source: vbox/trunk/src/VBox/Additions/solaris/SharedFolders/vboxfs_vnode.c@ 29074

/** @file
 * VirtualBox File System for Solaris Guests, vnode implementation.
 */

/*
 * Copyright (C) 2009 Oracle Corporation
 *
 * This file is part of VirtualBox Open Source Edition (OSE), as
 * available from http://www.virtualbox.org. This file is free software;
 * you can redistribute it and/or modify it under the terms of the GNU
 * General Public License (GPL) as published by the Free Software
 * Foundation, in version 2 as it comes in the "COPYING" file of the
 * VirtualBox OSE distribution. VirtualBox OSE is distributed in the
 * hope that it will be useful, but WITHOUT ANY WARRANTY of any kind.
 */

/*
 * Shared Folder File System is used from Solaris when run as a guest operating
 * system on VirtualBox, though is meant to be usable with any hypervisor that
 * can provide similar functionality. The sffs code handles all the Solaris
 * specific semantics and relies on a provider module to actually access
 * directories, files, etc. The provider interfaces are described in
 * "vboxfs_prov.h" and the module implementing them is shipped as part of the
 * VirtualBox Guest Additions for Solaris.
 *
 * The shared folder file system is similar to a networked file system,
 * but with some caveats. The sffs code caches minimal information and proxies
 * out to the provider whenever possible. Here are some things that are
 * handled in this code and not by the proxy:
 *
 * - a way to open ".." from any already open directory
 * - st_ino numbers
 * - detecting directory changes that happened on the host.
 *
 * The implementation builds a cache of information for every file/directory
 * ever accessed in all mounted sffs filesystems using sf_node structures.
 *
 * This information for both open or closed files can become invalid if
 * asynchronous changes are made on the host. Solaris should not panic() in
 * this event, but some file system operations may return unexpected errors.
 * Information for such directories or files while they have active vnodes
 * is removed from the regular cache and stored in a "stale" bucket until
 * the vnode becomes completely inactive.
 *
 * No file data is cached in the guest. This means we don't support mmap() yet.
 * A future version could relatively easily add support for read-only
 * mmap(MAP_SHARED) and any mmap(MAP_PRIVATE). But a la ZFS, this data caching
 * would not be coherent with normal simultaneous read()/write() operations,
 * nor will it be coherent with data access on the host. Writable
 * mmap(MAP_SHARED) access is possible, but guaranteeing any kind of coherency
 * with concurrent activity on the host would be near impossible with the
 * existing interfaces.
 *
 * A note about locking. sffs is not a high performance file system.
 * No fine grained locking is done. The one sffs_lock protects just about
 * everything.
 */

#include <VBox/log.h>

#include <sys/types.h>
#include <sys/stat.h>
#include <sys/mntent.h>
#include <sys/param.h>
#include <sys/modctl.h>
#include <sys/mount.h>
#include <sys/policy.h>
#include <sys/atomic.h>
#include <sys/sysmacros.h>
#include <sys/ddi.h>
#include <sys/sunddi.h>
#include <sys/vfs.h>
#if !defined(VBOX_VFS_SOLARIS_10U6)
#include <sys/vfs_opreg.h>
#endif
#include <sys/pathname.h>
#include <sys/dirent.h>
#include <sys/fs_subr.h>
#include "vboxfs_prov.h"
#include "vboxfs_vnode.h"
#include "vboxfs_vfs.h"

static struct vnodeops *sffs_ops = NULL;

kmutex_t sffs_lock;
static avl_tree_t sfnodes;
static avl_tree_t stale_sfnodes;

/*
 * For now we'll use an I/O buffer that doesn't page fault for VirtualBox
 * to transfer data into.
 */
char *sffs_buffer;

/*
 * sfnode_compare() is needed for AVL tree functionality.
 * The nodes are sorted by mounted filesystem, then path. If the
 * nodes are stale, the node pointer itself is used to force uniqueness.
 */
static int
sfnode_compare(const void *a, const void *b)
{
        sfnode_t *x = (sfnode_t *)a;
        sfnode_t *y = (sfnode_t *)b;
        int diff;

        if (x->sf_is_stale) {
                ASSERT(y->sf_is_stale);
                diff = strcmp(x->sf_path, y->sf_path);
                if (diff == 0)
                        diff = (uintptr_t)y - (uintptr_t)x;
        } else {
                ASSERT(!y->sf_is_stale);
                diff = (uintptr_t)y->sf_sffs - (uintptr_t)x->sf_sffs;
                if (diff == 0)
                        diff = strcmp(x->sf_path, y->sf_path);
        }
        if (diff < 0)
                return (-1);
        if (diff > 0)
                return (1);
        return (0);
}

/*
 * Construct a new pathname given an sfnode plus an optional tail component.
 * This handles ".." and "."
 */
static char *
sfnode_construct_path(sfnode_t *node, char *tail)
{
        char *p;

        if (strcmp(tail, ".") == 0 || strcmp(tail, "..") == 0)
                panic("construct path for %s", tail);
        p = kmem_alloc(strlen(node->sf_path) + 1 + strlen(tail) + 1, KM_SLEEP);
        strcpy(p, node->sf_path);
        strcat(p, "/");
        strcat(p, tail);
        return (p);
}

/*
 * Open the provider file associated with a vnode. Holding the file open is
 * the only way we have of trying to have a vnode continue to refer to the
 * same host file in the host in light of the possibility of host side renames.
 */
static void
sfnode_open(sfnode_t *node)
{
        int error;
        sfp_file_t *fp;

        if (node->sf_file != NULL)
                return;
        error = sfprov_open(node->sf_sffs->sf_handle, node->sf_path, &fp);
        if (error == 0)
                node->sf_file = fp;
}

/*
 * get a new vnode reference for an sfnode
 */
vnode_t *
sfnode_get_vnode(sfnode_t *node)
{
        vnode_t *vp;

        if (node->sf_vnode != NULL) {
                VN_HOLD(node->sf_vnode);
        } else {
                vp = vn_alloc(KM_SLEEP);
        LogFlowFunc(("  %s gets vnode 0x%p\n", node->sf_path, vp));
                vp->v_type = node->sf_type;
                vp->v_vfsp = node->sf_sffs->sf_vfsp;
                VFS_HOLD(vp->v_vfsp);
                vn_setops(vp, sffs_ops);
                vp->v_flag = VNOMAP | VNOSWAP;
                vn_exists(vp);
                vp->v_data = node;
                node->sf_vnode = vp;
        }
        sfnode_open(node);
        return (node->sf_vnode);
}

/*
 * Allocate and initialize a new sfnode and assign it a vnode
 */
sfnode_t *
sfnode_make(
        sffs_data_t     *sffs,
        char            *path,
        vtype_t         type,
        sfp_file_t      *fp,
        sfnode_t        *parent)        /* can be NULL for root */
{
        sfnode_t        *node;
        avl_index_t     where;

        ASSERT(MUTEX_HELD(&sffs_lock));
        ASSERT(path != NULL);

        /*
         * build the sfnode
         */
    LogFlowFunc(("sffs_make(%s)\n", path));
        node = kmem_alloc(sizeof (*node), KM_SLEEP);
        node->sf_sffs = sffs;
        node->sf_path = path;
        node->sf_ino = sffs->sf_ino++;
        node->sf_type = type;
        node->sf_is_stale = 0;  /* never stale at creation */
        node->sf_file = fp;
        node->sf_vnode = NULL;  /* do this before any sfnode_get_vnode() */
        node->sf_children = 0;
        node->sf_parent = parent;
        if (parent)
                ++parent->sf_children;

        /*
         * add the new node to our cache
         */
        if (avl_find(&sfnodes, node, &where) != NULL)
                panic("sffs_create_sfnode(%s): duplicate sfnode_t", path);
        avl_insert(&sfnodes, node, where);
        return (node);
}

/*
 * destroy an sfnode
 */
static void
sfnode_destroy(sfnode_t *node)
{
        avl_index_t where;
        avl_tree_t *tree;
        sfnode_t *parent;
top:
        parent = node->sf_parent;
        ASSERT(MUTEX_HELD(&sffs_lock));
        ASSERT(node->sf_path != NULL);
    LogFlowFunc(("sffs_destroy(%s)%s\n", node->sf_path, node->sf_is_stale ? " stale": ""));
        if (node->sf_children != 0)
                panic("sfnode_destroy(%s) has %d children", node->sf_children);
        if (node->sf_vnode != NULL)
                panic("sfnode_destroy(%s) has active vnode", node->sf_path);

        if (node->sf_is_stale)
                tree = &stale_sfnodes;
        else
                tree = &sfnodes;
        if (avl_find(tree, node, &where) == NULL)
                panic("sfnode_destroy(%s) not found", node->sf_path);
        avl_remove(tree, node);

        VFS_RELE(node->sf_sffs->sf_vfsp);
        kmem_free(node->sf_path, strlen(node->sf_path) + 1);
        kmem_free(node, sizeof (*node));
        if (parent != NULL) {
                if (parent->sf_children == 0)
                        panic("sfnode_destroy(%s) parent has no child");
                --parent->sf_children;
                if (parent->sf_children == 0 &&
                    parent->sf_is_stale &&
                    parent->sf_vnode == NULL) {
                        node = parent;
                        goto top;
                }
        }
}

/*
 * Some sort of host operation on an sfnode has failed or it has been
 * deleted. Mark this node and any children as stale, deleting knowledge
 * about any which do not have active vnodes or children
 * This also handle deleting an inactive node that was already stale.
 */
static void
sfnode_make_stale(sfnode_t *node)
{
        sfnode_t *n;
        int len;
        ASSERT(MUTEX_HELD(&sffs_lock));
        avl_index_t where;

        /*
         * First deal with any children of a directory node.
         * If a directory becomes stale, anything below it becomes stale too.
         */
        if (!node->sf_is_stale && node->sf_type == VDIR) {
                len = strlen(node->sf_path);

                n = node;
                while ((n = AVL_NEXT(&sfnodes, node)) != NULL) {
                        ASSERT(!n->sf_is_stale);

                        /*
                         * quit when no longer seeing children of node
                         */
                        if (n->sf_sffs != node->sf_sffs ||
                            strncmp(node->sf_path, n->sf_path, len) != 0 ||
                            n->sf_path[len] != '/')
                                break;

                        /*
                         * Either mark the child as stale or destroy it
                         */
                        if (n->sf_vnode == NULL && n->sf_children == 0) {
                                sfnode_destroy(n);
                        } else {
                LogFlowFunc(("sffs_make_stale(%s) sub\n", n->sf_path));
                                if (avl_find(&sfnodes, n, &where) == NULL)
                                        panic("sfnode_make_stale(%s)"
                                            " not in sfnodes", n->sf_path);
                                avl_remove(&sfnodes, n);
                                n->sf_is_stale = 1;
                                if (avl_find(&stale_sfnodes, n, &where) != NULL)
                                        panic("sffs_make_stale(%s) duplicates",
                                            n->sf_path);
                                avl_insert(&stale_sfnodes, n, where);
                        }
                }
        }

        /*
         * Now deal with the given node.
         */
        if (node->sf_vnode == NULL && node->sf_children == 0) {
                sfnode_destroy(node);
        } else if (!node->sf_is_stale) {
        LogFlowFunc(("sffs_make_stale(%s)\n", node->sf_path));
                if (avl_find(&sfnodes, node, &where) == NULL)
                        panic("sfnode_make_stale(%s) not in sfnodes",
                            node->sf_path);
                avl_remove(&sfnodes, node);
                node->sf_is_stale = 1;
                if (avl_find(&stale_sfnodes, node, &where) != NULL)
                        panic("sffs_make_stale(%s) duplicates", node->sf_path);
                avl_insert(&stale_sfnodes, node, where);
        }
}

/*
 * Rename a file or a directory
 */
static void
sfnode_rename(sfnode_t *node, sfnode_t *newparent, char *path)
{
        sfnode_t *n;
        sfnode_t template;
        avl_index_t where;
        int len = strlen(path);
        int old_len;
        char *new_path;
        char *tail;
        ASSERT(MUTEX_HELD(&sffs_lock));

        ASSERT(!node->sf_is_stale);

        /*
         * Have to remove anything existing that had the new name.
         */
        template.sf_sffs = node->sf_sffs;
        template.sf_path = path;
        template.sf_is_stale = 0;
        n = avl_find(&sfnodes, &template, &where);
        if (n != NULL)
                sfnode_make_stale(n);

        /*
         * Do the renaming, deal with any children of this node first.
         */
        if (node->sf_type == VDIR) {
                old_len = strlen(node->sf_path);
                while ((n = AVL_NEXT(&sfnodes, node)) != NULL) {

                        /*
                         * quit when no longer seeing children of node
                         */
                        if (n->sf_sffs != node->sf_sffs ||
                            strncmp(node->sf_path, n->sf_path, old_len) != 0 ||
                            n->sf_path[old_len] != '/')
                                break;

                        /*
                         * Rename the child:
                         * - build the new path name
                         * - unlink the AVL node
                         * - assign the new name
                         * - re-insert the AVL name
                         */
                        ASSERT(strlen(n->sf_path) > old_len);
                        tail = n->sf_path + old_len; /* includes intial "/" */
                        new_path = kmem_alloc(len + strlen(tail) + 1,
                            KM_SLEEP);
                        strcpy(new_path, path);
                        strcat(new_path, tail);
                        if (avl_find(&sfnodes, n, &where) == NULL)
                                panic("sfnode_rename(%s) not in sfnodes",
                                   n->sf_path);
                        avl_remove(&sfnodes, n);
            LogFlowFunc(("sfnode_rname(%s to %s) sub\n", n->sf_path, new_path));
                        kmem_free(n->sf_path, strlen(n->sf_path) + 1);
                        n->sf_path = new_path;
                        if (avl_find(&sfnodes, n, &where) != NULL)
                                panic("sfnode_rename(%s) duplicates",
                                    n->sf_path);
                        avl_insert(&sfnodes, n, where);
                }
        }

        /*
         * Deal with the given node.
         */
        if (avl_find(&sfnodes, node, &where) == NULL)
                panic("sfnode_rename(%s) not in sfnodes", node->sf_path);
        avl_remove(&sfnodes, node);
    LogFlowFunc(("sfnode_rname(%s to %s)\n", node->sf_path, path));
        kmem_free(node->sf_path, strlen(node->sf_path) + 1);
        node->sf_path = path;
        if (avl_find(&sfnodes, node, &where) != NULL)
                panic("sfnode_rename(%s) duplicates", node->sf_path);
        avl_insert(&sfnodes, node, where);

        /*
         * change the parent
         */
        if (node->sf_parent == NULL)
                panic("sfnode_rename(%s) no parent", node->sf_path);
        if (node->sf_parent->sf_children == 0)
                panic("sfnode_rename(%s) parent has no child", node->sf_path);
        --node->sf_parent->sf_children;
        node->sf_parent = newparent;
        ++newparent->sf_children;
}

/*
 * Look for a cached node, if not found either handle ".." or try looking
 * via the provider. Create an entry in sfnodes if found but not cached yet.
 * If the create flag is set, a file or directory is created. If the file
 * already existed, an error is returned.
 * Nodes returned from this routine always have a vnode with its ref count
 * bumped by 1.
 */
static sfnode_t *
sfnode_lookup(sfnode_t *dir, char *name, vtype_t create)
{
        avl_index_t     where;
        sfnode_t        template;
        sfnode_t        *node;
        int             error;
        int             type;
        char            *fullpath;
        sfp_file_t      *fp;

        ASSERT(MUTEX_HELD(&sffs_lock));

        /*
         * handle referencing myself
         */
        if (strcmp(name, "") == 0 || strcmp(name, ".") == 0)
                return (dir);

        /*
         * deal with parent
         */
        if (strcmp(name, "..") == 0)
                return (dir->sf_parent);

        /*
         * Look for an existing node.
         */
        fullpath = sfnode_construct_path(dir, name);
        template.sf_sffs = dir->sf_sffs;
        template.sf_path = fullpath;
        template.sf_is_stale = 0;
        node = avl_find(&sfnodes, &template, &where);
        if (node != NULL) {
                kmem_free(fullpath, strlen(fullpath) + 1);
                if (create != VNON)
                        return (NULL);
                return (node);
        }

        /*
         * No entry for this path currently.
         * Check if the file exists with the provider and get the type from
         * there.
         */
        if (create == VREG) {
                type = VREG;
                error = sfprov_create(dir->sf_sffs->sf_handle, fullpath, &fp);
        } else if (create == VDIR) {
                type = VDIR;
                error = sfprov_mkdir(dir->sf_sffs->sf_handle, fullpath, &fp);
        } else {
                mode_t m;
                fp = NULL;
                type = VNON;
                error =
                    sfprov_get_mode(dir->sf_sffs->sf_handle, fullpath, &m);
                if (error != 0)
                        error = ENOENT;
                else if (S_ISDIR(m))
                        type = VDIR;
                else if (S_ISREG(m))
                        type = VREG;
        }

        /*
         * If no errors, make a new node and return it.
         */
        if (error) {
                kmem_free(fullpath, strlen(fullpath) + 1);
                return (NULL);
        }
        node = sfnode_make(dir->sf_sffs, fullpath, type, fp, dir);
        return (node);
}


/*
 * uid and gid in sffs determine owner and group for all files.
 */
static int
sfnode_access(sfnode_t *node, mode_t mode, cred_t *cr)
{
        sffs_data_t *sffs = node->sf_sffs;
        mode_t m;
        int shift = 0;
        int error;
        vnode_t *vp;

        ASSERT(MUTEX_HELD(&sffs_lock));

        /*
         * get the mode from the provider
         */
        error = sfprov_get_mode(node->sf_sffs->sf_handle, node->sf_path, &m);
        if (error != 0) {
                m = 0;
                if (error == ENOENT)
                        sfnode_make_stale(node);
        }

        /*
         * mask off the permissions based on uid/gid
         */
        if (crgetuid(cr) != sffs->sf_uid) {
                shift += 3;
                if (groupmember(sffs->sf_gid, cr) == 0)
                        shift += 3;
        }
        mode &= ~(m << shift);

        if (mode == 0) {
                error = 0;
        } else {
                vp = sfnode_get_vnode(node);
                error = secpolicy_vnode_access(cr, vp, sffs->sf_uid, mode);
                VN_RELE(vp);
        }
        return (error);
}


/*
 *
 * Everything below this point are the vnode operations used by Solaris VFS
 */
static int
sffs_readdir(
        vnode_t         *vp,
        uio_t           *uiop,
        cred_t          *cred,
        int             *eofp,
        caller_context_t *ct,
        int             flags)
{
        sfnode_t *dir = VN2SFN(vp);
        sfnode_t *node;
        struct dirent64 *dirent;
        int dummy_eof;
        int error = 0;
        int namelen;
        void *prov_buff = NULL;
        size_t prov_buff_size;
        char **names;
        uint32_t nents;
        uint32_t index;

        if (uiop->uio_iovcnt != 1)
                return (EINVAL);

        if (vp->v_type != VDIR)
                return (ENOTDIR);

        if (eofp == NULL)
                eofp = &dummy_eof;
        *eofp = 0;

        if (uiop->uio_loffset >= MAXOFF_T) {
                *eofp = 1;
                return (0);
        }

        dirent = kmem_zalloc(DIRENT64_RECLEN(MAXNAMELEN), KM_SLEEP);

        /*
         * Get the directory entry names from the host. This gets all
         * entries, so add in starting offset. Max the caller can expect
         * would be the size of the UIO buffer / sizeof of a dirent for
         * file with name of length 1
         */
        mutex_enter(&sffs_lock);
        index = uiop->uio_loffset;
        nents = index + (uiop->uio_resid / DIRENT64_RECLEN(1));
        error = sfprov_readdir(dir->sf_sffs->sf_handle, dir->sf_path,
            &prov_buff, &prov_buff_size, &nents);
        if (error != 0)
                goto done;
        if (nents <= index) {
                *eofp = 1;
                goto done;
        }
        names = (void *)prov_buff;

        /*
         * Lookup each of the names, so that we have ino's.
         */
        for (; index < nents; ++index) {
                if (strcmp(names[index], ".") == 0) {
                        node = dir;
                } else if (strcmp(names[index], "..") == 0) {
                        node = dir->sf_parent;
                        if (node == NULL)
                                node = dir;
                } else {
                        node = sfnode_lookup(dir, names[index], VNON);
                        if (node == NULL)
                                panic("sffs_readdir() lookup failed");
                }
                namelen = strlen(names[index]);
                strcpy(&dirent->d_name[0], names[index]);
                dirent->d_reclen = DIRENT64_RECLEN(namelen);
                dirent->d_off = index;
                dirent->d_ino = node->sf_ino;
                if (dirent->d_reclen > uiop->uio_resid) {
                        error = ENOSPC;
                        break;
                }
                error = uiomove(dirent, dirent->d_reclen, UIO_READ, uiop);
                if (error != 0)
                        break;
                bzero(&dirent->d_name[0], namelen);
        }
        if (error == 0 && index >= nents)
                *eofp = 1;
done:
        mutex_exit(&sffs_lock);
        if (prov_buff != NULL)
                kmem_free(prov_buff, prov_buff_size);
        kmem_free(dirent, DIRENT64_RECLEN(MAXNAMELEN));
        return (error);
}


#if defined(VBOX_VFS_SOLARIS_10U6)
/*
 * HERE JOE.. this may need more logic, need to look at other file systems
 */
static int
sffs_pathconf(
        vnode_t *vp,
        int     cmd,
        ulong_t *valp,
        cred_t  *cr)
{
        return (fs_pathconf(vp, cmd, valp, cr));
}
#else
/*
 * HERE JOE.. this may need more logic, need to look at other file systems
 */
static int
sffs_pathconf(
        vnode_t *vp,
        int     cmd,
        ulong_t *valp,
        cred_t  *cr,
        caller_context_t *ct)
{
        return (fs_pathconf(vp, cmd, valp, cr, ct));
}
#endif

static int
sffs_getattr(
        vnode_t         *vp,
        vattr_t         *vap,
        int             flags,
        cred_t          *cred,
        caller_context_t *ct)
{
        sfnode_t        *node = VN2SFN(vp);
        sffs_data_t     *sffs = node->sf_sffs;
        mode_t          mode;
        timestruc_t     time;
        uint64_t        x;
        int             error;

        mutex_enter(&sffs_lock);
        vap->va_type = vp->v_type;
        vap->va_uid = sffs->sf_uid;
        vap->va_gid = sffs->sf_gid;
        vap->va_fsid = sffs->sf_vfsp->vfs_dev;
        vap->va_nodeid = node->sf_ino;
        vap->va_nlink = 1;
        vap->va_rdev =  sffs->sf_vfsp->vfs_dev;
        vap->va_seq = 0;

        error = sfprov_get_mode(node->sf_sffs->sf_handle, node->sf_path, &mode);
        if (error == ENOENT)
                sfnode_make_stale(node);
        if (error != 0)
                goto done;
        vap->va_mode = mode & MODEMASK;

        error = sfprov_get_size(node->sf_sffs->sf_handle, node->sf_path, &x);
        if (error == ENOENT)
                sfnode_make_stale(node);
        if (error != 0)
                goto done;
        vap->va_size = x;
        vap->va_blksize = 512;
        vap->va_nblocks = (x + 511) / 512;

        error =
            sfprov_get_atime(node->sf_sffs->sf_handle, node->sf_path, &time);
        if (error == ENOENT)
                sfnode_make_stale(node);
        if (error != 0)
                goto done;
        vap->va_atime = time;

        error =
            sfprov_get_mtime(node->sf_sffs->sf_handle, node->sf_path, &time);
        if (error == ENOENT)
                sfnode_make_stale(node);
        if (error != 0)
                goto done;
        vap->va_mtime = time;

        error =
            sfprov_get_ctime(node->sf_sffs->sf_handle, node->sf_path, &time);
        if (error == ENOENT)
                sfnode_make_stale(node);
        if (error != 0)
                goto done;
        vap->va_ctime = time;

done:
        mutex_exit(&sffs_lock);
        return (error);
}

/*ARGSUSED*/
static int
sffs_read(
        vnode_t         *vp,
        struct uio      *uio,
        int             ioflag,
        cred_t          *cred,
        caller_context_t *ct)
{
        sfnode_t        *node = VN2SFN(vp);
        int             error = 0;
        uint32_t        bytes;
        uint32_t        done;
        ulong_t         offset;
        ssize_t         total;

        if (vp->v_type == VDIR)
                return (EISDIR);
        if (vp->v_type != VREG)
                return (EINVAL);
        if (uio->uio_loffset >= MAXOFF_T)
                return (0);
        if (uio->uio_loffset < 0)
                return (EINVAL);
        total = uio->uio_resid;
        if (total == 0)
                return (0);

        mutex_enter(&sffs_lock);
        sfnode_open(node);
        if (node->sf_file == NULL) {
                mutex_exit(&sffs_lock);
                return (EINVAL);
        }

        do {
                offset = uio->uio_offset;
                done = bytes = MIN(PAGESIZE, uio->uio_resid);
                error = sfprov_read(node->sf_file, sffs_buffer, offset, &done);
                if (error == 0 && done > 0)
                        error = uiomove(sffs_buffer, done, UIO_READ, uio);
        } while (error == 0 && uio->uio_resid > 0 && done > 0);

        mutex_exit(&sffs_lock);

        /*
         * a partial read is never an error
         */
        if (total != uio->uio_resid)
                error = 0;
        return (error);
}

/*ARGSUSED*/
static int
sffs_write(
        vnode_t         *vp,
        struct uio      *uiop,
        int             ioflag,
        cred_t          *cred,
        caller_context_t *ct)
{
        sfnode_t        *node = VN2SFN(vp);
        int             error = 0;
        uint32_t        bytes;
        uint32_t        done;
        ulong_t         offset;
        ssize_t         total;
        rlim64_t        limit = uiop->uio_llimit;

        if (vp->v_type == VDIR)
                return (EISDIR);
        if (vp->v_type != VREG)
                return (EINVAL);

        /*
         * We have to hold this lock for a long time to keep
         * multiple FAPPEND writes from intermixing
         */
        mutex_enter(&sffs_lock);
        sfnode_open(node);
        if (node->sf_file == NULL) {
                mutex_exit(&sffs_lock);
                return (EINVAL);
        }
        if (ioflag & FAPPEND) {
                uint64_t endoffile;

                error = sfprov_get_size(node->sf_sffs->sf_handle,
                    node->sf_path, &endoffile);
                if (error == ENOENT)
                        sfnode_make_stale(node);
                if (error != 0) {
                        mutex_exit(&sffs_lock);
                        return (error);
                }
                uiop->uio_loffset = endoffile;
        }

        if (vp->v_type != VREG || uiop->uio_loffset < 0) {
                mutex_exit(&sffs_lock);
                return (EINVAL);
        }
        if (limit == RLIM64_INFINITY || limit > MAXOFFSET_T)
                limit = MAXOFFSET_T;
        if (limit > MAXOFF_T)
                limit = MAXOFF_T;

        if (uiop->uio_loffset >= limit) {
                proc_t *p = ttoproc(curthread);
                mutex_enter(&p->p_lock);
                (void) rctl_action(rctlproc_legacy[RLIMIT_FSIZE], p->p_rctls,
                    p, RCA_UNSAFE_SIGINFO);
                mutex_exit(&p->p_lock);
                mutex_exit(&sffs_lock);
                return (EFBIG);
        }

        if (uiop->uio_loffset >= MAXOFF_T) {
                mutex_exit(&sffs_lock);
                return (EFBIG);
        }


        total = uiop->uio_resid;
        if (total == 0) {
                mutex_exit(&sffs_lock);
                return (0);
        }

        do {
                offset = uiop->uio_offset;
                bytes = MIN(PAGESIZE, uiop->uio_resid);
                if (offset + bytes >= limit) {
                        if (offset >= limit) {
                                error = EFBIG;
                                break;
                        }
                        bytes = limit - offset;
                }
                error = uiomove(sffs_buffer, bytes, UIO_WRITE, uiop);
                if (error != 0)
                        break;
                done = bytes;
                if (error == 0)
                        error = sfprov_write(node->sf_file, sffs_buffer,
                            offset, &done);
                total -= done;
                if (done != bytes) {
                        uiop->uio_resid += bytes - done;
                        break;
                }
        } while (error == 0 && uiop->uio_resid > 0 && done > 0);

        mutex_exit(&sffs_lock);

        /*
         * A short write is never really an error.
         */
        if (total != uiop->uio_resid)
                error = 0;
        return (error);
}

/*ARGSUSED*/
static int
sffs_access(vnode_t *vp, int mode, int flags, cred_t *cr, caller_context_t *ct)
{
        sfnode_t *node = VN2SFN(vp);
        int error;

        mutex_enter(&sffs_lock);
        error = sfnode_access(node, mode, cr);
        mutex_exit(&sffs_lock);
        return (error);
}

/*
 * Lookup an entry in a directory and create a new vnode if found.
 */
/* ARGSUSED3 */
static int
sffs_lookup(
        vnode_t         *dvp,           /* the directory vnode */
        char            *name,          /* the name of the file or directory */
        vnode_t         **vpp,          /* the vnode we found or NULL */
        struct pathname *pnp,
        int             flags,
        vnode_t         *rdir,
        cred_t          *cred,
        caller_context_t *ct,
        int             *direntflags,
        struct pathname *realpnp)
{
        int             error;
        sfnode_t        *node;

        /*
         * dvp must be a directory
         */
        if (dvp->v_type != VDIR)
                return (ENOTDIR);

        /*
         * An empty component name or just "." means the directory itself.
         * Don't do any further lookup or checking.
         */
        if (strcmp(name, "") == 0 || strcmp(name, ".") == 0) {
                VN_HOLD(dvp);
                *vpp = dvp;
                return (0);
        }

        /*
         * Check permission to look at this directory. We always allow "..".
         */
        mutex_enter(&sffs_lock);
        if (strcmp(name, "..") != 0) {
                error = sfnode_access(VN2SFN(dvp), VEXEC, cred);
                if (error) {
                        mutex_exit(&sffs_lock);
                        return (error);
                }
        }

        /*
         * Lookup the node.
         */
        node = sfnode_lookup(VN2SFN(dvp), name, VNON);
        if (node != NULL)
                *vpp = sfnode_get_vnode(node);
        mutex_exit(&sffs_lock);
        return ((node == NULL) ? ENOENT : 0);
}

/*ARGSUSED*/
static int
sffs_create(
        vnode_t         *dvp,
        char            *name,
        struct vattr    *vap,
        vcexcl_t        exclusive,
        int             mode,
        vnode_t         **vpp,
        cred_t          *cr,
        int             flag,
        caller_context_t *ct,
        vsecattr_t      *vsecp)
{
        vnode_t         *vp;
        sfnode_t        *node;
        int             error;

        ASSERT(name != NULL);

        /*
         * this is used for regular files, not mkdir
         */
        if (vap->va_type == VDIR)
                return (EISDIR);
        if (vap->va_type != VREG)
                return (EINVAL);

        /*
         * is this a pre-existing file?
         */
        error = sffs_lookup(dvp, name, &vp,
            NULL, 0, NULL, cr, ct, NULL, NULL);
        if (error == ENOENT)
                vp = NULL;
        else if (error != 0)
                return (error);

        /*
         * Operation on a pre-existing file.
         */
        if (vp != NULL) {
                if (exclusive == EXCL) {
                        VN_RELE(vp);
                        return (EEXIST);
                }
                if (vp->v_type == VDIR && (mode & VWRITE) == VWRITE) {
                        VN_RELE(vp);
                        return (EISDIR);
                }

                mutex_enter(&sffs_lock);
                node = VN2SFN(vp);
                error = sfnode_access(node, mode, cr);
                if (error != 0) {
                        mutex_exit(&sffs_lock);
                        VN_RELE(vp);
                        return (error);
                }

                /*
                 * handle truncating an existing file
                 */
                if (vp->v_type == VREG && (vap->va_mask & AT_SIZE) &&
                    vap->va_size == 0) {
                        sfnode_open(node);
                        if (node->sf_path == NULL)
                                error = ENOENT;
                        else
                                error = sfprov_trunc(node->sf_sffs->sf_handle,
                                    node->sf_path);
                        if (error) {
                                mutex_exit(&sffs_lock);
                                VN_RELE(vp);
                                return (error);
                        }
                }
                mutex_exit(&sffs_lock);
                *vpp = vp;
                return (0);
        }

        /*
         * Create a new node. First check for a race creating it.
         */
        mutex_enter(&sffs_lock);
        node = sfnode_lookup(VN2SFN(dvp), name, VNON);
        if (node != NULL) {
                mutex_exit(&sffs_lock);
                return (EEXIST);
        }

        /*
         * Doesn't exist yet and we have the lock, so create it.
         */
        node = sfnode_lookup(VN2SFN(dvp), name, VREG);
        mutex_exit(&sffs_lock);
        if (node == NULL)
                return (EINVAL);
        *vpp = sfnode_get_vnode(node);
        return (0);
}

/*ARGSUSED*/
static int
sffs_mkdir(
        vnode_t         *dvp,
        char            *nm,
        vattr_t         *va,
        vnode_t         **vpp,
        cred_t          *cred,
        caller_context_t *ct,
        int             flags,
        vsecattr_t      *vsecp)
{
        sfnode_t        *node;
        vnode_t         *vp;
        int             error;

        /*
         * These should never happen
         */
        ASSERT(nm != NULL);
        ASSERT(strcmp(nm, "") != 0);
        ASSERT(strcmp(nm, ".") != 0);
        ASSERT(strcmp(nm, "..") != 0);

        /*
         * Do an unlocked look up first
         */
        error = sffs_lookup(dvp, nm, &vp, NULL, 0, NULL, cred, ct, NULL, NULL);
        if (error == 0) {
                VN_RELE(vp);
                return (EEXIST);
        }
        if (error != ENOENT)
                return (error);

        /*
         * Must be able to write in current directory
         */
        mutex_enter(&sffs_lock);
        error = sfnode_access(VN2SFN(dvp), VWRITE, cred);
        if (error) {
                mutex_exit(&sffs_lock);
                return (error);
        }

        node = sfnode_lookup(VN2SFN(dvp), nm, VDIR);
        mutex_exit(&sffs_lock);
        if (node == NULL)
                return (EACCES);
        *vpp = sfnode_get_vnode(node);
        return (0);
}

/*ARGSUSED*/
static int
sffs_rmdir(
        struct vnode    *dvp,
        char            *nm,
        vnode_t         *cdir,
        cred_t          *cred,
        caller_context_t *ct,
        int             flags)
{
        sfnode_t         *node;
        vnode_t         *vp;
        int             error;

        /*
         * Return error when removing . and ..
         */
        if (strcmp(nm, ".") == 0 || strcmp(nm, "") == 0)
                return (EINVAL);
        if (strcmp(nm, "..") == 0)
                return (EEXIST);

        error = sffs_lookup(dvp, nm, &vp, NULL, 0, NULL, cred, ct, NULL, NULL);
        if (error)
                return (error);
        if (vp->v_type != VDIR) {
                VN_RELE(vp);
                return (ENOTDIR);
        }
        if (vn_vfswlock(vp) || vn_mountedvfs(vp)) {
                VN_RELE(vp);
                return (EBUSY);
        }

        node = VN2SFN(vp);

        mutex_enter(&sffs_lock);
        error = sfnode_access(VN2SFN(dvp), VEXEC | VWRITE, cred);
        if (error)
                goto done;

        /*
         * If anything else is using this vnode, then fail the remove.
         * Why?  Windows hosts can't remove something that is open,
         * so we have to sfprov_close() it first.
         * There is no errno for this - since it's not a problem on UNIX,
         * but EINVAL is the closest.
         */
        if (node->sf_file != NULL) {
                if (vp->v_count > 1) {
                        error = EINVAL;
                        goto done;
                }
                (void)sfprov_close(node->sf_file);
                node->sf_file = NULL;
        }

        /*
         * Remove the directory on the host and mark the node as stale.
         */
        error = sfprov_rmdir(node->sf_sffs->sf_handle, node->sf_path);
        if (error == ENOENT || error == 0)
                sfnode_make_stale(node);
done:
        mutex_exit(&sffs_lock);
        VN_RELE(vp);
        return (error);
}


/*ARGSUSED*/
static int
sffs_remove(
        vnode_t         *dvp,
        char            *name,
        cred_t          *cred,
        caller_context_t *ct,
        int             flags)
{
        vnode_t         *vp;
        sfnode_t        *node;
        int             error;

        /*
         * These should never happen
         */
        ASSERT(name != NULL);
        ASSERT(strcmp(name, "..") != 0);

        error = sffs_lookup(dvp, name, &vp,
            NULL, 0, NULL, cred, ct, NULL, NULL);
        if (error)
                return (error);
        node = VN2SFN(vp);

        mutex_enter(&sffs_lock);
        error = sfnode_access(VN2SFN(dvp), VEXEC | VWRITE, cred);
        if (error)
                goto done;

        /*
         * If anything else is using this vnode, then fail the remove.
         * Why?  Windows hosts can't sfprov_remove() a file that is open,
         * so we have to sfprov_close() it first.
         * There is no errno for this - since it's not a problem on UNIX,
         * but ETXTBSY is the closest.
         */
        if (node->sf_file != NULL) {
                if (vp->v_count > 1) {
                        error = ETXTBSY;
                        goto done;
                }
                (void)sfprov_close(node->sf_file);
                node->sf_file = NULL;
        }

        /*
         * Remove the file on the host and mark the node as stale.
         */
        error = sfprov_remove(node->sf_sffs->sf_handle, node->sf_path);
        if (error == ENOENT || error == 0)
                sfnode_make_stale(node);
done:
        mutex_exit(&sffs_lock);
        VN_RELE(vp);
        return (error);
}

/*ARGSUSED*/
static int
sffs_rename(
        vnode_t         *old_dir,
        char            *old_nm,
        vnode_t         *new_dir,
        char            *new_nm,
        cred_t          *cred,
        caller_context_t *ct,
        int             flags)
{
        char            *newpath;
        int             error;
        sfnode_t        *node;

        if (strcmp(new_nm, "") == 0 ||
            strcmp(new_nm, ".") == 0 ||
            strcmp(new_nm, "..") == 0 ||
            strcmp(old_nm, "") == 0 ||
            strcmp(old_nm, ".") == 0 ||
            strcmp(old_nm, "..") == 0)
                return (EINVAL);

        /*
         * make sure we have permission to do the rename
         */
        mutex_enter(&sffs_lock);
        error = sfnode_access(VN2SFN(old_dir), VEXEC | VWRITE, cred);
        if (error == 0 && new_dir != old_dir)
                error = sfnode_access(VN2SFN(new_dir), VEXEC | VWRITE, cred);
        if (error)
                goto done;

        node = sfnode_lookup(VN2SFN(old_dir), old_nm, VNON);
        if (node == NULL) {
                error = ENOENT;
                goto done;
        }


        /*
         * Rename the file on the host and in our caches.
         */
        newpath = sfnode_construct_path(VN2SFN(new_dir), new_nm);
        error = sfprov_rename(node->sf_sffs->sf_handle, node->sf_path, newpath,
            node->sf_type == VDIR);
        if (error == 0)
                sfnode_rename(node, VN2SFN(new_dir), newpath);
        else {
                kmem_free(newpath, strlen(newpath) + 1);
                if (error == ENOENT)
                        sfnode_make_stale(node);
        }
done:
        mutex_exit(&sffs_lock);
        return (error);
}


/*ARGSUSED*/
static int
sffs_fsync(vnode_t *vp, int flag, cred_t *cr, caller_context_t *ct)
{
#if 0
        sfnode_t *node;

        /*
         * Ask the host to sync any data it may have cached for open files.
         * I don't think we care about errors.
         */
        mutex_enter(&sffs_lock);
        node = VN2SFN(vp);
        if (node->sf_file != NULL)
                (void) sfprov_fsync(node->sf_file);
        mutex_exit(&sffs_lock);
#endif
        return (0);
}

/*
 * This may be the last reference, possibly time to close the file and
 * destroy the vnode. If the sfnode is stale, we'll destroy that too.
 */
/*ARGSUSED*/
static void
sffs_inactive(vnode_t *vp, cred_t *cr, caller_context_t *ct)
{
        sfnode_t *node;

        /*
         * nothing to do if this isn't the last use
         */
        mutex_enter(&sffs_lock);
        node = VN2SFN(vp);
        mutex_enter(&vp->v_lock);
        if (vp->v_count > 1) {
                --vp->v_count;
                mutex_exit(&vp->v_lock);
                mutex_exit(&sffs_lock);
                return;
        }

        /*
         * There should never be cached data, since we don't support mmap().
         */
        if (vn_has_cached_data(vp))
                panic("sffs_inactive() found cached data");

        /*
         * destroy the vnode
         */
        node->sf_vnode = NULL;
        mutex_exit(&vp->v_lock);
        vn_invalid(vp);
        vn_free(vp);
    LogFlowFunc(("  %s vnode cleared\n", node->sf_path));

        /*
         * Close the sf_file for the node.
         */
        if (node->sf_file != NULL) {
                (void)sfprov_close(node->sf_file);
                node->sf_file = NULL;
        }

        /*
         * If the node is stale, we can also destroy it.
         */
        if (node->sf_is_stale && node->sf_children == 0)
                sfnode_destroy(node);

        mutex_exit(&sffs_lock);
        return;
}

/*
 * All the work for this is really done in lookup.
 */
/*ARGSUSED*/
static int
sffs_open(vnode_t **vpp, int flag, cred_t *cr, caller_context_t *ct)
{
        return (0);
}

/*
 * All the work for this is really done in inactive.
 */
/*ARGSUSED*/
static int
sffs_close(
        vnode_t *vp,
        int flag,
        int count,
        offset_t offset,
        cred_t *cr,
        caller_context_t *ct)
{
        return (0);
}

/* ARGSUSED */
static int
sffs_seek(vnode_t *v, offset_t o, offset_t *no, caller_context_t *ct)
{
        if (*no < 0 || *no > MAXOFFSET_T)
                return (EINVAL);
        return (0);
}



/*
 * By returning an error for this, we prevent anything in sffs from
 * being re-exported by NFS
 */
/* ARGSUSED */
static int
sffs_fid(vnode_t *vp, fid_t *fidp, caller_context_t *ct)
{
        return (ENOTSUP);
}

/*
 * vnode operations for regular files
 */
const fs_operation_def_t sffs_ops_template[] = {
#if defined(VBOX_VFS_SOLARIS_10U6)
        VOPNAME_ACCESS,         sffs_access,
        VOPNAME_CLOSE,          sffs_close,
        VOPNAME_CREATE,         sffs_create,
        VOPNAME_FID,            sffs_fid,
        VOPNAME_FSYNC,          sffs_fsync,
        VOPNAME_GETATTR,        sffs_getattr,
        VOPNAME_INACTIVE,       sffs_inactive,
        VOPNAME_LOOKUP,         sffs_lookup,
        VOPNAME_MKDIR,          sffs_mkdir,
        VOPNAME_OPEN,           sffs_open,
        VOPNAME_PATHCONF,       sffs_pathconf,
        VOPNAME_READ,           sffs_read,
        VOPNAME_READDIR,        sffs_readdir,
        VOPNAME_REMOVE,         sffs_remove,
        VOPNAME_RENAME,         sffs_rename,
        VOPNAME_RMDIR,          sffs_rmdir,
        VOPNAME_SEEK,           sffs_seek,
        VOPNAME_WRITE,          sffs_write,
        NULL,                   NULL
#else
        VOPNAME_ACCESS,         { .vop_access = sffs_access },
        VOPNAME_CLOSE,          { .vop_close = sffs_close },
        VOPNAME_CREATE,         { .vop_create = sffs_create },
        VOPNAME_FID,            { .vop_fid = sffs_fid },
        VOPNAME_FSYNC,          { .vop_fsync = sffs_fsync },
        VOPNAME_GETATTR,        { .vop_getattr = sffs_getattr },
        VOPNAME_INACTIVE,       { .vop_inactive = sffs_inactive },
        VOPNAME_LOOKUP,         { .vop_lookup = sffs_lookup },
        VOPNAME_MKDIR,          { .vop_mkdir = sffs_mkdir },
        VOPNAME_OPEN,           { .vop_open = sffs_open },
        VOPNAME_PATHCONF,       { .vop_pathconf = sffs_pathconf },
        VOPNAME_READ,           { .vop_read = sffs_read },
        VOPNAME_READDIR,        { .vop_readdir = sffs_readdir },
        VOPNAME_REMOVE,         { .vop_remove = sffs_remove },
        VOPNAME_RENAME,         { .vop_rename = sffs_rename },
        VOPNAME_RMDIR,          { .vop_rmdir = sffs_rmdir },
        VOPNAME_SEEK,           { .vop_seek = sffs_seek },
        VOPNAME_WRITE,          { .vop_write = sffs_write },
        NULL,                   NULL
#endif
};

/*
 * Also, init and fini functions...
 */
int
sffs_vnode_init(void)
{
        int err;

        err = vn_make_ops("sffs", sffs_ops_template, &sffs_ops);
        if (err)
                return (err);

        avl_create(&sfnodes, sfnode_compare, sizeof (sfnode_t),
            offsetof(sfnode_t, sf_linkage));
        avl_create(&stale_sfnodes, sfnode_compare, sizeof (sfnode_t),
            offsetof(sfnode_t, sf_linkage));

        sffs_buffer = kmem_alloc(PAGESIZE, KM_SLEEP);

        return (0);
}

void
sffs_vnode_fini(void)
{
        if (sffs_ops)
                vn_freevnodeops(sffs_ops);
        ASSERT(avl_first(&sfnodes) == NULL);
        avl_destroy(&sfnodes);
        if (sffs_buffer != NULL) {
                kmem_free(sffs_buffer, PAGESIZE);
                sffs_buffer = NULL;
        }
}

/*
 * Utility at unmount to get all nodes in that mounted filesystem removed.
 */
int
sffs_purge(struct sffs_data *sffs)
{
        sfnode_t *node;
        sfnode_t *prev;

        /*
         * Check that no vnodes are active.
         */
        if (sffs->sf_rootnode->v_count > 1)
                return (-1);
        for (node = avl_first(&sfnodes); node;
            node = AVL_NEXT(&sfnodes, node)) {
                if (node->sf_sffs == sffs && node->sf_vnode &&
                    node->sf_vnode != sffs->sf_rootnode)
                        return (-1);
        }
        for (node = avl_first(&stale_sfnodes); node;
            node = AVL_NEXT(&stale_sfnodes, node)) {
                if (node->sf_sffs == sffs && node->sf_vnode &&
                    node->sf_vnode != sffs->sf_rootnode)
                        return (-1);
        }

        /*
         * All clear to destroy all node information. Since there are no
         * vnodes, the make stale will cause deletion.
         */
        VN_RELE(sffs->sf_rootnode);
        mutex_enter(&sffs_lock);
        for (prev = NULL;;) {
                if (prev == NULL)
                        node = avl_first(&sfnodes);
                else
                        node = AVL_NEXT(&sfnodes, prev);

                if (node == NULL)
                        break;

                if (node->sf_sffs == sffs) {
                        if (node->sf_vnode != NULL)
                                panic("vboxfs: purge hit active vnode");
                        sfnode_make_stale(node);
                } else {
                        prev = node;
                }
        }
done:
        mutex_exit(&sffs_lock);
        return (0);
}

static void
sfnode_print(sfnode_t *node)
{
        Log(("0x%p", node));
        Log((" type=%s (%d)",
            node->sf_type == VDIR ? "VDIR" :
            node->sf_type == VNON ? "VNON" :
            node->sf_type == VREG ? "VREG" : "other", node->sf_type));
        Log((" ino=%d", (uint_t)node->sf_ino));
        Log((" path=%s", node->sf_path));
        Log((" parent=0x%p", node->sf_parent));
        if (node->sf_children)
                Log((" children=%d", node->sf_children));
        if (node->sf_vnode)
                Log((" vnode=0x%p", node->sf_vnode));
        Log(("%s\n", node->sf_is_stale ? " STALE" : ""));
}

void
sfnode_list()
{
        sfnode_t *n;
        for (n = avl_first(&sfnodes); n != NULL; n = AVL_NEXT(&sfnodes, n))
                sfnode_print(n);
        for (n = avl_first(&stale_sfnodes); n != NULL;
            n = AVL_NEXT(&stale_sfnodes, n))
                sfnode_print(n);
}