ufs/ffs/ffs_vfsops.c

/*
 * Copyright (c) 1989, 1991 The Regents of the University of California.
 * All rights reserved.
 *
 * %sccs.include.redist.c%
 *
 *	@(#)ffs_vfsops.c	7.81 (Berkeley) 10/08/92
 */

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/namei.h>
#include <sys/proc.h>
#include <sys/kernel.h>
#include <sys/vnode.h>
#include <sys/socket.h>
#include <sys/mount.h>
#include <sys/buf.h>
#include <sys/mbuf.h>
#include <sys/file.h>
#include <sys/disklabel.h>
#include <sys/ioctl.h>
#include <sys/errno.h>
#include <sys/malloc.h>

#include <miscfs/specfs/specdev.h>

#include <ufs/ufs/quota.h>
#include <ufs/ufs/ufsmount.h>
#include <ufs/ufs/inode.h>
#include <ufs/ufs/ufs_extern.h>

#include <ufs/ffs/fs.h>
#include <ufs/ffs/ffs_extern.h>

int ffs_sbupdate __P((struct ufsmount *, int));

struct vfsops ufs_vfsops = {
	ffs_mount,
	ufs_start,
	ffs_unmount,
	ffs_root,
	ufs_quotactl,
	ffs_statfs,
	ffs_sync,
	ffs_vget,
	ffs_fhtovp,
	ffs_vptofh,
	ffs_init,
};

extern u_long nextgennumber;

/*
 * Called by main() when ufs is going to be mounted as root.
 *
 * Name is updated by mount(8) after booting.
 */
#define ROOTNAME	"root_device"

ffs_mountroot()
{
	extern struct vnode *rootvp;
	register struct fs *fs;
	register struct mount *mp;
	struct proc *p = curproc;	/* XXX */
	struct ufsmount *ump;
	u_int size;
	int error;

	/*
	 * Get vnodes for swapdev and rootdev.
	 */
	if (bdevvp(swapdev, &swapdev_vp) || bdevvp(rootdev, &rootvp))
		panic("ffs_mountroot: can't setup bdevvp's");

	mp = malloc((u_long)sizeof(struct mount), M_MOUNT, M_WAITOK);
	bzero((char *)mp, (u_long)sizeof(struct mount));
	mp->mnt_op = &ufs_vfsops;
	mp->mnt_flag = MNT_RDONLY;
	if (error = ffs_mountfs(rootvp, mp, p)) {
		free(mp, M_MOUNT);
		return (error);
	}
	if (error = vfs_lock(mp)) {
		(void)ffs_unmount(mp, 0, p);
		free(mp, M_MOUNT);
		return (error);
	}
	rootfs = mp;
	mp->mnt_next = mp;
	mp->mnt_prev = mp;
	mp->mnt_vnodecovered = NULLVP;
	ump = VFSTOUFS(mp);
	fs = ump->um_fs;
	bzero(fs->fs_fsmnt, sizeof(fs->fs_fsmnt));
	fs->fs_fsmnt[0] = '/';
	bcopy((caddr_t)fs->fs_fsmnt, (caddr_t)mp->mnt_stat.f_mntonname,
	    MNAMELEN);
	(void) copystr(ROOTNAME, mp->mnt_stat.f_mntfromname, MNAMELEN - 1,
	    &size);
	bzero(mp->mnt_stat.f_mntfromname + size, MNAMELEN - size);
	(void)ffs_statfs(mp, &mp->mnt_stat, p);
	vfs_unlock(mp);
	inittodr(fs->fs_time);
	return (0);
}

/*
 * VFS Operations.
 *
 * mount system call
 */
int
ffs_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;
	register struct fs *fs;
	u_int size;
	int error;

	if (error = copyin(data, (caddr_t)&args, sizeof (struct ufs_args)))
		return (error);
	/*
	 * 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);
		fs = ump->um_fs;
		if (fs->fs_ronly && (mp->mnt_flag & MNT_RDONLY) == 0)
			fs->fs_ronly = 0;
		if (args.fspec == 0) {
			/*
			 * Process export requests.
			 */
			if (args.exflags & MNT_EXPORTED) {
				if (error = ufs_hang_addrlist(mp, &args))
					return (error);
				mp->mnt_flag |= MNT_EXPORTED;
			}
			if (args.exflags & MNT_DELEXPORT) {
				ufs_free_addrlist(ump);
				mp->mnt_flag &=
				    ~(MNT_EXPORTED | MNT_DEFEXPORTED);
			}
			return (0);
		}
	}
	/*
	 * 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 (error = namei(ndp))
		return (error);
	devvp = ndp->ni_vp;

	if (devvp->v_type != VBLK) {
		vrele(devvp);
		return (ENOTBLK);
	}
	if (major(devvp->v_rdev) >= nblkdev) {
		vrele(devvp);
		return (ENXIO);
	}
	if ((mp->mnt_flag & MNT_UPDATE) == 0)
		error = ffs_mountfs(devvp, mp, p);
	else {
		if (devvp != ump->um_devvp)
			error = EINVAL;	/* needs translation */
		else
			vrele(devvp);
	}
	if (error) {
		vrele(devvp);
		return (error);
	}
	ump = VFSTOUFS(mp);
	fs = ump->um_fs;
	(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)ffs_statfs(mp, &mp->mnt_stat, p);
	return (0);
}

/*
 * Common code for mount and mountroot
 */
int
ffs_mountfs(devvp, mp, p)
	register struct vnode *devvp;
	struct mount *mp;
	struct proc *p;
{
	register struct ufsmount *ump;
	struct buf *bp;
	register struct fs *fs;
	dev_t dev = devvp->v_rdev;
	struct partinfo dpart;
	caddr_t base, space;
	int havepart = 0, blks;
	int error, i, size;
	int ronly;
	extern struct vnode *rootvp;

	/*
	 * 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 = ufs_mountedon(devvp))
		return (error);
	if (vcount(devvp) > 1 && devvp != rootvp)
		return (EBUSY);
	if (error = vinvalbuf(devvp, V_SAVE, p->p_ucred, p))
		return (error);

	ronly = (mp->mnt_flag & MNT_RDONLY) != 0;
	if (error = VOP_OPEN(devvp, ronly ? FREAD : FREAD|FWRITE, NOCRED, p))
		return (error);
	if (VOP_IOCTL(devvp, DIOCGPART, (caddr_t)&dpart, FREAD, NOCRED, p) != 0)
		size = DEV_BSIZE;
	else {
		havepart = 1;
		size = dpart.disklab->d_secsize;
	}

	bp = NULL;
	ump = NULL;
	if (error = bread(devvp, SBLOCK, SBSIZE, NOCRED, &bp))
		goto out;
	fs = bp->b_un.b_fs;
	if (fs->fs_magic != FS_MAGIC || fs->fs_bsize > MAXBSIZE ||
	    fs->fs_bsize < sizeof(struct fs)) {
		error = EINVAL;		/* XXX needs translation */
		goto out;
	}
	ump = malloc(sizeof *ump, M_UFSMNT, M_WAITOK);
	bzero((caddr_t)ump, sizeof *ump);
	ump->um_fs = malloc((u_long)fs->fs_sbsize, M_UFSMNT,
	    M_WAITOK);
	bcopy((caddr_t)bp->b_un.b_addr, (caddr_t)ump->um_fs,
	   (u_int)fs->fs_sbsize);
	if (fs->fs_sbsize < SBSIZE)
		bp->b_flags |= B_INVAL;
	brelse(bp);
	bp = NULL;
	fs = ump->um_fs;
	fs->fs_ronly = ronly;
	if (ronly == 0)
		fs->fs_fmod = 1;
	if (havepart) {
		dpart.part->p_fstype = FS_BSDFFS;
		dpart.part->p_fsize = fs->fs_fsize;
		dpart.part->p_frag = fs->fs_frag;
		dpart.part->p_cpg = fs->fs_cpg;
	}
	blks = howmany(fs->fs_cssize, fs->fs_fsize);
	base = space = malloc((u_long)fs->fs_cssize, M_UFSMNT,
	    M_WAITOK);
	for (i = 0; i < blks; i += fs->fs_frag) {
		size = fs->fs_bsize;
		if (i + fs->fs_frag > blks)
			size = (blks - i) * fs->fs_fsize;
		error = bread(devvp, fsbtodb(fs, fs->fs_csaddr + i), size,
			NOCRED, &bp);
		if (error) {
			free(base, M_UFSMNT);
			goto out;
		}
		bcopy((caddr_t)bp->b_un.b_addr, space, (u_int)size);
		fs->fs_csp[fragstoblks(fs, i)] = (struct csum *)space;
		space += size;
		brelse(bp);
		bp = NULL;
	}
	mp->mnt_data = (qaddr_t)ump;
	mp->mnt_stat.f_fsid.val[0] = (long)dev;
	mp->mnt_stat.f_fsid.val[1] = MOUNT_UFS;
	mp->mnt_maxsymlinklen = fs->fs_maxsymlinklen;
	mp->mnt_flag |= MNT_LOCAL;
	ump->um_mountp = mp;
	ump->um_dev = dev;
	ump->um_devvp = devvp;
	ump->um_nindir = fs->fs_nindir;
	ump->um_bptrtodb = fs->fs_fsbtodb;
	ump->um_seqinc = fs->fs_frag;
	for (i = 0; i < MAXQUOTAS; i++)
		ump->um_quotas[i] = NULLVP;
	devvp->v_specflags |= SI_MOUNTEDON;

	/* Sanity checks for old file systems.			   XXX */
	fs->fs_npsect = max(fs->fs_npsect, fs->fs_nsect);	/* XXX */
	fs->fs_interleave = max(fs->fs_interleave, 1);		/* XXX */
	if (fs->fs_postblformat == FS_42POSTBLFMT)		/* XXX */
		fs->fs_nrpos = 8;				/* XXX */
	if (fs->fs_inodefmt < FS_44INODEFMT) {			/* XXX */
		quad_t sizepb = fs->fs_bsize;			/* XXX */
								/* XXX */
		fs->fs_maxfilesize = fs->fs_bsize * NDADDR - 1;	/* XXX */
		for (i = 0; i < NIADDR; i++) {			/* XXX */
			sizepb *= NINDIR(fs);			/* XXX */
			fs->fs_maxfilesize += sizepb;		/* XXX */
		}						/* XXX */
		fs->fs_qbmask = ~fs->fs_bmask;			/* XXX */
		fs->fs_qfmask = ~fs->fs_fmask;			/* XXX */
	}							/* XXX */
	return (0);
out:
	if (bp)
		brelse(bp);
	(void)VOP_CLOSE(devvp, ronly ? FREAD : FREAD|FWRITE, NOCRED, p);
	if (ump) {
		free(ump->um_fs, M_UFSMNT);
		free(ump, M_UFSMNT);
		mp->mnt_data = (qaddr_t)0;
	}
	return (error);
}

/*
 * unmount system call
 */
int
ffs_unmount(mp, mntflags, p)
	struct mount *mp;
	int mntflags;
	struct proc *p;
{
	extern int doforce;
	register struct ufsmount *ump;
	register struct fs *fs;
	int i, error, flags, ronly;

	flags = 0;
	if (mntflags & MNT_FORCE) {
		if (!doforce || mp == rootfs)
			return (EINVAL);
		flags |= FORCECLOSE;
	}
	ump = VFSTOUFS(mp);
#ifdef QUOTA
	if (mp->mnt_flag & MNT_QUOTA) {
		if (error = vflush(mp, NULLVP, 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, NULLVP, flags))
		return (error);
	fs = ump->um_fs;
	ronly = !fs->fs_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->fs_csp[0], M_UFSMNT);
	free(fs, M_UFSMNT);
	free(ump, M_UFSMNT);
	mp->mnt_data = (qaddr_t)0;
	mp->mnt_flag &= ~MNT_LOCAL;
	return (error);
}

/*
 * Return root of a filesystem
 */
int
ffs_root(mp, vpp)
	struct mount *mp;
	struct vnode **vpp;
{
	struct vnode *nvp;
	int error;

	if (error = VFS_VGET(mp, (ino_t)ROOTINO, &nvp))
		return (error);
	*vpp = nvp;
	return (0);
}

/*
 * Get file system statistics.
 */
int
ffs_statfs(mp, sbp, p)
	struct mount *mp;
	register struct statfs *sbp;
	struct proc *p;
{
	register struct ufsmount *ump;
	register struct fs *fs;

	ump = VFSTOUFS(mp);
	fs = ump->um_fs;
	if (fs->fs_magic != FS_MAGIC)
		panic("ffs_statfs");
	sbp->f_type = MOUNT_UFS;
	sbp->f_bsize = fs->fs_fsize;
	sbp->f_iosize = fs->fs_bsize;
	sbp->f_blocks = fs->fs_dsize;
	sbp->f_bfree = fs->fs_cstotal.cs_nbfree * fs->fs_frag +
		fs->fs_cstotal.cs_nffree;
	sbp->f_bavail = (fs->fs_dsize * (100 - fs->fs_minfree) / 100) -
		(fs->fs_dsize - sbp->f_bfree);
	sbp->f_files =  fs->fs_ncg * fs->fs_ipg - ROOTINO;
	sbp->f_ffree = fs->fs_cstotal.cs_nifree;
	if (sbp != &mp->mnt_stat) {
		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'.
 */
int
ffs_sync(mp, waitfor, cred, p)
	struct mount *mp;
	int waitfor;
	struct ucred *cred;
	struct proc *p;
{
	register struct vnode *vp;
	register struct inode *ip;
	register struct ufsmount *ump = VFSTOUFS(mp);
	register struct fs *fs;
	int error, allerror = 0;

	fs = ump->um_fs;
	/*
	 * Write back modified superblock.
	 * Consistency check that the superblock
	 * is still in the buffer cache.
	 */
	if (fs->fs_fmod != 0) {
		if (fs->fs_ronly != 0) {		/* XXX */
			printf("fs = %s\n", fs->fs_fsmnt);
			panic("update: rofs mod");
		}
		fs->fs_fmod = 0;
		fs->fs_time = time.tv_sec;
		allerror = ffs_sbupdate(ump, waitfor);
	}
	/*
	 * Write back each (modified) inode.
	 */
loop:
	for (vp = mp->mnt_mounth; vp; vp = vp->v_mountf) {
		/*
		 * If the vnode that we are about to sync is no longer
		 * associated with this mount point, start over.
		 */
		if (vp->v_mount != mp)
			goto loop;
		if (VOP_ISLOCKED(vp))
			continue;
		ip = VTOI(vp);
		if ((ip->i_flag & (IMOD|IACC|IUPD|ICHG)) == 0 &&
		    vp->v_dirtyblkhd.le_next == NULL)
			continue;
		if (vget(vp))
			goto loop;
		if (error = VOP_FSYNC(vp, cred, waitfor, p))
			allerror = error;
		vput(vp);
	}
	/*
	 * Force stale file system control information to be flushed.
	 */
	if (error = VOP_FSYNC(ump->um_devvp, cred, waitfor, p))
		allerror = error;
#ifdef QUOTA
	qsync(mp);
#endif
	return (allerror);
}

/*
 * Look up a FFS dinode number to find its incore vnode.
 * If it is not in core, read it in from the specified device.
 * If it is in core, wait for the lock bit to clear, then
 * return the inode locked. Detection and handling of mount
 * points must be done by the calling routine.
 */
int
ffs_vget(mp, ino, vpp)
	struct mount *mp;
	ino_t ino;
	struct vnode **vpp;
{
	register struct fs *fs;
	register struct inode *ip;
	struct ufsmount *ump;
	struct buf *bp;
	struct dinode *dp;
	struct vnode *vp;
	union ihead *ih;
	dev_t dev;
	int i, type, error;

	ump = VFSTOUFS(mp);
	dev = ump->um_dev;
	if ((*vpp = ufs_ihashget(dev, ino)) != NULL)
		return (0);

	/* Allocate a new vnode/inode. */
	if (error = getnewvnode(VT_UFS, mp, ffs_vnodeop_p, &vp)) {
		*vpp = NULL;
		return (error);
	}
	type = ump->um_devvp->v_tag == VT_MFS ? M_MFSNODE : M_FFSNODE; /* XXX */
	MALLOC(ip, struct inode *, sizeof(struct inode), type, M_WAITOK);
	vp->v_data = ip;
	ip->i_vnode = vp;
	ip->i_flag = 0;
	ip->i_devvp = 0;
	ip->i_mode = 0;
	ip->i_diroff = 0;
	ip->i_lockf = 0;
	ip->i_fs = fs = ump->um_fs;
	ip->i_dev = dev;
	ip->i_number = ino;
#ifdef QUOTA
	for (i = 0; i < MAXQUOTAS; i++)
		ip->i_dquot[i] = NODQUOT;
#endif
	/*
	 * 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.
	 */
	ufs_ihashins(ip);

	/* Read in the disk contents for the inode, copy into the inode. */
	if (error = bread(ump->um_devvp, fsbtodb(fs, itod(fs, ino)),
	    (int)fs->fs_bsize, NOCRED, &bp)) {
		/*
		 * The inode does not contain anything useful, so it would
		 * be misleading to leave it on its hash chain. It will be
		 * returned to the free list by ufs_iput().
		 */
		ufs_ihashrem(ip);

		/* Unlock and discard unneeded inode. */
		ufs_iput(ip);
		brelse(bp);
		*vpp = NULL;
		return (error);
	}
	dp = bp->b_un.b_dino;
	dp += itoo(fs, ino);
	ip->i_din = *dp;
	brelse(bp);

	/*
	 * Initialize the vnode from the inode, check for aliases.
	 * Note that the underlying vnode may have changed.
	 */
	if (error = ufs_vinit(mp, ffs_specop_p, FFS_FIFOOPS, &vp)) {
		ufs_iput(ip);
		*vpp = NULL;
		return (error);
	}
	/*
	 * Finish inode initialization now that aliasing has been resolved.
	 */
	ip->i_devvp = ump->um_devvp;
	VREF(ip->i_devvp);
	/*
	 * Set up a generation number for this inode if it does not
	 * already have one. This should only happen on old filesystems.
	 */
	if (ip->i_gen == 0) {
		if (++nextgennumber < (u_long)time.tv_sec)
			nextgennumber = time.tv_sec;
		ip->i_gen = nextgennumber;
		if ((vp->v_mount->mnt_flag & MNT_RDONLY) == 0)
			ip->i_flag |= IMOD;
	}
	/*
	 * Ensure that uid and gid are correct. This is a temporary
	 * fix until fsck has been changed to do the update.
	 */
	if (fs->fs_inodefmt < FS_44INODEFMT) {		/* XXX */
		ip->i_uid = ip->i_din.di_ouid;		/* XXX */
		ip->i_gid = ip->i_din.di_ogid;		/* XXX */
	}						/* XXX */

	*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 ffs_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
 */
int
ffs_fhtovp(mp, fhp, nam, vpp, exflagsp, credanonp)
	register struct mount *mp;
	struct fid *fhp;
	struct mbuf *nam;
	struct vnode **vpp;
	int *exflagsp;
	struct ucred **credanonp;
{
	register struct ufid *ufhp;
	struct fs *fs;

	ufhp = (struct ufid *)fhp;
	fs = VFSTOUFS(mp)->um_fs;
	if (ufhp->ufid_ino < ROOTINO ||
	    ufhp->ufid_ino >= fs->fs_ncg * fs->fs_ipg)
		return (ESTALE);
	return (ufs_check_export(mp, ufhp, nam, vpp, exflagsp, credanonp));
}

/*
 * Vnode pointer to File handle
 */
/* ARGSUSED */
ffs_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);
}

/*
 * Write a superblock and associated information back to disk.
 */
int
ffs_sbupdate(mp, waitfor)
	struct ufsmount *mp;
	int waitfor;
{
	register struct fs *fs = mp->um_fs;
	register struct buf *bp;
	int blks;
	caddr_t space;
	int i, size, error = 0;

	bp = getblk(mp->um_devvp, SBLOCK, (int)fs->fs_sbsize);
	bcopy((caddr_t)fs, bp->b_un.b_addr, (u_int)fs->fs_sbsize);
	/* Restore compatibility to old file systems.		   XXX */
	if (fs->fs_postblformat == FS_42POSTBLFMT)		/* XXX */
		bp->b_un.b_fs->fs_nrpos = -1;			/* XXX */
	if (waitfor == MNT_WAIT)
		error = bwrite(bp);
	else
		bawrite(bp);
	blks = howmany(fs->fs_cssize, fs->fs_fsize);
	space = (caddr_t)fs->fs_csp[0];
	for (i = 0; i < blks; i += fs->fs_frag) {
		size = fs->fs_bsize;
		if (i + fs->fs_frag > blks)
			size = (blks - i) * fs->fs_fsize;
		bp = getblk(mp->um_devvp, fsbtodb(fs, fs->fs_csaddr + i), size);
		bcopy(space, bp->b_un.b_addr, (u_int)size);
		space += size;
		if (waitfor == MNT_WAIT)
			error = bwrite(bp);
		else
			bawrite(bp);
	}
	return (error);
}