2 * modified for Lites 1.1
4 * Aug 1995, Godmar Back (gback@cs.utah.edu)
5 * University of Utah, Department of Computer Science
9 * The Regents of the University of California. All rights reserved.
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * 3. All advertising materials mentioning features or use of this software
20 * must display the following acknowledgement:
21 * This product includes software developed by the University of
22 * California, Berkeley and its contributors.
23 * 4. Neither the name of the University nor the names of its contributors
24 * may be used to endorse or promote products derived from this software
25 * without specific prior written permission.
27 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
28 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
29 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
30 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
31 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
32 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
33 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
34 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
35 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
36 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
39 * @(#)ufs_readwrite.c 8.7 (Berkeley) 1/21/94
40 * $FreeBSD: src/sys/gnu/ext2fs/ext2_readwrite.c,v 1.18.2.2 2000/12/22 18:44:33 dillon Exp $
43 #define BLKSIZE(a, b, c) blksize(a, b, c)
46 * Vnode op for reading.
48 * ext2_read(struct vnode *a_vp, struct uio *a_uio, int a_ioflag,
49 * struct ucred *a_cred)
53 ext2_read(struct vop_read_args *ap)
58 struct ext2_sb_info *fs;
63 long size, xfersize, blkoffset;
64 int error, orig_resid;
65 int seqcount = ap->a_ioflag >> 16;
72 if (uio->uio_rw != UIO_READ)
73 panic("ext2_read: mode");
75 if (vp->v_type == VLNK) {
76 if ((int)ip->i_size < vp->v_mount->mnt_maxsymlinklen)
77 panic("ext2_read: short symlink");
78 } else if (vp->v_type != VREG && vp->v_type != VDIR)
79 panic("ext2_read: type %d", vp->v_type);
83 if ((u_quad_t)uio->uio_offset > fs->fs_maxfilesize)
87 orig_resid = uio->uio_resid;
88 for (error = 0, bp = NULL; uio->uio_resid > 0; bp = NULL) {
89 if ((bytesinfile = ip->i_size - uio->uio_offset) <= 0)
91 lbn = lblkno(fs, uio->uio_offset);
93 nextloffset = lblktodoff(fs, nextlbn);
94 size = BLKSIZE(fs, ip, lbn);
95 blkoffset = blkoff(fs, uio->uio_offset);
97 xfersize = fs->s_frag_size - blkoffset;
98 if (uio->uio_resid < xfersize)
99 xfersize = uio->uio_resid;
100 if (bytesinfile < xfersize)
101 xfersize = bytesinfile;
103 if (nextloffset >= ip->i_size) {
104 error = bread(vp, lblktodoff(fs, lbn), size, &bp);
105 } else if ((vp->v_mount->mnt_flag & MNT_NOCLUSTERR) == 0) {
106 error = cluster_read(vp, (off_t)ip->i_size,
107 lblktodoff(fs, lbn), size,
109 (ap->a_ioflag >> 16) * BKVASIZE,
111 } else if (seqcount > 1) {
112 int nextsize = BLKSIZE(fs, ip, nextlbn);
113 error = breadn(vp, lblktodoff(fs, lbn),
114 size, &nextloffset, &nextsize, 1, &bp);
116 error = bread(vp, lblktodoff(fs, lbn), size, &bp);
125 * We should only get non-zero b_resid when an I/O error
126 * has occurred, which should cause us to break above.
127 * However, if the short read did not cause an error,
128 * then we want to ensure that we do not uiomove bad
129 * or uninitialized data.
132 if (size < xfersize) {
137 error = uiomove((char *)bp->b_data + blkoffset,
146 if (orig_resid > 0 && (error == 0 || uio->uio_resid != orig_resid) &&
147 (vp->v_mount->mnt_flag & MNT_NOATIME) == 0)
148 ip->i_flag |= IN_ACCESS;
153 * Vnode op for writing.
155 * ext2_write(struct vnode *a_vp, struct uio *a_uio, int a_ioflag,
156 * struct ucred *a_cred)
159 ext2_write(struct vop_write_args *ap)
164 struct ext2_sb_info *fs;
170 int blkoffset, error, flags, ioflag, resid, size, xfersize;
172 ioflag = ap->a_ioflag;
173 seqcount = ap->a_ioflag >> 16;
179 if (uio->uio_rw != UIO_WRITE)
180 panic("ext2_write: mode");
183 switch (vp->v_type) {
185 if (ioflag & IO_APPEND)
186 uio->uio_offset = ip->i_size;
187 if ((ip->i_flags & APPEND) && uio->uio_offset != ip->i_size)
193 if ((ioflag & IO_SYNC) == 0)
194 panic("ext2_write: nonsync dir write");
197 panic("ext2_write: type");
202 if (uio->uio_offset < 0 ||
203 (u_quad_t)uio->uio_offset + uio->uio_resid > fs->fs_maxfilesize)
207 * Maybe this should be above the vnode op call, but so long as
208 * file servers have no limits, I don't think it matters.
211 if (vp->v_type == VREG && td && td->td_proc &&
212 uio->uio_offset + uio->uio_resid >
213 td->td_proc->p_rlimit[RLIMIT_FSIZE].rlim_cur) {
214 lwpsignal(td->td_proc, td->td_lwp, SIGXFSZ);
218 resid = uio->uio_resid;
220 flags = ioflag & IO_SYNC ? B_SYNC : 0;
222 for (error = 0; uio->uio_resid > 0;) {
223 lbn = lblkno(fs, uio->uio_offset);
224 blkoffset = blkoff(fs, uio->uio_offset);
225 xfersize = fs->s_frag_size - blkoffset;
226 if (uio->uio_resid < xfersize)
227 xfersize = uio->uio_resid;
229 if (uio->uio_offset + xfersize > ip->i_size)
230 vnode_pager_setsize(vp, uio->uio_offset + xfersize);
233 * Avoid a data-consistency race between write() and mmap()
234 * by ensuring that newly allocated blocks are zerod. The
235 * race can occur even in the case where the write covers
238 * Just set B_CLRBUF unconditionally, even if the write
239 * covers the whole block. This also handles the UIO_NOCOPY
240 * case. See ffs_write() in ufs for a more sophisticated
244 error = ext2_balloc(ip, lbn, blkoffset + xfersize,
245 ap->a_cred, &bp, flags);
249 if (uio->uio_offset + xfersize > ip->i_size) {
250 ip->i_size = uio->uio_offset + xfersize;
253 size = BLKSIZE(fs, ip, lbn) - bp->b_resid;
257 error = uiomove((char *)bp->b_data + blkoffset, xfersize, uio);
258 if ((ioflag & IO_VMIO) &&
259 LIST_FIRST(&bp->b_dep) == NULL) /* in ext2fs? */
260 bp->b_flags |= B_RELBUF;
262 if (ioflag & IO_SYNC) {
264 } else if (xfersize + blkoffset == fs->s_frag_size) {
265 if ((vp->v_mount->mnt_flag & MNT_NOCLUSTERW) == 0) {
266 bp->b_flags |= B_CLUSTEROK;
267 cluster_write(bp, (off_t)ip->i_size, vp->v_mount->mnt_stat.f_iosize, seqcount);
272 bp->b_flags |= B_CLUSTEROK;
275 if (error || xfersize == 0)
277 ip->i_flag |= IN_CHANGE | IN_UPDATE;
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
284 if (resid > uio->uio_resid && ap->a_cred && ap->a_cred->cr_uid != 0)
285 ip->i_mode &= ~(ISUID | ISGID);
287 if (ioflag & IO_UNIT) {
288 EXT2_TRUNCATE(vp, osize, ioflag & IO_SYNC, ap->a_cred);
289 uio->uio_offset -= resid - uio->uio_resid;
290 uio->uio_resid = resid;
292 } else if (resid > uio->uio_resid && (ioflag & IO_SYNC))
293 error = EXT2_UPDATE(vp, 1);