r15237: Add primitive Samba 4 backend for cifs posix clients (based on simple backend).
authorSteve French <sfrench@samba.org>
Tue, 25 Apr 2006 11:36:15 +0000 (11:36 +0000)
committerGerald (Jerry) Carter <jerry@samba.org>
Wed, 10 Oct 2007 19:04:23 +0000 (14:04 -0500)
(This used to be commit f054e1a9e6101f22559f5c4a546b537531b60f94)

source4/ntvfs/cifs_posix_cli/README [new file with mode: 0644]
source4/ntvfs/cifs_posix_cli/cvfs.h [new file with mode: 0644]
source4/ntvfs/cifs_posix_cli/svfs_util.c [new file with mode: 0644]
source4/ntvfs/cifs_posix_cli/vfs_simple.c [new file with mode: 0644]
source4/ntvfs/config.mk

diff --git a/source4/ntvfs/cifs_posix_cli/README b/source4/ntvfs/cifs_posix_cli/README
new file mode 100644 (file)
index 0000000..f1de5d9
--- /dev/null
@@ -0,0 +1,10 @@
+This module (ntvfs 'simple') provides a very, very simple posix backend.
+
+WARNING: All file access is done as user 'root'!!!
+        Only use this module for testing, with only test data!!! 
+
+For activating this module use:
+
+[testshare]
+       path = /tmp/testshare
+       nfvfs handler = simple
diff --git a/source4/ntvfs/cifs_posix_cli/cvfs.h b/source4/ntvfs/cifs_posix_cli/cvfs.h
new file mode 100644 (file)
index 0000000..8e462d9
--- /dev/null
@@ -0,0 +1,35 @@
+
+struct svfs_private {
+       /* the base directory */
+       char *connectpath;
+
+       /* a linked list of open searches */
+       struct search_state *search;
+
+       /* next available search handle */
+       uint16_t next_search_handle;
+
+       struct svfs_file *open_files;
+};
+
+struct svfs_dir {
+       uint_t count;
+       char *unix_dir;
+       struct svfs_dirfile {
+               char *name;
+               struct stat st;
+       } *files;
+};
+
+struct svfs_file {
+       struct svfs_file *next, *prev;
+       int fd;
+       char *name;
+};
+
+struct search_state {
+       struct search_state *next, *prev;
+       uint16_t handle;
+       uint_t current_index;
+       struct svfs_dir *dir;
+};
diff --git a/source4/ntvfs/cifs_posix_cli/svfs_util.c b/source4/ntvfs/cifs_posix_cli/svfs_util.c
new file mode 100644 (file)
index 0000000..b85c43b
--- /dev/null
@@ -0,0 +1,185 @@
+/* 
+   Unix SMB/CIFS implementation.
+
+   simple NTVFS filesystem backend
+
+   Copyright (C) Andrew Tridgell 2003
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 2 of the License, or
+   (at your option) any later version.
+   
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+   
+   You should have received a copy of the GNU General Public License
+   along with this program; if not, write to the Free Software
+   Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+*/
+/*
+  utility functions for simple backend
+*/
+
+#include "includes.h"
+#include "system/filesys.h"
+#include "cvfs.h"
+#include "system/time.h"
+#include "system/dir.h"
+#include "ntvfs/ntvfs.h"
+
+/*
+  convert a windows path to a unix path - don't do any manging or case sensitive handling
+*/
+char *cvfs_unix_path(struct ntvfs_module_context *ntvfs,
+                    struct ntvfs_request *req, const char *name)
+{
+       struct svfs_private *private = ntvfs->private_data;
+       char *ret;
+
+       if (*name != '\\') {
+               ret = talloc_asprintf(req, "%s/%s", private->connectpath, name);
+       } else {
+               ret = talloc_asprintf(req, "%s%s", private->connectpath, name);
+       }
+       all_string_sub(ret, "\\", "/", 0);
+
+       strlower(ret + strlen(private->connectpath));
+
+       return ret;
+}
+
+
+/*
+  read a directory and find all matching file names and stat info
+  returned names are separate unix and DOS names. The returned names
+  are relative to the directory
+*/
+struct svfs_dir *cvfs_list_unix(TALLOC_CTX *mem_ctx, struct ntvfs_request *req, const char *unix_path)
+{
+       char *p, *mask;
+       struct svfs_dir *dir;
+       DIR *odir;
+       struct dirent *dent;
+       uint_t allocated = 0;
+       char *low_mask;
+
+       dir = talloc(mem_ctx, struct svfs_dir);
+       if (!dir) { return NULL; }
+
+       dir->count = 0;
+       dir->files = 0;
+
+       /* find the base directory */
+       p = strrchr(unix_path, '/');
+       if (!p) { return NULL; }
+
+       dir->unix_dir = talloc_strndup(mem_ctx, unix_path, PTR_DIFF(p, unix_path));
+       if (!dir->unix_dir) { return NULL; }
+
+       /* the wildcard pattern is the last part */
+       mask = p+1;
+
+       low_mask = talloc_strdup(mem_ctx, mask);
+       if (!low_mask) { return NULL; }
+       strlower(low_mask);
+
+       odir = opendir(dir->unix_dir);
+       if (!odir) { return NULL; }
+
+       while ((dent = readdir(odir))) {
+               uint_t i = dir->count;
+               char *full_name;
+               char *low_name;
+
+               if (strchr(dent->d_name, ':') && !strchr(unix_path, ':')) {
+                       /* don't show streams in dir listing */
+                       continue;
+               }
+
+               low_name = talloc_strdup(mem_ctx, dent->d_name);
+               if (!low_name) { continue; }
+               strlower(low_name);
+
+               /* check it matches the wildcard pattern */
+               if (ms_fnmatch(low_mask, low_name, PROTOCOL_NT1) != 0) {
+                       continue;
+               }
+               
+               if (dir->count >= allocated) {
+                       allocated = (allocated + 100) * 1.2;
+                       dir->files = talloc_realloc(dir, dir->files, struct svfs_dirfile, allocated);
+                       if (!dir->files) { 
+                               closedir(odir);
+                               return NULL;
+                       }
+               }
+
+               dir->files[i].name = low_name;
+               if (!dir->files[i].name) { continue; }
+
+               asprintf(&full_name, "%s/%s", dir->unix_dir, dir->files[i].name);
+               if (!full_name) { continue; }
+
+               if (stat(full_name, &dir->files[i].st) == 0) { 
+                       dir->count++;
+               }
+
+               free(full_name); 
+       }
+
+       closedir(odir);
+
+       return dir;
+}
+
+/*
+  read a directory and find all matching file names and stat info
+  returned names are separate unix and DOS names. The returned names
+  are relative to the directory
+*/
+struct svfs_dir *cvfs_list(struct ntvfs_module_context *ntvfs, struct ntvfs_request *req, const char *pattern)
+{
+       struct svfs_private *private = ntvfs->private_data;
+       char *unix_path;
+
+       unix_path = cvfs_unix_path(ntvfs, req, pattern);
+       if (!unix_path) { return NULL; }
+
+       return cvfs_list_unix(private, req, unix_path);
+}
+
+
+/*******************************************************************
+set the time on a file via file descriptor
+*******************************************************************/
+int cvfs_file_utime(int fd, struct utimbuf *times)
+{
+       char *fd_path = NULL;
+       int ret;
+
+       asprintf(&fd_path, "/proc/self/%d", fd);
+       if (!fd_path) {
+               errno = ENOMEM;
+               return -1;
+       }
+       
+       ret = utime(fd_path, times);
+       free(fd_path);
+       return ret;
+}
+
+
+/*
+  map a unix file attrib to a DOS attribute
+*/
+uint16_t cvfs_unix_to_dos_attrib(mode_t mode)
+{
+       uint16_t ret = 0;
+       if (S_ISDIR(mode)) ret |= FILE_ATTRIBUTE_DIRECTORY;
+       if (!(mode & S_IWUSR)) ret |= FILE_ATTRIBUTE_READONLY;
+       return ret;
+}
+
diff --git a/source4/ntvfs/cifs_posix_cli/vfs_simple.c b/source4/ntvfs/cifs_posix_cli/vfs_simple.c
new file mode 100644 (file)
index 0000000..cdecf5f
--- /dev/null
@@ -0,0 +1,1022 @@
+/* 
+   Unix SMB/CIFS implementation.
+
+   simple NTVFS filesystem backend
+
+   Copyright (C) Andrew Tridgell 2003
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 2 of the License, or
+   (at your option) any later version.
+   
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+   
+   You should have received a copy of the GNU General Public License
+   along with this program; if not, write to the Free Software
+   Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+*/
+/*
+  this implements a very simple NTVFS filesystem backend. 
+  
+  this backend largely ignores the POSIX -> CIFS mappings, just doing absolutely
+  minimal work to give a working backend.
+*/
+
+#include "includes.h"
+#include "system/dir.h"
+#include "system/filesys.h"
+#include "cvfs.h"
+#include "system/time.h"
+#include "dlinklist.h"
+#include "ntvfs/ntvfs.h"
+#include "ntvfs/cifs_posix_cli/proto.h"
+
+#ifndef O_DIRECTORY
+#define O_DIRECTORY 0
+#endif
+
+#define CHECK_READ_ONLY(req) do { if (lp_readonly(ntvfs->ctx->config.snum)) return NT_STATUS_ACCESS_DENIED; } while (0)
+
+/*
+  connect to a share - used when a tree_connect operation comes
+  in. For a disk based backend we needs to ensure that the base
+  directory exists (tho it doesn't need to be accessible by the user,
+  that comes later)
+*/
+static NTSTATUS svfs_connect(struct ntvfs_module_context *ntvfs,
+                            struct ntvfs_request *req, const char *sharename)
+{
+       struct stat st;
+       struct svfs_private *private;
+       int snum = ntvfs->ctx->config.snum;
+
+       private = talloc(ntvfs, struct svfs_private);
+
+       private->next_search_handle = 0;
+       private->connectpath = talloc_strdup(private, lp_pathname(snum));
+       private->open_files = NULL;
+       private->search = NULL;
+
+       /* the directory must exist */
+       if (stat(private->connectpath, &st) != 0 || !S_ISDIR(st.st_mode)) {
+               DEBUG(0,("'%s' is not a directory, when connecting to [%s]\n", 
+                        private->connectpath, sharename));
+               return NT_STATUS_BAD_NETWORK_NAME;
+       }
+
+       ntvfs->ctx->fs_type = talloc_strdup(ntvfs->ctx, "NTFS");
+       NT_STATUS_HAVE_NO_MEMORY(ntvfs->ctx->fs_type);
+       ntvfs->ctx->dev_type = talloc_strdup(ntvfs->ctx, "A:");
+       NT_STATUS_HAVE_NO_MEMORY(ntvfs->ctx->dev_type);
+
+       ntvfs->private_data = private;
+
+       DEBUG(0,("WARNING: ntvfs simple: connect to share [%s] with ROOT privileges!!!\n",sharename));
+
+       return NT_STATUS_OK;
+}
+
+/*
+  disconnect from a share
+*/
+static NTSTATUS svfs_disconnect(struct ntvfs_module_context *ntvfs)
+{
+       return NT_STATUS_OK;
+}
+
+/*
+  find open file handle given fd
+*/
+static struct svfs_file *find_fd(struct svfs_private *private, int fd)
+{
+       struct svfs_file *f;
+       for (f=private->open_files;f;f=f->next) {
+               if (f->fd == fd) {
+                       return f;
+               }
+       }
+       return NULL;
+}
+
+/*
+  delete a file - the dirtype specifies the file types to include in the search. 
+  The name can contain CIFS wildcards, but rarely does (except with OS/2 clients)
+*/
+static NTSTATUS svfs_unlink(struct ntvfs_module_context *ntvfs,
+                           struct ntvfs_request *req,
+                           union smb_unlink *unl)
+{
+       char *unix_path;
+
+       CHECK_READ_ONLY(req);
+
+       unix_path = cvfs_unix_path(ntvfs, req, unl->unlink.in.pattern);
+
+       /* ignoring wildcards ... */
+       if (unlink(unix_path) == -1) {
+               return map_nt_error_from_unix(errno);
+       }
+
+       return NT_STATUS_OK;
+}
+
+
+/*
+  ioctl interface - we don't do any
+*/
+static NTSTATUS svfs_ioctl(struct ntvfs_module_context *ntvfs,
+                          struct ntvfs_request *req, union smb_ioctl *io)
+{
+       return NT_STATUS_INVALID_PARAMETER;
+}
+
+/*
+  check if a directory exists
+*/
+static NTSTATUS svfs_chkpath(struct ntvfs_module_context *ntvfs,
+                            struct ntvfs_request *req,
+                            union smb_chkpath *cp)
+{
+       char *unix_path;
+       struct stat st;
+
+       unix_path = cvfs_unix_path(ntvfs, req, cp->chkpath.in.path);
+
+       if (stat(unix_path, &st) == -1) {
+               return map_nt_error_from_unix(errno);
+       }
+
+       if (!S_ISDIR(st.st_mode)) {
+               return NT_STATUS_NOT_A_DIRECTORY;
+       }
+
+       return NT_STATUS_OK;
+}
+
+/*
+  build a file_id from a stat struct
+*/
+static uint64_t svfs_file_id(struct stat *st)
+{
+       uint64_t ret = st->st_ino;
+       ret <<= 32;
+       ret |= st->st_dev;
+       return ret;
+}
+
+/*
+  approximately map a struct stat to a generic fileinfo struct
+*/
+static NTSTATUS svfs_map_fileinfo(struct ntvfs_module_context *ntvfs,
+                                 struct ntvfs_request *req, union smb_fileinfo *info, 
+                                 struct stat *st, const char *unix_path)
+{
+       struct svfs_dir *dir = NULL;
+       char *pattern = NULL;
+       int i;
+       const char *s, *short_name;
+
+       s = strrchr(unix_path, '/');
+       if (s) {
+               short_name = s+1;
+       } else {
+               short_name = "";
+       }
+
+       asprintf(&pattern, "%s:*", unix_path);
+       
+       if (pattern) {
+               dir = cvfs_list_unix(req, req, pattern);
+       }
+
+       unix_to_nt_time(&info->generic.out.create_time, st->st_ctime);
+       unix_to_nt_time(&info->generic.out.access_time, st->st_atime);
+       unix_to_nt_time(&info->generic.out.write_time,  st->st_mtime);
+       unix_to_nt_time(&info->generic.out.change_time, st->st_mtime);
+       info->generic.out.alloc_size = st->st_size;
+       info->generic.out.size = st->st_size;
+       info->generic.out.attrib = cvfs_unix_to_dos_attrib(st->st_mode);
+       info->generic.out.alloc_size = st->st_blksize * st->st_blocks;
+       info->generic.out.nlink = st->st_nlink;
+       info->generic.out.directory = S_ISDIR(st->st_mode) ? 1 : 0;
+       info->generic.out.file_id = svfs_file_id(st);
+       /* REWRITE: TODO stuff in here */
+       info->generic.out.delete_pending = 0;
+       info->generic.out.ea_size = 0;
+       info->generic.out.num_eas = 0;
+       info->generic.out.fname.s = talloc_strdup(req, short_name);
+       info->generic.out.alt_fname.s = talloc_strdup(req, short_name);
+       info->generic.out.compressed_size = 0;
+       info->generic.out.format = 0;
+       info->generic.out.unit_shift = 0;
+       info->generic.out.chunk_shift = 0;
+       info->generic.out.cluster_shift = 0;
+       
+       info->generic.out.access_flags = 0;
+       info->generic.out.position = 0;
+       info->generic.out.mode = 0;
+       info->generic.out.alignment_requirement = 0;
+       info->generic.out.reparse_tag = 0;
+       info->generic.out.num_streams = 0;
+       /* setup a single data stream */
+       info->generic.out.num_streams = 1 + (dir?dir->count:0);
+       info->generic.out.streams = talloc_array(req, 
+                                                  struct stream_struct,
+                                                  info->generic.out.num_streams);
+       if (!info->generic.out.streams) {
+               return NT_STATUS_NO_MEMORY;
+       }
+       info->generic.out.streams[0].size = st->st_size;
+       info->generic.out.streams[0].alloc_size = st->st_size;
+       info->generic.out.streams[0].stream_name.s = talloc_strdup(req,"::$DATA");
+
+       for (i=0;dir && i<dir->count;i++) {
+               s = strchr(dir->files[i].name, ':');
+               info->generic.out.streams[1+i].size = dir->files[i].st.st_size;
+               info->generic.out.streams[1+i].alloc_size = dir->files[i].st.st_size;
+               info->generic.out.streams[1+i].stream_name.s = s?s:dir->files[i].name;
+       }
+
+       return NT_STATUS_OK;
+}
+
+/*
+  return info on a pathname
+*/
+static NTSTATUS svfs_qpathinfo(struct ntvfs_module_context *ntvfs,
+                              struct ntvfs_request *req, union smb_fileinfo *info)
+{
+       char *unix_path;
+       struct stat st;
+
+       DEBUG(19,("svfs_qpathinfo: file %s level 0x%x\n", info->generic.in.file.path, info->generic.level));
+       if (info->generic.level != RAW_FILEINFO_GENERIC) {
+               return ntvfs_map_qpathinfo(ntvfs, req, info);
+       }
+       
+       unix_path = cvfs_unix_path(ntvfs, req, info->generic.in.file.path);
+       DEBUG(19,("svfs_qpathinfo: file %s\n", unix_path));
+       if (stat(unix_path, &st) == -1) {
+               DEBUG(19,("svfs_qpathinfo: file %s errno=%d\n", unix_path, errno));
+               return map_nt_error_from_unix(errno);
+       }
+       DEBUG(19,("svfs_qpathinfo: file %s, stat done\n", unix_path));
+       return svfs_map_fileinfo(ntvfs, req, info, &st, unix_path);
+}
+
+/*
+  query info on a open file
+*/
+static NTSTATUS svfs_qfileinfo(struct ntvfs_module_context *ntvfs,
+                              struct ntvfs_request *req, union smb_fileinfo *info)
+{
+       struct svfs_private *private = ntvfs->private_data;
+       struct svfs_file *f;
+       struct stat st;
+
+       if (info->generic.level != RAW_FILEINFO_GENERIC) {
+               return ntvfs_map_qfileinfo(ntvfs, req, info);
+       }
+
+       f = find_fd(private, info->generic.in.file.fnum);
+       if (!f) {
+               return NT_STATUS_INVALID_HANDLE;
+       }
+       
+       if (fstat(info->generic.in.file.fnum, &st) == -1) {
+               return map_nt_error_from_unix(errno);
+       }
+
+       return svfs_map_fileinfo(ntvfs, req,info, &st, f->name);
+}
+
+
+/*
+  open a file
+*/
+static NTSTATUS svfs_open(struct ntvfs_module_context *ntvfs,
+                         struct ntvfs_request *req, union smb_open *io)
+{
+       struct svfs_private *private = ntvfs->private_data;
+       char *unix_path;
+       struct stat st;
+       int fd, flags;
+       struct svfs_file *f;
+       int create_flags, rdwr_flags;
+       BOOL readonly;
+       
+       if (io->generic.level != RAW_OPEN_GENERIC) {
+               return ntvfs_map_open(ntvfs, req, io);
+       }
+
+       readonly = lp_readonly(ntvfs->ctx->config.snum);
+       if (readonly) {
+               create_flags = 0;
+               rdwr_flags = O_RDONLY;
+       } else {
+               create_flags = O_CREAT;
+               rdwr_flags = O_RDWR;
+       }
+
+       unix_path = cvfs_unix_path(ntvfs, req, io->ntcreatex.in.fname);
+
+       switch (io->generic.in.open_disposition) {
+       case NTCREATEX_DISP_SUPERSEDE:
+       case NTCREATEX_DISP_OVERWRITE_IF:
+               flags = create_flags | O_TRUNC;
+               break;
+       case NTCREATEX_DISP_OPEN:
+       case NTCREATEX_DISP_OVERWRITE:
+               flags = 0;
+               break;
+       case NTCREATEX_DISP_CREATE:
+               flags = create_flags | O_EXCL;
+               break;
+       case NTCREATEX_DISP_OPEN_IF:
+               flags = create_flags;
+               break;
+       default:
+               flags = 0;
+               break;
+       }
+       
+       flags |= rdwr_flags;
+
+       if (io->generic.in.create_options & NTCREATEX_OPTIONS_DIRECTORY) {
+               flags = O_RDONLY | O_DIRECTORY;
+               if (readonly) {
+                       goto do_open;
+               }
+               switch (io->generic.in.open_disposition) {
+               case NTCREATEX_DISP_CREATE:
+                       if (mkdir(unix_path, 0755) == -1) {
+                               DEBUG(9,("svfs_open: mkdir %s errno=%d\n", unix_path, errno));
+                               return map_nt_error_from_unix(errno);
+                       }
+                       break;
+               case NTCREATEX_DISP_OPEN_IF:
+                       if (mkdir(unix_path, 0755) == -1 && errno != EEXIST) {
+                               DEBUG(9,("svfs_open: mkdir %s errno=%d\n", unix_path, errno));
+                               return map_nt_error_from_unix(errno);
+                       }
+                       break;
+               }
+       }
+
+do_open:
+       fd = open(unix_path, flags, 0644);
+       if (fd == -1) {
+               return map_nt_error_from_unix(errno);
+       }
+
+       if (fstat(fd, &st) == -1) {
+               DEBUG(9,("svfs_open: fstat errno=%d\n", errno));
+               close(fd);
+               return map_nt_error_from_unix(errno);
+       }
+
+       f = talloc(ntvfs, struct svfs_file);
+       NT_STATUS_HAVE_NO_MEMORY(f);
+       f->fd = fd;
+       f->name = talloc_strdup(f, unix_path);
+       NT_STATUS_HAVE_NO_MEMORY(f->name);
+
+       DLIST_ADD(private->open_files, f);
+
+       ZERO_STRUCT(io->generic.out);
+       
+       unix_to_nt_time(&io->generic.out.create_time, st.st_ctime);
+       unix_to_nt_time(&io->generic.out.access_time, st.st_atime);
+       unix_to_nt_time(&io->generic.out.write_time,  st.st_mtime);
+       unix_to_nt_time(&io->generic.out.change_time, st.st_mtime);
+       io->generic.out.file.fnum = fd;
+       io->generic.out.alloc_size = st.st_size;
+       io->generic.out.size = st.st_size;
+       io->generic.out.attrib = cvfs_unix_to_dos_attrib(st.st_mode);
+       io->generic.out.is_directory = S_ISDIR(st.st_mode) ? 1 : 0;
+
+       return NT_STATUS_OK;
+}
+
+/*
+  create a directory
+*/
+static NTSTATUS svfs_mkdir(struct ntvfs_module_context *ntvfs,
+                          struct ntvfs_request *req, union smb_mkdir *md)
+{
+       char *unix_path;
+
+       CHECK_READ_ONLY(req);
+
+       if (md->generic.level != RAW_MKDIR_MKDIR) {
+               return NT_STATUS_INVALID_LEVEL;
+       }
+
+       unix_path = cvfs_unix_path(ntvfs, req, md->mkdir.in.path);
+
+       if (mkdir(unix_path, 0777) == -1) {
+               return map_nt_error_from_unix(errno);
+       }
+
+       return NT_STATUS_OK;
+}
+
+/*
+  remove a directory
+*/
+static NTSTATUS svfs_rmdir(struct ntvfs_module_context *ntvfs,
+                          struct ntvfs_request *req, struct smb_rmdir *rd)
+{
+       char *unix_path;
+
+       CHECK_READ_ONLY(req);
+
+       unix_path = cvfs_unix_path(ntvfs, req, rd->in.path);
+
+       if (rmdir(unix_path) == -1) {
+               return map_nt_error_from_unix(errno);
+       }
+
+       return NT_STATUS_OK;
+}
+
+/*
+  rename a set of files
+*/
+static NTSTATUS svfs_rename(struct ntvfs_module_context *ntvfs,
+                           struct ntvfs_request *req, union smb_rename *ren)
+{
+       char *unix_path1, *unix_path2;
+
+       CHECK_READ_ONLY(req);
+
+       if (ren->generic.level != RAW_RENAME_RENAME) {
+               return NT_STATUS_INVALID_LEVEL;
+       }
+
+       unix_path1 = cvfs_unix_path(ntvfs, req, ren->rename.in.pattern1);
+       unix_path2 = cvfs_unix_path(ntvfs, req, ren->rename.in.pattern2);
+
+       if (rename(unix_path1, unix_path2) == -1) {
+               return map_nt_error_from_unix(errno);
+       }
+       
+       return NT_STATUS_OK;
+}
+
+/*
+  copy a set of files
+*/
+static NTSTATUS svfs_copy(struct ntvfs_module_context *ntvfs,
+                         struct ntvfs_request *req, struct smb_copy *cp)
+{
+       return NT_STATUS_NOT_SUPPORTED;
+}
+
+/*
+  read from a file
+*/
+static NTSTATUS svfs_read(struct ntvfs_module_context *ntvfs,
+                         struct ntvfs_request *req, union smb_read *rd)
+{
+       ssize_t ret;
+
+       if (rd->generic.level != RAW_READ_READX) {
+               return NT_STATUS_NOT_SUPPORTED;
+       }
+
+       ret = pread(rd->readx.in.file.fnum, 
+                   rd->readx.out.data, 
+                   rd->readx.in.maxcnt,
+                   rd->readx.in.offset);
+       if (ret == -1) {
+               return map_nt_error_from_unix(errno);
+       }
+
+       rd->readx.out.nread = ret;
+       rd->readx.out.remaining = 0; /* should fill this in? */
+       rd->readx.out.compaction_mode = 0; 
+
+       return NT_STATUS_OK;
+}
+
+/*
+  write to a file
+*/
+static NTSTATUS svfs_write(struct ntvfs_module_context *ntvfs,
+                          struct ntvfs_request *req, union smb_write *wr)
+{
+       ssize_t ret;
+
+       if (wr->generic.level != RAW_WRITE_WRITEX) {
+               return ntvfs_map_write(ntvfs, req, wr);
+       }
+
+       CHECK_READ_ONLY(req);
+
+       ret = pwrite(wr->writex.in.file.fnum, 
+                    wr->writex.in.data, 
+                    wr->writex.in.count,
+                    wr->writex.in.offset);
+       if (ret == -1) {
+               return map_nt_error_from_unix(errno);
+       }
+               
+       wr->writex.out.nwritten = ret;
+       wr->writex.out.remaining = 0; /* should fill this in? */
+       
+       return NT_STATUS_OK;
+}
+
+/*
+  seek in a file
+*/
+static NTSTATUS svfs_seek(struct ntvfs_module_context *ntvfs,
+                         struct ntvfs_request *req,
+                         union smb_seek *io)
+{
+       return NT_STATUS_NOT_SUPPORTED;
+}
+
+/*
+  flush a file
+*/
+static NTSTATUS svfs_flush(struct ntvfs_module_context *ntvfs,
+                          struct ntvfs_request *req,
+                          union smb_flush *io)
+{
+       fsync(io->flush.in.file.fnum);
+       return NT_STATUS_OK;
+}
+
+/*
+  close a file
+*/
+static NTSTATUS svfs_close(struct ntvfs_module_context *ntvfs,
+                          struct ntvfs_request *req,
+                          union smb_close *io)
+{
+       struct svfs_private *private = ntvfs->private_data;
+       struct svfs_file *f;
+
+       if (io->generic.level != RAW_CLOSE_CLOSE) {
+               /* we need a mapping function */
+               return NT_STATUS_INVALID_LEVEL;
+       }
+
+       f = find_fd(private, io->close.in.file.fnum);
+       if (!f) {
+               return NT_STATUS_INVALID_HANDLE;
+       }
+
+       if (close(io->close.in.file.fnum) == -1) {
+               return map_nt_error_from_unix(errno);
+       }
+
+       DLIST_REMOVE(private->open_files, f);
+       talloc_free(f->name);
+       talloc_free(f);
+
+       return NT_STATUS_OK;
+}
+
+/*
+  exit - closing files
+*/
+static NTSTATUS svfs_exit(struct ntvfs_module_context *ntvfs,
+                         struct ntvfs_request *req)
+{
+       return NT_STATUS_NOT_SUPPORTED;
+}
+
+/*
+  logoff - closing files
+*/
+static NTSTATUS svfs_logoff(struct ntvfs_module_context *ntvfs,
+                           struct ntvfs_request *req)
+{
+       return NT_STATUS_NOT_SUPPORTED;
+}
+
+/*
+  setup for an async call
+*/
+static NTSTATUS svfs_async_setup(struct ntvfs_module_context *ntvfs,
+                                struct ntvfs_request *req, 
+                                void *private)
+{
+       return NT_STATUS_OK;
+}
+
+/*
+  cancel an async call
+*/
+static NTSTATUS svfs_cancel(struct ntvfs_module_context *ntvfs, struct ntvfs_request *req)
+{
+       return NT_STATUS_UNSUCCESSFUL;
+}
+
+/*
+  lock a byte range
+*/
+static NTSTATUS svfs_lock(struct ntvfs_module_context *ntvfs,
+                         struct ntvfs_request *req, union smb_lock *lck)
+{
+       DEBUG(0,("REWRITE: not doing byte range locking!\n"));
+       return NT_STATUS_OK;
+}
+
+/*
+  set info on a pathname
+*/
+static NTSTATUS svfs_setpathinfo(struct ntvfs_module_context *ntvfs,
+                                struct ntvfs_request *req, union smb_setfileinfo *st)
+{
+       CHECK_READ_ONLY(req);
+
+       return NT_STATUS_NOT_SUPPORTED;
+}
+
+/*
+  set info on a open file
+*/
+static NTSTATUS svfs_setfileinfo(struct ntvfs_module_context *ntvfs,
+                                struct ntvfs_request *req, 
+                                union smb_setfileinfo *info)
+{
+       struct utimbuf unix_times;
+       int fd;
+
+       CHECK_READ_ONLY(req);
+               
+       switch (info->generic.level) {
+       case RAW_SFILEINFO_END_OF_FILE_INFO:
+       case RAW_SFILEINFO_END_OF_FILE_INFORMATION:
+               if (ftruncate(info->end_of_file_info.in.file.fnum, 
+                             info->end_of_file_info.in.size) == -1) {
+                       return map_nt_error_from_unix(errno);
+               }
+               break;
+       case RAW_SFILEINFO_SETATTRE:
+               unix_times.actime = info->setattre.in.access_time;
+               unix_times.modtime = info->setattre.in.write_time;
+               fd = info->setattre.in.file.fnum;
+       
+               if (unix_times.actime == 0 && unix_times.modtime == 0) {
+                       break;
+               } 
+
+               /* set modify time = to access time if modify time was 0 */
+               if (unix_times.actime != 0 && unix_times.modtime == 0) {
+                       unix_times.modtime = unix_times.actime;
+               }
+
+               /* Set the date on this file */
+               if (cvfs_file_utime(fd, &unix_times) != 0) {
+                       return NT_STATUS_ACCESS_DENIED;
+               }
+               break;
+       default:
+               DEBUG(2,("svfs_setfileinfo: level %d not implemented\n", 
+                        info->generic.level));
+               return NT_STATUS_NOT_IMPLEMENTED;
+       }
+       return NT_STATUS_OK;
+}
+
+
+/*
+  return filesystem space info
+*/
+static NTSTATUS svfs_fsinfo(struct ntvfs_module_context *ntvfs,
+                           struct ntvfs_request *req, union smb_fsinfo *fs)
+{
+       struct svfs_private *private = ntvfs->private_data;
+       struct stat st;
+
+       if (fs->generic.level != RAW_QFS_GENERIC) {
+               return ntvfs_map_fsinfo(ntvfs, req, fs);
+       }
+
+       if (sys_fsusage(private->connectpath, 
+                       &fs->generic.out.blocks_free, 
+                       &fs->generic.out.blocks_total) == -1) {
+               return map_nt_error_from_unix(errno);
+       }
+
+       fs->generic.out.block_size = 512;
+
+       if (stat(private->connectpath, &st) != 0) {
+               return NT_STATUS_DISK_CORRUPT_ERROR;
+       }
+       
+       fs->generic.out.fs_id = st.st_ino;
+       unix_to_nt_time(&fs->generic.out.create_time, st.st_ctime);
+       fs->generic.out.serial_number = st.st_ino;
+       fs->generic.out.fs_attr = 0;
+       fs->generic.out.max_file_component_length = 255;
+       fs->generic.out.device_type = 0;
+       fs->generic.out.device_characteristics = 0;
+       fs->generic.out.quota_soft = 0;
+       fs->generic.out.quota_hard = 0;
+       fs->generic.out.quota_flags = 0;
+       fs->generic.out.volume_name = talloc_strdup(req, lp_servicename(ntvfs->ctx->config.snum));
+       fs->generic.out.fs_type = ntvfs->ctx->fs_type;
+
+       return NT_STATUS_OK;
+}
+
+#if 0
+/*
+  return filesystem attribute info
+*/
+static NTSTATUS svfs_fsattr(struct ntvfs_module_context *ntvfs,
+                           struct ntvfs_request *req, union smb_fsattr *fs)
+{
+       struct stat st;
+       struct svfs_private *private = ntvfs->private_data;
+
+       if (fs->generic.level != RAW_FSATTR_GENERIC) {
+               return ntvfs_map_fsattr(ntvfs, req, fs);
+       }
+
+       if (stat(private->connectpath, &st) == -1) {
+               return map_nt_error_from_unix(errno);
+       }
+
+       unix_to_nt_time(&fs->generic.out.create_time, st.st_ctime);
+       fs->generic.out.fs_attr = 
+               FILE_CASE_PRESERVED_NAMES | 
+               FILE_CASE_SENSITIVE_SEARCH | 
+               FILE_PERSISTENT_ACLS;
+       fs->generic.out.max_file_component_length = 255;
+       fs->generic.out.serial_number = 1;
+       fs->generic.out.fs_type = talloc_strdup(req, "NTFS");
+       fs->generic.out.volume_name = talloc_strdup(req, 
+                                                   lp_servicename(req->tcon->service));
+
+       return NT_STATUS_OK;
+}
+#endif
+
+/*
+  return print queue info
+*/
+static NTSTATUS svfs_lpq(struct ntvfs_module_context *ntvfs,
+                        struct ntvfs_request *req, union smb_lpq *lpq)
+{
+       return NT_STATUS_NOT_SUPPORTED;
+}
+
+/* 
+   list files in a directory matching a wildcard pattern
+*/
+static NTSTATUS svfs_search_first(struct ntvfs_module_context *ntvfs,
+                                 struct ntvfs_request *req, union smb_search_first *io, 
+                                 void *search_private, 
+                                 BOOL (*callback)(void *, union smb_search_data *))
+{
+       struct svfs_dir *dir;
+       int i;
+       struct svfs_private *private = ntvfs->private_data;
+       struct search_state *search;
+       union smb_search_data file;
+       uint_t max_count;
+
+       if (io->generic.level != RAW_SEARCH_BOTH_DIRECTORY_INFO) {
+               return NT_STATUS_NOT_SUPPORTED;
+       }
+
+       search = talloc_zero(private, struct search_state);
+       if (!search) {
+               return NT_STATUS_NO_MEMORY;
+       }
+
+       max_count = io->t2ffirst.in.max_count;
+
+       dir = cvfs_list(ntvfs, req, io->t2ffirst.in.pattern);
+       if (!dir) {
+               return NT_STATUS_FOOBAR;
+       }
+
+       search->handle = private->next_search_handle;
+       search->dir = dir;
+
+       if (dir->count < max_count) {
+               max_count = dir->count;
+       }
+
+       for (i=0; i < max_count;i++) {
+               ZERO_STRUCT(file);
+               unix_to_nt_time(&file.both_directory_info.create_time, dir->files[i].st.st_ctime);
+               unix_to_nt_time(&file.both_directory_info.access_time, dir->files[i].st.st_atime);
+               unix_to_nt_time(&file.both_directory_info.write_time,  dir->files[i].st.st_mtime);
+               unix_to_nt_time(&file.both_directory_info.change_time, dir->files[i].st.st_mtime);
+               file.both_directory_info.name.s = dir->files[i].name;
+               file.both_directory_info.short_name.s = dir->files[i].name;
+               file.both_directory_info.size = dir->files[i].st.st_size;
+               file.both_directory_info.attrib = svfs_unix_to_dos_attrib(dir->files[i].st.st_mode);
+
+               if (!callback(search_private, &file)) {
+                       break;
+               }
+       }
+
+       search->current_index = i;
+
+       io->t2ffirst.out.count = i;
+       io->t2ffirst.out.handle = search->handle;
+       io->t2ffirst.out.end_of_search = (i == dir->count) ? 1 : 0;
+
+       /* work out if we are going to keep the search state */
+       if ((io->t2ffirst.in.flags & FLAG_TRANS2_FIND_CLOSE) ||
+           ((io->t2ffirst.in.flags & FLAG_TRANS2_FIND_CLOSE_IF_END) && (i == dir->count))) {
+               talloc_free(search);
+       } else {
+               private->next_search_handle++;
+               DLIST_ADD(private->search, search);
+       }
+
+       return NT_STATUS_OK;
+}
+
+/* continue a search */
+static NTSTATUS svfs_search_next(struct ntvfs_module_context *ntvfs,
+                                struct ntvfs_request *req, union smb_search_next *io, 
+                                void *search_private, 
+                                BOOL (*callback)(void *, union smb_search_data *))
+{
+       struct svfs_dir *dir;
+       int i;
+       struct svfs_private *private = ntvfs->private_data;
+       struct search_state *search;
+       union smb_search_data file;
+       uint_t max_count;
+
+       if (io->generic.level != RAW_SEARCH_BOTH_DIRECTORY_INFO) {
+               return NT_STATUS_NOT_SUPPORTED;
+       }
+
+       for (search=private->search; search; search = search->next) {
+               if (search->handle == io->t2fnext.in.handle) break;
+       }
+       
+       if (!search) {
+               /* we didn't find the search handle */
+               return NT_STATUS_FOOBAR;
+       }
+
+       dir = search->dir;
+
+       /* the client might be asking for something other than just continuing
+          with the search */
+       if (!(io->t2fnext.in.flags & FLAG_TRANS2_FIND_CONTINUE) &&
+           (io->t2fnext.in.flags & FLAG_TRANS2_FIND_REQUIRE_RESUME) &&
+           io->t2fnext.in.last_name && *io->t2fnext.in.last_name) {
+               /* look backwards first */
+               for (i=search->current_index; i > 0; i--) {
+                       if (strcmp(io->t2fnext.in.last_name, dir->files[i-1].name) == 0) {
+                               search->current_index = i;
+                               goto found;
+                       }
+               }
+
+               /* then look forwards */
+               for (i=search->current_index+1; i <= dir->count; i++) {
+                       if (strcmp(io->t2fnext.in.last_name, dir->files[i-1].name) == 0) {
+                               search->current_index = i;
+                               goto found;
+                       }
+               }
+       }
+
+found: 
+       max_count = search->current_index + io->t2fnext.in.max_count;
+
+       if (max_count > dir->count) {
+               max_count = dir->count;
+       }
+
+       for (i = search->current_index; i < max_count;i++) {
+               ZERO_STRUCT(file);
+               unix_to_nt_time(&file.both_directory_info.create_time, dir->files[i].st.st_ctime);
+               unix_to_nt_time(&file.both_directory_info.access_time, dir->files[i].st.st_atime);
+               unix_to_nt_time(&file.both_directory_info.write_time,  dir->files[i].st.st_mtime);
+               unix_to_nt_time(&file.both_directory_info.change_time, dir->files[i].st.st_mtime);
+               file.both_directory_info.name.s = dir->files[i].name;
+               file.both_directory_info.short_name.s = dir->files[i].name;
+               file.both_directory_info.size = dir->files[i].st.st_size;
+               file.both_directory_info.attrib = svfs_unix_to_dos_attrib(dir->files[i].st.st_mode);
+
+               if (!callback(search_private, &file)) {
+                       break;
+               }
+       }
+
+       io->t2fnext.out.count = i - search->current_index;
+       io->t2fnext.out.end_of_search = (i == dir->count) ? 1 : 0;
+
+       search->current_index = i;
+
+       /* work out if we are going to keep the search state */
+       if ((io->t2fnext.in.flags & FLAG_TRANS2_FIND_CLOSE) ||
+           ((io->t2fnext.in.flags & FLAG_TRANS2_FIND_CLOSE_IF_END) && (i == dir->count))) {
+               DLIST_REMOVE(private->search, search);
+               talloc_free(search);
+       }
+
+       return NT_STATUS_OK;
+}
+
+/* close a search */
+static NTSTATUS svfs_search_close(struct ntvfs_module_context *ntvfs,
+                                 struct ntvfs_request *req, union smb_search_close *io)
+{
+       struct svfs_private *private = ntvfs->private_data;
+       struct search_state *search;
+
+       for (search=private->search; search; search = search->next) {
+               if (search->handle == io->findclose.in.handle) break;
+       }
+       
+       if (!search) {
+               /* we didn't find the search handle */
+               return NT_STATUS_FOOBAR;
+       }
+
+       DLIST_REMOVE(private->search, search);
+       talloc_free(search);
+
+       return NT_STATUS_OK;
+}
+
+/* SMBtrans - not used on file shares */
+static NTSTATUS svfs_trans(struct ntvfs_module_context *ntvfs,
+                          struct ntvfs_request *req, struct smb_trans2 *trans2)
+{
+       return NT_STATUS_ACCESS_DENIED;
+}
+
+
+/*
+  initialialise the POSIX disk backend, registering ourselves with the ntvfs subsystem
+ */
+NTSTATUS ntvfs_cifs_posix_init(void)
+{
+       NTSTATUS ret;
+       struct ntvfs_ops ops;
+       NTVFS_CURRENT_CRITICAL_SIZES(vers);
+
+       ZERO_STRUCT(ops);
+
+       /* fill in all the operations */
+       ops.connect = svfs_connect;
+       ops.disconnect = svfs_disconnect;
+       ops.unlink = svfs_unlink;
+       ops.chkpath = svfs_chkpath;
+       ops.qpathinfo = svfs_qpathinfo;
+       ops.setpathinfo = svfs_setpathinfo;
+       ops.open = svfs_open;
+       ops.mkdir = svfs_mkdir;
+       ops.rmdir = svfs_rmdir;
+       ops.rename = svfs_rename;
+       ops.copy = svfs_copy;
+       ops.ioctl = svfs_ioctl;
+       ops.read = svfs_read;
+       ops.write = svfs_write;
+       ops.seek = svfs_seek;
+       ops.flush = svfs_flush; 
+       ops.close = svfs_close;
+       ops.exit = svfs_exit;
+       ops.lock = svfs_lock;
+       ops.setfileinfo = svfs_setfileinfo;
+       ops.qfileinfo = svfs_qfileinfo;
+       ops.fsinfo = svfs_fsinfo;
+       ops.lpq = svfs_lpq;
+       ops.search_first = svfs_search_first;
+       ops.search_next = svfs_search_next;
+       ops.search_close = svfs_search_close;
+       ops.trans = svfs_trans;
+       ops.logoff = svfs_logoff;
+       ops.async_setup = svfs_async_setup;
+       ops.cancel = svfs_cancel;
+
+       /* register ourselves with the NTVFS subsystem. We register
+          under names 'simple'
+       */
+
+       ops.type = NTVFS_DISK;
+       ops.name = "cifs-posix-cli";
+       ret = ntvfs_register(&ops, &vers);
+
+       if (!NT_STATUS_IS_OK(ret)) {
+               DEBUG(0,("Failed to register simple backend with name: %s!\n",
+                        ops.name));
+       }
+
+       return ret;
+}
index e17f649abb7700601315c42aceb064a1422cc267..ac96164347e77667825f062e631d3aae84e9ff03 100644 (file)
@@ -25,7 +25,19 @@ PRIVATE_PROTO_HEADER = simple/proto.h
 OBJ_FILES = \
                simple/vfs_simple.o \
                simple/svfs_util.o
-# End MODULE ntvfs_cifs
+# End MODULE ntvfs_simple
+################################################
+
+################################################
+# Start MODULE ntvfs_cifs_posix_cli
+[MODULE::ntvfs_cifs_posix]
+INIT_FUNCTION = ntvfs_cifs_posix_init
+SUBSYSTEM = ntvfs
+PRIVATE_PROTO_HEADER = cifs_posix_cli/proto.h
+OBJ_FILES = \
+                cifs_posix_cli/vfs_simple.o \
+                cifs_posix_cli/svfs_util.o
+# End MODULE ntvfs_cifs_posix_cli
 ################################################
 
 ################################################