1
0
mirror of https://github.com/samba-team/samba.git synced 2024-12-23 17:34:34 +03:00
samba-mirror/source3/smbd/server.c
Jones Syue 4f3dfb2029 s3: SIGHUP handlers use consistent log level 3
When turn-on 'log level = 3', sending SIGHUP to samba processes, for
example: smbd parent/children, smbd-notifyd, and smbd-cleanupd. Then
monitor log.smbd in order to parse sighup logs, it looks like the log level
is inconsistent among these processes: smbd parent/children use level 1,
and smbd-notifyd/smbd-cleanupd use level 3.

This patch raises sighup handler's log level from level 1 to level 3, which
is more consistent with smbd-notifyd by Commit 6e5bff80a0 ("s3:notifyd:
Handle sigup in notifyd to reparse smb.conf"), and smbd-cleanupd by Commit
57c1e115ec ("smbd: reopen logs on SIGHUP for notifyd and cleanupd").

BUG: https://bugzilla.samba.org/show_bug.cgi?id=15706

Signed-off-by: Jones Syue <jonessyue@qnap.com>
Reviewed-by: Jeremy Allison <jra@samba.org>
Reviewed-by: Martin Schwenke <martin@meltin.net>

Autobuild-User(master): Martin Schwenke <martins@samba.org>
Autobuild-Date(master): Wed Sep 25 01:38:02 UTC 2024 on atb-devel-224
2024-09-25 01:38:02 +00:00

2366 lines
57 KiB
C

/*
Unix SMB/CIFS implementation.
Main SMB server routines
Copyright (C) Andrew Tridgell 1992-1998
Copyright (C) Martin Pool 2002
Copyright (C) Jelmer Vernooij 2002-2003
Copyright (C) Volker Lendecke 1993-2007
Copyright (C) Jeremy Allison 1993-2007
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 3 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, see <http://www.gnu.org/licenses/>.
*/
#include "includes.h"
#include "system/filesys.h"
#include "lib/util/server_id.h"
#include "lib/util/close_low_fd.h"
#include "lib/cmdline/cmdline.h"
#include "locking/share_mode_lock.h"
#include "smbd/smbd.h"
#include "smbd/globals.h"
#include "source3/smbd/smbXsrv_session.h"
#include "smbd/smbXsrv_open.h"
#include "registry/reg_init_full.h"
#include "libcli/auth/schannel.h"
#include "secrets.h"
#include "../lib/util/memcache.h"
#include "ctdbd_conn.h"
#include "lib/util/util_process.h"
#include "util_cluster.h"
#include "printing/queue_process.h"
#include "rpc_server/rpc_config.h"
#include "passdb.h"
#include "auth.h"
#include "messages.h"
#include "messages_ctdb.h"
#include "smbprofile.h"
#include "lib/id_cache.h"
#include "lib/param/param.h"
#include "lib/background.h"
#include "../lib/util/pidfile.h"
#include "lib/smbd_shim.h"
#include "scavenger.h"
#include "locking/leases_db.h"
#include "smbd/notifyd/notifyd.h"
#include "smbd/smbd_cleanupd.h"
#include "lib/util/sys_rw.h"
#include "cleanupdb.h"
#include "g_lock.h"
#include "lib/global_contexts.h"
#include "source3/lib/substitute.h"
#include "lib/addrchange.h"
#ifdef CLUSTER_SUPPORT
#include "ctdb_protocol.h"
#endif
struct smbd_open_socket;
struct smbd_child_pid;
struct smbd_parent_context {
bool interactive;
struct tevent_context *ev_ctx;
struct messaging_context *msg_ctx;
/* the list of listening sockets */
struct smbd_open_socket *sockets;
/* the list of current child processes */
struct smbd_child_pid *children;
size_t num_children;
struct server_id cleanupd;
struct server_id notifyd;
struct tevent_timer *cleanup_te;
};
struct smbd_open_socket {
struct smbd_open_socket *prev, *next;
struct smbd_parent_context *parent;
int fd;
struct tevent_fd *fde;
};
struct smbd_child_pid {
struct smbd_child_pid *prev, *next;
pid_t pid;
};
/*******************************************************************
What to do when smb.conf is updated.
********************************************************************/
static NTSTATUS messaging_send_to_children(struct messaging_context *msg_ctx,
uint32_t msg_type, DATA_BLOB* data);
static void smbd_parent_conf_updated(struct messaging_context *msg,
void *private_data,
uint32_t msg_type,
struct server_id server_id,
DATA_BLOB *data)
{
bool ok;
DEBUG(10,("smbd_parent_conf_updated: Got message saying smb.conf was "
"updated. Reloading.\n"));
change_to_root_user();
reload_services(NULL, NULL, false);
ok = reinit_guest_session_info(NULL);
if (!ok) {
DBG_ERR("Failed to reinit guest info\n");
}
messaging_send_to_children(msg, MSG_SMB_CONF_UPDATED, NULL);
}
/****************************************************************************
Send a SIGTERM to our process group.
*****************************************************************************/
static void killkids(void)
{
if(am_parent) kill(0,SIGTERM);
}
static void msg_exit_server(struct messaging_context *msg,
void *private_data,
uint32_t msg_type,
struct server_id server_id,
DATA_BLOB *data)
{
DEBUG(3, ("got a SHUTDOWN message\n"));
exit_server_cleanly(NULL);
}
#ifdef DEVELOPER
static void msg_inject_fault(struct messaging_context *msg,
void *private_data,
uint32_t msg_type,
struct server_id src,
DATA_BLOB *data)
{
int sig;
struct server_id_buf tmp;
if (data->length != sizeof(sig)) {
DEBUG(0, ("Process %s sent bogus signal injection request\n",
server_id_str_buf(src, &tmp)));
return;
}
sig = *(int *)data->data;
if (sig == -1) {
exit_server("internal error injected");
return;
}
#ifdef HAVE_STRSIGNAL
DEBUG(0, ("Process %s requested injection of signal %d (%s)\n",
server_id_str_buf(src, &tmp), sig, strsignal(sig)));
#else
DEBUG(0, ("Process %s requested injection of signal %d\n",
server_id_str_buf(src, &tmp), sig));
#endif
kill(getpid(), sig);
}
#endif /* DEVELOPER */
#if defined(DEVELOPER) || defined(ENABLE_SELFTEST)
/*
* Sleep for the specified number of seconds.
*/
static void msg_sleep(struct messaging_context *msg,
void *private_data,
uint32_t msg_type,
struct server_id src,
DATA_BLOB *data)
{
unsigned int seconds;
struct server_id_buf tmp;
if (data->length != sizeof(seconds)) {
DBG_ERR("Process %s sent bogus sleep request\n",
server_id_str_buf(src, &tmp));
return;
}
seconds = *(unsigned int *)data->data;
DBG_ERR("Process %s request a sleep of %u seconds\n",
server_id_str_buf(src, &tmp),
seconds);
sleep(seconds);
DBG_ERR("Restarting after %u second sleep requested by process %s\n",
seconds,
server_id_str_buf(src, &tmp));
}
#endif /* DEVELOPER */
static NTSTATUS messaging_send_to_children(struct messaging_context *msg_ctx,
uint32_t msg_type, DATA_BLOB* data)
{
NTSTATUS status;
struct smbd_parent_context *parent = am_parent;
struct smbd_child_pid *child;
if (parent == NULL) {
return NT_STATUS_INTERNAL_ERROR;
}
for (child = parent->children; child != NULL; child = child->next) {
status = messaging_send(parent->msg_ctx,
pid_to_procid(child->pid),
msg_type, data);
if (!NT_STATUS_IS_OK(status)) {
DBG_DEBUG("messaging_send(%d) failed: %s\n",
(int)child->pid, nt_errstr(status));
}
}
return NT_STATUS_OK;
}
static void smb_parent_send_to_children(struct messaging_context *ctx,
void* data,
uint32_t msg_type,
struct server_id srv_id,
DATA_BLOB* msg_data)
{
messaging_send_to_children(ctx, msg_type, msg_data);
}
/*
* Parent smbd process sets its own debug level first and then
* sends a message to all the smbd children to adjust their debug
* level to that of the parent.
*/
static void smbd_msg_debug(struct messaging_context *msg_ctx,
void *private_data,
uint32_t msg_type,
struct server_id server_id,
DATA_BLOB *data)
{
debug_message(msg_ctx, private_data, MSG_DEBUG, server_id, data);
messaging_send_to_children(msg_ctx, MSG_DEBUG, data);
}
static void smbd_parent_id_cache_kill(struct messaging_context *msg_ctx,
void *private_data,
uint32_t msg_type,
struct server_id server_id,
DATA_BLOB* data)
{
const char *msg = (data && data->data)
? (const char *)data->data : "<NULL>";
struct id_cache_ref id;
if (!id_cache_ref_parse(msg, &id)) {
DEBUG(0, ("Invalid ?ID: %s\n", msg));
return;
}
id_cache_delete_from_cache(&id);
messaging_send_to_children(msg_ctx, msg_type, data);
}
static void smbd_parent_id_cache_delete(struct messaging_context *ctx,
void* data,
uint32_t msg_type,
struct server_id srv_id,
DATA_BLOB* msg_data)
{
id_cache_delete_message(ctx, data, msg_type, srv_id, msg_data);
messaging_send_to_children(ctx, msg_type, msg_data);
}
static void add_child_pid(struct smbd_parent_context *parent,
pid_t pid)
{
struct smbd_child_pid *child;
child = talloc_zero(parent, struct smbd_child_pid);
if (child == NULL) {
DEBUG(0, ("Could not add child struct -- malloc failed\n"));
return;
}
child->pid = pid;
DLIST_ADD(parent->children, child);
parent->num_children += 1;
}
static void smb_tell_num_children(struct messaging_context *ctx, void *data,
uint32_t msg_type, struct server_id srv_id,
DATA_BLOB *msg_data)
{
uint8_t buf[sizeof(uint32_t)];
if (am_parent) {
SIVAL(buf, 0, am_parent->num_children);
messaging_send_buf(ctx, srv_id, MSG_SMB_NUM_CHILDREN,
buf, sizeof(buf));
}
}
static void notifyd_stopped(struct tevent_req *req);
static struct tevent_req *notifyd_req(struct messaging_context *msg_ctx,
struct tevent_context *ev)
{
struct tevent_req *req;
sys_notify_watch_fn sys_notify_watch = NULL;
struct sys_notify_context *sys_notify_ctx = NULL;
struct ctdbd_connection *ctdbd_conn = NULL;
if (lp_kernel_change_notify()) {
#ifdef HAVE_INOTIFY
if (lp_parm_bool(-1, "notify", "inotify", true)) {
sys_notify_watch = inotify_watch;
}
#endif
#ifdef HAVE_FAM
if (lp_parm_bool(-1, "notify", "fam",
(sys_notify_watch == NULL))) {
sys_notify_watch = fam_watch;
}
#endif
}
if (sys_notify_watch != NULL) {
sys_notify_ctx = sys_notify_context_create(msg_ctx, ev);
if (sys_notify_ctx == NULL) {
return NULL;
}
}
if (lp_clustering()) {
ctdbd_conn = messaging_ctdb_connection();
}
req = notifyd_send(msg_ctx, ev, msg_ctx, ctdbd_conn,
sys_notify_watch, sys_notify_ctx);
if (req == NULL) {
TALLOC_FREE(sys_notify_ctx);
return NULL;
}
tevent_req_set_callback(req, notifyd_stopped, msg_ctx);
return req;
}
static void notifyd_stopped(struct tevent_req *req)
{
int ret;
ret = notifyd_recv(req);
TALLOC_FREE(req);
DEBUG(1, ("notifyd stopped: %s\n", strerror(ret)));
}
static void notifyd_sig_hup_handler(struct tevent_context *ev,
struct tevent_signal *se,
int signum,
int count,
void *siginfo,
void *pvt)
{
DBG_NOTICE("notifyd: Reloading services after SIGHUP\n");
reload_services(NULL, NULL, false);
reopen_logs();
}
static bool smbd_notifyd_init(struct messaging_context *msg, bool interactive,
struct server_id *ppid)
{
struct tevent_context *ev = messaging_tevent_context(msg);
struct tevent_req *req;
struct tevent_signal *se = NULL;
pid_t pid;
NTSTATUS status;
bool ok;
if (interactive) {
req = notifyd_req(msg, ev);
return (req != NULL);
}
pid = fork();
if (pid == -1) {
DEBUG(1, ("%s: fork failed: %s\n", __func__,
strerror(errno)));
return false;
}
if (pid != 0) {
if (am_parent != NULL) {
add_child_pid(am_parent, pid);
}
*ppid = pid_to_procid(pid);
return true;
}
status = smbd_reinit_after_fork(msg, ev, true);
if (!NT_STATUS_IS_OK(status)) {
DEBUG(1, ("%s: reinit_after_fork failed: %s\n",
__func__, nt_errstr(status)));
exit(1);
}
process_set_title("smbd-notifyd", "notifyd");
reopen_logs();
/* Set up sighup handler for notifyd */
se = tevent_add_signal(ev,
ev,
SIGHUP, 0,
notifyd_sig_hup_handler,
NULL);
if (!se) {
DEBUG(0, ("failed to setup notifyd SIGHUP handler\n"));
exit(1);
}
req = notifyd_req(msg, ev);
if (req == NULL) {
exit(1);
}
tevent_req_set_callback(req, notifyd_stopped, msg);
/* Block those signals that we are not handling */
BlockSignals(True, SIGUSR1);
messaging_send(msg, pid_to_procid(getppid()), MSG_SMB_NOTIFY_STARTED,
NULL);
ok = tevent_req_poll(req, ev);
if (!ok) {
DBG_WARNING("tevent_req_poll returned %s\n", strerror(errno));
exit(1);
}
exit(0);
}
static void notifyd_init_trigger(struct tevent_req *req);
struct notifyd_init_state {
bool ok;
struct tevent_context *ev;
struct messaging_context *msg;
struct server_id *ppid;
};
static struct tevent_req *notifyd_init_send(struct tevent_context *ev,
TALLOC_CTX *mem_ctx,
struct messaging_context *msg,
struct server_id *ppid)
{
struct tevent_req *req = NULL;
struct tevent_req *subreq = NULL;
struct notifyd_init_state *state = NULL;
req = tevent_req_create(mem_ctx, &state, struct notifyd_init_state);
if (req == NULL) {
return NULL;
}
*state = (struct notifyd_init_state) {
.msg = msg,
.ev = ev,
.ppid = ppid
};
subreq = tevent_wakeup_send(state, ev, tevent_timeval_current_ofs(1, 0));
if (tevent_req_nomem(subreq, req)) {
return tevent_req_post(req, ev);
}
tevent_req_set_callback(subreq, notifyd_init_trigger, req);
return req;
}
static void notifyd_init_trigger(struct tevent_req *subreq)
{
struct tevent_req *req = tevent_req_callback_data(
subreq, struct tevent_req);
struct notifyd_init_state *state = tevent_req_data(
req, struct notifyd_init_state);
bool ok;
DBG_NOTICE("Triggering notifyd startup\n");
ok = tevent_wakeup_recv(subreq);
TALLOC_FREE(subreq);
if (!ok) {
tevent_req_error(req, ENOMEM);
return;
}
state->ok = smbd_notifyd_init(state->msg, false, state->ppid);
if (state->ok) {
DBG_WARNING("notifyd restarted\n");
tevent_req_done(req);
return;
}
DBG_NOTICE("notifyd startup failed, rescheduling\n");
subreq = tevent_wakeup_send(state, state->ev,
tevent_timeval_current_ofs(1, 0));
if (tevent_req_nomem(subreq, req)) {
DBG_ERR("scheduling notifyd restart failed, giving up\n");
return;
}
tevent_req_set_callback(subreq, notifyd_init_trigger, req);
return;
}
static bool notifyd_init_recv(struct tevent_req *req)
{
struct notifyd_init_state *state = tevent_req_data(
req, struct notifyd_init_state);
return state->ok;
}
static void notifyd_started(struct tevent_req *req)
{
bool ok;
ok = notifyd_init_recv(req);
TALLOC_FREE(req);
if (!ok) {
DBG_ERR("Failed to restart notifyd, giving up\n");
return;
}
}
static void cleanupd_sig_hup_handler(struct tevent_context *ev,
struct tevent_signal *se,
int signum,
int count,
void *siginfo,
void *pvt)
{
DBG_NOTICE("cleanupd: Reloading services after SIGHUP\n");
reopen_logs();
}
static void cleanupd_stopped(struct tevent_req *req);
static bool cleanupd_init(struct messaging_context *msg, bool interactive,
struct server_id *ppid)
{
struct tevent_context *ev = messaging_tevent_context(msg);
struct server_id parent_id = messaging_server_id(msg);
struct tevent_signal *se = NULL;
struct tevent_req *req;
pid_t pid;
NTSTATUS status;
ssize_t rwret;
int ret;
bool ok;
char c;
int up_pipe[2];
if (interactive) {
req = smbd_cleanupd_send(msg, ev, msg, parent_id.pid);
*ppid = messaging_server_id(msg);
return (req != NULL);
}
ret = pipe(up_pipe);
if (ret == -1) {
DBG_WARNING("pipe failed: %s\n", strerror(errno));
return false;
}
pid = fork();
if (pid == -1) {
DBG_WARNING("fork failed: %s\n", strerror(errno));
close(up_pipe[0]);
close(up_pipe[1]);
return false;
}
if (pid != 0) {
close(up_pipe[1]);
rwret = sys_read(up_pipe[0], &c, 1);
close(up_pipe[0]);
if (rwret == -1) {
DBG_WARNING("sys_read failed: %s\n", strerror(errno));
return false;
}
if (rwret == 0) {
DBG_WARNING("cleanupd could not start\n");
return false;
}
if (c != 0) {
DBG_WARNING("cleanupd returned %d\n", (int)c);
return false;
}
DBG_DEBUG("Started cleanupd pid=%d\n", (int)pid);
if (am_parent != NULL) {
add_child_pid(am_parent, pid);
}
*ppid = pid_to_procid(pid);
return true;
}
close(up_pipe[0]);
status = smbd_reinit_after_fork(msg, ev, true);
if (!NT_STATUS_IS_OK(status)) {
DBG_WARNING("reinit_after_fork failed: %s\n",
nt_errstr(status));
c = 1;
sys_write(up_pipe[1], &c, 1);
exit(1);
}
process_set_title("smbd-cleanupd", "cleanupd");
se = tevent_add_signal(ev,
ev,
SIGHUP,
0,
cleanupd_sig_hup_handler,
NULL);
if (se == NULL) {
DBG_ERR("Could not add SIGHUP handler\n");
exit(1);
}
req = smbd_cleanupd_send(msg, ev, msg, parent_id.pid);
if (req == NULL) {
DBG_WARNING("smbd_cleanupd_send failed\n");
c = 2;
sys_write(up_pipe[1], &c, 1);
exit(1);
}
tevent_req_set_callback(req, cleanupd_stopped, msg);
c = 0;
rwret = sys_write(up_pipe[1], &c, 1);
close(up_pipe[1]);
if (rwret == -1) {
DBG_WARNING("sys_write failed: %s\n", strerror(errno));
exit(1);
}
if (rwret != 1) {
DBG_WARNING("sys_write could not write result\n");
exit(1);
}
ok = tevent_req_poll(req, ev);
if (!ok) {
DBG_WARNING("tevent_req_poll returned %s\n", strerror(errno));
}
exit(0);
}
static void cleanupd_stopped(struct tevent_req *req)
{
NTSTATUS status;
status = smbd_cleanupd_recv(req);
DBG_WARNING("cleanupd stopped: %s\n", nt_errstr(status));
}
static void cleanupd_init_trigger(struct tevent_req *req);
struct cleanup_init_state {
bool ok;
struct tevent_context *ev;
struct messaging_context *msg;
struct server_id *ppid;
};
static struct tevent_req *cleanupd_init_send(struct tevent_context *ev,
TALLOC_CTX *mem_ctx,
struct messaging_context *msg,
struct server_id *ppid)
{
struct tevent_req *req = NULL;
struct tevent_req *subreq = NULL;
struct cleanup_init_state *state = NULL;
req = tevent_req_create(mem_ctx, &state, struct cleanup_init_state);
if (req == NULL) {
return NULL;
}
*state = (struct cleanup_init_state) {
.msg = msg,
.ev = ev,
.ppid = ppid
};
subreq = tevent_wakeup_send(state, ev, tevent_timeval_current_ofs(0, 0));
if (tevent_req_nomem(subreq, req)) {
return tevent_req_post(req, ev);
}
tevent_req_set_callback(subreq, cleanupd_init_trigger, req);
return req;
}
static void cleanupd_init_trigger(struct tevent_req *subreq)
{
struct tevent_req *req = tevent_req_callback_data(
subreq, struct tevent_req);
struct cleanup_init_state *state = tevent_req_data(
req, struct cleanup_init_state);
bool ok;
DBG_NOTICE("Triggering cleanupd startup\n");
ok = tevent_wakeup_recv(subreq);
TALLOC_FREE(subreq);
if (!ok) {
tevent_req_error(req, ENOMEM);
return;
}
state->ok = cleanupd_init(state->msg, false, state->ppid);
if (state->ok) {
DBG_WARNING("cleanupd restarted\n");
tevent_req_done(req);
return;
}
DBG_NOTICE("cleanupd startup failed, rescheduling\n");
subreq = tevent_wakeup_send(state, state->ev,
tevent_timeval_current_ofs(1, 0));
if (tevent_req_nomem(subreq, req)) {
DBG_ERR("scheduling cleanupd restart failed, giving up\n");
return;
}
tevent_req_set_callback(subreq, cleanupd_init_trigger, req);
return;
}
static bool cleanupd_init_recv(struct tevent_req *req)
{
struct cleanup_init_state *state = tevent_req_data(
req, struct cleanup_init_state);
return state->ok;
}
static void cleanupd_started(struct tevent_req *req)
{
bool ok;
NTSTATUS status;
struct smbd_parent_context *parent = tevent_req_callback_data(
req, struct smbd_parent_context);
ok = cleanupd_init_recv(req);
TALLOC_FREE(req);
if (!ok) {
DBG_ERR("Failed to restart cleanupd, giving up\n");
return;
}
status = messaging_send(parent->msg_ctx,
parent->cleanupd,
MSG_SMB_NOTIFY_CLEANUP,
&data_blob_null);
if (!NT_STATUS_IS_OK(status)) {
DBG_ERR("messaging_send returned %s\n",
nt_errstr(status));
}
}
static void remove_child_pid(struct smbd_parent_context *parent,
pid_t pid,
bool unclean_shutdown)
{
struct smbd_child_pid *child;
NTSTATUS status;
bool ok;
for (child = parent->children; child != NULL; child = child->next) {
if (child->pid == pid) {
struct smbd_child_pid *tmp = child;
DLIST_REMOVE(parent->children, child);
TALLOC_FREE(tmp);
parent->num_children -= 1;
break;
}
}
if (child == NULL) {
/* not all forked child processes are added to the children list */
DEBUG(2, ("Could not find child %d -- ignoring\n", (int)pid));
return;
}
if (pid == procid_to_pid(&parent->cleanupd)) {
struct tevent_req *req;
server_id_set_disconnected(&parent->cleanupd);
DBG_WARNING("Restarting cleanupd\n");
req = cleanupd_init_send(messaging_tevent_context(parent->msg_ctx),
parent,
parent->msg_ctx,
&parent->cleanupd);
if (req == NULL) {
DBG_ERR("Failed to restart cleanupd\n");
return;
}
tevent_req_set_callback(req, cleanupd_started, parent);
return;
}
if (pid == procid_to_pid(&parent->notifyd)) {
struct tevent_req *req;
struct tevent_context *ev = messaging_tevent_context(
parent->msg_ctx);
server_id_set_disconnected(&parent->notifyd);
DBG_WARNING("Restarting notifyd\n");
req = notifyd_init_send(ev,
parent,
parent->msg_ctx,
&parent->notifyd);
if (req == NULL) {
DBG_ERR("Failed to restart notifyd\n");
return;
}
tevent_req_set_callback(req, notifyd_started, parent);
return;
}
ok = cleanupdb_store_child(pid, unclean_shutdown);
if (!ok) {
DBG_ERR("cleanupdb_store_child failed\n");
return;
}
if (!server_id_is_disconnected(&parent->cleanupd)) {
status = messaging_send(parent->msg_ctx,
parent->cleanupd,
MSG_SMB_NOTIFY_CLEANUP,
&data_blob_null);
if (!NT_STATUS_IS_OK(status)) {
DBG_ERR("messaging_send returned %s\n",
nt_errstr(status));
}
}
}
/****************************************************************************
Have we reached the process limit ?
****************************************************************************/
static bool allowable_number_of_smbd_processes(struct smbd_parent_context *parent)
{
int max_processes = lp_max_smbd_processes();
if (!max_processes)
return True;
return parent->num_children < max_processes;
}
static void smbd_sig_chld_handler(struct tevent_context *ev,
struct tevent_signal *se,
int signum,
int count,
void *siginfo,
void *private_data)
{
pid_t pid;
int status;
struct smbd_parent_context *parent =
talloc_get_type_abort(private_data,
struct smbd_parent_context);
while ((pid = waitpid(-1, &status, WNOHANG)) > 0) {
bool unclean_shutdown = False;
/* If the child terminated normally, assume
it was an unclean shutdown unless the
status is 0
*/
if (WIFEXITED(status)) {
unclean_shutdown = WEXITSTATUS(status);
}
/* If the child terminated due to a signal
we always assume it was unclean.
*/
if (WIFSIGNALED(status)) {
unclean_shutdown = True;
}
remove_child_pid(parent, pid, unclean_shutdown);
}
}
static void smbd_setup_sig_chld_handler(struct smbd_parent_context *parent)
{
struct tevent_signal *se;
se = tevent_add_signal(parent->ev_ctx,
parent, /* mem_ctx */
SIGCHLD, 0,
smbd_sig_chld_handler,
parent);
if (!se) {
exit_server("failed to setup SIGCHLD handler");
}
}
static void smbd_open_socket_close_fn(struct tevent_context *ev,
struct tevent_fd *fde,
int fd,
void *private_data)
{
/* this might be the socket_wrapper swrap_close() */
close(fd);
}
static void smbd_accept_connection(struct tevent_context *ev,
struct tevent_fd *fde,
uint16_t flags,
void *private_data)
{
struct smbd_open_socket *s = talloc_get_type_abort(private_data,
struct smbd_open_socket);
struct messaging_context *msg_ctx = s->parent->msg_ctx;
struct sockaddr_storage addr;
socklen_t in_addrlen = sizeof(addr);
int fd;
pid_t pid = 0;
fd = accept(s->fd, (struct sockaddr *)(void *)&addr,&in_addrlen);
if (fd == -1 && errno == EINTR)
return;
if (fd == -1) {
DEBUG(0,("accept: %s\n",
strerror(errno)));
return;
}
smb_set_close_on_exec(fd);
if (s->parent->interactive) {
NTSTATUS status;
status = reinit_after_fork(msg_ctx, ev, true);
if (!NT_STATUS_IS_OK(status)) {
exit_server("reinit_after_fork() failed");
return;
}
smbd_process(ev, msg_ctx, fd, true);
exit_server_cleanly("end of interactive mode");
return;
}
if (!allowable_number_of_smbd_processes(s->parent)) {
close(fd);
return;
}
pid = fork();
if (pid == 0) {
char addrstr[INET6_ADDRSTRLEN];
NTSTATUS status = NT_STATUS_OK;
/*
* Can't use TALLOC_FREE here. Nulling out the argument to it
* would overwrite memory we've just freed.
*/
talloc_free(s->parent);
s = NULL;
/* Stop zombies, the parent explicitly handles
* them, counting worker smbds. */
CatchChild();
status = smbd_reinit_after_fork(msg_ctx, ev, true);
if (!NT_STATUS_IS_OK(status)) {
if (NT_STATUS_EQUAL(status,
NT_STATUS_TOO_MANY_OPENED_FILES)) {
DEBUG(0,("child process cannot initialize "
"because too many files are open\n"));
goto exit;
}
if (lp_clustering() &&
(NT_STATUS_EQUAL(
status, NT_STATUS_INTERNAL_DB_ERROR) ||
NT_STATUS_EQUAL(
status, NT_STATUS_CONNECTION_REFUSED))) {
DEBUG(1, ("child process cannot initialize "
"because connection to CTDB "
"has failed: %s\n",
nt_errstr(status)));
goto exit;
}
DEBUG(0,("reinit_after_fork() failed\n"));
smb_panic("reinit_after_fork() failed");
}
print_sockaddr(addrstr, sizeof(addrstr), &addr);
process_set_title("smbd[%s]", "client [%s]", addrstr);
smbd_process(ev, msg_ctx, fd, false);
exit:
exit_server_cleanly("end of child");
return;
}
if (pid < 0) {
DEBUG(0,("smbd_accept_connection: fork() failed: %s\n",
strerror(errno)));
}
/* The parent doesn't need this socket */
close(fd);
/* Sun May 6 18:56:14 2001 ackley@cs.unm.edu:
Clear the closed fd info out of server_fd --
and more importantly, out of client_fd in
util_sock.c, to avoid a possible
getpeername failure if we reopen the logs
and use %I in the filename.
*/
if (pid != 0) {
add_child_pid(s->parent, pid);
}
/* Force parent to check log size after
* spawning child. Fix from
* klausr@ITAP.Physik.Uni-Stuttgart.De. The
* parent smbd will log to logserver.smb. It
* writes only two messages for each child
* started/finished. But each child writes,
* say, 50 messages also in logserver.smb,
* beginning with the debug_count of the
* parent, before the child opens its own log
* file logserver.client. In a worst case
* scenario the size of logserver.smb would be
* checked after about 50*50=2500 messages
* (ca. 100kb).
* */
force_check_log_size();
}
static bool smbd_open_one_socket(struct smbd_parent_context *parent,
struct tevent_context *ev_ctx,
const struct sockaddr_storage *ifss,
uint16_t port)
{
struct smbd_open_socket *s;
s = talloc(parent, struct smbd_open_socket);
if (!s) {
return false;
}
s->parent = parent;
s->fd = open_socket_in(SOCK_STREAM, ifss, port, true);
if (s->fd < 0) {
int err = -(s->fd);
DBG_ERR("open_socket_in failed: %s\n", strerror(err));
TALLOC_FREE(s);
/*
* We ignore an error here, as we've done before
*/
return true;
}
/* ready to listen */
set_socket_options(s->fd, "SO_KEEPALIVE");
set_socket_options(s->fd, lp_socket_options());
/* Set server socket to
* non-blocking for the accept. */
set_blocking(s->fd, False);
if (listen(s->fd, SMBD_LISTEN_BACKLOG) == -1) {
DEBUG(0,("smbd_open_one_socket: listen: "
"%s\n", strerror(errno)));
close(s->fd);
TALLOC_FREE(s);
return false;
}
s->fde = tevent_add_fd(ev_ctx,
s,
s->fd, TEVENT_FD_READ,
smbd_accept_connection,
s);
if (!s->fde) {
DEBUG(0,("smbd_open_one_socket: "
"tevent_add_fd: %s\n",
strerror(errno)));
close(s->fd);
TALLOC_FREE(s);
return false;
}
tevent_fd_set_close_fn(s->fde, smbd_open_socket_close_fn);
DLIST_ADD_END(parent->sockets, s);
return true;
}
/****************************************************************************
Open the socket communication.
****************************************************************************/
static bool open_sockets_smbd(struct smbd_parent_context *parent,
struct tevent_context *ev_ctx,
struct messaging_context *msg_ctx,
const char *smb_ports)
{
int num_interfaces = iface_count();
int i,j;
const char **ports;
unsigned dns_port = 0;
#ifdef HAVE_ATEXIT
atexit(killkids);
#endif
/* Stop zombies */
smbd_setup_sig_chld_handler(parent);
ports = lp_smb_ports();
/* use a reasonable default set of ports - listing on 445 and 139 */
if (smb_ports) {
char **l;
l = str_list_make_v3(talloc_tos(), smb_ports, NULL);
ports = discard_const_p(const char *, l);
}
for (j = 0; ports && ports[j]; j++) {
unsigned port = atoi(ports[j]);
if (port == 0 || port > 0xffff) {
exit_server_cleanly("Invalid port in the config or on "
"the commandline specified!");
}
}
if (lp_interfaces() && lp_bind_interfaces_only()) {
/* We have been given an interfaces line, and been
told to only bind to those interfaces. Create a
socket per interface and bind to only these.
*/
/* Now open a listen socket for each of the
interfaces. */
for(i = 0; i < num_interfaces; i++) {
const struct sockaddr_storage *ifss =
iface_n_sockaddr_storage(i);
if (ifss == NULL) {
DEBUG(0,("open_sockets_smbd: "
"interface %d has NULL IP address !\n",
i));
continue;
}
for (j = 0; ports && ports[j]; j++) {
unsigned port = atoi(ports[j]);
/* Keep the first port for mDNS service
* registration.
*/
if (dns_port == 0) {
dns_port = port;
}
if (!smbd_open_one_socket(parent,
ev_ctx,
ifss,
port)) {
return false;
}
}
}
} else {
/* Just bind to 0.0.0.0 - accept connections
from anywhere. */
const char *sock_addr;
char *sock_tok;
const char *sock_ptr;
#ifdef HAVE_IPV6
sock_addr = "::,0.0.0.0";
#else
sock_addr = "0.0.0.0";
#endif
for (sock_ptr=sock_addr;
next_token_talloc(talloc_tos(), &sock_ptr, &sock_tok, " \t,"); ) {
for (j = 0; ports && ports[j]; j++) {
struct sockaddr_storage ss;
unsigned port = atoi(ports[j]);
/* Keep the first port for mDNS service
* registration.
*/
if (dns_port == 0) {
dns_port = port;
}
/* open an incoming socket */
if (!interpret_string_addr(&ss, sock_tok,
AI_NUMERICHOST|AI_PASSIVE)) {
continue;
}
/*
* If we fail to open any sockets
* in this loop the parent-sockets == NULL
* case below will prevent us from starting.
*/
(void)smbd_open_one_socket(parent,
ev_ctx,
&ss,
port);
}
}
}
if (parent->sockets == NULL) {
DEBUG(0,("open_sockets_smbd: No "
"sockets available to bind to.\n"));
return false;
}
/* Listen to messages */
messaging_register(msg_ctx, NULL, MSG_SHUTDOWN, msg_exit_server);
messaging_register(msg_ctx, ev_ctx, MSG_SMB_CONF_UPDATED,
smbd_parent_conf_updated);
messaging_register(msg_ctx, NULL, MSG_DEBUG, smbd_msg_debug);
messaging_register(msg_ctx, NULL, MSG_SMB_FORCE_TDIS,
smb_parent_send_to_children);
messaging_register(msg_ctx, NULL, MSG_SMB_FORCE_TDIS_DENIED,
smb_parent_send_to_children);
messaging_register(msg_ctx, NULL, MSG_SMB_KILL_CLIENT_IP,
smb_parent_send_to_children);
messaging_register(msg_ctx, NULL, MSG_SMB_TELL_NUM_CHILDREN,
smb_tell_num_children);
messaging_register(msg_ctx, NULL,
ID_CACHE_DELETE, smbd_parent_id_cache_delete);
messaging_register(msg_ctx, NULL,
ID_CACHE_KILL, smbd_parent_id_cache_kill);
messaging_register(msg_ctx, NULL, MSG_SMB_NOTIFY_STARTED,
smb_parent_send_to_children);
if (lp_interfaces() && lp_bind_interfaces_only()) {
messaging_register(msg_ctx,
NULL,
MSG_SMB_IP_DROPPED,
smb_parent_send_to_children);
}
#ifdef DEVELOPER
messaging_register(msg_ctx, NULL, MSG_SMB_INJECT_FAULT,
msg_inject_fault);
#endif
#if defined(DEVELOPER) || defined(ENABLE_SELFTEST)
messaging_register(msg_ctx, NULL, MSG_SMB_SLEEP, msg_sleep);
#endif
if (lp_multicast_dns_register() && (dns_port != 0)) {
#ifdef WITH_DNSSD_SUPPORT
smbd_setup_mdns_registration(ev_ctx,
parent, dns_port);
#endif
#ifdef WITH_AVAHI_SUPPORT
void *avahi_conn;
avahi_conn = avahi_start_register(ev_ctx,
ev_ctx,
dns_port);
if (avahi_conn == NULL) {
DEBUG(10, ("avahi_start_register failed\n"));
}
#endif
}
return true;
}
/*
handle stdin becoming readable when we are in --foreground mode
*/
static void smbd_stdin_handler(struct tevent_context *ev,
struct tevent_fd *fde,
uint16_t flags,
void *private_data)
{
char c;
if (read(0, &c, 1) != 1) {
/* we have reached EOF on stdin, which means the
parent has exited. Shutdown the server */
exit_server_cleanly("EOF on stdin");
}
}
struct smbd_parent_tevent_trace_state {
TALLOC_CTX *frame;
};
static void smbd_parent_tevent_trace_callback(enum tevent_trace_point point,
void *private_data)
{
struct smbd_parent_tevent_trace_state *state =
(struct smbd_parent_tevent_trace_state *)private_data;
switch (point) {
case TEVENT_TRACE_BEFORE_WAIT:
break;
case TEVENT_TRACE_AFTER_WAIT:
break;
case TEVENT_TRACE_BEFORE_LOOP_ONCE:
TALLOC_FREE(state->frame);
state->frame = talloc_stackframe();
break;
case TEVENT_TRACE_AFTER_LOOP_ONCE:
TALLOC_FREE(state->frame);
break;
}
errno = 0;
}
static void smbd_parent_loop(struct tevent_context *ev_ctx,
struct smbd_parent_context *parent)
{
struct smbd_parent_tevent_trace_state trace_state = {
.frame = NULL,
};
int ret = 0;
tevent_set_trace_callback(ev_ctx, smbd_parent_tevent_trace_callback,
&trace_state);
/* now accept incoming connections - forking a new process
for each incoming connection */
DEBUG(2,("waiting for connections\n"));
ret = tevent_loop_wait(ev_ctx);
if (ret != 0) {
DEBUG(0, ("tevent_loop_wait failed: %d, %s, exiting\n",
ret, strerror(errno)));
}
TALLOC_FREE(trace_state.frame);
/* NOTREACHED return True; */
}
/****************************************************************************
Initialise connect, service and file structs.
****************************************************************************/
static bool init_structs(void )
{
/*
* Set the machine NETBIOS name if not already
* set from the config file.
*/
if (!secrets_init())
return False;
return True;
}
static void smbd_parent_sig_term_handler(struct tevent_context *ev,
struct tevent_signal *se,
int signum,
int count,
void *siginfo,
void *private_data)
{
exit_server_cleanly("termination signal");
}
static void smbd_parent_sig_hup_handler(struct tevent_context *ev,
struct tevent_signal *se,
int signum,
int count,
void *siginfo,
void *private_data)
{
change_to_root_user();
DBG_NOTICE("parent: Reloading services after SIGHUP\n");
reload_services(NULL, NULL, false);
}
struct smbd_claim_version_state {
TALLOC_CTX *mem_ctx;
char *version;
};
static void smbd_claim_version_parser(struct server_id exclusive,
size_t num_shared,
const struct server_id *shared,
const uint8_t *data,
size_t datalen,
void *private_data)
{
struct smbd_claim_version_state *state = private_data;
if (datalen == 0) {
state->version = NULL;
return;
}
if (data[datalen-1] != '\0') {
DBG_WARNING("Invalid samba version\n");
dump_data(DBGLVL_WARNING, data, datalen);
state->version = NULL;
return;
}
state->version = talloc_strdup(state->mem_ctx, (const char *)data);
}
static NTSTATUS smbd_claim_version(struct messaging_context *msg,
const char *version)
{
const char *name = "samba_version_string";
const TDB_DATA key = string_term_tdb_data(name);
struct smbd_claim_version_state state;
struct g_lock_ctx *ctx;
NTSTATUS status;
ctx = g_lock_ctx_init(msg, msg);
if (ctx == NULL) {
DBG_WARNING("g_lock_ctx_init failed\n");
return NT_STATUS_UNSUCCESSFUL;
}
status = g_lock_lock(ctx,
key,
G_LOCK_READ,
(struct timeval) { .tv_sec = 60 },
NULL,
NULL);
if (!NT_STATUS_IS_OK(status)) {
DBG_WARNING("g_lock_lock(G_LOCK_READ) failed: %s\n",
nt_errstr(status));
TALLOC_FREE(ctx);
return status;
}
state = (struct smbd_claim_version_state) { .mem_ctx = ctx };
status = g_lock_dump(ctx, key, smbd_claim_version_parser, &state);
if (!NT_STATUS_IS_OK(status) &&
!NT_STATUS_EQUAL(status, NT_STATUS_NOT_FOUND)) {
DBG_ERR("Could not read samba_version_string\n");
g_lock_unlock(ctx, key);
TALLOC_FREE(ctx);
return status;
}
if ((state.version != NULL) && (strcmp(version, state.version) == 0)) {
/*
* Leave the read lock for us around. Someone else already
* set the version correctly
*/
TALLOC_FREE(ctx);
return NT_STATUS_OK;
}
status = g_lock_lock(ctx,
key,
G_LOCK_UPGRADE,
(struct timeval) { .tv_sec = 60 },
NULL,
NULL);
if (!NT_STATUS_IS_OK(status)) {
DBG_WARNING("g_lock_lock(G_LOCK_UPGRADE) failed: %s\n",
nt_errstr(status));
DBG_ERR("smbd %s already running, refusing to start "
"version %s\n", state.version, version);
TALLOC_FREE(ctx);
return NT_STATUS_SXS_VERSION_CONFLICT;
}
status = g_lock_write_data(
ctx, key, (const uint8_t *)version, strlen(version)+1);
if (!NT_STATUS_IS_OK(status)) {
DBG_WARNING("g_lock_write_data failed: %s\n",
nt_errstr(status));
TALLOC_FREE(ctx);
return status;
}
status = g_lock_lock(ctx,
key,
G_LOCK_DOWNGRADE,
(struct timeval) { .tv_sec = 60 },
NULL,
NULL);
if (!NT_STATUS_IS_OK(status)) {
DBG_WARNING("g_lock_lock(G_LOCK_DOWNGRADE) failed: %s\n",
nt_errstr(status));
TALLOC_FREE(ctx);
return status;
}
/*
* Leave "ctx" dangling so that g_lock.tdb keeps opened.
*/
return NT_STATUS_OK;
}
/****************************************************************************
Open socket communication on given ip address
****************************************************************************/
static bool smbd_open_socket_for_ip(struct smbd_parent_context *parent,
struct tevent_context *ev_ctx,
struct messaging_context *msg_ctx,
const char *smb_ports,
const struct sockaddr_storage *ifss)
{
int j;
const char **ports;
unsigned dns_port = 0;
TALLOC_CTX *ctx;
bool status = true;
ports = lp_smb_ports();
ctx = talloc_stackframe();
/* use a reasonable default set of ports - listing on 445 and 139 */
if (smb_ports) {
char **l;
l = str_list_make_v3(ctx, smb_ports, NULL);
ports = discard_const_p(const char *, l);
}
for (j = 0; ports && ports[j]; j++) {
unsigned port = atoi(ports[j]);
/* Keep the first port for mDNS service
* registration.
*/
if (dns_port == 0) {
dns_port = port;
}
if (!smbd_open_one_socket(parent,
ev_ctx,
ifss,
port)) {
status = false;
goto out_free;
}
}
out_free:
TALLOC_FREE(ctx);
return status;
}
struct smbd_addrchanged_state {
struct addrchange_context *ctx;
struct tevent_context *ev;
struct messaging_context *msg_ctx;
struct smbd_parent_context *parent;
char *ports;
};
static void smbd_addr_changed(struct tevent_req *req);
static void smbd_init_addrchange(TALLOC_CTX *mem_ctx,
struct tevent_context *ev,
struct messaging_context *msg_ctx,
struct smbd_parent_context *parent,
char *ports)
{
struct smbd_addrchanged_state *state;
struct tevent_req *req;
NTSTATUS status;
state = talloc(mem_ctx, struct smbd_addrchanged_state);
if (state == NULL) {
DBG_DEBUG("talloc failed\n");
return;
}
*state = (struct smbd_addrchanged_state) {
.ev = ev,
.msg_ctx = msg_ctx,
.parent = parent,
.ports = ports
};
status = addrchange_context_create(state, &state->ctx);
if (!NT_STATUS_IS_OK(status)) {
DBG_DEBUG("addrchange_context_create failed: %s\n",
nt_errstr(status));
TALLOC_FREE(state);
return;
}
req = addrchange_send(state, ev, state->ctx);
if (req == NULL) {
DBG_ERR("addrchange_send failed\n");
TALLOC_FREE(state);
return;
}
tevent_req_set_callback(req, smbd_addr_changed, state);
}
static void smbd_close_socket_for_ip(struct smbd_parent_context *parent,
struct messaging_context *msg_ctx,
struct sockaddr_storage *addr)
{
struct smbd_open_socket *s = NULL;
for (s = parent->sockets; s != NULL; s = s->next) {
struct sockaddr_storage a;
socklen_t addr_len = sizeof(a);
if (getsockname(s->fd, (struct sockaddr *)&a, &addr_len) < 0) {
DBG_NOTICE("smbd: Unable to get address - skip\n");
continue;
}
if (sockaddr_equal((struct sockaddr *)&a,
(struct sockaddr *)addr)) {
char addrstr[INET6_ADDRSTRLEN];
DATA_BLOB blob;
NTSTATUS status;
DLIST_REMOVE(parent->sockets, s);
TALLOC_FREE(s);
print_sockaddr(addrstr, sizeof(addrstr), addr);
DBG_NOTICE("smbd: Closed listening socket for %s\n",
addrstr);
blob = data_blob_const(addrstr, strlen(addrstr)+1);
status = messaging_send(msg_ctx,
messaging_server_id(msg_ctx),
MSG_SMB_IP_DROPPED,
&blob);
if (!NT_STATUS_IS_OK(status)) {
DBG_NOTICE(
"messaging_send failed: %s - ignoring\n",
nt_errstr(status));
}
return;
}
}
}
static void smbd_addr_changed(struct tevent_req *req)
{
struct smbd_addrchanged_state *state = tevent_req_callback_data(
req, struct smbd_addrchanged_state);
enum addrchange_type type;
struct sockaddr_storage addr;
NTSTATUS status;
uint32_t if_index;
bool match;
status = addrchange_recv(req, &type, &addr, &if_index);
TALLOC_FREE(req);
if (!NT_STATUS_IS_OK(status)) {
DBG_DEBUG("addrchange_recv failed: %s, stop listening\n",
nt_errstr(status));
TALLOC_FREE(state);
return;
}
match = interface_ifindex_exists_with_options(if_index,
IFACE_DYNAMIC_OPTION);
if (!match) {
DBG_NOTICE(
"smbd: No interface present for if_index %u "
"with dynamic option\n",
if_index);
goto rearm;
}
if (type == ADDRCHANGE_DEL) {
char addrstr[INET6_ADDRSTRLEN];
print_sockaddr(addrstr, sizeof(addrstr), &addr);
DBG_NOTICE("smbd: kernel (AF_NETLINK) dropped ip %s "
"on if_index %u\n",
addrstr, if_index);
smbd_close_socket_for_ip(state->parent, state->msg_ctx, &addr);
goto rearm;
}
if (type == ADDRCHANGE_ADD) {
char addrstr[INET6_ADDRSTRLEN];
print_sockaddr(addrstr, sizeof(addrstr), &addr);
DBG_NOTICE("smbd: kernel (AF_NETLINK) added ip %s "
"on if_index %u\n",
addrstr, if_index);
if (!smbd_open_socket_for_ip(state->parent,
state->ev,
state->msg_ctx,
state->ports,
&addr)) {
DBG_NOTICE("smbd: Unable to open socket on %s\n",
addrstr);
}
}
rearm:
req = addrchange_send(state, state->ev, state->ctx);
if (req == NULL) {
DBG_ERR("addrchange_send failed\n");
TALLOC_FREE(state);
return;
}
tevent_req_set_callback(req, smbd_addr_changed, state);
}
/****************************************************************************
main program.
****************************************************************************/
/* Declare prototype for build_options() to avoid having to run it through
mkproto.h. Mixing $(builddir) and $(srcdir) source files in the current
prototype generation system is too complicated. */
extern void build_options(bool screen);
int main(int argc,const char *argv[])
{
/* shall I run as a daemon */
struct samba_cmdline_daemon_cfg *cmdline_daemon_cfg = NULL;
bool log_stdout = false;
char *ports = NULL;
char *profile_level = NULL;
int opt;
poptContext pc;
struct server_id main_server_id = {0};
struct poptOption long_options[] = {
POPT_AUTOHELP
{
.longName = "build-options",
.shortName = 'b',
.argInfo = POPT_ARG_NONE,
.arg = NULL,
.val = 'b',
.descrip = "Print build options" ,
},
{
.longName = "port",
.shortName = 'p',
.argInfo = POPT_ARG_STRING,
.arg = &ports,
.val = 0,
.descrip = "Listen on the specified ports",
},
{
.longName = "profiling-level",
.shortName = 'P',
.argInfo = POPT_ARG_STRING,
.arg = &profile_level,
.val = 0,
.descrip = "Set profiling level","PROFILE_LEVEL",
},
POPT_COMMON_SAMBA
POPT_COMMON_DAEMON
POPT_COMMON_VERSION
POPT_TABLEEND
};
struct smbd_parent_context *parent = NULL;
TALLOC_CTX *frame;
NTSTATUS status;
struct tevent_context *ev_ctx;
struct messaging_context *msg_ctx;
struct server_id server_id;
struct tevent_signal *se;
int profiling_level;
char *np_dir = NULL;
const struct loadparm_substitution *lp_sub =
loadparm_s3_global_substitution();
static const struct smbd_shim smbd_shim_fns =
{
.change_to_root_user = smbd_change_to_root_user,
.become_authenticated_pipe_user = smbd_become_authenticated_pipe_user,
.unbecome_authenticated_pipe_user = smbd_unbecome_authenticated_pipe_user,
.contend_level2_oplocks_begin = smbd_contend_level2_oplocks_begin,
.contend_level2_oplocks_end = smbd_contend_level2_oplocks_end,
.become_root = smbd_become_root,
.unbecome_root = smbd_unbecome_root,
.exit_server = smbd_exit_server,
.exit_server_cleanly = smbd_exit_server_cleanly,
};
bool ok;
setproctitle_init(argc, discard_const(argv), environ);
/*
* Do this before any other talloc operation
*/
talloc_enable_null_tracking();
frame = talloc_stackframe();
smb_init_locale();
set_smbd_shim(&smbd_shim_fns);
smbd_init_globals();
TimeInit();
#ifdef HAVE_SET_AUTH_PARAMETERS
set_auth_parameters(argc,argv);
#endif
ok = samba_cmdline_init(frame,
SAMBA_CMDLINE_CONFIG_SERVER,
true /* require_smbconf */);
if (!ok) {
DBG_ERR("Failed to setup cmdline parser!\n");
exit(ENOMEM);
}
cmdline_daemon_cfg = samba_cmdline_get_daemon_cfg();
pc = samba_popt_get_context(getprogname(),
argc,
argv,
long_options,
0);
if (pc == NULL) {
DBG_ERR("Failed to get popt context!\n");
exit(ENOMEM);
}
while((opt = poptGetNextOpt(pc)) != -1) {
switch (opt) {
case 'b':
build_options(true); /* Display output to screen as well as debug */
exit(0);
break;
default:
d_fprintf(stderr, "\nInvalid option %s: %s\n\n",
poptBadOption(pc, 0), poptStrerror(opt));
poptPrintUsage(pc, stderr, 0);
exit(1);
}
}
poptFreeContext(pc);
log_stdout = (debug_get_log_type() == DEBUG_STDOUT);
if (cmdline_daemon_cfg->interactive) {
log_stdout = True;
}
#ifdef HAVE_SETLUID
/* needed for SecureWare on SCO */
setluid(0);
#endif
set_remote_machine_name("smbd", False);
if (cmdline_daemon_cfg->interactive && (DEBUGLEVEL >= 9)) {
talloc_enable_leak_report();
}
if (log_stdout && cmdline_daemon_cfg->fork) {
DEBUG(0,("ERROR: Can't log to stdout (-S) unless daemon is in foreground (-F) or interactive (-i)\n"));
exit(1);
}
/*
* We want to die early if we can't open /dev/urandom
*/
generate_random_buffer(NULL, 0);
/* get initial effective uid and gid */
sec_init();
/* make absolutely sure we run as root - to handle cases where people
are crazy enough to have it setuid */
gain_root_privilege();
gain_root_group_privilege();
dump_core_setup("smbd", lp_logfile(talloc_tos(), lp_sub));
/* we are never interested in SIGPIPE */
BlockSignals(True,SIGPIPE);
#if defined(SIGFPE)
/* we are never interested in SIGFPE */
BlockSignals(True,SIGFPE);
#endif
#if defined(SIGUSR2)
/* We are no longer interested in USR2 */
BlockSignals(True,SIGUSR2);
#endif
/*
* POSIX demands that signals are inherited. If the invoking
* process has these signals masked, we will have problems, as
* we won't receive them.
*/
BlockSignals(False, SIGHUP);
BlockSignals(False, SIGUSR1);
BlockSignals(False, SIGTERM);
/* Ensure we leave no zombies until we
* correctly set up child handling below. */
CatchChild();
/* we want total control over the permissions on created files,
so set our umask to 0 */
umask(0);
reopen_logs();
DBG_STARTUP_NOTICE("smbd version %s started.\n%s\n",
samba_version_string(),
samba_copyright_string());
DEBUG(2,("uid=%d gid=%d euid=%d egid=%d\n",
(int)getuid(),(int)getgid(),(int)geteuid(),(int)getegid()));
/* Output the build options to the debug log */
build_options(False);
if (sizeof(uint16_t) < 2 || sizeof(uint32_t) < 4) {
DEBUG(0,("ERROR: Samba is not configured correctly for the word size on your machine\n"));
exit(1);
}
/*
* This calls unshare(CLONE_FS); on linux
* in order to check if the running kernel/container
* environment supports it.
*/
per_thread_cwd_check();
if (!cluster_probe_ok()) {
exit(1);
}
/* Init the security context and global current_user */
init_sec_ctx();
/*
* Initialize the event context. The event context needs to be
* initialized before the messaging context, cause the messaging
* context holds an event context.
*/
ev_ctx = global_event_context();
if (ev_ctx == NULL) {
exit(1);
}
/*
* Init the messaging context
* FIXME: This should only call messaging_init()
*/
msg_ctx = global_messaging_context();
if (msg_ctx == NULL) {
exit(1);
}
/*
* Reloading of the printers will not work here as we don't have a
* server info and rpc services set up. It will be called later.
*/
if (!reload_services(NULL, NULL, false)) {
exit(1);
}
if (lp_server_role() == ROLE_ACTIVE_DIRECTORY_DC) {
if (!lp_parm_bool(-1, "server role check", "inhibit", false)) {
DBG_ERR("server role = 'active directory domain controller' not compatible with running smbd standalone. \n");
DEBUGADD(0, ("You should start 'samba' instead, and it will control starting smbd if required\n"));
exit(1);
}
/* Main 'samba' daemon will notify */
daemon_sd_notifications(false);
}
/* ...NOTE... Log files are working from this point! */
DEBUG(3,("loaded services\n"));
init_structs();
if (!profile_setup(msg_ctx, False)) {
DEBUG(0,("ERROR: failed to setup profiling\n"));
return -1;
}
if (profile_level != NULL) {
profiling_level = atoi(profile_level);
} else {
profiling_level = lp_smbd_profiling_level();
}
main_server_id = messaging_server_id(msg_ctx);
set_profile_level(profiling_level, &main_server_id);
if (!cmdline_daemon_cfg->daemon && !is_a_socket(0)) {
if (!cmdline_daemon_cfg->interactive) {
DEBUG(3, ("Standard input is not a socket, "
"assuming -D option\n"));
}
/*
* Setting "daemon" here prevents us from eventually calling
* the open_sockets_inetd()
*/
cmdline_daemon_cfg->daemon = true;
}
if (cmdline_daemon_cfg->daemon && !cmdline_daemon_cfg->interactive) {
DEBUG(3, ("Becoming a daemon.\n"));
become_daemon(cmdline_daemon_cfg->fork,
cmdline_daemon_cfg->no_process_group,
log_stdout);
} else {
daemon_status("smbd", "Starting process ...");
}
#ifdef HAVE_SETPGID
/*
* If we're interactive we want to set our own process group for
* signal management.
*/
if (cmdline_daemon_cfg->interactive &&
!cmdline_daemon_cfg->no_process_group)
{
setpgid( (pid_t)0, (pid_t)0);
}
#endif
if (!directory_exist(lp_lock_directory()))
mkdir(lp_lock_directory(), 0755);
if (!directory_exist(lp_pid_directory()))
mkdir(lp_pid_directory(), 0755);
if (cmdline_daemon_cfg->daemon)
pidfile_create(lp_pid_directory(), "smbd");
status = reinit_after_fork(msg_ctx, ev_ctx, false);
if (!NT_STATUS_IS_OK(status)) {
exit_daemon("reinit_after_fork() failed", map_errno_from_nt_status(status));
}
if (!cmdline_daemon_cfg->interactive) {
/*
* Do not initialize the parent-child-pipe before becoming a
* daemon: this is used to detect a died parent in the child
* process.
*/
status = init_before_fork();
if (!NT_STATUS_IS_OK(status)) {
exit_daemon(nt_errstr(status), map_errno_from_nt_status(status));
}
}
parent = talloc_zero(ev_ctx, struct smbd_parent_context);
if (!parent) {
exit_server("talloc(struct smbd_parent_context) failed");
}
parent->interactive = cmdline_daemon_cfg->interactive;
parent->ev_ctx = ev_ctx;
parent->msg_ctx = msg_ctx;
am_parent = parent;
se = tevent_add_signal(parent->ev_ctx,
parent,
SIGTERM, 0,
smbd_parent_sig_term_handler,
parent);
if (!se) {
exit_server("failed to setup SIGTERM handler");
}
se = tevent_add_signal(parent->ev_ctx,
parent,
SIGHUP, 0,
smbd_parent_sig_hup_handler,
parent);
if (!se) {
exit_server("failed to setup SIGHUP handler");
}
/* Setup all the TDB's - including CLEAR_IF_FIRST tdb's. */
if (smbd_memcache() == NULL) {
exit_daemon("no memcache available", EACCES);
}
memcache_set_global(smbd_memcache());
/* Initialise the password backed before the global_sam_sid
to ensure that we fetch from ldap before we make a domain sid up */
if(!initialize_password_db(false, ev_ctx))
exit(1);
if (!secrets_init()) {
exit_daemon("smbd can not open secrets.tdb", EACCES);
}
if (lp_server_role() == ROLE_DOMAIN_BDC || lp_server_role() == ROLE_DOMAIN_PDC || lp_server_role() == ROLE_IPA_DC) {
struct loadparm_context *lp_ctx = loadparm_init_s3(NULL, loadparm_s3_helpers());
if (!open_schannel_session_store(NULL, lp_ctx)) {
exit_daemon("ERROR: Samba cannot open schannel store for secured NETLOGON operations.", EACCES);
}
TALLOC_FREE(lp_ctx);
}
if(!get_global_sam_sid()) {
exit_daemon("Samba cannot create a SAM SID", EACCES);
}
server_id = messaging_server_id(msg_ctx);
status = smbXsrv_version_global_init(&server_id);
if (!NT_STATUS_IS_OK(status)) {
exit_daemon("Samba cannot init server context", EACCES);
}
status = smbXsrv_client_global_init();
if (!NT_STATUS_IS_OK(status)) {
exit_daemon("Samba cannot init clients context", EACCES);
}
status = smbXsrv_session_global_init(msg_ctx);
if (!NT_STATUS_IS_OK(status)) {
exit_daemon("Samba cannot init session context", EACCES);
}
status = smbXsrv_tcon_global_init();
if (!NT_STATUS_IS_OK(status)) {
exit_daemon("Samba cannot init tcon context", EACCES);
}
if (!locking_init())
exit_daemon("Samba cannot init locking", EACCES);
if (!leases_db_init(false)) {
exit_daemon("Samba cannot init leases", EACCES);
}
if (!smbd_notifyd_init(
msg_ctx,
cmdline_daemon_cfg->interactive,
&parent->notifyd)) {
exit_daemon("Samba cannot init notification", EACCES);
}
if (!cleanupd_init(
msg_ctx,
cmdline_daemon_cfg->interactive,
&parent->cleanupd)) {
exit_daemon("Samba cannot init the cleanupd", EACCES);
}
if (!messaging_parent_dgm_cleanup_init(msg_ctx)) {
exit(1);
}
if (!smbd_scavenger_init(NULL, msg_ctx, ev_ctx)) {
exit_daemon("Samba cannot init scavenging", EACCES);
}
if (!W_ERROR_IS_OK(registry_init_full()))
exit_daemon("Samba cannot init registry", EACCES);
/* Open the share_info.tdb here, so we don't have to open
after the fork on every single connection. This is a small
performance improvement and reduces the total number of system
fds used. */
status = share_info_db_init();
if (!NT_STATUS_IS_OK(status)) {
exit_daemon("ERROR: failed to load share info db.", EACCES);
}
status = init_system_session_info(NULL);
if (!NT_STATUS_IS_OK(status)) {
DEBUG(1, ("ERROR: failed to setup system user info: %s.\n",
nt_errstr(status)));
return -1;
}
if (!init_guest_session_info(NULL)) {
DEBUG(0,("ERROR: failed to setup guest info.\n"));
return -1;
}
if (!file_init_global()) {
DEBUG(0, ("ERROR: file_init_global() failed\n"));
return -1;
}
status = smbXsrv_open_global_init();
if (!NT_STATUS_IS_OK(status)) {
exit_daemon("Samba cannot init global open", map_errno_from_nt_status(status));
}
if (lp_clustering() && !lp_allow_unsafe_cluster_upgrade()) {
status = smbd_claim_version(msg_ctx, samba_version_string());
if (!NT_STATUS_IS_OK(status)) {
DBG_ERR("Could not claim version: %s\n",
nt_errstr(status));
return -1;
}
}
/* This MUST be done before start_epmd() because otherwise
* start_epmd() forks and races against dcesrv_ep_setup() to
* call directory_create_or_exist() */
if (!directory_create_or_exist(lp_ncalrpc_dir(), 0755)) {
DEBUG(0, ("Failed to create pipe directory %s - %s\n",
lp_ncalrpc_dir(), strerror(errno)));
return -1;
}
np_dir = talloc_asprintf(talloc_tos(), "%s/np", lp_ncalrpc_dir());
if (!np_dir) {
DEBUG(0, ("%s: Out of memory\n", __location__));
return -1;
}
if (!directory_create_or_exist_strict(np_dir, geteuid(), 0700)) {
DEBUG(0, ("Failed to create pipe directory %s - %s\n",
np_dir, strerror(errno)));
return -1;
}
if (!cmdline_daemon_cfg->interactive) {
daemon_ready("smbd");
}
if (!cmdline_daemon_cfg->daemon) {
int ret, sock;
/* inetd mode */
TALLOC_FREE(frame);
/* Started from inetd. fd 0 is the socket. */
/* We will abort gracefully when the client or remote system
goes away */
sock = dup(0);
/* close stdin, stdout (if not logging to it), but not stderr */
ret = close_low_fd(0);
if (ret != 0) {
DBG_ERR("close_low_fd(0) failed: %s\n", strerror(ret));
return 1;
}
if (!debug_get_output_is_stdout()) {
ret = close_low_fd(1);
if (ret != 0) {
DBG_ERR("close_low_fd(1) failed: %s\n",
strerror(ret));
return 1;
}
}
#ifdef HAVE_ATEXIT
atexit(killkids);
#endif
/* Stop zombies */
smbd_setup_sig_chld_handler(parent);
smbd_process(ev_ctx, msg_ctx, sock, true);
exit_server_cleanly(NULL);
return(0);
}
if (lp_interfaces() && lp_bind_interfaces_only()) {
smbd_init_addrchange(NULL, ev_ctx, msg_ctx, parent, ports);
}
if (!open_sockets_smbd(parent, ev_ctx, msg_ctx, ports))
exit_server("open_sockets_smbd() failed");
TALLOC_FREE(frame);
/* make sure we always have a valid stackframe */
frame = talloc_stackframe();
if (!cmdline_daemon_cfg->fork) {
/* if we are running in the foreground then look for
EOF on stdin, and exit if it happens. This allows
us to die if the parent process dies
Only do this on a pipe or socket, no other device.
*/
struct stat st;
if (fstat(0, &st) != 0) {
return 1;
}
if (S_ISFIFO(st.st_mode) || S_ISSOCK(st.st_mode)) {
tevent_add_fd(ev_ctx,
parent,
0,
TEVENT_FD_READ,
smbd_stdin_handler,
NULL);
}
}
smbd_parent_loop(ev_ctx, parent);
exit_server_cleanly(NULL);
TALLOC_FREE(frame);
return(0);
}