xref: /netbsd-src/sys/ufs/ext2fs/ext2fs_readwrite.c (revision 481fca6e59249d8ffcf24fef7cfbe7b131bfb080)
1 /*	$NetBSD: ext2fs_readwrite.c,v 1.13 2000/06/28 14:16:38 mrg Exp $	*/
2 
3 /*-
4  * Copyright (c) 1997 Manuel Bouyer.
5  * Copyright (c) 1993
6  *	The Regents of the University of California.  All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  * 3. All advertising materials mentioning features or use of this software
17  *    must display the following acknowledgement:
18  *	This product includes software developed by the University of
19  *	California, Berkeley and its contributors.
20  * 4. Neither the name of the University nor the names of its contributors
21  *    may be used to endorse or promote products derived from this software
22  *    without specific prior written permission.
23  *
24  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
25  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
29  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
30  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
31  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
33  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34  * SUCH DAMAGE.
35  *
36  *	@(#)ufs_readwrite.c	8.8 (Berkeley) 8/4/94
37  * Modified for ext2fs by Manuel Bouyer.
38  */
39 
40 #include <sys/param.h>
41 #include <sys/systm.h>
42 #include <sys/resourcevar.h>
43 #include <sys/kernel.h>
44 #include <sys/file.h>
45 #include <sys/stat.h>
46 #include <sys/buf.h>
47 #include <sys/proc.h>
48 #include <sys/conf.h>
49 #include <sys/mount.h>
50 #include <sys/vnode.h>
51 #include <sys/malloc.h>
52 #include <sys/signalvar.h>
53 
54 #include <ufs/ufs/quota.h>
55 #include <ufs/ufs/inode.h>
56 #include <ufs/ext2fs/ext2fs.h>
57 #include <ufs/ext2fs/ext2fs_extern.h>
58 
59 
60 #define doclusterread 0 /* XXX underway */
61 #define doclusterwrite 0
62 
63 /*
64  * Vnode op for reading.
65  */
66 /* ARGSUSED */
67 int
68 ext2fs_read(v)
69 	void *v;
70 {
71 	struct vop_read_args /* {
72 		struct vnode *a_vp;
73 		struct uio *a_uio;
74 		int a_ioflag;
75 		struct ucred *a_cred;
76 	} */ *ap = v;
77 	struct vnode *vp;
78 	struct inode *ip;
79 	struct uio *uio;
80 	struct m_ext2fs *fs;
81 	struct buf *bp;
82 	ufs_daddr_t lbn, nextlbn;
83 	off_t bytesinfile;
84 	long size, xfersize, blkoffset;
85 	int error;
86 
87 	vp = ap->a_vp;
88 	ip = VTOI(vp);
89 	uio = ap->a_uio;
90 
91 #ifdef DIAGNOSTIC
92 	if (uio->uio_rw != UIO_READ)
93 		panic("%s: mode", "ext2fs_read");
94 
95 	if (vp->v_type == VLNK) {
96 		if ((int)ip->i_e2fs_size < vp->v_mount->mnt_maxsymlinklen ||
97 			(vp->v_mount->mnt_maxsymlinklen == 0 &&
98 			 ip->i_e2fs_nblock == 0))
99 			panic("%s: short symlink", "ext2fs_read");
100 	} else if (vp->v_type != VREG && vp->v_type != VDIR)
101 		panic("%s: type %d", "ext2fs_read", vp->v_type);
102 #endif
103 	fs = ip->i_e2fs;
104 	if ((u_int64_t)uio->uio_offset >
105 		((u_int64_t)0x80000000 * fs->e2fs_bsize - 1))
106 		return (EFBIG);
107 	if (uio->uio_resid == 0)
108 		return (0);
109 
110 	for (error = 0, bp = NULL; uio->uio_resid > 0; bp = NULL) {
111 		if ((bytesinfile = ip->i_e2fs_size - uio->uio_offset) <= 0)
112 			break;
113 		lbn = lblkno(fs, uio->uio_offset);
114 		nextlbn = lbn + 1;
115 		size = fs->e2fs_bsize;
116 		blkoffset = blkoff(fs, uio->uio_offset);
117 		xfersize = fs->e2fs_bsize - blkoffset;
118 		if (uio->uio_resid < xfersize)
119 			xfersize = uio->uio_resid;
120 		if (bytesinfile < xfersize)
121 			xfersize = bytesinfile;
122 
123 		if (lblktosize(fs, nextlbn) >= ip->i_e2fs_size)
124 			error = bread(vp, lbn, size, NOCRED, &bp);
125 		else if (doclusterread)
126 			error = cluster_read(vp,
127 				ip->i_e2fs_size, lbn, size, NOCRED, &bp);
128 		else if (lbn - 1 == vp->v_lastr) {
129 			int nextsize = fs->e2fs_bsize;
130 			error = breadn(vp, lbn,
131 				size, &nextlbn, &nextsize, 1, NOCRED, &bp);
132 		} else
133 			error = bread(vp, lbn, size, NOCRED, &bp);
134 		if (error)
135 			break;
136 		vp->v_lastr = lbn;
137 
138 		/*
139 		 * We should only get non-zero b_resid when an I/O error
140 		 * has occurred, which should cause us to break above.
141 		 * However, if the short read did not cause an error,
142 		 * then we want to ensure that we do not uiomove bad
143 		 * or uninitialized data.
144 		 */
145 		size -= bp->b_resid;
146 		if (size < xfersize) {
147 			if (size == 0)
148 				break;
149 			xfersize = size;
150 		}
151 		error = uiomove((char *)bp->b_data + blkoffset, (int)xfersize,
152 				uio);
153 		if (error)
154 			break;
155 		brelse(bp);
156 	}
157 	if (bp != NULL)
158 		brelse(bp);
159 	if (!(vp->v_mount->mnt_flag & MNT_NOATIME)) {
160 		ip->i_flag |= IN_ACCESS;
161 		if ((ap->a_ioflag & IO_SYNC) == IO_SYNC)
162 			error = VOP_UPDATE(vp, NULL, NULL, UPDATE_WAIT);
163 	}
164 	return (error);
165 }
166 
167 /*
168  * Vnode op for writing.
169  */
170 int
171 ext2fs_write(v)
172 	void *v;
173 {
174 	struct vop_write_args /* {
175 		struct vnode *a_vp;
176 		struct uio *a_uio;
177 		int a_ioflag;
178 		struct ucred *a_cred;
179 	} */ *ap = v;
180 	struct vnode *vp;
181 	struct uio *uio;
182 	struct inode *ip;
183 	struct m_ext2fs *fs;
184 	struct buf *bp;
185 	struct proc *p;
186 	ufs_daddr_t lbn;
187 	off_t osize;
188 	int blkoffset, error, flags, ioflag, resid, size, xfersize;
189 
190 	ioflag = ap->a_ioflag;
191 	uio = ap->a_uio;
192 	vp = ap->a_vp;
193 	ip = VTOI(vp);
194 
195 #ifdef DIAGNOSTIC
196 	if (uio->uio_rw != UIO_WRITE)
197 		panic("%s: mode", "ext2fs_write");
198 #endif
199 
200 	switch (vp->v_type) {
201 	case VREG:
202 		if (ioflag & IO_APPEND)
203 			uio->uio_offset = ip->i_e2fs_size;
204 		if ((ip->i_e2fs_flags & EXT2_APPEND) &&
205 			uio->uio_offset != ip->i_e2fs_size)
206 			return (EPERM);
207 		/* FALLTHROUGH */
208 	case VLNK:
209 		break;
210 	case VDIR:
211 		if ((ioflag & IO_SYNC) == 0)
212 			panic("%s: nonsync dir write", "ext2fs_write");
213 		break;
214 	default:
215 		panic("%s: type", "ext2fs_write");
216 	}
217 
218 	fs = ip->i_e2fs;
219 	if (uio->uio_offset < 0 ||
220 		(u_int64_t)uio->uio_offset + uio->uio_resid >
221 		((u_int64_t)0x80000000 * fs->e2fs_bsize - 1))
222 		return (EFBIG);
223 	/*
224 	 * Maybe this should be above the vnode op call, but so long as
225 	 * file servers have no limits, I don't think it matters.
226 	 */
227 	p = uio->uio_procp;
228 	if (vp->v_type == VREG && p &&
229 		uio->uio_offset + uio->uio_resid >
230 		p->p_rlimit[RLIMIT_FSIZE].rlim_cur) {
231 		psignal(p, SIGXFSZ);
232 		return (EFBIG);
233 	}
234 
235 	resid = uio->uio_resid;
236 	osize = ip->i_e2fs_size;
237 	flags = ioflag & IO_SYNC ? B_SYNC : 0;
238 
239 	for (error = 0; uio->uio_resid > 0;) {
240 		lbn = lblkno(fs, uio->uio_offset);
241 		blkoffset = blkoff(fs, uio->uio_offset);
242 		xfersize = fs->e2fs_bsize - blkoffset;
243 		if (uio->uio_resid < xfersize)
244 			xfersize = uio->uio_resid;
245 		if (fs->e2fs_bsize > xfersize)
246 			flags |= B_CLRBUF;
247 		else
248 			flags &= ~B_CLRBUF;
249 
250 		error = ext2fs_balloc(ip,
251 			lbn, blkoffset + xfersize, ap->a_cred, &bp, flags);
252 		if (error)
253 			break;
254 		if (uio->uio_offset + xfersize > ip->i_e2fs_size) {
255 			ip->i_e2fs_size = uio->uio_offset + xfersize;
256 			uvm_vnp_setsize(vp, ip->i_e2fs_size);
257 		}
258 		(void)uvm_vnp_uncache(vp);
259 
260 		size = fs->e2fs_bsize - bp->b_resid;
261 		if (size < xfersize)
262 			xfersize = size;
263 
264 		error =
265 			uiomove((char *)bp->b_data + blkoffset, (int)xfersize, uio);
266 		if (ioflag & IO_SYNC)
267 			(void)bwrite(bp);
268 		else if (xfersize + blkoffset == fs->e2fs_bsize)
269 			if (doclusterwrite)
270 				cluster_write(bp, ip->i_e2fs_size);
271 			else
272 				bawrite(bp);
273 		else
274 			bdwrite(bp);
275 		if (error || xfersize == 0)
276 			break;
277 		ip->i_flag |= IN_CHANGE | IN_UPDATE;
278 	}
279 	/*
280 	 * If we successfully wrote any data, and we are not the superuser
281 	 * we clear the setuid and setgid bits as a precaution against
282 	 * tampering.
283 	 */
284 	if (resid > uio->uio_resid && ap->a_cred && ap->a_cred->cr_uid != 0)
285 		ip->i_e2fs_mode &= ~(ISUID | ISGID);
286 	if (error) {
287 		if (ioflag & IO_UNIT) {
288 			(void)VOP_TRUNCATE(vp, osize,
289 				ioflag & IO_SYNC, ap->a_cred, uio->uio_procp);
290 			uio->uio_offset -= resid - uio->uio_resid;
291 			uio->uio_resid = resid;
292 		}
293 	} else if (resid > uio->uio_resid && (ioflag & IO_SYNC) == IO_SYNC)
294 		error = VOP_UPDATE(vp, NULL, NULL, UPDATE_WAIT);
295 	return (error);
296 }
297