statfs.f_bsize => statfs.f_iosize; statfs.f_fsize => statfs.f_bsize (for SunOS)
[unix-history] / usr / src / sys / ufs / ffs / ufs_lookup.c
index d850468..124a0ca 100644 (file)
-/*     ufs_lookup.c    4.11    82/03/06        */
+/*
+ * Copyright (c) 1989 The Regents of the University of California.
+ * All rights reserved.
+ *
+ * %sccs.include.redist.c%
+ *
+ *     @(#)ufs_lookup.c        7.36 (Berkeley) %G%
+ */
+
+#include <sys/param.h>
+#include <sys/namei.h>
+#include <sys/buf.h>
+#include <sys/file.h>
+#include <sys/mount.h>
+#include <sys/vnode.h>
+
+#include <ufs/ufs/quota.h>
+#include <ufs/ufs/inode.h>
+#include <ufs/ufs/dir.h>
+#include <ufs/ufs/ufsmount.h>
+#include <ufs/ufs/ufs_extern.h>
 
 
-#include "../h/param.h"
-#include "../h/systm.h"
-#include "../h/inode.h"
-#include "../h/mount.h"
-#include "../h/dir.h"
-#include "../h/user.h"
-#include "../h/buf.h"
-#include "../h/conf.h"
+struct nchstats nchstats;
+#ifdef DIAGNOSTIC
+int    dirchk = 1;
+#else
+int    dirchk = 0;
+#endif
 
 /*
 
 /*
- * Convert a pathname into a pointer to
- * a locked inode.
+ * Convert a component of a pathname into a pointer to a locked inode.
+ * This is a very central and rather complicated routine.
+ * If the file system is not maintained in a strict tree hierarchy,
+ * this can result in a deadlock situation (see comments in code below).
+ *
+ * The flag argument is LOOKUP, CREATE, RENAME, or DELETE depending on
+ * whether the name is to be looked up, created, renamed, or deleted.
+ * When CREATE, RENAME, or DELETE is specified, information usable in
+ * creating, renaming, or deleting a directory entry may be calculated.
+ * If flag has LOCKPARENT or'ed into it and the target of the pathname
+ * exists, lookup returns both the target and its parent directory locked.
+ * When creating or renaming and LOCKPARENT is specified, the target may
+ * not be ".".  When deleting and LOCKPARENT is specified, the target may
+ * be "."., but the caller must check to ensure it does an vrele and iput
+ * instead of two iputs.
  *
  *
- * func = function called to get next char of name
- *     &uchar if name is in user space
- *     &schar if name is in system space
- * flag = 0 if name is sought
- *     1 if name is to be created
- *     2 if name is to be deleted
- * follow = 1 if links are to be followed at the end of the name
+ * Overall outline of ufs_lookup:
+ *
+ *     check accessibility of directory
+ *     look for name in cache, if found, then if at end of path
+ *       and deleting or creating, drop it, else return name
+ *     search for name in directory, to found or notfound
+ * notfound:
+ *     if creating, return locked directory, leaving info on available slots
+ *     else return error
+ * found:
+ *     if at end of path and deleting, return information to allow delete
+ *     if at end of path and rewriting (RENAME and LOCKPARENT), lock target
+ *       inode and return info to allow rewrite
+ *     if not at end, add name to cache; if at end and neither creating
+ *       nor deleting, add name to cache
+ *
+ * NOTE: (LOOKUP | LOCKPARENT) currently returns the parent inode unlocked.
  */
  */
-struct inode *
-namei(func, flag, follow)
-       int (*func)(), flag, follow;
+int
+ufs_lookup(vdp, ndp, p)
+       register struct vnode *vdp;
+       register struct nameidata *ndp;
+       struct proc *p;
 {
 {
-       register struct inode *dp;
-       register char *cp;
-       register struct buf *bp, *nbp;
-       register struct direct *ep;
-       struct inode *pdp;
-       int i, nlink;
-       dev_t d;
-       ino_t ino;
-       off_t eo;
+       register struct inode *dp;      /* the directory we are searching */
+       struct buf *bp;                 /* a buffer of directory entries */
+       register struct direct *ep;     /* the current directory entry */
+       int entryoffsetinblock;         /* offset of ep in bp's buffer */
+       enum {NONE, COMPACT, FOUND} slotstatus;
+       int slotoffset;                 /* offset of area with free space */
+       int slotsize;                   /* size of area at slotoffset */
+       int slotfreespace;              /* amount of space free in slot */
+       int slotneeded;                 /* size of the entry we're seeking */
+       int numdirpasses;               /* strategy for directory search */
+       int endsearch;                  /* offset to end directory search */
+       int prevoff;                    /* prev entry ndp->ni_ufs.ufs_offset */
+       struct inode *pdp;              /* saved dp during symlink work */
+       struct vnode *tdp;              /* returned by VOP_VGET */
+       off_t enduseful;                /* pointer past last used dir slot */
+       u_long bmask;                   /* block offset mask */
+       int flag;                       /* LOOKUP, CREATE, RENAME, or DELETE */
+       int lockparent;                 /* 1 => lockparent flag is set */
+       int wantparent;                 /* 1 => wantparent or lockparent flag */
+       int error;
+
+       bp = NULL;
+       slotoffset = -1;
+       ndp->ni_dvp = vdp;
+       ndp->ni_vp = NULL;
+       dp = VTOI(vdp);
+       lockparent = ndp->ni_nameiop & LOCKPARENT;
+       flag = ndp->ni_nameiop & OPMASK;
+       wantparent = ndp->ni_nameiop & (LOCKPARENT|WANTPARENT);
 
        /*
 
        /*
-        * allocate name buffer; copy name
+        * Check accessiblity of directory.
         */
         */
-       nbp = geteblk();
-       nlink = 0;
-       for (i=0, cp = nbp->b_un.b_addr; *cp = (*func)(); i++) {
-               if ((*cp&0377) == ('/'|0200)) {
-                       u.u_error = EPERM;
-                       break;
-               }
-#ifdef notdef
-               if (*cp++&0200 && flag==1 || cp >= nbp->b_un.b_addr+BSIZE) {
-#else
-               cp++;
-               if (cp >= nbp->b_un.b_addr+BSIZE) {
+       if ((dp->i_mode&IFMT) != IFDIR)
+               return (ENOTDIR);
+       if (error = ufs_access(vdp, VEXEC, ndp->ni_cred, p))
+               return (error);
+
+       /*
+        * We now have a segment name to search for, and a directory to search.
+        *
+        * Before tediously performing a linear scan of the directory,
+        * check the name cache to see if the directory/name pair
+        * we are looking for is known already.
+        */
+       if (error = cache_lookup(ndp)) {
+               int vpid;       /* capability number of vnode */
+
+               if (error == ENOENT)
+                       return (error);
+#ifdef PARANOID
+               if (vdp == ndp->ni_rdir && ndp->ni_isdotdot)
+                       panic("ufs_lookup: .. through root");
 #endif
 #endif
-                       u.u_error = ENOENT;
-                       break;
+               /*
+                * Get the next vnode in the path.
+                * See comment below starting `Step through' for
+                * an explaination of the locking protocol.
+                */
+               pdp = dp;
+               dp = VTOI(ndp->ni_vp);
+               vdp = ndp->ni_vp;
+               vpid = vdp->v_id;
+               if (pdp == dp) {
+                       VREF(vdp);
+                       error = 0;
+               } else if (ndp->ni_isdotdot) {
+                       IUNLOCK(pdp);
+                       error = vget(vdp);
+                       if (!error && lockparent && *ndp->ni_next == '\0')
+                               ILOCK(pdp);
+               } else {
+                       error = vget(vdp);
+                       if (!lockparent || error || *ndp->ni_next != '\0')
+                               IUNLOCK(pdp);
                }
                }
+               /*
+                * Check that the capability number did not change
+                * while we were waiting for the lock.
+                */
+               if (!error) {
+                       if (vpid == vdp->v_id)
+                               return (0);
+                       ufs_iput(dp);
+                       if (lockparent && pdp != dp && *ndp->ni_next == '\0')
+                               IUNLOCK(pdp);
+               }
+               ILOCK(pdp);
+               dp = pdp;
+               vdp = ITOV(dp);
+               ndp->ni_vp = NULL;
        }
        }
-       if (u.u_error) {
-               dp = NULL;
-               goto out1;
-       }
-       cp = nbp->b_un.b_addr;
+
        /*
        /*
-        * If name starts with '/' start from
-        * root; otherwise start from current dir.
+        * Suppress search for slots unless creating
+        * file and at end of pathname, in which case
+        * we watch for a place to put the new file in
+        * case it doesn't already exist.
         */
         */
-       dp = u.u_cdir;
-       if (*cp == '/') {
-               while (*cp == '/')
-                       cp++;
-               if ((dp = u.u_rdir) == NULL)
-                       dp = rootdir;
+       slotstatus = FOUND;
+       if ((flag == CREATE || flag == RENAME) && *ndp->ni_next == 0) {
+               slotstatus = NONE;
+               slotfreespace = 0;
+               slotneeded = ((sizeof (struct direct) - (MAXNAMLEN + 1)) +
+                       ((ndp->ni_namelen + 1 + 3) &~ 3));
        }
        }
-       ilock(dp);
-       dp->i_count++;
 
        /*
 
        /*
-        * dp must be a directory and
-        * must have X permission.
-        * cp is a path name relative to that directory.
+        * If there is cached information on a previous search of
+        * this directory, pick up where we last left off.
+        * We cache only lookups as these are the most common
+        * and have the greatest payoff. Caching CREATE has little
+        * benefit as it usually must search the entire directory
+        * to determine that the entry does not exist. Caching the
+        * location of the last DELETE or RENAME has not reduced
+        * profiling time and hence has been removed in the interest
+        * of simplicity.
         */
         */
-
-dirloop:
-       if ((dp->i_mode&IFMT) != IFDIR)
-               u.u_error = ENOTDIR;
-       (void) access(dp, IEXEC);
-       for (i=0; *cp!='\0' && *cp!='/'; i++) {
-#ifdef notdef
-               if (i >= DIRSIZ) {
-                       u.u_error = ENOENT;
-                       break;
-               }
-               u.u_dbuf[i] = *cp++;
-#else
-               if (i < DIRSIZ)
-                       u.u_dbuf[i] = *cp;
-               cp++;
-#endif
-       }
-       if (u.u_error)
-               goto out;
-       u.u_pdir = dp;
-       while (i < DIRSIZ)
-               u.u_dbuf[i++] = '\0';
-       if (u.u_dbuf[0] == '\0') {              /* null name, e.g. "/" or "" */
-               if (flag) {
-                       u.u_error = ENOENT;
-                       goto out;
-               }
-               goto out1;
+       bmask = VFSTOUFS(vdp->v_mount)->um_mountp->mnt_stat.f_iosize - 1;
+       if (flag != LOOKUP || dp->i_diroff == 0 || dp->i_diroff > dp->i_size) {
+               ndp->ni_ufs.ufs_offset = 0;
+               numdirpasses = 1;
+       } else {
+               ndp->ni_ufs.ufs_offset = dp->i_diroff;
+               if ((entryoffsetinblock = ndp->ni_ufs.ufs_offset & bmask) &&
+                   (error = VOP_BLKATOFF(vdp, ndp->ni_ufs.ufs_offset, NULL,
+                    &bp)))
+                       return (error);
+               numdirpasses = 2;
+               nchstats.ncs_2passes++;
        }
        }
-       u.u_segflg = 1;
-       eo = -1;
-       bp = NULL;
+       endsearch = roundup(dp->i_size, DIRBLKSIZ);
+       enduseful = 0;
 
 
-       for (u.u_offset=0; u.u_offset < dp->i_size;
-          u.u_offset += sizeof(struct direct), ep++) {
+searchloop:
+       while (ndp->ni_ufs.ufs_offset < endsearch) {
                /*
                /*
-                * If offset is on a block boundary,
-                * read the next directory block.
-                * Release previous if it exists.
+                * If offset is on a block boundary, read the next directory
+                * block.  Release previous if it exists.
                 */
                 */
-               if ((u.u_offset&BMASK) == 0) {
+               if ((ndp->ni_ufs.ufs_offset & bmask) == 0) {
                        if (bp != NULL)
                                brelse(bp);
                        if (bp != NULL)
                                brelse(bp);
-                       bp = bread(dp->i_dev,
-                               bmap(dp,(daddr_t)(u.u_offset>>BSHIFT), B_READ));
-                       if (bp->b_flags & B_ERROR) {
-                               brelse(bp);
-                               goto out;
-                       }
-                       ep = (struct direct *)bp->b_un.b_addr;
+                       if (error = VOP_BLKATOFF(vdp, ndp->ni_ufs.ufs_offset,
+                           NULL, &bp))
+                               return (error);
+                       entryoffsetinblock = 0;
                }
                /*
                }
                /*
-                * Note first empty directory slot
-                * in eo for possible creat.
-                * String compare the directory entry
-                * and the current component.
+                * If still looking for a slot, and at a DIRBLKSIZE
+                * boundary, have to start looking for free space again.
                 */
                 */
-               if (ep->d_ino == 0) {
-                       if (eo < 0)
-                               eo = u.u_offset;
-                       continue;
+               if (slotstatus == NONE &&
+                   (entryoffsetinblock & (DIRBLKSIZ - 1)) == 0) {
+                       slotoffset = -1;
+                       slotfreespace = 0;
                }
                }
-               if (strncmp(u.u_dbuf, ep->d_name, DIRSIZ) != 0)
-                       continue;
                /*
                /*
-                * Here a component matched in a directory.
-                * If there is more pathname, go back to
-                * dirloop, otherwise return.
+                * Get pointer to next entry.
+                * Full validation checks are slow, so we only check
+                * enough to insure forward progress through the
+                * directory. Complete checks can be run by patching
+                * "dirchk" to be true.
                 */
                 */
-               bcopy((caddr_t)ep, (caddr_t)&u.u_dent, sizeof(struct direct));
-               brelse(bp);
-               if (flag==2 && *cp=='\0') {
-                       if (access(dp, IWRITE))
-                               goto out;
-                       /* should fix unlink */
-                       u.u_offset += sizeof(struct direct);
-                       goto out1;
+               ep = (struct direct *)(bp->b_un.b_addr + entryoffsetinblock);
+               if (ep->d_reclen == 0 ||
+                   dirchk && ufs_dirbadentry(ep, entryoffsetinblock)) {
+                       int i;
+
+                       ufs_dirbad(dp, ndp->ni_ufs.ufs_offset, "mangled entry");
+                       i = DIRBLKSIZ - (entryoffsetinblock & (DIRBLKSIZ - 1));
+                       ndp->ni_ufs.ufs_offset += i;
+                       entryoffsetinblock += i;
+                       continue;
                }
                }
+
                /*
                /*
-                * Special handling for ".."
+                * If an appropriate sized slot has not yet been found,
+                * check to see if one is available. Also accumulate space
+                * in the current block so that we can determine if
+                * compaction is viable.
                 */
                 */
-               if (u.u_dent.d_name[0]=='.' && u.u_dent.d_name[1]=='.' &&
-                   u.u_dent.d_name[2]=='\0') {
-                       if (dp == u.u_rdir)
-                               u.u_dent.d_ino = dp->i_number;
-                       else if (u.u_dent.d_ino==ROOTINO &&
-                          dp->i_number == ROOTINO) {
-                               for(i=1; i<NMOUNT; i++)
-                                       if (mount[i].m_bufp != NULL &&
-                                          mount[i].m_dev == dp->i_dev) {
-                                               iput(dp);
-                                               dp = mount[i].m_inodp;
-                                               ilock(dp);
-                                               dp->i_count++;
-                                               cp -= 2;     /* back over .. */
-                                               goto dirloop;
+               if (slotstatus != FOUND) {
+                       int size = ep->d_reclen;
+
+                       if (ep->d_ino != 0)
+                               size -= DIRSIZ(ep);
+                       if (size > 0) {
+                               if (size >= slotneeded) {
+                                       slotstatus = FOUND;
+                                       slotoffset = ndp->ni_ufs.ufs_offset;
+                                       slotsize = ep->d_reclen;
+                               } else if (slotstatus == NONE) {
+                                       slotfreespace += size;
+                                       if (slotoffset == -1)
+                                               slotoffset =
+                                                     ndp->ni_ufs.ufs_offset;
+                                       if (slotfreespace >= slotneeded) {
+                                               slotstatus = COMPACT;
+                                               slotsize =
+                                                     ndp->ni_ufs.ufs_offset +
+                                                     ep->d_reclen - slotoffset;
                                        }
                                        }
+                               }
                        }
                }
                        }
                }
-               d = dp->i_dev;
-               ino = dp->i_number;
-               irele(dp);
-               pdp = dp;
-               dp = iget(d, u.u_dent.d_ino);
-               if (dp == NULL)  {
-                       iput(pdp);
-                       goto out1;
-               }
+
                /*
                /*
-                * Check for symbolic link
+                * Check for a name match.
                 */
                 */
-               if ((dp->i_mode&IFMT)==IFLNK && (follow || *cp=='/')) {
-                       char *ocp;
-
-                       ocp = cp;
-                       while (*cp++)
-                               ;
-                       if (dp->i_size + (cp-ocp) >= BSIZE-1 || ++nlink>8) {
-                               u.u_error = ELOOP;
-                               iput(pdp);
-                               goto out;
-                       }
-                       bcopy(ocp, nbp->b_un.b_addr+dp->i_size, cp-ocp);
-                       bp = bread(dp->i_dev, bmap(dp, (daddr_t)0, B_READ));
-                       if (bp->b_flags & B_ERROR) {
+               if (ep->d_ino) {
+                       if (ep->d_namlen == ndp->ni_namelen &&
+                           !bcmp(ndp->ni_ptr, ep->d_name,
+                               (unsigned)ep->d_namlen)) {
+                               /*
+                                * Save directory entry's inode number and
+                                * reclen in ndp->ni_ufs area, and release
+                                * directory buffer.
+                                */
+                               ndp->ni_ufs.ufs_ino = ep->d_ino;
+                               ndp->ni_ufs.ufs_reclen = ep->d_reclen;
                                brelse(bp);
                                brelse(bp);
-                               iput(pdp);
-                               goto out;
-                       }
-                       bcopy(bp->b_un.b_addr, nbp->b_un.b_addr, dp->i_size);
-                       brelse(bp);
-                       cp = nbp->b_un.b_addr;
-                       iput(dp);
-                       if (*cp == '/') {
-                               iput(pdp);
-                               while (*cp == '/')
-                                       cp++;
-                               if ((dp = u.u_rdir) == NULL)
-                                       dp = rootdir;
-                               ilock(dp);
-                               dp->i_count++;
-                       } else {
-                               dp = pdp;
-                               ilock(dp);
+                               goto found;
                        }
                        }
-                       goto dirloop;
-               }
-               iput(pdp);
-               if (*cp == '/') {
-                       while (*cp == '/')
-                               cp++;
-                       goto dirloop;
                }
                }
-               goto out1;
+               prevoff = ndp->ni_ufs.ufs_offset;
+               ndp->ni_ufs.ufs_offset += ep->d_reclen;
+               entryoffsetinblock += ep->d_reclen;
+               if (ep->d_ino)
+                       enduseful = ndp->ni_ufs.ufs_offset;
        }
        }
+/* notfound: */
        /*
        /*
-        * Search failed.
+        * If we started in the middle of the directory and failed
+        * to find our target, we must check the beginning as well.
         */
         */
+       if (numdirpasses == 2) {
+               numdirpasses--;
+               ndp->ni_ufs.ufs_offset = 0;
+               endsearch = dp->i_diroff;
+               goto searchloop;
+       }
        if (bp != NULL)
                brelse(bp);
        if (bp != NULL)
                brelse(bp);
-       if (flag==1 && *cp=='\0' && dp->i_nlink) {
-               if (access(dp, IWRITE))
-                       goto out;
-               if (eo>=0)
-                       u.u_offset = eo;
+       /*
+        * If creating, and at end of pathname and current
+        * directory has not been removed, then can consider
+        * allowing file to be created.
+        */
+       if ((flag == CREATE || flag == RENAME) &&
+           *ndp->ni_next == 0 && dp->i_nlink != 0) {
+               /*
+                * Access for write is interpreted as allowing
+                * creation of files in the directory.
+                */
+               if (error = ufs_access(vdp, VWRITE, ndp->ni_cred, p))
+                       return (error);
+               /*
+                * Return an indication of where the new directory
+                * entry should be put.  If we didn't find a slot,
+                * then set ndp->ni_ufs.ufs_count to 0 indicating
+                * that the new slot belongs at the end of the
+                * directory. If we found a slot, then the new entry
+                * can be put in the range from ndp->ni_ufs.ufs_offset
+                * to ndp->ni_ufs.ufs_offset + ndp->ni_ufs.ufs_count.
+                */
+               if (slotstatus == NONE) {
+                       ndp->ni_ufs.ufs_offset = roundup(dp->i_size, DIRBLKSIZ);
+                       ndp->ni_ufs.ufs_count = 0;
+                       enduseful = ndp->ni_ufs.ufs_offset;
+               } else {
+                       ndp->ni_ufs.ufs_offset = slotoffset;
+                       ndp->ni_ufs.ufs_count = slotsize;
+                       if (enduseful < slotoffset + slotsize)
+                               enduseful = slotoffset + slotsize;
+               }
+               ndp->ni_ufs.ufs_endoff = roundup(enduseful, DIRBLKSIZ);
                dp->i_flag |= IUPD|ICHG;
                dp->i_flag |= IUPD|ICHG;
-               dp = NULL;
-               goto out1;
+               /*
+                * We return with the directory locked, so that
+                * the parameters we set up above will still be
+                * valid if we actually decide to do a direnter().
+                * We return ni_vp == NULL to indicate that the entry
+                * does not currently exist; we leave a pointer to
+                * the (locked) directory inode in ndp->ni_dvp.
+                * The pathname buffer is saved so that the name
+                * can be obtained later.
+                *
+                * NB - if the directory is unlocked, then this
+                * information cannot be used.
+                */
+               ndp->ni_nameiop |= SAVENAME;
+               if (!lockparent)
+                       IUNLOCK(dp);
        }
        }
-       u.u_error = ENOENT;
-out:
-       iput(dp);
-       dp = NULL;
-out1:
-       brelse(nbp);
-       return (dp);
+       /*
+        * Insert name into cache (as non-existent) if appropriate.
+        */
+       if (ndp->ni_makeentry && flag != CREATE)
+               cache_enter(ndp);
+       return (ENOENT);
+
+found:
+       if (numdirpasses == 2)
+               nchstats.ncs_pass2++;
+       /*
+        * Check that directory length properly reflects presence
+        * of this entry.
+        */
+       if (entryoffsetinblock + DIRSIZ(ep) > dp->i_size) {
+               ufs_dirbad(dp, ndp->ni_ufs.ufs_offset, "i_size too small");
+               dp->i_size = entryoffsetinblock + DIRSIZ(ep);
+               dp->i_flag |= IUPD|ICHG;
+       }
+
+       /*
+        * Found component in pathname.
+        * If the final component of path name, save information
+        * in the cache as to where the entry was found.
+        */
+       if (*ndp->ni_next == '\0' && flag == LOOKUP)
+               dp->i_diroff = ndp->ni_ufs.ufs_offset &~ (DIRBLKSIZ - 1);
+
+       /*
+        * If deleting, and at end of pathname, return
+        * parameters which can be used to remove file.
+        * If the wantparent flag isn't set, we return only
+        * the directory (in ndp->ni_dvp), otherwise we go
+        * on and lock the inode, being careful with ".".
+        */
+       if (flag == DELETE && *ndp->ni_next == 0) {
+               /*
+                * Write access to directory required to delete files.
+                */
+               if (error = ufs_access(vdp, VWRITE, ndp->ni_cred, p))
+                       return (error);
+               /*
+                * Return pointer to current entry in ndp->ni_ufs.ufs_offset,
+                * and distance past previous entry (if there
+                * is a previous entry in this block) in ndp->ni_ufs.ufs_count.
+                * Save directory inode pointer in ndp->ni_dvp for dirremove().
+                */
+               if ((ndp->ni_ufs.ufs_offset&(DIRBLKSIZ-1)) == 0)
+                       ndp->ni_ufs.ufs_count = 0;
+               else
+                       ndp->ni_ufs.ufs_count =
+                           ndp->ni_ufs.ufs_offset - prevoff;
+               if (dp->i_number == ndp->ni_ufs.ufs_ino) {
+                       VREF(vdp);
+                       ndp->ni_vp = vdp;
+                       return (0);
+               }
+               if (error = VOP_VGET(vdp, ndp->ni_ufs.ufs_ino, &tdp))
+                       return (error);
+               /*
+                * If directory is "sticky", then user must own
+                * the directory, or the file in it, else she
+                * may not delete it (unless she's root). This
+                * implements append-only directories.
+                */
+               if ((dp->i_mode & ISVTX) &&
+                   ndp->ni_cred->cr_uid != 0 &&
+                   ndp->ni_cred->cr_uid != dp->i_uid &&
+                   VTOI(tdp)->i_uid != ndp->ni_cred->cr_uid) {
+                       vput(tdp);
+                       return (EPERM);
+               }
+               ndp->ni_vp = tdp;
+               if (!lockparent)
+                       IUNLOCK(dp);
+               return (0);
+       }
+
+       /*
+        * If rewriting (RENAME), return the inode and the
+        * information required to rewrite the present directory
+        * Must get inode of directory entry to verify it's a
+        * regular file, or empty directory.
+        */
+       if (flag == RENAME && wantparent && *ndp->ni_next == 0) {
+               if (error = ufs_access(vdp, VWRITE, ndp->ni_cred, p))
+                       return (error);
+               /*
+                * Careful about locking second inode.
+                * This can only occur if the target is ".".
+                */
+               if (dp->i_number == ndp->ni_ufs.ufs_ino)
+                       return (EISDIR);
+               if (error = VOP_VGET(vdp, ndp->ni_ufs.ufs_ino, &tdp))
+                       return (error);
+               ndp->ni_vp = tdp;
+               ndp->ni_nameiop |= SAVENAME;
+               if (!lockparent)
+                       IUNLOCK(dp);
+               return (0);
+       }
+
+       /*
+        * Step through the translation in the name.  We do not `iput' the
+        * directory because we may need it again if a symbolic link
+        * is relative to the current directory.  Instead we save it
+        * unlocked as "pdp".  We must get the target inode before unlocking
+        * the directory to insure that the inode will not be removed
+        * before we get it.  We prevent deadlock by always fetching
+        * inodes from the root, moving down the directory tree. Thus
+        * when following backward pointers ".." we must unlock the
+        * parent directory before getting the requested directory.
+        * There is a potential race condition here if both the current
+        * and parent directories are removed before the `iget' for the
+        * inode associated with ".." returns.  We hope that this occurs
+        * infrequently since we cannot avoid this race condition without
+        * implementing a sophisticated deadlock detection algorithm.
+        * Note also that this simple deadlock detection scheme will not
+        * work if the file system has any hard links other than ".."
+        * that point backwards in the directory structure.
+        */
+       pdp = dp;
+       if (ndp->ni_isdotdot) {
+               IUNLOCK(pdp);   /* race to get the inode */
+               if (error = VOP_VGET(vdp, ndp->ni_ufs.ufs_ino, &tdp)) {
+                       ILOCK(pdp);
+                       return (error);
+               }
+               if (lockparent && *ndp->ni_next == '\0')
+                       ILOCK(pdp);
+               ndp->ni_vp = tdp;
+       } else if (dp->i_number == ndp->ni_ufs.ufs_ino) {
+               VREF(vdp);      /* we want ourself, ie "." */
+               ndp->ni_vp = vdp;
+       } else {
+               if (error = VOP_VGET(vdp, ndp->ni_ufs.ufs_ino, &tdp))
+                       return (error);
+               if (!lockparent || *ndp->ni_next != '\0')
+                       IUNLOCK(pdp);
+               ndp->ni_vp = tdp;
+       }
+
+       /*
+        * Insert name into cache if appropriate.
+        */
+       if (ndp->ni_makeentry)
+               cache_enter(ndp);
+       return (0);
+}
+
+void
+ufs_dirbad(ip, offset, how)
+       struct inode *ip;
+       off_t offset;
+       char *how;
+{
+       struct mount *mp;
+
+       mp = ITOV(ip)->v_mount;
+       (void)printf("%s: bad dir ino %d at offset %d: %s\n",
+           mp->mnt_stat.f_mntonname, ip->i_number, offset, how);
+       if ((mp->mnt_stat.f_flags & MNT_RDONLY) == 0)
+               panic("bad dir");
 }
 
 /*
 }
 
 /*
- * Return the next character from the
- * kernel string pointed at by dirp.
+ * Do consistency checking on a directory entry:
+ *     record length must be multiple of 4
+ *     entry must fit in rest of its DIRBLKSIZ block
+ *     record must be large enough to contain entry
+ *     name is not longer than MAXNAMLEN
+ *     name must be as long as advertised, and null terminated
  */
  */
-schar()
+int
+ufs_dirbadentry(ep, entryoffsetinblock)
+       register struct direct *ep;
+       int entryoffsetinblock;
 {
 {
+       register int i;
 
 
-       return (*u.u_dirp++ & 0377);
+       if ((ep->d_reclen & 0x3) != 0 ||
+           ep->d_reclen > DIRBLKSIZ - (entryoffsetinblock & (DIRBLKSIZ - 1)) ||
+           ep->d_reclen < DIRSIZ(ep) || ep->d_namlen > MAXNAMLEN) {
+               /*return (1); */
+               printf("First bad\n");
+               goto bad;
+       }
+       for (i = 0; i < ep->d_namlen; i++)
+               if (ep->d_name[i] == '\0') {
+                       /*return (1); */
+                       printf("Second bad\n");
+                       goto bad;
+       }
+       if (ep->d_name[i])
+               goto bad;
+       return (ep->d_name[i]);
+bad:
+printf("ufs_dirbadentry: jumping out: reclen: %d namlen %d ino %d name %s\n",
+       ep->d_reclen, ep->d_namlen, ep->d_ino, ep->d_name );
+       return(1);
 }
 
 /*
 }
 
 /*
- * Return the next character from the
- * user string pointed at by dirp.
+ * Write a directory entry after a call to namei, using the parameters
+ * that it left in nameidata.  The argument ip is the inode which the new
+ * directory entry will refer to.  The nameidata field ndp->ni_dvp is a
+ * pointer to the directory to be written, which was left locked by namei.
+ * Remaining parameters (ndp->ni_ufs.ufs_offset, ndp->ni_ufs.ufs_count)
+ * indicate how the space for the new entry is to be obtained.
  */
  */
-uchar()
+int
+ufs_direnter(ip, ndp)
+       struct inode *ip;
+       register struct nameidata *ndp;
 {
 {
-       register c;
+       register struct direct *ep, *nep;
+       register struct inode *dp;
+       register struct vnode *dvp;
+       struct buf *bp;
+       struct direct newdir;
+       struct iovec aiov;
+       struct uio auio;
+       u_int dsize;
+       int error, loc, newentrysize, spacefree;
+       char *dirbuf;
+
+#ifdef DIAGNOSTIC
+       if ((ndp->ni_nameiop & SAVENAME) == 0)
+               panic("direnter: missing name");
+#endif
+       dvp = ndp->ni_dvp;
+       dp = VTOI(dvp);
+       newdir.d_ino = ip->i_number;
+       newdir.d_namlen = ndp->ni_namelen;
+       bcopy(ndp->ni_ptr, newdir.d_name, (unsigned)ndp->ni_namelen + 1);
+       newentrysize = DIRSIZ(&newdir);
+       if (ndp->ni_ufs.ufs_count == 0) {
+               /*
+                * If ndp->ni_ufs.ufs_count is 0, then namei could find no
+                * space in the directory. Here, ndp->ni_ufs.ufs_offset will
+                * be on a directory block boundary and we will write the
+                * new entry into a fresh block.
+                */
+               if (ndp->ni_ufs.ufs_offset & (DIRBLKSIZ - 1))
+                       panic("wdir: newblk");
+               auio.uio_offset = ndp->ni_ufs.ufs_offset;
+               newdir.d_reclen = DIRBLKSIZ;
+               auio.uio_resid = newentrysize;
+               aiov.iov_len = newentrysize;
+               aiov.iov_base = (caddr_t)&newdir;
+               auio.uio_iov = &aiov;
+               auio.uio_iovcnt = 1;
+               auio.uio_rw = UIO_WRITE;
+               auio.uio_segflg = UIO_SYSSPACE;
+               auio.uio_procp = (struct proc *)0;
+               error = VOP_WRITE(dvp, &auio, IO_SYNC, ndp->ni_cred);
+               if (DIRBLKSIZ >
+                   VFSTOUFS(dvp->v_mount)->um_mountp->mnt_stat.f_bsize)
+                       /* XXX should grow with balloc() */
+                       panic("ufs_direnter: frag size");
+               else if (!error) {
+                       dp->i_size = roundup(dp->i_size, DIRBLKSIZ);
+                       dp->i_flag |= ICHG;
+               }
+               return (error);
+       }
 
 
-       c = fubyte(u.u_dirp++);
-       if (c == -1) {
-               u.u_error = EFAULT;
-               c = 0;
+       /*
+        * If ndp->ni_ufs.ufs_count is non-zero, then namei found space
+        * for the new entry in the range ndp->ni_ufs.ufs_offset to
+        * ndp->ni_ufs.ufs_offset + ndp->ni_ufs.ufs_count in the directory.
+        * To use this space, we may have to compact the entries located
+        * there, by copying them together towards the beginning of the
+        * block, leaving the free space in one usable chunk at the end.
+        */
+
+       /*
+        * Increase size of directory if entry eats into new space.
+        * This should never push the size past a new multiple of
+        * DIRBLKSIZE.
+        *
+        * N.B. - THIS IS AN ARTIFACT OF 4.2 AND SHOULD NEVER HAPPEN.
+        */
+       if (ndp->ni_ufs.ufs_offset + ndp->ni_ufs.ufs_count > dp->i_size)
+               dp->i_size = ndp->ni_ufs.ufs_offset + ndp->ni_ufs.ufs_count;
+       /*
+        * Get the block containing the space for the new directory entry.
+        */
+       if (error = VOP_BLKATOFF(dvp, ndp->ni_ufs.ufs_offset, &dirbuf, &bp))
+               return (error);
+       /*
+        * Find space for the new entry. In the simple case, the entry at
+        * offset base will have the space. If it does not, then namei
+        * arranged that compacting the region ndp->ni_ufs.ufs_offset to
+        * ndp->ni_ufs.ufs_offset + ndp->ni_ufs.ufs_count would yield the
+        * space.
+        */
+       ep = (struct direct *)dirbuf;
+       dsize = DIRSIZ(ep);
+       spacefree = ep->d_reclen - dsize;
+       for (loc = ep->d_reclen; loc < ndp->ni_ufs.ufs_count; ) {
+               nep = (struct direct *)(dirbuf + loc);
+               if (ep->d_ino) {
+                       /* trim the existing slot */
+                       ep->d_reclen = dsize;
+                       ep = (struct direct *)((char *)ep + dsize);
+               } else {
+                       /* overwrite; nothing there; header is ours */
+                       spacefree += dsize;
+               }
+               dsize = DIRSIZ(nep);
+               spacefree += nep->d_reclen - dsize;
+               loc += nep->d_reclen;
+               bcopy((caddr_t)nep, (caddr_t)ep, dsize);
+       }
+       /*
+        * Update the pointer fields in the previous entry (if any),
+        * copy in the new entry, and write out the block.
+        */
+       if (ep->d_ino == 0) {
+               if (spacefree + dsize < newentrysize)
+                       panic("wdir: compact1");
+               newdir.d_reclen = spacefree + dsize;
+       } else {
+               if (spacefree < newentrysize)
+                       panic("wdir: compact2");
+               newdir.d_reclen = spacefree;
+               ep->d_reclen = dsize;
+               ep = (struct direct *)((char *)ep + dsize);
        }
        }
-       return (c);
+       bcopy((caddr_t)&newdir, (caddr_t)ep, (u_int)newentrysize);
+       error = VOP_BWRITE(bp);
+       dp->i_flag |= IUPD|ICHG;
+       if (!error && ndp->ni_ufs.ufs_endoff &&
+           ndp->ni_ufs.ufs_endoff < dp->i_size)
+               error = VOP_TRUNCATE(dvp, (u_long)ndp->ni_ufs.ufs_endoff,
+                   IO_SYNC);
+       return (error);
 }
 
 }
 
-#ifndef vax
-strncmp(s1, s2, len)
-       register char *s1, *s2;
-       register len;
+/*
+ * Remove a directory entry after a call to namei, using
+ * the parameters which it left in nameidata. The entry
+ * ni_ufs.ufs_offset contains the offset into the directory of the
+ * entry to be eliminated.  The ni_ufs.ufs_count field contains the
+ * size of the previous record in the directory.  If this
+ * is 0, the first entry is being deleted, so we need only
+ * zero the inode number to mark the entry as free.  If the
+ * entry is not the first in the directory, we must reclaim
+ * the space of the now empty record by adding the record size
+ * to the size of the previous entry.
+ */
+int
+ufs_dirremove(ndp)
+       register struct nameidata *ndp;
 {
 {
+       register struct inode *dp;
+       struct direct *ep;
+       struct buf *bp;
+       int error;
+
+       dp = VTOI(ndp->ni_dvp);
+       if (ndp->ni_ufs.ufs_count == 0) {
+               /*
+                * First entry in block: set d_ino to zero.
+                */
+               if (error = VOP_BLKATOFF(ndp->ni_dvp, ndp->ni_ufs.ufs_offset,
+                   (char **)&ep, &bp))
+                       return (error);
+               ep->d_ino = 0;
+               error = VOP_BWRITE(bp);
+               dp->i_flag |= IUPD|ICHG;
+               return (error);
+       }
+       /*
+        * Collapse new free space into previous entry.
+        */
+       if (error = VOP_BLKATOFF(ndp->ni_dvp,
+           ndp->ni_ufs.ufs_offset - ndp->ni_ufs.ufs_count, (char **)&ep, &bp))
+               return (error);
+       ep->d_reclen += ndp->ni_ufs.ufs_reclen;
+       error = VOP_BWRITE(bp);
+       dp->i_flag |= IUPD|ICHG;
+       return (error);
+}
 
 
-       do {
-               if (*s1 != *s2++)
-                       return (1);
-               if (*s1++ == '\0')
+/*
+ * Rewrite an existing directory entry to point at the inode
+ * supplied.  The parameters describing the directory entry are
+ * set up by a call to namei.
+ */
+int
+ufs_dirrewrite(dp, ip, ndp)
+       struct inode *dp, *ip;
+       struct nameidata *ndp;
+{
+       struct buf *bp;
+       struct direct *ep;
+       int error;
+
+       if (error = VOP_BLKATOFF(ITOV(dp), ndp->ni_ufs.ufs_offset,
+           (char **)&ep, &bp))
+               return (error);
+       ep->d_ino = ip->i_number;
+       error = VOP_BWRITE(bp);
+       dp->i_flag |= IUPD|ICHG;
+       return (error);
+}
+
+/*
+ * Check if a directory is empty or not.
+ * Inode supplied must be locked.
+ *
+ * Using a struct dirtemplate here is not precisely
+ * what we want, but better than using a struct direct.
+ *
+ * NB: does not handle corrupted directories.
+ */
+int
+ufs_dirempty(ip, parentino, cred)
+       register struct inode *ip;
+       ino_t parentino;
+       struct ucred *cred;
+{
+       register off_t off;
+       struct dirtemplate dbuf;
+       register struct direct *dp = (struct direct *)&dbuf;
+       int error, count;
+#define        MINDIRSIZ (sizeof (struct dirtemplate) / 2)
+
+       for (off = 0; off < ip->i_size; off += dp->d_reclen) {
+               error = vn_rdwr(UIO_READ, ITOV(ip), (caddr_t)dp, MINDIRSIZ, off,
+                  UIO_SYSSPACE, IO_NODELOCKED, cred, &count, (struct proc *)0);
+               /*
+                * Since we read MINDIRSIZ, residual must
+                * be 0 unless we're at end of file.
+                */
+               if (error || count != 0)
                        return (0);
                        return (0);
-       } while (--len);
-       return (0);
+               /* avoid infinite loops */
+               if (dp->d_reclen == 0)
+                       return (0);
+               /* skip empty entries */
+               if (dp->d_ino == 0)
+                       continue;
+               /* accept only "." and ".." */
+               if (dp->d_namlen > 2)
+                       return (0);
+               if (dp->d_name[0] != '.')
+                       return (0);
+               /*
+                * At this point d_namlen must be 1 or 2.
+                * 1 implies ".", 2 implies ".." if second
+                * char is also "."
+                */
+               if (dp->d_namlen == 1)
+                       continue;
+               if (dp->d_name[1] == '.' && dp->d_ino == parentino)
+                       continue;
+               return (0);
+       }
+       return (1);
+}
+
+/*
+ * Check if source directory is in the path of the target directory.
+ * Target is supplied locked, source is unlocked.
+ * The target is always iput before returning.
+ */
+int
+ufs_checkpath(source, target, cred)
+       struct inode *source, *target;
+       struct ucred *cred;
+{
+       struct dirtemplate dirbuf;
+       register struct inode *ip;
+       struct vnode *vp;
+       int error, rootino;
+
+       ip = target;
+       if (ip->i_number == source->i_number) {
+               error = EEXIST;
+               goto out;
+       }
+       rootino = ROOTINO;
+       error = 0;
+       if (ip->i_number == rootino)
+               goto out;
+
+       for (;;) {
+               if ((ip->i_mode&IFMT) != IFDIR) {
+                       error = ENOTDIR;
+                       break;
+               }
+               vp = ITOV(ip);
+               error = vn_rdwr(UIO_READ, vp, (caddr_t)&dirbuf,
+                       sizeof (struct dirtemplate), (off_t)0, UIO_SYSSPACE,
+                       IO_NODELOCKED, cred, (int *)0, (struct proc *)0);
+               if (error != 0)
+                       break;
+               if (dirbuf.dotdot_namlen != 2 ||
+                   dirbuf.dotdot_name[0] != '.' ||
+                   dirbuf.dotdot_name[1] != '.') {
+                       error = ENOTDIR;
+                       break;
+               }
+               if (dirbuf.dotdot_ino == source->i_number) {
+                       error = EINVAL;
+                       break;
+               }
+               if (dirbuf.dotdot_ino == rootino)
+                       break;
+               ufs_iput(ip);
+               if (error = VOP_VGET(vp, dirbuf.dotdot_ino, &vp))
+                       break;
+               ip = VTOI(vp);
+       }
+
+out:
+       if (error == ENOTDIR)
+               printf("checkpath: .. not a directory\n");
+       if (ip != NULL)
+               ufs_iput(ip);
+       return (error);
 }
 }
-#endif