/*
 * Copyright (c) 1989 The Regents of the University of California.
 * All rights reserved.
 *
 * %sccs.include.redist.c%
 *
 *	@(#)spec_vnops.c	7.42 (Berkeley) 02/04/92
 */

#include <sys/param.h>
#include <sys/proc.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/conf.h>
#include <sys/buf.h>
#include <sys/mount.h>
#include <sys/namei.h>
#include <sys/vnode.h>
#include <sys/specdev.h>
#include <sys/stat.h>
#include <sys/errno.h>
#include <sys/ioctl.h>
#include <sys/file.h>
#include <sys/disklabel.h>

/* symbolic sleep message strings for devices */
char	devopn[] = "devopn";
char	devio[] = "devio";
char	devwait[] = "devwait";
char	devin[] = "devin";
char	devout[] = "devout";
char	devioc[] = "devioc";
char	devcls[] = "devcls";

struct vnodeops spec_vnodeops = {
	spec_lookup,		/* lookup */
	spec_create,		/* create */
	spec_mknod,		/* mknod */
	spec_open,		/* open */
	spec_close,		/* close */
	spec_access,		/* access */
	spec_getattr,		/* getattr */
	spec_setattr,		/* setattr */
	spec_read,		/* read */
	spec_write,		/* write */
	spec_ioctl,		/* ioctl */
	spec_select,		/* select */
	spec_mmap,		/* mmap */
	spec_fsync,		/* fsync */
	spec_seek,		/* seek */
	spec_remove,		/* remove */
	spec_link,		/* link */
	spec_rename,		/* rename */
	spec_mkdir,		/* mkdir */
	spec_rmdir,		/* rmdir */
	spec_symlink,		/* symlink */
	spec_readdir,		/* readdir */
	spec_readlink,		/* readlink */
	spec_abortop,		/* abortop */
	spec_inactive,		/* inactive */
	spec_reclaim,		/* reclaim */
	spec_lock,		/* lock */
	spec_unlock,		/* unlock */
	spec_bmap,		/* bmap */
	spec_strategy,		/* strategy */
	spec_print,		/* print */
	spec_islocked,		/* islocked */
	spec_advlock,		/* advlock */
	spec_blkatoff,		/* blkatoff */
	spec_vget,		/* vget */
	spec_valloc,		/* valloc */
	spec_vfree,		/* vfree */
	spec_truncate,		/* truncate */
	spec_update,		/* update */
	spec_bwrite,		/* bwrite */
};

/*
 * Trivial lookup routine that always fails.
 */
int
spec_lookup(dvp, vpp, cnp)
	struct vnode *dvp;
	struct vnode **vpp;
	struct componentname *cnp;
{

	*vpp = NULL;
	return (ENOTDIR);
}

/*
 * Open a special file: Don't allow open if fs is mounted -nodev,
 * and don't allow opens of block devices that are currently mounted.
 * Otherwise, call device driver open function.
 */
/* ARGSUSED */
spec_open(vp, mode, cred, p)
	register struct vnode *vp;
	int mode;
	struct ucred *cred;
	struct proc *p;
{
	dev_t dev = (dev_t)vp->v_rdev;
	register int maj = major(dev);
	int error;

	if (vp->v_mount && (vp->v_mount->mnt_flag & MNT_NODEV))
		return (ENXIO);

	switch (vp->v_type) {

	case VCHR:
		if ((u_int)maj >= nchrdev)
			return (ENXIO);
		VOP_UNLOCK(vp);
		error = (*cdevsw[maj].d_open)(dev, mode, S_IFCHR, p);
		VOP_LOCK(vp);
		return (error);

	case VBLK:
		if ((u_int)maj >= nblkdev)
			return (ENXIO);
		if (error = ufs_mountedon(vp))
			return (error);
		return ((*bdevsw[maj].d_open)(dev, mode, S_IFBLK, p));
	}
	return (0);
}

/*
 * Vnode op for read
 */
/* ARGSUSED */
spec_read(vp, uio, ioflag, cred)
	register struct vnode *vp;
	register struct uio *uio;
	int ioflag;
	struct ucred *cred;
{
	struct proc *p = uio->uio_procp;
	struct buf *bp;
	daddr_t bn, nextbn;
	long bsize, bscale;
	struct partinfo dpart;
	register int n, on;
	int error = 0;

#ifdef DIAGNOSTIC
	if (uio->uio_rw != UIO_READ)
		panic("spec_read mode");
	if (uio->uio_segflg == UIO_USERSPACE && uio->uio_procp != curproc)
		panic("spec_read proc");
#endif
	if (uio->uio_resid == 0)
		return (0);

	switch (vp->v_type) {

	case VCHR:
		VOP_UNLOCK(vp);
		error = (*cdevsw[major(vp->v_rdev)].d_read)
			(vp->v_rdev, uio, ioflag);
		VOP_LOCK(vp);
		return (error);

	case VBLK:
		if (uio->uio_offset < 0)
			return (EINVAL);
		bsize = BLKDEV_IOSIZE;
		if ((*bdevsw[major(vp->v_rdev)].d_ioctl)(vp->v_rdev, DIOCGPART,
		    (caddr_t)&dpart, FREAD, p) == 0) {
			if (dpart.part->p_fstype == FS_BSDFFS &&
			    dpart.part->p_frag != 0 && dpart.part->p_fsize != 0)
				bsize = dpart.part->p_frag *
				    dpart.part->p_fsize;
		}
		bscale = bsize / DEV_BSIZE;
		do {
			bn = (uio->uio_offset / DEV_BSIZE) &~ (bscale - 1);
			on = uio->uio_offset % bsize;
			n = MIN((unsigned)(bsize - on), uio->uio_resid);
			if (vp->v_lastr + bscale == bn) {
				nextbn = bn + bscale;
				error = breadn(vp, bn, (int)bsize, &nextbn,
					(int *)&bsize, 1, NOCRED, &bp);
			} else
				error = bread(vp, bn, (int)bsize, NOCRED, &bp);
			vp->v_lastr = bn;
			n = MIN(n, bsize - bp->b_resid);
			if (error) {
				brelse(bp);
				return (error);
			}
			error = uiomove(bp->b_un.b_addr + on, n, uio);
			if (n + on == bsize)
				bp->b_flags |= B_AGE;
			brelse(bp);
		} while (error == 0 && uio->uio_resid > 0 && n != 0);
		return (error);

	default:
		panic("spec_read type");
	}
	/* NOTREACHED */
}

/*
 * Vnode op for write
 */
/* ARGSUSED */
spec_write(vp, uio, ioflag, cred)
	register struct vnode *vp;
	register struct uio *uio;
	int ioflag;
	struct ucred *cred;
{
	struct proc *p = uio->uio_procp;
	struct buf *bp;
	daddr_t bn;
	int bsize, blkmask;
	struct partinfo dpart;
	register int n, on;
	int error = 0;

#ifdef DIAGNOSTIC
	if (uio->uio_rw != UIO_WRITE)
		panic("spec_write mode");
	if (uio->uio_segflg == UIO_USERSPACE && uio->uio_procp != curproc)
		panic("spec_write proc");
#endif

	switch (vp->v_type) {

	case VCHR:
		VOP_UNLOCK(vp);
		error = (*cdevsw[major(vp->v_rdev)].d_write)
			(vp->v_rdev, uio, ioflag);
		VOP_LOCK(vp);
		return (error);

	case VBLK:
		if (uio->uio_resid == 0)
			return (0);
		if (uio->uio_offset < 0)
			return (EINVAL);
		bsize = BLKDEV_IOSIZE;
		if ((*bdevsw[major(vp->v_rdev)].d_ioctl)(vp->v_rdev, DIOCGPART,
		    (caddr_t)&dpart, FREAD, p) == 0) {
			if (dpart.part->p_fstype == FS_BSDFFS &&
			    dpart.part->p_frag != 0 && dpart.part->p_fsize != 0)
				bsize = dpart.part->p_frag *
				    dpart.part->p_fsize;
		}
		blkmask = (bsize / DEV_BSIZE) - 1;
		do {
			bn = (uio->uio_offset / DEV_BSIZE) &~ blkmask;
			on = uio->uio_offset % bsize;
			n = MIN((unsigned)(bsize - on), uio->uio_resid);
			if (n == bsize)
				bp = getblk(vp, bn, bsize);
			else
				error = bread(vp, bn, bsize, NOCRED, &bp);
			n = MIN(n, bsize - bp->b_resid);
			if (error) {
				brelse(bp);
				return (error);
			}
			error = uiomove(bp->b_un.b_addr + on, n, uio);
			if (n + on == bsize) {
				bp->b_flags |= B_AGE;
				bawrite(bp);
			} else
				bdwrite(bp);
		} while (error == 0 && uio->uio_resid > 0 && n != 0);
		return (error);

	default:
		panic("spec_write type");
	}
	/* NOTREACHED */
}

/*
 * Device ioctl operation.
 */
/* ARGSUSED */
spec_ioctl(vp, com, data, fflag, cred, p)
	struct vnode *vp;
	int com;
	caddr_t data;
	int fflag;
	struct ucred *cred;
	struct proc *p;
{
	dev_t dev = vp->v_rdev;

	switch (vp->v_type) {

	case VCHR:
		return ((*cdevsw[major(dev)].d_ioctl)(dev, com, data,
		    fflag, p));

	case VBLK:
		if (com == 0 && (int)data == B_TAPE)
			if (bdevsw[major(dev)].d_flags & B_TAPE)
				return (0);
			else
				return (1);
		return ((*bdevsw[major(dev)].d_ioctl)(dev, com, data,
		   fflag, p));

	default:
		panic("spec_ioctl");
		/* NOTREACHED */
	}
}

/* ARGSUSED */
spec_select(vp, which, fflags, cred, p)
	struct vnode *vp;
	int which, fflags;
	struct ucred *cred;
	struct proc *p;
{
	register dev_t dev;

	switch (vp->v_type) {

	default:
		return (1);		/* XXX */

	case VCHR:
		dev = vp->v_rdev;
		return (*cdevsw[major(dev)].d_select)(dev, which, p);
	}
}

/*
 * Just call the device strategy routine
 */
spec_strategy(bp)
	register struct buf *bp;
{

	(*bdevsw[major(bp->b_dev)].d_strategy)(bp);
	return (0);
}

/*
 * This is a noop, simply returning what one has been given.
 */
spec_bmap(vp, bn, vpp, bnp)
	struct vnode *vp;
	daddr_t bn;
	struct vnode **vpp;
	daddr_t *bnp;
{

	if (vpp != NULL)
		*vpp = vp;
	if (bnp != NULL)
		*bnp = bn;
	return (0);
}

/*
 * At the moment we do not do any locking.
 */
/* ARGSUSED */
spec_lock(vp)
	struct vnode *vp;
{

	return (0);
}

/* ARGSUSED */
spec_unlock(vp)
	struct vnode *vp;
{

	return (0);
}

/*
 * Device close routine
 */
/* ARGSUSED */
spec_close(vp, flag, cred, p)
	register struct vnode *vp;
	int flag;
	struct ucred *cred;
	struct proc *p;
{
	dev_t dev = vp->v_rdev;
	int (*devclose) __P((dev_t, int, int, struct proc *));
	int mode;

	switch (vp->v_type) {

	case VCHR:
		/*
		 * If the vnode is locked, then we are in the midst
		 * of forcably closing the device, otherwise we only
		 * close on last reference.
		 */
		if (vcount(vp) > 1 && (vp->v_flag & VXLOCK) == 0)
			return (0);
		devclose = cdevsw[major(dev)].d_close;
		mode = S_IFCHR;
		break;

	case VBLK:
		/*
		 * On last close of a block device (that isn't mounted)
		 * we must invalidate any in core blocks, so that
		 * we can, for instance, change floppy disks.
		 */
		vflushbuf(vp, 0);
		if (vinvalbuf(vp, 1))
			return (0);
		/*
		 * We do not want to really close the device if it
		 * is still in use unless we are trying to close it
		 * forcibly. Since every use (buffer, vnode, swap, cmap)
		 * holds a reference to the vnode, and because we mark
		 * any other vnodes that alias this device, when the
		 * sum of the reference counts on all the aliased
		 * vnodes descends to one, we are on last close.
		 */
		if (vcount(vp) > 1 && (vp->v_flag & VXLOCK) == 0)
			return (0);
		devclose = bdevsw[major(dev)].d_close;
		mode = S_IFBLK;
		break;

	default:
		panic("spec_close: not special");
	}

	return ((*devclose)(dev, flag, mode, p));
}

/*
 * Print out the contents of a special device vnode.
 */
spec_print(vp)
	struct vnode *vp;
{

	printf("tag VT_NON, dev %d, %d\n", major(vp->v_rdev),
		minor(vp->v_rdev));
}

/*
 * Special device advisory byte-level locks.
 */
/* ARGSUSED */
spec_advlock(vp, id, op, fl, flags)
	struct vnode *vp;
	caddr_t id;
	int op;
	struct flock *fl;
	int flags;
{

	return (EOPNOTSUPP);
}

/*
 * Special device failed operation
 */
spec_ebadf()
{

	return (EBADF);
}

/*
 * Special device bad operation
 */
spec_badop()
{

	panic("spec_badop called");
	/* NOTREACHED */
}