ext2fs_readwrite.c revision 1.71 1 /* $NetBSD: ext2fs_readwrite.c,v 1.71 2015/03/28 17:06:15 riastradh Exp $ */
2
3 /*-
4 * Copyright (c) 1993
5 * The Regents of the University of California. All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of the University nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * SUCH DAMAGE.
30 *
31 * @(#)ufs_readwrite.c 8.8 (Berkeley) 8/4/94
32 * Modified for ext2fs by Manuel Bouyer.
33 */
34
35 /*-
36 * Copyright (c) 1997 Manuel Bouyer.
37 *
38 * Redistribution and use in source and binary forms, with or without
39 * modification, are permitted provided that the following conditions
40 * are met:
41 * 1. Redistributions of source code must retain the above copyright
42 * notice, this list of conditions and the following disclaimer.
43 * 2. Redistributions in binary form must reproduce the above copyright
44 * notice, this list of conditions and the following disclaimer in the
45 * documentation and/or other materials provided with the distribution.
46 *
47 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
48 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
49 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
50 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
51 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
52 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
53 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
54 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
55 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
56 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
57 *
58 * @(#)ufs_readwrite.c 8.8 (Berkeley) 8/4/94
59 * Modified for ext2fs by Manuel Bouyer.
60 */
61
62 #include <sys/cdefs.h>
63 __KERNEL_RCSID(0, "$NetBSD: ext2fs_readwrite.c,v 1.71 2015/03/28 17:06:15 riastradh Exp $");
64
65 #include <sys/param.h>
66 #include <sys/systm.h>
67 #include <sys/resourcevar.h>
68 #include <sys/kernel.h>
69 #include <sys/file.h>
70 #include <sys/stat.h>
71 #include <sys/buf.h>
72 #include <sys/proc.h>
73 #include <sys/mount.h>
74 #include <sys/vnode.h>
75 #include <sys/signalvar.h>
76 #include <sys/kauth.h>
77
78 #include <ufs/ufs/inode.h>
79 #include <ufs/ufs/ufsmount.h>
80 #include <ufs/ufs/ufs_extern.h>
81 #include <ufs/ext2fs/ext2fs.h>
82 #include <ufs/ext2fs/ext2fs_extern.h>
83
84 static int ext2fs_post_read_update(struct vnode *, int, int);
85 static int ext2fs_post_write_update(struct vnode *, struct uio *, int,
86 kauth_cred_t, off_t, int, int, int);
87
88 /*
89 * Vnode op for reading.
90 */
91 /* ARGSUSED */
92 int
93 ext2fs_read(void *v)
94 {
95 struct vop_read_args /* {
96 struct vnode *a_vp;
97 struct uio *a_uio;
98 int a_ioflag;
99 kauth_cred_t a_cred;
100 } */ *ap = v;
101 struct vnode *vp;
102 struct inode *ip;
103 struct uio *uio;
104 struct ufsmount *ump;
105 vsize_t bytelen;
106 int advice;
107 int error;
108
109 vp = ap->a_vp;
110 ip = VTOI(vp);
111 ump = ip->i_ump;
112 uio = ap->a_uio;
113 error = 0;
114
115 KASSERT(uio->uio_rw == UIO_READ);
116 KASSERT(vp->v_type == VREG || vp->v_type == VDIR);
117
118 /* XXX Eliminate me by refusing directory reads from userland. */
119 if (vp->v_type == VDIR)
120 return ext2fs_bufrd(vp, uio, ap->a_ioflag, ap->a_cred);
121
122 if ((uint64_t)uio->uio_offset > ump->um_maxfilesize)
123 return (EFBIG);
124 if (uio->uio_resid == 0)
125 return (0);
126 if (uio->uio_offset >= ext2fs_size(ip))
127 goto out;
128
129 KASSERT(vp->v_type == VREG);
130 advice = IO_ADV_DECODE(ap->a_ioflag);
131 while (uio->uio_resid > 0) {
132 bytelen = MIN(ext2fs_size(ip) - uio->uio_offset,
133 uio->uio_resid);
134 if (bytelen == 0)
135 break;
136
137 error = ubc_uiomove(&vp->v_uobj, uio, bytelen, advice,
138 UBC_READ | UBC_PARTIALOK | UBC_UNMAP_FLAG(vp));
139 if (error)
140 break;
141 }
142
143 out:
144 error = ext2fs_post_read_update(vp, ap->a_ioflag, error);
145 return (error);
146 }
147
148 /*
149 * UFS op for reading via the buffer cache
150 */
151 int
152 ext2fs_bufrd(struct vnode *vp, struct uio *uio, int ioflag, kauth_cred_t cred)
153 {
154 struct inode *ip;
155 struct ufsmount *ump;
156 struct m_ext2fs *fs;
157 struct buf *bp;
158 off_t bytesinfile;
159 daddr_t lbn, nextlbn;
160 long size, xfersize, blkoffset;
161 int error;
162
163 KASSERT(uio->uio_rw == UIO_READ);
164 KASSERT(VOP_ISLOCKED(vp));
165 KASSERT(vp->v_type == VDIR || vp->v_type == VLNK);
166
167 ip = VTOI(vp);
168 ump = ip->i_ump;
169 fs = ip->i_e2fs;
170 error = 0;
171
172 KASSERT(vp->v_type != VLNK ||
173 ext2fs_size(ip) >= ump->um_maxsymlinklen);
174 KASSERT(vp->v_type != VLNK || ump->um_maxsymlinklen != 0 ||
175 ext2fs_nblock(ip) != 0);
176
177 if (uio->uio_offset > ump->um_maxfilesize)
178 return EFBIG;
179 if (uio->uio_resid == 0)
180 return 0;
181 if (uio->uio_offset >= ext2fs_size(ip))
182 goto out;
183
184 for (error = 0, bp = NULL; uio->uio_resid > 0; bp = NULL) {
185 bytesinfile = ext2fs_size(ip) - uio->uio_offset;
186 if (bytesinfile <= 0)
187 break;
188 lbn = ext2_lblkno(fs, uio->uio_offset);
189 nextlbn = lbn + 1;
190 size = fs->e2fs_bsize;
191 blkoffset = ext2_blkoff(fs, uio->uio_offset);
192 xfersize = fs->e2fs_bsize - blkoffset;
193 if (uio->uio_resid < xfersize)
194 xfersize = uio->uio_resid;
195 if (bytesinfile < xfersize)
196 xfersize = bytesinfile;
197
198 if (ext2_lblktosize(fs, nextlbn) >= ext2fs_size(ip))
199 error = bread(vp, lbn, size, NOCRED, 0, &bp);
200 else {
201 int nextsize = fs->e2fs_bsize;
202 error = breadn(vp, lbn,
203 size, &nextlbn, &nextsize, 1, NOCRED, 0, &bp);
204 }
205 if (error)
206 break;
207
208 /*
209 * We should only get non-zero b_resid when an I/O error
210 * has occurred, which should cause us to break above.
211 * However, if the short read did not cause an error,
212 * then we want to ensure that we do not uiomove bad
213 * or uninitialized data.
214 */
215 size -= bp->b_resid;
216 if (size < xfersize) {
217 if (size == 0)
218 break;
219 xfersize = size;
220 }
221 error = uiomove((char *)bp->b_data + blkoffset, xfersize, uio);
222 if (error)
223 break;
224 brelse(bp, 0);
225 }
226 if (bp != NULL)
227 brelse(bp, 0);
228
229 out:
230 error = ext2fs_post_read_update(vp, ioflag, error);
231 return (error);
232 }
233
234 static int
235 ext2fs_post_read_update(struct vnode *vp, int ioflag, int error)
236 {
237 struct inode *ip = VTOI(vp);
238
239 if (!(vp->v_mount->mnt_flag & MNT_NOATIME)) {
240 ip->i_flag |= IN_ACCESS;
241 if ((ioflag & IO_SYNC) == IO_SYNC)
242 error = ext2fs_update(vp, NULL, NULL, UPDATE_WAIT);
243 }
244
245 return error;
246 }
247
248 /*
249 * Vnode op for writing.
250 */
251 int
252 ext2fs_write(void *v)
253 {
254 struct vop_write_args /* {
255 struct vnode *a_vp;
256 struct uio *a_uio;
257 int a_ioflag;
258 kauth_cred_t a_cred;
259 } */ *ap = v;
260 struct vnode *vp;
261 struct uio *uio;
262 struct inode *ip;
263 struct m_ext2fs *fs;
264 struct ufsmount *ump;
265 off_t osize;
266 int blkoffset, error, ioflag, resid;
267 vsize_t bytelen;
268 off_t oldoff = 0; /* XXX */
269 bool async;
270 int extended = 0;
271 int advice;
272
273 ioflag = ap->a_ioflag;
274 advice = IO_ADV_DECODE(ioflag);
275 uio = ap->a_uio;
276 vp = ap->a_vp;
277 ip = VTOI(vp);
278 ump = ip->i_ump;
279 error = 0;
280
281 KASSERT(uio->uio_rw == UIO_WRITE);
282 KASSERT(vp->v_type == VREG);
283
284 if (ioflag & IO_APPEND)
285 uio->uio_offset = ext2fs_size(ip);
286 if ((ip->i_e2fs_flags & EXT2_APPEND) &&
287 uio->uio_offset != ext2fs_size(ip))
288 return (EPERM);
289
290 fs = ip->i_e2fs;
291 if (uio->uio_offset < 0 ||
292 (uint64_t)uio->uio_offset + uio->uio_resid > ump->um_maxfilesize)
293 return (EFBIG);
294 if (uio->uio_resid == 0)
295 return (0);
296
297 async = vp->v_mount->mnt_flag & MNT_ASYNC;
298 resid = uio->uio_resid;
299 osize = ext2fs_size(ip);
300
301 KASSERT(vp->v_type == VREG);
302 while (uio->uio_resid > 0) {
303 oldoff = uio->uio_offset;
304 blkoffset = ext2_blkoff(fs, uio->uio_offset);
305 bytelen = MIN(fs->e2fs_bsize - blkoffset, uio->uio_resid);
306
307 if (vp->v_size < oldoff + bytelen) {
308 uvm_vnp_setwritesize(vp, oldoff + bytelen);
309 }
310 error = ufs_balloc_range(vp, uio->uio_offset, bytelen,
311 ap->a_cred, 0);
312 if (error)
313 break;
314 error = ubc_uiomove(&vp->v_uobj, uio, bytelen, advice,
315 UBC_WRITE | UBC_UNMAP_FLAG(vp));
316 if (error)
317 break;
318
319 /*
320 * update UVM's notion of the size now that we've
321 * copied the data into the vnode's pages.
322 */
323
324 if (vp->v_size < uio->uio_offset) {
325 uvm_vnp_setsize(vp, uio->uio_offset);
326 extended = 1;
327 }
328
329 /*
330 * flush what we just wrote if necessary.
331 * XXXUBC simplistic async flushing.
332 */
333
334 if (!async && oldoff >> 16 != uio->uio_offset >> 16) {
335 mutex_enter(vp->v_interlock);
336 error = VOP_PUTPAGES(vp, (oldoff >> 16) << 16,
337 (uio->uio_offset >> 16) << 16,
338 PGO_CLEANIT | PGO_LAZY);
339 }
340 }
341 if (error == 0 && ioflag & IO_SYNC) {
342 mutex_enter(vp->v_interlock);
343 error = VOP_PUTPAGES(vp, trunc_page(oldoff),
344 round_page(ext2_blkroundup(fs, uio->uio_offset)),
345 PGO_CLEANIT | PGO_SYNCIO);
346 }
347
348 error = ext2fs_post_write_update(vp, uio, ioflag, ap->a_cred, osize,
349 resid, extended, error);
350 return (error);
351 }
352
353 /*
354 * UFS op for writing via the buffer cache
355 */
356 int
357 ext2fs_bufwr(struct vnode *vp, struct uio *uio, int ioflag, kauth_cred_t cred)
358 {
359 struct inode *ip;
360 struct ufsmount *ump;
361 struct m_ext2fs *fs;
362 struct buf *bp;
363 int flags;
364 off_t osize;
365 daddr_t lbn;
366 int resid, blkoffset, xfersize;
367 int extended = 0;
368 int error;
369
370 KASSERT(VOP_ISLOCKED(vp) == LK_EXCLUSIVE);
371 KASSERT(vp->v_type == VDIR || vp->v_type == VLNK);
372 KASSERT(vp->v_type != VDIR || ISSET(ioflag, IO_SYNC));
373 KASSERT(uio->uio_rw == UIO_WRITE);
374
375 ip = VTOI(vp);
376 ump = ip->i_ump;
377 fs = ip->i_e2fs;
378 error = 0;
379
380 if (uio->uio_offset < 0 ||
381 uio->uio_resid > ump->um_maxfilesize ||
382 uio->uio_offset > (ump->um_maxfilesize - uio->uio_resid))
383 return EFBIG;
384 if (uio->uio_resid == 0)
385 return 0;
386
387 flags = ioflag & IO_SYNC ? B_SYNC : 0;
388 resid = uio->uio_resid;
389 osize = ext2fs_size(ip);
390
391 for (error = 0; uio->uio_resid > 0;) {
392 lbn = ext2_lblkno(fs, uio->uio_offset);
393 blkoffset = ext2_blkoff(fs, uio->uio_offset);
394 xfersize = MIN(fs->e2fs_bsize - blkoffset, uio->uio_resid);
395 if (xfersize < fs->e2fs_bsize)
396 flags |= B_CLRBUF;
397 else
398 flags &= ~B_CLRBUF;
399 error = ext2fs_balloc(ip, lbn, blkoffset + xfersize, cred, &bp,
400 flags);
401 if (error)
402 break;
403 if (ext2fs_size(ip) < uio->uio_offset + xfersize) {
404 error = ext2fs_setsize(ip, uio->uio_offset + xfersize);
405 if (error)
406 break;
407 }
408 error = uiomove((char *)bp->b_data + blkoffset, xfersize, uio);
409
410 /*
411 * update UVM's notion of the size now that we've
412 * copied the data into the vnode's pages.
413 */
414
415 if (vp->v_size < uio->uio_offset) {
416 uvm_vnp_setsize(vp, uio->uio_offset);
417 extended = 1;
418 }
419
420 if (ioflag & IO_SYNC)
421 (void)bwrite(bp);
422 else if (xfersize + blkoffset == fs->e2fs_bsize)
423 bawrite(bp);
424 else
425 bdwrite(bp);
426 if (error || xfersize == 0)
427 break;
428 }
429
430 error = ext2fs_post_write_update(vp, uio, ioflag, cred, osize, resid,
431 extended, error);
432 return (error);
433 }
434
435 static int
436 ext2fs_post_write_update(struct vnode *vp, struct uio *uio, int ioflag,
437 kauth_cred_t cred, off_t osize, int resid, int extended, int error)
438 {
439 struct inode *ip = VTOI(vp);
440
441 /* Trigger ctime and mtime updates, and atime if MNT_RELATIME. */
442 ip->i_flag |= IN_CHANGE | IN_UPDATE;
443 if (vp->v_mount->mnt_flag & MNT_RELATIME)
444 ip->i_flag |= IN_ACCESS;
445
446 /*
447 * If we successfully wrote any data and we are not the superuser,
448 * we clear the setuid and setgid bits as a precaution against
449 * tampering.
450 */
451 if (resid > uio->uio_resid && cred) {
452 if (ip->i_e2fs_mode & ISUID) {
453 if (kauth_authorize_vnode(cred,
454 KAUTH_VNODE_RETAIN_SUID, vp, NULL, EPERM) != 0)
455 ip->i_e2fs_mode &= ISUID;
456 }
457
458 if (ip->i_e2fs_mode & ISGID) {
459 if (kauth_authorize_vnode(cred,
460 KAUTH_VNODE_RETAIN_SGID, vp, NULL, EPERM) != 0)
461 ip->i_e2fs_mode &= ~ISGID;
462 }
463 }
464
465 /* If we successfully wrote anything, notify kevent listeners. */
466 if (resid > uio->uio_resid)
467 VN_KNOTE(vp, NOTE_WRITE | (extended ? NOTE_EXTEND : 0));
468
469 /*
470 * Update the size on disk: truncate back to original size on
471 * error, or reflect the new size on success.
472 */
473 if (error) {
474 (void) ext2fs_truncate(vp, osize, ioflag & IO_SYNC, cred);
475 uio->uio_offset -= resid - uio->uio_resid;
476 uio->uio_resid = resid;
477 } else if (resid > uio->uio_resid && (ioflag & IO_SYNC) == IO_SYNC)
478 error = ext2fs_update(vp, NULL, NULL, UPDATE_WAIT);
479
480 /* Make sure the vnode uvm size matches the inode file size. */
481 KASSERT(vp->v_size == ext2fs_size(ip));
482
483 return error;
484 }
485