1 /* 2 * Copyright (c) 1982, 1986, 1989, 1993 3 * The Regents of the University of California. All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 3. All advertising materials mentioning features or use of this software 14 * must display the following acknowledgement: 15 * This product includes software developed by the University of 16 * California, Berkeley and its contributors. 17 * 4. Neither the name of the University nor the names of its contributors 18 * may be used to endorse or promote products derived from this software 19 * without specific prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 * 33 * @(#)ffs_vnops.c 8.15 (Berkeley) 5/14/95 34 * $FreeBSD$ 35 */ 36 37 #include "opt_ffs.h" 38 39 #include <sys/param.h> 40 #include <sys/systm.h> 41 #include <sys/resourcevar.h> 42 #include <sys/signalvar.h> 43 #include <sys/kernel.h> 44 #include <sys/stat.h> 45 #include <sys/bio.h> 46 #include <sys/buf.h> 47 #include <sys/proc.h> 48 #include <sys/mount.h> 49 #include <sys/vnode.h> 50 #include <sys/conf.h> 51 52 #include <machine/limits.h> 53 54 #include <vm/vm.h> 55 #include <vm/vm_page.h> 56 #include <vm/vm_object.h> 57 #include <vm/vm_extern.h> 58 59 #include <ufs/ufs/extattr.h> 60 #include <ufs/ufs/quota.h> 61 #include <ufs/ufs/inode.h> 62 #include <ufs/ufs/ufsmount.h> 63 #include <ufs/ufs/ufs_extern.h> 64 65 #include <ufs/ffs/fs.h> 66 #include <ufs/ffs/ffs_extern.h> 67 68 int ffs_fsync __P((struct vop_fsync_args *)); 69 static int ffs_getpages __P((struct vop_getpages_args *)); 70 static int ffs_putpages __P((struct vop_putpages_args *)); 71 static int ffs_read __P((struct vop_read_args *)); 72 static int ffs_write __P((struct vop_write_args *)); 73 74 /* Global vfs data structures for ufs. */ 75 vop_t **ffs_vnodeop_p; 76 static struct vnodeopv_entry_desc ffs_vnodeop_entries[] = { 77 { &vop_default_desc, (vop_t *) ufs_vnoperate }, 78 { &vop_fsync_desc, (vop_t *) ffs_fsync }, 79 { &vop_getpages_desc, (vop_t *) ffs_getpages }, 80 { &vop_putpages_desc, (vop_t *) ffs_putpages }, 81 { &vop_read_desc, (vop_t *) ffs_read }, 82 { &vop_balloc_desc, (vop_t *) ffs_balloc }, 83 { &vop_reallocblks_desc, (vop_t *) ffs_reallocblks }, 84 { &vop_write_desc, (vop_t *) ffs_write }, 85 #ifdef FFS_EXTATTR 86 { &vop_getextattr_desc, (vop_t *) ufs_vop_getextattr }, 87 { &vop_setextattr_desc, (vop_t *) ufs_vop_setextattr }, 88 #endif 89 { NULL, NULL } 90 }; 91 static struct vnodeopv_desc ffs_vnodeop_opv_desc = 92 { &ffs_vnodeop_p, ffs_vnodeop_entries }; 93 94 vop_t **ffs_specop_p; 95 static struct vnodeopv_entry_desc ffs_specop_entries[] = { 96 { &vop_default_desc, (vop_t *) ufs_vnoperatespec }, 97 { &vop_fsync_desc, (vop_t *) ffs_fsync }, 98 { &vop_copyonwrite_desc, (vop_t *) ffs_copyonwrite }, 99 #ifdef FFS_EXTATTR 100 { &vop_getextattr_desc, (vop_t *) ufs_vop_getextattr }, 101 { &vop_setextattr_desc, (vop_t *) ufs_vop_setextattr }, 102 #endif 103 { NULL, NULL } 104 }; 105 static struct vnodeopv_desc ffs_specop_opv_desc = 106 { &ffs_specop_p, ffs_specop_entries }; 107 108 vop_t **ffs_fifoop_p; 109 static struct vnodeopv_entry_desc ffs_fifoop_entries[] = { 110 { &vop_default_desc, (vop_t *) ufs_vnoperatefifo }, 111 #ifdef FFS_EXTATTR 112 { &vop_getextattr_desc, (vop_t *) ufs_vop_getextattr }, 113 { &vop_setextattr_desc, (vop_t *) ufs_vop_setextattr }, 114 #endif 115 { &vop_fsync_desc, (vop_t *) ffs_fsync }, 116 { NULL, NULL } 117 }; 118 static struct vnodeopv_desc ffs_fifoop_opv_desc = 119 { &ffs_fifoop_p, ffs_fifoop_entries }; 120 121 VNODEOP_SET(ffs_vnodeop_opv_desc); 122 VNODEOP_SET(ffs_specop_opv_desc); 123 VNODEOP_SET(ffs_fifoop_opv_desc); 124 125 #include <ufs/ufs/ufs_readwrite.c> 126 127 /* 128 * Synch an open file. 129 */ 130 /* ARGSUSED */ 131 int 132 ffs_fsync(ap) 133 struct vop_fsync_args /* { 134 struct vnode *a_vp; 135 struct ucred *a_cred; 136 int a_waitfor; 137 struct proc *a_p; 138 } */ *ap; 139 { 140 struct vnode *vp = ap->a_vp; 141 struct inode *ip = VTOI(vp); 142 struct buf *bp; 143 struct buf *nbp; 144 int s, error, wait, passes, skipmeta; 145 daddr_t lbn; 146 147 /* 148 * Snapshots have to be unlocked so they do not deadlock 149 * checking whether they need to copy their written buffers. 150 * We always hold a reference, so they cannot be removed 151 * out from underneath us. 152 */ 153 if (ip->i_flags & SF_SNAPSHOT) 154 VOP_UNLOCK(vp, 0, ap->a_p); 155 wait = (ap->a_waitfor == MNT_WAIT); 156 if (vn_isdisk(vp, NULL)) { 157 lbn = INT_MAX; 158 if (vp->v_rdev->si_mountpoint != NULL && 159 (vp->v_rdev->si_mountpoint->mnt_flag & MNT_SOFTDEP)) 160 softdep_fsync_mountdev(vp); 161 } else { 162 lbn = lblkno(ip->i_fs, (ip->i_size + ip->i_fs->fs_bsize - 1)); 163 } 164 165 /* 166 * Flush all dirty buffers associated with a vnode. 167 */ 168 passes = NIADDR + 1; 169 skipmeta = 0; 170 if (wait) 171 skipmeta = 1; 172 s = splbio(); 173 loop: 174 for (bp = TAILQ_FIRST(&vp->v_dirtyblkhd); bp; 175 bp = TAILQ_NEXT(bp, b_vnbufs)) 176 bp->b_flags &= ~B_SCANNED; 177 for (bp = TAILQ_FIRST(&vp->v_dirtyblkhd); bp; bp = nbp) { 178 nbp = TAILQ_NEXT(bp, b_vnbufs); 179 /* 180 * Reasons to skip this buffer: it has already been considered 181 * on this pass, this pass is the first time through on a 182 * synchronous flush request and the buffer being considered 183 * is metadata, the buffer has dependencies that will cause 184 * it to be redirtied and it has not already been deferred, 185 * or it is already being written. 186 */ 187 if ((bp->b_flags & B_SCANNED) != 0) 188 continue; 189 bp->b_flags |= B_SCANNED; 190 if ((skipmeta == 1 && bp->b_lblkno < 0)) 191 continue; 192 if (!wait && LIST_FIRST(&bp->b_dep) != NULL && 193 (bp->b_flags & B_DEFERRED) == 0 && 194 buf_countdeps(bp, 0)) { 195 bp->b_flags |= B_DEFERRED; 196 continue; 197 } 198 if (BUF_LOCK(bp, LK_EXCLUSIVE | LK_NOWAIT)) 199 continue; 200 if ((bp->b_flags & B_DELWRI) == 0) 201 panic("ffs_fsync: not dirty"); 202 if (vp != bp->b_vp) 203 panic("ffs_fsync: vp != vp->b_vp"); 204 /* 205 * If this is a synchronous flush request, or it is not a 206 * file or device, start the write on this buffer immediatly. 207 */ 208 if (wait || (vp->v_type != VREG && vp->v_type != VBLK)) { 209 210 /* 211 * On our final pass through, do all I/O synchronously 212 * so that we can find out if our flush is failing 213 * because of write errors. 214 */ 215 if (passes > 0 || !wait) { 216 if ((bp->b_flags & B_CLUSTEROK) && !wait) { 217 BUF_UNLOCK(bp); 218 (void) vfs_bio_awrite(bp); 219 } else { 220 bremfree(bp); 221 splx(s); 222 (void) bawrite(bp); 223 s = splbio(); 224 } 225 } else { 226 bremfree(bp); 227 splx(s); 228 if ((error = bwrite(bp)) != 0) 229 return (error); 230 s = splbio(); 231 } 232 } else if ((vp->v_type == VREG) && (bp->b_lblkno >= lbn)) { 233 /* 234 * If the buffer is for data that has been truncated 235 * off the file, then throw it away. 236 */ 237 bremfree(bp); 238 bp->b_flags |= B_INVAL | B_NOCACHE; 239 splx(s); 240 brelse(bp); 241 s = splbio(); 242 } else { 243 BUF_UNLOCK(bp); 244 vfs_bio_awrite(bp); 245 } 246 /* 247 * Since we may have slept during the I/O, we need 248 * to start from a known point. 249 */ 250 nbp = TAILQ_FIRST(&vp->v_dirtyblkhd); 251 } 252 /* 253 * If we were asked to do this synchronously, then go back for 254 * another pass, this time doing the metadata. 255 */ 256 if (skipmeta) { 257 skipmeta = 0; 258 goto loop; 259 } 260 261 if (wait) { 262 while (vp->v_numoutput) { 263 vp->v_flag |= VBWAIT; 264 (void) tsleep((caddr_t)&vp->v_numoutput, 265 PRIBIO + 4, "ffsfsn", 0); 266 } 267 268 /* 269 * Ensure that any filesystem metatdata associated 270 * with the vnode has been written. 271 */ 272 splx(s); 273 if ((error = softdep_sync_metadata(ap)) != 0) 274 return (error); 275 s = splbio(); 276 277 if (!TAILQ_EMPTY(&vp->v_dirtyblkhd)) { 278 /* 279 * Block devices associated with filesystems may 280 * have new I/O requests posted for them even if 281 * the vnode is locked, so no amount of trying will 282 * get them clean. Thus we give block devices a 283 * good effort, then just give up. For all other file 284 * types, go around and try again until it is clean. 285 */ 286 if (passes > 0) { 287 passes -= 1; 288 goto loop; 289 } 290 #ifdef DIAGNOSTIC 291 if (!vn_isdisk(vp, NULL)) 292 vprint("ffs_fsync: dirty", vp); 293 #endif 294 } 295 } 296 splx(s); 297 error = UFS_UPDATE(vp, wait); 298 if (ip->i_flags & SF_SNAPSHOT) 299 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, ap->a_p); 300 return (error); 301 } 302