1
0
mirror of https://github.com/samba-team/samba.git synced 2025-02-08 05:57:51 +03:00
Andrew Tridgell a3cec511bb r2561: completely redid the ntvfs module chaining code, You can now do something like:
ntvfs handler = nbench posix

and the nbench pass-thru module will be called before the posix
module. The chaining logic is now much saner, and less racy, with each
level in the chain getting its own private pointer rather than relying
on save/restore logic in the pass-thru module.

The only pass-thru module we have at the moment is the nbench one
(which records all traffic in a nbench compatibe format), but I plan
on soon writing a "unixuid" pass-thru module that will implement the
setegid()/setgroups()/seteuid() logic for standard posix uid
handling. This separation of the posix backend from the uid handling
should simplify the code, and make development easier.

I also modified the nbench module so it can do multiple chaining, so
if you want to you can do:

   ntvfs module = nbench nbench posix

and it will save 2 copies of the log file in /tmp. This is really only
useful for testing at the moment until we have more than one pass-thru
module.
(This used to be commit f84c0af35cb54c8fdc4933afefc18fa4c062aae4)
2007-10-10 12:59:06 -05:00

717 lines
17 KiB
C

/*
Unix SMB/CIFS implementation.
CIFS-on-CIFS NTVFS filesystem backend
Copyright (C) Andrew Tridgell 2003
Copyright (C) James J Myers 2003 <myersjj@samba.org>
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 CIFS->CIFS NTVFS filesystem backend.
*/
#include "includes.h"
/* this is stored in ntvfs_private */
struct cvfs_private {
struct smbcli_tree *tree;
struct smbcli_transport *transport;
struct smbsrv_tcon *tcon;
const struct ntvfs_ops *ops;
};
/* a structure used to pass information to an async handler */
struct async_info {
struct smbsrv_request *req;
void *parms;
};
/*
an idle function to cope with messages from the smbd client while
waiting for a reply from the server
this function won't be needed once all of the cifs backend
and the core of smbd is converted to use async calls
*/
static void idle_func(struct smbcli_transport *transport, void *p_private)
{
struct cvfs_private *private = p_private;
int fd = socket_get_fd(private->tcon->smb_conn->connection->socket);
if (socket_pending(fd)) {
smbd_process_async(private->tcon->smb_conn);
}
}
/*
a handler for oplock break events from the server - these need to be passed
along to the client
*/
static BOOL oplock_handler(struct smbcli_transport *transport, uint16_t tid, uint16_t fnum, uint8_t level, void *p_private)
{
struct cvfs_private *private = p_private;
DEBUG(5,("vfs_cifs: sending oplock break level %d for fnum %d\n", level, fnum));
return req_send_oplock_break(private->tcon, fnum, level);
}
/*
a handler for read events on a connection to a backend server
*/
static void cifs_socket_handler(struct event_context *ev, struct fd_event *fde, time_t t, uint16_t flags)
{
struct cvfs_private *private = fde->private;
struct smbsrv_tcon *tcon = private->tcon;
DEBUG(5,("cifs_socket_handler event on fd %d\n", fde->fd));
if (!smbcli_transport_process(private->transport)) {
/* the connection to our server is dead */
close_cnum(tcon);
}
}
/*
connect to a share - used when a tree_connect operation comes in.
*/
static NTSTATUS cvfs_connect(struct smbsrv_request *req, const char *sharename, int depth)
{
struct smbsrv_tcon *tcon = req->tcon;
NTSTATUS status;
struct cvfs_private *private;
const char *host, *user, *pass, *domain, *remote_share;
/* Here we need to determine which server to connect to.
* For now we use parametric options, type cifs.
* Later we will use security=server and auth_server.c.
*/
host = lp_parm_string(req->tcon->service, "cifs", "server");
user = lp_parm_string(req->tcon->service, "cifs", "user");
pass = lp_parm_string(req->tcon->service, "cifs", "password");
domain = lp_parm_string(req->tcon->service, "cifs", "domain");
remote_share = lp_parm_string(req->tcon->service, "cifs", "share");
if (!remote_share) {
remote_share = sharename;
}
if (!host || !user || !pass || !domain) {
DEBUG(1,("CIFS backend: You must supply server, user, password and domain\n"));
return NT_STATUS_INVALID_PARAMETER;
}
private = talloc(req->tcon, sizeof(struct cvfs_private));
if (!private) {
return NT_STATUS_NO_MEMORY;
}
ZERO_STRUCTP(private);
ntvfs_set_private(req->tcon, depth, private);
status = smbcli_tree_full_connection(&private->tree,
"vfs_cifs",
host,
0,
remote_share, "?????",
user, domain,
pass);
if (!NT_STATUS_IS_OK(status)) {
return status;
}
private->transport = private->tree->session->transport;
private->tree->session->pid = SVAL(req->in.hdr, HDR_PID);
private->tcon = req->tcon;
private->ops = ntvfs_backend_byname("cifs", NTVFS_DISK);
tcon->fs_type = talloc_strdup(tcon, "NTFS");
tcon->dev_type = talloc_strdup(tcon, "A:");
/* we need to receive oplock break requests from the server */
smbcli_oplock_handler(private->transport, oplock_handler, private);
smbcli_transport_idle_handler(private->transport, idle_func, 1, private);
private->transport->event.fde->handler = cifs_socket_handler;
private->transport->event.fde->private = private;
private->transport->event.ctx = event_context_merge(tcon->smb_conn->connection->event.ctx,
private->transport->event.ctx);
return NT_STATUS_OK;
}
/*
disconnect from a share
*/
static NTSTATUS cvfs_disconnect(struct smbsrv_tcon *tcon, int depth)
{
struct cvfs_private *private = tcon->ntvfs_private_list[depth];
smb_tree_disconnect(private->tree);
smbcli_tree_close(private->tree);
return NT_STATUS_OK;
}
/*
a handler for simple async replies
this handler can only be used for functions that don't return any
parameters (those that just return a status code)
*/
static void async_simple(struct smbcli_request *c_req)
{
struct async_info *async = c_req->async.private;
struct smbsrv_request *req = async->req;
req->async.status = smbcli_request_simple_recv(c_req);
req->async.send_fn(req);
}
/* save some typing for the simple functions */
#define ASYNC_RECV_TAIL(io, async_fn) do { \
if (!c_req) return NT_STATUS_UNSUCCESSFUL; \
{ \
struct async_info *async; \
async = talloc_p(req, struct async_info); \
if (!async) return NT_STATUS_NO_MEMORY; \
async->parms = io; \
async->req = req; \
c_req->async.private = async; \
} \
c_req->async.fn = async_fn; \
req->control_flags |= REQ_CONTROL_ASYNC; \
return NT_STATUS_OK; \
} while (0)
#define SIMPLE_ASYNC_TAIL ASYNC_RECV_TAIL(NULL, async_simple)
/*
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 cvfs_unlink(struct smbsrv_request *req, struct smb_unlink *unl)
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
struct smbcli_request *c_req;
/* see if the front end will allow us to perform this
function asynchronously. */
if (!req->async.send_fn) {
return smb_raw_unlink(private->tree, unl);
}
c_req = smb_raw_unlink_send(private->tree, unl);
SIMPLE_ASYNC_TAIL;
}
/*
a handler for async ioctl replies
*/
static void async_ioctl(struct smbcli_request *c_req)
{
struct async_info *async = c_req->async.private;
struct smbsrv_request *req = async->req;
req->async.status = smb_raw_ioctl_recv(c_req, req, async->parms);
req->async.send_fn(req);
}
/*
ioctl interface
*/
static NTSTATUS cvfs_ioctl(struct smbsrv_request *req, union smb_ioctl *io)
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
struct smbcli_request *c_req;
/* see if the front end will allow us to perform this
function asynchronously. */
if (!req->async.send_fn) {
return smb_raw_ioctl(private->tree, req, io);
}
c_req = smb_raw_ioctl_send(private->tree, io);
ASYNC_RECV_TAIL(io, async_ioctl);
}
/*
check if a directory exists
*/
static NTSTATUS cvfs_chkpath(struct smbsrv_request *req, struct smb_chkpath *cp)
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
struct smbcli_request *c_req;
if (!req->async.send_fn) {
return smb_raw_chkpath(private->tree, cp);
}
c_req = smb_raw_chkpath_send(private->tree, cp);
SIMPLE_ASYNC_TAIL;
}
/*
a handler for async qpathinfo replies
*/
static void async_qpathinfo(struct smbcli_request *c_req)
{
struct async_info *async = c_req->async.private;
struct smbsrv_request *req = async->req;
req->async.status = smb_raw_pathinfo_recv(c_req, req, async->parms);
req->async.send_fn(req);
}
/*
return info on a pathname
*/
static NTSTATUS cvfs_qpathinfo(struct smbsrv_request *req, union smb_fileinfo *info)
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
struct smbcli_request *c_req;
if (!req->async.send_fn) {
return smb_raw_pathinfo(private->tree, req, info);
}
c_req = smb_raw_pathinfo_send(private->tree, info);
ASYNC_RECV_TAIL(info, async_qpathinfo);
}
/*
a handler for async qfileinfo replies
*/
static void async_qfileinfo(struct smbcli_request *c_req)
{
struct async_info *async = c_req->async.private;
struct smbsrv_request *req = async->req;
req->async.status = smb_raw_fileinfo_recv(c_req, req, async->parms);
req->async.send_fn(req);
}
/*
query info on a open file
*/
static NTSTATUS cvfs_qfileinfo(struct smbsrv_request *req, union smb_fileinfo *info)
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
struct smbcli_request *c_req;
if (!req->async.send_fn) {
return smb_raw_fileinfo(private->tree, req, info);
}
c_req = smb_raw_fileinfo_send(private->tree, info);
ASYNC_RECV_TAIL(info, async_qfileinfo);
}
/*
set info on a pathname
*/
static NTSTATUS cvfs_setpathinfo(struct smbsrv_request *req, union smb_setfileinfo *st)
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
struct smbcli_request *c_req;
if (!req->async.send_fn) {
return smb_raw_setpathinfo(private->tree, st);
}
c_req = smb_raw_setpathinfo_send(private->tree, st);
SIMPLE_ASYNC_TAIL;
}
/*
a handler for async open replies
*/
static void async_open(struct smbcli_request *c_req)
{
struct async_info *async = c_req->async.private;
struct smbsrv_request *req = async->req;
req->async.status = smb_raw_open_recv(c_req, req, async->parms);
req->async.send_fn(req);
}
/*
open a file
*/
static NTSTATUS cvfs_open(struct smbsrv_request *req, union smb_open *io)
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
struct smbcli_request *c_req;
if (!req->async.send_fn) {
return smb_raw_open(private->tree, req, io);
}
c_req = smb_raw_open_send(private->tree, io);
ASYNC_RECV_TAIL(io, async_open);
}
/*
create a directory
*/
static NTSTATUS cvfs_mkdir(struct smbsrv_request *req, union smb_mkdir *md)
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
struct smbcli_request *c_req;
if (!req->async.send_fn) {
return smb_raw_mkdir(private->tree, md);
}
c_req = smb_raw_mkdir_send(private->tree, md);
SIMPLE_ASYNC_TAIL;
}
/*
remove a directory
*/
static NTSTATUS cvfs_rmdir(struct smbsrv_request *req, struct smb_rmdir *rd)
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
struct smbcli_request *c_req;
if (!req->async.send_fn) {
return smb_raw_rmdir(private->tree, rd);
}
c_req = smb_raw_rmdir_send(private->tree, rd);
SIMPLE_ASYNC_TAIL;
}
/*
rename a set of files
*/
static NTSTATUS cvfs_rename(struct smbsrv_request *req, union smb_rename *ren)
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
struct smbcli_request *c_req;
if (!req->async.send_fn) {
return smb_raw_rename(private->tree, ren);
}
c_req = smb_raw_rename_send(private->tree, ren);
SIMPLE_ASYNC_TAIL;
}
/*
copy a set of files
*/
static NTSTATUS cvfs_copy(struct smbsrv_request *req, struct smb_copy *cp)
{
return NT_STATUS_NOT_SUPPORTED;
}
/*
a handler for async read replies
*/
static void async_read(struct smbcli_request *c_req)
{
struct async_info *async = c_req->async.private;
struct smbsrv_request *req = async->req;
req->async.status = smb_raw_read_recv(c_req, async->parms);
req->async.send_fn(req);
}
/*
read from a file
*/
static NTSTATUS cvfs_read(struct smbsrv_request *req, union smb_read *rd)
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
struct smbcli_request *c_req;
if (!req->async.send_fn) {
return smb_raw_read(private->tree, rd);
}
c_req = smb_raw_read_send(private->tree, rd);
ASYNC_RECV_TAIL(rd, async_read);
}
/*
a handler for async write replies
*/
static void async_write(struct smbcli_request *c_req)
{
struct async_info *async = c_req->async.private;
struct smbsrv_request *req = async->req;
req->async.status = smb_raw_write_recv(c_req, async->parms);
req->async.send_fn(req);
}
/*
write to a file
*/
static NTSTATUS cvfs_write(struct smbsrv_request *req, union smb_write *wr)
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
struct smbcli_request *c_req;
if (!req->async.send_fn) {
return smb_raw_write(private->tree, wr);
}
c_req = smb_raw_write_send(private->tree, wr);
ASYNC_RECV_TAIL(wr, async_write);
}
/*
seek in a file
*/
static NTSTATUS cvfs_seek(struct smbsrv_request *req, struct smb_seek *io)
{
return NT_STATUS_NOT_SUPPORTED;
}
/*
flush a file
*/
static NTSTATUS cvfs_flush(struct smbsrv_request *req, struct smb_flush *io)
{
return NT_STATUS_OK;
}
/*
close a file
*/
static NTSTATUS cvfs_close(struct smbsrv_request *req, union smb_close *io)
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
struct smbcli_request *c_req;
if (!req->async.send_fn) {
return smb_raw_close(private->tree, io);
}
c_req = smb_raw_close_send(private->tree, io);
SIMPLE_ASYNC_TAIL;
}
/*
exit - closing files?
*/
static NTSTATUS cvfs_exit(struct smbsrv_request *req)
{
return NT_STATUS_NOT_SUPPORTED;
}
/*
lock a byte range
*/
static NTSTATUS cvfs_lock(struct smbsrv_request *req, union smb_lock *lck)
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
struct smbcli_request *c_req;
if (!req->async.send_fn) {
return smb_raw_lock(private->tree, lck);
}
c_req = smb_raw_lock_send(private->tree, lck);
SIMPLE_ASYNC_TAIL;
}
/*
set info on a open file
*/
static NTSTATUS cvfs_setfileinfo(struct smbsrv_request *req,
union smb_setfileinfo *info)
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
struct smbcli_request *c_req;
if (!req->async.send_fn) {
return smb_raw_setfileinfo(private->tree, info);
}
c_req = smb_raw_setfileinfo_send(private->tree, info);
SIMPLE_ASYNC_TAIL;
}
/*
a handler for async fsinfo replies
*/
static void async_fsinfo(struct smbcli_request *c_req)
{
struct async_info *async = c_req->async.private;
struct smbsrv_request *req = async->req;
req->async.status = smb_raw_fsinfo_recv(c_req, req, async->parms);
req->async.send_fn(req);
}
/*
return filesystem space info
*/
static NTSTATUS cvfs_fsinfo(struct smbsrv_request *req, union smb_fsinfo *fs)
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
struct smbcli_request *c_req;
if (!req->async.send_fn) {
return smb_raw_fsinfo(private->tree, req, fs);
}
c_req = smb_raw_fsinfo_send(private->tree, req, fs);
ASYNC_RECV_TAIL(fs, async_fsinfo);
}
/*
return print queue info
*/
static NTSTATUS cvfs_lpq(struct smbsrv_request *req, union smb_lpq *lpq)
{
return NT_STATUS_NOT_SUPPORTED;
}
/*
list files in a directory matching a wildcard pattern
*/
static NTSTATUS cvfs_search_first(struct smbsrv_request *req, union smb_search_first *io,
void *search_private,
BOOL (*callback)(void *, union smb_search_data *))
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
return smb_raw_search_first(private->tree, req, io, search_private, callback);
}
/* continue a search */
static NTSTATUS cvfs_search_next(struct smbsrv_request *req, union smb_search_next *io,
void *search_private,
BOOL (*callback)(void *, union smb_search_data *))
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
return smb_raw_search_next(private->tree, req, io, search_private, callback);
}
/* close a search */
static NTSTATUS cvfs_search_close(struct smbsrv_request *req, union smb_search_close *io)
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
return smb_raw_search_close(private->tree, io);
}
/*
a handler for async trans2 replies
*/
static void async_trans2(struct smbcli_request *c_req)
{
struct async_info *async = c_req->async.private;
struct smbsrv_request *req = async->req;
req->async.status = smb_raw_trans2_recv(c_req, req, async->parms);
req->async.send_fn(req);
}
/* raw trans2 */
static NTSTATUS cvfs_trans2(struct smbsrv_request *req, struct smb_trans2 *trans2)
{
NTVFS_GET_PRIVATE(cvfs_private, private, req);
struct smbcli_request *c_req;
if (!req->async.send_fn) {
return smb_raw_trans2(private->tree, req, trans2);
}
c_req = smb_raw_trans2_send(private->tree, trans2);
ASYNC_RECV_TAIL(trans2, async_trans2);
}
/* SMBtrans - not used on file shares */
static NTSTATUS cvfs_trans(struct smbsrv_request *req, struct smb_trans2 *trans2)
{
return NT_STATUS_ACCESS_DENIED;
}
/*
initialise the CIFS->CIFS backend, registering ourselves with the ntvfs subsystem
*/
NTSTATUS ntvfs_cifs_init(void)
{
NTSTATUS ret;
struct ntvfs_ops ops;
ZERO_STRUCT(ops);
/* fill in the name and type */
ops.name = "cifs";
ops.type = NTVFS_DISK;
/* fill in all the operations */
ops.connect = cvfs_connect;
ops.disconnect = cvfs_disconnect;
ops.unlink = cvfs_unlink;
ops.chkpath = cvfs_chkpath;
ops.qpathinfo = cvfs_qpathinfo;
ops.setpathinfo = cvfs_setpathinfo;
ops.open = cvfs_open;
ops.mkdir = cvfs_mkdir;
ops.rmdir = cvfs_rmdir;
ops.rename = cvfs_rename;
ops.copy = cvfs_copy;
ops.ioctl = cvfs_ioctl;
ops.read = cvfs_read;
ops.write = cvfs_write;
ops.seek = cvfs_seek;
ops.flush = cvfs_flush;
ops.close = cvfs_close;
ops.exit = cvfs_exit;
ops.lock = cvfs_lock;
ops.setfileinfo = cvfs_setfileinfo;
ops.qfileinfo = cvfs_qfileinfo;
ops.fsinfo = cvfs_fsinfo;
ops.lpq = cvfs_lpq;
ops.search_first = cvfs_search_first;
ops.search_next = cvfs_search_next;
ops.search_close = cvfs_search_close;
ops.trans = cvfs_trans;
if (lp_parm_bool(-1, "cifs", "maptrans2", False)) {
ops.trans2 = cvfs_trans2;
}
/* register ourselves with the NTVFS subsystem. We register
under the name 'cifs'. */
ret = register_backend("ntvfs", &ops);
if (!NT_STATUS_IS_OK(ret)) {
DEBUG(0,("Failed to register CIFS backend!\n"));
}
return ret;
}