X-Git-Url: https://git.subgeniuskitty.com/unix-history/.git/blobdiff_plain/68f215625447a44285aa05489c39e39d64fd2214..d7961b1db4e049660ccbe7026f356c9ee87c2cda:/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 7f801ffc2d..85c4a232e4 100644 --- a/usr/src/sys/ufs/ffs/ufs_vnops.c +++ b/usr/src/sys/ufs/ffs/ufs_vnops.c @@ -1,770 +1,778 @@ -/* ufs_vnops.c 6.13 84/07/27 */ - -#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. + * + * %sccs.include.redist.c% + * + * @(#)ufs_vnops.c 7.74 (Berkeley) %G% */ -chdir() -{ - chdirec(&u.u_cdir); +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#include +#include + +int ufs_chmod __P((struct vnode *, int, struct proc *)); +int ufs_chown __P((struct vnode *, u_int, u_int, struct proc *)); + +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, +}; + +#ifdef _NOQUAD +#define SETHIGH(q, h) (q).val[_QUAD_HIGHWORD] = (h) +#define SETLOW(q, l) (q).val[_QUAD_LOWWORD] = (l) +#else /* QUAD */ +union _qcvt { + quad_t qcvt; + long val[2]; +}; +#define SETHIGH(q, h) { \ + union _qcvt tmp; \ + tmp.qcvt = (q); \ + tmp.val[_QUAD_HIGHWORD] = (h); \ + (q) = tmp.qcvt; \ } +#define SETLOW(q, l) { \ + union _qcvt tmp; \ + tmp.qcvt = (q); \ + tmp.val[_QUAD_LOWWORD] = (l); \ + (q) = tmp.qcvt; \ +} +#endif /* QUAD */ /* - * Change notion of root (``/'') directory. + * Create a regular file */ -chroot() +int +ufs_create(dvp, vpp, cnp, vap) + struct vnode *dvp; + struct vnode **vpp; + struct componentname *cnp; + struct vattr *vap; { + int error; - if (suser()) - chdirec(&u.u_rdir); + if (error = + ufs_makeinode(MAKEIMODE(vap->va_type, vap->va_mode), dvp, vpp, cnp)) + return (error); + return (0); } /* - * Common routine for chroot and chdir. + * Mknod vnode call */ -chdirec(ipp) - register struct inode **ipp; +/* ARGSUSED */ +int +ufs_mknod(dvp, vpp, cnp, vap) + struct vnode *dvp; + struct vnode **vpp; + struct componentname *cnp; + struct vattr *vap; { 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 error; -bad: - iput(ip); + if (error = + ufs_makeinode(MAKEIMODE(vap->va_type, vap->va_mode), dvp, vpp, cnp)) + return (error); + ip = VTOI(*vpp); + 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; + } + /* + * 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. + */ + vput(*vpp); + (*vpp)->v_type = VNON; + vgone(*vpp); + *vpp = 0; + return (0); } /* - * Open system call. + * Open called. + * + * Nothing to do. */ -open() +/* ARGSUSED */ +int +ufs_open(vp, mode, cred, p) + struct vnode *vp; + int mode; + struct ucred *cred; + struct proc *p; { - struct a { - char *fname; - int mode; - int crtmode; - } *uap = (struct a *) u.u_ap; - copen(uap->mode-FOPEN, uap->crtmode, uap->fname); + return (0); } /* - * Creat system call. + * Close called + * + * Update the times on the inode. */ -creat() +/* ARGSUSED */ +int +ufs_close(vp, fflag, cred, p) + struct vnode *vp; + int fflag; + struct ucred *cred; + struct proc *p; { - struct a { - char *fname; - int fmode; - } *uap = (struct a *)u.u_ap; + register struct inode *ip; - copen(FWRITE|FCREAT|FTRUNC, uap->fmode, uap->fname); + ip = VTOI(vp); + if (vp->v_usecount > 1 && !(ip->i_flag & ILOCKED)) + ITIMES(ip, &time, &time); + return (0); } /* - * Common code for open and creat. - * Check permissions, allocate an open file structure, - * and call the device open routine if any. + * Check mode permission on inode pointer. Mode is READ, WRITE or EXEC. + * The mode is shifted to select the owner/group/other fields. The + * super user is granted all permissions. */ -copen(mode, arg, fname) +int +ufs_access(vp, mode, cred, p) + struct vnode *vp; register int mode; - int arg; - caddr_t fname; + struct ucred *cred; + struct proc *p; { - register struct inode *ip; - register struct file *fp; - register struct nameidata *ndp = &u.u_nd; - int i; - -#ifdef notdef - if ((mode&(FREAD|FWRITE)) == 0) { - u.u_error = EINVAL; - return; + register struct inode *ip = VTOI(vp); + register gid_t *gp; + int i, error; + +#ifdef DIAGNOSTIC + if (!VOP_ISLOCKED(vp)) { + vprint("ufs_access: not locked", vp); + panic("ufs_access: not locked"); } #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; - } +#ifdef QUOTA + if (mode & VWRITE) { + switch (vp->v_type) { + case VREG: case VDIR: case VLNK: + if (error = getinoquota(ip)) + return (error); } } - 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; +#endif /* QUOTA */ + /* + * If you're the super-user, you always get access. + */ + if (cred->cr_uid == 0) + return (0); + /* + * Access check is based on only one of owner, group, public. + * If not owner, then check group. If not a member of the + * group, then check public access. + */ + if (cred->cr_uid != ip->i_uid) { + mode >>= 3; + gp = cred->cr_groups; + for (i = 0; i < cred->cr_ngroups; i++, gp++) + if (ip->i_gid == *gp) + goto found; + mode >>= 3; +found: + ; } - 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); + if ((ip->i_mode & mode) != 0) + return (0); + return (EACCES); } -/* - * Mknod system call - */ -mknod() +/* ARGSUSED */ +int +ufs_getattr(vp, vap, cred, p) + struct vnode *vp; + register struct vattr *vap; + struct ucred *cred; + struct proc *p; { 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); + ip = VTOI(vp); + ITIMES(ip, &time, &time); + /* + * Copy from inode table + */ + 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; +#ifdef tahoe + vap->va_size = ip->i_size; + vap->va_size_rsv = 0; +#else + vap->va_qsize = ip->i_din.di_qsize; +#endif + 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 + vap->va_blocksize = vp->v_mount->mnt_stat.f_iosize; + vap->va_bytes = dbtob(ip->i_blocks); +#ifdef _NOQUAD + vap->va_bytes_rsv = 0; +#endif + vap->va_type = vp->v_type; + vap->va_filerev = ip->i_modrev; + return (0); } /* - * link system call + * Set attribute vnode op. called from several syscalls */ -link() +int +ufs_setattr(vp, vap, cred, p) + register struct vnode *vp; + register struct vattr *vap; + register struct ucred *cred; + struct proc *p; { - 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; + register struct inode *ip; + int error; + + /* + * Check for unsettable 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); } - 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; + /* + * Go through the fields and update iff not VNOVAL. + */ + if (vap->va_uid != (u_short)VNOVAL || vap->va_gid != (u_short)VNOVAL) + if (error = ufs_chown(vp, vap->va_uid, vap->va_gid, p)) + return (error); + if (vap->va_size != VNOVAL) { + if (vp->v_type == VDIR) + return (EISDIR); + if (error = VOP_TRUNCATE(vp, vap->va_size, 0)) /* IO_SYNC? */ + return (error); } - 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; + ip = VTOI(vp); + if (vap->va_atime.tv_sec != VNOVAL || vap->va_mtime.tv_sec != VNOVAL) { + if (cred->cr_uid != ip->i_uid && + (error = suser(cred, &p->p_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 = VOP_UPDATE(vp, &vap->va_atime, &vap->va_mtime, 1)) + return (error); } - u.u_error = direnter(ip, ndp); -out: - if (u.u_error) { - ip->i_nlink--; + error = 0; + if (vap->va_mode != (u_short)VNOVAL) + error = ufs_chmod(vp, (int)vap->va_mode, p); + if (vap->va_flags != VNOVAL) { + if (cred->cr_uid != ip->i_uid && + (error = suser(cred, &p->p_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; } - irele(ip); + return (error); } /* - * symlink -- make a symbolic link + * Change the mode on a file. + * Inode must be locked before calling. */ -symlink() +static int +ufs_chmod(vp, mode, p) + register struct vnode *vp; + register int mode; + struct proc *p; { - 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; + register struct ucred *cred = p->p_ucred; + register struct inode *ip = VTOI(vp); + int error; + + if (cred->cr_uid != ip->i_uid && + (error = suser(cred, &p->p_acflag))) + return (error); + if (cred->cr_uid) { + if (vp->v_type != VDIR && (mode & ISVTX)) + return (EFTYPE); + if (!groupmember(ip->i_gid, cred) && (mode & ISGID)) + return (EPERM); } - 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); + ip->i_mode &= ~07777; + ip->i_mode |= mode & 07777; + ip->i_flag |= ICHG; + if ((vp->v_flag & VTEXT) && (ip->i_mode & ISVTX) == 0) + (void) vnode_pager_uncache(vp); + return (0); } /* - * Unlink system call. - * Hard to avoid races here, especially - * in unlinking directories. + * Perform chown operation on inode ip; + * inode must be locked prior to call. */ -unlink() +static int +ufs_chown(vp, uid, gid, p) + register struct vnode *vp; + u_int uid; + u_int gid; + struct proc *p; { - struct a { - char *fname; - } *uap = (struct a *)u.u_ap; - register struct inode *ip, *dp; - register struct nameidata *ndp = &u.u_nd; - - 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; + register struct inode *ip = VTOI(vp); + register struct ucred *cred = p->p_ucred; + uid_t ouid; + gid_t ogid; + int error = 0; +#ifdef QUOTA + register int i; + long change; +#endif + + if (uid == (u_short)VNOVAL) + uid = ip->i_uid; + if (gid == (u_short)VNOVAL) + gid = ip->i_gid; /* - * Don't unlink a mounted file. + * 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 (ip->i_dev != dp->i_dev) { - u.u_error = EBUSY; - goto out; + if ((cred->cr_uid != ip->i_uid || uid != ip->i_uid || + !groupmember((gid_t)gid, cred)) && + (error = suser(cred, &p->p_acflag))) + return (error); + ouid = ip->i_uid; + ogid = ip->i_gid; +#ifdef QUOTA + if (error = getinoquota(ip)) + return (error); + if (ouid == uid) { + dqrele(vp, ip->i_dquot[USRQUOTA]); + ip->i_dquot[USRQUOTA] = NODQUOT; } - if (ip->i_flag&ITEXT) - xrele(ip); /* try once to free text */ - if (dirremove(ndp)) { - ip->i_nlink--; - ip->i_flag |= ICHG; + if (ogid == gid) { + dqrele(vp, ip->i_dquot[GRPQUOTA]); + ip->i_dquot[GRPQUOTA] = NODQUOT; } -out: - if (dp == ip) - irele(ip); - else - iput(ip); - iput(dp); -} - -/* - * Seek system call - */ -lseek() -{ - 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; + change = ip->i_blocks; + (void) chkdq(ip, -change, cred, CHOWN); + (void) chkiq(ip, -1, cred, CHOWN); + for (i = 0; i < MAXQUOTAS; i++) { + dqrele(vp, ip->i_dquot[i]); + ip->i_dquot[i] = NODQUOT; } - 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; - - default: - u.u_error = EINVAL; - return; +#endif + ip->i_uid = uid; + ip->i_gid = gid; +#ifdef QUOTA + if ((error = getinoquota(ip)) == 0) { + if (ouid == uid) { + dqrele(vp, ip->i_dquot[USRQUOTA]); + ip->i_dquot[USRQUOTA] = NODQUOT; + } + if (ogid == gid) { + dqrele(vp, ip->i_dquot[GRPQUOTA]); + ip->i_dquot[GRPQUOTA] = NODQUOT; + } + if ((error = chkdq(ip, change, cred, CHOWN)) == 0) { + if ((error = chkiq(ip, 1, cred, CHOWN)) == 0) + goto good; + else + (void) chkdq(ip, -change, cred, CHOWN|FORCE); + } + for (i = 0; i < MAXQUOTAS; i++) { + dqrele(vp, ip->i_dquot[i]); + ip->i_dquot[i] = NODQUOT; + } } - 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); + ip->i_uid = ouid; + ip->i_gid = ogid; + if (getinoquota(ip) == 0) { + if (ouid == uid) { + dqrele(vp, ip->i_dquot[USRQUOTA]); + ip->i_dquot[USRQUOTA] = NODQUOT; + } + if (ogid == gid) { + dqrele(vp, ip->i_dquot[GRPQUOTA]); + ip->i_dquot[GRPQUOTA] = NODQUOT; + } + (void) chkdq(ip, change, cred, FORCE|CHOWN); + (void) chkiq(ip, 1, cred, FORCE|CHOWN); + (void) getinoquota(ip); } - u.u_uid = svuid; - u.u_gid = svgid; + return (error); +good: + if (getinoquota(ip)) + panic("chown: lost quota"); +#endif /* QUOTA */ + if (ouid != uid || ogid != gid) + ip->i_flag |= ICHG; + if (ouid != uid && cred->cr_uid != 0) + ip->i_mode &= ~ISUID; + if (ogid != gid && cred->cr_uid != 0) + ip->i_mode &= ~ISGID; + return (0); } -/* - * Stat system call. This version follows links. - */ -stat() +/* ARGSUSED */ +int +ufs_ioctl(vp, com, data, fflag, cred, p) + struct vnode *vp; + int com; + caddr_t data; + int fflag; + struct ucred *cred; + struct proc *p; { - stat1(FOLLOW); + return (ENOTTY); } -/* - * Lstat system call. This version does not follow links. - */ -lstat() +/* ARGSUSED */ +int +ufs_select(vp, which, fflags, cred, p) + struct vnode *vp; + int which, fflags; + struct ucred *cred; + struct proc *p; { - 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)); + /* + * We should really check to see if I/O is possible. + */ + return (1); } /* - * Return target name of a symbolic link + * Mmap a file + * + * NB Currently unsupported. */ -readlink() +/* ARGSUSED */ +int +ufs_mmap(vp, fflags, cred, p) + struct vnode *vp; + int fflags; + struct ucred *cred; + struct proc *p; { - 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; - } - 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; + + return (EINVAL); } /* - * Change mode of a file given path name. + * Seek on a file + * + * Nothing to do, so just return. */ -chmod() +/* ARGSUSED */ +int +ufs_seek(vp, oldoff, newoff, cred) + struct vnode *vp; + off_t oldoff, newoff; + struct ucred *cred; { - 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); + + return (0); } /* - * Change mode of a file given a file descriptor. + * ufs remove + * Hard to avoid races here, especially + * in unlinking directories. */ -fchmod() +int +ufs_remove(dvp, vp, cnp) + struct vnode *dvp, *vp; + struct componentname *cnp; { - 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); + register struct inode *ip, *dp; + int error; + + ip = VTOI(vp); + dp = VTOI(dvp); + error = ufs_dirremove(dvp, cnp); + if (!error) { + ip->i_nlink--; + ip->i_flag |= ICHG; + } + if (dp == ip) + vrele(ITOV(ip)); + else + ufs_iput(ip); + ufs_iput(dp); + return (error); } /* - * Change the mode on a file. - * Inode must be locked before calling. + * link vnode call */ -chmod1(ip, mode) - register struct inode *ip; - register int mode; +int +ufs_link(vp, tdvp, cnp) + register struct vnode *vp; /* source vnode */ + struct vnode *tdvp; + struct componentname *cnp; { + register struct inode *ip; + int error; - ip->i_mode &= ~07777; - if (u.u_uid) { - mode &= ~ISVTX; - if (!groupmember(ip->i_gid)) - mode &= ~ISGID; +#ifdef DIANOSTIC + if ((cnp->cn_flags & HASBUF) == 0) + panic("ufs_link: no name"); +#endif + ip = VTOI(vp); + if ((unsigned short)ip->i_nlink >= LINK_MAX) { + free(cnp->cn_pnbuf, M_NAMEI); + return (EMLINK); } - ip->i_mode |= mode&07777; + if (tdvp != vp) + ILOCK(ip); + ip->i_nlink++; ip->i_flag |= ICHG; - if (ip->i_flag&ITEXT && (ip->i_mode&ISVTX)==0) - xrele(ip); + error = VOP_UPDATE(vp, &time, &time, 1); + if (!error) + error = ufs_direnter(ip, tdvp, cnp); + if (tdvp != vp) + IUNLOCK(ip); + FREE(cnp->cn_pnbuf, M_NAMEI); + vput(tdvp); + if (error) { + ip->i_nlink--; + ip->i_flag |= ICHG; + } + return (error); } -/* - * 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); -} /* - * Perform chown operation on inode ip; - * inode must be locked prior to call. + * relookup - lookup a path name component + * Used by lookup to re-aquire things. */ -chown1(ip, uid, gid) - register struct inode *ip; - int uid, gid; +int +relookup(dvp, vpp, cnp) + struct vnode *dvp, **vpp; + struct componentname *cnp; { -#ifdef QUOTA - register long change; + register char *cp; /* pointer into pathname argument */ + register struct vnode *dp = 0; /* the directory we are searching */ + struct vnode *tdp; /* saved dp */ + struct mount *mp; /* mount table entry */ + int docache; /* == 0 do not cache last component */ + int wantparent; /* 1 => wantparent or lockparent flag */ + int rdonly; /* lookup read-only flag bit */ + int error = 0; + int newhash; + + /* + * Setup: break out flag bits into variables. + */ + wantparent = cnp->cn_flags & (LOCKPARENT|WANTPARENT); + docache = (cnp->cn_flags & NOCACHE) ^ NOCACHE; + if (cnp->cn_nameiop == DELETE || + (wantparent && cnp->cn_nameiop != CREATE)) + docache = 0; + rdonly = cnp->cn_flags & RDONLY; + cnp->cn_flags &= ~ISSYMLINK; + dp = dvp; + VOP_LOCK(dp); + +/* dirloop: */ + /* + * Search a new directory. + * + * The cn_hash value is for use by vfs_cache. + * The last component of the filename is left accessible via + * cnp->cn_nameptr for callers that need the name. Callers needing + * the name set the SAVENAME flag. When done, they assume + * responsibility for freeing the pathname buffer. + */ +#ifdef NAMEI_DIAGNOSTIC + newhash = 0; + for (cp = cnp->cn_nameptr; *cp != 0 && *cp != '/'; cp++) + newhash += (unsigned char)*cp; + if (newhash != cnp->cn_hash) + panic("relookup: bad hash"); + if (cnp->cn_namelen != cp - cnp->cn_nameptr) + panic ("relookup: bad len"); + { char c = *cp; + *cp = '\0'; + printf("{%s}: ", cnp->cn_nameptr); + *cp = c; } #endif - if (uid == -1) - uid = ip->i_uid; - if (gid == -1) - gid = ip->i_gid; -#ifdef QUOTA - if (ip->i_uid == uid) /* this just speeds things a little */ - change = 0; - else - change = ip->i_blocks; - (void) chkdq(ip, -change, 1); - (void) chkiq(ip->i_dev, ip, ip->i_uid, 1); - dqrele(ip->i_dquot); + /* + * Check for degenerate name (e.g. / or "") + * which is a way of talking about a directory, + * e.g. like "/." or ".". + */ + if (cnp->cn_nameptr[0] == '\0') { + if (cnp->cn_nameiop != LOOKUP || wantparent) { + error = EISDIR; + goto bad; + } + if (dp->v_type != VDIR) { + error = ENOTDIR; + goto bad; + } + if (!(cnp->cn_flags & LOCKLEAF)) + VOP_UNLOCK(dp); + *vpp = dp; + if (cnp->cn_flags & SAVESTART) + panic("lookup: SAVESTART"); + return (0); + } + + if (cnp->cn_flags & ISDOTDOT) + panic ("relookup: lookup on dot-dot"); + + /* + * We now have a segment name to search for, and a directory to search. + */ + if (error = VOP_LOOKUP(dp, vpp, cnp)) { +#ifdef DIAGNOSTIC + if (*vpp != NULL) + panic("leaf should be empty"); #endif - 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); - return (u.u_error); /* should == 0 ALWAYS !! */ -#else - return (0); +#ifdef NAMEI_DIAGNOSTIC + printf("not found\n"); #endif -} - -utimes() -{ - register struct a { - char *fname; - struct timeval *tptr; - } *uap = (struct a *)u.u_ap; - register struct inode *ip; - struct timeval tv[2]; - - 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); + if (cnp->cn_nameiop == LOOKUP || cnp->cn_nameiop == DELETE || + error != ENOENT || *cp != 0) + goto bad; + /* + * If creating and at end of pathname, then can consider + * allowing file to be created. + */ + if (rdonly || (dvp->v_mount->mnt_flag & MNT_RDONLY)) { + error = EROFS; + goto bad; + } + /* + * We return with ni_vp NULL to indicate that the entry + * doesn't currently exist, leaving a pointer to the + * (possibly locked) directory inode in ndp->ni_dvp. + */ + if (cnp->cn_flags & SAVESTART) { + /* + * startdir == dvp, always + */ + VREF(dvp); + } + return (0); } - iput(ip); -} +#ifdef NAMEI_DIAGNOSTIC + printf("found\n"); +#endif -/* - * Flush any pending I/O. - */ -sync() -{ + dp = *vpp; +#ifdef DIAGNOSTIC + /* + * Check for symbolic link + */ + if (dp->v_type == VLNK) { + panic ("relookup: symlink found.\n"); + }; - update(); -} + /* + * Check to see if the vnode has been mounted on; + * if so find the root of the mounted file system. + */ +#endif -/* - * Truncate a file given its path name. - */ -truncate() -{ - struct a { - char *fname; - u_long length; - } *uap = (struct a *)u.u_ap; - struct inode *ip; - 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 (access(ip, IWRITE)) - goto bad; - if ((ip->i_mode&IFMT) == IFDIR) { - u.u_error = EISDIR; - goto bad; - } - itrunc(ip, uap->length); -bad: - iput(ip); -} -/* - * Truncate a file given a file descriptor. - */ -ftruncate() -{ - 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; +nextname: + /* + * Check for read-only file systems. + */ + if (cnp->cn_nameiop == DELETE || cnp->cn_nameiop == RENAME) { + /* + * Disallow directory write attempts on read-only + * file systems. + */ + if (rdonly || (dp->v_mount->mnt_flag & MNT_RDONLY) || + (wantparent && + (dvp->v_mount->mnt_flag & MNT_RDONLY))) { + error = EROFS; + goto bad2; + } } - ip = (struct inode *)fp->f_data; - ILOCK(ip); - itrunc(ip, uap->length); - IUNLOCK(ip); -} + if (cnp->cn_flags & SAVESTART) { + /* ASSERT(dvp==ndp->ni_startdir) */ + VREF(dvp); + } + + if (!wantparent) + vrele(dvp); + if ((cnp->cn_flags & LOCKLEAF) == 0) + VOP_UNLOCK(dp); + return (0); -/* - * Synch an open file. - */ -fsync() -{ - struct a { - int fd; - } *uap = (struct a *)u.u_ap; - struct inode *ip; - struct file *fp; - - fp = getinode(uap->fd); - if (fp == NULL) - return; - ip = (struct inode *)fp->f_data; - ILOCK(ip); - syncip(ip); - IUNLOCK(ip); +bad2: + if ((cnp->cn_flags & LOCKPARENT) && (cnp->cn_flags & ISLASTCN)) + VOP_UNLOCK(dvp); + vrele(dvp); +bad: + vput(dp); + *vpp = NULL; + return (error); } + /* * Rename system call. * rename("foo", "bar"); @@ -779,7 +787,7 @@ 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 + * 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, @@ -788,52 +796,65 @@ fsync() * 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() +int +ufs_rename(fdvp, fvp, fcnp, + tdvp, tvp, tcnp) + struct vnode *fdvp, *fvp; + struct componentname *fcnp; + struct vnode *tdvp, *tvp; + struct componentname *tcnp; { - struct a { - char *from; - char *to; - } *uap = (struct a *)u.u_ap; register struct inode *ip, *xp, *dp; struct dirtemplate dirbuf; int doingdirectory = 0, oldparent = 0, newparent = 0; - register struct nameidata *ndp = &u.u_nd; int error = 0; + int fdvpneedsrele = 1, tdvpneedsrele = 1; - 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; +#ifdef DIANOSTIC + if ((tcnp->cn_flags & HASBUF) == 0 || + (fcnp->cn_flags & HASBUF) == 0) + panic("ufs_rename: no name"); +#endif + dp = VTOI(fdvp); + ip = VTOI(fvp); + /* + * Check if just deleting a link name. + */ + if (fvp == tvp) { + VOP_ABORTOP(tdvp, tcnp); + vput(tdvp); + vput(tvp); + vrele(fdvp); + if ((ip->i_mode&IFMT) == IFDIR) { + VOP_ABORTOP(fdvp, fcnp); + vrele(fvp); + return (EINVAL); + } + doingdirectory = 0; + goto unlinkit; + } + ILOCK(ip); if ((ip->i_mode&IFMT) == IFDIR) { - register struct direct *d; - - 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) || (ip->i_flag & IRENAME)) { - iput(dp); - if (dp == ip) - irele(ip); - else - iput(ip); - u.u_error = EINVAL; - return; + if ((fcnp->cn_namelen == 1 && fcnp->cn_nameptr[0] == '.') || + dp == ip || (fcnp->cn_flags&ISDOTDOT) || (ip->i_flag & IRENAME)) { + VOP_ABORTOP(tdvp, tcnp); + vput(tdvp); + if (tvp) + vput(tvp); + VOP_ABORTOP(fdvp, fcnp); + vrele(fdvp); + vput(fvp); + return (EINVAL); } ip->i_flag |= IRENAME; oldparent = dp->i_number; doingdirectory++; } - iput(dp); + vrele(fdvp); /* * 1) Bump link count while we're moving stuff @@ -843,21 +864,17 @@ rename() */ ip->i_nlink++; ip->i_flag |= ICHG; - iupdat(ip, &time, &time, 1); + error = VOP_UPDATE(fvp, &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(tdvp); + xp = NULL; + if (tvp) + xp = VTOI(tvp); /* * If ".." must be changed (ie the directory gets a new * parent) then the source directory must not be in the @@ -871,21 +888,24 @@ rename() if (oldparent != dp->i_number) newparent = dp->i_number; if (doingdirectory && newparent) { - if (access(ip, IWRITE)) + VOP_LOCK(fvp); + error = ufs_access(fvp, VWRITE, tcnp->cn_cred, tcnp->cn_proc); + VOP_UNLOCK(fvp); + if (error) goto bad; - do { - dp = ndp->ni_pdir; - if (xp != NULL) - iput(xp); - u.u_error = checkpath(ip, dp); - if (u.u_error) - goto out; - xp = namei(ndp); - if (u.u_error) { - error = u.u_error; - goto out; - } - } while (dp != ndp->ni_pdir); + if (xp != NULL) + ufs_iput(xp); + if (error = ufs_checkpath(ip, dp, tcnp->cn_cred)) + goto out; + if ((tcnp->cn_flags & SAVESTART) == 0) + panic("ufs_rename: lost to startdir"); + p->p_spare[1]--; + if (error = relookup(tdvp, &tvp, tcnp)) + goto out; + dp = VTOI(tdvp); + xp = NULL; + if (tvp) + xp = VTOI(tvp); } /* * 2) If target doesn't exist, link the target @@ -895,42 +915,60 @@ 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 new directory. * When source and destination have the same * parent we don't fool with the link count. */ if (doingdirectory && newparent) { + if ((unsigned short)dp->i_nlink >= LINK_MAX) { + error = EMLINK; + goto bad; + } dp->i_nlink++; dp->i_flag |= ICHG; - iupdat(dp, &time, &time, 1); + if (error = VOP_UPDATE(ITOV(dp), &time, &time, 1)) + goto bad; } - error = direnter(ip, ndp); - if (error) - goto out; - } else { - if (xp->i_dev != dp->i_dev || xp->i_dev != ip->i_dev) { - error = EXDEV; + if (error = ufs_direnter(ip, tdvp, tcnp)) { + if (doingdirectory && newparent) { + dp->i_nlink--; + dp->i_flag |= ICHG; + (void)VOP_UPDATE(ITOV(dp), &time, &time, 1); + } goto bad; } + ufs_iput(dp); + } else { + 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) + 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) && tcnp->cn_cred->cr_uid != 0 && + tcnp->cn_cred->cr_uid != dp->i_uid && + xp->i_uid != tcnp->cn_cred->cr_uid) { + error = EPERM; goto bad; + } /* - * Target must be empty if a directory - * and have no links to it. - * Also, insure source and target are - * compatible (both directories, or both - * not directories). + * Target must be empty if a directory and have no links + * to it. Also, ensure source and target are compatible + * (both directories, or both not directories). */ if ((xp->i_mode&IFMT) == IFDIR) { - if (!dirempty(xp, dp->i_number) || xp->i_nlink > 2) { + if (!ufs_dirempty(xp, dp->i_number, tcnp->cn_cred) || + xp->i_nlink > 2) { error = ENOTEMPTY; goto bad; } @@ -938,16 +976,24 @@ rename() error = ENOTDIR; goto bad; } - cacheinval(dp); + 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 = ufs_dirrewrite(dp, ip, tcnp)) + goto bad; + /* + * If the target directory is in the same + * directory as the source directory, + * decrement the link count on the parent + * of the target directory. + */ + if (doingdirectory && !newparent) { + dp->i_nlink--; + dp->i_flag |= ICHG; } + ufs_iput(dp); /* * Adjust the link count of the target to * reflect the dirrewrite above. If this is @@ -962,34 +1008,48 @@ rename() if (doingdirectory) { if (--xp->i_nlink != 0) panic("rename: linked directory"); - itrunc(xp, (u_long)0); + error = VOP_TRUNCATE(ITOV(xp), (u_long)0, IO_SYNC); } xp->i_flag |= ICHG; - iput(xp); + ufs_iput(xp); xp = NULL; } /* * 3) Unlink the source. */ - ndp->ni_nameiop = DELETE | LOCKPARENT; - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = uap->from; - xp = namei(ndp); - dp = ndp->ni_pdir; +unlinkit: + fcnp->cn_flags &= ~MODMASK; + fcnp->cn_flags |= LOCKPARENT | LOCKLEAF; + if ((fcnp->cn_flags & SAVESTART) == 0) + panic("ufs_rename: lost from startdir"); + p->p_spare[1]--; + (void) relookup(fdvp, &fvp, fcnp); /* NEEDSWORK: startdir stuff */ + if (fvp != NULL) { + xp = VTOI(fvp); + dp = VTOI(fdvp); + } else { + /* + * From name has disappeared. + */ + if (doingdirectory) + panic("rename: lost dir entry"); + vrele(ITOV(ip)); + return (0); + } /* - * Insure that the directory entry still exists and has not + * 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 insures that it cannot be moved by another + * The IRENAME flag ensures that it cannot be moved by another * rename. */ - if (dp == NULL || xp != ip) { + if (xp != ip) { if (doingdirectory) - panic("rename: lost entry"); + panic("rename: lost dir entry"); } else { /* * If the source is a directory with a @@ -1000,114 +1060,58 @@ rename() if (doingdirectory && newparent) { dp->i_nlink--; dp->i_flag |= ICHG; - error = rdwri(UIO_READ, xp, (caddr_t)&dirbuf, - sizeof (struct dirtemplate), (off_t)0, 1, - (int *)0); + error = vn_rdwr(UIO_READ, ITOV(xp), (caddr_t)&dirbuf, + sizeof (struct dirtemplate), (off_t)0, + UIO_SYSSPACE, IO_NODELOCKED, + tcnp->cn_cred, (int *)0, (struct proc *)0); if (error == 0) { if (dirbuf.dotdot_namlen != 2 || dirbuf.dotdot_name[0] != '.' || dirbuf.dotdot_name[1] != '.') { - printf("rename: mangled dir\n"); + ufs_dirbad(xp, 12, + "rename: mangled dir"); } else { dirbuf.dotdot_ino = newparent; - (void) rdwri(UIO_WRITE, xp, + (void) vn_rdwr(UIO_WRITE, ITOV(xp), (caddr_t)&dirbuf, sizeof (struct dirtemplate), - (off_t)0, 1, (int *)0); - cacheinval(dp); + (off_t)0, UIO_SYSSPACE, + IO_NODELOCKED|IO_SYNC, + tcnp->cn_cred, (int *)0, + (struct proc *)0); + cache_purge(ITOV(dp)); } } } - if (dirremove(ndp)) { + error = ufs_dirremove(fdvp, fcnp); + if (!error) { xp->i_nlink--; xp->i_flag |= ICHG; } xp->i_flag &= ~IRENAME; - if (error == 0) /* XXX conservative */ - error = u.u_error; } if (dp) - iput(dp); + vput(ITOV(dp)); if (xp) - iput(xp); - irele(ip); - if (error) - u.u_error = error; - return; + 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); - 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); } /* * A virgin directory (no blushing please). */ -struct dirtemplate mastertemplate = { +static struct dirtemplate mastertemplate = { 0, 12, 1, ".", 0, DIRBLKSIZ - 12, 2, ".." }; @@ -1115,55 +1119,59 @@ struct dirtemplate mastertemplate = { /* * Mkdir system call */ -mkdir() +int +ufs_mkdir(dvp, vpp, cnp, vap) + struct vnode *dvp; + struct vnode **vpp; + struct componentname *cnp; + struct vattr *vap; { - struct a { - char *name; - int dmode; - } *uap = (struct a *)u.u_ap; register struct inode *ip, *dp; + struct vnode *tvp; struct dirtemplate dirtemplate; - register struct nameidata *ndp = &u.u_nd; - - 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; + int error; + int dmode; + +#ifdef DIANOSTIC + if ((cnp->cn_flags & HASBUF) == 0) + panic("ufs_mkdir: no name"); +#endif + dp = VTOI(dvp); + if ((unsigned short)dp->i_nlink >= LINK_MAX) { + free(cnp->cn_pnbuf, M_NAMEI); + ufs_iput(dp); + return (EMLINK); } - dp = ndp->ni_pdir; - uap->dmode &= 0777; - uap->dmode |= IFDIR; + dmode = vap->va_mode&0777; + dmode |= IFDIR; /* - * Must simulate part of maknode here - * in order to acquire the inode, but - * not have it entered in the parent - * directory. The entry is made later - * after writing "." and ".." entries out. + * Must simulate part of maknode here to acquire the inode, but + * not have it entered in the parent directory. The entry is made + * later after writing "." and ".." entries. */ - ip = ialloc(dp, dirpref(dp->i_fs), uap->dmode); - if (ip == NULL) { - iput(dp); - return; + if (error = VOP_VALLOC(dvp, dmode, cnp->cn_cred, &tvp)) { + free(cnp->cn_pnbuf, M_NAMEI); + ufs_iput(dp); + return (error); } + ip = VTOI(tvp); + ip->i_uid = cnp->cn_cred->cr_uid; + ip->i_gid = dp->i_gid; #ifdef QUOTA - if (ip->i_dquot != NODQUOT) - panic("mkdir: dquot"); + if ((error = getinoquota(ip)) || + (error = chkiq(ip, 1, cnp->cn_cred, 0))) { + free(cnp->cn_pnbuf, M_NAMEI); + VOP_VFREE(tvp, ip->i_number, dmode); + ufs_iput(ip); + ufs_iput(dp); + return (error); + } #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_gid = dp->i_gid; -#ifdef QUOTA - ip->i_dquot = inoquota(ip); -#endif - iupdat(ip, &time, &time, 1); + error = VOP_UPDATE(ITOV(ip), &time, &time, 1); /* * Bump link count in parent directory @@ -1173,91 +1181,70 @@ mkdir() */ dp->i_nlink++; dp->i_flag |= ICHG; - iupdat(dp, &time, &time, 1); + if (error = VOP_UPDATE(ITOV(dp), &time, &time, 1)) + goto bad; - /* - * Initialize directory with "." - * and ".." from static template. - */ + /* Initialize directory with "." and ".." from static template. */ 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, cnp->cn_cred, (int *)0, (struct proc *)0); + if (error) { dp->i_nlink--; dp->i_flag |= ICHG; goto bad; } - /* - * Directory all set up, now - * install the entry for it in - * the parent directory. - */ - u.u_error = direnter(ip, ndp); - dp = NULL; - if (u.u_error) { - ndp->ni_nameiop = LOOKUP | NOCACHE; - ndp->ni_segflg = UIO_USERSPACE; - ndp->ni_dirp = uap->name; - dp = namei(ndp); - if (dp) { - dp->i_nlink--; - dp->i_flag |= ICHG; - } + if (DIRBLKSIZ > VFSTOUFS(dvp->v_mount)->um_mountp->mnt_stat.f_bsize) + panic("ufs_mkdir: blksize"); /* XXX should grow with balloc() */ + else { + ip->i_size = DIRBLKSIZ; + ip->i_flag |= ICHG; + } + + /* Directory set up, now install it's entry in the parent directory. */ + if (error = ufs_direnter(ip, dvp, cnp)) { + dp->i_nlink--; + dp->i_flag |= ICHG; } bad: /* - * No need to do an explicit itrunc here, - * irele will do this for us because we set - * the link count to 0. + * No need to do an explicit VOP_TRUNCATE here, 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; - } - if (dp) - iput(dp); - iput(ip); + ufs_iput(ip); + } else + *vpp = ITOV(ip); + FREE(cnp->cn_pnbuf, M_NAMEI); + ufs_iput(dp); + return (error); } /* * Rmdir system call. */ -rmdir() +int +ufs_rmdir(dvp, vp, cnp) + struct vnode *dvp; + struct vnode *vp; + struct componentname *cnp; { - struct a { - char *name; - } *uap = (struct a *)u.u_ap; register struct inode *ip, *dp; - register struct nameidata *ndp = &u.u_nd; - - 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; + int error; + + ip = VTOI(vp); + dp = VTOI(dvp); /* * No rmdir "." please. */ if (dp == ip) { - irele(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; + vrele(dvp); + ufs_iput(ip); + return (EINVAL); } /* * Verify the directory is empty (and valid). @@ -1266,8 +1253,10 @@ rmdir() * the current directory and thus be * non-empty.) */ - if (ip->i_nlink != 2 || !dirempty(ip, dp->i_number)) { - u.u_error = ENOTEMPTY; + error = 0; + if (ip->i_nlink != 2 || + !ufs_dirempty(ip, dp->i_number, cnp->cn_cred)) { + error = ENOTEMPTY; goto out; } /* @@ -1275,13 +1264,13 @@ rmdir() * inode. If we crash in between, the directory * will be reattached to lost+found, */ - if (dirremove(ndp) == 0) + if (error = ufs_dirremove(dvp, cnp)) goto out; dp->i_nlink--; dp->i_flag |= ICHG; - cacheinval(dp); - iput(dp); - dp = NULL; + cache_purge(dvp); + ufs_iput(dp); + dvp = NULL; /* * Truncate inode. The only stuff left * in the directory is "." and "..". The @@ -1294,40 +1283,544 @@ rmdir() * worry about them later. */ ip->i_nlink -= 2; - itrunc(ip, (u_long)0); - cacheinval(ip); + error = VOP_TRUNCATE(vp, (u_long)0, IO_SYNC); + cache_purge(ITOV(ip)); out: - if (dp) - iput(dp); - iput(ip); + if (dvp) + ufs_iput(dp); + ufs_iput(ip); + return (error); +} + +/* + * symlink -- make a symbolic link + */ +int +ufs_symlink(dvp, vpp, cnp, vap, target) + struct vnode *dvp; + struct vnode **vpp; + struct componentname *cnp; + struct vattr *vap; + char *target; +{ + int error; + + if (error = ufs_makeinode(IFLNK | vap->va_mode, dvp, vpp, cnp)) + return (error); + error = vn_rdwr(UIO_WRITE, *vpp, target, strlen(target), (off_t)0, + UIO_SYSSPACE, IO_NODELOCKED, cnp->cn_cred, (int *)0, + (struct proc *)0); + vput(*vpp); + return (error); +} + +/* + * Vnode op for read and write + */ +int +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 = VOP_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 + */ +int +ufs_readlink(vp, uiop, cred) + struct vnode *vp; + struct uio *uiop; + struct ucred *cred; +{ + + return (VOP_READ(vp, uiop, 0, cred)); +} + +/* + * Ufs abort op, called after namei() when a CREATE/DELETE isn't actually + * done. If a buffer has been saved in anticipation of a CREATE, delete it. + */ +/* ARGSUSED */ +int +ufs_abortop(dvp, cnp) + struct vnode *dvp; + struct componentname *cnp; +{ + if ((cnp->cn_flags & (HASBUF | SAVESTART)) == HASBUF) + FREE(cnp->cn_pnbuf, M_NAMEI); + return (0); +} + +/* + * Lock an inode. + */ +int +ufs_lock(vp) + struct vnode *vp; +{ + register struct inode *ip = VTOI(vp); + + ILOCK(ip); + return (0); +} + +/* + * Unlock an inode. + */ +int +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. + */ +int +ufs_islocked(vp) + struct vnode *vp; +{ + + if (VTOI(vp)->i_flag & ILOCKED) + return (1); + return (0); +} + +/* + * Calculate the logical to physical mapping if not done already, + * then call the device strategy routine. + */ +int checkoverlap = 0; + +int +ufs_strategy(bp) + register struct buf *bp; +{ + register struct inode *ip; + struct vnode *vp; + int error; + + ip = VTOI(bp->b_vp); + 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 = + VOP_BMAP(bp->b_vp, bp->b_lblkno, NULL, &bp->b_blkno)) + return (error); + if ((long)bp->b_blkno == -1) + clrbuf(bp); + } + if ((long)bp->b_blkno == -1) { + biodone(bp); + return (0); + } +#ifdef DIAGNOSTIC + if (checkoverlap && bp->b_vp->v_mount->mnt_stat.f_type == MOUNT_UFS) + ffs_checkoverlap(bp, ip); +#endif + + vp = ip->i_devvp; + bp->b_dev = vp->v_rdev; + (vp->v_op->vop_strategy)(bp); + return (0); +} + +/* + * Print out the contents of an inode. + */ +int +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_lockholder == 0) + return (0); + printf("\towner pid %d", ip->i_lockholder); + if (ip->i_lockwaiter) + printf(" waiting pid %d", ip->i_lockwaiter); + printf("\n"); + return (0); +} + +/* + * Read wrapper for special devices. + */ +int +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. + */ +int +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. + */ +int +ufsspec_close(vp, fflag, cred, p) + struct vnode *vp; + int fflag; + struct ucred *cred; + struct proc *p; +{ + 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, p)); +} + +#ifdef FIFO +/* + * Read wrapper for fifo's + */ +int +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. + */ +int +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, p) + struct vnode *vp; + int fflag; + struct ucred *cred; + struct proc *p; +{ + 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, p)); } +#endif /* FIFO */ -struct file * -getinode(fdes) - int fdes; +/* + * Advisory record locking support + */ +int +ufs_advlock(vp, id, op, fl, flags) + struct vnode *vp; + caddr_t id; + int op; + register struct flock *fl; + int flags; { - struct file *fp; + register struct inode *ip = VTOI(vp); + register struct lockf *lock; + off_t start, end; + int error; + + /* + * Avoid the common case of unlocking when inode has no locks. + */ + if (ip->i_lockf == (struct lockf *)0) { + if (op != F_SETLK) { + fl->l_type = F_UNLCK; + return (0); + } + } + /* + * Convert the flock structure into a start and end. + */ + switch (fl->l_whence) { + + case SEEK_SET: + case SEEK_CUR: + /* + * Caller is responsible for adding any necessary offset + * when SEEK_CUR is used. + */ + start = fl->l_start; + break; + + case SEEK_END: + start = ip->i_size + fl->l_start; + break; - if ((unsigned)fdes >= NOFILE || (fp = u.u_ofile[fdes]) == NULL) { - u.u_error = EBADF; - return ((struct file *)0); + default: + return (EINVAL); + } + if (start < 0) + return (EINVAL); + if (fl->l_len == 0) + end = -1; + else + end = start + fl->l_len - 1; + /* + * Create the lockf structure + */ + MALLOC(lock, struct lockf *, sizeof *lock, M_LOCKF, M_WAITOK); + lock->lf_start = start; + lock->lf_end = end; + lock->lf_id = id; + lock->lf_inode = ip; + lock->lf_type = fl->l_type; + lock->lf_next = (struct lockf *)0; + lock->lf_block = (struct lockf *)0; + lock->lf_flags = flags; + /* + * Do the requested operation. + */ + switch(op) { + case F_SETLK: + return (lf_setlock(lock)); + + case F_UNLCK: + error = lf_clearlock(lock); + FREE(lock, M_LOCKF); + return (error); + + case F_GETLK: + error = lf_getlock(lock, fl); + FREE(lock, M_LOCKF); + return (error); + + default: + free(lock, M_LOCKF); + return (EINVAL); } - if (fp->f_type != DTYPE_INODE) { - u.u_error = EINVAL; - return ((struct file *)0); + /* NOTREACHED */ +} + +/* + * Initialize the vnode associated with a new inode, handle aliased + * vnodes. + */ +int +ufs_vinit(mntp, specops, fifoops, vpp) + struct mount *mntp; + struct vnodeops *specops, *fifoops; + struct vnode **vpp; +{ + struct inode *ip, *nip; + struct vnode *vp, *nvp; + extern struct vnodeops spec_vnodeops; + + vp = *vpp; + ip = VTOI(vp); + switch(vp->v_type = IFTOVT(ip->i_mode)) { + case VCHR: + case VBLK: + vp->v_op = specops; + if (nvp = checkalias(vp, ip->i_rdev, mntp)) { + /* + * Discard unneeded vnode, but save its inode. + */ + remque(ip); + IUNLOCK(ip); + nvp->v_data = vp->v_data; + vp->v_data = NULL; + vp->v_op = &spec_vnodeops; + vrele(vp); + vgone(vp); + /* + * Reinitialize aliased inode. + */ + vp = nvp; + ip->i_vnode = vp; + ufs_ihashins(ip); + } + break; + case VFIFO: +#ifdef FIFO + vp->v_op = fifoops; + break; +#else + return (EOPNOTSUPP); +#endif } - return (fp); + if (ip->i_number == ROOTINO) + vp->v_flag |= VROOT; + /* + * Initialize modrev times + */ + SETHIGH(ip->i_modrev, mono_time.tv_sec); + SETLOW(ip->i_modrev, mono_time.tv_usec * 4294); + *vpp = vp; + return (0); } /* - * mode mask for creation of files + * Allocate a new inode. */ -umask() +int +ufs_makeinode(mode, dvp, vpp, cnp) + int mode; + struct vnode *dvp; + struct vnode **vpp; + struct componentname *cnp; { - register struct a { - int mask; - } *uap = (struct a *)u.u_ap; + register struct inode *ip, *pdir; + struct vnode *tvp; + int error; + + pdir = VTOI(dvp); +#ifdef DIANOSTIC + if ((cnp->cn_flags & HASBUF) == 0) + panic("ufs_makeinode: no name"); +#endif + *vpp = NULL; + if ((mode & IFMT) == 0) + mode |= IFREG; + + if (error = VOP_VALLOC(dvp, mode, cnp->cn_cred, &tvp)) { + free(cnp->cn_pnbuf, M_NAMEI); + ufs_iput(pdir); + return (error); + } + ip = VTOI(tvp); + ip->i_uid = cnp->cn_cred->cr_uid; + ip->i_gid = pdir->i_gid; +#ifdef QUOTA + if ((error = getinoquota(ip)) || + (error = chkiq(ip, 1, cnp->cn_cred, 0))) { + free(cnp->cn_pnbuf, M_NAMEI); + VOP_VFREE(tvp, ip->i_number, mode); + ufs_iput(ip); + ufs_iput(pdir); + return (error); + } +#endif + ip->i_flag |= IACC|IUPD|ICHG; + ip->i_mode = mode; + tvp->v_type = IFTOVT(mode); /* Rest init'd in iget() */ + ip->i_nlink = 1; + if ((ip->i_mode & ISGID) && !groupmember(ip->i_gid, cnp->cn_cred) && + suser(cnp->cn_cred, NULL)) + ip->i_mode &= ~ISGID; + + /* + * Make sure inode goes to disk before directory entry. + */ + if (error = VOP_UPDATE(tvp, &time, &time, 1)) + goto bad; + if (error = ufs_direnter(ip, dvp, cnp)) + goto bad; + if ((cnp->cn_flags & SAVESTART) == 0) + FREE(cnp->cn_pnbuf, M_NAMEI); + ufs_iput(pdir); + *vpp = tvp; + return (0); + +bad: + /* + * Write error occurred trying to update the inode + * or the directory so must deallocate the inode. + */ + free(cnp->cn_pnbuf, M_NAMEI); + ufs_iput(pdir); + ip->i_nlink = 0; + ip->i_flag |= ICHG; + ufs_iput(ip); + return (error); +} - u.u_r.r_val1 = u.u_cmask; - u.u_cmask = uap->mask & 07777; + +#if defined(JOHNH) && 0 +/* + * A hack to get the kernel to compile. + */ +int +hang_addrlist() +{ + return 0; } +int +free_addrlist() +{ + return 0; +} +#endif +