]> git.saurik.com Git - apple/xnu.git/blobdiff - bsd/nfs/nfs_vfsops.c
xnu-792.tar.gz
[apple/xnu.git] / bsd / nfs / nfs_vfsops.c
index 7e852dde977378bf587487028661650dd385c577..6d72e61e05bd9c4fb370f1bd6b3c7d44d6e1f1db 100644 (file)
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2000-2002 Apple Computer, Inc. All rights reserved.
+ * Copyright (c) 2000-2005 Apple Computer, Inc. All rights reserved.
  *
  * @APPLE_LICENSE_HEADER_START@
  * 
 #include <sys/conf.h>
 #include <sys/ioctl.h>
 #include <sys/signal.h>
-#include <sys/proc.h>
-#include <sys/namei.h>
-#include <sys/vnode.h>
+#include <sys/proc_internal.h> /* for fs rooting to update rootdir in fdp */
+#include <sys/kauth.h>
+#include <sys/vnode_internal.h>
 #include <sys/malloc.h>
 #include <sys/kernel.h>
 #include <sys/sysctl.h>
-#include <sys/mount.h>
-#include <sys/buf.h>
-#include <sys/mbuf.h>
+#include <sys/mount_internal.h>
+#include <sys/kpi_mbuf.h>
 #include <sys/socket.h>
 #include <sys/socketvar.h>
+#include <sys/fcntl.h>
+#include <libkern/OSAtomic.h>
 
 #include <sys/vm.h>
 #include <sys/vmparam.h>
 #include <nfs/xdr_subs.h>
 #include <nfs/nfsm_subs.h>
 #include <nfs/nfsdiskless.h>
-#include <nfs/nqnfs.h>
+#include <nfs/nfs_lock.h>
 
-extern int     nfs_mountroot __P((void));
+extern int     nfs_mountroot(void);
 
 extern int     nfs_ticks;
+extern int     nfs_mount_type;
+extern int     nfs_resv_mounts;
 
 struct nfsstats        nfsstats;
-static int nfs_sysctl(int *, u_int, void *, size_t *, void *, size_t,
-                     struct proc *);
+static int nfs_sysctl(int *, u_int, user_addr_t, size_t *, user_addr_t, size_t, vfs_context_t);
 /* XXX CSM 11/25/97 Upgrade sysctl.h someday */
 #ifdef notyet
 SYSCTL_NODE(_vfs, MOUNT_NFS, nfs, CTLFLAG_RW, 0, "NFS filesystem");
 SYSCTL_STRUCT(_vfs_nfs, NFS_NFSSTATS, nfsstats, CTLFLAG_RD,
        &nfsstats, nfsstats, "");
 #endif
-#if NFSDIAG
-int nfs_debug;
-/* XXX CSM 11/25/97 Upgrade sysctl.h someday */
-#ifdef notyet
-SYSCTL_INT(_vfs_nfs, OID_AUTO, debug, CTLFLAG_RW, &nfs_debug, 0, "");
-#endif
-#endif
 
-static int     nfs_iosize __P((struct nfsmount *nmp));
-static int     mountnfs __P((struct nfs_args *,struct mount *,
-                       struct mbuf *,char *,char *,struct vnode **));
-static int     nfs_mount __P(( struct mount *mp, char *path, caddr_t data,
-                       struct nameidata *ndp, struct proc *p));
-static int     nfs_start __P(( struct mount *mp, int flags,
-                       struct proc *p));
-static int     nfs_unmount __P(( struct mount *mp, int mntflags,
-                       struct proc *p));
-static int     nfs_root __P(( struct mount *mp, struct vnode **vpp));
-static int     nfs_quotactl __P(( struct mount *mp, int cmds, uid_t uid,
-                       caddr_t arg, struct proc *p));
-static int     nfs_statfs __P(( struct mount *mp, struct statfs *sbp,
-                       struct proc *p));
-static int     nfs_sync __P(( struct mount *mp, int waitfor,
-                       struct ucred *cred, struct proc *p));
-static int     nfs_vptofh __P(( struct vnode *vp, struct fid *fhp));
-static int     nfs_fhtovp __P((struct mount *mp, struct fid *fhp,
-                       struct mbuf *nam, struct vnode **vpp,
-                       int *exflagsp, struct ucred **credanonp));
-static int     nfs_vget __P((struct mount *, ino_t, struct vnode **));
+SYSCTL_DECL(_vfs_generic_nfs);
+SYSCTL_NODE(_vfs_generic_nfs, OID_AUTO, client, CTLFLAG_RW, 0,
+    "nfs client hinge");
+/* how long NFS will wait before signalling vfs that it's down. */
+static int nfs_tprintf_initial_delay = NFS_TPRINTF_INITIAL_DELAY;
+SYSCTL_INT(_vfs_generic_nfs_client, NFS_TPRINTF_INITIAL_DELAY,
+    initialdowndelay, CTLFLAG_RW, &nfs_tprintf_initial_delay, 0, "");
+/* how long between console messages "nfs server foo not responding" */
+static int nfs_tprintf_delay = NFS_TPRINTF_DELAY;
+SYSCTL_INT(_vfs_generic_nfs_client, NFS_TPRINTF_DELAY,
+    nextdowndelay, CTLFLAG_RW, &nfs_tprintf_delay, 0, "");
+
+static int     nfs_iosize(struct nfsmount *nmp);
+static int     mountnfs(struct user_nfs_args *,mount_t,mbuf_t,proc_t,vnode_t *);
+static int     nfs_mount(mount_t mp, vnode_t vp, user_addr_t data, vfs_context_t context);
+static int     nfs_start(mount_t mp, int flags, vfs_context_t context);
+static int     nfs_unmount(mount_t mp, int mntflags, vfs_context_t context);
+static int     nfs_root(mount_t mp, vnode_t *vpp, vfs_context_t context);
+static int     nfs_statfs(mount_t mp, struct vfsstatfs *sbp, vfs_context_t context);
+static int     nfs_vfs_getattr(mount_t mp, struct vfs_attr *fsap, vfs_context_t context);
+static int     nfs_sync( mount_t mp, int waitfor, vfs_context_t context);
+static int     nfs_vptofh(vnode_t vp, int *fhlenp, unsigned char *fhp, vfs_context_t context);
+static int     nfs_fhtovp(mount_t mp, int fhlen, unsigned char *fhp, vnode_t *vpp, vfs_context_t context);
+static int     nfs_vget(mount_t , ino64_t, vnode_t *, vfs_context_t context);
 
 
 /*
@@ -149,36 +147,24 @@ struct vfsops nfs_vfsops = {
        nfs_start,
        nfs_unmount,
        nfs_root,
-       nfs_quotactl,
-       nfs_statfs,
+       NULL,           /* quotactl */
+       nfs_vfs_getattr,
        nfs_sync,
        nfs_vget,
        nfs_fhtovp,
        nfs_vptofh,
        nfs_init,
-       nfs_sysctl
+       nfs_sysctl,
+       NULL            /* setattr */
 };
-/* XXX CSM 11/25/97 Mysterious kernel.h ld crud */
-#ifdef notyet
-VFS_SET(nfs_vfsops, nfs, MOUNT_NFS, VFCF_NETWORK);
-#endif
 
 
-void nfsargs_ntoh __P((struct nfs_args *));
 static int
-nfs_mount_diskless __P((struct nfs_dlmount *, char *, int, struct vnode **,
-                                               struct mount **));
+nfs_mount_diskless(struct nfs_dlmount *, const char *, int, vnode_t *, mount_t *);
 #if !defined(NO_MOUNT_PRIVATE)
 static int
-nfs_mount_diskless_private __P((struct nfs_dlmount *, char *, int,
-                               struct vnode **, struct mount **));
+nfs_mount_diskless_private(struct nfs_dlmount *, const char *, int, vnode_t *, mount_t *);
 #endif /* NO_MOUNT_PRIVATE */
-static void nfs_convert_oargs __P((struct nfs_args *args,
-                                  struct onfs_args *oargs));
-#if NFSDIAG
-int nfsreqqusers = 0;
-extern int nfsbtlen, nfsbtcpu, nfsbtthread, nfsbt[32];
-#endif
 
 static int nfs_iosize(nmp)
        struct nfsmount* nmp;
@@ -189,99 +175,100 @@ static int nfs_iosize(nmp)
         * Calculate the size used for io buffers.  Use the larger
         * of the two sizes to minimise nfs requests but make sure
         * that it is at least one VM page to avoid wasting buffer
-        * space.
+        * space and to allow easy mmapping of I/O buffers.
+        * The read/write rpc calls handle the splitting up of
+        * buffers into multiple requests if the buffer size is
+        * larger than the I/O size.
         */
        iosize = max(nmp->nm_rsize, nmp->nm_wsize);
        if (iosize < PAGE_SIZE)
                iosize = PAGE_SIZE;
-       return (trunc_page(iosize));
-}
-
-static void nfs_convert_oargs(args,oargs)
-     struct nfs_args *args;
-     struct onfs_args *oargs;
-{
-  args->version = NFS_ARGSVERSION;
-  args->addr = oargs->addr;
-  args->addrlen = oargs->addrlen;
-  args->sotype = oargs->sotype;
-  args->proto = oargs->proto;
-  args->fh = oargs->fh;
-  args->fhsize = oargs->fhsize;
-  args->flags = oargs->flags;
-  args->wsize = oargs->wsize;
-  args->rsize = oargs->rsize;
-  args->readdirsize = oargs->readdirsize;
-  args->timeo = oargs->timeo;
-  args->retrans = oargs->retrans;
-  args->maxgrouplist = oargs->maxgrouplist;
-  args->readahead = oargs->readahead;
-  args->leaseterm = oargs->leaseterm;
-  args->deadthresh = oargs->deadthresh;
-  args->hostname = oargs->hostname;
+       return (trunc_page_32(iosize));
 }
 
 /*
  * nfs statfs call
  */
 int
-nfs_statfs(mp, sbp, p)
-       struct mount *mp;
-       register struct statfs *sbp;
-       struct proc *p;
+nfs_statfs(mount_t mp, struct vfsstatfs *sbp, vfs_context_t context)
 {
-       register struct vnode *vp;
-       register struct nfs_statfs *sfp;
-       register caddr_t cp;
-       register u_long *tl;
-       register long t1, t2;
+       proc_t p = vfs_context_proc(context);
+       vnode_t vp;
+       struct nfs_statfs *sfp;
+       caddr_t cp;
+       u_long *tl;
+       long t1, t2;
        caddr_t bpos, dpos, cp2;
        struct nfsmount *nmp = VFSTONFS(mp);
        int error = 0, v3 = (nmp->nm_flag & NFSMNT_NFSV3), retattr;
-       struct mbuf *mreq, *mrep, *md, *mb, *mb2;
-       struct ucred *cred;
-       u_quad_t tquad;
-       extern int nfs_mount_type;
+       mbuf_t mreq, mrep, md, mb, mb2;
        u_int64_t xid;
+       kauth_cred_t cred;
+       struct ucred temp_cred;
 
 #ifndef nolint
        sfp = (struct nfs_statfs *)0;
 #endif
        vp = nmp->nm_dvp;
-       if (error = vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p))
+       if ((error = vnode_get(vp)))
                return(error);
-       cred = crget();
-       cred->cr_ngroups = 1;
-       if (v3 && (nmp->nm_flag & NFSMNT_GOTFSINFO) == 0)
-               (void)nfs_fsinfo(nmp, vp, cred, p);
-       nfsstats.rpccnt[NFSPROC_FSSTAT]++;
-       nfsm_reqhead(vp, NFSPROC_FSSTAT, NFSX_FH(v3));
+
+       bzero(&temp_cred, sizeof(temp_cred));
+       temp_cred.cr_ngroups = 1;
+       cred = kauth_cred_create(&temp_cred);
+
+       if (v3 && (nmp->nm_state & NFSSTA_GOTFSINFO) == 0)
+               nfs_fsinfo(nmp, vp, cred, p);
+       nfsm_reqhead(NFSX_FH(v3));
+       if (error) {
+               kauth_cred_rele(cred);
+               vnode_put(vp);
+               return (error);
+       }
+       OSAddAtomic(1, (SInt32*)&nfsstats.rpccnt[NFSPROC_FSSTAT]);
        nfsm_fhtom(vp, v3);
        nfsm_request(vp, NFSPROC_FSSTAT, p, cred, &xid);
-       if (v3)
-               nfsm_postop_attr(vp, retattr, &xid);
+       if (v3 && mrep)
+               nfsm_postop_attr_update(vp, v3, retattr, &xid);
        nfsm_dissect(sfp, struct nfs_statfs *, NFSX_STATFS(v3));
 
-/* XXX CSM 12/2/97 Cleanup when/if we integrate FreeBSD mount.h */
-#ifdef notyet
-       sbp->f_type = MOUNT_NFS;
-#else
-       sbp->f_type = nfs_mount_type;
-#endif
        sbp->f_flags = nmp->nm_flag;
        sbp->f_iosize = nfs_iosize(nmp);
        if (v3) {
-               sbp->f_bsize = NFS_FABLKSIZE;
+               /*
+                * Adjust block size to get total block count to fit in a long.
+                * If we can't increase block size enough, clamp to max long.
+                */
+               u_quad_t tquad, tquad2, bsize;
+               bsize = NFS_FABLKSIZE;
+
                fxdr_hyper(&sfp->sf_tbytes, &tquad);
-               sbp->f_blocks = (long)(tquad / ((u_quad_t)NFS_FABLKSIZE));
+               tquad /= bsize;
+               while ((tquad & ~0x7fffffff) && (bsize < 0x40000000)) {
+                       bsize <<= 1;
+                       tquad >>= 1;
+               }
+               sbp->f_blocks = (tquad & ~0x7fffffff) ? 0x7fffffff : (long)tquad;
+
                fxdr_hyper(&sfp->sf_fbytes, &tquad);
-               sbp->f_bfree = (long)(tquad / ((u_quad_t)NFS_FABLKSIZE));
+               tquad /= bsize;
+               sbp->f_bfree = (tquad & ~0x7fffffff) ? 0x7fffffff : (long)tquad;
+
                fxdr_hyper(&sfp->sf_abytes, &tquad);
-               sbp->f_bavail = (long)(tquad / ((u_quad_t)NFS_FABLKSIZE));
-               sbp->f_files = (fxdr_unsigned(long, sfp->sf_tfiles.nfsuquad[1])
-                       & 0x7fffffff);
-               sbp->f_ffree = (fxdr_unsigned(long, sfp->sf_ffiles.nfsuquad[1])
-                       & 0x7fffffff);
+               tquad /= bsize;
+               sbp->f_bavail = (tquad & ~0x7fffffff) ? 0x7fffffff : (long)tquad;
+
+               sbp->f_bsize = (long)bsize;
+
+               /* adjust file slots too... */
+               fxdr_hyper(&sfp->sf_tfiles, &tquad);
+               fxdr_hyper(&sfp->sf_ffiles, &tquad2);
+               while (tquad & ~0x7fffffff) {
+                       tquad >>= 1;
+                       tquad2 >>= 1;
+               }
+               sbp->f_files = tquad;
+               sbp->f_ffree = tquad2;
        } else {
                sbp->f_bsize = fxdr_unsigned(long, sfp->sf_bsize);
                sbp->f_blocks = fxdr_unsigned(long, sfp->sf_blocks);
@@ -290,13 +277,226 @@ nfs_statfs(mp, sbp, p)
                sbp->f_files = 0;
                sbp->f_ffree = 0;
        }
-       if (sbp != &mp->mnt_stat) {
-               bcopy(mp->mnt_stat.f_mntonname, sbp->f_mntonname, MNAMELEN);
-               bcopy(mp->mnt_stat.f_mntfromname, sbp->f_mntfromname, MNAMELEN);
-       }
        nfsm_reqdone;
-       VOP_UNLOCK(vp, 0, p);
-       crfree(cred);
+       kauth_cred_rele(cred);
+       vnode_put(vp);
+       return (error);
+}
+
+/*
+ * The nfs_statfs code is complicated, and used by mountnfs(), so leave it as-is
+ * and handle VFS_GETATTR by calling nfs_statfs and copying fields.
+ */
+static int
+nfs_vfs_getattr(mount_t mp, struct vfs_attr *fsap, vfs_context_t context)
+{
+       int error = 0;
+       
+       if (VFSATTR_IS_ACTIVE(fsap, f_bsize)  ||
+           VFSATTR_IS_ACTIVE(fsap, f_iosize) ||
+           VFSATTR_IS_ACTIVE(fsap, f_blocks) ||
+           VFSATTR_IS_ACTIVE(fsap, f_bfree)  ||
+           VFSATTR_IS_ACTIVE(fsap, f_bavail) ||
+           VFSATTR_IS_ACTIVE(fsap, f_bused)  ||
+           VFSATTR_IS_ACTIVE(fsap, f_files)  ||
+           VFSATTR_IS_ACTIVE(fsap, f_ffree)) {
+               struct vfsstatfs sb;
+
+               error = nfs_statfs(mp, &sb, context);
+               if (!error) {
+                       VFSATTR_RETURN(fsap, f_bsize, sb.f_bsize);
+                       VFSATTR_RETURN(fsap, f_iosize, sb.f_iosize);
+                       VFSATTR_RETURN(fsap, f_blocks, sb.f_blocks);
+                       VFSATTR_RETURN(fsap, f_bfree, sb.f_bfree);
+                       VFSATTR_RETURN(fsap, f_bavail, sb.f_bavail);
+                       VFSATTR_RETURN(fsap, f_bused, sb.f_blocks - sb.f_bfree);
+                       VFSATTR_RETURN(fsap, f_files, sb.f_files);
+                       VFSATTR_RETURN(fsap, f_ffree, sb.f_ffree);
+               }
+       }
+
+       if (VFSATTR_IS_ACTIVE(fsap, f_capabilities)) {
+               struct nfsmount *nmp;
+               struct nfsv3_pathconf pc;
+               u_int32_t caps, valid;
+               vnode_t vp;
+               int v3;
+
+               if (!(nmp = VFSTONFS(mp)))
+                       return (ENXIO);
+               vp = nmp->nm_dvp;
+               v3 = (nmp->nm_flag & NFSMNT_NFSV3);
+
+               /*
+                * The capabilities[] array defines what this volume supports.
+                *
+                * The valid[] array defines which bits this code understands
+                * the meaning of (whether the volume has that capability or not).
+                * Any zero bits here means "I don't know what you're asking about"
+                * and the caller cannot tell whether that capability is
+                * present or not.
+                */
+               caps = valid = 0;
+               if (v3) {
+                       /* try to get fsinfo if we haven't already */
+                       if (!(nmp->nm_state & NFSSTA_GOTFSINFO)) {
+                               nfs_fsinfo(nmp, vp, vfs_context_ucred(context),
+                                       vfs_context_proc(context));
+                               if (!(nmp = VFSTONFS(vnode_mount(vp))))
+                                       return (ENXIO);
+                       }
+                       if (nmp->nm_state & NFSSTA_GOTFSINFO) {
+                               /* fsinfo indicates (non)support of links and symlinks */
+                               valid |= VOL_CAP_FMT_SYMBOLICLINKS |
+                                        VOL_CAP_FMT_HARDLINKS;
+                               if (nmp->nm_fsinfo.fsproperties & NFSV3FSINFO_SYMLINK)
+                                       caps |= VOL_CAP_FMT_SYMBOLICLINKS;
+                               if (nmp->nm_fsinfo.fsproperties & NFSV3FSINFO_LINK)
+                                       caps |= VOL_CAP_FMT_HARDLINKS;
+                               /* if fsinfo indicates all pathconf info is the same, */
+                               /* we can use it to report case attributes */
+                               if ((nmp->nm_fsinfo.fsproperties & NFSV3FSINFO_HOMOGENEOUS) &&
+                                   !(nmp->nm_state & NFSSTA_GOTPATHCONF)) {
+                                       /* no cached pathconf info, try to get now */
+                                       error = nfs_pathconfrpc(vp, &pc,
+                                                       vfs_context_ucred(context),
+                                                       vfs_context_proc(context));
+                                       if (!(nmp = VFSTONFS(vnode_mount(vp))))
+                                               return (ENXIO);
+                                       if (!error) {
+                                               /* all files have the same pathconf info, */
+                                               /* so cache a copy of the results */
+                                               nfs_pathconf_cache(nmp, &pc);
+                                       }
+                               }
+                               if (nmp->nm_state & NFSSTA_GOTPATHCONF) {
+                                       valid |= VOL_CAP_FMT_CASE_SENSITIVE |
+                                                VOL_CAP_FMT_CASE_PRESERVING;
+                                       if (!(nmp->nm_fsinfo.pcflags &
+                                               NFSPCINFO_CASE_INSENSITIVE))
+                                               caps |= VOL_CAP_FMT_CASE_SENSITIVE;
+                                       if (nmp->nm_fsinfo.pcflags &
+                                               NFSPCINFO_CASE_PRESERVING)
+                                               caps |= VOL_CAP_FMT_CASE_PRESERVING;
+                               }
+                               /* Is server's max file size at least 2TB? */
+                               if (nmp->nm_fsinfo.maxfilesize >= 0x20000000000ULL)
+                                       caps |= VOL_CAP_FMT_2TB_FILESIZE;
+                       } else {
+                               /*
+                                * NFSv3 supports 64 bits of file size.
+                                * Without FSINFO from the server, we'll
+                                * just assume maxfilesize >= 2TB
+                                */
+                               caps |= VOL_CAP_FMT_2TB_FILESIZE;
+                       }
+               }
+               fsap->f_capabilities.capabilities[VOL_CAPABILITIES_FORMAT] =
+                       // VOL_CAP_FMT_PERSISTENTOBJECTIDS |
+                       // VOL_CAP_FMT_SYMBOLICLINKS |
+                       // VOL_CAP_FMT_HARDLINKS |
+                       // VOL_CAP_FMT_JOURNAL |
+                       // VOL_CAP_FMT_JOURNAL_ACTIVE |
+                       // VOL_CAP_FMT_NO_ROOT_TIMES |
+                       // VOL_CAP_FMT_SPARSE_FILES |
+                       // VOL_CAP_FMT_ZERO_RUNS |
+                       // VOL_CAP_FMT_CASE_SENSITIVE |
+                       // VOL_CAP_FMT_CASE_PRESERVING |
+                       // VOL_CAP_FMT_FAST_STATFS |
+                       // VOL_CAP_FMT_2TB_FILESIZE |
+                       caps;
+               fsap->f_capabilities.valid[VOL_CAPABILITIES_FORMAT] =
+                       VOL_CAP_FMT_PERSISTENTOBJECTIDS |
+                       // VOL_CAP_FMT_SYMBOLICLINKS |
+                       // VOL_CAP_FMT_HARDLINKS |
+                       // VOL_CAP_FMT_JOURNAL |
+                       // VOL_CAP_FMT_JOURNAL_ACTIVE |
+                       // VOL_CAP_FMT_NO_ROOT_TIMES |
+                       // VOL_CAP_FMT_SPARSE_FILES |
+                       // VOL_CAP_FMT_ZERO_RUNS |
+                       // VOL_CAP_FMT_CASE_SENSITIVE |
+                       // VOL_CAP_FMT_CASE_PRESERVING |
+                       VOL_CAP_FMT_FAST_STATFS |
+                       VOL_CAP_FMT_2TB_FILESIZE |
+                       valid;
+
+               /*
+                * We don't support most of the interfaces.
+                *
+                * We MAY support locking, but we don't have any easy way of probing.
+                * We can tell if there's no lockd running or if locks have been
+                * disabled for a mount, so we can definitely answer NO in that case.
+                * Any attempt to send a request to lockd to test for locking support
+                * may cause the lazily-launched locking daemons to be started
+                * unnecessarily.  So we avoid that.  However, we do record if we ever
+                * successfully perform a lock operation on a mount point, so if it
+                * looks like lock ops have worked, we do report that we support them.
+                */
+               caps = valid = 0;
+               if ((!nfslockdvnode && !nfslockdwaiting) ||
+                   (nmp->nm_flag & NFSMNT_NOLOCKS)) {
+                       /* locks disabled on this mount, so they definitely won't work */
+                       valid = VOL_CAP_INT_ADVLOCK | VOL_CAP_INT_FLOCK;
+               } else if (nmp->nm_state & NFSSTA_LOCKSWORK) {
+                       caps = VOL_CAP_INT_ADVLOCK | VOL_CAP_INT_FLOCK;
+                       valid = VOL_CAP_INT_ADVLOCK | VOL_CAP_INT_FLOCK;
+               }
+               fsap->f_capabilities.capabilities[VOL_CAPABILITIES_INTERFACES] =
+                       // VOL_CAP_INT_SEARCHFS |
+                       // VOL_CAP_INT_ATTRLIST |
+                       // VOL_CAP_INT_NFSEXPORT |
+                       // VOL_CAP_INT_READDIRATTR |
+                       // VOL_CAP_INT_EXCHANGEDATA |
+                       // VOL_CAP_INT_COPYFILE |
+                       // VOL_CAP_INT_ALLOCATE |
+                       // VOL_CAP_INT_VOL_RENAME |
+                       // VOL_CAP_INT_ADVLOCK |
+                       // VOL_CAP_INT_FLOCK |
+                       // VOL_CAP_INT_EXTENDED_SECURITY |
+                       // VOL_CAP_INT_USERACCESS |
+                       caps;
+               fsap->f_capabilities.valid[VOL_CAPABILITIES_INTERFACES] =
+                       VOL_CAP_INT_SEARCHFS |
+                       VOL_CAP_INT_ATTRLIST |
+                       VOL_CAP_INT_NFSEXPORT |
+                       VOL_CAP_INT_READDIRATTR |
+                       VOL_CAP_INT_EXCHANGEDATA |
+                       VOL_CAP_INT_COPYFILE |
+                       VOL_CAP_INT_ALLOCATE |
+                       VOL_CAP_INT_VOL_RENAME |
+                       // VOL_CAP_INT_ADVLOCK |
+                       // VOL_CAP_INT_FLOCK |
+                       // VOL_CAP_INT_EXTENDED_SECURITY |
+                       // VOL_CAP_INT_USERACCESS |
+                       valid;
+
+               fsap->f_capabilities.capabilities[VOL_CAPABILITIES_RESERVED1] = 0;
+               fsap->f_capabilities.valid[VOL_CAPABILITIES_RESERVED1] = 0;
+
+               fsap->f_capabilities.capabilities[VOL_CAPABILITIES_RESERVED2] = 0;
+               fsap->f_capabilities.valid[VOL_CAPABILITIES_RESERVED2] = 0;
+
+               VFSATTR_SET_SUPPORTED(fsap, f_capabilities);
+       }
+
+       if (VFSATTR_IS_ACTIVE(fsap, f_attributes)) {
+               fsap->f_attributes.validattr.commonattr = 0;
+               fsap->f_attributes.validattr.volattr =
+                       ATTR_VOL_CAPABILITIES | ATTR_VOL_ATTRIBUTES;
+               fsap->f_attributes.validattr.dirattr = 0;
+               fsap->f_attributes.validattr.fileattr = 0;
+               fsap->f_attributes.validattr.forkattr = 0;
+
+               fsap->f_attributes.nativeattr.commonattr = 0;
+               fsap->f_attributes.nativeattr.volattr =
+                       ATTR_VOL_CAPABILITIES | ATTR_VOL_ATTRIBUTES;
+               fsap->f_attributes.nativeattr.dirattr = 0;
+               fsap->f_attributes.nativeattr.fileattr = 0;
+               fsap->f_attributes.nativeattr.forkattr = 0;
+
+               VFSATTR_SET_SUPPORTED(fsap, f_attributes);
+       }
+
        return (error);
 }
 
@@ -305,54 +505,61 @@ nfs_statfs(mp, sbp, p)
  */
 int
 nfs_fsinfo(nmp, vp, cred, p)
-       register struct nfsmount *nmp;
-       register struct vnode *vp;
-       struct ucred *cred;
-       struct proc *p;
+       struct nfsmount *nmp;
+       vnode_t vp;
+       kauth_cred_t cred;
+       proc_t p;
 {
-       register struct nfsv3_fsinfo *fsp;
-       register caddr_t cp;
-       register long t1, t2;
-       register u_long *tl, pref, max;
+       struct nfsv3_fsinfo *fsp;
+       caddr_t cp;
+       long t1, t2;
+       u_long *tl;
+       int prefsize, maxsize;
        caddr_t bpos, dpos, cp2;
        int error = 0, retattr;
-       struct mbuf *mreq, *mrep, *md, *mb, *mb2;
+       mbuf_t mreq, mrep, md, mb, mb2;
        u_int64_t xid;
 
-       nfsstats.rpccnt[NFSPROC_FSINFO]++;
-       nfsm_reqhead(vp, NFSPROC_FSINFO, NFSX_FH(1));
+       nfsm_reqhead(NFSX_FH(1));
+       if (error)
+               return (error);
+       OSAddAtomic(1, (SInt32*)&nfsstats.rpccnt[NFSPROC_FSINFO]);
        nfsm_fhtom(vp, 1);
        nfsm_request(vp, NFSPROC_FSINFO, p, cred, &xid);
-       nfsm_postop_attr(vp, retattr, &xid);
+       if (mrep) {
+               nfsm_postop_attr_update(vp, 1, retattr, &xid);
+       }
        if (!error) {
                nfsm_dissect(fsp, struct nfsv3_fsinfo *, NFSX_V3FSINFO);
-               pref = fxdr_unsigned(u_long, fsp->fs_wtpref);
-               if (pref < nmp->nm_wsize)
-                       nmp->nm_wsize = (pref + NFS_FABLKSIZE - 1) &
+               prefsize = fxdr_unsigned(u_long, fsp->fs_wtpref);
+               if (prefsize < nmp->nm_wsize)
+                       nmp->nm_wsize = (prefsize + NFS_FABLKSIZE - 1) &
                                ~(NFS_FABLKSIZE - 1);
-               max = fxdr_unsigned(u_long, fsp->fs_wtmax);
-               if (max < nmp->nm_wsize) {
-                       nmp->nm_wsize = max & ~(NFS_FABLKSIZE - 1);
+               maxsize = fxdr_unsigned(u_long, fsp->fs_wtmax);
+               if (maxsize < nmp->nm_wsize) {
+                       nmp->nm_wsize = maxsize & ~(NFS_FABLKSIZE - 1);
                        if (nmp->nm_wsize == 0)
-                               nmp->nm_wsize = max;
+                               nmp->nm_wsize = maxsize;
                }
-               pref = fxdr_unsigned(u_long, fsp->fs_rtpref);
-               if (pref < nmp->nm_rsize)
-                       nmp->nm_rsize = (pref + NFS_FABLKSIZE - 1) &
+               prefsize = fxdr_unsigned(u_long, fsp->fs_rtpref);
+               if (prefsize < nmp->nm_rsize)
+                       nmp->nm_rsize = (prefsize + NFS_FABLKSIZE - 1) &
                                ~(NFS_FABLKSIZE - 1);
-               max = fxdr_unsigned(u_long, fsp->fs_rtmax);
-               if (max < nmp->nm_rsize) {
-                       nmp->nm_rsize = max & ~(NFS_FABLKSIZE - 1);
+               maxsize = fxdr_unsigned(u_long, fsp->fs_rtmax);
+               if (maxsize < nmp->nm_rsize) {
+                       nmp->nm_rsize = maxsize & ~(NFS_FABLKSIZE - 1);
                        if (nmp->nm_rsize == 0)
-                               nmp->nm_rsize = max;
+                               nmp->nm_rsize = maxsize;
                }
-               pref = fxdr_unsigned(u_long, fsp->fs_dtpref);
-               if (pref < nmp->nm_readdirsize)
-                       nmp->nm_readdirsize = pref;
-               if (max < nmp->nm_readdirsize) {
-                       nmp->nm_readdirsize = max;
+               prefsize = fxdr_unsigned(u_long, fsp->fs_dtpref);
+               if (prefsize < nmp->nm_readdirsize)
+                       nmp->nm_readdirsize = prefsize;
+               if (maxsize < nmp->nm_readdirsize) {
+                       nmp->nm_readdirsize = maxsize;
                }
-               nmp->nm_flag |= NFSMNT_GOTFSINFO;
+               fxdr_hyper(&fsp->fs_maxfilesize, &nmp->nm_fsinfo.maxfilesize);
+               nmp->nm_fsinfo.fsproperties = fxdr_unsigned(u_long, fsp->fs_properties);
+               nmp->nm_state |= NFSSTA_GOTFSINFO;
        }
        nfsm_reqdone;
        return (error);
@@ -375,17 +582,16 @@ int
 nfs_mountroot()
 {
        struct nfs_diskless nd;
-       struct vattr attr;
-       struct mount *mp;
-       struct vnode *vp;
-       struct proc *procp;
-       long n;
+       struct nfs_vattr nvattr;
+       mount_t mp;
+       vnode_t vp;
+       proc_t procp;
        int error;
 #if !defined(NO_MOUNT_PRIVATE)
-       struct mount *mppriv;
-       struct vnode *vppriv;
+       mount_t mppriv;
+       vnode_t vppriv;
 #endif /* NO_MOUNT_PRIVATE */
-       int v3;
+       int v3, sotype;
 
        procp = current_proc(); /* XXX */
 
@@ -400,56 +606,86 @@ nfs_mountroot()
                panic("nfs_boot_init failed with %d\n", error);
        }
 
-       /* try NFSv3 first, if that fails then try NFSv2 */
+       /*
+        * Try NFSv3 first, then fallback to NFSv2.
+        * Likewise, try TCP first, then fall back to UDP.
+        */
        v3 = 1;
+       sotype = SOCK_STREAM;
 
 tryagain:
-       error = nfs_boot_getfh(&nd, procp, v3);
+       error = nfs_boot_getfh(&nd, procp, v3, sotype);
        if (error) {
+               if (error == EHOSTDOWN || error == EHOSTUNREACH) {
+                       if (nd.nd_root.ndm_path)
+                               FREE_ZONE(nd.nd_root.ndm_path, 
+                                         MAXPATHLEN, M_NAMEI);
+                       if (nd.nd_private.ndm_path)
+                               FREE_ZONE(nd.nd_private.ndm_path, 
+                                         MAXPATHLEN, M_NAMEI);
+                       return (error);
+               }
                if (v3) {
-                       printf("nfs_boot_getfh(v3) failed with %d, trying v2...\n", error);
+                       if (sotype == SOCK_STREAM) {
+                               printf("nfs_boot_getfh(v3,TCP) failed with %d, trying UDP...\n", error);
+                               sotype = SOCK_DGRAM;
+                               goto tryagain;
+                       }
+                       printf("nfs_boot_getfh(v3,UDP) failed with %d, trying v2...\n", error);
                        v3 = 0;
+                       sotype = SOCK_STREAM;
+                       goto tryagain;
+               } else if (sotype == SOCK_STREAM) {
+                       printf("nfs_boot_getfh(v2,TCP) failed with %d, trying UDP...\n", error);
+                       sotype = SOCK_DGRAM;
                        goto tryagain;
                }
-               panic("nfs_boot_getfh(v2) failed with %d\n", error);
+               panic("nfs_boot_getfh(v2,UDP) failed with %d\n", error);
        }
 
        /*
         * Create the root mount point.
         */
 #if !defined(NO_MOUNT_PRIVATE)
-       if ((error = nfs_mount_diskless(&nd.nd_root, "/", MNT_RDONLY, &vp, &mp))) {
+       if ((error = nfs_mount_diskless(&nd.nd_root, "/", MNT_RDONLY|MNT_ROOTFS, &vp, &mp)))
 #else
-       if (error = nfs_mount_diskless(&nd.nd_root, "/", NULL, &vp, &mp)) {
+       if ((error = nfs_mount_diskless(&nd.nd_root, "/", MNT_ROOTFS, &vp, &mp)))
 #endif /* NO_MOUNT_PRIVATE */
+       {
                if (v3) {
-                       printf("nfs_mount_diskless(v3) failed with %d, trying v2...\n", error);
+                       if (sotype == SOCK_STREAM) {
+                               printf("nfs_mount_diskless(v3,TCP) failed with %d, trying UDP...\n", error);
+                               sotype = SOCK_DGRAM;
+                               goto tryagain;
+                       }
+                       printf("nfs_mount_diskless(v3,UDP) failed with %d, trying v2...\n", error);
                        v3 = 0;
+                       sotype = SOCK_STREAM;
+                       goto tryagain;
+               } else if (sotype == SOCK_STREAM) {
+                       printf("nfs_mount_diskless(v2,TCP) failed with %d, trying UDP...\n", error);
+                       sotype = SOCK_DGRAM;
                        goto tryagain;
                }
-               panic("nfs_mount_diskless root failed with %d\n", error);
+               panic("nfs_mount_diskless(v2,UDP) root failed with %d\n", error);
        }
        printf("root on %s\n", (char *)&nd.nd_root.ndm_host);
 
-       simple_lock(&mountlist_slock);
-       CIRCLEQ_INSERT_TAIL(&mountlist, mp, mnt_list);
-       simple_unlock(&mountlist_slock);
-       vfs_unbusy(mp, procp);
+       vfs_unbusy(mp);
+       mount_list_add(mp);
        rootvp = vp;
        
 #if !defined(NO_MOUNT_PRIVATE)
        if (nd.nd_private.ndm_saddr.sin_addr.s_addr) {
            error = nfs_mount_diskless_private(&nd.nd_private, "/private",
-                                              NULL, &vppriv, &mppriv);
+                                              0, &vppriv, &mppriv);
            if (error) {
                panic("nfs_mount_diskless private failed with %d\n", error);
            }
            printf("private on %s\n", (char *)&nd.nd_private.ndm_host);
            
-           simple_lock(&mountlist_slock);
-           CIRCLEQ_INSERT_TAIL(&mountlist, mppriv, mnt_list);
-           simple_unlock(&mountlist_slock);
-           vfs_unbusy(mppriv, procp);
+           vfs_unbusy(mppriv);
+           mount_list_add(mppriv);
        }
 
 #endif /* NO_MOUNT_PRIVATE */
@@ -460,10 +696,8 @@ tryagain:
                FREE_ZONE(nd.nd_private.ndm_path, MAXPATHLEN, M_NAMEI);
 
        /* Get root attributes (for the time). */
-       error = VOP_GETATTR(vp, &attr, procp->p_ucred, procp);
+       error = nfs_getattr(vp, &nvattr, kauth_cred_get(), procp);
        if (error) panic("nfs_mountroot: getattr for root");
-       n = attr.va_mtime.tv_sec;
-       inittodr(n);
        return (0);
 }
 
@@ -471,52 +705,60 @@ tryagain:
  * Internal version of mount system call for diskless setup.
  */
 static int
-nfs_mount_diskless(ndmntp, mntname, mntflag, vpp, mpp)
-       struct nfs_dlmount *ndmntp;
-       char *mntname;
-       int mntflag;
-       struct vnode **vpp;
-       struct mount **mpp;
+nfs_mount_diskless(
+       struct nfs_dlmount *ndmntp,
+       const char *mntname,
+       int mntflag,
+       vnode_t *vpp,
+       mount_t *mpp)
 {
-       struct nfs_args args;
-       struct mount *mp;
-       struct mbuf *m;
+       struct user_nfs_args args;
+       mount_t mp;
+       mbuf_t m;
        int error;
-       struct proc *procp;
+       proc_t procp;
 
        procp = current_proc(); /* XXX */
 
        if ((error = vfs_rootmountalloc("nfs", ndmntp->ndm_host, &mp))) {
-               printf("nfs_mountroot: NFS not configured");
+               printf("nfs_mount_diskless: NFS not configured");
                return (error);
        }
-       mp->mnt_flag = mntflag;
+
+       mp->mnt_flag |= mntflag;
+       if (!(mntflag & MNT_RDONLY))
+               mp->mnt_flag &= ~MNT_RDONLY;
 
        /* Initialize mount args. */
        bzero((caddr_t) &args, sizeof(args));
-       args.addr     = (struct sockaddr *)&ndmntp->ndm_saddr;
-       args.addrlen  = args.addr->sa_len;
-       args.sotype   = SOCK_DGRAM;
-       args.fh       = ndmntp->ndm_fh;
+       args.addr     = CAST_USER_ADDR_T(&ndmntp->ndm_saddr);
+       args.addrlen  = ndmntp->ndm_saddr.sin_len;
+       args.sotype   = ndmntp->ndm_sotype;
+       args.fh       = CAST_USER_ADDR_T(&ndmntp->ndm_fh[0]);
        args.fhsize   = ndmntp->ndm_fhlen;
-       args.hostname = ndmntp->ndm_host;
+       args.hostname = CAST_USER_ADDR_T(ndmntp->ndm_host);
        args.flags    = NFSMNT_RESVPORT;
        if (ndmntp->ndm_nfsv3)
                args.flags |= NFSMNT_NFSV3;
 
-       MGET(m, M_DONTWAIT, MT_SONAME);
-       bcopy((caddr_t)args.addr, mtod(m, caddr_t), 
-                               (m->m_len = args.addr->sa_len));
-       if ((error = mountnfs(&args, mp, m, mntname, args.hostname, vpp))) {
-               printf("nfs_mountroot: mount %s failed: %d", mntname, error);
-               mp->mnt_vfc->vfc_refcount--;
-               vfs_unbusy(mp, procp);
-               _FREE_ZONE(mp, sizeof (struct mount), M_MOUNT);
+       error = mbuf_get(MBUF_WAITOK, MBUF_TYPE_SONAME, &m);
+       if (error) {
+               printf("nfs_mount_diskless: mbuf_get(soname) failed");
+               return (error);
+       }
+       mbuf_setlen(m, ndmntp->ndm_saddr.sin_len);
+       bcopy((caddr_t)args.addr, mbuf_data(m), ndmntp->ndm_saddr.sin_len);
+       if ((error = mountnfs(&args, mp, m, procp, vpp))) {
+               printf("nfs_mountroot: mount %s failed: %d\n", mntname, error);
+               // XXX vfs_rootmountfailed(mp);
+               mount_list_lock();
+               mp->mnt_vtable->vfc_refcount--;
+               mount_list_unlock();
+               vfs_unbusy(mp);
+               mount_lock_destroy(mp);
+               FREE_ZONE(mp, sizeof(struct mount), M_MOUNT);
                return (error);
        }
-#if 0  /* Causes incorrect reporting of "mounted on" */
-       (void) copystr(args.hostname, mp->mnt_stat.f_mntonname, MNAMELEN - 1, 0);
-#endif /* 0 */
        *mpp = mp;
        return (0);
 }
@@ -527,23 +769,26 @@ nfs_mount_diskless(ndmntp, mntname, mntflag, vpp, mpp)
  * separately in diskless setup
  */
 static int
-nfs_mount_diskless_private(ndmntp, mntname, mntflag, vpp, mpp)
-       struct nfs_dlmount *ndmntp;
-       char *mntname;
-       int mntflag;
-       struct vnode **vpp;
-       struct mount **mpp;
+nfs_mount_diskless_private(
+       struct nfs_dlmount *ndmntp,
+       const char *mntname,
+       int mntflag,
+       vnode_t *vpp,
+       mount_t *mpp)
 {
-       struct nfs_args args;
-       struct mount *mp;
-       struct mbuf *m;
+       struct user_nfs_args args;
+       mount_t mp;
+       mbuf_t m;
        int error;
-       struct proc *procp;
-       struct vfsconf *vfsp;
+       proc_t procp;
+       struct vfstable *vfsp;
        struct nameidata nd;
-       struct vnode *vp;
+       vnode_t vp;
+       struct vfs_context context;
 
        procp = current_proc(); /* XXX */
+       context.vc_proc = procp;
+       context.vc_ucred = kauth_cred_get();
 
        {
        /*
@@ -552,49 +797,55 @@ nfs_mount_diskless_private(ndmntp, mntname, mntflag, vpp, mpp)
         */
                struct filedesc *fdp;   /* pointer to file descriptor state */
                fdp = procp->p_fd;
-               mountlist.cqh_first->mnt_flag |= MNT_ROOTFS;
+               mountlist.tqh_first->mnt_flag |= MNT_ROOTFS;
 
                /* Get the vnode for '/'. Set fdp->fd_cdir to reference it. */
-               if (VFS_ROOT(mountlist.cqh_first, &rootvnode))
+               if (VFS_ROOT(mountlist.tqh_first, &rootvnode, NULL))
                        panic("cannot find root vnode");
-               VREF(rootvnode);
+               error = vnode_ref(rootvnode);
+               if (error) {
+                       printf("nfs_mountroot: vnode_ref() failed on root vnode!\n");
+                       return (error);
+               }
                fdp->fd_cdir = rootvnode;
-               VOP_UNLOCK(rootvnode, 0, procp);
                fdp->fd_rdir = NULL;
        }
 
        /*
         * Get vnode to be covered
         */
-       NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE,
-           mntname, procp);
+       NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE32,
+           mntname, &context);
        if ((error = namei(&nd))) {
-               printf("nfs_mountroot: private namei failed!");
+               printf("nfs_mountroot: private namei failed!\n");
                return (error);
        }
        {
-               /* undo VREF in mimic main()! */
-               vrele(rootvnode);
+               /* undo vnode_ref() in mimic main()! */
+               vnode_rele(rootvnode);
        }
+       nameidone(&nd);
        vp = nd.ni_vp;
-       if ((error = vinvalbuf(vp, V_SAVE, procp->p_ucred, procp, 0, 0))) {
-               vput(vp);
+
+       if ((error = VNOP_FSYNC(vp, MNT_WAIT, &context)) ||
+           (error = buf_invalidateblks(vp, BUF_WRITE_DATA, 0, 0))) {
+               vnode_put(vp);
                return (error);
        }
-       if (vp->v_type != VDIR) {
-               vput(vp);
+       if (vnode_vtype(vp) != VDIR) {
+               vnode_put(vp);
                return (ENOTDIR);
        }
        for (vfsp = vfsconf; vfsp; vfsp = vfsp->vfc_next)
                if (!strcmp(vfsp->vfc_name, "nfs"))
                        break;
        if (vfsp == NULL) {
-               printf("nfs_mountroot: private NFS not configured");
-               vput(vp);
+               printf("nfs_mountroot: private NFS not configured\n");
+               vnode_put(vp);
                return (ENODEV);
        }
-       if (vp->v_mountedhere != NULL) {
-               vput(vp);
+       if (vnode_mountedhere(vp) != NULL) {
+               vnode_put(vp);
                return (EBUSY);
        }
 
@@ -602,48 +853,65 @@ nfs_mount_diskless_private(ndmntp, mntname, mntflag, vpp, mpp)
         * Allocate and initialize the filesystem.
         */
        mp = _MALLOC_ZONE((u_long)sizeof(struct mount), M_MOUNT, M_WAITOK);
+       if (!mp) {
+               printf("nfs_mountroot: unable to allocate mount structure\n");
+               vnode_put(vp);
+               return (ENOMEM);
+       }
        bzero((char *)mp, (u_long)sizeof(struct mount));
 
-    /* Initialize the default IO constraints */
-    mp->mnt_maxreadcnt = mp->mnt_maxwritecnt = MAXPHYS;
-    mp->mnt_segreadcnt = mp->mnt_segwritecnt = 32;
-
-       lockinit(&mp->mnt_lock, PVFS, "vfslock", 0, 0);
-       (void)vfs_busy(mp, LK_NOWAIT, 0, procp);
-       LIST_INIT(&mp->mnt_vnodelist);
-       mp->mnt_op = vfsp->vfc_vfsops;
-       mp->mnt_vfc = vfsp;
+       /* Initialize the default IO constraints */
+       mp->mnt_maxreadcnt = mp->mnt_maxwritecnt = MAXPHYS;
+       mp->mnt_segreadcnt = mp->mnt_segwritecnt = 32;
+
+       mount_lock_init(mp);
+       TAILQ_INIT(&mp->mnt_vnodelist);
+       TAILQ_INIT(&mp->mnt_workerqueue);
+       TAILQ_INIT(&mp->mnt_newvnodes);
+       (void)vfs_busy(mp, LK_NOWAIT);
+       TAILQ_INIT(&mp->mnt_vnodelist);
+       mount_list_lock();
        vfsp->vfc_refcount++;
-       mp->mnt_stat.f_type = vfsp->vfc_typenum;
+       mount_list_unlock();
+       mp->mnt_vtable = vfsp;
+       mp->mnt_op = vfsp->vfc_vfsops;
+       // mp->mnt_stat.f_type = vfsp->vfc_typenum;
        mp->mnt_flag = mntflag;
        mp->mnt_flag |= vfsp->vfc_flags & MNT_VISFLAGMASK;
-       strncpy(mp->mnt_stat.f_fstypename, vfsp->vfc_name, MFSNAMELEN);
+       strncpy(mp->mnt_vfsstat.f_fstypename, vfsp->vfc_name, MFSNAMELEN-1);
        vp->v_mountedhere = mp;
        mp->mnt_vnodecovered = vp;
-       mp->mnt_stat.f_owner = procp->p_ucred->cr_uid;
-       (void) copystr(mntname, mp->mnt_stat.f_mntonname, MNAMELEN - 1, 0);
-       (void) copystr(ndmntp->ndm_host, mp->mnt_stat.f_mntfromname, MNAMELEN - 1, 0);
+       mp->mnt_vfsstat.f_owner = kauth_cred_getuid(kauth_cred_get());
+       (void) copystr(mntname, mp->mnt_vfsstat.f_mntonname, MNAMELEN - 1, 0);
+       (void) copystr(ndmntp->ndm_host, mp->mnt_vfsstat.f_mntfromname, MNAMELEN - 1, 0);
 
        /* Initialize mount args. */
        bzero((caddr_t) &args, sizeof(args));
-       args.addr     = (struct sockaddr *)&ndmntp->ndm_saddr;
-       args.addrlen  = args.addr->sa_len;
-       args.sotype   = SOCK_DGRAM;
-       args.fh       = ndmntp->ndm_fh;
+       args.addr     = CAST_USER_ADDR_T(&ndmntp->ndm_saddr);
+       args.addrlen  = ndmntp->ndm_saddr.sin_len;
+       args.sotype   = ndmntp->ndm_sotype;
+       args.fh       = CAST_USER_ADDR_T(ndmntp->ndm_fh);
        args.fhsize   = ndmntp->ndm_fhlen;
-       args.hostname = ndmntp->ndm_host;
+       args.hostname = CAST_USER_ADDR_T(ndmntp->ndm_host);
        args.flags    = NFSMNT_RESVPORT;
        if (ndmntp->ndm_nfsv3)
                args.flags |= NFSMNT_NFSV3;
 
-       MGET(m, M_DONTWAIT, MT_SONAME);
-       bcopy((caddr_t)args.addr, mtod(m, caddr_t), 
-                               (m->m_len = args.addr->sa_len));
-       if ((error = mountnfs(&args, mp, m, mntname, args.hostname, &vp))) {
-               printf("nfs_mountroot: mount %s failed: %d", mntname, error);
-               mp->mnt_vfc->vfc_refcount--;
-               vfs_unbusy(mp, procp);
-               _FREE_ZONE(mp, sizeof (struct mount), M_MOUNT);
+       error = mbuf_get(MBUF_WAITOK, MBUF_TYPE_SONAME, &m);
+       if (error) {
+               printf("nfs_mount_diskless_private: mbuf_get(soname) failed");
+               return (error);
+       }
+       mbuf_setlen(m, ndmntp->ndm_saddr.sin_len);
+       bcopy((caddr_t)args.addr, mbuf_data(m), ndmntp->ndm_saddr.sin_len);
+       if ((error = mountnfs(&args, mp, m, procp, &vp))) {
+               printf("nfs_mountroot: mount %s failed: %d\n", mntname, error);
+               mount_list_lock();
+               vfsp->vfc_refcount--;
+               mount_list_unlock();
+               vfs_unbusy(mp);
+               mount_lock_destroy(mp);
+               FREE_ZONE(mp, sizeof (struct mount), M_MOUNT);
                return (error);
        }
 
@@ -657,66 +925,88 @@ nfs_mount_diskless_private(ndmntp, mntname, mntflag, vpp, mpp)
  * VFS Operations.
  *
  * mount system call
- * It seems a bit dumb to copyinstr() the host and path here and then
- * bcopy() them in mountnfs(), but I wanted to detect errors before
- * doing the sockargs() call because sockargs() allocates an mbuf and
- * an error after that means that I have to release the mbuf.
  */
-/* ARGSUSED */
 static int
-nfs_mount(mp, path, data, ndp, p)
-       struct mount *mp;
-       char *path;
-       caddr_t data;
-       struct nameidata *ndp;
-       struct proc *p;
+nfs_mount(mount_t mp, vnode_t vp, user_addr_t data, vfs_context_t context)
 {
-       int error;
-       struct nfs_args args;
-       struct mbuf *nam;
-       struct vnode *vp;
-       char pth[MNAMELEN], hst[MNAMELEN];
-       u_int len;
+       proc_t p = vfs_context_proc(context);
+       int error, argsvers;
+       struct user_nfs_args args;
+       struct nfs_args tempargs;
+       mbuf_t nam;
+       size_t len;
        u_char nfh[NFSX_V3FHMAX];
+       char *mntfrom;
 
-       error = copyin(data, (caddr_t)&args, sizeof (struct nfs_args));
+       error = copyin(data, (caddr_t)&argsvers, sizeof (argsvers));
        if (error)
                return (error);
-       if (args.version != NFS_ARGSVERSION) {
-#ifndef NO_COMPAT_PRELITE2
-               /*
-                * If the argument version is unknown, then assume the
-                * caller is a pre-lite2 4.4BSD client and convert its
-                * arguments.
-                */
-               struct onfs_args oargs;
-               error = copyin(data, (caddr_t)&oargs, sizeof (struct onfs_args));
-               if (error)
-                       return (error);
-               nfs_convert_oargs(&args,&oargs);
-#else /* NO_COMPAT_PRELITE2 */
+
+       switch (argsvers) {
+       case 3:
+               if (vfs_context_is64bit(context))
+                       error = copyin(data, (caddr_t)&args, sizeof (struct user_nfs_args3));
+               else
+                       error = copyin(data, (caddr_t)&tempargs, sizeof (struct nfs_args3));
+               break;
+       case 4:
+               if (vfs_context_is64bit(context))
+                       error = copyin(data, (caddr_t)&args, sizeof (args));
+               else
+                       error = copyin(data, (caddr_t)&tempargs, sizeof (tempargs));
+               break;
+       default:
                return (EPROGMISMATCH);
-#endif /* !NO_COMPAT_PRELITE2 */
        }
-       if (args.fhsize > NFSX_V3FHMAX)
-               return (EINVAL);
-       error = copyin((caddr_t)args.fh, (caddr_t)nfh, args.fhsize);
        if (error)
                return (error);
-       error = copyinstr(path, pth, MNAMELEN-1, &len);
+
+       if (!vfs_context_is64bit(context)) {
+               args.version = tempargs.version;
+               args.addrlen = tempargs.addrlen;
+               args.sotype = tempargs.sotype;
+               args.proto = tempargs.proto;
+               args.fhsize = tempargs.fhsize;
+               args.flags = tempargs.flags;
+               args.wsize = tempargs.wsize;
+               args.rsize = tempargs.rsize;
+               args.readdirsize = tempargs.readdirsize;
+               args.timeo = tempargs.timeo;
+               args.retrans = tempargs.retrans;
+               args.maxgrouplist = tempargs.maxgrouplist;
+               args.readahead = tempargs.readahead;
+               args.leaseterm = tempargs.leaseterm;
+               args.deadthresh = tempargs.deadthresh;
+               args.addr = CAST_USER_ADDR_T(tempargs.addr);
+               args.fh = CAST_USER_ADDR_T(tempargs.fh);
+               args.hostname = CAST_USER_ADDR_T(tempargs.hostname);
+               if (argsvers >= 4) {
+                       args.acregmin = tempargs.acregmin;
+                       args.acregmax = tempargs.acregmax;
+                       args.acdirmin = tempargs.acdirmin;
+                       args.acdirmax = tempargs.acdirmax;
+               }
+       }
+
+       if (args.fhsize > NFSX_V3FHMAX)
+               return (EINVAL);
+       error = copyin(args.fh, (caddr_t)nfh, args.fhsize);
        if (error)
                return (error);
-       bzero(&pth[len], MNAMELEN - len);
-       error = copyinstr(args.hostname, hst, MNAMELEN-1, &len);
+
+       mntfrom = &vfs_statfs(mp)->f_mntfromname[0];
+       error = copyinstr(args.hostname, mntfrom, MAXPATHLEN-1, &len);
        if (error)
                return (error);
-       bzero(&hst[len], MNAMELEN - len);
+       bzero(&mntfrom[len], MAXPATHLEN - len);
+
        /* sockargs() call must be after above copyin() calls */
-       error = sockargs(&nam, (caddr_t)args.addr, args.addrlen, MT_SONAME);
+       error = sockargs(&nam, args.addr, args.addrlen, MBUF_TYPE_SONAME);
        if (error)
                return (error);
-       args.fh = nfh;
-       error = mountnfs(&args, mp, nam, pth, hst, &vp);
+
+       args.fh = CAST_USER_ADDR_T(&nfh[0]);
+       error = mountnfs(&args, mp, nam, p, &vp);
        return (error);
 }
 
@@ -724,75 +1014,74 @@ nfs_mount(mp, path, data, ndp, p)
  * Common code for mount and mountroot
  */
 static int
-mountnfs(argp, mp, nam, pth, hst, vpp)
-       register struct nfs_args *argp;
-       register struct mount *mp;
-       struct mbuf *nam;
-       char *pth, *hst;
-       struct vnode **vpp;
+mountnfs(
+       struct user_nfs_args *argp,
+       mount_t mp,
+       mbuf_t nam,
+       proc_t p,
+       vnode_t *vpp)
 {
-       register struct nfsmount *nmp;
+       struct nfsmount *nmp;
        struct nfsnode *np;
        int error, maxio;
-       struct vattr attrs;
-       struct proc *curproc;
+       struct nfs_vattr nvattrs;
+       struct vfs_context context; /* XXX get from caller? */
+       u_int64_t xid;
 
        /*
-        * turning off NQNFS until we have further testing
-        * with UBC changes, in particular, nfs_pagein and nfs_pageout.
-        * Those have NQNFS defined out in conjunction with this
-        * returning an error. Remove when fully tested.
+        * Silently clear NFSMNT_NOCONN if it's a TCP mount, it makes
+        * no sense in that context.
         */
-       if (argp->flags & NFSMNT_NQNFS) {
-               error = NFSERR_NOTSUPP;
-               goto bad2;
-       }
+       if (argp->sotype == SOCK_STREAM)
+               argp->flags &= ~NFSMNT_NOCONN;
         
-       if (mp->mnt_flag & MNT_UPDATE) {
+       if (vfs_flags(mp) & MNT_UPDATE) {
                nmp = VFSTONFS(mp);
                /* update paths, file handles, etc, here        XXX */
-               m_freem(nam);
+               mbuf_freem(nam);
                return (0);
        } else {
                MALLOC_ZONE(nmp, struct nfsmount *,
                                sizeof (struct nfsmount), M_NFSMNT, M_WAITOK);
+               if (!nmp) {
+                       mbuf_freem(nam);
+                       return (ENOMEM);
+               }
                bzero((caddr_t)nmp, sizeof (struct nfsmount));
                TAILQ_INIT(&nmp->nm_uidlruhead);
                TAILQ_INIT(&nmp->nm_bufq);
-               mp->mnt_data = (qaddr_t)nmp;
+               vfs_setfsprivate(mp, nmp);
        }
-       vfs_getnewfsid(mp);
-       nmp->nm_mountp = mp;
-       nmp->nm_flag = argp->flags;
-       if (nmp->nm_flag & NFSMNT_NQNFS)
-               /*
-                * We have to set mnt_maxsymlink to a non-zero value so
-                * that COMPAT_43 routines will know that we are setting
-                * the d_type field in directories (and can zero it for
-                * unsuspecting binaries).
-                */
-               mp->mnt_maxsymlinklen = 1;
+
+       /* setup defaults */
        nmp->nm_timeo = NFS_TIMEO;
        nmp->nm_retry = NFS_RETRANS;
-       nmp->nm_wsize = NFS_WSIZE;
-       nmp->nm_rsize = NFS_RSIZE;
+       if (argp->sotype == SOCK_DGRAM) {
+               nmp->nm_wsize = NFS_DGRAM_WSIZE;
+               nmp->nm_rsize = NFS_DGRAM_RSIZE;
+       } else {
+               nmp->nm_wsize = NFS_WSIZE;
+               nmp->nm_rsize = NFS_RSIZE;
+       }
        nmp->nm_readdirsize = NFS_READDIRSIZE;
        nmp->nm_numgrps = NFS_MAXGRPS;
        nmp->nm_readahead = NFS_DEFRAHEAD;
-       nmp->nm_leaseterm = NQ_DEFLEASE;
-       nmp->nm_deadthresh = NQ_DEADTHRESH;
-       CIRCLEQ_INIT(&nmp->nm_timerhead);
-       nmp->nm_inprog = NULLVP;
-       bcopy(hst, mp->mnt_stat.f_mntfromname, MNAMELEN);
-       bcopy(pth, mp->mnt_stat.f_mntonname, MNAMELEN);
-       nmp->nm_nam = nam;
+       nmp->nm_tprintf_delay = nfs_tprintf_delay;
+       if (nmp->nm_tprintf_delay < 0)
+               nmp->nm_tprintf_delay = 0;
+       nmp->nm_tprintf_initial_delay = nfs_tprintf_initial_delay;
+       if (nmp->nm_tprintf_initial_delay < 0)
+               nmp->nm_tprintf_initial_delay = 0;
+       nmp->nm_acregmin = NFS_MINATTRTIMO;
+       nmp->nm_acregmax = NFS_MAXATTRTIMO;
+       nmp->nm_acdirmin = NFS_MINDIRATTRTIMO;
+       nmp->nm_acdirmax = NFS_MAXDIRATTRTIMO;
 
-       /*
-        * Silently clear NFSMNT_NOCONN if it's a TCP mount, it makes
-        * no sense in that context.
-        */
-       if (argp->sotype == SOCK_STREAM)
-               argp->flags &= ~NFSMNT_NOCONN;
+       vfs_getnewfsid(mp);
+       nmp->nm_mountp = mp;
+       vfs_setauthopaque(mp);
+       nmp->nm_flag = argp->flags;
+       nmp->nm_nam = nam;
 
        if ((argp->flags & NFSMNT_TIMEO) && argp->timeo > 0) {
                nmp->nm_timeo = (argp->timeo * NFS_HZ + 5) / 10;
@@ -854,16 +1143,30 @@ mountnfs(argp, mp, nam, pth, hst, vpp)
        if ((argp->flags & NFSMNT_READAHEAD) && argp->readahead >= 0 &&
                argp->readahead <= NFS_MAXRAHEAD)
                nmp->nm_readahead = argp->readahead;
-       if ((argp->flags & NFSMNT_LEASETERM) && argp->leaseterm >= 2 &&
-               argp->leaseterm <= NQ_MAXLEASE)
-               nmp->nm_leaseterm = argp->leaseterm;
-       if ((argp->flags & NFSMNT_DEADTHRESH) && argp->deadthresh >= 1 &&
-               argp->deadthresh <= NQ_NEVERDEAD)
-               nmp->nm_deadthresh = argp->deadthresh;
+
+       if (argp->version >= 4) {
+               if ((argp->flags & NFSMNT_ACREGMIN) && argp->acregmin >= 0)
+                       nmp->nm_acregmin = argp->acregmin;
+               if ((argp->flags & NFSMNT_ACREGMAX) && argp->acregmax >= 0)
+                       nmp->nm_acregmax = argp->acregmax;
+               if ((argp->flags & NFSMNT_ACDIRMIN) && argp->acdirmin >= 0)
+                       nmp->nm_acdirmin = argp->acdirmin;
+               if ((argp->flags & NFSMNT_ACDIRMAX) && argp->acdirmax >= 0)
+                       nmp->nm_acdirmax = argp->acdirmax;
+               if (nmp->nm_acregmin > nmp->nm_acregmax)
+                       nmp->nm_acregmin = nmp->nm_acregmax;
+               if (nmp->nm_acdirmin > nmp->nm_acdirmax)
+                       nmp->nm_acdirmin = nmp->nm_acdirmax;
+       }
+
        /* Set up the sockets and per-host congestion */
        nmp->nm_sotype = argp->sotype;
        nmp->nm_soproto = argp->proto;
 
+       /* make sure mbuf constants are set up */
+       if (!nfs_mbuf_mlen)
+               nfs_mbuf_init();
+
        /*
         * For Connection based sockets (TCP,...) defer the connect until
         * the first request, in case the server is not responding.
@@ -873,12 +1176,20 @@ mountnfs(argp, mp, nam, pth, hst, vpp)
                goto bad;
 
        /*
-        * This is silly, but it has to be set so that vinifod() works.
-        * We do not want to do an nfs_statfs() here since we can get
-        * stuck on a dead server and we are holding a lock on the mount
-        * point.
+        * Get file attributes for the mountpoint.  These are needed
+        * in order to properly create the root vnode.
         */
-       mp->mnt_stat.f_iosize = nfs_iosize(nmp);
+       // LP64todo - fix CAST_DOWN of argp->fh
+       error = nfs_getattr_no_vnode(mp, CAST_DOWN(caddr_t, argp->fh), argp->fhsize,
+                       proc_ucred(p), p, &nvattrs, &xid);
+       if (error) {
+               /*
+                * we got problems... we couldn't get the attributes
+                * from the NFS server... so the mount fails.
+                */
+               goto bad;
+       }
+
        /*
         * A reference count is needed on the nfsnode representing the
         * remote root.  If this object is not persistent, then backward
@@ -887,35 +1198,59 @@ mountnfs(argp, mp, nam, pth, hst, vpp)
         * this problem, because one can identify root inodes by their
         * number == ROOTINO (2).
         */
-       error = nfs_nget(mp, (nfsfh_t *)argp->fh, argp->fhsize, &np);
+       error = nfs_nget(mp, NULL, NULL, CAST_DOWN(caddr_t, argp->fh), argp->fhsize,
+                       &nvattrs, &xid, NG_MARKROOT, &np);
        if (error)
                goto bad;
 
        /*
         * save this vnode pointer. That way nfs_unmount()
-        * does not need to call nfs_net() just get it to drop
+        * does not need to call nfs_nget() just get it to drop
         * this vnode reference.
         */
        nmp->nm_dvp = *vpp = NFSTOV(np);
+       /* get usecount and drop iocount */
+       error = vnode_ref(*vpp);
+       if (error) {
+               vnode_put(*vpp);
+               goto bad;
+       }
+       vnode_put(*vpp);
+
+       /*
+        * Set the mount point's block I/O size.
+        * We really need to do this after we get info back from
+        * the server about what its preferred I/O sizes are.
+        */
+       if (nmp->nm_flag & NFSMNT_NFSV3)
+               nfs_fsinfo(nmp, *vpp, proc_ucred(p), p);
+       vfs_statfs(mp)->f_iosize = nfs_iosize(nmp);
 
        /*
-        * Get file attributes for the mountpoint.  This has the side
-        * effect of filling in (*vpp)->v_type with the correct value.
+        * V3 mounts give us a (relatively) reliable remote access(2)
+        * call, so advertise the fact.
+        *
+        * XXX this may not be the best way to go, as the granularity
+        *     offered isn't a good match to our needs.
         */
-       curproc = current_proc();
-       VOP_GETATTR(*vpp, &attrs, curproc->p_ucred, curproc);
+       if (nmp->nm_flag & NFSMNT_NFSV3)
+               vfs_setauthopaqueaccess(mp);
 
        /*
-        * Lose the lock but keep the ref.
+        * Do statfs to ensure static info gets set to reasonable values.
         */
-       VOP_UNLOCK(*vpp, 0, curproc);
+       context.vc_proc = p;
+       context.vc_ucred = proc_ucred(p);
+       nfs_statfs(mp, vfs_statfs(mp), &context);
 
+       if (nmp->nm_flag & NFSMNT_RESVPORT)
+               nfs_resv_mounts++;
+       nmp->nm_state |= NFSSTA_MOUNTED;
        return (0);
 bad:
        nfs_disconnect(nmp);
-       _FREE_ZONE((caddr_t)nmp, sizeof (struct nfsmount), M_NFSMNT);
-bad2:
-       m_freem(nam);
+       FREE_ZONE((caddr_t)nmp, sizeof (struct nfsmount), M_NFSMNT);
+       mbuf_freem(nam);
        return (error);
 }
 
@@ -924,98 +1259,84 @@ bad2:
  * unmount system call
  */
 static int
-nfs_unmount(mp, mntflags, p)
-       struct mount *mp;
-       int mntflags;
-       struct proc *p;
+nfs_unmount(
+       mount_t mp,
+       int mntflags,
+       __unused vfs_context_t context)
 {
        register struct nfsmount *nmp;
-       struct vnode *vp;
+       vnode_t vp;
        int error, flags = 0;
 
-       if (mntflags & MNT_FORCE)
-               flags |= FORCECLOSE;
        nmp = VFSTONFS(mp);
+       /*
+        * During a force unmount we want to...
+        *   Mark that we are doing a force unmount.
+        *   Make the mountpoint soft.
+        */
+       if (mntflags & MNT_FORCE) {
+               flags |= FORCECLOSE;
+               nmp->nm_state |= NFSSTA_FORCE;
+               nmp->nm_flag |= NFSMNT_SOFT;
+       }
        /*
         * Goes something like this..
         * - Call vflush() to clear out vnodes for this file system,
         *   except for the swap files. Deal with them in 2nd pass.
-        *   It will do vgone making the vnode VBAD at that time.
         * - Decrement reference on the vnode representing remote root.
         * - Close the socket
         * - Free up the data structures
         */
        vp = nmp->nm_dvp;
        
-       /*
-        * Must handshake with nqnfs_clientd() if it is active.
-        */
-       nmp->nm_flag |= NFSMNT_DISMINPROG;
-       while (nmp->nm_inprog != NULLVP)
-               (void) tsleep((caddr_t)&lbolt, PSOCK, "nfsdism", 0);
        /*
         * vflush will check for busy vnodes on mountpoint. 
         * Will do the right thing for MNT_FORCE. That is, we should
         * not get EBUSY back.
         */
        error = vflush(mp, vp, SKIPSWAP | flags);
-       if (mntflags & MNT_FORCE) 
+       if (mntflags & MNT_FORCE) {
                error = vflush(mp, NULLVP, flags); /* locks vp in the process */
-       else {
-               if (vp->v_usecount > 1) {
-                       nmp->nm_flag &= ~NFSMNT_DISMINPROG;
+       } else {
+               if (vnode_isinuse(vp, 1))
                        return (EBUSY);
-               }
                error = vflush(mp, vp, flags);
        }
-
-       if (error) {
-               nmp->nm_flag &= ~NFSMNT_DISMINPROG;
+       if (error)
                return (error);
-       }
 
-       /*
-        * We are now committed to the unmount.
-        * For NQNFS, let the server daemon free the nfsmount structure.
-        */
-       if (nmp->nm_flag & (NFSMNT_NQNFS | NFSMNT_KERB))
-               nmp->nm_flag |= NFSMNT_DISMNT;
+       nmp->nm_state &= ~NFSSTA_MOUNTED;
+       if (nmp->nm_flag & NFSMNT_RESVPORT) {
+               if (--nfs_resv_mounts == 0)
+                       nfs_bind_resv_thread_wake();
+       }
 
        /*
         * Release the root vnode reference held by mountnfs()
-        * vflush did the vgone for us when we didn't skip over
-        * it in the MNT_FORCE case. (Thus vp can't be locked when
-        * called vflush in non-skip vp case.)
         */
-       vrele(vp);
-       if (!(mntflags & MNT_FORCE))
-               vgone(vp);
-       mp->mnt_data = 0; /* don't want to end up using stale vp */
+       vnode_rele(vp);
+
+       (void)vflush(mp, NULLVP, FORCECLOSE);
+       vfs_setfsprivate(mp, 0); /* don't want to end up using stale vp */
+
        nfs_disconnect(nmp);
-       m_freem(nmp->nm_nam);
+       mbuf_freem(nmp->nm_nam);
 
-       if ((nmp->nm_flag & (NFSMNT_NQNFS | NFSMNT_KERB)) == 0) {
-               register struct nfsreq *rp;
+       if ((nmp->nm_flag & NFSMNT_KERB) == 0) {
+               struct nfsreq *rp;
                /*
                 * Loop through outstanding request list and remove dangling
                 * references to defunct nfsmount struct
                 */
-#if NFSDIAG && 0
-               if (hw_atomic_add(&nfsreqqusers, 1) != 1)
-                       nfsatompanic("unmount add");
-               nfsbtlen = backtrace(&nfsbt, sizeof(nfsbt));
-               nfsbtcpu = cpu_number();
-               nfsbtthread = (int)(current_thread());
-#endif
-
                for (rp = nfs_reqq.tqh_first; rp; rp = rp->r_chain.tqe_next)
                        if (rp->r_nmp == nmp)
                                rp->r_nmp = (struct nfsmount *)0;
-#if NFSDIAG && 0
-               if (hw_atomic_sub(&nfsreqqusers, 1) != 0)
-                       nfsatompanic("unmount sub");
-#endif
-               _FREE_ZONE((caddr_t)nmp, sizeof (struct nfsmount), M_NFSMNT);
+               /* Need to wake up any rcvlock waiters so they notice the unmount. */
+               if (nmp->nm_state & NFSSTA_WANTRCV) {
+                       nmp->nm_state &= ~NFSSTA_WANTRCV;
+                       wakeup(&nmp->nm_state);
+               }
+               FREE_ZONE((caddr_t)nmp, sizeof (struct nfsmount), M_NFSMNT);
        }
        return (0);
 }
@@ -1024,155 +1345,147 @@ nfs_unmount(mp, mntflags, p)
  * Return root of a filesystem
  */
 static int
-nfs_root(mp, vpp)
-       struct mount *mp;
-       struct vnode **vpp;
+nfs_root(mount_t mp, vnode_t *vpp, __unused vfs_context_t context)
 {
-       register struct vnode *vp;
+       vnode_t vp;
        struct nfsmount *nmp;
        int error;
+       u_long vpid;
 
        nmp = VFSTONFS(mp);
        vp = nmp->nm_dvp;
-       error = vget(vp, LK_EXCLUSIVE, current_proc());
-       if (error)
-               return (error);
-       if (vp->v_type == VNON)
-           vp->v_type = VDIR;
-       vp->v_flag |= VROOT;
+       vpid = vnode_vid(vp);
+       while ((error = vnode_getwithvid(vp, vpid))) {
+               /* vnode_get() may return ENOENT if the dir changes. */
+               /* If that happens, just try it again, else return the error. */
+               if ((error != ENOENT) || (vnode_vid(vp) == vpid))
+                       return (error);
+               vpid = vnode_vid(vp);
+       }
        *vpp = vp;
        return (0);
 }
 
-extern int syncprt;
-
 /*
  * Flush out the buffer cache
  */
-/* ARGSUSED */
+
+struct nfs_sync_cargs {
+        vfs_context_t context;
+        int    waitfor;
+        int    error;
+};
+
 static int
-nfs_sync(mp, waitfor, cred, p)
-       struct mount *mp;
-       int waitfor;
-       struct ucred *cred;
-       struct proc *p;
+nfs_sync_callout(vnode_t vp, void *arg)
 {
-       register struct vnode *vp;
-       int error, allerror = 0;
+       struct nfs_sync_cargs *cargs = (struct nfs_sync_cargs*)arg;
+       int error;
 
-       /*
-        * Force stale buffer cache information to be flushed.
-        */
-loop:
-       for (vp = mp->mnt_vnodelist.lh_first;
-            vp != NULL;
-            vp = vp->v_mntvnodes.le_next) {
-                int didhold = 0;
-               /*
-                * If the vnode that we are about to sync is no longer
-                * associated with this mount point, start over.
-                */
-               if (vp->v_mount != mp)
-                       goto loop;
-               if (VOP_ISLOCKED(vp) || vp->v_dirtyblkhd.lh_first == NULL)
-                       continue;
-               if (vget(vp, LK_EXCLUSIVE, p))
-                       goto loop;
-               didhold = ubc_hold(vp);
-               error = VOP_FSYNC(vp, cred, waitfor, p);
-               if (error)
-                       allerror = error;
-               VOP_UNLOCK(vp, 0, p);
-               if (didhold)
-                       ubc_rele(vp);
-               vrele(vp);
-       }
-       return (allerror);
+       if (LIST_EMPTY(&VTONFS(vp)->n_dirtyblkhd))
+               return (VNODE_RETURNED);
+       if (VTONFS(vp)->n_flag & NWRBUSY)
+               return (VNODE_RETURNED);
+
+       error = nfs_flush(vp, cargs->waitfor,
+                       vfs_context_ucred(cargs->context),
+                       vfs_context_proc(cargs->context), 0);
+       if (error)
+               cargs->error = error;
+
+       return (VNODE_RETURNED);
+}
+
+static int
+nfs_sync(mount_t mp, int waitfor, vfs_context_t context)
+{
+       struct nfs_sync_cargs cargs;
+
+       cargs.waitfor = waitfor;
+       cargs.context = context;
+       cargs.error = 0;
+
+       vnode_iterate(mp, 0, nfs_sync_callout, &cargs);
+
+       return (cargs.error);
 }
 
 /*
  * NFS flat namespace lookup.
  * Currently unsupported.
  */
-/* ARGSUSED */
+/*ARGSUSED*/
 static int
-nfs_vget(mp, ino, vpp)
-       struct mount *mp;
-       ino_t ino;
-       struct vnode **vpp;
+nfs_vget(
+       __unused mount_t mp,
+       __unused ino64_t ino,
+       __unused vnode_t *vpp,
+       __unused vfs_context_t context)
 {
 
-       return (EOPNOTSUPP);
+       return (ENOTSUP);
 }
 
 /*
  * At this point, this should never happen
  */
-/* ARGSUSED */
+/*ARGSUSED*/
 static int
-nfs_fhtovp(mp, fhp, nam, vpp, exflagsp, credanonp)
-       register struct mount *mp;
-       struct fid *fhp;
-       struct mbuf *nam;
-       struct vnode **vpp;
-       int *exflagsp;
-       struct ucred **credanonp;
+nfs_fhtovp(
+       __unused mount_t mp,
+       __unused int fhlen,
+       __unused unsigned char *fhp,
+       __unused vnode_t *vpp,
+       __unused vfs_context_t context)
 {
 
-       return (EINVAL);
+       return (ENOTSUP);
 }
 
 /*
  * Vnode pointer to File handle, should never happen either
  */
-/* ARGSUSED */
+/*ARGSUSED*/
 static int
-nfs_vptofh(vp, fhp)
-       struct vnode *vp;
-       struct fid *fhp;
+nfs_vptofh(
+       __unused vnode_t vp,
+       __unused int *fhlenp,
+       __unused unsigned char *fhp,
+       __unused vfs_context_t context)
 {
 
-       return (EINVAL);
+       return (ENOTSUP);
 }
 
 /*
  * Vfs start routine, a no-op.
  */
-/* ARGSUSED */
+/*ARGSUSED*/
 static int
-nfs_start(mp, flags, p)
-       struct mount *mp;
-       int flags;
-       struct proc *p;
+nfs_start(
+       __unused mount_t mp,
+       __unused int flags,
+       __unused vfs_context_t context)
 {
 
        return (0);
 }
 
-/*
- * Do operations associated with quotas, not supported
- */
-/* ARGSUSED */
-static int
-nfs_quotactl(mp, cmd, uid, arg, p)
-       struct mount *mp;
-       int cmd;
-       uid_t uid;
-       caddr_t arg;
-       struct proc *p;
-{
-
-       return (EOPNOTSUPP);
-}
-
 /*
  * Do that sysctl thang...
  */
 static int
-nfs_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp, void *newp,
-          size_t newlen, struct proc *p)
+nfs_sysctl(int *name, u_int namelen, user_addr_t oldp, size_t *oldlenp, 
+           user_addr_t newp, size_t newlen, vfs_context_t context)
 {
-       int rv;
+       int error = 0, val;
+       struct sysctl_req *req = NULL;
+       struct vfsidctl vc;
+       struct user_vfsidctl user_vc;
+       mount_t mp;
+       struct nfsmount *nmp = NULL;
+       struct vfsquery vq;
+       boolean_t is_64_bit;
 
        /*
         * All names at this level are terminal.
@@ -1180,6 +1493,43 @@ nfs_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp, void *newp,
        if(namelen > 1)
                return ENOTDIR; /* overloaded */
 
+       is_64_bit = vfs_context_is64bit(context);
+
+       /* common code for "new style" VFS_CTL sysctl, get the mount. */
+       switch (name[0]) {
+       case VFS_CTL_TIMEO:
+       case VFS_CTL_QUERY:
+       case VFS_CTL_NOLOCKS:
+               req = CAST_DOWN(struct sysctl_req *, oldp);
+               if (is_64_bit) {
+                       error = SYSCTL_IN(req, &user_vc, sizeof(user_vc));
+                       if (error)
+                                return (error);
+                       mp = vfs_getvfs(&user_vc.vc_fsid);
+               } 
+               else {
+                       error = SYSCTL_IN(req, &vc, sizeof(vc));
+                       if (error)
+                               return (error);
+                       mp = vfs_getvfs(&vc.vc_fsid);
+               }
+               if (mp == NULL)
+                       return (ENOENT);
+               nmp = VFSTONFS(mp);
+               if (nmp == NULL)
+                       return (ENOENT);
+               bzero(&vq, sizeof(vq));
+               req->newidx = 0;
+               if (is_64_bit) {
+                       req->newptr = user_vc.vc_ptr;
+                       req->newlen = (size_t)user_vc.vc_len;
+               }
+               else {
+                       req->newptr = CAST_USER_ADDR_T(vc.vc_ptr);
+                       req->newlen = vc.vc_len;
+               }
+       }
+
        switch(name[0]) {
        case NFS_NFSSTATS:
                if(!oldp) {
@@ -1192,8 +1542,9 @@ nfs_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp, void *newp,
                        return ENOMEM;
                }
 
-               rv = copyout(&nfsstats, oldp, sizeof nfsstats);
-               if(rv) return rv;
+               error = copyout(&nfsstats, oldp, sizeof nfsstats);
+               if (error)
+                       return (error);
 
                if(newp && newlen != sizeof nfsstats)
                        return EINVAL;
@@ -1202,9 +1553,50 @@ nfs_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp, void *newp,
                        return copyin(newp, &nfsstats, sizeof nfsstats);
                }
                return 0;
-
+       case VFS_CTL_NOLOCKS:
+               val = (nmp->nm_flag & NFSMNT_NOLOCKS) ? 1 : 0;
+               if (req->oldptr != USER_ADDR_NULL) {
+                       error = SYSCTL_OUT(req, &val, sizeof(val));
+                       if (error)
+                               return (error);
+               }
+               if (req->newptr != USER_ADDR_NULL) {
+                       error = SYSCTL_IN(req, &val, sizeof(val));
+                       if (error)
+                               return (error);
+                       if (val)
+                               nmp->nm_flag |= NFSMNT_NOLOCKS;
+                       else
+                               nmp->nm_flag &= ~NFSMNT_NOLOCKS;
+               }
+               break;
+       case VFS_CTL_QUERY:
+               if (nmp->nm_state & NFSSTA_TIMEO)
+                       vq.vq_flags |= VQ_NOTRESP;
+               if (!(nmp->nm_flag & NFSMNT_NOLOCKS) &&
+                   (nmp->nm_state & NFSSTA_LOCKTIMEO))
+                       vq.vq_flags |= VQ_NOTRESPLOCK;
+               error = SYSCTL_OUT(req, &vq, sizeof(vq));
+               break;
+       case VFS_CTL_TIMEO:
+               if (req->oldptr != USER_ADDR_NULL) {
+                       error = SYSCTL_OUT(req, &nmp->nm_tprintf_initial_delay,
+                           sizeof(nmp->nm_tprintf_initial_delay));
+                       if (error)
+                               return (error);
+               }
+               if (req->newptr != USER_ADDR_NULL) {
+                       error = SYSCTL_IN(req, &nmp->nm_tprintf_initial_delay,
+                           sizeof(nmp->nm_tprintf_initial_delay));
+                       if (error)
+                               return (error);
+                       if (nmp->nm_tprintf_initial_delay < 0)
+                               nmp->nm_tprintf_initial_delay = 0;
+               }
+               break;
        default:
-               return EOPNOTSUPP;
+               return (ENOTSUP);
        }
+       return (error);
 }