]> git.saurik.com Git - apple/xnu.git/blobdiff - bsd/vfs/vfs_cache.c
xnu-1699.22.73.tar.gz
[apple/xnu.git] / bsd / vfs / vfs_cache.c
index 8cb282de6a33eb09f5009eada93f0f717cb429c7..3096d12941242a58deaa997d4d735b7d6af68b37 100644 (file)
@@ -1,23 +1,29 @@
 /*
- * Copyright (c) 2000-2003 Apple Computer, Inc. All rights reserved.
+ * Copyright (c) 2000-2008 Apple Inc. All rights reserved.
  *
- * @APPLE_LICENSE_HEADER_START@
+ * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
  * 
- * The contents of this file constitute Original Code as defined in and
- * are subject to the Apple Public Source License Version 1.1 (the
- * "License").  You may not use this file except in compliance with the
- * License.  Please obtain a copy of the License at
- * http://www.apple.com/publicsource and read it before using this file.
+ * This file contains Original Code and/or Modifications of Original Code
+ * as defined in and that are subject to the Apple Public Source License
+ * Version 2.0 (the 'License'). You may not use this file except in
+ * compliance with the License. The rights granted to you under the License
+ * may not be used to create, or enable the creation or redistribution of,
+ * unlawful or unlicensed copies of an Apple operating system, or to
+ * circumvent, violate, or enable the circumvention or violation of, any
+ * terms of an Apple operating system software license agreement.
  * 
- * This Original Code and all software distributed under the License are
- * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
+ * Please obtain a copy of the License at
+ * http://www.opensource.apple.com/apsl/ and read it before using this file.
+ * 
+ * The Original Code and all software distributed under the License are
+ * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT.  Please see the
- * License for the specific language governing rights and limitations
- * under the License.
+ * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
+ * Please see the License for the specific language governing rights and
+ * limitations under the License.
  * 
- * @APPLE_LICENSE_HEADER_END@
+ * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
  */
 /* Copyright (c) 1995 NeXT Computer, Inc. All Rights Reserved */
 /*
  *
  *     @(#)vfs_cache.c 8.5 (Berkeley) 3/22/95
  */
+/*
+ * NOTICE: This file was modified by SPARTA, Inc. in 2005 to introduce
+ * support for mandatory and extensible security protections.  This notice
+ * is included in support of clause 2.2 (b) of the Apple Public License,
+ * Version 2.0.
+ */
 #include <sys/param.h>
 #include <sys/systm.h>
 #include <sys/time.h>
 #include <sys/malloc.h>
 #include <sys/kauth.h>
 #include <sys/user.h>
+#include <sys/paths.h>
+
+#if CONFIG_MACF
+#include <security/mac_framework.h>
+#endif
 
 /*
  * Name caching works as follows:
@@ -96,24 +113,64 @@ u_long     nchash;                         /* size of hash table - 1 */
 long   numcache;                       /* number of cache entries allocated */
 int    desiredNodes;
 int    desiredNegNodes;
+int    ncs_negtotal;
+int    nc_disabled = 0;
 TAILQ_HEAD(, namecache) nchead;                /* chain of all name cache entries */
 TAILQ_HEAD(, namecache) neghead;       /* chain of only negative cache entries */
+
+
+#if COLLECT_STATS
+
 struct nchstats nchstats;              /* cache effectiveness statistics */
 
+#define        NCHSTAT(v) {            \
+        nchstats.v++;          \
+}
+#define NAME_CACHE_LOCK()              name_cache_lock()
+#define NAME_CACHE_UNLOCK()            name_cache_unlock()
+#define        NAME_CACHE_LOCK_SHARED()        name_cache_lock()
+
+#else
+
+#define NCHSTAT(v)
+#define NAME_CACHE_LOCK()              name_cache_lock()
+#define NAME_CACHE_UNLOCK()            name_cache_unlock()
+#define        NAME_CACHE_LOCK_SHARED()        name_cache_lock_shared()
+
+#endif
+
+
 /* vars for name cache list lock */
 lck_grp_t * namecache_lck_grp;
 lck_grp_attr_t * namecache_lck_grp_attr;
 lck_attr_t * namecache_lck_attr;
-lck_mtx_t * namecache_mtx_lock;
+
+lck_grp_t * strcache_lck_grp;
+lck_grp_attr_t * strcache_lck_grp_attr;
+lck_attr_t * strcache_lck_attr;
+
+lck_rw_t  * namecache_rw_lock;
+lck_rw_t  * strtable_rw_lock;
+
+#define NUM_STRCACHE_LOCKS 1024
+
+lck_mtx_t strcache_mtx_locks[NUM_STRCACHE_LOCKS];
+
 
 static vnode_t cache_lookup_locked(vnode_t dvp, struct componentname *cnp);
-static int  remove_name_locked(const char *);
-static char *add_name_locked(const char *, size_t, u_int, u_int);
-static void init_string_table(void);
+static const char *add_name_internal(const char *, uint32_t, u_int, boolean_t, u_int);
+static void init_string_table(void) __attribute__((section("__TEXT, initcode")));
 static void cache_delete(struct namecache *, int);
-static void dump_string_table(void);
+static void cache_enter_locked(vnode_t dvp, vnode_t vp, struct componentname *cnp, const char *strname);
+
+#ifdef DUMP_STRING_TABLE
+/*
+ * Internal dump function used for debugging
+ */
+void dump_string_table(void);
+#endif /* DUMP_STRING_TABLE */
 
-static void init_crc32(void);
+static void init_crc32(void) __attribute__((section("__TEXT, initcode")));
 static unsigned int crc32tab[256];
 
 
@@ -122,112 +179,320 @@ static unsigned int crc32tab[256];
 
 
 
-//
-// This function builds the path to a filename in "buff".  The
-// length of the buffer *INCLUDING* the trailing zero byte is
-// returned in outlen.  NOTE: the length includes the trailing
-// zero byte and thus the length is one greater than what strlen
-// would return.  This is important and lots of code elsewhere
-// in the kernel assumes this behavior.
-//
+/*
+ * This function builds the path to a filename in "buff".  The
+ * length of the buffer *INCLUDING* the trailing zero byte is
+ * returned in outlen.  NOTE: the length includes the trailing
+ * zero byte and thus the length is one greater than what strlen
+ * would return.  This is important and lots of code elsewhere
+ * in the kernel assumes this behavior.
+ * 
+ * This function can call vnop in file system if the parent vnode 
+ * does not exist or when called for hardlinks via volfs path.  
+ * If BUILDPATH_NO_FS_ENTER is set in flags, it only uses values present
+ * in the name cache and does not enter the file system.
+ *
+ * passed in vp must have a valid io_count reference
+ */
 int
-build_path(vnode_t first_vp, char *buff, int buflen, int *outlen)
+build_path(vnode_t first_vp, char *buff, int buflen, int *outlen, int flags, vfs_context_t ctx)
 {
-        vnode_t vp = first_vp;
-       char *end, *str;
-       int   len, ret=0, counter=0;
+        vnode_t vp, tvp;
+       vnode_t vp_with_iocount;
+        vnode_t proc_root_dir_vp;
+       char *end;
+       const char *str;
+       int  len;
+       int  ret = 0;
+       int  fixhardlink;
+
+       if (first_vp == NULLVP)
+               return (EINVAL);
+
+       /*
+        * Grab the process fd so we can evaluate fd_rdir.
+        */
+       if (vfs_context_proc(ctx)->p_fd)
+               proc_root_dir_vp = vfs_context_proc(ctx)->p_fd->fd_rdir;
+       else
+               proc_root_dir_vp = NULL;
+
+       vp_with_iocount = NULLVP;
+again:
+       vp = first_vp;
 
        end = &buff[buflen-1];
        *end = '\0';
 
        /*
-        * if this is the root dir of a file system...
+        * holding the NAME_CACHE_LOCK in shared mode is
+        * sufficient to stabilize both the vp->v_parent chain
+        * and the 'vp->v_mount->mnt_vnodecovered' chain
+        *
+        * if we need to drop this lock, we must first grab the v_id
+        * from the vnode we're currently working with... if that
+        * vnode doesn't already have an io_count reference (the vp
+        * passed in comes with one), we must grab a reference
+        * after we drop the NAME_CACHE_LOCK via vnode_getwithvid...
+        * deadlocks may result if you call vnode_get while holding
+        * the NAME_CACHE_LOCK... we lazily release the reference
+        * we pick up the next time we encounter a need to drop 
+        * the NAME_CACHE_LOCK or before we return from this routine
         */
-       if (vp && (vp->v_flag & VROOT) && vp->v_mount) {
-               /*
-                * then if it's the root fs, just put in a '/' and get out of here
-                */
-               if (vp->v_mount->mnt_flag & MNT_ROOTFS) {
+       NAME_CACHE_LOCK_SHARED();
+
+       /*
+        * Check if this is the root of a file system.
+        */
+       while (vp && vp->v_flag & VROOT) {
+               if (vp->v_mount == NULL) {
+                       ret = EINVAL;
+                       goto out_unlock;
+               }
+               if ((vp->v_mount->mnt_flag & MNT_ROOTFS) || (vp == proc_root_dir_vp)) {
+                       /*
+                        * It's the root of the root file system, so it's
+                        * just "/".
+                        */
                        *--end = '/';
-                       goto out;
+
+                       goto out_unlock;
                } else {
-                       /*
-                        * else just use the covered vnode to get the mount path
-                        */
                        vp = vp->v_mount->mnt_vnodecovered;
                }
        }
-       name_cache_lock();
 
-       while (vp && vp->v_parent != vp) {
-               /*
-                * the maximum depth of a file system hierarchy is MAXPATHLEN/2
-                * (with single-char names separated by slashes).  we panic if
-                * we've ever looped more than that.
-                */
-               if (counter++ > MAXPATHLEN/2) {
-                       panic("build_path: vnode parent chain is too long! vp 0x%x\n", vp);
-               }
-               str = vp->v_name;
-
-               if (str == NULL) {
-                       if (vp->v_parent != NULL) {
-                               ret = EINVAL;
-                       }
-                       break;
-               }
-               len = strlen(str);
+       while ((vp != NULLVP) && (vp->v_parent != vp)) {
+               int  vid;
 
                /*
-                * check that there's enough space (make sure to include space for the '/')
+                * For hardlinks the v_name may be stale, so if its OK
+                * to enter a file system, ask the file system for the
+                * name and parent (below).
                 */
-               if ((end - buff) < (len + 1)) {
-                       ret = ENOSPC;
-                       break;
+               fixhardlink = (vp->v_flag & VISHARDLINK) &&
+                             (vp->v_mount->mnt_kern_flag & MNTK_PATH_FROM_ID) &&
+                             !(flags & BUILDPATH_NO_FS_ENTER);
+
+               if (!fixhardlink) {
+                       str = vp->v_name;
+
+                       if (str == NULL || *str == '\0') {
+                               if (vp->v_parent != NULL)
+                                       ret = EINVAL;
+                               else
+                                       ret = ENOENT;
+                               goto out_unlock;
+                       }
+                       len = strlen(str);
+                       /*
+                        * Check that there's enough space (including space for the '/')
+                        */
+                       if ((end - buff) < (len + 1)) {
+                               ret = ENOSPC;
+                               goto out_unlock;
+                       }
+                       /*
+                        * Copy the name backwards.
+                        */
+                       str += len;
+       
+                       for (; len > 0; len--)
+                              *--end = *--str;
+                       /*
+                        * Add a path separator.
+                        */
+                       *--end = '/';
                }
-               /*
-                * copy it backwards
-                */
-               str += len;
 
-               for (; len > 0; len--) {
-                      *--end = *--str;
-               }
                /*
-                * put in the path separator
+                * Walk up the parent chain.
                 */
-               *--end = '/';
+               if (((vp->v_parent != NULLVP) && !fixhardlink) ||
+                   (flags & BUILDPATH_NO_FS_ENTER)) {
+                       /*
+                        * In this if () block we are not allowed to enter the filesystem
+                        * to conclusively get the most accurate parent identifier.
+                        * As a result, if 'vp' does not identify '/' and it
+                        * does not have a valid v_parent, then error out
+                        * and disallow further path construction
+                        */
+                       if ((vp->v_parent == NULLVP) && (rootvnode != vp)) {
+                               /* Only '/' is allowed to have a NULL parent pointer */
+                               ret = EINVAL;
 
-               /*
-                * walk up the chain (as long as we're not the root)  
-                */
-               if (vp == first_vp && (vp->v_flag & VROOT)) {
-                       if (vp->v_mount && vp->v_mount->mnt_vnodecovered) {
-                               vp = vp->v_mount->mnt_vnodecovered->v_parent;
-                       } else {
-                               vp = NULLVP;
+                               /* The code below will exit early if 'tvp = vp' == NULL */
                        }
+
+                       vp = vp->v_parent;
+                       
+                       /*
+                        * if the vnode we have in hand isn't a directory and it
+                        * has a v_parent, then we started with the resource fork
+                        * so skip up to avoid getting a duplicate copy of the
+                        * file name in the path.
+                        */
+                       if (vp && !vnode_isdir(vp) && vp->v_parent)
+                               vp = vp->v_parent;
                } else {
-                       vp = vp->v_parent;
+                       /*
+                        * No parent, go get it if supported.
+                        */
+                       struct vnode_attr  va;
+                       vnode_t  dvp;
+
+                       /*
+                        * Make sure file system supports obtaining a path from id.
+                        */
+                       if (!(vp->v_mount->mnt_kern_flag & MNTK_PATH_FROM_ID)) {
+                               ret = ENOENT;
+                               goto out_unlock;
+                       }
+                       vid = vp->v_id;
+
+                       NAME_CACHE_UNLOCK();
+
+                       if (vp != first_vp && vp != vp_with_iocount) {
+                               if (vp_with_iocount) {
+                                       vnode_put(vp_with_iocount);
+                                       vp_with_iocount = NULLVP;
+                               }
+                               if (vnode_getwithvid(vp, vid))
+                                       goto again;
+                               vp_with_iocount = vp;
+                       }
+                       VATTR_INIT(&va);
+                       VATTR_WANTED(&va, va_parentid);
+
+                       if (fixhardlink) {
+                               VATTR_WANTED(&va, va_name);
+                               MALLOC_ZONE(va.va_name, caddr_t, MAXPATHLEN, M_NAMEI, M_WAITOK);
+                       } else {
+                               va.va_name = NULL;
+                       }
+                       /*
+                        * Ask the file system for its parent id and for its name (optional).
+                        */
+                       ret = vnode_getattr(vp, &va, ctx);
+
+                       if (fixhardlink) {
+                               if ((ret == 0) && (VATTR_IS_SUPPORTED(&va, va_name))) {
+                                       str = va.va_name;
+                                       vnode_update_identity(vp, NULL, str, strlen(str), 0, VNODE_UPDATE_NAME);
+                               } else if (vp->v_name) {
+                                       str = vp->v_name;
+                                       ret = 0;
+                               } else {
+                                       ret = ENOENT;
+                                       goto bad_news;
+                               }
+                               len = strlen(str);
+
+                               /*
+                                * Check that there's enough space.
+                                */
+                               if ((end - buff) < (len + 1)) {
+                                       ret = ENOSPC;
+                               } else {
+                                       /* Copy the name backwards. */
+                                       str += len;
+
+                                       for (; len > 0; len--) {
+                                               *--end = *--str;
+                                       }
+                                       /*
+                                        * Add a path separator.
+                                        */
+                                       *--end = '/';
+                               }
+bad_news:
+                               FREE_ZONE(va.va_name, MAXPATHLEN, M_NAMEI);
+                       }
+                       if (ret || !VATTR_IS_SUPPORTED(&va, va_parentid)) {
+                               ret = ENOENT;
+                               goto out;
+                       }
+                       /*
+                        * Ask the file system for the parent vnode.
+                        */
+                       if ((ret = VFS_VGET(vp->v_mount, (ino64_t)va.va_parentid, &dvp, ctx)))
+                               goto out;
+
+                       if (!fixhardlink && (vp->v_parent != dvp))
+                               vnode_update_identity(vp, dvp, NULL, 0, 0, VNODE_UPDATE_PARENT);
+
+                       if (vp_with_iocount)
+                               vnode_put(vp_with_iocount);
+                       vp = dvp;
+                       vp_with_iocount = vp;
+
+                       NAME_CACHE_LOCK_SHARED();
+
+                       /*
+                        * if the vnode we have in hand isn't a directory and it
+                        * has a v_parent, then we started with the resource fork
+                        * so skip up to avoid getting a duplicate copy of the
+                        * file name in the path.
+                        */
+                       if (vp && !vnode_isdir(vp) && vp->v_parent)
+                               vp = vp->v_parent;
                }
                /*
-                * check if we're crossing a mount point and
-                * switch the vp if we are.
+                * When a mount point is crossed switch the vp.
+                * Continue until we find the root or we find
+                * a vnode that's not the root of a mounted
+                * file system.
                 */
-               if (vp && (vp->v_flag & VROOT) && vp->v_mount) {
-                       vp = vp->v_mount->mnt_vnodecovered;
+               tvp = vp;
+
+               while (tvp) {
+                       if (tvp == proc_root_dir_vp)
+                               goto out_unlock;        /* encountered the root */
+
+                       if (!(tvp->v_flag & VROOT) || !tvp->v_mount)
+                               break;                  /* not the root of a mounted FS */
+                       tvp = tvp->v_mount->mnt_vnodecovered;
+               }
+               if (tvp == NULLVP)
+                       goto out_unlock;
+               vp = tvp;
+
+               if (vp && (flags & BUILDPATH_CHECKACCESS)) {
+                       vid = vp->v_id;
+
+                       NAME_CACHE_UNLOCK();
+
+                       if (vp != first_vp && vp != vp_with_iocount) {
+                               if (vp_with_iocount) {
+                                       vnode_put(vp_with_iocount);
+                                       vp_with_iocount = NULLVP;
+                               }
+                               if (vnode_getwithvid(vp, vid))
+                                       goto again;
+                               vp_with_iocount = vp;
+                       }
+                       if ((ret = vnode_authorize(vp, NULL, KAUTH_VNODE_SEARCH, ctx)))
+                               goto out;       /* no peeking */
+
+                       NAME_CACHE_LOCK_SHARED();
                }
        }
-       name_cache_unlock();
+out_unlock:
+       NAME_CACHE_UNLOCK();
 out:
+       if (vp_with_iocount)
+               vnode_put(vp_with_iocount);
        /*
-        * slide it down to the beginning of the buffer
+        * Slide the name down to the beginning of the buffer.
         */
        memmove(buff, end, &buff[buflen] - end);
-    
-       *outlen = &buff[buflen] - end;  // length includes the trailing zero byte
+
+       /*
+        * length includes the trailing zero byte
+        */
+       *outlen = &buff[buflen] - end;
  
-       return ret;
+       return (ret);
 }
 
 
@@ -242,7 +507,7 @@ vnode_getparent(vnode_t vp)
         vnode_t pvp = NULLVP;
        int     pvid;
 
-        name_cache_lock();
+       NAME_CACHE_LOCK_SHARED();
        /*
         * v_parent is stable behind the name_cache lock
         * however, the only thing we can really guarantee
@@ -253,38 +518,33 @@ vnode_getparent(vnode_t vp)
        if ( (pvp = vp->v_parent) != NULLVP ) {
                pvid = pvp->v_id;
 
-               name_cache_unlock();
+               NAME_CACHE_UNLOCK();
 
                if (vnode_getwithvid(pvp, pvid) != 0)
                        pvp = NULL;
        } else
-               name_cache_unlock();
-
+               NAME_CACHE_UNLOCK();
        return (pvp);
 }
 
-char *
+const char *
 vnode_getname(vnode_t vp)
 {
-        char *name = NULL;
-
-        name_cache_lock();
+        const char *name = NULL;
+       
+       NAME_CACHE_LOCK_SHARED();
        
        if (vp->v_name)
-               name = add_name_locked(vp->v_name, strlen(vp->v_name), 0, 0);
-       name_cache_unlock();
+               name = vfs_addname(vp->v_name, strlen(vp->v_name), 0, 0);
+       NAME_CACHE_UNLOCK();
 
        return (name);
 }
 
 void
-vnode_putname(char *name)
+vnode_putname(const char *name)
 {
-       name_cache_lock();
-
-       remove_name_locked(name);
-
-       name_cache_unlock();
+       vfs_removename(name);
 }
 
 
@@ -302,52 +562,115 @@ vnode_putname(char *name)
  * if VNODE_UPDATE_CACHE, flush the name cache entries associated with vp
  */
 void
-vnode_update_identity(vnode_t vp, vnode_t dvp, char *name, int name_len, int name_hashval, int flags)
+vnode_update_identity(vnode_t vp, vnode_t dvp, const char *name, int name_len, uint32_t name_hashval, int flags)
 {
        struct  namecache *ncp;
         vnode_t        old_parentvp = NULLVP;
-
+#if NAMEDSTREAMS
+       int isstream = (vp->v_flag & VISNAMEDSTREAM);
+       int kusecountbumped = 0;
+#endif
+       kauth_cred_t tcred = NULL;
+       const char *vname = NULL;
+       const char *tname = NULL;
 
        if (flags & VNODE_UPDATE_PARENT) {
-               if (dvp && vnode_ref(dvp) != 0)
-                       dvp = NULLVP;
-       } else
+               if (dvp && vnode_ref(dvp) != 0) {
+                       dvp = NULLVP;
+               }
+#if NAMEDSTREAMS
+               /* Don't count a stream's parent ref during unmounts */
+               if (isstream && dvp && (dvp != vp) && (dvp != vp->v_parent) && (dvp->v_type == VREG)) {
+                       vnode_lock_spin(dvp);
+                       ++dvp->v_kusecount;
+                       kusecountbumped = 1;
+                       vnode_unlock(dvp);
+               }
+#endif
+       } else {
                dvp = NULLVP;
-       name_cache_lock();
+       }
+       if ( (flags & VNODE_UPDATE_NAME) ) {
+               if (name != vp->v_name) {
+                       if (name && *name) {
+                               if (name_len == 0)
+                                       name_len = strlen(name);
+                               tname = vfs_addname(name, name_len, name_hashval, 0);
+                       }
+               } else
+                       flags &= ~VNODE_UPDATE_NAME;
+       }
+       if ( (flags & (VNODE_UPDATE_PURGE | VNODE_UPDATE_PARENT | VNODE_UPDATE_CACHE | VNODE_UPDATE_NAME)) ) {
+
+               NAME_CACHE_LOCK();
 
-       if ( (flags & VNODE_UPDATE_NAME) && (name != vp->v_name) ) {
-               if (vp->v_name != NULL) {
-                       remove_name_locked(vp->v_name);
-                       vp->v_name = NULL;
+               if ( (flags & VNODE_UPDATE_PURGE) ) {
+
+                       if (vp->v_parent)
+                               vp->v_parent->v_nc_generation++;
+
+                       while ( (ncp = LIST_FIRST(&vp->v_nclinks)) )
+                               cache_delete(ncp, 1);
+
+                       while ( (ncp = LIST_FIRST(&vp->v_ncchildren)) )
+                               cache_delete(ncp, 1);
+
+                       /*
+                        * Use a temp variable to avoid kauth_cred_unref() while NAME_CACHE_LOCK is held
+                        */
+                       tcred = vp->v_cred;
+                       vp->v_cred = NOCRED;
+                       vp->v_authorized_actions = 0;
                }
-               if (name && *name) {
-                       if (name_len == 0)
-                               name_len = strlen(name);
-                       vp->v_name = add_name_locked(name, name_len, name_hashval, 0);
+               if ( (flags & VNODE_UPDATE_NAME) ) {
+                       vname = vp->v_name;
+                       vp->v_name = tname;
                }
-       }
-       if (flags & VNODE_UPDATE_PARENT) {
-               if (dvp != vp && dvp != vp->v_parent) {
-                       old_parentvp = vp->v_parent;
-                       vp->v_parent = dvp;
-                       dvp = NULLVP;
+               if (flags & VNODE_UPDATE_PARENT) {
+                       if (dvp != vp && dvp != vp->v_parent) {
+                               old_parentvp = vp->v_parent;
+                               vp->v_parent = dvp;
+                               dvp = NULLVP;
 
-                       if (old_parentvp)
-                               flags |= VNODE_UPDATE_CACHE;
+                               if (old_parentvp)
+                                       flags |= VNODE_UPDATE_CACHE;
+                       }
                }
-       }
-       if (flags & VNODE_UPDATE_CACHE) {
-               while ( (ncp = LIST_FIRST(&vp->v_nclinks)) )
-                       cache_delete(ncp, 1);
-       }
-       name_cache_unlock();
+               if (flags & VNODE_UPDATE_CACHE) {
+                       while ( (ncp = LIST_FIRST(&vp->v_nclinks)) )
+                               cache_delete(ncp, 1);
+               }
+               NAME_CACHE_UNLOCK();
        
-       if (dvp != NULLVP)
+               if (vname != NULL)
+                       vfs_removename(vname);
+
+               if (IS_VALID_CRED(tcred))
+                       kauth_cred_unref(&tcred);
+       }
+       if (dvp != NULLVP) {
+#if NAMEDSTREAMS
+               /* Back-out the ref we took if we lost a race for vp->v_parent. */
+               if (kusecountbumped) {
+                       vnode_lock_spin(dvp);
+                       if (dvp->v_kusecount > 0)
+                               --dvp->v_kusecount;  
+                       vnode_unlock(dvp);
+               }
+#endif
                vnode_rele(dvp);
-       
+       }
        if (old_parentvp) {
                struct  uthread *ut;
 
+#if NAMEDSTREAMS
+               if (isstream) {
+                       vnode_lock_spin(old_parentvp);
+                       if ((old_parentvp->v_type != VDIR) && (old_parentvp->v_kusecount > 0))
+                               --old_parentvp->v_kusecount;
+                       vnode_unlock(old_parentvp);
+               }
+#endif
                ut = get_bsdthread_info(current_thread());
 
                /*
@@ -362,8 +685,7 @@ vnode_update_identity(vnode_t vp, vnode_t dvp, char *name, int name_len, int nam
 
                while ( (vp = old_parentvp) != NULLVP ) {
          
-                       vnode_lock(vp);
-
+                       vnode_lock_spin(vp);
                        vnode_rele_internal(vp, 0, 0, 1);
 
                        /*
@@ -377,7 +699,7 @@ vnode_update_identity(vnode_t vp, vnode_t dvp, char *name, int name_len, int nam
                         * we'll sit in this loop until we run into
                         * a parent in this chain that is not in this state
                         *
-                        * make our check and the node_rele atomic
+                        * make our check and the vnode_rele atomic
                         * with respect to the current vnode we're working on
                         * by holding the vnode lock
                         * if vnode_rele deferred the vnode_reclaim and has put
@@ -392,11 +714,17 @@ vnode_update_identity(vnode_t vp, vnode_t dvp, char *name, int name_len, int nam
                                 * pull the parent pointer now so that when we do the
                                 * vnode_reclaim for each of the vnodes in the uu_vreclaims
                                 * list, we won't recurse back through here
+                                *
+                                * need to do a convert here in case vnode_rele_internal
+                                * returns with the lock held in the spin mode... it 
+                                * can drop and retake the lock under certain circumstances
                                 */
-                               name_cache_lock();
+                               vnode_lock_convert(vp);
+
+                               NAME_CACHE_LOCK();
                                old_parentvp = vp->v_parent;
                                vp->v_parent = NULLVP;
-                               name_cache_unlock();
+                               NAME_CACHE_UNLOCK();
                        } else {
                                /*
                                 * we're done... we ran into a vnode that isn't
@@ -430,9 +758,9 @@ vnode_update_identity(vnode_t vp, vnode_t dvp, char *name, int name_len, int nam
  * so that HFS can post-process the lookup.  Also, volfs will call
  * VNOP_GETATTR2 to determine the parent, instead of using v_parent.
  */
-void vnode_set_hard_link(vnode_t vp)
+void vnode_setmultipath(vnode_t vp)
 {
-       vnode_lock(vp);
+       vnode_lock_spin(vp);
 
        /*
         * In theory, we're changing the vnode's identity as far as the
@@ -451,137 +779,211 @@ void vnode_set_hard_link(vnode_t vp)
 }
 
 
+
+/*
+ * backwards compatibility
+ */
 void vnode_uncache_credentials(vnode_t vp)
 {
-        kauth_cred_t ucred = NULL;
+        vnode_uncache_authorized_action(vp, KAUTH_INVALIDATE_CACHED_RIGHTS);
+}
 
-        if (vp->v_cred) {
-               vnode_lock(vp);
 
-               ucred = vp->v_cred;
-               vp->v_cred = NULL;
+/*
+ * use the exclusive form of NAME_CACHE_LOCK to protect the update of the
+ * following fields in the vnode: v_cred_timestamp, v_cred, v_authorized_actions
+ * we use this lock so that we can look at the v_cred and v_authorized_actions
+ * atomically while behind the NAME_CACHE_LOCK in shared mode in 'cache_lookup_path',
+ * which is the super-hot path... if we are updating the authorized actions for this
+ * vnode, we are already in the super-slow and far less frequented path so its not
+ * that bad that we take the lock exclusive for this case... of course we strive
+ * to hold it for the minimum amount of time possible
+ */
 
-               vnode_unlock(vp);
+void vnode_uncache_authorized_action(vnode_t vp, kauth_action_t action)
+{
+        kauth_cred_t tcred = NOCRED;
 
-               if (ucred)
-                       kauth_cred_rele(ucred);
+       NAME_CACHE_LOCK();
+
+       vp->v_authorized_actions &= ~action;
+
+       if (action == KAUTH_INVALIDATE_CACHED_RIGHTS &&
+           IS_VALID_CRED(vp->v_cred)) {
+               /*
+                * Use a temp variable to avoid kauth_cred_unref() while NAME_CACHE_LOCK is held
+                */
+               tcred = vp->v_cred;
+               vp->v_cred = NOCRED;
        }
+       NAME_CACHE_UNLOCK();
+
+       if (tcred != NOCRED)
+               kauth_cred_unref(&tcred);
 }
 
 
-void vnode_cache_credentials(vnode_t vp, vfs_context_t context)
+extern int bootarg_vnode_cache_defeat; /* default = 0, from bsd_init.c */
+
+boolean_t
+vnode_cache_is_authorized(vnode_t vp, vfs_context_t ctx, kauth_action_t action)
 {
-       kauth_cred_t ucred;
-       kauth_cred_t tcred = NOCRED;
-       struct timeval tv;
+       kauth_cred_t    ucred;
+       boolean_t       retval = FALSE;
 
-       ucred = vfs_context_ucred(context);
+       /* Boot argument to defeat rights caching */
+       if (bootarg_vnode_cache_defeat)
+               return FALSE;
 
-       if (vp->v_cred != ucred || (vp->v_mount->mnt_kern_flag & MNTK_AUTH_OPAQUE)) {
-               vnode_lock(vp);
+       if ( (vp->v_mount->mnt_kern_flag & (MNTK_AUTH_OPAQUE | MNTK_AUTH_CACHE_TTL)) ) {
+               /*
+                * a TTL is enabled on the rights cache... handle it here
+                * a TTL of 0 indicates that no rights should be cached
+                */
+               if (vp->v_mount->mnt_authcache_ttl) {
+                       if ( !(vp->v_mount->mnt_kern_flag & MNTK_AUTH_CACHE_TTL) ) {
+                               /*
+                                * For filesystems marked only MNTK_AUTH_OPAQUE (generally network ones),
+                                * we will only allow a SEARCH right on a directory to be cached...
+                                * that cached right always has a default TTL associated with it
+                                */
+                               if (action != KAUTH_VNODE_SEARCH || vp->v_type != VDIR)
+                                       vp = NULLVP;
+                       }
+                       if (vp != NULLVP && vnode_cache_is_stale(vp) == TRUE) {
+                               vnode_uncache_authorized_action(vp, vp->v_authorized_actions);
+                               vp = NULLVP;
+                       }
+               } else
+                       vp = NULLVP;
+       }
+       if (vp != NULLVP) {
+               ucred = vfs_context_ucred(ctx);
 
-               microuptime(&tv);
-               vp->v_cred_timestamp = tv.tv_sec;
+               NAME_CACHE_LOCK_SHARED();
 
-               if (vp->v_cred != ucred) {
-                       kauth_cred_ref(ucred);
-       
-                       tcred = vp->v_cred;
-                       vp->v_cred = ucred;
-               }
-               vnode_unlock(vp);
-       
-               if (tcred)
-                       kauth_cred_rele(tcred);
+               if (vp->v_cred == ucred && (vp->v_authorized_actions & action) == action)
+                       retval = TRUE;
+               
+               NAME_CACHE_UNLOCK();
        }
+       return retval;
 }
 
-/*     reverse_lookup - lookup by walking back up the parent chain while leveraging
- *     use of the name cache lock in order to protect our starting vnode.
- *     NOTE - assumes you already have search access to starting point.
- *  returns 0 when we have reached the root, current working dir, or chroot root 
- *
- */
-int
-reverse_lookup(vnode_t start_vp, vnode_t *lookup_vpp, struct filedesc *fdp, vfs_context_t context, int *dp_authorized)
+
+void vnode_cache_authorized_action(vnode_t vp, vfs_context_t ctx, kauth_action_t action)
 {
-       int                             vid, done = 0;
-       int                             auth_opaque = 0;
-       vnode_t                 dp = start_vp;
-       vnode_t                 vp = NULLVP;
-       kauth_cred_t    ucred;
-       struct timeval  tv;
+       kauth_cred_t tcred = NOCRED;
+       kauth_cred_t ucred;
+       struct timeval tv;
+       boolean_t ttl_active = FALSE;
 
-       ucred = vfs_context_ucred(context);
-       *lookup_vpp = start_vp;
+       ucred = vfs_context_ucred(ctx);
 
-       name_cache_lock();
+       if (!IS_VALID_CRED(ucred) || action == 0)
+               return;
+
+       if ( (vp->v_mount->mnt_kern_flag & (MNTK_AUTH_OPAQUE | MNTK_AUTH_CACHE_TTL)) ) {
+               /*
+                * a TTL is enabled on the rights cache... handle it here
+                * a TTL of 0 indicates that no rights should be cached
+                */
+               if (vp->v_mount->mnt_authcache_ttl == 0) 
+                       return;
+
+               if ( !(vp->v_mount->mnt_kern_flag & MNTK_AUTH_CACHE_TTL) ) {
+                       /*
+                        * only cache SEARCH action for filesystems marked
+                        * MNTK_AUTH_OPAQUE on VDIRs...
+                        * the lookup_path code will time these out
+                        */
+                       if ( (action & ~KAUTH_VNODE_SEARCH) || vp->v_type != VDIR )
+                               return;
+               }
+               ttl_active = TRUE;
 
-       if ( dp->v_mount && (dp->v_mount->mnt_kern_flag & MNTK_AUTH_OPAQUE) ) {
-               auth_opaque = 1;
                microuptime(&tv);
        }
-       for (;;) {
-               *dp_authorized = 0;
+       NAME_CACHE_LOCK();
 
-               if (auth_opaque && ((tv.tv_sec - dp->v_cred_timestamp) > VCRED_EXPIRED))
-                       break;
-               if (dp->v_cred != ucred)
-                       break;
-               /*
-                * indicate that we're allowed to traverse this directory...
-                * even if we bail for some reason, this information is valid and is used
-                * to avoid doing a vnode_authorize
+       if (vp->v_cred != ucred) {
+               kauth_cred_ref(ucred);
+               /*
+                * Use a temp variable to avoid kauth_cred_unref() while NAME_CACHE_LOCK is held
                 */
-               *dp_authorized = 1;
+               tcred = vp->v_cred;
+               vp->v_cred = ucred;
+               vp->v_authorized_actions = 0;
+       }
+       if (ttl_active == TRUE && vp->v_authorized_actions == 0) {
+               /*
+                * only reset the timestamnp on the
+                * first authorization cached after the previous
+                * timer has expired or we're switching creds...
+                * 'vnode_cache_is_authorized' will clear the 
+                * authorized actions if the TTL is active and
+                * it has expired
+                */
+               vp->v_cred_timestamp = tv.tv_sec;
+       }
+       vp->v_authorized_actions |= action;
 
-               if ((dp->v_flag & VROOT) != 0   ||              /* Hit "/" */
-                   (dp == fdp->fd_cdir)        ||              /* Hit process's working directory */
-                   (dp == fdp->fd_rdir)) {                     /* Hit process chroot()-ed root */
-                       done = 1;
-                       break;
-               }
+       NAME_CACHE_UNLOCK();
 
-               if ( (vp = dp->v_parent) == NULLVP)
-                       break;
+       if (IS_VALID_CRED(tcred))
+               kauth_cred_unref(&tcred);
+}
 
-               dp = vp;
-               *lookup_vpp = dp;
-       } /* for (;;) */
 
-       vid = dp->v_id;
-       
-       name_cache_unlock();
-       
-       if (done == 0 && dp != start_vp) {
-               if (vnode_getwithvid(dp, vid) != 0) {
-                       *lookup_vpp = start_vp;
-               }
-       }
+boolean_t vnode_cache_is_stale(vnode_t vp)
+{
+       struct timeval  tv;
+       boolean_t       retval;
+
+       microuptime(&tv);
+
+       if ((tv.tv_sec - vp->v_cred_timestamp) > vp->v_mount->mnt_authcache_ttl)
+               retval = TRUE;
+       else
+               retval = FALSE;
 
-       return((done == 1) ? 0 : -1);
+       return retval;
 }
 
+
+
+/*
+ * Returns:    0                       Success
+ *             ERECYCLE                vnode was recycled from underneath us.  Force lookup to be re-driven from namei.
+ *                                             This errno value should not be seen by anyone outside of the kernel.
+ */
 int 
-cache_lookup_path(struct nameidata *ndp, struct componentname *cnp, vnode_t dp, vfs_context_t context, int *trailing_slash, int *dp_authorized)
+cache_lookup_path(struct nameidata *ndp, struct componentname *cnp, vnode_t dp, 
+               vfs_context_t ctx, int *dp_authorized, vnode_t last_dp)
 {
        char            *cp;            /* pointer into pathname argument */
-       int             vid, vvid;
-       int             auth_opaque = 0;
+       int             vid;
+       int             vvid = 0;       /* protected by vp != NULLVP */
        vnode_t         vp = NULLVP;
        vnode_t         tdp = NULLVP;
        kauth_cred_t    ucred;
-       struct timeval tv;
+       boolean_t       ttl_enabled = FALSE;
+       struct timeval  tv;
+        mount_t                mp;
        unsigned int    hash;
+       int             error = 0;
 
-       ucred = vfs_context_ucred(context);
-       *trailing_slash = 0;
+#if CONFIG_TRIGGERS
+       vnode_t         trigger_vp;
+#endif /* CONFIG_TRIGGERS */
 
-       name_cache_lock();
+       ucred = vfs_context_ucred(ctx);
+       ndp->ni_flag &= ~(NAMEI_TRAILINGSLASH);
 
+       NAME_CACHE_LOCK_SHARED();
 
-       if ( dp->v_mount && (dp->v_mount->mnt_kern_flag & MNTK_AUTH_OPAQUE) ) {
-               auth_opaque = 1;
+       if ( dp->v_mount && (dp->v_mount->mnt_kern_flag & (MNTK_AUTH_OPAQUE | MNTK_AUTH_CACHE_TTL)) ) {
+               ttl_enabled = TRUE;
                microuptime(&tv);
        }
        for (;;) {
@@ -596,7 +998,7 @@ cache_lookup_path(struct nameidata *ndp, struct componentname *cnp, vnode_t dp,
                cp = cnp->cn_nameptr;
 
                while (*cp && (*cp != '/')) {
-                       hash ^= crc32tab[((hash >> 24) ^ (unsigned char)*cp++)];
+                       hash = crc32tab[((hash >> 24) ^ (unsigned char)*cp++)] ^ hash << 8;
                }
                /*
                 * the crc generator can legitimately generate
@@ -623,7 +1025,7 @@ cache_lookup_path(struct nameidata *ndp, struct componentname *cnp, vnode_t dp,
                        ndp->ni_pathlen--;
 
                        if (*cp == '\0') {
-                               *trailing_slash = 1;
+                               ndp->ni_flag |= NAMEI_TRAILINGSLASH;
                                *ndp->ni_next = '\0';
                        }
                }
@@ -638,12 +1040,56 @@ cache_lookup_path(struct nameidata *ndp, struct componentname *cnp, vnode_t dp,
                        cnp->cn_flags |= ISDOTDOT;
 
                *dp_authorized = 0;
+#if NAMEDRSRCFORK
+               /*
+                * Process a request for a file's resource fork.
+                *
+                * Consume the _PATH_RSRCFORKSPEC suffix and tag the path.
+                */
+               if ((ndp->ni_pathlen == sizeof(_PATH_RSRCFORKSPEC)) &&
+                   (cp[1] == '.' && cp[2] == '.') &&
+                   bcmp(cp, _PATH_RSRCFORKSPEC, sizeof(_PATH_RSRCFORKSPEC)) == 0) {
+                       /* Skip volfs file systems that don't support native streams. */
+                       if ((dp->v_mount != NULL) &&
+                           (dp->v_mount->mnt_flag & MNT_DOVOLFS) &&
+                           (dp->v_mount->mnt_kern_flag & MNTK_NAMED_STREAMS) == 0) {
+                               goto skiprsrcfork;
+                       }
+                       cnp->cn_flags |= CN_WANTSRSRCFORK;
+                       cnp->cn_flags |= ISLASTCN;
+                       ndp->ni_next[0] = '\0';
+                       ndp->ni_pathlen = 1;
+               }
+skiprsrcfork:
+#endif
 
-               if (auth_opaque && ((tv.tv_sec - dp->v_cred_timestamp) > VCRED_EXPIRED))
+#if CONFIG_MACF
+
+               /* 
+                * Name cache provides authorization caching (see below)
+                * that will short circuit MAC checks in lookup().
+                * We must perform MAC check here.  On denial
+                * dp_authorized will remain 0 and second check will
+                * be perfomed in lookup().
+                */
+               if (!(cnp->cn_flags & DONOTAUTH)) {
+                       error = mac_vnode_check_lookup(ctx, dp, cnp);
+                       if (error) {
+                               NAME_CACHE_UNLOCK();
+                               goto errorout;
+                       }
+               }
+#endif /* MAC */
+               if (ttl_enabled && ((tv.tv_sec - dp->v_cred_timestamp) > dp->v_mount->mnt_authcache_ttl))
                        break;
 
-               if (dp->v_cred != ucred)
+               /*
+                * NAME_CACHE_LOCK holds these fields stable
+                */
+               if ((dp->v_cred != ucred || !(dp->v_authorized_actions & KAUTH_VNODE_SEARCH)) &&
+                   !(dp->v_authorized_actions & KAUTH_VNODE_SEARCHBYANYONE))
                        break;
+
                /*
                 * indicate that we're allowed to traverse this directory...
                 * even if we fail the cache lookup or decide to bail for
@@ -653,14 +1099,53 @@ cache_lookup_path(struct nameidata *ndp, struct componentname *cnp, vnode_t dp,
                *dp_authorized = 1;
 
                if ( (cnp->cn_flags & (ISLASTCN | ISDOTDOT)) ) {
-                       if (cnp->cn_nameiop != LOOKUP)
-                               break;
-                       if (cnp->cn_flags & (LOCKPARENT | NOCACHE | ISDOTDOT))
-                               break;
+                       if (cnp->cn_nameiop != LOOKUP)
+                               break;
+                       if (cnp->cn_flags & LOCKPARENT) 
+                               break;
+                       if (cnp->cn_flags & NOCACHE)
+                               break;
+                       if (cnp->cn_flags & ISDOTDOT) {
+                               /*
+                                * Force directory hardlinks to go to
+                                * file system for ".." requests.
+                                */
+                               if (dp && (dp->v_flag & VISHARDLINK)) {
+                                       break;
+                               }
+                               /*
+                                * Quit here only if we can't use
+                                * the parent directory pointer or
+                                * don't have one.  Otherwise, we'll
+                                * use it below.
+                                */
+                               if ((dp->v_flag & VROOT)  ||
+                                   dp == ndp->ni_rootdir ||
+                                   dp->v_parent == NULLVP)
+                                       break;
+                       }
                }
-               if ( (vp = cache_lookup_locked(dp, cnp)) == NULLVP)
-                       break;
 
+               /*
+                * "." and ".." aren't supposed to be cached, so check
+                * for them before checking the cache.
+                */
+               if (cnp->cn_namelen == 1 && cnp->cn_nameptr[0] == '.')
+                       vp = dp;
+               else if ( (cnp->cn_flags & ISDOTDOT) )
+                       vp = dp->v_parent;
+               else {
+                       if ( (vp = cache_lookup_locked(dp, cnp)) == NULLVP)
+                               break;
+
+                       if ( (vp->v_flag & VISHARDLINK) ) {
+                               /*
+                                * The file system wants a VNOP_LOOKUP on this vnode
+                                */
+                               vp = NULL;
+                               break;
+                       }
+               }
                if ( (cnp->cn_flags & ISLASTCN) )
                        break;
 
@@ -669,8 +1154,27 @@ cache_lookup_path(struct nameidata *ndp, struct componentname *cnp, vnode_t dp,
                                vp = NULL;
                        break;
                }
-               if (vp->v_mountedhere && ((cnp->cn_flags & NOCROSSMOUNT) == 0))
-                       break;
+
+               if ( (mp = vp->v_mountedhere) && ((cnp->cn_flags & NOCROSSMOUNT) == 0)) {
+
+                       if (mp->mnt_realrootvp == NULLVP || mp->mnt_generation != mount_generation ||
+                               mp->mnt_realrootvp_vid != mp->mnt_realrootvp->v_id)
+                               break;
+                       vp = mp->mnt_realrootvp;
+               }
+
+#if CONFIG_TRIGGERS
+               /*
+                * After traversing all mountpoints stacked here, if we have a
+                * trigger in hand, resolve it.  Note that we don't need to 
+                * leave the fast path if the mount has already happened.
+                */
+               if ((vp->v_resolve != NULL) && 
+                               (vp->v_resolve->vr_resolve_func != NULL)) {
+                       break;
+               } 
+#endif /* CONFIG_TRIGGERS */
+
 
                dp = vp;
                vp = NULLVP;
@@ -686,8 +1190,7 @@ cache_lookup_path(struct nameidata *ndp, struct componentname *cnp, vnode_t dp,
                vvid = vp->v_id;
        vid = dp->v_id;
        
-       name_cache_unlock();
-
+       NAME_CACHE_UNLOCK();
 
        if ((vp != NULLVP) && (vp->v_type != VLNK) &&
            ((cnp->cn_flags & (ISLASTCN | LOCKPARENT | WANTPARENT | SAVESTART)) == ISLASTCN)) {
@@ -704,34 +1207,45 @@ cache_lookup_path(struct nameidata *ndp, struct componentname *cnp, vnode_t dp,
                dp = NULLVP;
        } else {
 need_dp:
-               /*
+               /*
                 * return the last directory we looked at
-                * with an io reference held
+                * with an io reference held. If it was the one passed
+                * in as a result of the last iteration of VNOP_LOOKUP,
+                * it should already hold an io ref. No need to increase ref.
                 */
-               if (dp == ndp->ni_usedvp) {
-                       /*
-                        * if this vnode matches the one passed in via USEDVP
-                        * than this context already holds an io_count... just
-                        * use vnode_get to get an extra ref for lookup to play
-                        * with... can't use the getwithvid variant here because
-                        * it will block behind a vnode_drain which would result
-                        * in a deadlock (since we already own an io_count that the
-                        * vnode_drain is waiting on)... vnode_get grabs the io_count
-                        * immediately w/o waiting... it always succeeds
-                        */
-                       vnode_get(dp);
-               } else if ( (vnode_getwithvid(dp, vid)) ) {
-                       /*
-                        * failure indicates the vnode
-                        * changed identity or is being
-                        * TERMINATED... in either case
-                        * punt this lookup
-                        */
-                       return (ENOENT);
+               if (last_dp != dp){
+                       
+                       if (dp == ndp->ni_usedvp) {
+                               /*
+                                * if this vnode matches the one passed in via USEDVP
+                                * than this context already holds an io_count... just
+                                * use vnode_get to get an extra ref for lookup to play
+                                * with... can't use the getwithvid variant here because
+                                * it will block behind a vnode_drain which would result
+                                * in a deadlock (since we already own an io_count that the
+                                * vnode_drain is waiting on)... vnode_get grabs the io_count
+                                * immediately w/o waiting... it always succeeds
+                                */
+                               vnode_get(dp);
+                       } else if ( (vnode_getwithvid_drainok(dp, vid)) ) {
+                               /*
+                                * failure indicates the vnode
+                                * changed identity or is being
+                                * TERMINATED... in either case
+                                * punt this lookup.
+                                * 
+                                * don't necessarily return ENOENT, though, because
+                                * we really want to go back to disk and make sure it's
+                                * there or not if someone else is changing this
+                                * vnode.
+                                */
+                               error = ERECYCLE;
+                               goto errorout;
+                       }
                }
        }
        if (vp != NULLVP) {
-               if ( (vnode_getwithvid(vp, vvid)) ) {
+               if ( (vnode_getwithvid_drainok(vp, vvid)) ) {
                        vp = NULLVP;
 
                        /*
@@ -748,45 +1262,72 @@ need_dp:
                        }
                }
        }
+
        ndp->ni_dvp = dp;
        ndp->ni_vp  = vp;
 
-       return (0);
+#if CONFIG_TRIGGERS
+       trigger_vp = vp ? vp : dp;
+       if ((error == 0) && (trigger_vp != NULLVP) && vnode_isdir(trigger_vp)) {
+               error = vnode_trigger_resolve(trigger_vp, ndp, ctx);
+               if (error) {
+                       if (vp)
+                               vnode_put(vp);
+                       if (dp) 
+                               vnode_put(dp);
+                       goto errorout;
+               }
+       } 
+#endif /* CONFIG_TRIGGERS */
+
+errorout:
+       /* 
+        * If we came into cache_lookup_path after an iteration of the lookup loop that
+        * resulted in a call to VNOP_LOOKUP, then VNOP_LOOKUP returned a vnode with a io ref
+        * on it.  It is now the job of cache_lookup_path to drop the ref on this vnode 
+        * when it is no longer needed.  If we get to this point, and last_dp is not NULL
+        * and it is ALSO not the dvp we want to return to caller of this function, it MUST be
+        * the case that we got to a subsequent path component and this previous vnode is 
+        * no longer needed.  We can then drop the io ref on it.
+        */
+       if ((last_dp != NULLVP) && (last_dp != ndp->ni_dvp)){
+               vnode_put(last_dp);
+       }
+       
+       //initialized to 0, should be the same if no error cases occurred.
+       return error;
 }
 
 
 static vnode_t
 cache_lookup_locked(vnode_t dvp, struct componentname *cnp)
 {
-       register struct namecache *ncp;
-       register struct nchashhead *ncpp;
-       register long namelen = cnp->cn_namelen;
-       char *nameptr = cnp->cn_nameptr;
+       struct namecache *ncp;
+       struct nchashhead *ncpp;
+       long namelen = cnp->cn_namelen;
        unsigned int hashval = (cnp->cn_hash & NCHASHMASK);
-       vnode_t vp;
        
+       if (nc_disabled) {
+               return NULL;
+       }
+
        ncpp = NCHHASH(dvp, cnp->cn_hash);
        LIST_FOREACH(ncp, ncpp, nc_hash) {
                if ((ncp->nc_dvp == dvp) && (ncp->nc_hashval == hashval)) {
-                       if (memcmp(ncp->nc_name, nameptr, namelen) == 0 && ncp->nc_name[namelen] == 0)
+                       if (memcmp(ncp->nc_name, cnp->cn_nameptr, namelen) == 0 && ncp->nc_name[namelen] == 0)
                                break;
                }
        }
-       if (ncp == 0)
+       if (ncp == 0) {
                /*
                 * We failed to find an entry
                 */
+               NCHSTAT(ncs_miss);
                return (NULL);
-
-       vp = ncp->nc_vp;
-       if (vp && (vp->v_flag & VISHARDLINK)) {
-                       /*
-                        * The file system wants a VNOP_LOOKUP on this vnode
-                        */
-                       vp = NULL;
        }
-       
-       return (vp);
+       NCHSTAT(ncs_goodhits);
+
+       return (ncp->nc_vp);
 }
 
 
@@ -801,11 +1342,11 @@ hash_string(const char *cp, int len)
 
     if (len) {
             while (len--) {
-                   hash ^= crc32tab[((hash >> 24) ^ (unsigned char)*cp++)];
+                   hash = crc32tab[((hash >> 24) ^ (unsigned char)*cp++)] ^ hash << 8;
            }
     } else {
             while (*cp != '\0') {
-                   hash ^= crc32tab[((hash >> 24) ^ (unsigned char)*cp++)];
+                   hash = crc32tab[((hash >> 24) ^ (unsigned char)*cp++)] ^ hash << 8;
            }
     }
     /*
@@ -835,55 +1376,69 @@ hash_string(const char *cp, int len)
  */
 
 int
-cache_lookup(dvp, vpp, cnp)
-       struct vnode *dvp;
-       struct vnode **vpp;
-       struct componentname *cnp;
+cache_lookup(struct vnode *dvp, struct vnode **vpp, struct componentname *cnp)
 {
-       register struct namecache *ncp;
-       register struct nchashhead *ncpp;
-       register long namelen = cnp->cn_namelen;
-       char *nameptr = cnp->cn_nameptr;
-       unsigned int hashval = (cnp->cn_hash & NCHASHMASK);
+       struct namecache *ncp;
+       struct nchashhead *ncpp;
+       long namelen = cnp->cn_namelen;
+       unsigned int hashval;
+       boolean_t       have_exclusive = FALSE;
        uint32_t vid;
        vnode_t  vp;
 
-       name_cache_lock();
+       if (cnp->cn_hash == 0)
+               cnp->cn_hash = hash_string(cnp->cn_nameptr, cnp->cn_namelen);
+       hashval = (cnp->cn_hash & NCHASHMASK);
+
+       if (nc_disabled) {
+               return 0;
+       }
+
+       NAME_CACHE_LOCK_SHARED();
 
+relook:
        ncpp = NCHHASH(dvp, cnp->cn_hash);
        LIST_FOREACH(ncp, ncpp, nc_hash) {
                if ((ncp->nc_dvp == dvp) && (ncp->nc_hashval == hashval)) {
-                       if (memcmp(ncp->nc_name, nameptr, namelen) == 0 && ncp->nc_name[namelen] == 0)
+                       if (memcmp(ncp->nc_name, cnp->cn_nameptr, namelen) == 0 && ncp->nc_name[namelen] == 0)
                                break;
                }
        }
        /* We failed to find an entry */
        if (ncp == 0) {
-               nchstats.ncs_miss++;
-               name_cache_unlock();
+               NCHSTAT(ncs_miss);
+               NAME_CACHE_UNLOCK();
                return (0);
        }
 
        /* We don't want to have an entry, so dump it */
        if ((cnp->cn_flags & MAKEENTRY) == 0) {
-               nchstats.ncs_badhits++;
-               cache_delete(ncp, 1);
-               name_cache_unlock();
-               return (0);
+               if (have_exclusive == TRUE) {
+                       NCHSTAT(ncs_badhits);
+                       cache_delete(ncp, 1);
+                       NAME_CACHE_UNLOCK();
+                       return (0);
+               }
+               NAME_CACHE_UNLOCK();
+               NAME_CACHE_LOCK();
+               have_exclusive = TRUE;
+               goto relook;
        } 
        vp = ncp->nc_vp;
 
        /* We found a "positive" match, return the vnode */
         if (vp) {
-               nchstats.ncs_goodhits++;
+               NCHSTAT(ncs_goodhits);
 
                vid = vp->v_id;
-               name_cache_unlock();
+               NAME_CACHE_UNLOCK();
 
                if (vnode_getwithvid(vp, vid)) {
-                       name_cache_lock();
-                       nchstats.ncs_badvid++;
-                       name_cache_unlock();
+#if COLLECT_STATS
+                       NAME_CACHE_LOCK();
+                       NCHSTAT(ncs_badvid);
+                       NAME_CACHE_UNLOCK();
+#endif
                        return (0);
                }
                *vpp = vp;
@@ -892,49 +1447,124 @@ cache_lookup(dvp, vpp, cnp)
 
        /* We found a negative match, and want to create it, so purge */
        if (cnp->cn_nameiop == CREATE || cnp->cn_nameiop == RENAME) {
-               nchstats.ncs_badhits++;
-               cache_delete(ncp, 1);
-               name_cache_unlock();
-               return (0);
+               if (have_exclusive == TRUE) {
+                       NCHSTAT(ncs_badhits);
+                       cache_delete(ncp, 1);
+                       NAME_CACHE_UNLOCK();
+                       return (0);
+               }
+               NAME_CACHE_UNLOCK();
+               NAME_CACHE_LOCK();
+               have_exclusive = TRUE;
+               goto relook;
        }
 
        /*
         * We found a "negative" match, ENOENT notifies client of this match.
         * The nc_whiteout field records whether this is a whiteout.
         */
-       nchstats.ncs_neghits++;
+       NCHSTAT(ncs_neghits);
 
        if (ncp->nc_whiteout)
                cnp->cn_flags |= ISWHITEOUT;
-       name_cache_unlock();
+       NAME_CACHE_UNLOCK();
        return (ENOENT);
 }
 
+const char *
+cache_enter_create(vnode_t dvp, vnode_t vp, struct componentname *cnp)
+{
+       const char *strname;
+
+        if (cnp->cn_hash == 0)
+               cnp->cn_hash = hash_string(cnp->cn_nameptr, cnp->cn_namelen);
+
+       /*
+        * grab 2 references on the string entered
+        * one for the cache_enter_locked to consume
+        * and the second to be consumed by v_name (vnode_create call point)
+        */
+       strname = add_name_internal(cnp->cn_nameptr, cnp->cn_namelen, cnp->cn_hash, TRUE, 0);
+
+       NAME_CACHE_LOCK();
+
+       cache_enter_locked(dvp, vp, cnp, strname);
+
+       NAME_CACHE_UNLOCK();
+
+       return (strname);
+}
+
+
+/*
+ * Add an entry to the cache...
+ * but first check to see if the directory
+ * that this entry is to be associated with has
+ * had any cache_purges applied since we took
+ * our identity snapshot... this check needs to
+ * be done behind the name cache lock
+ */
+void
+cache_enter_with_gen(struct vnode *dvp, struct vnode *vp, struct componentname *cnp, int gen)
+{
+
+        if (cnp->cn_hash == 0)
+               cnp->cn_hash = hash_string(cnp->cn_nameptr, cnp->cn_namelen);
+
+       NAME_CACHE_LOCK();
+
+       if (dvp->v_nc_generation == gen)
+               (void)cache_enter_locked(dvp, vp, cnp, NULL);
+
+       NAME_CACHE_UNLOCK();
+}
+
+
 /*
  * Add an entry to the cache.
  */
 void
-cache_enter(dvp, vp, cnp)
-       struct vnode *dvp;
-       struct vnode *vp;
-       struct componentname *cnp;
+cache_enter(struct vnode *dvp, struct vnode *vp, struct componentname *cnp)
 {
-        register struct namecache *ncp, *negp;
-       register struct nchashhead *ncpp;
+       const char *strname;
 
         if (cnp->cn_hash == 0)
                cnp->cn_hash = hash_string(cnp->cn_nameptr, cnp->cn_namelen);
 
-       name_cache_lock();
+       /*
+        * grab 1 reference on the string entered
+        * for the cache_enter_locked to consume
+        */
+       strname = add_name_internal(cnp->cn_nameptr, cnp->cn_namelen, cnp->cn_hash, FALSE, 0);
+
+       NAME_CACHE_LOCK();
 
-       /* if the entry is for -ve caching vp is null */
+       cache_enter_locked(dvp, vp, cnp, strname);
+
+       NAME_CACHE_UNLOCK();
+}
+
+
+static void
+cache_enter_locked(struct vnode *dvp, struct vnode *vp, struct componentname *cnp, const char *strname)
+{
+        struct namecache *ncp, *negp;
+       struct nchashhead *ncpp;
+
+       if (nc_disabled) 
+               return;
+
+       /*
+        * if the entry is for -ve caching vp is null
+        */
        if ((vp != NULLVP) && (LIST_FIRST(&vp->v_nclinks))) {
                /*
                 * someone beat us to the punch..
                 * this vnode is already in the cache
                 */
-               name_cache_unlock();
-                       return;
+               if (strname != NULL)
+                       vfs_removename(strname);
+               return;
        }
        /*
         * We allocate a new entry if we are less than the maximum
@@ -947,7 +1577,7 @@ cache_enter(dvp, vp, cnp)
                /*
                 * Allocate one more entry
                 */
-               ncp = (struct namecache *)_MALLOC_ZONE((u_long)sizeof *ncp, M_CACHE, M_WAITOK);
+               ncp = (struct namecache *)_MALLOC_ZONE(sizeof(*ncp), M_CACHE, M_WAITOK);
                numcache++;
        } else {
                /*
@@ -961,11 +1591,11 @@ cache_enter(dvp, vp, cnp)
                        * still in use... we need to
                        * delete it before re-using it
                        */
-                       nchstats.ncs_stolen++;
+                       NCHSTAT(ncs_stolen);
                        cache_delete(ncp, 0);
                }
        }
-       nchstats.ncs_enters++;
+       NCHSTAT(ncs_enters);
 
        /*
         * Fill in cache info, if vp is NULL this is a "negative" cache entry.
@@ -974,8 +1604,11 @@ cache_enter(dvp, vp, cnp)
        ncp->nc_dvp = dvp;
        ncp->nc_hashval = cnp->cn_hash;
        ncp->nc_whiteout = FALSE;
-       ncp->nc_name = add_name_locked(cnp->cn_nameptr, cnp->cn_namelen, cnp->cn_hash, 0);
 
+       if (strname == NULL)
+               ncp->nc_name = add_name_internal(cnp->cn_nameptr, cnp->cn_namelen, cnp->cn_hash, FALSE, 0);
+       else
+               ncp->nc_name = strname;
        /*
         * make us the newest entry in the cache
         * i.e. we'll be the last to be stolen
@@ -985,7 +1618,7 @@ cache_enter(dvp, vp, cnp)
        ncpp = NCHHASH(dvp, cnp->cn_hash);
 #if DIAGNOSTIC
        {
-               register struct namecache *p;
+               struct namecache *p;
 
                for (p = ncpp->lh_first; p != 0; p = p->nc_hash.le_next)
                        if (p == ncp)
@@ -1013,17 +1646,15 @@ cache_enter(dvp, vp, cnp)
          
                if (cnp->cn_flags & ISWHITEOUT)
                        ncp->nc_whiteout = TRUE;
-               nchstats.ncs_negtotal++;
+               ncs_negtotal++;
 
-               if (nchstats.ncs_negtotal > desiredNegNodes) {
+               if (ncs_negtotal > desiredNegNodes) {
                       /*
                        * if we've reached our desired limit
                        * of negative cache entries, delete
                        * the oldest
                        */
                        negp = TAILQ_FIRST(&neghead);
-                       TAILQ_REMOVE(&neghead, negp, nc_un.nc_negentry);
-
                        cache_delete(negp, 1);
                }
        }
@@ -1032,8 +1663,6 @@ cache_enter(dvp, vp, cnp)
         * are children of dvp
         */
        LIST_INSERT_HEAD(&dvp->v_ncchildren, ncp, nc_child);
-
-       name_cache_unlock();
 }
 
 
@@ -1073,6 +1702,8 @@ static void init_crc32(void)
 void
 nchinit(void)
 {
+       int     i;
+
        desiredNegNodes = (desiredvnodes / 10);
        desiredNodes = desiredvnodes + desiredNegNodes;
 
@@ -1081,39 +1712,55 @@ nchinit(void)
 
        init_crc32();
 
-       nchashtbl = hashinit(MAX(4096, (2 *desiredNodes)), M_CACHE, &nchash);
+       nchashtbl = hashinit(MAX(CONFIG_NC_HASH, (2 *desiredNodes)), M_CACHE, &nchash);
        nchashmask = nchash;
        nchash++;
 
        init_string_table();
        
-       /* Allocate mount list lock group attribute and group */
+       /* Allocate name cache lock group attribute and group */
        namecache_lck_grp_attr= lck_grp_attr_alloc_init();
-       lck_grp_attr_setstat(namecache_lck_grp_attr);
 
        namecache_lck_grp = lck_grp_alloc_init("Name Cache",  namecache_lck_grp_attr);
        
-       /* Allocate mount list lock attribute */
+       /* Allocate name cache lock attribute */
        namecache_lck_attr = lck_attr_alloc_init();
-       //lck_attr_setdebug(namecache_lck_attr);
 
-       /* Allocate mount list lock */
-       namecache_mtx_lock = lck_mtx_alloc_init(namecache_lck_grp, namecache_lck_attr);
+       /* Allocate name cache lock */
+       namecache_rw_lock = lck_rw_alloc_init(namecache_lck_grp, namecache_lck_attr);
 
 
+       /* Allocate string cache lock group attribute and group */
+       strcache_lck_grp_attr= lck_grp_attr_alloc_init();
+
+       strcache_lck_grp = lck_grp_alloc_init("String Cache",  strcache_lck_grp_attr);
+       
+       /* Allocate string cache lock attribute */
+       strcache_lck_attr = lck_attr_alloc_init();
+
+       /* Allocate string cache lock */
+       strtable_rw_lock = lck_rw_alloc_init(strcache_lck_grp, strcache_lck_attr);
+
+       for (i = 0; i < NUM_STRCACHE_LOCKS; i++)
+               lck_mtx_init(&strcache_mtx_locks[i], strcache_lck_grp, strcache_lck_attr);
+}
+
+void
+name_cache_lock_shared(void)
+{
+       lck_rw_lock_shared(namecache_rw_lock);
 }
 
 void
 name_cache_lock(void)
 {
-       lck_mtx_lock(namecache_mtx_lock);
+       lck_rw_lock_exclusive(namecache_rw_lock);
 }
 
 void
 name_cache_unlock(void)
 {
-       lck_mtx_unlock(namecache_mtx_lock);
-
+       lck_rw_done(namecache_rw_lock);
 }
 
 
@@ -1132,7 +1779,7 @@ resize_namecache(u_int newsize)
     dNodes = newsize + dNegNodes;
 
     // we don't support shrinking yet
-    if (dNodes < desiredNodes) {
+    if (dNodes <= desiredNodes) {
        return 0;
     }
     new_table = hashinit(2 * dNodes, M_CACHE, &nchashmask);
@@ -1142,7 +1789,7 @@ resize_namecache(u_int newsize)
        return ENOMEM;
     }
 
-    name_cache_lock();
+    NAME_CACHE_LOCK();
     // do the switch!
     old_table = nchashtbl;
     nchashtbl = new_table;
@@ -1170,7 +1817,7 @@ resize_namecache(u_int newsize)
     desiredNodes = dNodes;
     desiredNegNodes = dNegNodes;
     
-    name_cache_unlock();
+    NAME_CACHE_UNLOCK();
     FREE(old_table, M_CACHE);
 
     return 0;
@@ -1179,13 +1826,13 @@ resize_namecache(u_int newsize)
 static void
 cache_delete(struct namecache *ncp, int age_entry)
 {
-        nchstats.ncs_deletes++;
+        NCHSTAT(ncs_deletes);
 
         if (ncp->nc_vp) {
                LIST_REMOVE(ncp, nc_un.nc_link);
        } else {
                TAILQ_REMOVE(&neghead, ncp, nc_un.nc_negentry);
-               nchstats.ncs_negtotal--;
+               ncs_negtotal--;
        }
         LIST_REMOVE(ncp, nc_child);
 
@@ -1206,7 +1853,7 @@ cache_delete(struct namecache *ncp, int age_entry)
                TAILQ_REMOVE(&nchead, ncp, nc_entry);
                TAILQ_INSERT_HEAD(&nchead, ncp, nc_entry);
        }
-       remove_name_locked(ncp->nc_name);
+       vfs_removename(ncp->nc_name);
        ncp->nc_name = NULL;
 }
 
@@ -1219,11 +1866,18 @@ void
 cache_purge(vnode_t vp)
 {
         struct namecache *ncp;
+       kauth_cred_t tcred = NULL;
 
-       if ((LIST_FIRST(&vp->v_nclinks) == NULL) && (LIST_FIRST(&vp->v_ncchildren) == NULL))
+       if ((LIST_FIRST(&vp->v_nclinks) == NULL) && 
+                       (LIST_FIRST(&vp->v_ncchildren) == NULL) && 
+                       (vp->v_cred == NOCRED) &&
+                       (vp->v_parent == NULLVP))
                return;
 
-       name_cache_lock();
+       NAME_CACHE_LOCK();
+
+       if (vp->v_parent)
+               vp->v_parent->v_nc_generation++;
 
        while ( (ncp = LIST_FIRST(&vp->v_nclinks)) )
                cache_delete(ncp, 1);
@@ -1231,7 +1885,17 @@ cache_purge(vnode_t vp)
        while ( (ncp = LIST_FIRST(&vp->v_ncchildren)) )
                cache_delete(ncp, 1);
 
-       name_cache_unlock();
+       /*
+        * Use a temp variable to avoid kauth_cred_unref() while NAME_CACHE_LOCK is held
+        */
+       tcred = vp->v_cred;
+       vp->v_cred = NOCRED;
+       vp->v_authorized_actions = 0;
+
+       NAME_CACHE_UNLOCK();
+
+       if (IS_VALID_CRED(tcred))
+               kauth_cred_unref(&tcred);
 }
 
 /*
@@ -1244,15 +1908,15 @@ cache_purge(vnode_t vp)
 void
 cache_purge_negatives(vnode_t vp)
 {
-       struct namecache *ncp;
+       struct namecache *ncp, *next_ncp;
 
-       name_cache_lock();
+       NAME_CACHE_LOCK();
 
-       LIST_FOREACH(ncp, &vp->v_ncchildren, nc_child)
+       LIST_FOREACH_SAFE(ncp, &vp->v_ncchildren, nc_child, next_ncp)
                if (ncp->nc_vp == NULL)
                        cache_delete(ncp , 1);
 
-       name_cache_unlock();
+       NAME_CACHE_UNLOCK();
 }
 
 /*
@@ -1262,13 +1926,12 @@ cache_purge_negatives(vnode_t vp)
  * entries at the same time.
  */
 void
-cache_purgevfs(mp)
-       struct mount *mp;
+cache_purgevfs(struct mount *mp)
 {
        struct nchashhead *ncpp;
        struct namecache *ncp;
 
-       name_cache_lock();
+       NAME_CACHE_LOCK();
        /* Scan hash tables for applicable entries */
        for (ncpp = &nchashtbl[nchash - 1]; ncpp >= nchashtbl; ncpp--) {
 restart:         
@@ -1279,7 +1942,7 @@ restart:
                        }
                }
        }
-       name_cache_unlock();
+       NAME_CACHE_UNLOCK();
 }
 
 
@@ -1289,195 +1952,226 @@ restart:
 //
 static LIST_HEAD(stringhead, string_t) *string_ref_table;
 static u_long   string_table_mask;
-static uint32_t max_chain_len=0;
-static struct stringhead *long_chain_head=NULL;
 static uint32_t filled_buckets=0;
-static uint32_t num_dups=0;
-static uint32_t nstrings=0;
+
 
 typedef struct string_t {
     LIST_ENTRY(string_t)  hash_chain;
-    unsigned char        *str;
+    const char *str;
     uint32_t              refcount;
 } string_t;
 
 
-
-static int
+static void
 resize_string_ref_table(void)
 {
-    struct stringhead *new_table;
-    struct stringhead *old_table;
-    struct stringhead *old_head, *head;
-    string_t          *entry, *next;
-    uint32_t           i, hashval;
-    u_long             new_mask, old_mask;
-
-    new_table = hashinit((string_table_mask + 1) * 2, M_CACHE, &new_mask);
-    if (new_table == NULL) {
-       return ENOMEM;
-    }
+       struct stringhead *new_table;
+       struct stringhead *old_table;
+       struct stringhead *old_head, *head;
+       string_t          *entry, *next;
+       uint32_t           i, hashval;
+       u_long             new_mask, old_mask;
 
-    // do the switch!
-    old_table         = string_ref_table;
-    string_ref_table  = new_table;
-    old_mask          = string_table_mask;
-    string_table_mask = new_mask;
+       /*
+        * need to hold the table lock exclusively
+        * in order to grow the table... need to recheck
+        * the need to resize again after we've taken
+        * the lock exclusively in case some other thread
+        * beat us to the punch
+        */
+       lck_rw_lock_exclusive(strtable_rw_lock);
 
-    printf("resize: max chain len %d, new table size %d\n",
-          max_chain_len, new_mask + 1);
-    max_chain_len   = 0;
-    long_chain_head = NULL;
-    filled_buckets  = 0;
+       if (4 * filled_buckets < ((string_table_mask + 1) * 3)) {
+               lck_rw_done(strtable_rw_lock);
+               return;
+       }
+       new_table = hashinit((string_table_mask + 1) * 2, M_CACHE, &new_mask);
 
-    // walk the old table and insert all the entries into
-    // the new table
-    //
-    for(i=0; i <= old_mask; i++) {
-       old_head = &old_table[i];
-       for (entry=old_head->lh_first; entry != NULL; entry=next) {
-           hashval = hash_string(entry->str, 0);
-           head = &string_ref_table[hashval & string_table_mask];
-           if (head->lh_first == NULL) {
-               filled_buckets++;
-           }
+       if (new_table == NULL) {
+               printf("failed to resize the hash table.\n");
+               lck_rw_done(strtable_rw_lock);
+               return;
+       }
 
-           next = entry->hash_chain.le_next;
-           LIST_INSERT_HEAD(head, entry, hash_chain);
+       // do the switch!
+       old_table         = string_ref_table;
+       string_ref_table  = new_table;
+       old_mask          = string_table_mask;
+       string_table_mask = new_mask;
+       filled_buckets    = 0;
+
+       // walk the old table and insert all the entries into
+       // the new table
+       //
+       for (i = 0; i <= old_mask; i++) {
+               old_head = &old_table[i];
+               for (entry = old_head->lh_first; entry != NULL; entry = next) {
+                       hashval = hash_string((const char *)entry->str, 0);
+                       head = &string_ref_table[hashval & string_table_mask];
+                       if (head->lh_first == NULL) {
+                               filled_buckets++;
+                       }
+                       next = entry->hash_chain.le_next;
+                       LIST_INSERT_HEAD(head, entry, hash_chain);
+               }
        }
-    }
-    
-    FREE(old_table, M_CACHE);
+       lck_rw_done(strtable_rw_lock);
 
-    return 0;
+       FREE(old_table, M_CACHE);
 }
 
 
 static void
 init_string_table(void)
 {
-    string_ref_table = hashinit(4096, M_CACHE, &string_table_mask);
+       string_ref_table = hashinit(CONFIG_VFS_NAMES, M_CACHE, &string_table_mask);
 }
 
 
-char *
-vfs_addname(const char *name, size_t len, u_int hashval, u_int flags)
+const char *
+vfs_addname(const char *name, uint32_t len, u_int hashval, u_int flags)
 {
-        char * ptr;
-
-       name_cache_lock();
-       ptr = add_name_locked(name, len, hashval, flags);
-       name_cache_unlock();
-
-       return(ptr);
+       return (add_name_internal(name, len, hashval, FALSE, flags));
 }
 
-static char *
-add_name_locked(const char *name, size_t len, u_int hashval, __unused u_int flags)
+
+static const char *
+add_name_internal(const char *name, uint32_t len, u_int hashval, boolean_t need_extra_ref, __unused u_int flags)
 {
-    struct stringhead *head;
-    string_t          *entry;
-    uint32_t          chain_len = 0;
+       struct stringhead *head;
+       string_t          *entry;
+       uint32_t          chain_len = 0;
+       uint32_t          hash_index;
+        uint32_t         lock_index;
+       char              *ptr;
     
-    //
-    // If the table gets more than 3/4 full, resize it
-    //
-    if (4*filled_buckets >= ((string_table_mask + 1) * 3)) {
-               if (resize_string_ref_table() != 0) {
-                       printf("failed to resize the hash table.\n");
-               }
-    }
-    if (hashval == 0) {
-       hashval = hash_string(name, 0);
-    }
-
-    head = &string_ref_table[hashval & string_table_mask];
-    for (entry=head->lh_first; entry != NULL; chain_len++, entry=entry->hash_chain.le_next) {
-       if (memcmp(entry->str, name, len) == 0 && entry->str[len] == '\0') {
-           entry->refcount++;
-           num_dups++;
-           break;
+       /*
+        * if the length already accounts for the null-byte, then
+        * subtract one so later on we don't index past the end
+        * of the string.
+        */
+       if (len > 0 && name[len-1] == '\0') {
+               len--;
+       }
+       if (hashval == 0) {
+               hashval = hash_string(name, len);
        }
-    }
 
-    if (entry == NULL) {
-       // it wasn't already there so add it.
-       MALLOC(entry, string_t *, sizeof(string_t) + len + 1, M_TEMP, M_WAITOK);
+       /*
+        * take this lock 'shared' to keep the hash stable
+        * if someone else decides to grow the pool they
+        * will take this lock exclusively
+        */
+       lck_rw_lock_shared(strtable_rw_lock);
 
-       // have to get "head" again because we could have blocked
-       // in malloc and thus head could have changed.
-       //
-       head = &string_ref_table[hashval & string_table_mask];
-       if (head->lh_first == NULL) {
-           filled_buckets++;
+       /*
+        * If the table gets more than 3/4 full, resize it
+        */
+       if (4 * filled_buckets >= ((string_table_mask + 1) * 3)) {
+               lck_rw_done(strtable_rw_lock);
+
+               resize_string_ref_table();
+
+               lck_rw_lock_shared(strtable_rw_lock);
        }
+       hash_index = hashval & string_table_mask;
+       lock_index = hash_index % NUM_STRCACHE_LOCKS;
 
-       entry->str = (char *)((char *)entry + sizeof(string_t));
-       strncpy(entry->str, name, len);
-       entry->str[len] = '\0';
-       entry->refcount = 1;
-       LIST_INSERT_HEAD(head, entry, hash_chain);
+       head = &string_ref_table[hash_index];
 
-       if (chain_len > max_chain_len) {
-           max_chain_len   = chain_len;
-           long_chain_head = head;
+       lck_mtx_lock_spin(&strcache_mtx_locks[lock_index]);
+
+       for (entry = head->lh_first; entry != NULL; chain_len++, entry = entry->hash_chain.le_next) {
+               if (memcmp(entry->str, name, len) == 0 && entry->str[len] == 0) {
+                       entry->refcount++;
+                       break;
+               }
        }
+       if (entry == NULL) {
+               lck_mtx_convert_spin(&strcache_mtx_locks[lock_index]);
+               /*
+                * it wasn't already there so add it.
+                */
+               MALLOC(entry, string_t *, sizeof(string_t) + len + 1, M_TEMP, M_WAITOK);
 
-       nstrings++;
-    }
+               if (head->lh_first == NULL) {
+                       OSAddAtomic(1, &filled_buckets);
+               }
+               ptr = (char *)((char *)entry + sizeof(string_t));
+               strncpy(ptr, name, len);
+               ptr[len] = '\0';
+               entry->str = ptr;
+               entry->refcount = 1;
+               LIST_INSERT_HEAD(head, entry, hash_chain);
+       }
+       if (need_extra_ref == TRUE)
+               entry->refcount++;
     
-    return entry->str;
+       lck_mtx_unlock(&strcache_mtx_locks[lock_index]);
+       lck_rw_done(strtable_rw_lock);
+
+       return (const char *)entry->str;
 }
 
+
 int
 vfs_removename(const char *nameref)
 {
-       int i;
+       struct stringhead *head;
+       string_t          *entry;
+       uint32_t           hashval;
+       uint32_t           hash_index;
+        uint32_t          lock_index;
+       int                retval = ENOENT;
 
-       name_cache_lock();
-       i = remove_name_locked(nameref);
-       name_cache_unlock();
+       hashval = hash_string(nameref, 0);
 
-       return(i);
-       
-}
+       /*
+        * take this lock 'shared' to keep the hash stable
+        * if someone else decides to grow the pool they
+        * will take this lock exclusively
+        */
+       lck_rw_lock_shared(strtable_rw_lock);
+       /*
+        * must compute the head behind the table lock
+        * since the size and location of the table
+        * can change on the fly
+        */
+       hash_index = hashval & string_table_mask;
+       lock_index = hash_index % NUM_STRCACHE_LOCKS;
 
+       head = &string_ref_table[hash_index];
 
-static int
-remove_name_locked(const char *nameref)
-{
-    struct stringhead *head;
-    string_t          *entry;
-    uint32_t           hashval;
-    char * ptr;
-
-    hashval = hash_string(nameref, 0);
-    head = &string_ref_table[hashval & string_table_mask];
-    for (entry=head->lh_first; entry != NULL; entry=entry->hash_chain.le_next) {
-       if (entry->str == (unsigned char *)nameref) {
-           entry->refcount--;
-           if (entry->refcount == 0) {
-               LIST_REMOVE(entry, hash_chain);
-               if (head->lh_first == NULL) {
-                   filled_buckets--;
-               }
-               ptr = entry->str;
-               entry->str = NULL;
-               nstrings--;
+       lck_mtx_lock_spin(&strcache_mtx_locks[lock_index]);
 
-               FREE(entry, M_TEMP);
-           } else {
-               num_dups--;
-           }
+       for (entry = head->lh_first; entry != NULL; entry = entry->hash_chain.le_next) {
+               if (entry->str == nameref) {
+                       entry->refcount--;
 
-           return 0;
+                       if (entry->refcount == 0) {
+                               LIST_REMOVE(entry, hash_chain);
+
+                               if (head->lh_first == NULL) {
+                                       OSAddAtomic(-1, &filled_buckets);
+                               }
+                       } else {
+                               entry = NULL;
+                       }
+                       retval = 0;
+                       break;
+               }
        }
-    }
+       lck_mtx_unlock(&strcache_mtx_locks[lock_index]);
+       lck_rw_done(strtable_rw_lock);
 
-    return ENOENT;
+       if (entry != NULL)
+               FREE(entry, M_TEMP);
+
+       return retval;
 }
 
 
+#ifdef DUMP_STRING_TABLE
 void
 dump_string_table(void)
 {
@@ -1485,12 +2179,14 @@ dump_string_table(void)
     string_t          *entry;
     u_long            i;
     
-    name_cache_lock();
+    lck_rw_lock_shared(strtable_rw_lock);
+
     for (i = 0; i <= string_table_mask; i++) {
        head = &string_ref_table[i];
        for (entry=head->lh_first; entry != NULL; entry=entry->hash_chain.le_next) {
            printf("%6d - %s\n", entry->refcount, entry->str);
        }
     }
-    name_cache_unlock();
+    lck_rw_done(strtable_rw_lock);
 }
+#endif /* DUMP_STRING_TABLE */