xref: /netbsd-src/sys/ufs/ext2fs/ext2fs_readwrite.c (revision 404fbe5fb94ca1e054339640cabb2801ce52dd30)
1 /*	$NetBSD: ext2fs_readwrite.c,v 1.53 2008/11/26 20:17:33 pooka 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  * 3. All advertising materials mentioning features or use of this software
47  *    must display the following acknowledgement:
48  *	This product includes software developed by Manuel Bouyer.
49  * 4. The name of the author may not be used to endorse or promote products
50  *    derived from this software without specific prior written permission.
51  *
52  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
53  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
54  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
55  * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
56  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
57  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
58  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
59  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
60  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
61  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
62  *
63  *	@(#)ufs_readwrite.c	8.8 (Berkeley) 8/4/94
64  * Modified for ext2fs by Manuel Bouyer.
65  */
66 
67 #include <sys/cdefs.h>
68 __KERNEL_RCSID(0, "$NetBSD: ext2fs_readwrite.c,v 1.53 2008/11/26 20:17:33 pooka Exp $");
69 
70 #include <sys/param.h>
71 #include <sys/systm.h>
72 #include <sys/resourcevar.h>
73 #include <sys/kernel.h>
74 #include <sys/file.h>
75 #include <sys/stat.h>
76 #include <sys/buf.h>
77 #include <sys/proc.h>
78 #include <sys/mount.h>
79 #include <sys/vnode.h>
80 #include <sys/malloc.h>
81 #include <sys/signalvar.h>
82 #include <sys/kauth.h>
83 
84 #include <ufs/ufs/inode.h>
85 #include <ufs/ufs/ufsmount.h>
86 #include <ufs/ufs/ufs_extern.h>
87 #include <ufs/ext2fs/ext2fs.h>
88 #include <ufs/ext2fs/ext2fs_extern.h>
89 
90 
91 #define doclusterread 0 /* XXX underway */
92 #define doclusterwrite 0
93 
94 /*
95  * Vnode op for reading.
96  */
97 /* ARGSUSED */
98 int
99 ext2fs_read(void *v)
100 {
101 	struct vop_read_args /* {
102 		struct vnode *a_vp;
103 		struct uio *a_uio;
104 		int a_ioflag;
105 		kauth_cred_t a_cred;
106 	} */ *ap = v;
107 	struct vnode *vp;
108 	struct inode *ip;
109 	struct uio *uio;
110 	struct m_ext2fs *fs;
111 	struct buf *bp;
112 	struct ufsmount *ump;
113 	vsize_t bytelen;
114 	daddr_t lbn, nextlbn;
115 	off_t bytesinfile;
116 	long size, xfersize, blkoffset;
117 	int error;
118 
119 	vp = ap->a_vp;
120 	ip = VTOI(vp);
121 	ump = ip->i_ump;
122 	uio = ap->a_uio;
123 	error = 0;
124 
125 #ifdef DIAGNOSTIC
126 	if (uio->uio_rw != UIO_READ)
127 		panic("%s: mode", "ext2fs_read");
128 
129 	if (vp->v_type == VLNK) {
130 		if (ext2fs_size(ip) < ump->um_maxsymlinklen ||
131 		    (ump->um_maxsymlinklen == 0 && ip->i_e2fs_nblock == 0))
132 			panic("%s: short symlink", "ext2fs_read");
133 	} else if (vp->v_type != VREG && vp->v_type != VDIR)
134 		panic("%s: type %d", "ext2fs_read", vp->v_type);
135 #endif
136 	fs = ip->i_e2fs;
137 	if ((u_int64_t)uio->uio_offset > ump->um_maxfilesize)
138 		return (EFBIG);
139 	if (uio->uio_resid == 0)
140 		return (0);
141 	if (uio->uio_offset >= ext2fs_size(ip))
142 		goto out;
143 
144 	if (vp->v_type == VREG) {
145 		const int advice = IO_ADV_DECODE(ap->a_ioflag);
146 
147 		while (uio->uio_resid > 0) {
148 			bytelen = MIN(ext2fs_size(ip) - uio->uio_offset,
149 			    uio->uio_resid);
150 			if (bytelen == 0)
151 				break;
152 
153 			error = ubc_uiomove(&vp->v_uobj, uio, bytelen, advice,
154 			    UBC_READ | UBC_PARTIALOK | UBC_UNMAP_FLAG(vp));
155 			if (error)
156 				break;
157 		}
158 		goto out;
159 	}
160 
161 	for (error = 0, bp = NULL; uio->uio_resid > 0; bp = NULL) {
162 		bytesinfile = ext2fs_size(ip) - uio->uio_offset;
163 		if (bytesinfile <= 0)
164 			break;
165 		lbn = lblkno(fs, uio->uio_offset);
166 		nextlbn = lbn + 1;
167 		size = fs->e2fs_bsize;
168 		blkoffset = blkoff(fs, uio->uio_offset);
169 		xfersize = fs->e2fs_bsize - blkoffset;
170 		if (uio->uio_resid < xfersize)
171 			xfersize = uio->uio_resid;
172 		if (bytesinfile < xfersize)
173 			xfersize = bytesinfile;
174 
175 		if (lblktosize(fs, nextlbn) >= ext2fs_size(ip))
176 			error = bread(vp, lbn, size, NOCRED, 0, &bp);
177 		else {
178 			int nextsize = fs->e2fs_bsize;
179 			error = breadn(vp, lbn,
180 				size, &nextlbn, &nextsize, 1, NOCRED, 0, &bp);
181 		}
182 		if (error)
183 			break;
184 
185 		/*
186 		 * We should only get non-zero b_resid when an I/O error
187 		 * has occurred, which should cause us to break above.
188 		 * However, if the short read did not cause an error,
189 		 * then we want to ensure that we do not uiomove bad
190 		 * or uninitialized data.
191 		 */
192 		size -= bp->b_resid;
193 		if (size < xfersize) {
194 			if (size == 0)
195 				break;
196 			xfersize = size;
197 		}
198 		error = uiomove((char *)bp->b_data + blkoffset, xfersize, uio);
199 		if (error)
200 			break;
201 		brelse(bp, 0);
202 	}
203 	if (bp != NULL)
204 		brelse(bp, 0);
205 
206 out:
207 	if (!(vp->v_mount->mnt_flag & MNT_NOATIME)) {
208 		ip->i_flag |= IN_ACCESS;
209 		if ((ap->a_ioflag & IO_SYNC) == IO_SYNC)
210 			error = ext2fs_update(vp, NULL, NULL, UPDATE_WAIT);
211 	}
212 	return (error);
213 }
214 
215 /*
216  * Vnode op for writing.
217  */
218 int
219 ext2fs_write(void *v)
220 {
221 	struct vop_write_args /* {
222 		struct vnode *a_vp;
223 		struct uio *a_uio;
224 		int a_ioflag;
225 		kauth_cred_t a_cred;
226 	} */ *ap = v;
227 	struct vnode *vp;
228 	struct uio *uio;
229 	struct inode *ip;
230 	struct m_ext2fs *fs;
231 	struct buf *bp;
232 	struct proc *p;
233 	struct ufsmount *ump;
234 	daddr_t lbn;
235 	off_t osize;
236 	int blkoffset, error, flags, ioflag, resid, xfersize;
237 	vsize_t bytelen;
238 	off_t oldoff = 0;					/* XXX */
239 	bool async;
240 	int extended = 0;
241 	int advice;
242 
243 	ioflag = ap->a_ioflag;
244 	advice = IO_ADV_DECODE(ioflag);
245 	uio = ap->a_uio;
246 	vp = ap->a_vp;
247 	ip = VTOI(vp);
248 	ump = ip->i_ump;
249 	error = 0;
250 
251 #ifdef DIAGNOSTIC
252 	if (uio->uio_rw != UIO_WRITE)
253 		panic("%s: mode", "ext2fs_write");
254 #endif
255 
256 	switch (vp->v_type) {
257 	case VREG:
258 		if (ioflag & IO_APPEND)
259 			uio->uio_offset = ext2fs_size(ip);
260 		if ((ip->i_e2fs_flags & EXT2_APPEND) &&
261 		    uio->uio_offset != ext2fs_size(ip))
262 			return (EPERM);
263 		/* FALLTHROUGH */
264 	case VLNK:
265 		break;
266 	case VDIR:
267 		if ((ioflag & IO_SYNC) == 0)
268 			panic("%s: nonsync dir write", "ext2fs_write");
269 		break;
270 	default:
271 		panic("%s: type", "ext2fs_write");
272 	}
273 
274 	fs = ip->i_e2fs;
275 	if (uio->uio_offset < 0 ||
276 	    (u_int64_t)uio->uio_offset + uio->uio_resid > ump->um_maxfilesize)
277 		return (EFBIG);
278 	/*
279 	 * Maybe this should be above the vnode op call, but so long as
280 	 * file servers have no limits, I don't think it matters.
281 	 */
282 	p = curproc;
283 	if (vp->v_type == VREG && p &&
284 	    uio->uio_offset + uio->uio_resid >
285 	    p->p_rlimit[RLIMIT_FSIZE].rlim_cur) {
286 		mutex_enter(proc_lock);
287 		psignal(p, SIGXFSZ);
288 		mutex_exit(proc_lock);
289 		return (EFBIG);
290 	}
291 	if (uio->uio_resid == 0)
292 		return (0);
293 
294 	async = vp->v_mount->mnt_flag & MNT_ASYNC;
295 	resid = uio->uio_resid;
296 	osize = ext2fs_size(ip);
297 
298 	if (vp->v_type == VREG) {
299 		while (uio->uio_resid > 0) {
300 			oldoff = uio->uio_offset;
301 			blkoffset = blkoff(fs, uio->uio_offset);
302 			bytelen = MIN(fs->e2fs_bsize - blkoffset,
303 			    uio->uio_resid);
304 
305 			if (vp->v_size < oldoff + bytelen) {
306 				uvm_vnp_setwritesize(vp, oldoff + bytelen);
307 			}
308 			error = ufs_balloc_range(vp, uio->uio_offset,
309 			    bytelen, ap->a_cred, 0);
310 			if (error)
311 				break;
312 			error = ubc_uiomove(&vp->v_uobj, uio, bytelen, advice,
313 			    UBC_WRITE | UBC_UNMAP_FLAG(vp));
314 			if (error)
315 				break;
316 
317 			/*
318 			 * update UVM's notion of the size now that we've
319 			 * copied the data into the vnode's pages.
320 			 */
321 
322 			if (vp->v_size < uio->uio_offset) {
323 				uvm_vnp_setsize(vp, uio->uio_offset);
324 				extended = 1;
325 			}
326 
327 			/*
328 			 * flush what we just wrote if necessary.
329 			 * XXXUBC simplistic async flushing.
330 			 */
331 
332 			if (!async && oldoff >> 16 != uio->uio_offset >> 16) {
333 				mutex_enter(&vp->v_interlock);
334 				error = VOP_PUTPAGES(vp, (oldoff >> 16) << 16,
335 				    (uio->uio_offset >> 16) << 16, PGO_CLEANIT);
336 			}
337 		}
338 		if (error == 0 && ioflag & IO_SYNC) {
339 			mutex_enter(&vp->v_interlock);
340 			error = VOP_PUTPAGES(vp, trunc_page(oldoff),
341 			    round_page(blkroundup(fs, uio->uio_offset)),
342 			    PGO_CLEANIT | PGO_SYNCIO);
343 		}
344 
345 		goto out;
346 	}
347 
348 	flags = ioflag & IO_SYNC ? B_SYNC : 0;
349 	for (error = 0; uio->uio_resid > 0;) {
350 		lbn = lblkno(fs, uio->uio_offset);
351 		blkoffset = blkoff(fs, uio->uio_offset);
352 		xfersize = MIN(fs->e2fs_bsize - blkoffset, uio->uio_resid);
353 		if (xfersize < fs->e2fs_bsize)
354 			flags |= B_CLRBUF;
355 		else
356 			flags &= ~B_CLRBUF;
357 		error = ext2fs_balloc(ip,
358 		    lbn, blkoffset + xfersize, ap->a_cred, &bp, flags);
359 		if (error)
360 			break;
361 		if (ext2fs_size(ip) < uio->uio_offset + xfersize) {
362 			error = ext2fs_setsize(ip, uio->uio_offset + xfersize);
363 			if (error)
364 				break;
365 		}
366 		error = uiomove((char *)bp->b_data + blkoffset, xfersize, uio);
367 
368 		/*
369 		 * update UVM's notion of the size now that we've
370 		 * copied the data into the vnode's pages.
371 		 */
372 
373 		if (vp->v_size < uio->uio_offset) {
374 			uvm_vnp_setsize(vp, uio->uio_offset);
375 			extended = 1;
376 		}
377 
378 		if (ioflag & IO_SYNC)
379 			(void)bwrite(bp);
380 		else if (xfersize + blkoffset == fs->e2fs_bsize)
381 			bawrite(bp);
382 		else
383 			bdwrite(bp);
384 		if (error || xfersize == 0)
385 			break;
386 	}
387 
388 	/*
389 	 * If we successfully wrote any data, and we are not the superuser
390 	 * we clear the setuid and setgid bits as a precaution against
391 	 * tampering.
392 	 */
393 
394 out:
395 	ip->i_flag |= IN_CHANGE | IN_UPDATE;
396 	if (resid > uio->uio_resid && ap->a_cred &&
397 	    kauth_authorize_generic(ap->a_cred, KAUTH_GENERIC_ISSUSER, NULL))
398 		ip->i_e2fs_mode &= ~(ISUID | ISGID);
399 	if (resid > uio->uio_resid)
400 		VN_KNOTE(vp, NOTE_WRITE | (extended ? NOTE_EXTEND : 0));
401 	if (error) {
402 		(void) ext2fs_truncate(vp, osize, ioflag & IO_SYNC, ap->a_cred);
403 		uio->uio_offset -= resid - uio->uio_resid;
404 		uio->uio_resid = resid;
405 	} else if (resid > uio->uio_resid && (ioflag & IO_SYNC) == IO_SYNC)
406 		error = ext2fs_update(vp, NULL, NULL, UPDATE_WAIT);
407 	KASSERT(vp->v_size == ext2fs_size(ip));
408 	return (error);
409 }
410