X-Git-Url: https://git.subgeniuskitty.com/unix-history/.git/blobdiff_plain/86cdabf6f2d7da31eac857bc4799229c6254a582..c43604c7566af0a35bc20cea474e88e0ad5fde69:/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 173b89a6cc..a1818b5d2e 100644 --- a/usr/src/sys/ufs/ffs/ufs_vnops.c +++ b/usr/src/sys/ufs/ffs/ufs_vnops.c @@ -2,171 +2,76 @@ * 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. + * %sccs.include.redist.c% * - * @(#)ufs_vnops.c 7.21 (Berkeley) %G% + * @(#)ufs_vnops.c 7.109 (Berkeley) %G% */ -#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" - -/* - * Global vfs data structures for ufs - */ - -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(); - -struct vnodeops ufs_vnodeops = { - 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, -}; - -int spec_lookup(), - spec_open(), - spec_read(), - spec_write(), - spec_strategy(), - spec_ioctl(), - spec_select(), - spec_close(), - spec_badop(), - spec_nullop(); - -struct vnodeops spec_inodeops = { - spec_lookup, /* lookup */ - spec_badop, /* create */ - spec_badop, /* mknod */ - spec_open, /* open */ - spec_close, /* close */ - ufs_access, /* access */ - ufs_getattr, /* getattr */ - ufs_setattr, /* setattr */ - spec_read, /* read */ - spec_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_badop, /* bmap */ - spec_strategy, /* strategy */ -}; - -enum vtype iftovt_tab[8] = { - VNON, VCHR, VDIR, VBLK, VREG, VLNK, VSOCK, VBAD, -}; -int vttoif_tab[8] = { - 0, IFREG, IFDIR, IFBLK, IFCHR, IFLNK, IFSOCK, IFMT, +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include + +#include + +#include +#include +#include +#include +#include +#include + +static int ufs_chmod __P((struct vnode *, int, struct ucred *, struct proc *)); +static int ufs_chown + __P((struct vnode *, uid_t, gid_t, struct ucred *, struct proc *)); + +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; \ +} /* * Create a regular file */ -ufs_create(ndp, vap) - struct nameidata *ndp; - struct vattr *vap; +int +ufs_create(ap) + struct vop_create_args /* { + struct vnode *a_dvp; + struct vnode **a_vpp; + struct componentname *a_cnp; + struct vattr *a_vap; + } */ *ap; { - struct inode *ip; int error; - if (error = maknode(MAKEIMODE(vap->va_type, vap->va_mode), ndp, &ip)) + if (error = + ufs_makeinode(MAKEIMODE(ap->a_vap->va_type, ap->a_vap->va_mode), + ap->a_dvp, ap->a_vpp, ap->a_cnp)) return (error); - ndp->ni_vp = ITOV(ip); return (0); } @@ -174,34 +79,42 @@ ufs_create(ndp, vap) * Mknod vnode call */ /* ARGSUSED */ -ufs_mknod(ndp, vap, cred) - struct nameidata *ndp; - struct ucred *cred; - struct vattr *vap; +int +ufs_mknod(ap) + struct vop_mknod_args /* { + struct vnode *a_dvp; + struct vnode **a_vpp; + struct componentname *a_cnp; + struct vattr *a_vap; + } */ *ap; { - register struct vnode *vp; - struct inode *ip; + register struct vattr *vap = ap->a_vap; + register struct vnode **vpp = ap->a_vpp; + register struct inode *ip; int error; - if (error = maknode(MAKEIMODE(vap->va_type, vap->va_mode), ndp, &ip)) + if (error = + ufs_makeinode(MAKEIMODE(vap->va_type, vap->va_mode), + ap->a_dvp, vpp, ap->a_cnp)) return (error); - vp = ITOV(ip); - if (vap->va_rdev) { + 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. */ - vp->v_rdev = ip->i_rdev = vap->va_rdev; - ip->i_flag |= IACC|IUPD|ICHG; + 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. */ - iput(ip); - vp->v_type = VNON; - vgone(vp); + vput(*vpp); + (*vpp)->v_type = VNON; + vgone(*vpp); + *vpp = 0; return (0); } @@ -211,10 +124,14 @@ ufs_mknod(ndp, vap, cred) * Nothing to do. */ /* ARGSUSED */ -ufs_open(vp, mode, cred) - struct vnode *vp; - int mode; - struct ucred *cred; +int +ufs_open(ap) + struct vop_open_args /* { + struct vnode *a_vp; + int a_mode; + struct ucred *a_cred; + struct proc *a_p; + } */ *ap; { return (0); @@ -226,34 +143,97 @@ ufs_open(vp, mode, cred) * Update the times on the inode. */ /* ARGSUSED */ -ufs_close(vp, fflag, cred) - struct vnode *vp; - int fflag; - struct ucred *cred; +int +ufs_close(ap) + struct vop_close_args /* { + struct vnode *a_vp; + int a_fflag; + struct ucred *a_cred; + struct proc *a_p; + } */ *ap; { + register struct vnode *vp = ap->a_vp; register struct inode *ip = VTOI(vp); - if (vp->v_count > 1 && !(ip->i_flag & ILOCKED)) + if (vp->v_usecount > 1 && !(ip->i_flag & ILOCKED)) ITIMES(ip, &time, &time); return (0); } -ufs_access(vp, mode, cred) - struct vnode *vp; - int mode; - struct ucred *cred; +/* + * 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. + */ +int +ufs_access(ap) + struct vop_access_args /* { + struct vnode *a_vp; + int a_mode; + struct ucred *a_cred; + struct proc *a_p; + } */ *ap; { - - return (iaccess(VTOI(vp), mode, cred)); + register struct vnode *vp = ap->a_vp; + register struct inode *ip = VTOI(vp); + register struct ucred *cred = ap->a_cred; + mode_t mode = ap->a_mode; + 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 +#ifdef QUOTA + if (mode & VWRITE) { + switch (vp->v_type) { + case VREG: case VDIR: case VLNK: + if (error = getinoquota(ip)) + return (error); + } + } +#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: + ; + } + if ((ip->i_mode & mode) != 0) + return (0); + return (EACCES); } /* ARGSUSED */ -ufs_getattr(vp, vap, cred) - struct vnode *vp; - register struct vattr *vap; - struct ucred *cred; +int +ufs_getattr(ap) + struct vop_getattr_args /* { + struct vnode *a_vp; + struct vattr *a_vap; + struct ucred *a_cred; + struct proc *a_p; + } */ *ap; { + register struct vnode *vp = ap->a_vp; register struct inode *ip = VTOI(vp); + register struct vattr *vap = ap->a_vap; ITIMES(ip, &time, &time); /* @@ -266,14 +246,10 @@ ufs_getattr(vp, vap, cred) 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_size = ip->i_din.di_size; + vap->va_atime = ip->i_atime; + vap->va_mtime = ip->i_mtime; + vap->va_ctime = ip->i_ctime; vap->va_flags = ip->i_flags; vap->va_gen = ip->i_gen; /* this doesn't belong here */ @@ -282,26 +258,35 @@ ufs_getattr(vp, vap, cred) else if (vp->v_type == VCHR) vap->va_blocksize = MAXBSIZE; else - vap->va_blocksize = ip->i_fs->fs_bsize; + vap->va_blocksize = vp->v_mount->mnt_stat.f_iosize; vap->va_bytes = dbtob(ip->i_blocks); - vap->va_bytes1 = -1; vap->va_type = vp->v_type; + vap->va_filerev = ip->i_modrev; return (0); } /* * Set attribute vnode op. called from several syscalls */ -ufs_setattr(vp, vap, cred) - register struct vnode *vp; - register struct vattr *vap; - register struct ucred *cred; +int +ufs_setattr(ap) + struct vop_setattr_args /* { + struct vnode *a_vp; + struct vattr *a_vap; + struct ucred *a_cred; + struct proc *a_p; + } */ *ap; { + register struct vattr *vap = ap->a_vap; + register struct vnode *vp = ap->a_vp; register struct inode *ip = VTOI(vp); - int error = 0; + register struct ucred *cred = ap->a_cred; + register struct proc *p = ap->a_p; + struct timeval atimeval, mtimeval; + int error; /* - * Check for unsetable attributes. + * Check for unsettable attributes. */ if ((vap->va_type != VNON) || (vap->va_nlink != VNOVAL) || (vap->va_fsid != VNOVAL) || (vap->va_fileid != VNOVAL) || @@ -312,32 +297,37 @@ ufs_setattr(vp, vap, cred) /* * 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)) + if (vap->va_uid != (uid_t)VNOVAL || vap->va_gid != (gid_t)VNOVAL) + if (error = ufs_chown(vp, vap->va_uid, vap->va_gid, cred, p)) return (error); if (vap->va_size != VNOVAL) { if (vp->v_type == VDIR) return (EISDIR); - if (error = itrunc(ip, vap->va_size)) + if (error = VOP_TRUNCATE(vp, vap->va_size, 0, cred, p)) return (error); } - if (vap->va_atime.tv_sec != VNOVAL || vap->va_mtime.tv_sec != VNOVAL) { + ip = VTOI(vp); + if (vap->va_atime.ts_sec != VNOVAL || vap->va_mtime.ts_sec != VNOVAL) { if (cred->cr_uid != ip->i_uid && - (error = suser(cred, &u.u_acflag))) + (error = suser(cred, &p->p_acflag))) return (error); - if (vap->va_atime.tv_sec != VNOVAL) + if (vap->va_atime.ts_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)) + if (vap->va_mtime.ts_sec != VNOVAL) + ip->i_flag |= IUPD | ICHG; + atimeval.tv_sec = vap->va_atime.ts_sec; + atimeval.tv_usec = vap->va_atime.ts_nsec / 1000; + mtimeval.tv_sec = vap->va_mtime.ts_sec; + mtimeval.tv_usec = vap->va_mtime.ts_nsec / 1000; + if (error = VOP_UPDATE(vp, &atimeval, &mtimeval, 1)) return (error); } - if (vap->va_mode != (u_short)VNOVAL) - error = chmod1(vp, (int)vap->va_mode, cred); + error = 0; + if (vap->va_mode != (mode_t)VNOVAL) + error = ufs_chmod(vp, (int)vap->va_mode, cred, p); if (vap->va_flags != VNOVAL) { if (cred->cr_uid != ip->i_uid && - (error = suser(cred, &u.u_acflag))) + (error = suser(cred, &p->p_acflag))) return (error); if (cred->cr_uid == 0) { ip->i_flags = vap->va_flags; @@ -354,28 +344,30 @@ ufs_setattr(vp, vap, cred) * Change the mode on a file. * Inode must be locked before calling. */ -chmod1(vp, mode, cred) +static int +ufs_chmod(vp, mode, cred, p) register struct vnode *vp; register int mode; - struct ucred *cred; + register struct ucred *cred; + struct proc *p; { register struct inode *ip = VTOI(vp); int error; if (cred->cr_uid != ip->i_uid && - (error = suser(cred, &u.u_acflag))) + (error = suser(cred, &p->p_acflag))) return (error); - ip->i_mode &= ~07777; if (cred->cr_uid) { - if (vp->v_type != VDIR) - mode &= ~ISVTX; - if (!groupmember(ip->i_gid, cred)) - mode &= ~ISGID; + if (vp->v_type != VDIR && (mode & ISVTX)) + return (EFTYPE); + if (!groupmember(ip->i_gid, cred) && (mode & ISGID)) + return (EPERM); } + ip->i_mode &= ~07777; ip->i_mode |= mode & 07777; ip->i_flag |= ICHG; if ((vp->v_flag & VTEXT) && (ip->i_mode & ISVTX) == 0) - xrele(vp); + (void) vnode_pager_uncache(vp); return (0); } @@ -383,21 +375,26 @@ chmod1(vp, mode, cred) * Perform chown operation on inode ip; * inode must be locked prior to call. */ -chown1(vp, uid, gid, cred) +static int +ufs_chown(vp, uid, gid, cred, p) register struct vnode *vp; uid_t uid; gid_t gid; struct ucred *cred; + struct proc *p; { register struct inode *ip = VTOI(vp); + uid_t ouid; + gid_t ogid; + int error = 0; #ifdef QUOTA - register long change; + register int i; + long change; #endif - int error; - if (uid == (u_short)VNOVAL) + if (uid == (uid_t)VNOVAL) uid = ip->i_uid; - if (gid == (u_short)VNOVAL) + if (gid == (gid_t)VNOVAL) gid = ip->i_gid; /* * If we don't own the file, are trying to change the owner @@ -406,56 +403,113 @@ chown1(vp, uid, gid, cred) */ if ((cred->cr_uid != ip->i_uid || uid != ip->i_uid || !groupmember((gid_t)gid, cred)) && - (error = suser(cred, &u.u_acflag))) + (error = suser(cred, &p->p_acflag))) return (error); + ouid = ip->i_uid; + ogid = 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); + if (error = getinoquota(ip)) + return (error); + 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; + } + 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; + } #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; #ifdef QUOTA - ip->i_dquot = inoquota(ip); - (void) chkdq(ip, change, 1); - (void) chkiq(ip->i_dev, (struct inode *)NULL, (uid_t)uid, 1); - return (u.u_error); /* should == 0 ALWAYS !! */ -#else + 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; + } + } + 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); + } + 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); -#endif } /* ARGSUSED */ -ufs_ioctl(vp, com, data, fflag, cred) - struct vnode *vp; - int com; - caddr_t data; - int fflag; - struct ucred *cred; +int +ufs_ioctl(ap) + struct vop_ioctl_args /* { + struct vnode *a_vp; + int a_command; + caddr_t a_data; + int a_fflag; + struct ucred *a_cred; + struct proc *a_p; + } */ *ap; { - printf("ufs_ioctl called with type %d\n", vp->v_type); return (ENOTTY); } /* ARGSUSED */ -ufs_select(vp, which, cred) - struct vnode *vp; - int which; - struct ucred *cred; +int +ufs_select(ap) + struct vop_select_args /* { + struct vnode *a_vp; + int a_which; + int a_fflags; + struct ucred *a_cred; + struct proc *a_p; + } */ *ap; { - printf("ufs_select called with type %d\n", vp->v_type); - return (1); /* XXX */ + /* + * We should really check to see if I/O is possible. + */ + return (1); } /* @@ -464,42 +518,33 @@ ufs_select(vp, which, cred) * NB Currently unsupported. */ /* ARGSUSED */ -ufs_mmap(vp, fflags, cred) - struct vnode *vp; - int fflags; - struct ucred *cred; +int +ufs_mmap(ap) + struct vop_mmap_args /* { + struct vnode *a_vp; + int a_fflags; + struct ucred *a_cred; + struct proc *a_p; + } */ *ap; { return (EINVAL); } -/* - * Synch an open file. - */ -/* ARGSUSED */ -ufs_fsync(vp, fflags, cred, waitfor) - struct vnode *vp; - int fflags; - struct ucred *cred; - int waitfor; -{ - struct inode *ip = VTOI(vp); - - if (fflags&FWRITE) - ip->i_flag |= ICHG; - return (syncip(ip, waitfor)); -} - /* * Seek on a file * * Nothing to do, so just return. */ /* ARGSUSED */ -ufs_seek(vp, oldoff, newoff, cred) - struct vnode *vp; - off_t oldoff, newoff; - struct ucred *cred; +int +ufs_seek(ap) + struct vop_seek_args /* { + struct vnode *a_vp; + off_t a_oldoff; + off_t a_newoff; + struct ucred *a_cred; + } */ *ap; { return (0); @@ -510,15 +555,20 @@ ufs_seek(vp, oldoff, newoff, cred) * Hard to avoid races here, especially * in unlinking directories. */ -ufs_remove(ndp) - struct nameidata *ndp; +int +ufs_remove(ap) + struct vop_remove_args /* { + struct vnode *a_dvp; + struct vnode *a_vp; + struct componentname *a_cnp; + } */ *ap; { register struct inode *ip, *dp; int error; - ip = VTOI(ndp->ni_vp); - dp = VTOI(ndp->ni_dvp); - error = dirremove(ndp); + ip = VTOI(ap->a_vp); + dp = VTOI(ap->a_dvp); + error = ufs_dirremove(ap->a_dvp, ap->a_cnp); if (!error) { ip->i_nlink--; ip->i_flag |= ICHG; @@ -526,35 +576,59 @@ ufs_remove(ndp) if (dp == ip) vrele(ITOV(ip)); else - iput(ip); - iput(dp); + ufs_iput(ip); + ufs_iput(dp); return (error); } /* * link vnode call */ -ufs_link(vp, ndp) - register struct vnode *vp; - register struct nameidata *ndp; +int +ufs_link(ap) + struct vop_link_args /* { + struct vnode *a_vp; + struct vnode *a_tdvp; + struct componentname *a_cnp; + } */ *ap; { - register struct inode *ip = VTOI(vp); + register struct vnode *vp = ap->a_vp; + register struct vnode *tdvp = ap->a_tdvp; + register struct componentname *cnp = ap->a_cnp; + register struct inode *ip; + struct timeval tv; int error; - if (ndp->ni_dvp != vp) - ILOCK(ip); - if (ip->i_nlink == LINK_MAX - 1) { - error = EMLINK; - goto out; + if (vp->v_mount != tdvp->v_mount) { + VOP_ABORTOP(vp, cnp); + if (tdvp == vp) + vrele(vp); + else + vput(vp); + return (EXDEV); } + +#ifdef DIAGNOSTIC + if ((cnp->cn_flags & HASBUF) == 0) + panic("ufs_link: no name"); +#endif + ip = VTOI(tdvp); + if ((nlink_t)ip->i_nlink >= LINK_MAX) { + free(cnp->cn_pnbuf, M_NAMEI); + return (EMLINK); + } + if (vp != tdvp) + ILOCK(ip); ip->i_nlink++; ip->i_flag |= ICHG; - error = iupdat(ip, &time, &time, 1); + tv = time; + error = VOP_UPDATE(tdvp, &tv, &tv, 1); if (!error) - error = direnter(ip, ndp); -out: - if (ndp->ni_dvp != vp) + error = ufs_direnter(ip, vp, cnp); + if (vp != tdvp) IUNLOCK(ip); + FREE(cnp->cn_pnbuf, M_NAMEI); + vput(vp); if (error) { ip->i_nlink--; ip->i_flag |= ICHG; @@ -562,6 +636,162 @@ out: return (error); } + + +/* + * relookup - lookup a path name component + * Used by lookup to re-aquire things. + */ +int +relookup(dvp, vpp, cnp) + struct vnode *dvp, **vpp; + struct componentname *cnp; +{ + 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 */ + char *cp; /* DEBUG: check name ptr/len */ + int newhash; /* DEBUG: check name hash */ + int error = 0; + + /* + * 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 + for (newhash = 0, 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"); + if (*cp != 0) + panic("relookup: not last component"); + printf("{%s}: ", cnp->cn_nameptr); +#endif + + /* + * 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 + if (error != EJUSTRETURN) + 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; + } + /* ASSERT(dvp == ndp->ni_startdir) */ + if (cnp->cn_flags & SAVESTART) + VREF(dvp); + /* + * 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. + */ + return (0); + } + dp = *vpp; + +#ifdef DIAGNOSTIC + /* + * Check for symbolic link + */ + if (dp->v_type == VLNK && (cnp->cn_flags & FOLLOW)) + panic ("relookup: symlink found.\n"); +#endif + +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; + } + } + /* ASSERT(dvp == ndp->ni_startdir) */ + if (cnp->cn_flags & SAVESTART) + VREF(dvp); + + if (!wantparent) + vrele(dvp); + if ((cnp->cn_flags & LOCKLEAF) == 0) + VOP_UNLOCK(dp); + return (0); + +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"); @@ -586,35 +816,92 @@ out: * is different from the source, patch the ".." entry in the * directory. */ -ufs_rename(fndp, tndp) - register struct nameidata *fndp, *tndp; +int +ufs_rename(ap) + struct vop_rename_args /* { + struct vnode *a_fdvp; + struct vnode *a_fvp; + struct componentname *a_fcnp; + struct vnode *a_tdvp; + struct vnode *a_tvp; + struct componentname *a_tcnp; + } */ *ap; { + struct vnode *tvp = ap->a_tvp; + register struct vnode *tdvp = ap->a_tdvp; + struct vnode *fvp = ap->a_fvp; + register struct vnode *fdvp = ap->a_fdvp; + register struct componentname *tcnp = ap->a_tcnp; + register struct componentname *fcnp = ap->a_fcnp; register struct inode *ip, *xp, *dp; struct dirtemplate dirbuf; + struct timeval tv; int doingdirectory = 0, oldparent = 0, newparent = 0; int error = 0; + int fdvpneedsrele = 1, tdvpneedsrele = 1; + u_char namlen; + + /* Check for cross-device rename */ + if ((fvp->v_mount != tdvp->v_mount) || + (tvp && (fvp->v_mount != tvp->v_mount))) { + VOP_ABORTOP(tdvp, tcnp); /* XXX, why not in NFS? */ + if (tdvp == tvp) + vrele(tdvp); + else + vput(tdvp); + if (tvp) + vput(tvp); + VOP_ABORTOP(fdvp, fcnp); /* XXX, why not in NFS? */ + vrele(fdvp); + vrele(fvp); + return (EXDEV); + } - dp = VTOI(fndp->ni_dvp); - ip = VTOI(fndp->ni_vp); +#ifdef DIAGNOSTIC + 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 = &fndp->ni_dent; - /* * Avoid ".", "..", and aliases of "." for obvious reasons. */ - 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); + 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++; } - vrele(fndp->ni_dvp); + vrele(fdvp); /* * 1) Bump link count while we're moving stuff @@ -624,17 +911,18 @@ ufs_rename(fndp, tndp) */ ip->i_nlink++; ip->i_flag |= ICHG; - error = iupdat(ip, &time, &time, 1); + tv = time; + error = VOP_UPDATE(fvp, &tv, &tv, 1); IUNLOCK(ip); /* * When the target exists, both the directory * and target vnodes are returned locked. */ - dp = VTOI(tndp->ni_dvp); + dp = VTOI(tdvp); xp = NULL; - if (tndp->ni_vp) - xp = VTOI(tndp->ni_vp); + 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 @@ -648,21 +936,24 @@ ufs_rename(fndp, tndp) if (oldparent != dp->i_number) newparent = dp->i_number; if (doingdirectory && newparent) { - if (error = iaccess(ip, IWRITE, tndp->ni_cred)) + VOP_LOCK(fvp); + error = VOP_ACCESS(fvp, VWRITE, tcnp->cn_cred, tcnp->cn_proc); + VOP_UNLOCK(fvp); + if (error) goto bad; - tndp->ni_nameiop = RENAME | LOCKPARENT | LOCKLEAF | NOCACHE; - do { - dp = VTOI(tndp->ni_dvp); - if (xp != NULL) - iput(xp); - if (error = checkpath(ip, dp, tndp->ni_cred)) - goto out; - if (error = namei(tndp)) - goto out; - xp = NULL; - if (tndp->ni_vp) - xp = VTOI(tndp->ni_vp); - } while (dp != VTOI(tndp->ni_dvp)); + 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 @@ -680,12 +971,24 @@ ufs_rename(fndp, tndp) * parent we don't fool with the link count. */ if (doingdirectory && newparent) { + if ((nlink_t)dp->i_nlink >= LINK_MAX) { + error = EMLINK; + goto bad; + } dp->i_nlink++; dp->i_flag |= ICHG; - error = iupdat(dp, &time, &time, 1); + if (error = VOP_UPDATE(ITOV(dp), &tv, &tv, 1)) + goto bad; } - if (error = direnter(ip, tndp)) - goto out; + if (error = ufs_direnter(ip, tdvp, tcnp)) { + if (doingdirectory && newparent) { + dp->i_nlink--; + dp->i_flag |= ICHG; + (void)VOP_UPDATE(ITOV(dp), &tv, &tv, 1); + } + goto bad; + } + ufs_iput(dp); } else { if (xp->i_dev != dp->i_dev || xp->i_dev != ip->i_dev) panic("rename: EXDEV"); @@ -700,21 +1003,19 @@ ufs_rename(fndp, tndp) * 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) { + 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, tndp->ni_cred) || + if (!ufs_dirempty(xp, dp->i_number, tcnp->cn_cred) || xp->i_nlink > 2) { error = ENOTEMPTY; goto bad; @@ -728,9 +1029,19 @@ ufs_rename(fndp, tndp) error = EISDIR; goto bad; } - if (error = dirrewrite(dp, ip, tndp)) + if (error = ufs_dirrewrite(dp, ip, tcnp)) goto bad; - vput(ITOV(dp)); + /* + * 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 @@ -745,26 +1056,35 @@ ufs_rename(fndp, tndp) if (doingdirectory) { if (--xp->i_nlink != 0) panic("rename: linked directory"); - error = itrunc(xp, (u_long)0); + error = VOP_TRUNCATE(ITOV(xp), (off_t)0, IO_SYNC, + tcnp->cn_cred, tcnp->cn_proc); } xp->i_flag |= ICHG; - iput(xp); + ufs_iput(xp); xp = NULL; } /* * 3) Unlink the source. */ - fndp->ni_nameiop = DELETE | LOCKPARENT | LOCKLEAF; - (void)namei(fndp); - if (fndp->ni_vp != NULL) { - xp = VTOI(fndp->ni_vp); - dp = VTOI(fndp->ni_dvp); +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); + if (fvp != NULL) { + xp = VTOI(fvp); + dp = VTOI(fdvp); } else { - if (fndp->ni_dvp != NULL) - vput(fndp->ni_dvp); - xp = NULL; - dp = NULL; + /* + * From name has disappeared. + */ + if (doingdirectory) + panic("rename: lost dir entry"); + vrele(ITOV(ip)); + return (0); } /* * Ensure that the directory entry still exists and has not @@ -792,12 +1112,21 @@ ufs_rename(fndp, tndp) 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); + tcnp->cn_cred, (int *)0, (struct proc *)0); if (error == 0) { - if (dirbuf.dotdot_namlen != 2 || +# if (BYTE_ORDER == LITTLE_ENDIAN) + if (fvp->v_mount->mnt_maxsymlinklen <= 0) + namlen = dirbuf.dotdot_type; + else + namlen = dirbuf.dotdot_namlen; +# else + namlen = dirbuf.dotdot_namlen; +# endif + if (namlen != 2 || dirbuf.dotdot_name[0] != '.' || dirbuf.dotdot_name[1] != '.') { - printf("rename: mangled dir\n"); + ufs_dirbad(xp, (doff_t)12, + "rename: mangled dir"); } else { dirbuf.dotdot_ino = newparent; (void) vn_rdwr(UIO_WRITE, ITOV(xp), @@ -805,12 +1134,13 @@ ufs_rename(fndp, tndp) sizeof (struct dirtemplate), (off_t)0, UIO_SYSSPACE, IO_NODELOCKED|IO_SYNC, - tndp->ni_cred, (int *)0); + tcnp->cn_cred, (int *)0, + (struct proc *)0); cache_purge(ITOV(dp)); } } } - error = dirremove(fndp); + error = ufs_dirremove(fdvp, fcnp); if (!error) { xp->i_nlink--; xp->i_flag |= ICHG; @@ -838,7 +1168,11 @@ out: /* * A virgin directory (no blushing please). */ -struct dirtemplate mastertemplate = { +static struct dirtemplate mastertemplate = { + 0, 12, DT_DIR, 1, ".", + 0, DIRBLKSIZ - 12, DT_DIR, 2, ".." +}; +static struct odirtemplate omastertemplate = { 0, 12, 1, ".", 0, DIRBLKSIZ - 12, 2, ".." }; @@ -846,48 +1180,65 @@ struct dirtemplate mastertemplate = { /* * Mkdir system call */ -ufs_mkdir(ndp, vap) - struct nameidata *ndp; - struct vattr *vap; +int +ufs_mkdir(ap) + struct vop_mkdir_args /* { + struct vnode *a_dvp; + struct vnode **a_vpp; + struct componentname *a_cnp; + struct vattr *a_vap; + } */ *ap; { + register struct vnode *dvp = ap->a_dvp; + register struct vattr *vap = ap->a_vap; + register struct componentname *cnp = ap->a_cnp; register struct inode *ip, *dp; - struct inode *tip; - struct vnode *dvp; - struct dirtemplate dirtemplate; - int error; - int dmode; - - dvp = ndp->ni_dvp; + struct vnode *tvp; + struct dirtemplate dirtemplate, *dtp; + struct timeval tv; + int error, dmode; + +#ifdef DIAGNOSTIC + if ((cnp->cn_flags & HASBUF) == 0) + panic("ufs_mkdir: no name"); +#endif dp = VTOI(dvp); + if ((nlink_t)dp->i_nlink >= LINK_MAX) { + free(cnp->cn_pnbuf, M_NAMEI); + ufs_iput(dp); + return (EMLINK); + } 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. */ - error = ialloc(dp, dirpref(dp->i_fs), dmode, &tip); - if (error) { - iput(dp); + if (error = VOP_VALLOC(dvp, dmode, cnp->cn_cred, &tvp)) { + free(cnp->cn_pnbuf, M_NAMEI); + ufs_iput(dp); return (error); } - ip = tip; + 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 = dmode; ITOV(ip)->v_type = VDIR; /* Rest init'd in iget() */ ip->i_nlink = 2; - ip->i_uid = ndp->ni_cred->cr_uid; - ip->i_gid = dp->i_gid; -#ifdef QUOTA - ip->i_dquot = inoquota(ip); -#endif - error = iupdat(ip, &time, &time, 1); + tv = time; + error = VOP_UPDATE(ITOV(ip), &tv, &tv, 1); /* * Bump link count in parent directory @@ -897,77 +1248,77 @@ ufs_mkdir(ndp, vap) */ dp->i_nlink++; dp->i_flag |= ICHG; - error = iupdat(dp, &time, &time, 1); + if (error = VOP_UPDATE(ITOV(dp), &tv, &tv, 1)) + goto bad; - /* - * Initialize directory with "." - * and ".." from static template. - */ - dirtemplate = mastertemplate; + /* Initialize directory with "." and ".." from static template. */ + if (dvp->v_mount->mnt_maxsymlinklen > 0) + dtp = &mastertemplate; + else + dtp = (struct dirtemplate *)&omastertemplate; + dirtemplate = *dtp; dirtemplate.dot_ino = ip->i_number; dirtemplate.dotdot_ino = dp->i_number; 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); + 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; } - if (DIRBLKSIZ > dp->i_fs->fs_fsize) - panic("mkdir: blksize"); /* XXX - should grow w/balloc() */ - else + 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; - /* - * Directory all set up, now - * install the entry for it in - * the parent directory. - */ - error = direnter(ip, ndp); - dp = NULL; - if (error) { - ndp->ni_nameiop = LOOKUP | NOCACHE; - error = namei(ndp); - if (!error) { - dp = VTOI(ndp->ni_vp); - dp->i_nlink--; - dp->i_flag |= ICHG; - } + 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, - * vrele 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 (error) { ip->i_nlink = 0; ip->i_flag |= ICHG; - iput(ip); + ufs_iput(ip); } else - ndp->ni_vp = ITOV(ip); - if (dp) - iput(dp); + *ap->a_vpp = ITOV(ip); + FREE(cnp->cn_pnbuf, M_NAMEI); + ufs_iput(dp); return (error); } /* * Rmdir system call. */ -ufs_rmdir(ndp) - register struct nameidata *ndp; +int +ufs_rmdir(ap) + struct vop_rmdir_args /* { + struct vnode *a_dvp; + struct vnode *a_vp; + struct componentname *a_cnp; + } */ *ap; { + register struct vnode *dvp = ap->a_dvp; + register struct componentname *cnp = ap->a_cnp; register struct inode *ip, *dp; - int error = 0; + int error; - ip = VTOI(ndp->ni_vp); - dp = VTOI(ndp->ni_dvp); + ip = VTOI(ap->a_vp); + dp = VTOI(dvp); /* * No rmdir "." please. */ if (dp == ip) { - vrele(ITOV(dp)); - iput(ip); + vrele(dvp); + ufs_iput(ip); return (EINVAL); } /* @@ -977,7 +1328,9 @@ ufs_rmdir(ndp) * the current directory and thus be * non-empty.) */ - if (ip->i_nlink != 2 || !dirempty(ip, dp->i_number, ndp->ni_cred)) { + error = 0; + if (ip->i_nlink != 2 || + !ufs_dirempty(ip, dp->i_number, cnp->cn_cred)) { error = ENOTEMPTY; goto out; } @@ -986,13 +1339,13 @@ ufs_rmdir(ndp) * inode. If we crash in between, the directory * will be reattached to lost+found, */ - if (error = dirremove(ndp)) + if (error = ufs_dirremove(dvp, cnp)) goto out; dp->i_nlink--; dp->i_flag |= ICHG; - cache_purge(ITOV(dp)); - iput(dp); - ndp->ni_dvp = NULL; + cache_purge(dvp); + ufs_iput(dp); + dvp = NULL; /* * Truncate inode. The only stuff left * in the directory is "." and "..". The @@ -1005,43 +1358,69 @@ ufs_rmdir(ndp) * worry about them later. */ ip->i_nlink -= 2; - error = itrunc(ip, (u_long)0); + error = VOP_TRUNCATE(ap->a_vp, (off_t)0, IO_SYNC, cnp->cn_cred, + cnp->cn_proc); cache_purge(ITOV(ip)); out: - if (ndp->ni_dvp) - iput(dp); - iput(ip); + if (dvp) + ufs_iput(dp); + ufs_iput(ip); return (error); } /* * symlink -- make a symbolic link */ -ufs_symlink(ndp, vap, target) - struct nameidata *ndp; - struct vattr *vap; - char *target; +int +ufs_symlink(ap) + struct vop_symlink_args /* { + struct vnode *a_dvp; + struct vnode **a_vpp; + struct componentname *a_cnp; + struct vattr *a_vap; + char *a_target; + } */ *ap; { - struct inode *ip; - int error; + register struct vnode *vp, **vpp = ap->a_vpp; + register struct inode *ip; + int len, error; - error = maknode(IFLNK | vap->va_mode, ndp, &ip); - if (error) + if (error = ufs_makeinode(IFLNK | ap->a_vap->va_mode, ap->a_dvp, + vpp, ap->a_cnp)) 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); + vp = *vpp; + len = strlen(ap->a_target); + if (len < vp->v_mount->mnt_maxsymlinklen) { + ip = VTOI(vp); + bcopy(ap->a_target, (char *)ip->i_shortlink, len); + ip->i_size = len; + ip->i_flag |= IUPD|ICHG; + } else + error = vn_rdwr(UIO_WRITE, vp, ap->a_target, len, (off_t)0, + UIO_SYSSPACE, IO_NODELOCKED, ap->a_cnp->cn_cred, (int *)0, + (struct proc *)0); + vput(vp); return (error); } /* - * Vnode op for read and write + * Vnode op for reading directories. + * + * The routine below assumes that the on-disk format of a directory + * is the same as that defined by . If the on-disk + * format changes, then it will be necessary to do a conversion + * from the on-disk format that read returns to the format defined + * by . */ -ufs_readdir(vp, uio, cred) - struct vnode *vp; - register struct uio *uio; - struct ucred *cred; +int +ufs_readdir(ap) + struct vop_readdir_args /* { + struct vnode *a_vp; + struct uio *a_uio; + struct ucred *a_cred; + } */ *ap; { + register struct uio *uio = ap->a_uio; int count, lost, error; count = uio->uio_resid; @@ -1051,7 +1430,48 @@ ufs_readdir(vp, uio, cred) return (EINVAL); uio->uio_resid = count; uio->uio_iov->iov_len = count; - error = ufs_read(vp, uio, 0, cred); +# if (BYTE_ORDER == LITTLE_ENDIAN) + if (ap->a_vp->v_mount->mnt_maxsymlinklen > 0) { + error = VOP_READ(ap->a_vp, uio, 0, ap->a_cred); + } else { + struct dirent *dp, *edp; + struct uio auio; + struct iovec aiov; + caddr_t dirbuf; + int readcnt; + u_char tmp; + + auio = *uio; + auio.uio_iov = &aiov; + auio.uio_iovcnt = 1; + auio.uio_segflg = UIO_SYSSPACE; + aiov.iov_len = count; + MALLOC(dirbuf, caddr_t, count, M_TEMP, M_WAITOK); + aiov.iov_base = dirbuf; + error = VOP_READ(ap->a_vp, &auio, 0, ap->a_cred); + if (error == 0) { + readcnt = count - auio.uio_resid; + edp = (struct dirent *)&dirbuf[readcnt]; + for (dp = (struct dirent *)dirbuf; dp < edp; ) { + tmp = dp->d_namlen; + dp->d_namlen = dp->d_type; + dp->d_type = tmp; + if (dp->d_reclen > 0) { + dp = (struct dirent *) + ((char *)dp + dp->d_reclen); + } else { + error = EIO; + break; + } + } + if (dp >= edp) + error = uiomove(dirbuf, readcnt, uio); + } + FREE(dirbuf, M_TEMP); + } +# else + error = VOP_READ(ap->a_vp, uio, 0, ap->a_cred); +# endif uio->uio_resid += lost; return (error); } @@ -1059,52 +1479,68 @@ ufs_readdir(vp, uio, cred) /* * Return target name of a symbolic link */ -ufs_readlink(vp, uiop, cred) - struct vnode *vp; - struct uio *uiop; - struct ucred *cred; +int +ufs_readlink(ap) + struct vop_readlink_args /* { + struct vnode *a_vp; + struct uio *a_uio; + struct ucred *a_cred; + } */ *ap; { + register struct vnode *vp = ap->a_vp; + register struct inode *ip = VTOI(vp); + int isize; - return (ufs_read(vp, uiop, 0, cred)); + isize = ip->i_size; + if (isize < vp->v_mount->mnt_maxsymlinklen) { + uiomove((char *)ip->i_shortlink, isize, ap->a_uio); + return (0); + } + return (VOP_READ(vp, ap->a_uio, 0, ap->a_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. + * done. If a buffer has been saved in anticipation of a CREATE, delete it. */ -ufs_abortop(ndp) - register struct nameidata *ndp; +/* ARGSUSED */ +int +ufs_abortop(ap) + struct vop_abortop_args /* { + struct vnode *a_dvp; + struct componentname *a_cnp; + } */ *ap; { - register struct inode *ip; - - if (ndp->ni_vp) { - ip = VTOI(ndp->ni_vp); - if (ip->i_flag & ILOCKED) - IUNLOCK(ip); - vrele(ndp->ni_vp); - } - if (ndp->ni_dvp) { - ip = VTOI(ndp->ni_dvp); - if (ip->i_flag & ILOCKED) - IUNLOCK(ip); - vrele(ndp->ni_dvp); - } - return; + if ((ap->a_cnp->cn_flags & (HASBUF | SAVESTART)) == HASBUF) + FREE(ap->a_cnp->cn_pnbuf, M_NAMEI); + return (0); } -ufs_lock(vp) - struct vnode *vp; +/* + * Lock an inode. + */ +int +ufs_lock(ap) + struct vop_lock_args /* { + struct vnode *a_vp; + } */ *ap; { - register struct inode *ip = VTOI(vp); + register struct inode *ip = VTOI(ap->a_vp); ILOCK(ip); return (0); } -ufs_unlock(vp) - struct vnode *vp; +/* + * Unlock an inode. + */ +int +ufs_unlock(ap) + struct vop_unlock_args /* { + struct vnode *a_vp; + } */ *ap; { - register struct inode *ip = VTOI(vp); + register struct inode *ip = VTOI(ap->a_vp); if (!(ip->i_flag & ILOCKED)) panic("ufs_unlock NOT LOCKED"); @@ -1113,91 +1549,435 @@ ufs_unlock(vp) } /* - * Get access to bmap + * Check for a locked inode. */ -ufs_bmap(vp, bn, vpp, bnp) - struct vnode *vp; - daddr_t bn; - struct vnode **vpp; - daddr_t *bnp; +int +ufs_islocked(ap) + struct vop_islocked_args /* { + struct vnode *a_vp; + } */ *ap; +{ + + if (VTOI(ap->a_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 +ufs_strategy(ap) + struct vop_strategy_args /* { + struct buf *a_bp; + } */ *ap; { - struct inode *ip = VTOI(vp); + register struct buf *bp = ap->a_bp; + register struct vnode *vp = bp->b_vp; + register struct inode *ip; + int error; - if (vpp != NULL) - *vpp = ip->i_devvp; - if (bnp == NULL) + ip = VTOI(vp); + if (vp->v_type == VBLK || vp->v_type == VCHR) + panic("ufs_strategy: spec"); + if (bp->b_blkno == bp->b_lblkno) { + if (error = + VOP_BMAP(vp, bp->b_lblkno, NULL, &bp->b_blkno, NULL)) { + bp->b_error = error; + bp->b_flags |= B_ERROR; + biodone(bp); + return (error); + } + if ((long)bp->b_blkno == -1) + clrbuf(bp); + } + if ((long)bp->b_blkno == -1) { + biodone(bp); return (0); - return (bmap(ip, bn, bnp, (daddr_t *)0, (int *)0)); + } + vp = ip->i_devvp; + bp->b_dev = vp->v_rdev; + VOCALL (vp->v_op, VOFFSET(vop_strategy), ap); + return (0); } /* - * Just call the device strategy routine + * Print out the contents of an inode. */ -ufs_strategy(bp) - register struct buf *bp; +int +ufs_print(ap) + struct vop_print_args /* { + struct vnode *a_vp; + } */ *ap; { - (*bdevsw[major(bp->b_dev)].d_strategy)(bp); + register struct vnode *vp = ap->a_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); } /* - * Make a new file. + * Read wrapper for special devices. */ -maknode(mode, ndp, ipp) - int mode; - register struct nameidata *ndp; - struct inode **ipp; +int +ufsspec_read(ap) + struct vop_read_args /* { + struct vnode *a_vp; + struct uio *a_uio; + int a_ioflag; + struct ucred *a_cred; + } */ *ap; { - register struct inode *ip; - struct inode *tip; - register struct inode *pdir = VTOI(ndp->ni_dvp); - ino_t ipref; + + /* + * Set access flag. + */ + VTOI(ap->a_vp)->i_flag |= IACC; + return (VOCALL (spec_vnodeop_p, VOFFSET(vop_read), ap)); +} + +/* + * Write wrapper for special devices. + */ +int +ufsspec_write(ap) + struct vop_write_args /* { + struct vnode *a_vp; + struct uio *a_uio; + int a_ioflag; + struct ucred *a_cred; + } */ *ap; +{ + + /* + * Set update and change flags. + */ + VTOI(ap->a_vp)->i_flag |= IUPD|ICHG; + return (VOCALL (spec_vnodeop_p, VOFFSET(vop_write), ap)); +} + +/* + * Close wrapper for special devices. + * + * Update the times on the inode then do device close. + */ +int +ufsspec_close(ap) + struct vop_close_args /* { + struct vnode *a_vp; + int a_fflag; + struct ucred *a_cred; + struct proc *a_p; + } */ *ap; +{ + register struct inode *ip = VTOI(ap->a_vp); + + if (ap->a_vp->v_usecount > 1 && !(ip->i_flag & ILOCKED)) + ITIMES(ip, &time, &time); + return (VOCALL (spec_vnodeop_p, VOFFSET(vop_close), ap)); +} + +#ifdef FIFO +/* + * Read wrapper for fifo's + */ +int +ufsfifo_read(ap) + struct vop_read_args /* { + struct vnode *a_vp; + struct uio *a_uio; + int a_ioflag; + struct ucred *a_cred; + } */ *ap; +{ + extern int (**fifo_vnodeop_p)(); + + /* + * Set access flag. + */ + VTOI(ap->a_vp)->i_flag |= IACC; + return (VOCALL (fifo_vnodeop_p, VOFFSET(vop_read), ap)); +} + +/* + * Write wrapper for fifo's. + */ +int +ufsfifo_write(ap) + struct vop_write_args /* { + struct vnode *a_vp; + struct uio *a_uio; + int a_ioflag; + struct ucred *a_cred; + } */ *ap; +{ + extern int (**fifo_vnodeop_p)(); + + /* + * Set update and change flags. + */ + VTOI(ap->a_vp)->i_flag |= IUPD|ICHG; + return (VOCALL (fifo_vnodeop_p, VOFFSET(vop_write), ap)); +} + +/* + * Close wrapper for fifo's. + * + * Update the times on the inode then do device close. + */ +ufsfifo_close(ap) + struct vop_close_args /* { + struct vnode *a_vp; + int a_fflag; + struct ucred *a_cred; + struct proc *a_p; + } */ *ap; +{ + extern int (**fifo_vnodeop_p)(); + register struct inode *ip = VTOI(ap->a_vp); + + if (ap->a_vp->v_usecount > 1 && !(ip->i_flag & ILOCKED)) + ITIMES(ip, &time, &time); + return (VOCALL (fifo_vnodeop_p, VOFFSET(vop_close), ap)); +} +#endif /* FIFO */ + +/* + * Advisory record locking support + */ +int +ufs_advlock(ap) + struct vop_advlock_args /* { + struct vnode *a_vp; + caddr_t a_id; + int a_op; + struct flock *a_fl; + int a_flags; + } */ *ap; +{ + register struct inode *ip = VTOI(ap->a_vp); + register struct flock *fl = ap->a_fl; + register struct lockf *lock; + off_t start, end; int error; - *ipp = 0; - if ((mode & IFMT) == IFDIR) - ipref = dirpref(pdir->i_fs); + /* + * Avoid the common case of unlocking when inode has no locks. + */ + if (ip->i_lockf == (struct lockf *)0) { + if (ap->a_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; + + default: + return (EINVAL); + } + if (start < 0) + return (EINVAL); + if (fl->l_len == 0) + end = -1; else - ipref = pdir->i_number; - error = ialloc(pdir, ipref, mode, &tip); - if (error) { - iput(pdir); + 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 = ap->a_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 = ap->a_flags; + /* + * Do the requested operation. + */ + switch(ap->a_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); } - ip = tip; -#ifdef QUOTA - if (ip->i_dquot != NODQUOT) - panic("maknode: dquot"); + /* NOTREACHED */ +} + +/* + * Initialize the vnode associated with a new inode, handle aliased + * vnodes. + */ +int +ufs_vinit(mntp, specops, fifoops, vpp) + struct mount *mntp; + int (**specops)(); + int (**fifoops)(); + struct vnode **vpp; +{ + struct inode *ip; + struct vnode *vp, *nvp; + + 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. + */ + ufs_ihashrem(ip); + IUNLOCK(ip); + nvp->v_data = vp->v_data; + vp->v_data = NULL; + vp->v_op = spec_vnodeop_p; + 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 - ip->i_flag |= IACC|IUPD|ICHG; + } + 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); +} + +/* + * Allocate a new inode. + */ +int +ufs_makeinode(mode, dvp, vpp, cnp) + int mode; + struct vnode *dvp; + struct vnode **vpp; + struct componentname *cnp; +{ + register struct inode *ip, *pdir; + struct timeval tv; + struct vnode *tvp; + int error; + + pdir = VTOI(dvp); +#ifdef DIAGNOSTIC + if ((cnp->cn_flags & HASBUF) == 0) + panic("ufs_makeinode: no name"); +#endif + *vpp = NULL; 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; + + 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; - 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); + 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 = 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; + tv = time; + if (error = VOP_UPDATE(tvp, &tv, &tv, 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); }