iovcnt = 1;
 263         aiov.iov_base = base;
 264         aiov.iov_len = len;
 265         auio.uio_resid = len;
 266         auio.uio_offset = offset;
 267         auio.uio_segflg = segflg;
 268         auio.uio_rw = rw;
 269         auio.uio_procp = p;
 270
 271         if (rw == UIO_READ)
 272                 error = VOP_READ(vp, &auio, ioflg, cred);
 273          else
 274                 error = VOP_WRITE(vp, &auio, ioflg, cred);
 275
 276         if (aresid)
 277                 *aresid = auio.uio_resid;
 278         else
 279                 if (auio.uio_resid && error == 0)
 280                         error = EIO;
 281         if ((ioflg & IO_NODELOCKED) == 0)
 282                 VOP_UNLOCK(vp, 0, p);
 283         return (error);
 284 }
 285
 286 /*
 287  * File table vnode read routine.
 288  */
 289 vn_read(fp, uio, cred)
 290         struct file *fp;
 291         struct uio *uio;
 292         struct ucred *cred;
 293 {
 294         struct vnode *vp = (struct vnode *)fp->f_data;
 295         struct proc *p = uio->uio_procp;
 296         int error;
 297         off_t count;
 298
 299         VOP_LEASE(vp, p, cred, LEASE_READ);
 300         (void)vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
 301         uio->uio_offset = fp->f_offset;
 302         count = uio->uio_resid;
 303
 304         error = VOP_READ(vp, uio, (fp->f_flag & FNONBLOCK) ? IO_NDELAY : 0, cred);
 305
 306         fp->f_offset += count - uio->uio_resid;
 307         VOP_UNLOCK(vp, 0, p);
 308         return (error);
 309 }
 310
 311
 312 /*
 313  * File table vnode write routine.
 314  */
 315 vn_write(fp, uio, cred)
 316         struct file *fp;
 317         struct uio *uio;
 318         struct ucred *cred;
 319 {
 320         struct vnode *vp = (struct vnode *)fp->f_data;
 321         struct proc *p = uio->uio_procp;
 322         int error, ioflag = IO_UNIT;
 323         off_t count;
 324
 325         if (vp->v_type == VREG && (fp->f_flag & O_APPEND))
 326                 ioflag |= IO_APPEND;
 327         if (fp->f_flag & FNONBLOCK)
 328                 ioflag |= IO_NDELAY;
 329         if ((fp->f_flag & O_FSYNC) ||
 330                 (vp->v_mount && (vp->v_mount->mnt_flag & MNT_SYNCHRONOUS)))
 331                 ioflag |= IO_SYNC;
 332         VOP_LEASE(vp, p, cred, LEASE_WRITE);
 333         (void)vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
 334         uio->uio_offset = fp->f_offset;
 335         count = uio->uio_resid;
 336
 337         error = VOP_WRITE(vp, uio, ioflag, cred);
 338
 339         if (ioflag & IO_APPEND)
 340                 fp->f_offset = uio->uio_offset;
 341         else
 342                 fp->f_offset += count - uio->uio_resid;
 343         /*
 344          * Set the credentials on successful writes
 345          */
 346         if ((error == 0) && (vp->v_tag == VT_NFS) && (UBCINFOEXISTS(vp))) {
 347                 ubc_setcred(vp, p);
 348         }
 349
 350         VOP_UNLOCK(vp, 0, p);
 351         return (error);
 352 }
 353
 354 /*
 355  * File table vnode stat routine.
 356  */
 357 vn_stat(vp, sb, p)
 358         struct vnode *vp;
 359         register struct stat *sb;
 360         struct proc *p;
 361 {
 362         struct vattr vattr;
 363         register struct vattr *vap;
 364         int error;
 365         u_short mode;
 366
 367         vap = &vattr;
 368         error = VOP_GETATTR(vp, vap, p->p_ucred, p);
 369         if (error)
 370                 return (error);
 371         /*
 372          * Copy from vattr table
 373          */
 374         sb->st_dev = vap->va_fsid;
 375         sb->st_ino = vap->va_fileid;
 376         mode = vap->va_mode;
 377         switch (vp->v_type) {
 378         case VREG:
 379                 mode |= S_IFREG;
 380                 break;
 381         case VDIR:
 382                 mode |= S_IFDIR;
 383                 break;
 384         case VBLK:
 385                 mode |= S_IFBLK;
 386                 break;
 387         case VCHR:
 388                 mode |= S_IFCHR;
 389                 break;
 390         case VLNK:
 391                 mode |= S_IFLNK;
 392                 break;
 393         case VSOCK:
 394                 mode |= S_IFSOCK;
 395                 break;
 396         case VFIFO:
 397                 mode |= S_IFIFO;
 398                 break;
 399         default:
 400                 return (EBADF);
 401         };
 402         sb->st_mode = mode;
 403         sb->st_nlink = vap->va_nlink;
 404         sb->st_uid = vap->va_uid;
 405         sb->st_gid = vap->va_gid;
 406         sb->st_rdev = vap->va_rdev;
 407         sb->st_size = vap->va_size;
 408         sb->st_atimespec = vap->va_atime;
 409         sb->st_mtimespec = vap->va_mtime;
 410         sb->st_ctimespec = vap->va_ctime;
 411         sb->st_blksize = vap->va_blocksize;
 412         sb->st_flags = vap->va_flags;
 413         /* Do not give the generation number out to unpriviledged users */
 414         if (suser(p->p_ucred, &p->p_acflag))
 415                 sb->st_gen = 0;
 416         else
 417                 sb->st_gen = vap->va_gen;
 418         sb->st_blocks = vap->va_bytes / S_BLKSIZE;
 419         return (0);
 420 }
 421
 422 /*
 423  * File table vnode ioctl routine.
 424  */
 425 vn_ioctl(fp, com, data, p)
 426         struct file *fp;
 427         u_long com;
 428         caddr_t data;
 429         struct proc *p;
 430 {
 431         register struct vnode *vp = ((struct vnode *)fp->f_data);
 432         struct vattr vattr;
 433         int error;
 434         struct vnode *ttyvp;
 435
 436         switch (vp->v_type) {
 437
 438         case VREG:
 439         case VDIR:
 440                 if (com == FIONREAD) {
 441                         if (error = VOP_GETATTR(vp, &vattr, p->p_ucred, p))
 442                                 return (error);
 443                         *(int *)data = vattr.va_size - fp->f_offset;
 444                         return (0);
 445                 }
 446                 if (com == FIONBIO || com == FIOASYNC)  /* XXX */
 447                         return (0);                     /* XXX */
 448                 /* fall into ... */
 449
 450         default:
 451                 return (ENOTTY);
 452
 453         case VFIFO:
 454         case VCHR:
 455         case VBLK:
 456                 error = VOP_IOCTL(vp, com, data, fp->f_flag, p->p_ucred, p);
 457                 if (error == 0 && com == TIOCSCTTY) {
 458                         VREF(vp);
 459                         ttyvp = p->p_session->s_ttyvp;
 460                         p->p_session->s_ttyvp = vp;
 461                         if (ttyvp)
 462                                 vrele(ttyvp);
 463                 }
 464                 return (error);
 465         }
 466 }
 467
 468 /*
 469  * File table vnode select routine.
 470  */
 471 vn_select(fp, which, wql, p)
 472         struct file *fp;
 473         int which;
 474         void * wql;
 475         struct proc *p;
 476 {
 477
 478         return(VOP_SELECT(((struct vnode *)fp->f_data), which, fp->f_flag,
 479                 fp->f_cred, wql, p));
 480 }
 481
 482 /*
 483  * Check that the vnode is still valid, and if so
 484  * acquire requested lock.
 485  */
 486 int
 487 vn_lock(vp, flags, p)
 488         struct vnode *vp;
 489         int flags;
 490         struct proc *p;
 491 {
 492         int error;
 493
 494         do {
 495                 if ((flags & LK_INTERLOCK) == 0)
 496                         simple_lock(&vp->v_interlock);
 497                 if (vp->v_flag & VXLOCK) {
 498                         while (vp->v_flag & VXLOCK) {
 499                                 vp->v_flag |= VXWANT;
 500                                 simple_unlock(&vp->v_interlock);
 501                                 (void)tsleep((caddr_t)vp, PINOD, "vn_lock", 0);
 502                         }
 503                         error = ENOENT;
 504                 } else {
 505                         error = VOP_LOCK(vp, flags | LK_INTERLOCK, p);
 506                         if (error == 0)
 507                                 return (error);
 508                 }
 509                 flags &= ~LK_INTERLOCK;
 510         } while (flags & LK_RETRY);
 511         return (error);
 512 }
 513
 514 /*
 515  * File table vnode close routine.
 516  */
 517 vn_closefile(fp, p)
 518         struct file *fp;
 519         struct proc *p;
 520 {
 521
 522         return (vn_close(((struct vnode *)fp->f_data), fp->f_flag,
 523                 fp->f_cred, p));
 524 }