xref: /netbsd-src/sys/kern/vfs_vnops.c (revision 2a399c6883d870daece976daec6ffa7bb7f934ce)
1 /*	$NetBSD: vfs_vnops.c,v 1.23 1997/10/10 02:09:34 fvdl Exp $	*/
2 
3 /*
4  * Copyright (c) 1982, 1986, 1989, 1993
5  *	The Regents of the University of California.  All rights reserved.
6  * (c) UNIX System Laboratories, Inc.
7  * All or some portions of this file are derived from material licensed
8  * to the University of California by American Telephone and Telegraph
9  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
10  * the permission of UNIX System Laboratories, Inc.
11  *
12  * Redistribution and use in source and binary forms, with or without
13  * modification, are permitted provided that the following conditions
14  * are met:
15  * 1. Redistributions of source code must retain the above copyright
16  *    notice, this list of conditions and the following disclaimer.
17  * 2. Redistributions in binary form must reproduce the above copyright
18  *    notice, this list of conditions and the following disclaimer in the
19  *    documentation and/or other materials provided with the distribution.
20  * 3. All advertising materials mentioning features or use of this software
21  *    must display the following acknowledgement:
22  *	This product includes software developed by the University of
23  *	California, Berkeley and its contributors.
24  * 4. Neither the name of the University nor the names of its contributors
25  *    may be used to endorse or promote products derived from this software
26  *    without specific prior written permission.
27  *
28  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
29  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
30  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
31  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
32  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
33  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
34  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
35  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
36  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
37  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
38  * SUCH DAMAGE.
39  *
40  *	@(#)vfs_vnops.c	8.5 (Berkeley) 12/8/94
41  */
42 
43 #include <sys/param.h>
44 #include <sys/systm.h>
45 #include <sys/kernel.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/ioctl.h>
54 #include <sys/tty.h>
55 #include <sys/poll.h>
56 
57 #include <vm/vm.h>
58 
59 struct 	fileops vnops =
60 	{ vn_read, vn_write, vn_ioctl, vn_poll, vn_closefile };
61 
62 /*
63  * Common code for vnode open operations.
64  * Check permissions, and call the VOP_OPEN or VOP_CREATE routine.
65  */
66 int
67 vn_open(ndp, fmode, cmode)
68 	register struct nameidata *ndp;
69 	int fmode, cmode;
70 {
71 	register struct vnode *vp;
72 	register struct proc *p = ndp->ni_cnd.cn_proc;
73 	register struct ucred *cred = p->p_ucred;
74 	struct vattr va;
75 	int error;
76 
77 	if (fmode & O_CREAT) {
78 		ndp->ni_cnd.cn_nameiop = CREATE;
79 		ndp->ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF;
80 		if ((fmode & O_EXCL) == 0)
81 			ndp->ni_cnd.cn_flags |= FOLLOW;
82 		if ((error = namei(ndp)) != 0)
83 			return (error);
84 		if (ndp->ni_vp == NULL) {
85 			VATTR_NULL(&va);
86 			va.va_type = VREG;
87 			va.va_mode = cmode;
88 			VOP_LEASE(ndp->ni_dvp, p, cred, LEASE_WRITE);
89 			error = VOP_CREATE(ndp->ni_dvp, &ndp->ni_vp,
90 					   &ndp->ni_cnd, &va);
91 			if (error)
92 				return (error);
93 			fmode &= ~O_TRUNC;
94 			vp = ndp->ni_vp;
95 		} else {
96 			VOP_ABORTOP(ndp->ni_dvp, &ndp->ni_cnd);
97 			if (ndp->ni_dvp == ndp->ni_vp)
98 				vrele(ndp->ni_dvp);
99 			else
100 				vput(ndp->ni_dvp);
101 			ndp->ni_dvp = NULL;
102 			vp = ndp->ni_vp;
103 			if (fmode & O_EXCL) {
104 				error = EEXIST;
105 				goto bad;
106 			}
107 			fmode &= ~O_CREAT;
108 		}
109 	} else {
110 		ndp->ni_cnd.cn_nameiop = LOOKUP;
111 		ndp->ni_cnd.cn_flags = FOLLOW | LOCKLEAF;
112 		if ((error = namei(ndp)) != 0)
113 			return (error);
114 		vp = ndp->ni_vp;
115 	}
116 	if (vp->v_type == VSOCK) {
117 		error = EOPNOTSUPP;
118 		goto bad;
119 	}
120 	if ((fmode & O_CREAT) == 0) {
121 		if (fmode & FREAD) {
122 			if ((error = VOP_ACCESS(vp, VREAD, cred, p)) != 0)
123 				goto bad;
124 		}
125 		if (fmode & (FWRITE | O_TRUNC)) {
126 			if (vp->v_type == VDIR) {
127 				error = EISDIR;
128 				goto bad;
129 			}
130 			if ((error = vn_writechk(vp)) != 0 ||
131 			    (error = VOP_ACCESS(vp, VWRITE, cred, p)) != 0)
132 				goto bad;
133 		}
134 	}
135 	if (fmode & O_TRUNC) {
136 		VOP_UNLOCK(vp);				/* XXX */
137 		VOP_LEASE(vp, p, cred, LEASE_WRITE);
138 		VOP_LOCK(vp);				/* XXX */
139 		VATTR_NULL(&va);
140 		va.va_size = 0;
141 		if ((error = VOP_SETATTR(vp, &va, cred, p)) != 0)
142 			goto bad;
143 	}
144 	if ((error = VOP_OPEN(vp, fmode, cred, p)) != 0)
145 		goto bad;
146 	if (fmode & FWRITE)
147 		vp->v_writecount++;
148 	return (0);
149 bad:
150 	vput(vp);
151 	return (error);
152 }
153 
154 /*
155  * Check for write permissions on the specified vnode.
156  * The read-only status of the file system is checked.
157  * Also, prototype text segments cannot be written.
158  */
159 int
160 vn_writechk(vp)
161 	register struct vnode *vp;
162 {
163 
164 	/*
165 	 * Disallow write attempts on read-only file systems;
166 	 * unless the file is a socket or a block or character
167 	 * device resident on the file system.
168 	 */
169 	if (vp->v_mount->mnt_flag & MNT_RDONLY) {
170 		switch (vp->v_type) {
171 		case VREG: case VDIR: case VLNK:
172 			return (EROFS);
173 		case VNON: case VCHR: case VSOCK:
174 		case VFIFO: case VBAD: case VBLK:
175 			break;
176 		}
177 	}
178 	/*
179 	 * If there's shared text associated with
180 	 * the vnode, try to free it up once.  If
181 	 * we fail, we can't allow writing.
182 	 */
183 	if ((vp->v_flag & VTEXT) && !vnode_pager_uncache(vp))
184 		return (ETXTBSY);
185 	return (0);
186 }
187 
188 /*
189  * Vnode close call
190  */
191 int
192 vn_close(vp, flags, cred, p)
193 	register struct vnode *vp;
194 	int flags;
195 	struct ucred *cred;
196 	struct proc *p;
197 {
198 	int error;
199 
200 	if (flags & FWRITE)
201 		vp->v_writecount--;
202 	error = VOP_CLOSE(vp, flags, cred, p);
203 	vrele(vp);
204 	return (error);
205 }
206 
207 /*
208  * Package up an I/O request on a vnode into a uio and do it.
209  */
210 int
211 vn_rdwr(rw, vp, base, len, offset, segflg, ioflg, cred, aresid, p)
212 	enum uio_rw rw;
213 	struct vnode *vp;
214 	caddr_t base;
215 	int len;
216 	off_t offset;
217 	enum uio_seg segflg;
218 	int ioflg;
219 	struct ucred *cred;
220 	int *aresid;
221 	struct proc *p;
222 {
223 	struct uio auio;
224 	struct iovec aiov;
225 	int error;
226 
227 	if ((ioflg & IO_NODELOCKED) == 0)
228 		VOP_LOCK(vp);
229 	auio.uio_iov = &aiov;
230 	auio.uio_iovcnt = 1;
231 	aiov.iov_base = base;
232 	aiov.iov_len = len;
233 	auio.uio_resid = len;
234 	auio.uio_offset = offset;
235 	auio.uio_segflg = segflg;
236 	auio.uio_rw = rw;
237 	auio.uio_procp = p;
238 	if (rw == UIO_READ) {
239 		error = VOP_READ(vp, &auio, ioflg, cred);
240 	} else {
241 		error = VOP_WRITE(vp, &auio, ioflg, cred);
242 	}
243 	if (aresid)
244 		*aresid = auio.uio_resid;
245 	else
246 		if (auio.uio_resid && error == 0)
247 			error = EIO;
248 	if ((ioflg & IO_NODELOCKED) == 0)
249 		VOP_UNLOCK(vp);
250 	return (error);
251 }
252 
253 int
254 vn_readdir(fp, buf, segflg, count, done, p, cookies, ncookies)
255 	struct file *fp;
256 	char *buf;
257 	int segflg, *done, ncookies;
258 	u_int count;
259 	struct proc *p;
260 	off_t *cookies;
261 {
262 	struct vnode *vp = (struct vnode *)fp->f_data;
263 	struct iovec aiov;
264 	struct uio auio;
265 	int error, eofflag;
266 
267 unionread:
268 	if (vp->v_type != VDIR)
269 		return (EINVAL);
270 	aiov.iov_base = buf;
271 	aiov.iov_len = count;
272 	auio.uio_iov = &aiov;
273 	auio.uio_iovcnt = 1;
274 	auio.uio_rw = UIO_READ;
275 	auio.uio_segflg = segflg;
276 	auio.uio_procp = p;
277 	auio.uio_resid = count;
278 	VOP_LOCK(vp);
279 	auio.uio_offset = fp->f_offset;
280 	error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, (off_t *)cookies,
281 		    ncookies);
282 	fp->f_offset = auio.uio_offset;
283 	VOP_UNLOCK(vp);
284 	if (error)
285 		return (error);
286 
287 #ifdef UNION
288 {
289 	extern int (**union_vnodeop_p) __P((void *));
290 	extern struct vnode *union_dircache __P((struct vnode *));
291 
292 	if (count == auio.uio_resid && (vp->v_op == union_vnodeop_p)) {
293 		struct vnode *lvp;
294 
295 		lvp = union_dircache(vp);
296 		if (lvp != NULLVP) {
297 			struct vattr va;
298 
299 			/*
300 			 * If the directory is opaque,
301 			 * then don't show lower entries
302 			 */
303 			error = VOP_GETATTR(vp, &va, fp->f_cred, p);
304 			if (va.va_flags & OPAQUE) {
305 				vput(lvp);
306 				lvp = NULL;
307 			}
308 		}
309 
310 		if (lvp != NULLVP) {
311 			error = VOP_OPEN(lvp, FREAD, fp->f_cred, p);
312 			VOP_UNLOCK(lvp);
313 
314 			if (error) {
315 				vrele(lvp);
316 				return (error);
317 			}
318 			fp->f_data = (caddr_t) lvp;
319 			fp->f_offset = 0;
320 			error = vn_close(vp, FREAD, fp->f_cred, p);
321 			if (error)
322 				return (error);
323 			vp = lvp;
324 			goto unionread;
325 		}
326 	}
327 }
328 #endif /* UNION */
329 
330 	if (count == auio.uio_resid && (vp->v_flag & VROOT) &&
331 	    (vp->v_mount->mnt_flag & MNT_UNION)) {
332 		struct vnode *tvp = vp;
333 		vp = vp->v_mount->mnt_vnodecovered;
334 		VREF(vp);
335 		fp->f_data = (caddr_t) vp;
336 		fp->f_offset = 0;
337 		vrele(tvp);
338 		goto unionread;
339 	}
340 	*done = count - auio.uio_resid;
341 	return error;
342 }
343 
344 /*
345  * File table vnode read routine.
346  */
347 int
348 vn_read(fp, uio, cred)
349 	struct file *fp;
350 	struct uio *uio;
351 	struct ucred *cred;
352 {
353 	register struct vnode *vp = (struct vnode *)fp->f_data;
354 	int count, error;
355 
356 	VOP_LEASE(vp, uio->uio_procp, cred, LEASE_READ);
357 	VOP_LOCK(vp);
358 	uio->uio_offset = fp->f_offset;
359 	count = uio->uio_resid;
360 	error = VOP_READ(vp, uio, (fp->f_flag & FNONBLOCK) ? IO_NDELAY : 0,
361 		cred);
362 	fp->f_offset += count - uio->uio_resid;
363 	VOP_UNLOCK(vp);
364 	return (error);
365 }
366 
367 /*
368  * File table vnode write routine.
369  */
370 int
371 vn_write(fp, uio, cred)
372 	struct file *fp;
373 	struct uio *uio;
374 	struct ucred *cred;
375 {
376 	register struct vnode *vp = (struct vnode *)fp->f_data;
377 	int count, error, ioflag = IO_UNIT;
378 
379 	if (vp->v_type == VREG && (fp->f_flag & O_APPEND))
380 		ioflag |= IO_APPEND;
381 	if (fp->f_flag & FNONBLOCK)
382 		ioflag |= IO_NDELAY;
383 	VOP_LEASE(vp, uio->uio_procp, cred, LEASE_WRITE);
384 	VOP_LOCK(vp);
385 	uio->uio_offset = fp->f_offset;
386 	count = uio->uio_resid;
387 	error = VOP_WRITE(vp, uio, ioflag, cred);
388 	if (ioflag & IO_APPEND)
389 		fp->f_offset = uio->uio_offset;
390 	else
391 		fp->f_offset += count - uio->uio_resid;
392 	VOP_UNLOCK(vp);
393 	return (error);
394 }
395 
396 /*
397  * File table vnode stat routine.
398  */
399 int
400 vn_stat(vp, sb, p)
401 	struct vnode *vp;
402 	register struct stat *sb;
403 	struct proc *p;
404 {
405 	struct vattr va;
406 	int error;
407 	u_short mode;
408 
409 	error = VOP_GETATTR(vp, &va, p->p_ucred, p);
410 	if (error)
411 		return (error);
412 	/*
413 	 * Copy from vattr table
414 	 */
415 	sb->st_dev = va.va_fsid;
416 	sb->st_ino = va.va_fileid;
417 	mode = va.va_mode;
418 	switch (vp->v_type) {
419 	case VREG:
420 		mode |= S_IFREG;
421 		break;
422 	case VDIR:
423 		mode |= S_IFDIR;
424 		break;
425 	case VBLK:
426 		mode |= S_IFBLK;
427 		break;
428 	case VCHR:
429 		mode |= S_IFCHR;
430 		break;
431 	case VLNK:
432 		mode |= S_IFLNK;
433 		break;
434 	case VSOCK:
435 		mode |= S_IFSOCK;
436 		break;
437 	case VFIFO:
438 		mode |= S_IFIFO;
439 		break;
440 	default:
441 		return (EBADF);
442 	};
443 	sb->st_mode = mode;
444 	sb->st_nlink = va.va_nlink;
445 	sb->st_uid = va.va_uid;
446 	sb->st_gid = va.va_gid;
447 	sb->st_rdev = va.va_rdev;
448 	sb->st_size = va.va_size;
449 	sb->st_atimespec = va.va_atime;
450 	sb->st_mtimespec = va.va_mtime;
451 	sb->st_ctimespec = va.va_ctime;
452 	sb->st_blksize = va.va_blocksize;
453 	sb->st_flags = va.va_flags;
454 	sb->st_gen = 0;
455 	sb->st_blocks = va.va_bytes / S_BLKSIZE;
456 	return (0);
457 }
458 
459 /*
460  * File table vnode ioctl routine.
461  */
462 int
463 vn_ioctl(fp, com, data, p)
464 	struct file *fp;
465 	u_long com;
466 	caddr_t data;
467 	struct proc *p;
468 {
469 	register struct vnode *vp = ((struct vnode *)fp->f_data);
470 	struct vattr vattr;
471 	int error;
472 
473 	switch (vp->v_type) {
474 
475 	case VREG:
476 	case VDIR:
477 		if (com == FIONREAD) {
478 			error = VOP_GETATTR(vp, &vattr, p->p_ucred, p);
479 			if (error)
480 				return (error);
481 			*(int *)data = vattr.va_size - fp->f_offset;
482 			return (0);
483 		}
484 		if (com == FIONBIO || com == FIOASYNC)	/* XXX */
485 			return (0);			/* XXX */
486 		/* fall into ... */
487 
488 	default:
489 		return (ENOTTY);
490 
491 	case VFIFO:
492 	case VCHR:
493 	case VBLK:
494 		error = VOP_IOCTL(vp, com, data, fp->f_flag, p->p_ucred, p);
495 		if (error == 0 && com == TIOCSCTTY) {
496 			if (p->p_session->s_ttyvp)
497 				vrele(p->p_session->s_ttyvp);
498 			p->p_session->s_ttyvp = vp;
499 			VREF(vp);
500 		}
501 		return (error);
502 	}
503 }
504 
505 /*
506  * File table vnode poll routine.
507  */
508 int
509 vn_poll(fp, events, p)
510 	struct file *fp;
511 	int events;
512 	struct proc *p;
513 {
514 
515 	return (VOP_POLL(((struct vnode *)fp->f_data), events, p));
516 }
517 
518 /*
519  * File table vnode close routine.
520  */
521 int
522 vn_closefile(fp, p)
523 	struct file *fp;
524 	struct proc *p;
525 {
526 
527 	return (vn_close(((struct vnode *)fp->f_data), fp->f_flag,
528 		fp->f_cred, p));
529 }
530