X-Git-Url: https://git.subgeniuskitty.com/unix-history/.git/blobdiff_plain/1d00c1cfc6b163c3ebb35466de2a2e105d4f1ddc..930730fda4a84d270ddd788e2c31846b66b96054:/usr/src/sys/ufs/ffs/ufs_vnops.c diff --git a/usr/src/sys/ufs/ffs/ufs_vnops.c b/usr/src/sys/ufs/ffs/ufs_vnops.c index d3824595b1..3778eb892f 100644 --- a/usr/src/sys/ufs/ffs/ufs_vnops.c +++ b/usr/src/sys/ufs/ffs/ufs_vnops.c @@ -1,644 +1,470 @@ -/* ufs_vnops.c 6.12 84/07/20 */ - -#include "../h/param.h" -#include "../h/systm.h" -#include "../h/dir.h" -#include "../h/user.h" -#include "../h/kernel.h" -#include "../h/file.h" -#include "../h/stat.h" -#include "../h/inode.h" -#include "../h/fs.h" -#include "../h/buf.h" -#include "../h/proc.h" -#include "../h/quota.h" -#include "../h/uio.h" -#include "../h/socket.h" -#include "../h/socketvar.h" -#include "../h/mount.h" - -extern struct fileops inodeops; -struct file *getinode(); - /* - * Change current working directory (``.''). + * Copyright (c) 1982, 1986, 1989 Regents of the University of California. + * All rights reserved. + * + * Redistribution and use in source and binary forms are permitted + * provided that the above copyright notice and this paragraph are + * duplicated in all such forms and that any documentation, + * advertising materials, and other materials related to such + * distribution and use acknowledge that the software was developed + * by the University of California, Berkeley. The name of the + * University may not be used to endorse or promote products derived + * from this software without specific prior written permission. + * THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED + * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. + * + * @(#)ufs_vnops.c 7.33 (Berkeley) %G% */ -chdir() -{ - chdirec(&u.u_cdir); -} +#include "param.h" +#include "systm.h" +#include "user.h" +#include "kernel.h" +#include "file.h" +#include "stat.h" +#include "buf.h" +#include "proc.h" +#include "uio.h" +#include "socket.h" +#include "socketvar.h" +#include "conf.h" +#include "mount.h" +#include "vnode.h" +#include "../ufs/inode.h" +#include "../ufs/fs.h" +#include "../ufs/quota.h" /* - * Change notion of root (``/'') directory. + * Global vfs data structures for ufs */ -chroot() -{ - - if (suser()) - chdirec(&u.u_rdir); -} -/* - * Common routine for chroot and chdir. - */ -chdirec(ipp) - register struct inode **ipp; -{ - register struct inode *ip; - struct a { - char *fname; - } *uap = (struct a *)u.u_ap; - register struct nameidata *ndp = &u.u_nd; - - ndp->ni_nameiop = LOOKUP | FOLLOW; - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = uap->fname; - ip = namei(ndp); - if (ip == NULL) - return; - if ((ip->i_mode&IFMT) != IFDIR) { - u.u_error = ENOTDIR; - goto bad; - } - if (access(ip, IEXEC)) - goto bad; - IUNLOCK(ip); - if (*ipp) - irele(*ipp); - *ipp = ip; - return; +int ufs_lookup(), + ufs_create(), + ufs_mknod(), + ufs_open(), + ufs_close(), + ufs_access(), + ufs_getattr(), + ufs_setattr(), + ufs_read(), + ufs_write(), + ufs_ioctl(), + ufs_select(), + ufs_mmap(), + ufs_fsync(), + ufs_seek(), + ufs_remove(), + ufs_link(), + ufs_rename(), + ufs_mkdir(), + ufs_rmdir(), + ufs_symlink(), + ufs_readdir(), + ufs_readlink(), + ufs_abortop(), + ufs_inactive(), + ufs_reclaim(), + ufs_lock(), + ufs_unlock(), + ufs_bmap(), + ufs_strategy(), + ufs_print(), + ufs_islocked(); + +struct vnodeops ufs_vnodeops = { + ufs_lookup, /* lookup */ + ufs_create, /* create */ + ufs_mknod, /* mknod */ + ufs_open, /* open */ + ufs_close, /* close */ + ufs_access, /* access */ + ufs_getattr, /* getattr */ + ufs_setattr, /* setattr */ + ufs_read, /* read */ + ufs_write, /* write */ + ufs_ioctl, /* ioctl */ + ufs_select, /* select */ + ufs_mmap, /* mmap */ + ufs_fsync, /* fsync */ + ufs_seek, /* seek */ + ufs_remove, /* remove */ + ufs_link, /* link */ + ufs_rename, /* rename */ + ufs_mkdir, /* mkdir */ + ufs_rmdir, /* rmdir */ + ufs_symlink, /* symlink */ + ufs_readdir, /* readdir */ + ufs_readlink, /* readlink */ + ufs_abortop, /* abortop */ + ufs_inactive, /* inactive */ + ufs_reclaim, /* reclaim */ + ufs_lock, /* lock */ + ufs_unlock, /* unlock */ + ufs_bmap, /* bmap */ + ufs_strategy, /* strategy */ + ufs_print, /* print */ + ufs_islocked, /* islocked */ +}; -bad: - iput(ip); -} +int spec_lookup(), + spec_open(), + ufsspec_read(), + ufsspec_write(), + spec_strategy(), + spec_bmap(), + spec_ioctl(), + spec_select(), + ufsspec_close(), + spec_badop(), + spec_nullop(); + +struct vnodeops spec_inodeops = { + spec_lookup, /* lookup */ + spec_badop, /* create */ + spec_badop, /* mknod */ + spec_open, /* open */ + ufsspec_close, /* close */ + ufs_access, /* access */ + ufs_getattr, /* getattr */ + ufs_setattr, /* setattr */ + ufsspec_read, /* read */ + ufsspec_write, /* write */ + spec_ioctl, /* ioctl */ + spec_select, /* select */ + spec_badop, /* mmap */ + spec_nullop, /* fsync */ + spec_badop, /* seek */ + spec_badop, /* remove */ + spec_badop, /* link */ + spec_badop, /* rename */ + spec_badop, /* mkdir */ + spec_badop, /* rmdir */ + spec_badop, /* symlink */ + spec_badop, /* readdir */ + spec_badop, /* readlink */ + spec_badop, /* abortop */ + ufs_inactive, /* inactive */ + ufs_reclaim, /* reclaim */ + ufs_lock, /* lock */ + ufs_unlock, /* unlock */ + spec_bmap, /* bmap */ + spec_strategy, /* strategy */ + ufs_print, /* print */ + ufs_islocked, /* islocked */ +}; -/* - * Open system call. - */ -open() -{ - struct a { - char *fname; - int mode; - int crtmode; - } *uap = (struct a *) u.u_ap; +#ifdef FIFO +int fifo_lookup(), + fifo_open(), + ufsfifo_read(), + ufsfifo_write(), + fifo_bmap(), + fifo_ioctl(), + fifo_select(), + ufsfifo_close(), + fifo_print(), + fifo_badop(), + fifo_nullop(); + +struct vnodeops fifo_inodeops = { + fifo_lookup, /* lookup */ + fifo_badop, /* create */ + fifo_badop, /* mknod */ + fifo_open, /* open */ + ufsfifo_close, /* close */ + ufs_access, /* access */ + ufs_getattr, /* getattr */ + ufs_setattr, /* setattr */ + ufsfifo_read, /* read */ + ufsfifo_write, /* write */ + fifo_ioctl, /* ioctl */ + fifo_select, /* select */ + fifo_badop, /* mmap */ + fifo_nullop, /* fsync */ + fifo_badop, /* seek */ + fifo_badop, /* remove */ + fifo_badop, /* link */ + fifo_badop, /* rename */ + fifo_badop, /* mkdir */ + fifo_badop, /* rmdir */ + fifo_badop, /* symlink */ + fifo_badop, /* readdir */ + fifo_badop, /* readlink */ + fifo_badop, /* abortop */ + ufs_inactive, /* inactive */ + ufs_reclaim, /* reclaim */ + ufs_lock, /* lock */ + ufs_unlock, /* unlock */ + fifo_bmap, /* bmap */ + fifo_badop, /* strategy */ + ufs_print, /* print */ + ufs_islocked, /* islocked */ +}; +#endif /* FIFO */ - copen(uap->mode-FOPEN, uap->crtmode, uap->fname); -} +enum vtype iftovt_tab[16] = { + VNON, VFIFO, VCHR, VNON, VDIR, VNON, VBLK, VNON, + VREG, VNON, VLNK, VNON, VSOCK, VNON, VNON, VBAD, +}; +int vttoif_tab[9] = { + 0, IFREG, IFDIR, IFBLK, IFCHR, IFLNK, IFSOCK, IFIFO, IFMT, +}; /* - * Creat system call. + * Create a regular file */ -creat() +ufs_create(ndp, vap) + struct nameidata *ndp; + struct vattr *vap; { - struct a { - char *fname; - int fmode; - } *uap = (struct a *)u.u_ap; + struct inode *ip; + int error; - copen(FWRITE|FCREAT|FTRUNC, uap->fmode, uap->fname); + if (error = maknode(MAKEIMODE(vap->va_type, vap->va_mode), ndp, &ip)) + return (error); + ndp->ni_vp = ITOV(ip); + return (0); } /* - * Common code for open and creat. - * Check permissions, allocate an open file structure, - * and call the device open routine if any. + * Mknod vnode call */ -copen(mode, arg, fname) - register int mode; - int arg; - caddr_t fname; +/* ARGSUSED */ +ufs_mknod(ndp, vap, cred) + struct nameidata *ndp; + struct ucred *cred; + struct vattr *vap; { - register struct inode *ip; - register struct file *fp; - register struct nameidata *ndp = &u.u_nd; - int i; + register struct vnode *vp; + struct inode *ip; + int error; -#ifdef notdef - if ((mode&(FREAD|FWRITE)) == 0) { - u.u_error = EINVAL; - return; - } -#endif - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = fname; - if (mode&FCREAT) { - ndp->ni_nameiop = CREATE | FOLLOW; - ip = namei(ndp); - if (ip == NULL) { - if (u.u_error) - return; - ip = maknode(arg&07777&(~ISVTX), ndp); - if (ip == NULL) - return; - mode &= ~FTRUNC; - } else { - if (mode&FEXCL) { - u.u_error = EEXIST; - iput(ip); - return; - } - mode &= ~FCREAT; - } - } else { - ndp->ni_nameiop = LOOKUP | FOLLOW; - ip = namei(ndp); - if (ip == NULL) - return; - } - if ((ip->i_mode & IFMT) == IFSOCK) { - u.u_error = EOPNOTSUPP; - goto bad; - } - if ((mode&FCREAT) == 0) { - if (mode&FREAD) - if (access(ip, IREAD)) - goto bad; - if (mode&(FWRITE|FTRUNC)) { - if (access(ip, IWRITE)) - goto bad; - if ((ip->i_mode&IFMT) == IFDIR) { - u.u_error = EISDIR; - goto bad; - } - } - } - fp = falloc(); - if (fp == NULL) - goto bad; - if (mode&FTRUNC) - itrunc(ip, (u_long)0); - IUNLOCK(ip); - fp->f_flag = mode&FMASK; - fp->f_type = DTYPE_INODE; - fp->f_ops = &inodeops; - fp->f_data = (caddr_t)ip; - i = u.u_r.r_val1; - if (setjmp(&u.u_qsave)) { - if (u.u_error == 0) - u.u_error = EINTR; - u.u_ofile[i] = NULL; - closef(fp); - return; + if (error = maknode(MAKEIMODE(vap->va_type, vap->va_mode), ndp, &ip)) + return (error); + ip->i_flag |= IACC|IUPD|ICHG; + if (vap->va_rdev != VNOVAL) { + /* + * Want to be able to use this to make badblock + * inodes, so don't truncate the dev number. + */ + ip->i_rdev = vap->va_rdev; } - u.u_error = openi(ip, mode); - if (u.u_error == 0) - return; - u.u_ofile[i] = NULL; - fp->f_count--; - irele(ip); - return; -bad: - iput(ip); + /* + * Remove inode so that it will be reloaded by iget and + * checked to see if it is an alias of an existing entry + * in the inode cache. + */ + vp = ITOV(ip); + vput(vp); + vp->v_type = VNON; + vgone(vp); + return (0); } /* - * Mknod system call + * Open called. + * + * Nothing to do. */ -mknod() +/* ARGSUSED */ +ufs_open(vp, mode, cred) + struct vnode *vp; + int mode; + struct ucred *cred; { - register struct inode *ip; - register struct a { - char *fname; - int fmode; - int dev; - } *uap = (struct a *)u.u_ap; - register struct nameidata *ndp = &u.u_nd; - - if (!suser()) - return; - ndp->ni_nameiop = CREATE; - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = uap->fname; - ip = namei(ndp); - if (ip != NULL) { - u.u_error = EEXIST; - goto out; - } - if (u.u_error) - return; - ip = maknode(uap->fmode, ndp); - if (ip == NULL) - return; - switch (ip->i_mode & IFMT) { - - case IFMT: /* used by badsect to flag bad sectors */ - case IFCHR: - case IFBLK: - if (uap->dev) { - /* - * Want to be able to use this to make badblock - * inodes, so don't truncate the dev number. - */ - ip->i_rdev = uap->dev; - ip->i_flag |= IACC|IUPD|ICHG; - } - } -out: - iput(ip); + return (0); } /* - * link system call + * Close called + * + * Update the times on the inode. */ -link() +/* ARGSUSED */ +ufs_close(vp, fflag, cred) + struct vnode *vp; + int fflag; + struct ucred *cred; { - register struct inode *ip, *xp; - register struct a { - char *target; - char *linkname; - } *uap = (struct a *)u.u_ap; - register struct nameidata *ndp = &u.u_nd; - - ndp->ni_nameiop = LOOKUP | FOLLOW; - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = uap->target; - ip = namei(ndp); /* well, this routine is doomed anyhow */ - if (ip == NULL) - return; - if ((ip->i_mode&IFMT) == IFDIR && !suser()) { - iput(ip); - return; - } - ip->i_nlink++; - ip->i_flag |= ICHG; - iupdat(ip, &time, &time, 1); - IUNLOCK(ip); - ndp->ni_nameiop = CREATE; - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = (caddr_t)uap->linkname; - xp = namei(ndp); - if (xp != NULL) { - u.u_error = EEXIST; - iput(xp); - goto out; - } - if (u.u_error) - goto out; - if (ndp->ni_pdir->i_dev != ip->i_dev) { - iput(ndp->ni_pdir); - u.u_error = EXDEV; - goto out; - } - u.u_error = direnter(ip, ndp); -out: - if (u.u_error) { - ip->i_nlink--; - ip->i_flag |= ICHG; - } - irele(ip); + register struct inode *ip = VTOI(vp); + + if (vp->v_usecount > 1 && !(ip->i_flag & ILOCKED)) + ITIMES(ip, &time, &time); + return (0); } -/* - * symlink -- make a symbolic link - */ -symlink() +ufs_access(vp, mode, cred) + struct vnode *vp; + int mode; + struct ucred *cred; { - register struct a { - char *target; - char *linkname; - } *uap = (struct a *)u.u_ap; - register struct inode *ip; - register char *tp; - register c, nc; - register struct nameidata *ndp = &u.u_nd; - - tp = uap->target; - nc = 0; - while (c = fubyte(tp)) { - if (c < 0) { - u.u_error = EFAULT; - return; - } - tp++; - nc++; - } - ndp->ni_nameiop = CREATE; - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = uap->linkname; - ip = namei(ndp); - if (ip) { - iput(ip); - u.u_error = EEXIST; - return; - } - if (u.u_error) - return; - ip = maknode(IFLNK | 0777, ndp); - if (ip == NULL) - return; - u.u_error = rdwri(UIO_WRITE, ip, uap->target, nc, 0, 0, (int *)0); - /* handle u.u_error != 0 */ - iput(ip); + + return (iaccess(VTOI(vp), mode, cred)); } -/* - * Unlink system call. - * Hard to avoid races here, especially - * in unlinking directories. - */ -unlink() +/* ARGSUSED */ +ufs_getattr(vp, vap, cred) + struct vnode *vp; + register struct vattr *vap; + struct ucred *cred; { - struct a { - char *fname; - } *uap = (struct a *)u.u_ap; - register struct inode *ip, *dp; - register struct nameidata *ndp = &u.u_nd; + register struct inode *ip = VTOI(vp); - ndp->ni_nameiop = DELETE | LOCKPARENT; - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = uap->fname; - ip = namei(ndp); - if (ip == NULL) - return; - dp = ndp->ni_pdir; - if ((ip->i_mode&IFMT) == IFDIR && !suser()) - goto out; + ITIMES(ip, &time, &time); /* - * Don't unlink a mounted file. + * Copy from inode table */ - if (ip->i_dev != dp->i_dev) { - u.u_error = EBUSY; - goto out; - } - if (ip->i_flag&ITEXT) - xrele(ip); /* try once to free text */ - if (dirremove(ndp)) { - ip->i_nlink--; - ip->i_flag |= ICHG; - } -out: - if (dp == ip) - irele(ip); + vap->va_fsid = ip->i_dev; + vap->va_fileid = ip->i_number; + vap->va_mode = ip->i_mode & ~IFMT; + vap->va_nlink = ip->i_nlink; + vap->va_uid = ip->i_uid; + vap->va_gid = ip->i_gid; + vap->va_rdev = (dev_t)ip->i_rdev; + vap->va_size = ip->i_din.di_qsize.val[0]; + vap->va_size1 = ip->i_din.di_qsize.val[1]; + vap->va_atime.tv_sec = ip->i_atime; + vap->va_atime.tv_usec = 0; + vap->va_mtime.tv_sec = ip->i_mtime; + vap->va_mtime.tv_usec = 0; + vap->va_ctime.tv_sec = ip->i_ctime; + vap->va_ctime.tv_usec = 0; + vap->va_flags = ip->i_flags; + vap->va_gen = ip->i_gen; + /* this doesn't belong here */ + if (vp->v_type == VBLK) + vap->va_blocksize = BLKDEV_IOSIZE; + else if (vp->v_type == VCHR) + vap->va_blocksize = MAXBSIZE; else - iput(ip); - iput(dp); + vap->va_blocksize = ip->i_fs->fs_bsize; + vap->va_bytes = dbtob(ip->i_blocks); + vap->va_bytes1 = -1; + vap->va_type = vp->v_type; + return (0); } /* - * Seek system call + * Set attribute vnode op. called from several syscalls */ -lseek() +ufs_setattr(vp, vap, cred) + register struct vnode *vp; + register struct vattr *vap; + register struct ucred *cred; { - register struct file *fp; - register struct a { - int fd; - off_t off; - int sbase; - } *uap = (struct a *)u.u_ap; - - GETF(fp, uap->fd); - if (fp->f_type != DTYPE_INODE) { - u.u_error = ESPIPE; - return; - } - switch (uap->sbase) { - - case L_INCR: - fp->f_offset += uap->off; - break; - - case L_XTND: - fp->f_offset = uap->off + ((struct inode *)fp->f_data)->i_size; - break; - - case L_SET: - fp->f_offset = uap->off; - break; + register struct inode *ip = VTOI(vp); + int error = 0; - default: - u.u_error = EINVAL; - return; + /* + * Check for unsetable attributes. + */ + if ((vap->va_type != VNON) || (vap->va_nlink != VNOVAL) || + (vap->va_fsid != VNOVAL) || (vap->va_fileid != VNOVAL) || + (vap->va_blocksize != VNOVAL) || (vap->va_rdev != VNOVAL) || + ((int)vap->va_bytes != VNOVAL) || (vap->va_gen != VNOVAL)) { + return (EINVAL); } - u.u_r.r_off = fp->f_offset; -} - -/* - * Access system call - */ -saccess() -{ - register svuid, svgid; - register struct inode *ip; - register struct a { - char *fname; - int fmode; - } *uap = (struct a *)u.u_ap; - register struct nameidata *ndp = &u.u_nd; - - svuid = u.u_uid; - svgid = u.u_gid; - u.u_uid = u.u_ruid; - u.u_gid = u.u_rgid; - ndp->ni_nameiop = LOOKUP | FOLLOW; - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = uap->fname; - ip = namei(ndp); - if (ip != NULL) { - if ((uap->fmode&R_OK) && access(ip, IREAD)) - goto done; - if ((uap->fmode&W_OK) && access(ip, IWRITE)) - goto done; - if ((uap->fmode&X_OK) && access(ip, IEXEC)) - goto done; -done: - iput(ip); + /* + * Go through the fields and update iff not VNOVAL. + */ + if (vap->va_uid != (u_short)VNOVAL || vap->va_gid != (u_short)VNOVAL) + if (error = chown1(vp, vap->va_uid, vap->va_gid, cred)) + return (error); + if (vap->va_size != VNOVAL) { + if (vp->v_type == VDIR) + return (EISDIR); + if (error = itrunc(ip, vap->va_size, 0)) /* XXX IO_SYNC? */ + return (error); } - u.u_uid = svuid; - u.u_gid = svgid; -} - -/* - * Stat system call. This version follows links. - */ -stat() -{ - - stat1(FOLLOW); -} - -/* - * Lstat system call. This version does not follow links. - */ -lstat() -{ - - stat1(NOFOLLOW); -} - -stat1(follow) - int follow; -{ - register struct inode *ip; - register struct a { - char *fname; - struct stat *ub; - } *uap = (struct a *)u.u_ap; - struct stat sb; - register struct nameidata *ndp = &u.u_nd; - - ndp->ni_nameiop = LOOKUP | follow; - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = uap->fname; - ip = namei(ndp); - if (ip == NULL) - return; - (void) ino_stat(ip, &sb); - iput(ip); - u.u_error = copyout((caddr_t)&sb, (caddr_t)uap->ub, sizeof (sb)); -} - -/* - * Return target name of a symbolic link - */ -readlink() -{ - register struct inode *ip; - register struct a { - char *name; - char *buf; - int count; - } *uap = (struct a *)u.u_ap; - register struct nameidata *ndp = &u.u_nd; - int resid; - - ndp->ni_nameiop = LOOKUP; - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = uap->name; - ip = namei(ndp); - if (ip == NULL) - return; - if ((ip->i_mode&IFMT) != IFLNK) { - u.u_error = ENXIO; - goto out; + if (vap->va_atime.tv_sec != VNOVAL || vap->va_mtime.tv_sec != VNOVAL) { + if (cred->cr_uid != ip->i_uid && + (error = suser(cred, &u.u_acflag))) + return (error); + if (vap->va_atime.tv_sec != VNOVAL) + ip->i_flag |= IACC; + if (vap->va_mtime.tv_sec != VNOVAL) + ip->i_flag |= IUPD; + ip->i_flag |= ICHG; + if (error = iupdat(ip, &vap->va_atime, &vap->va_mtime, 1)) + return (error); } - u.u_error = rdwri(UIO_READ, ip, uap->buf, uap->count, 0, 0, &resid); -out: - iput(ip); - u.u_r.r_val1 = uap->count - resid; -} - -/* - * Change mode of a file given path name. - */ -chmod() -{ - struct inode *ip; - struct a { - char *fname; - int fmode; - } *uap = (struct a *)u.u_ap; - - if ((ip = owner(uap->fname, FOLLOW)) == NULL) - return; - chmod1(ip, uap->fmode); - iput(ip); -} - -/* - * Change mode of a file given a file descriptor. - */ -fchmod() -{ - struct a { - int fd; - int fmode; - } *uap = (struct a *)u.u_ap; - register struct inode *ip; - register struct file *fp; - - fp = getinode(uap->fd); - if (fp == NULL) - return; - ip = (struct inode *)fp->f_data; - if (u.u_uid != ip->i_uid && !suser()) - return; - ILOCK(ip); - chmod1(ip, uap->fmode); - IUNLOCK(ip); + if (vap->va_mode != (u_short)VNOVAL) + error = chmod1(vp, (int)vap->va_mode, cred); + if (vap->va_flags != VNOVAL) { + if (cred->cr_uid != ip->i_uid && + (error = suser(cred, &u.u_acflag))) + return (error); + if (cred->cr_uid == 0) { + ip->i_flags = vap->va_flags; + } else { + ip->i_flags &= 0xffff0000; + ip->i_flags |= (vap->va_flags & 0xffff); + } + ip->i_flag |= ICHG; + } + return (error); } /* * Change the mode on a file. * Inode must be locked before calling. */ -chmod1(ip, mode) - register struct inode *ip; +chmod1(vp, mode, cred) + register struct vnode *vp; register int mode; + struct ucred *cred; { + register struct inode *ip = VTOI(vp); + int error; + if (cred->cr_uid != ip->i_uid && + (error = suser(cred, &u.u_acflag))) + return (error); ip->i_mode &= ~07777; - if (u.u_uid) { - mode &= ~ISVTX; - if (!groupmember(ip->i_gid)) + if (cred->cr_uid) { + if (vp->v_type != VDIR) + mode &= ~ISVTX; + if (!groupmember(ip->i_gid, cred)) mode &= ~ISGID; } - ip->i_mode |= mode&07777; + ip->i_mode |= mode & 07777; ip->i_flag |= ICHG; - if (ip->i_flag&ITEXT && (ip->i_mode&ISVTX)==0) - xrele(ip); -} - -/* - * Set ownership given a path name. - */ -chown() -{ - struct inode *ip; - struct a { - char *fname; - int uid; - int gid; - } *uap = (struct a *)u.u_ap; - - if (!suser() || (ip = owner(uap->fname, NOFOLLOW)) == NULL) - return; - u.u_error = chown1(ip, uap->uid, uap->gid); - iput(ip); -} - -/* - * Set ownership given a file descriptor. - */ -fchown() -{ - struct a { - int fd; - int uid; - int gid; - } *uap = (struct a *)u.u_ap; - register struct inode *ip; - register struct file *fp; - - fp = getinode(uap->fd); - if (fp == NULL) - return; - ip = (struct inode *)fp->f_data; - if (!suser()) - return; - ILOCK(ip); - u.u_error = chown1(ip, uap->uid, uap->gid); - IUNLOCK(ip); + if ((vp->v_flag & VTEXT) && (ip->i_mode & ISVTX) == 0) + xrele(vp); + return (0); } /* * Perform chown operation on inode ip; * inode must be locked prior to call. */ -chown1(ip, uid, gid) - register struct inode *ip; - int uid, gid; +chown1(vp, uid, gid, cred) + register struct vnode *vp; + uid_t uid; + gid_t gid; + struct ucred *cred; { + register struct inode *ip = VTOI(vp); #ifdef QUOTA register long change; #endif + int error; - if (uid == -1) + if (uid == (u_short)VNOVAL) uid = ip->i_uid; - if (gid == -1) + if (gid == (u_short)VNOVAL) gid = ip->i_gid; + /* + * If we don't own the file, are trying to change the owner + * of the file, or are not a member of the target group, + * the caller must be superuser or the call fails. + */ + if ((cred->cr_uid != ip->i_uid || uid != ip->i_uid || + !groupmember((gid_t)gid, cred)) && + (error = suser(cred, &u.u_acflag))) + return (error); #ifdef QUOTA if (ip->i_uid == uid) /* this just speeds things a little */ change = 0; @@ -648,121 +474,300 @@ chown1(ip, uid, gid) (void) chkiq(ip->i_dev, ip, ip->i_uid, 1); dqrele(ip->i_dquot); #endif + if (ip->i_uid != uid && cred->cr_uid != 0) + ip->i_mode &= ~ISUID; + if (ip->i_gid != gid && cred->cr_uid != 0) + ip->i_mode &= ~ISGID; ip->i_uid = uid; ip->i_gid = gid; ip->i_flag |= ICHG; - if (u.u_ruid != 0) - ip->i_mode &= ~(ISUID|ISGID); #ifdef QUOTA ip->i_dquot = inoquota(ip); (void) chkdq(ip, change, 1); - (void) chkiq(ip->i_dev, (struct inode *)NULL, uid, 1); + (void) chkiq(ip->i_dev, (struct inode *)NULL, (uid_t)uid, 1); return (u.u_error); /* should == 0 ALWAYS !! */ #else return (0); #endif } -utimes() +/* + * Vnode op for reading. + */ +/* ARGSUSED */ +ufs_read(vp, uio, ioflag, cred) + struct vnode *vp; + register struct uio *uio; + int ioflag; + struct ucred *cred; { - register struct a { - char *fname; - struct timeval *tptr; - } *uap = (struct a *)u.u_ap; - register struct inode *ip; - struct timeval tv[2]; + register struct inode *ip = VTOI(vp); + register struct fs *fs; + struct buf *bp; + daddr_t lbn, bn, rablock; + int size, diff, error = 0; + long n, on, type; + + if (uio->uio_rw != UIO_READ) + panic("ufs_read mode"); + type = ip->i_mode & IFMT; + if (type != IFDIR && type != IFREG && type != IFLNK) + panic("ufs_read type"); + if (uio->uio_resid == 0) + return (0); + if (uio->uio_offset < 0) + return (EINVAL); + ip->i_flag |= IACC; + fs = ip->i_fs; + do { + lbn = lblkno(fs, uio->uio_offset); + on = blkoff(fs, uio->uio_offset); + n = MIN((unsigned)(fs->fs_bsize - on), uio->uio_resid); + diff = ip->i_size - uio->uio_offset; + if (diff <= 0) + return (0); + if (diff < n) + n = diff; + size = blksize(fs, ip, lbn); + rablock = lbn + 1; + if (vp->v_lastr + 1 == lbn && + lblktosize(fs, rablock) < ip->i_size) + error = breada(ITOV(ip), lbn, size, rablock, + blksize(fs, ip, rablock), NOCRED, &bp); + else + error = bread(ITOV(ip), lbn, size, NOCRED, &bp); + vp->v_lastr = lbn; + n = MIN(n, size - bp->b_resid); + if (error) { + brelse(bp); + return (error); + } + error = uiomove(bp->b_un.b_addr + on, (int)n, uio); + if (n + on == fs->fs_bsize || uio->uio_offset == ip->i_size) + bp->b_flags |= B_AGE; + brelse(bp); + } while (error == 0 && uio->uio_resid > 0 && n != 0); + return (error); +} - if ((ip = owner(uap->fname, FOLLOW)) == NULL) - return; - u.u_error = copyin((caddr_t)uap->tptr, (caddr_t)tv, sizeof (tv)); - if (u.u_error == 0) { - ip->i_flag |= IACC|IUPD|ICHG; - iupdat(ip, &tv[0], &tv[1], 0); +/* + * Vnode op for writing. + */ +ufs_write(vp, uio, ioflag, cred) + register struct vnode *vp; + struct uio *uio; + int ioflag; + struct ucred *cred; +{ + register struct inode *ip = VTOI(vp); + register struct fs *fs; + struct buf *bp; + daddr_t lbn, bn; + u_long osize; + int i, n, on, flags; + int count, size, resid, error = 0; + + if (uio->uio_rw != UIO_WRITE) + panic("ufs_write mode"); + switch (vp->v_type) { + case VREG: + if (ioflag & IO_APPEND) + uio->uio_offset = ip->i_size; + /* fall through */ + case VLNK: + break; + + case VDIR: + if ((ioflag & IO_SYNC) == 0) + panic("ufs_write nonsync dir write"); + break; + + default: + panic("ufs_write type"); } - iput(ip); + if (uio->uio_offset < 0) + return (EINVAL); + if (uio->uio_resid == 0) + return (0); + /* + * Maybe this should be above the vnode op call, but so long as + * file servers have no limits, i don't think it matters + */ + if (vp->v_type == VREG && + uio->uio_offset + uio->uio_resid > + u.u_rlimit[RLIMIT_FSIZE].rlim_cur) { + psignal(u.u_procp, SIGXFSZ); + return (EFBIG); + } + resid = uio->uio_resid; + osize = ip->i_size; + fs = ip->i_fs; + flags = 0; + if (ioflag & IO_SYNC) + flags = B_SYNC; + do { + lbn = lblkno(fs, uio->uio_offset); + on = blkoff(fs, uio->uio_offset); + n = MIN((unsigned)(fs->fs_bsize - on), uio->uio_resid); + if (n < fs->fs_bsize) + flags |= B_CLRBUF; + else + flags &= ~B_CLRBUF; + if (error = balloc(ip, lbn, (int)(on + n), &bp, flags)) + break; + bn = bp->b_blkno; + if (uio->uio_offset + n > ip->i_size) + ip->i_size = uio->uio_offset + n; + size = blksize(fs, ip, lbn); + count = howmany(size, CLBYTES); + for (i = 0; i < count; i++) + munhash(ip->i_devvp, bn + i * CLBYTES / DEV_BSIZE); + n = MIN(n, size - bp->b_resid); + error = uiomove(bp->b_un.b_addr + on, n, uio); + if (ioflag & IO_SYNC) + (void) bwrite(bp); + else if (n + on == fs->fs_bsize) { + bp->b_flags |= B_AGE; + bawrite(bp); + } else + bdwrite(bp); + ip->i_flag |= IUPD|ICHG; + if (cred->cr_uid != 0) + ip->i_mode &= ~(ISUID|ISGID); + } while (error == 0 && uio->uio_resid > 0 && n != 0); + if (error && (ioflag & IO_UNIT)) { + (void) itrunc(ip, osize, ioflag & IO_SYNC); + uio->uio_offset -= resid - uio->uio_resid; + uio->uio_resid = resid; + } + return (error); +} + +/* ARGSUSED */ +ufs_ioctl(vp, com, data, fflag, cred) + struct vnode *vp; + int com; + caddr_t data; + int fflag; + struct ucred *cred; +{ + + return (ENOTTY); +} + +/* ARGSUSED */ +ufs_select(vp, which, fflags, cred) + struct vnode *vp; + int which, fflags; + struct ucred *cred; +{ + + return (1); /* XXX */ } /* - * Flush any pending I/O. + * Mmap a file + * + * NB Currently unsupported. */ -sync() +/* ARGSUSED */ +ufs_mmap(vp, fflags, cred) + struct vnode *vp; + int fflags; + struct ucred *cred; { - update(); + return (EINVAL); } /* - * Truncate a file given its path name. + * Synch an open file. */ -truncate() +/* ARGSUSED */ +ufs_fsync(vp, fflags, cred, waitfor) + struct vnode *vp; + int fflags; + struct ucred *cred; + int waitfor; { - struct a { - char *fname; - u_long length; - } *uap = (struct a *)u.u_ap; - struct inode *ip; - register struct nameidata *ndp = &u.u_nd; + struct inode *ip = VTOI(vp); - ndp->ni_nameiop = LOOKUP | FOLLOW; - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = uap->fname; - ip = namei(ndp); - if (ip == NULL) - return; - if (access(ip, IWRITE)) - goto bad; - if ((ip->i_mode&IFMT) == IFDIR) { - u.u_error = EISDIR; - goto bad; - } - itrunc(ip, uap->length); -bad: - iput(ip); + if (fflags&FWRITE) + ip->i_flag |= ICHG; + vflushbuf(vp, waitfor == MNT_WAIT ? B_SYNC : 0); + return (iupdat(ip, &time, &time, waitfor == MNT_WAIT)); } /* - * Truncate a file given a file descriptor. + * Seek on a file + * + * Nothing to do, so just return. */ -ftruncate() +/* ARGSUSED */ +ufs_seek(vp, oldoff, newoff, cred) + struct vnode *vp; + off_t oldoff, newoff; + struct ucred *cred; { - struct a { - int fd; - u_long length; - } *uap = (struct a *)u.u_ap; - struct inode *ip; - struct file *fp; - fp = getinode(uap->fd); - if (fp == NULL) - return; - if ((fp->f_flag&FWRITE) == 0) { - u.u_error = EINVAL; - return; + return (0); +} + +/* + * ufs remove + * Hard to avoid races here, especially + * in unlinking directories. + */ +ufs_remove(ndp) + struct nameidata *ndp; +{ + register struct inode *ip, *dp; + int error; + + ip = VTOI(ndp->ni_vp); + dp = VTOI(ndp->ni_dvp); + error = dirremove(ndp); + if (!error) { + ip->i_nlink--; + ip->i_flag |= ICHG; } - ip = (struct inode *)fp->f_data; - ILOCK(ip); - itrunc(ip, uap->length); - IUNLOCK(ip); + if (dp == ip) + vrele(ITOV(ip)); + else + iput(ip); + iput(dp); + return (error); } /* - * Synch an open file. + * link vnode call */ -fsync() +ufs_link(vp, ndp) + register struct vnode *vp; + register struct nameidata *ndp; { - struct a { - int fd; - } *uap = (struct a *)u.u_ap; - struct inode *ip; - struct file *fp; + register struct inode *ip = VTOI(vp); + int error; - fp = getinode(uap->fd); - if (fp == NULL) - return; - ip = (struct inode *)fp->f_data; - ILOCK(ip); - syncip(ip); - IUNLOCK(ip); + if (ndp->ni_dvp != vp) + ILOCK(ip); + if (ip->i_nlink == LINK_MAX - 1) { + error = EMLINK; + goto out; + } + ip->i_nlink++; + ip->i_flag |= ICHG; + error = iupdat(ip, &time, &time, 1); + if (!error) + error = direnter(ip, ndp); +out: + if (ndp->ni_dvp != vp) + IUNLOCK(ip); + if (error) { + ip->i_nlink--; + ip->i_flag |= ICHG; + } + return (error); } /* @@ -779,60 +784,45 @@ fsync() * Basic algorithm is: * * 1) Bump link count on source while we're linking it to the - * target. This also insure the inode won't be deleted out - * from underneath us while we work. + * target. This also ensure the inode won't be deleted out + * from underneath us while we work (it may be truncated by + * a concurrent `trunc' or `open' for creation). * 2) Link source to destination. If destination already exists, * delete it first. - * 3) Unlink source reference to inode if still around. - * 4) If a directory was moved and the parent of the destination + * 3) Unlink source reference to inode if still around. If a + * directory was moved and the parent of the destination * is different from the source, patch the ".." entry in the * directory. - * - * Source and destination must either both be directories, or both - * not be directories. If target is a directory, it must be empty. */ -rename() +ufs_rename(fndp, tndp) + register struct nameidata *fndp, *tndp; { - struct a { - char *from; - char *to; - } *uap = (struct a *)u.u_ap; register struct inode *ip, *xp, *dp; - struct inode *zp; - int oldparent, parentdifferent, doingdirectory; - register struct nameidata *ndp = &u.u_nd; + struct dirtemplate dirbuf; + int doingdirectory = 0, oldparent = 0, newparent = 0; int error = 0; - ndp->ni_nameiop = DELETE | LOCKPARENT; - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = uap->from; - ip = namei(ndp); - if (ip == NULL) - return; - dp = ndp->ni_pdir; - oldparent = 0, doingdirectory = 0; + dp = VTOI(fndp->ni_dvp); + ip = VTOI(fndp->ni_vp); + ILOCK(ip); if ((ip->i_mode&IFMT) == IFDIR) { - register struct direct *d; + register struct direct *d = &fndp->ni_dent; - d = &ndp->ni_dent; /* * Avoid ".", "..", and aliases of "." for obvious reasons. */ - if ((d->d_namlen == 1 && d->d_name[0] == '.') || - (d->d_namlen == 2 && bcmp(d->d_name, "..", 2) == 0) || - (dp == ip)) { - iput(dp); - if (dp == ip) - irele(ip); - else - iput(ip); - u.u_error = EINVAL; - return; + if ((d->d_namlen == 1 && d->d_name[0] == '.') || dp == ip || + fndp->ni_isdotdot || (ip->i_flag & IRENAME)) { + IUNLOCK(ip); + ufs_abortop(fndp); + ufs_abortop(tndp); + return (EINVAL); } + ip->i_flag |= IRENAME; oldparent = dp->i_number; doingdirectory++; } - iput(dp); + vrele(fndp->ni_dvp); /* * 1) Bump link count while we're moving stuff @@ -842,21 +832,17 @@ rename() */ ip->i_nlink++; ip->i_flag |= ICHG; - iupdat(ip, &time, &time, 1); + error = iupdat(ip, &time, &time, 1); IUNLOCK(ip); /* * When the target exists, both the directory - * and target inodes are returned locked. + * and target vnodes are returned locked. */ - ndp->ni_nameiop = CREATE | LOCKPARENT | NOCACHE; - ndp->ni_dirp = (caddr_t)uap->to; - xp = namei(ndp); - if (u.u_error) { - error = u.u_error; - goto out; - } - dp = ndp->ni_pdir; + dp = VTOI(tndp->ni_dvp); + xp = NULL; + if (tndp->ni_vp) + xp = VTOI(tndp->ni_vp); /* * If ".." must be changed (ie the directory gets a new * parent) then the source directory must not be in the @@ -867,23 +853,24 @@ rename() * to namei, as the parent directory is unlocked by the * call to checkpath(). */ - parentdifferent = oldparent != dp->i_number; - if (doingdirectory && parentdifferent) { - if (access(ip, IWRITE)) + if (oldparent != dp->i_number) + newparent = dp->i_number; + if (doingdirectory && newparent) { + if (error = iaccess(ip, IWRITE, tndp->ni_cred)) goto bad; + tndp->ni_nameiop = RENAME | LOCKPARENT | LOCKLEAF | NOCACHE; do { - dp = ndp->ni_pdir; + dp = VTOI(tndp->ni_dvp); if (xp != NULL) iput(xp); - u.u_error = checkpath(ip, dp); - if (u.u_error) + if (error = checkpath(ip, dp, tndp->ni_cred)) goto out; - xp = namei(ndp); - if (u.u_error) { - error = u.u_error; + if (error = namei(tndp)) goto out; - } - } while (dp != ndp->ni_pdir); + xp = NULL; + if (tndp->ni_vp) + xp = VTOI(tndp->ni_vp); + } while (dp != VTOI(tndp->ni_dvp)); } /* * 2) If target doesn't exist, link the target @@ -893,35 +880,40 @@ rename() * expunge the original entry's existence. */ if (xp == NULL) { - if (dp->i_dev != ip->i_dev) { - error = EXDEV; - goto bad; - } + if (dp->i_dev != ip->i_dev) + panic("rename: EXDEV"); /* - * Account for ".." in directory. - * When source and destination have the - * same parent we don't fool with the - * link count -- this isn't required - * because we do a similar check below. + * Account for ".." in new directory. + * When source and destination have the same + * parent we don't fool with the link count. */ - if (doingdirectory && parentdifferent) { + if (doingdirectory && newparent) { dp->i_nlink++; dp->i_flag |= ICHG; - iupdat(dp, &time, &time, 1); + error = iupdat(dp, &time, &time, 1); } - error = direnter(ip, ndp); - if (error) + if (error = direnter(ip, tndp)) goto out; } else { - if (xp->i_dev != dp->i_dev || xp->i_dev != ip->i_dev) { - error = EXDEV; - goto bad; - } + if (xp->i_dev != dp->i_dev || xp->i_dev != ip->i_dev) + panic("rename: EXDEV"); /* * Short circuit rename(foo, foo). */ if (xp->i_number == ip->i_number) + panic("rename: same file"); + /* + * If the parent directory is "sticky", then the user must + * own the parent directory, or the destination of the rename, + * otherwise the destination may not be changed (except by + * root). This implements append-only directories. + */ + if ((dp->i_mode & ISVTX) && tndp->ni_cred->cr_uid != 0 && + tndp->ni_cred->cr_uid != dp->i_uid && + xp->i_uid != tndp->ni_cred->cr_uid) { + error = EPERM; goto bad; + } /* * Target must be empty if a directory * and have no links to it. @@ -930,7 +922,8 @@ rename() * not directories). */ if ((xp->i_mode&IFMT) == IFDIR) { - if (!dirempty(xp) || xp->i_nlink > 2) { + if (!dirempty(xp, dp->i_number, tndp->ni_cred) || + xp->i_nlink > 2) { error = ENOTEMPTY; goto bad; } @@ -938,15 +931,14 @@ rename() error = ENOTDIR; goto bad; } + cache_purge(ITOV(dp)); } else if (doingdirectory) { error = EISDIR; goto bad; } - dirrewrite(dp, ip, ndp); - if (u.u_error) { - error = u.u_error; - goto bad1; - } + if (error = dirrewrite(dp, ip, tndp)) + goto bad; + vput(ITOV(dp)); /* * Adjust the link count of the target to * reflect the dirrewrite above. If this is @@ -954,15 +946,14 @@ rename() * no links to it, so we can squash the inode and * any space associated with it. We disallowed * renaming over top of a directory with links to - * it above, as we've no way to determine if - * we've got a link or the directory itself, and - * if we get a link, then ".." will be screwed up. + * it above, as the remaining link would point to + * a directory without "." or ".." entries. */ xp->i_nlink--; if (doingdirectory) { if (--xp->i_nlink != 0) panic("rename: linked directory"); - itrunc(xp, (u_long)0); + error = itrunc(xp, (u_long)0, IO_SYNC); } xp->i_flag |= ICHG; iput(xp); @@ -972,151 +963,84 @@ rename() /* * 3) Unlink the source. */ - ndp->ni_nameiop = DELETE | LOCKPARENT; - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = uap->from; - zp = namei(ndp); - dp = ndp->ni_pdir; + fndp->ni_nameiop = DELETE | LOCKPARENT | LOCKLEAF; + (void)namei(fndp); + if (fndp->ni_vp != NULL) { + xp = VTOI(fndp->ni_vp); + dp = VTOI(fndp->ni_dvp); + } else { + if (fndp->ni_dvp != NULL) + vput(fndp->ni_dvp); + xp = NULL; + dp = NULL; + } /* - * Insure directory entry still exists and - * has not changed since the start of all - * this. If either has occured, forget about - * about deleting the original entry. + * Ensure that the directory entry still exists and has not + * changed while the new name has been entered. If the source is + * a file then the entry may have been unlinked or renamed. In + * either case there is no further work to be done. If the source + * is a directory then it cannot have been rmdir'ed; its link + * count of three would cause a rmdir to fail with ENOTEMPTY. + * The IRENAME flag ensures that it cannot be moved by another + * rename. */ - if (dp != NULL && zp == ip) { + if (xp != ip) { + if (doingdirectory) + panic("rename: lost dir entry"); + } else { /* - * If source is a directory, must adjust - * link count of parent directory also. - * If target didn't exist and source and - * target have the same parent, then we - * needn't touch the link count, it all - * balances out in the end. Otherwise, we - * must do so to reflect deletion of ".." - * done above. + * If the source is a directory with a + * new parent, the link count of the old + * parent directory must be decremented + * and ".." set to point to the new parent. */ - if (doingdirectory && (xp != NULL || parentdifferent)) { + if (doingdirectory && newparent) { dp->i_nlink--; dp->i_flag |= ICHG; + error = vn_rdwr(UIO_READ, ITOV(xp), (caddr_t)&dirbuf, + sizeof (struct dirtemplate), (off_t)0, + UIO_SYSSPACE, IO_NODELOCKED, + tndp->ni_cred, (int *)0); + if (error == 0) { + if (dirbuf.dotdot_namlen != 2 || + dirbuf.dotdot_name[0] != '.' || + dirbuf.dotdot_name[1] != '.') { + dirbad(xp, 12, "rename: mangled dir"); + } else { + dirbuf.dotdot_ino = newparent; + (void) vn_rdwr(UIO_WRITE, ITOV(xp), + (caddr_t)&dirbuf, + sizeof (struct dirtemplate), + (off_t)0, UIO_SYSSPACE, + IO_NODELOCKED|IO_SYNC, + tndp->ni_cred, (int *)0); + cache_purge(ITOV(dp)); + } + } } - if (dirremove(ndp)) { - zp->i_nlink--; - zp->i_flag |= ICHG; + error = dirremove(fndp); + if (!error) { + xp->i_nlink--; + xp->i_flag |= ICHG; } - if (error == 0) /* conservative */ - error = u.u_error; + xp->i_flag &= ~IRENAME; } - if (zp != NULL) - iput(zp); - irele(ip); if (dp) - iput(dp); - - /* - * 4) Renaming a directory with the parent - * different requires ".." to be rewritten. - * The window is still there for ".." to - * be inconsistent, but this is unavoidable, - * and a lot shorter than when it was done - * in a user process. - */ - if (doingdirectory && parentdifferent && error == 0) { - struct dirtemplate dirbuf; - - ndp->ni_nameiop = LOOKUP | LOCKPARENT; - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = uap->to; - ip = namei(ndp); - if (ip == NULL) { - printf("rename: .. went away\n"); - return; - } - dp = ndp->ni_pdir; - if ((ip->i_mode&IFMT) != IFDIR) { - printf("rename: .. not a directory\n"); - goto stuck; - } - error = rdwri(UIO_READ, ip, (caddr_t)&dirbuf, - sizeof (struct dirtemplate), (off_t)0, 1, (int *)0); - if (error == 0) { - dirbuf.dotdot_ino = dp->i_number; - cacheinval(dp); - (void) rdwri(UIO_WRITE, ip, (caddr_t)&dirbuf, - sizeof (struct dirtemplate), (off_t)0, 1, (int *)0); - } -stuck: - irele(dp); - iput(ip); - } - goto done; + vput(ITOV(dp)); + if (xp) + vput(ITOV(xp)); + vrele(ITOV(ip)); + return (error); bad: - iput(dp); -bad1: if (xp) - iput(xp); + vput(ITOV(xp)); + vput(ITOV(dp)); out: ip->i_nlink--; ip->i_flag |= ICHG; - irele(ip); -done: - if (error) - u.u_error = error; -} - -/* - * Make a new file. - */ -struct inode * -maknode(mode, ndp) - int mode; - register struct nameidata *ndp; -{ - register struct inode *ip; - register struct inode *pdir = ndp->ni_pdir; - ino_t ipref; - - if ((mode & IFMT) == IFDIR) - ipref = dirpref(pdir->i_fs); - else - ipref = pdir->i_number; - ip = ialloc(pdir, ipref, mode); - if (ip == NULL) { - iput(pdir); - return (NULL); - } -#ifdef QUOTA - if (ip->i_dquot != NODQUOT) - panic("maknode: dquot"); -#endif - ip->i_flag |= IACC|IUPD|ICHG; - if ((mode & IFMT) == 0) - mode |= IFREG; - ip->i_mode = mode & ~u.u_cmask; - ip->i_nlink = 1; - ip->i_uid = u.u_uid; - ip->i_gid = pdir->i_gid; - if (ip->i_mode & ISGID && !groupmember(ip->i_gid)) - ip->i_mode &= ~ISGID; -#ifdef QUOTA - ip->i_dquot = inoquota(ip); -#endif - - /* - * Make sure inode goes to disk before directory entry. - */ - iupdat(ip, &time, &time, 1); - u.u_error = direnter(ip, ndp); - if (u.u_error) { - /* - * Write error occurred trying to update directory - * so must deallocate the inode. - */ - ip->i_nlink = 0; - ip->i_flag |= ICHG; - iput(ip); - return (NULL); - } - return (ip); + vrele(ITOV(ip)); + return (error); } /* @@ -1130,30 +1054,21 @@ struct dirtemplate mastertemplate = { /* * Mkdir system call */ -mkdir() +ufs_mkdir(ndp, vap) + struct nameidata *ndp; + struct vattr *vap; { - struct a { - char *name; - int dmode; - } *uap = (struct a *)u.u_ap; register struct inode *ip, *dp; + struct inode *tip; + struct vnode *dvp; struct dirtemplate dirtemplate; - register struct nameidata *ndp = &u.u_nd; + int error; + int dmode; - ndp->ni_nameiop = CREATE; - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = uap->name; - ip = namei(ndp); - if (u.u_error) - return; - if (ip != NULL) { - iput(ip); - u.u_error = EEXIST; - return; - } - dp = ndp->ni_pdir; - uap->dmode &= 0777; - uap->dmode |= IFDIR; + dvp = ndp->ni_dvp; + dp = VTOI(dvp); + dmode = vap->va_mode&0777; + dmode |= IFDIR; /* * Must simulate part of maknode here * in order to acquire the inode, but @@ -1161,24 +1076,26 @@ mkdir() * directory. The entry is made later * after writing "." and ".." entries out. */ - ip = ialloc(dp, dirpref(dp->i_fs), uap->dmode); - if (ip == NULL) { + error = ialloc(dp, dirpref(dp->i_fs), dmode, &tip); + if (error) { iput(dp); - return; + return (error); } + ip = tip; #ifdef QUOTA if (ip->i_dquot != NODQUOT) panic("mkdir: dquot"); #endif ip->i_flag |= IACC|IUPD|ICHG; - ip->i_mode = uap->dmode & ~u.u_cmask; + ip->i_mode = dmode; + ITOV(ip)->v_type = VDIR; /* Rest init'd in iget() */ ip->i_nlink = 2; - ip->i_uid = u.u_uid; + ip->i_uid = ndp->ni_cred->cr_uid; ip->i_gid = dp->i_gid; #ifdef QUOTA ip->i_dquot = inoquota(ip); #endif - iupdat(ip, &time, &time, 1); + error = iupdat(ip, &time, &time, 1); /* * Bump link count in parent directory @@ -1188,7 +1105,7 @@ mkdir() */ dp->i_nlink++; dp->i_flag |= ICHG; - iupdat(dp, &time, &time, 1); + error = iupdat(dp, &time, &time, 1); /* * Initialize directory with "." @@ -1197,26 +1114,30 @@ mkdir() dirtemplate = mastertemplate; dirtemplate.dot_ino = ip->i_number; dirtemplate.dotdot_ino = dp->i_number; - u.u_error = rdwri(UIO_WRITE, ip, (caddr_t)&dirtemplate, - sizeof (dirtemplate), (off_t)0, 1, (int *)0); - if (u.u_error) { + error = vn_rdwr(UIO_WRITE, ITOV(ip), (caddr_t)&dirtemplate, + sizeof (dirtemplate), (off_t)0, UIO_SYSSPACE, + IO_NODELOCKED|IO_SYNC, ndp->ni_cred, (int *)0); + if (error) { dp->i_nlink--; dp->i_flag |= ICHG; goto bad; } + if (DIRBLKSIZ > dp->i_fs->fs_fsize) + panic("mkdir: blksize"); /* XXX - should grow w/balloc() */ + else + ip->i_size = DIRBLKSIZ; /* * Directory all set up, now * install the entry for it in * the parent directory. */ - u.u_error = direnter(ip, ndp); + error = direnter(ip, ndp); dp = NULL; - if (u.u_error) { + if (error) { ndp->ni_nameiop = LOOKUP | NOCACHE; - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = uap->name; - dp = namei(ndp); - if (dp) { + error = namei(ndp); + if (!error) { + dp = VTOI(ndp->ni_vp); dp->i_nlink--; dp->i_flag |= ICHG; } @@ -1224,55 +1145,38 @@ mkdir() bad: /* * No need to do an explicit itrunc here, - * irele will do this for us because we set + * vrele will do this for us because we set * the link count to 0. */ - if (u.u_error) { + if (error) { ip->i_nlink = 0; ip->i_flag |= ICHG; - } + iput(ip); + } else + ndp->ni_vp = ITOV(ip); if (dp) iput(dp); - iput(ip); + return (error); } /* * Rmdir system call. */ -rmdir() +ufs_rmdir(ndp) + register struct nameidata *ndp; { - struct a { - char *name; - } *uap = (struct a *)u.u_ap; register struct inode *ip, *dp; - register struct nameidata *ndp = &u.u_nd; + int error = 0; - ndp->ni_nameiop = DELETE | LOCKPARENT; - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = uap->name; - ip = namei(ndp); - if (ip == NULL) - return; - dp = ndp->ni_pdir; + ip = VTOI(ndp->ni_vp); + dp = VTOI(ndp->ni_dvp); /* * No rmdir "." please. */ if (dp == ip) { - irele(dp); + vrele(ITOV(dp)); iput(ip); - u.u_error = EINVAL; - return; - } - if ((ip->i_mode&IFMT) != IFDIR) { - u.u_error = ENOTDIR; - goto out; - } - /* - * Don't remove a mounted on directory. - */ - if (ip->i_dev != dp->i_dev) { - u.u_error = EBUSY; - goto out; + return (EINVAL); } /* * Verify the directory is empty (and valid). @@ -1281,8 +1185,8 @@ rmdir() * the current directory and thus be * non-empty.) */ - if (ip->i_nlink != 2 || !dirempty(ip)) { - u.u_error = ENOTEMPTY; + if (ip->i_nlink != 2 || !dirempty(ip, dp->i_number, ndp->ni_cred)) { + error = ENOTEMPTY; goto out; } /* @@ -1290,12 +1194,13 @@ rmdir() * inode. If we crash in between, the directory * will be reattached to lost+found, */ - if (dirremove(ndp) == 0) + if (error = dirremove(ndp)) goto out; dp->i_nlink--; dp->i_flag |= ICHG; + cache_purge(ITOV(dp)); iput(dp); - dp = NULL; + ndp->ni_dvp = NULL; /* * Truncate inode. The only stuff left * in the directory is "." and "..". The @@ -1308,40 +1213,391 @@ rmdir() * worry about them later. */ ip->i_nlink -= 2; - itrunc(ip, (u_long)0); - cacheinval(ip); + error = itrunc(ip, (u_long)0, IO_SYNC); + cache_purge(ITOV(ip)); out: - if (dp) + if (ndp->ni_dvp) iput(dp); iput(ip); + return (error); +} + +/* + * symlink -- make a symbolic link + */ +ufs_symlink(ndp, vap, target) + struct nameidata *ndp; + struct vattr *vap; + char *target; +{ + struct inode *ip; + int error; + + error = maknode(IFLNK | vap->va_mode, ndp, &ip); + if (error) + return (error); + error = vn_rdwr(UIO_WRITE, ITOV(ip), target, strlen(target), (off_t)0, + UIO_SYSSPACE, IO_NODELOCKED, ndp->ni_cred, (int *)0); + iput(ip); + return (error); +} + +/* + * Vnode op for read and write + */ +ufs_readdir(vp, uio, cred, eofflagp) + struct vnode *vp; + register struct uio *uio; + struct ucred *cred; + int *eofflagp; +{ + int count, lost, error; + + count = uio->uio_resid; + count &= ~(DIRBLKSIZ - 1); + lost = uio->uio_resid - count; + if (count < DIRBLKSIZ || (uio->uio_offset & (DIRBLKSIZ -1))) + return (EINVAL); + uio->uio_resid = count; + uio->uio_iov->iov_len = count; + error = ufs_read(vp, uio, 0, cred); + uio->uio_resid += lost; + if ((VTOI(vp)->i_size - uio->uio_offset) <= 0) + *eofflagp = 1; + else + *eofflagp = 0; + return (error); +} + +/* + * Return target name of a symbolic link + */ +ufs_readlink(vp, uiop, cred) + struct vnode *vp; + struct uio *uiop; + struct ucred *cred; +{ + + return (ufs_read(vp, uiop, 0, cred)); +} + +/* + * Ufs abort op, called after namei() when a CREATE/DELETE isn't actually + * done. Iff ni_vp/ni_dvp not null and locked, unlock. + */ +ufs_abortop(ndp) + register struct nameidata *ndp; +{ + + if (ndp->ni_dvp) { + if (VOP_ISLOCKED(ndp->ni_dvp)) + VOP_UNLOCK(ndp->ni_dvp); + vrele(ndp->ni_dvp); + } + if (ndp->ni_vp) { + if (VOP_ISLOCKED(ndp->ni_vp)) + VOP_UNLOCK(ndp->ni_vp); + vrele(ndp->ni_vp); + } + return; +} + +/* + * Lock an inode. + */ +ufs_lock(vp) + struct vnode *vp; +{ + register struct inode *ip = VTOI(vp); + + ILOCK(ip); + return (0); +} + +/* + * Unlock an inode. + */ +ufs_unlock(vp) + struct vnode *vp; +{ + register struct inode *ip = VTOI(vp); + + if (!(ip->i_flag & ILOCKED)) + panic("ufs_unlock NOT LOCKED"); + IUNLOCK(ip); + return (0); +} + +/* + * Check for a locked inode. + */ +ufs_islocked(vp) + struct vnode *vp; +{ + + if (VTOI(vp)->i_flag & ILOCKED) + return (1); + return (0); } -struct file * -getinode(fdes) - int fdes; +/* + * Get access to bmap + */ +ufs_bmap(vp, bn, vpp, bnp) + struct vnode *vp; + daddr_t bn; + struct vnode **vpp; + daddr_t *bnp; { - struct file *fp; + struct inode *ip = VTOI(vp); + + if (vpp != NULL) + *vpp = ip->i_devvp; + if (bnp == NULL) + return (0); + return (bmap(ip, bn, bnp, (daddr_t *)0, (int *)0)); +} + +/* + * Just call the device strategy routine + */ +int checkoverlap = 1; - if ((unsigned)fdes >= NOFILE || (fp = u.u_ofile[fdes]) == NULL) { - u.u_error = EBADF; - return ((struct file *)0); +ufs_strategy(bp) + register struct buf *bp; +{ + register struct inode *ip = VTOI(bp->b_vp); + register struct buf *ep; + struct vnode *vp; + struct buf *ebp; + daddr_t start, last; + int error; + + if (bp->b_vp->v_type == VBLK || bp->b_vp->v_type == VCHR) + panic("ufs_strategy: spec"); + if (bp->b_blkno == bp->b_lblkno) { + if (error = bmap(ip, bp->b_lblkno, &bp->b_blkno)) + return (error); + if ((long)bp->b_blkno == -1) + clrbuf(bp); + } + if ((long)bp->b_blkno == -1) { + biodone(bp); + return (0); } - if (fp->f_type != DTYPE_INODE) { - u.u_error = EINVAL; - return ((struct file *)0); + if (checkoverlap) { + ebp = &buf[nbuf]; + start = bp->b_blkno; + last = start + btodb(bp->b_bcount) - 1; + for (ep = buf; ep < ebp; ep++) { + if (ep == bp || (ep->b_flags & B_INVAL) || + ep->b_vp == (struct vnode *)0) + continue; + if (VOP_BMAP(ep->b_vp, (daddr_t)0, &vp, (daddr_t)0)) + continue; + if (vp != ip->i_devvp) + continue; + /* look for overlap */ + if (ep->b_bcount == 0 || ep->b_blkno > last || + ep->b_blkno + btodb(ep->b_bcount) <= start) + continue; + vprint("Disk overlap", vp); + printf("\tstart %d, end %d overlap start %d, end %d\n", + start, last, ep->b_blkno, + ep->b_blkno + btodb(ep->b_bcount) - 1); + } } - return (fp); + vp = ip->i_devvp; + bp->b_dev = vp->v_rdev; + (*(vp->v_op->vn_strategy))(bp); + return (0); +} + +/* + * Print out the contents of an inode. + */ +ufs_print(vp) + struct vnode *vp; +{ + register struct inode *ip = VTOI(vp); + + printf("tag VT_UFS, ino %d, on dev %d, %d", ip->i_number, + major(ip->i_dev), minor(ip->i_dev)); +#ifdef FIFO + if (vp->v_type == VFIFO) + fifo_printinfo(vp); +#endif /* FIFO */ + printf("%s\n", (ip->i_flag & ILOCKED) ? " (LOCKED)" : ""); + if (ip->i_spare0 == 0) + return; + printf("\towner pid %d", ip->i_spare0); + if (ip->i_spare1) + printf(" waiting pid %d", ip->i_spare1); + printf("\n"); +} + +/* + * Read wrapper for special devices. + */ +ufsspec_read(vp, uio, ioflag, cred) + struct vnode *vp; + struct uio *uio; + int ioflag; + struct ucred *cred; +{ + + /* + * Set access flag. + */ + VTOI(vp)->i_flag |= IACC; + return (spec_read(vp, uio, ioflag, cred)); +} + +/* + * Write wrapper for special devices. + */ +ufsspec_write(vp, uio, ioflag, cred) + struct vnode *vp; + struct uio *uio; + int ioflag; + struct ucred *cred; +{ + + /* + * Set update and change flags. + */ + VTOI(vp)->i_flag |= IUPD|ICHG; + return (spec_write(vp, uio, ioflag, cred)); +} + +/* + * Close wrapper for special devices. + * + * Update the times on the inode then do device close. + */ +ufsspec_close(vp, fflag, cred) + struct vnode *vp; + int fflag; + struct ucred *cred; +{ + register struct inode *ip = VTOI(vp); + + if (vp->v_usecount > 1 && !(ip->i_flag & ILOCKED)) + ITIMES(ip, &time, &time); + return (spec_close(vp, fflag, cred)); +} + +#ifdef FIFO +/* + * Read wrapper for fifo's + */ +ufsfifo_read(vp, uio, ioflag, cred) + struct vnode *vp; + struct uio *uio; + int ioflag; + struct ucred *cred; +{ + + /* + * Set access flag. + */ + VTOI(vp)->i_flag |= IACC; + return (fifo_read(vp, uio, ioflag, cred)); +} + +/* + * Write wrapper for fifo's. + */ +ufsfifo_write(vp, uio, ioflag, cred) + struct vnode *vp; + struct uio *uio; + int ioflag; + struct ucred *cred; +{ + + /* + * Set update and change flags. + */ + VTOI(vp)->i_flag |= IUPD|ICHG; + return (fifo_write(vp, uio, ioflag, cred)); +} + +/* + * Close wrapper for fifo's. + * + * Update the times on the inode then do device close. + */ +ufsfifo_close(vp, fflag, cred) + struct vnode *vp; + int fflag; + struct ucred *cred; +{ + register struct inode *ip = VTOI(vp); + + if (vp->v_usecount > 1 && !(ip->i_flag & ILOCKED)) + ITIMES(ip, &time, &time); + return (fifo_close(vp, fflag, cred)); } +#endif /* FIFO */ /* - * mode mask for creation of files + * Make a new file. */ -umask() +maknode(mode, ndp, ipp) + int mode; + register struct nameidata *ndp; + struct inode **ipp; { - register struct a { - int mask; - } *uap = (struct a *)u.u_ap; + register struct inode *ip; + struct inode *tip; + register struct inode *pdir = VTOI(ndp->ni_dvp); + ino_t ipref; + int error; + + *ipp = 0; + if ((mode & IFMT) == IFDIR) + ipref = dirpref(pdir->i_fs); + else + ipref = pdir->i_number; + error = ialloc(pdir, ipref, mode, &tip); + if (error) { + iput(pdir); + return (error); + } + ip = tip; +#ifdef QUOTA + if (ip->i_dquot != NODQUOT) + panic("maknode: dquot"); +#endif + ip->i_flag |= IACC|IUPD|ICHG; + if ((mode & IFMT) == 0) + mode |= IFREG; + ip->i_mode = mode; + ITOV(ip)->v_type = IFTOVT(mode); /* Rest init'd in iget() */ + ip->i_nlink = 1; + ip->i_uid = ndp->ni_cred->cr_uid; + ip->i_gid = pdir->i_gid; + if ((ip->i_mode & ISGID) && !groupmember(ip->i_gid, ndp->ni_cred) && + suser(ndp->ni_cred, NULL)) + ip->i_mode &= ~ISGID; +#ifdef QUOTA + ip->i_dquot = inoquota(ip); +#endif - u.u_r.r_val1 = u.u_cmask; - u.u_cmask = uap->mask & 07777; + /* + * Make sure inode goes to disk before directory entry. + */ + if ((error = iupdat(ip, &time, &time, 1)) || + (error = direnter(ip, ndp))) { + /* + * Write error occurred trying to update the inode + * or the directory so must deallocate the inode. + */ + ip->i_nlink = 0; + ip->i_flag |= ICHG; + iput(ip); + return (error); + } + *ipp = ip; + return (0); }