/* * Copyright (c) 1989, 1991, 1993, 1994 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the University of * California, Berkeley and its contributors. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)lfs_vfsops.c 8.20 (Berkeley) 6/10/95 * $Id: lfs_vfsops.c,v 1.24 1997/10/12 20:26:17 phk Exp $ */ #include "opt_quota.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include MALLOC_DEFINE(M_LFSNODE, "LFS node", "LFS vnode private part"); MALLOC_DEFINE(M_SEGMENT, "LFS segment", "Segment for LFS"); static int lfs_fhtovp __P((struct mount *, struct fid *, struct sockaddr *, struct vnode **, int *, struct ucred **)); static int lfs_mount __P((struct mount *, char *, caddr_t, struct nameidata *, struct proc *)); static int lfs_mountfs __P((struct vnode *, struct mount *, struct proc *)); static int lfs_statfs __P((struct mount *, struct statfs *, struct proc *)); static int lfs_sync __P((struct mount *, int, struct ucred *, struct proc *)); static int lfs_unmount __P((struct mount *, int, struct proc *)); static int lfs_vget __P((struct mount *, ino_t, struct vnode **)); static int lfs_vptofh __P((struct vnode *, struct fid *)); struct vfsops lfs_vfsops = { lfs_mount, ufs_start, lfs_unmount, ufs_root, ufs_quotactl, lfs_statfs, lfs_sync, lfs_vget, lfs_fhtovp, lfs_vptofh, lfs_init, #ifdef notyet lfs_sysctl, #endif }; VFS_SET(lfs_vfsops, lfs, MOUNT_LFS, 0); /* * Called by main() when ufs is going to be mounted as root. */ int lfs_mountroot() { struct fs *fs; struct mount *mp; struct proc *p = curproc; /* XXX */ int error; if ((error = bdevvp(rootdev, &rootvp))) { printf("lfs_mountroot: can't find rootvp"); return (error); } if (error = vfs_rootmountalloc("lfs", "root_device", &mp)) return (error); if (error = lfs_mountfs(rootvp, mp, p)) { mp->mnt_vfc->vfc_refcount--; vfs_unbusy(mp, p); free(mp, M_MOUNT); return (error); } simple_lock(&mountlist_slock); CIRCLEQ_INSERT_HEAD(&mountlist, mp, mnt_list); simple_unlock(&mountlist_slock); (void)lfs_statfs(mp, &mp->mnt_stat, p); vfs_unbusy(mp, p); return (0); } /* * lfs_mount * * Called when mounting local physical media * * PARAMETERS: * mountroot * mp mount point structure * path NULL (flag for root mount!!!) * data * ndp * p process (user credentials check [statfs]) * * mount * mp mount point structure * path path to mount point * data pointer to argument struct in user space * ndp mount point namei() return (used for * credentials on reload), reused to look * up block device. * p process (user credentials check) * * RETURNS: 0 Success * !0 error number (errno.h) * * LOCK STATE: * * ENTRY * mount point is locked * EXIT * mount point is locked * * NOTES: * A NULL path can be used for a flag since the mount * system call will fail with EFAULT in copyinstr in * namei() if it is a genuine NULL from the user. * * Root mounts are not currently supported. */ static int lfs_mount(mp, path, data, ndp, p) register struct mount *mp; char *path; caddr_t data; struct nameidata *ndp; struct proc *p; { struct vnode *devvp; struct ufs_args args; struct ufsmount *ump = 0; register struct lfs *fs; /* LFS */ u_int size; int err; mode_t accessmode; /* * Use NULL path to flag a root mount */ if( path == NULL) { /* *** * Mounting root file system *** */ /* XXX -- implement*/ panic("lfs_mount: NULL path not implemented"); } /* *** * Mounting non-root file system or updating a file system *** */ /* copy in user arguments*/ if (err = copyin(data, (caddr_t)&args, sizeof (struct ufs_args))) goto error_1; /* Until LFS can do NFS right. XXX */ if (args.export.ex_flags & MNT_EXPORTED) { err = EINVAL; goto error_1; } /* * If updating, check whether changing from read-only to * read/write; if there is no device name, that's all we do. */ if (mp->mnt_flag & MNT_UPDATE) { ump = VFSTOUFS(mp); if (fs->lfs_ronly && (mp->mnt_flag & MNT_WANTRDWR)) { /* * If upgrade to read-write by non-root, then verify * that user has necessary permissions on the device. */ if (p->p_ucred->cr_uid != 0) { vn_lock(ump->um_devvp, LK_EXCLUSIVE | LK_RETRY, p); if (err = VOP_ACCESS(ump->um_devvp, VREAD | VWRITE, p->p_ucred, p)) { VOP_UNLOCK(ump->um_devvp, 0, p); return (err); } VOP_UNLOCK(ump->um_devvp, 0, p); } fs->lfs_ronly = 0; } if (args.fspec == 0) { /* * Process export requests. Jumping to "success" * will return the vfs_export() error code. */ err = vfs_export(mp, &ump->um_export, &args.export); goto success; } } /* * Not an update, or updating the name: look up the name * and verify that it refers to a sensible block device. */ NDINIT(ndp, LOOKUP, FOLLOW, UIO_USERSPACE, args.fspec, p); if (err = namei(ndp)) goto error_1; devvp = ndp->ni_vp; if (devvp->v_type != VBLK) { err = ENOTBLK; goto error_2; } if (major(devvp->v_rdev) >= nblkdev) { err = ENXIO; goto error_2; } /* * If mount by non-root, then verify that user has necessary * permissions on the device. */ if (p->p_ucred->cr_uid != 0) { accessmode = VREAD; if ((mp->mnt_flag & MNT_RDONLY) == 0) accessmode |= VWRITE; vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY, p); if (err = VOP_ACCESS(devvp, accessmode, p->p_ucred, p)) { vput(devvp); return (err); } VOP_UNLOCK(devvp, 0, p); } if ((mp->mnt_flag & MNT_UPDATE) == 0) err = lfs_mountfs(devvp, mp, p); /* LFS */ else { if (devvp != ump->um_devvp) err = EINVAL; /* needs translation */ else vrele(devvp); } if (err) { goto error_2; } ump = VFSTOUFS(mp); fs = ump->um_lfs; /* LFS */ #ifdef NOTLFS /* LFS */ (void) copyinstr(path, fs->fs_fsmnt, sizeof(fs->fs_fsmnt) - 1, &size); bzero(fs->fs_fsmnt + size, sizeof(fs->fs_fsmnt) - size); bcopy((caddr_t)fs->fs_fsmnt, (caddr_t)mp->mnt_stat.f_mntonname, MNAMELEN); (void) copyinstr(args.fspec, mp->mnt_stat.f_mntfromname, MNAMELEN - 1, &size); bzero(mp->mnt_stat.f_mntfromname + size, MNAMELEN - size); (void) ufs_statfs(mp, &mp->mnt_stat, p); #else (void)copyinstr(path, fs->lfs_fsmnt, sizeof(fs->lfs_fsmnt) - 1, &size); bzero(fs->lfs_fsmnt + size, sizeof(fs->lfs_fsmnt) - size); bcopy((caddr_t)fs->lfs_fsmnt, (caddr_t)mp->mnt_stat.f_mntonname, MNAMELEN); (void) copyinstr(args.fspec, mp->mnt_stat.f_mntfromname, MNAMELEN - 1, &size); bzero(mp->mnt_stat.f_mntfromname + size, MNAMELEN - size); (void) lfs_statfs(mp, &mp->mnt_stat, p); #endif error_2: /* error with devvp held*/ /* release devvp before failing*/ vrele(devvp); error_1: /* no state to back out*/ success: return( err); } /* * Common code for mount and mountroot * LFS specific */ static int lfs_mountfs(devvp, mp, p) register struct vnode *devvp; struct mount *mp; struct proc *p; { register struct lfs *fs; register struct ufsmount *ump; struct vnode *vp; struct buf *bp; struct partinfo dpart; dev_t dev; int error, i, ronly, size; struct ucred *cred; cred = p ? p->p_ucred : NOCRED; /* * Disallow multiple mounts of the same device. * Disallow mounting of a device that is currently in use * (except for root, which might share swap device for miniroot). * Flush out any old buffers remaining from a previous use. */ if (error = vfs_mountedon(devvp)) return (error); if (vcount(devvp) > 1 && devvp != rootvp) return (EBUSY); if (error = vinvalbuf(devvp, V_SAVE, cred, p, 0, 0)) return (error); ronly = (mp->mnt_flag & MNT_RDONLY) != 0; if (error = VOP_OPEN(devvp, ronly ? FREAD : FREAD|FWRITE, FSCRED, p)) return (error); if (VOP_IOCTL(devvp, DIOCGPART, (caddr_t)&dpart, FREAD, cred, p) != 0) size = DEV_BSIZE; else { size = dpart.disklab->d_secsize; #ifdef NEVER_USED dpart.part->p_fstype = FS_LFS; dpart.part->p_fsize = fs->lfs_fsize; /* frag size */ dpart.part->p_frag = fs->lfs_frag; /* frags per block */ dpart.part->p_cpg = fs->lfs_segshift; /* segment shift */ #endif } /* Don't free random space on error. */ bp = NULL; ump = NULL; /* Read in the superblock. */ if (error = bread(devvp, LFS_LABELPAD / size, LFS_SBPAD, cred, &bp)) goto out; fs = (struct lfs *)bp->b_data; /* Check the basics. */ if (fs->lfs_magic != LFS_MAGIC || fs->lfs_bsize > MAXBSIZE || fs->lfs_bsize < sizeof(struct lfs)) { error = EINVAL; /* XXX needs translation */ goto out; } /* Allocate the mount structure, copy the superblock into it. */ ump = (struct ufsmount *)malloc(sizeof *ump, M_UFSMNT, M_WAITOK); bzero(ump, sizeof *ump); ump->um_malloctype = M_LFSNODE; fs = ump->um_lfs = malloc(sizeof(struct lfs), M_UFSMNT, M_WAITOK); bcopy(bp->b_data, fs, sizeof(struct lfs)); if (sizeof(struct lfs) < LFS_SBPAD) /* XXX why? */ bp->b_flags |= B_INVAL; brelse(bp); bp = NULL; /* Set up the I/O information */ fs->lfs_iocount = 0; /* Set up the ifile and lock aflags */ fs->lfs_doifile = 0; fs->lfs_writer = 0; fs->lfs_dirops = 0; fs->lfs_seglock = 0; /* Set the file system readonly/modify bits. */ fs->lfs_ronly = ronly; if (ronly == 0) fs->lfs_fmod = 1; /* Initialize the mount structure. */ dev = devvp->v_rdev; mp->mnt_data = (qaddr_t)ump; mp->mnt_stat.f_fsid.val[0] = (long)dev; mp->mnt_stat.f_fsid.val[1] = lfs_mount_type; mp->mnt_maxsymlinklen = fs->lfs_maxsymlinklen; mp->mnt_flag |= MNT_LOCAL; ump->um_mountp = mp; ump->um_dev = dev; ump->um_devvp = devvp; ump->um_bptrtodb = 0; ump->um_seqinc = 1 << fs->lfs_fsbtodb; ump->um_nindir = fs->lfs_nindir; for (i = 0; i < MAXQUOTAS; i++) ump->um_quotas[i] = NULLVP; devvp->v_specflags |= SI_MOUNTEDON; /* * We use the ifile vnode for almost every operation. Instead of * retrieving it from the hash table each time we retrieve it here, * artificially increment the reference count and keep a pointer * to it in the incore copy of the superblock. */ if (error = VFS_VGET(mp, LFS_IFILE_INUM, &vp)) goto out; fs->lfs_ivnode = vp; VREF(vp); vput(vp); return (0); out: if (bp) brelse(bp); (void)VOP_CLOSE(devvp, ronly ? FREAD : FREAD|FWRITE, cred, p); if (ump) { free(ump->um_lfs, M_UFSMNT); free(ump, M_UFSMNT); mp->mnt_data = (qaddr_t)0; } return (error); } /* * unmount system call */ static int lfs_unmount(mp, mntflags, p) struct mount *mp; int mntflags; struct proc *p; { register struct ufsmount *ump; register struct lfs *fs; int i, error, flags, ronly; flags = 0; if (mntflags & MNT_FORCE) flags |= FORCECLOSE; ump = VFSTOUFS(mp); fs = ump->um_lfs; #ifdef QUOTA if (mp->mnt_flag & MNT_QUOTA) { if (error = vflush(mp, fs->lfs_ivnode, SKIPSYSTEM|flags)) return (error); for (i = 0; i < MAXQUOTAS; i++) { if (ump->um_quotas[i] == NULLVP) continue; quotaoff(p, mp, i); } /* * Here we fall through to vflush again to ensure * that we have gotten rid of all the system vnodes. */ } #endif if (error = vflush(mp, fs->lfs_ivnode, flags)) return (error); fs->lfs_clean = 1; if (error = VFS_SYNC(mp, 1, p->p_ucred, p)) return (error); if (fs->lfs_ivnode->v_dirtyblkhd.lh_first) panic("lfs_unmount: still dirty blocks on ifile vnode"); vrele(fs->lfs_ivnode); vgone(fs->lfs_ivnode); ronly = fs->lfs_ronly; ump->um_devvp->v_specflags &= ~SI_MOUNTEDON; error = VOP_CLOSE(ump->um_devvp, ronly ? FREAD : FREAD|FWRITE, NOCRED, p); vrele(ump->um_devvp); free(fs, M_UFSMNT); free(ump, M_UFSMNT); mp->mnt_data = (qaddr_t)0; mp->mnt_flag &= ~MNT_LOCAL; return (error); } /* * Get file system statistics. */ static int lfs_statfs(mp, sbp, p) struct mount *mp; register struct statfs *sbp; struct proc *p; { register struct lfs *fs; register struct ufsmount *ump; ump = VFSTOUFS(mp); fs = ump->um_lfs; if (fs->lfs_magic != LFS_MAGIC) panic("lfs_statfs: magic"); sbp->f_bsize = fs->lfs_fsize; sbp->f_iosize = fs->lfs_bsize; sbp->f_blocks = dbtofrags(fs,fs->lfs_dsize); sbp->f_bfree = dbtofrags(fs, fs->lfs_bfree); /* * To compute the available space. Subtract the minimum free * from the total number of blocks in the file system. Set avail * to the smaller of this number and fs->lfs_bfree. */ sbp->f_bavail = fs->lfs_dsize * (100 - fs->lfs_minfree) / 100; sbp->f_bavail = sbp->f_bavail > fs->lfs_bfree ? fs->lfs_bfree : sbp->f_bavail; sbp->f_bavail = dbtofrags(fs, sbp->f_bavail); sbp->f_files = fs->lfs_nfiles; sbp->f_ffree = sbp->f_bfree * INOPB(fs); if (sbp != &mp->mnt_stat) { sbp->f_type = mp->mnt_vfc->vfc_typenum; bcopy((caddr_t)mp->mnt_stat.f_mntonname, (caddr_t)&sbp->f_mntonname[0], MNAMELEN); bcopy((caddr_t)mp->mnt_stat.f_mntfromname, (caddr_t)&sbp->f_mntfromname[0], MNAMELEN); } return (0); } /* * Go through the disk queues to initiate sandbagged IO; * go through the inodes to write those that have been modified; * initiate the writing of the super block if it has been modified. * * Note: we are always called with the filesystem marked `MPBUSY'. */ static int lfs_sync(mp, waitfor, cred, p) struct mount *mp; int waitfor; struct ucred *cred; struct proc *p; { int error; /* All syncs must be checkpoints until roll-forward is implemented. */ error = lfs_segwrite(mp, SEGM_CKP | (waitfor ? SEGM_SYNC : 0)); #ifdef QUOTA qsync(mp); #endif return (error); } /* * Look up an LFS dinode number to find its incore vnode. If not already * in core, read it in from the specified device. Return the inode locked. * Detection and handling of mount points must be done by the calling routine. */ static int lfs_vget(mp, ino, vpp) struct mount *mp; ino_t ino; struct vnode **vpp; { register struct lfs *fs; register struct inode *ip; struct buf *bp; struct ifile *ifp; struct vnode *vp; struct ufsmount *ump; ufs_daddr_t daddr; dev_t dev; int error; ump = VFSTOUFS(mp); dev = ump->um_dev; if ((*vpp = ufs_ihashget(dev, ino)) != NULL) return (0); /* Translate the inode number to a disk address. */ fs = ump->um_lfs; if (ino == LFS_IFILE_INUM) daddr = fs->lfs_idaddr; else { LFS_IENTRY(ifp, fs, ino, bp); daddr = ifp->if_daddr; brelse(bp); if (daddr == LFS_UNUSED_DADDR) return (ENOENT); } /* Allocate new vnode/inode. */ if (error = lfs_vcreate(mp, ino, &vp)) { *vpp = NULL; return (error); } /* * Put it onto its hash chain and lock it so that other requests for * this inode will block if they arrive while we are sleeping waiting * for old data structures to be purged or for the contents of the * disk portion of this inode to be read. */ ip = VTOI(vp); ufs_ihashins(ip); /* * XXX * This may not need to be here, logically it should go down with * the i_devvp initialization. * Ask Kirk. */ ip->i_lfs = ump->um_lfs; /* Read in the disk contents for the inode, copy into the inode. */ if (error = bread(ump->um_devvp, daddr, (int)fs->lfs_bsize, NOCRED, &bp)) { /* * The inode does not contain anything useful, so it would * be misleading to leave it on its hash chain. With mode * still zero, it will be unlinked and returned to the free * list by vput(). */ vput(vp); brelse(bp); *vpp = NULL; return (error); } ip->i_din = *lfs_ifind(fs, ino, (struct dinode *)bp->b_data); brelse(bp); /* * Initialize the vnode from the inode, check for aliases. In all * cases re-init ip, the underlying vnode/inode may have changed. */ if (error = ufs_vinit(mp, lfs_specop_p, LFS_FIFOOPS, &vp)) { vput(vp); *vpp = NULL; return (error); } /* * Finish inode initialization now that aliasing has been resolved. */ ip->i_devvp = ump->um_devvp; VREF(ip->i_devvp); *vpp = vp; return (0); } /* * File handle to vnode * * Have to be really careful about stale file handles: * - check that the inode number is valid * - call lfs_vget() to get the locked inode * - check for an unallocated inode (i_mode == 0) * - check that the given client host has export rights and return * those rights via. exflagsp and credanonp * * XXX * use ifile to see if inode is allocated instead of reading off disk * what is the relationship between my generational number and the NFS * generational number. */ static int lfs_fhtovp(mp, fhp, nam, vpp, exflagsp, credanonp) register struct mount *mp; struct fid *fhp; struct sockaddr *nam; struct vnode **vpp; int *exflagsp; struct ucred **credanonp; { register struct ufid *ufhp; ufhp = (struct ufid *)fhp; if (ufhp->ufid_ino < ROOTINO) return (ESTALE); return (ufs_check_export(mp, ufhp, nam, vpp, exflagsp, credanonp)); } /* * Vnode pointer to File handle */ /* ARGSUSED */ static int lfs_vptofh(vp, fhp) struct vnode *vp; struct fid *fhp; { register struct inode *ip; register struct ufid *ufhp; ip = VTOI(vp); ufhp = (struct ufid *)fhp; ufhp->ufid_len = sizeof(struct ufid); ufhp->ufid_ino = ip->i_number; ufhp->ufid_gen = ip->i_gen; return (0); } /* * Initialize the filesystem, most work done by ufs_init. */ int lfs_mount_type; int lfs_init(vfsp) struct vfsconf *vfsp; { lfs_mount_type = vfsp->vfc_typenum; return (ufs_init(vfsp)); }