ctdb-scripts: Move database handling to its own event script

[samba4-gss.git] / source3 / winbindd / winbindd_dual.c

/*
   Unix SMB/CIFS implementation.

   Winbind child daemons

   Copyright (C) Andrew Tridgell 2002
   Copyright (C) Volker Lendecke 2004,2005

   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/>.
*/

/*
 * We fork a child per domain to be able to act non-blocking in the main
 * winbind daemon. A domain controller thousands of miles away being being
 * slow replying with a 10.000 user list should not hold up netlogon calls
 * that can be handled locally.
 */

#include "includes.h"
#include "winbindd.h"
#include "rpc_client/rpc_client.h"
#include "nsswitch/wb_reqtrans.h"
#include "secrets.h"
#include "../lib/util/select.h"
#include "winbindd_traceid.h"
#include "../libcli/security/security.h"
#include "system/select.h"
#include "messages.h"
#include "../lib/util/tevent_unix.h"
#include "lib/param/loadparm.h"
#include "lib/util/sys_rw.h"
#include "lib/util/sys_rw_data.h"
#include "passdb.h"
#include "lib/util/string_wrappers.h"
#include "lib/global_contexts.h"
#include "idmap.h"
#include "libcli/auth/netlogon_creds_cli.h"
#include "../lib/util/pidfile.h"
#include "librpc/gen_ndr/ndr_winbind_c.h"
#include "lib/util/util_process.h"

#undef DBGC_CLASS
#define DBGC_CLASS DBGC_WINBIND

static void forall_domain_children(bool (*fn)(struct winbindd_child *c,
                                              void *private_data),
                                   void *private_data)
{
        struct winbindd_domain *d;

        for (d = domain_list(); d != NULL; d = d->next) {
                int i;

                for (i = 0; i < talloc_array_length(d->children); i++) {
                        struct winbindd_child *c = &d->children[i];
                        bool ok;

                        if (c->pid == 0) {
                                continue;
                        }

                        ok = fn(c, private_data);
                        if (!ok) {
                                return;
                        }
                }
        }
}

static void forall_children(bool (*fn)(struct winbindd_child *c,
                                       void *private_data),
                            void *private_data)
{
        struct winbindd_child *c;
        bool ok;

        c = idmap_child();
        if (c->pid != 0) {
                ok = fn(c, private_data);
                if (!ok) {
                        return;
                }
        }

        c = locator_child();
        if (c->pid != 0) {
                ok = fn(c, private_data);
                if (!ok) {
                        return;
                }
        }

        forall_domain_children(fn, private_data);
}

/* Read some data from a client connection */

static NTSTATUS child_read_request(int sock, struct winbindd_request *wreq)
{
        NTSTATUS status;

        status = read_data_ntstatus(sock, (char *)wreq, sizeof(*wreq));
        if (!NT_STATUS_IS_OK(status)) {
                DEBUG(3, ("child_read_request: read_data failed: %s\n",
                          nt_errstr(status)));
                return status;
        }

        if (wreq->extra_len == 0) {
                wreq->extra_data.data = NULL;
                return NT_STATUS_OK;
        }

        DEBUG(10, ("Need to read %d extra bytes\n", (int)wreq->extra_len));

        wreq->extra_data.data = SMB_MALLOC_ARRAY(char, wreq->extra_len + 1);
        if (wreq->extra_data.data == NULL) {
                DEBUG(0, ("malloc failed\n"));
                return NT_STATUS_NO_MEMORY;
        }

        /* Ensure null termination */
        wreq->extra_data.data[wreq->extra_len] = '\0';

        status = read_data_ntstatus(sock, wreq->extra_data.data,
                                    wreq->extra_len);
        if (!NT_STATUS_IS_OK(status)) {
                DEBUG(0, ("Could not read extra data: %s\n",
                          nt_errstr(status)));
        }
        return status;
}

static NTSTATUS child_write_response(int sock, struct winbindd_response *wrsp)
{
        struct iovec iov[2];
        int iov_count;

        iov[0].iov_base = (void *)wrsp;
        iov[0].iov_len = sizeof(struct winbindd_response);
        iov_count = 1;

        if (wrsp->length > sizeof(struct winbindd_response)) {
                iov[1].iov_base = (void *)wrsp->extra_data.data;
                iov[1].iov_len = wrsp->length-iov[0].iov_len;
                iov_count = 2;
        }

        DEBUG(10, ("Writing %d bytes to parent\n", (int)wrsp->length));

        if (write_data_iov(sock, iov, iov_count) != wrsp->length) {
                DEBUG(0, ("Could not write result\n"));
                return NT_STATUS_INVALID_HANDLE;
        }

        return NT_STATUS_OK;
}

/*
 * Do winbind child async request. This is not simply wb_simple_trans. We have
 * to do the queueing ourselves because while a request is queued, the child
 * might have crashed, and we have to re-fork it in the _trigger function.
 */

struct wb_child_request_state {
        struct tevent_context *ev;
        struct tevent_req *queue_subreq;
        struct tevent_req *subreq;
        struct winbindd_child *child;
        struct winbindd_request *request;
        struct winbindd_response *response;
};

static bool fork_domain_child(struct winbindd_child *child);

static void wb_child_request_waited(struct tevent_req *subreq);
static void wb_child_request_done(struct tevent_req *subreq);
static void wb_child_request_orphaned(struct tevent_req *subreq);

static void wb_child_request_cleanup(struct tevent_req *req,
                                     enum tevent_req_state req_state);

struct tevent_req *wb_child_request_send(TALLOC_CTX *mem_ctx,
                                         struct tevent_context *ev,
                                         struct winbindd_child *child,
                                         struct winbindd_request *request)
{
        struct tevent_req *req;
        struct wb_child_request_state *state;
        struct tevent_req *subreq;

        req = tevent_req_create(mem_ctx, &state,
                                struct wb_child_request_state);
        if (req == NULL) {
                return NULL;
        }

        state->ev = ev;
        state->child = child;

        /*
         * We have to make a copy of "request", because our caller
         * might drop us via talloc_free().
         *
         * The talloc_move() magic in wb_child_request_cleanup() keeps
         * all the requests, but if we are sitting deep within
         * writev_send() down to the client, we have given it the
         * pointer to "request". As our caller lost interest, it will
         * just free "request", while writev_send still references it.
         */

        state->request = talloc_memdup(state, request, sizeof(*request));
        if (tevent_req_nomem(state->request, req)) {
                return tevent_req_post(req, ev);
        }

        state->request->traceid = debug_traceid_get();

        if (request->extra_data.data != NULL) {
                state->request->extra_data.data = talloc_memdup(
                        state->request,
                        request->extra_data.data,
                        request->extra_len);
                if (tevent_req_nomem(state->request->extra_data.data, req)) {
                        return tevent_req_post(req, ev);
                }
        }

        subreq = tevent_queue_wait_send(state, ev, child->queue);
        if (tevent_req_nomem(subreq, req)) {
                return tevent_req_post(req, ev);
        }
        tevent_req_set_callback(subreq, wb_child_request_waited, req);
        state->queue_subreq = subreq;

        tevent_req_set_cleanup_fn(req, wb_child_request_cleanup);

        return req;
}

static void wb_child_request_waited(struct tevent_req *subreq)
{
        struct tevent_req *req = tevent_req_callback_data(
                subreq, struct tevent_req);
        struct wb_child_request_state *state = tevent_req_data(
                req, struct wb_child_request_state);
        bool ok;

        ok = tevent_queue_wait_recv(subreq);
        if (!ok) {
                tevent_req_oom(req);
                return;
        }
        /*
         * We need to keep state->queue_subreq
         * in order to block the queue.
         */
        subreq = NULL;

        if ((state->child->sock == -1) && (!fork_domain_child(state->child))) {
                tevent_req_error(req, errno);
                return;
        }

        tevent_fd_set_flags(state->child->monitor_fde, 0);

        subreq = wb_simple_trans_send(state, global_event_context(), NULL,
                                      state->child->sock, state->request);
        if (tevent_req_nomem(subreq, req)) {
                return;
        }

        state->subreq = subreq;
        tevent_req_set_callback(subreq, wb_child_request_done, req);
        tevent_req_set_endtime(req, state->ev, timeval_current_ofs(300, 0));
}

static void wb_child_request_done(struct tevent_req *subreq)
{
        struct tevent_req *req = tevent_req_callback_data(
                subreq, struct tevent_req);
        struct wb_child_request_state *state = tevent_req_data(
                req, struct wb_child_request_state);
        int ret, err;

        ret = wb_simple_trans_recv(subreq, state, &state->response, &err);
        /* Freeing the subrequest is deferred until the cleanup function,
         * which has to know whether a subrequest exists, and consequently
         * decide whether to shut down the pipe to the child process.
         */
        if (ret == -1) {
                tevent_req_error(req, err);
                return;
        }
        tevent_req_done(req);
}

static void wb_child_request_orphaned(struct tevent_req *subreq)
{
        struct winbindd_child *child =
                (struct winbindd_child *)tevent_req_callback_data_void(subreq);

        DBG_WARNING("cleanup orphaned subreq[%p]\n", subreq);
        TALLOC_FREE(subreq);

        if (child->domain != NULL) {
                /*
                 * If the child is attached to a domain,
                 * we need to make sure the domain queue
                 * can move forward, after the orphaned
                 * request is done.
                 */
                tevent_queue_start(child->domain->queue);
        }
}

int wb_child_request_recv(struct tevent_req *req, TALLOC_CTX *mem_ctx,
                          struct winbindd_response **presponse, int *err)
{
        struct wb_child_request_state *state = tevent_req_data(
                req, struct wb_child_request_state);

        if (tevent_req_is_unix_error(req, err)) {
                return -1;
        }
        *presponse = talloc_move(mem_ctx, &state->response);
        return 0;
}

static void wb_child_request_cleanup(struct tevent_req *req,
                                     enum tevent_req_state req_state)
{
        struct wb_child_request_state *state =
            tevent_req_data(req, struct wb_child_request_state);

        if (state->subreq == NULL) {
                /* nothing to cleanup */
                return;
        }

        if (req_state == TEVENT_REQ_RECEIVED) {
                struct tevent_req *subreq = NULL;

                /*
                 * Our caller gave up, but we need to keep
                 * the low level request (wb_simple_trans)
                 * in order to maintain the parent child protocol.
                 *
                 * We also need to keep the child queue blocked
                 * until we got the response from the child.
                 */

                subreq = talloc_move(state->child->queue, &state->subreq);
                talloc_move(subreq, &state->queue_subreq);
                talloc_move(subreq, &state->request);
                tevent_req_set_callback(subreq,
                                        wb_child_request_orphaned,
                                        state->child);

                DBG_WARNING("keep orphaned subreq[%p]\n", subreq);
                return;
        }

        TALLOC_FREE(state->subreq);
        TALLOC_FREE(state->queue_subreq);

        tevent_fd_set_flags(state->child->monitor_fde, TEVENT_FD_READ);

        if (state->child->domain != NULL) {
                /*
                 * If the child is attached to a domain,
                 * we need to make sure the domain queue
                 * can move forward, after the request
                 * is done.
                 */
                tevent_queue_start(state->child->domain->queue);
        }

        if (req_state == TEVENT_REQ_DONE) {
                /* transmitted request and got response */
                return;
        }

        /*
         * Failed to transmit and receive response, or request
         * cancelled while being serviced.
         * The basic parent/child communication broke, close
         * our socket
         */
        TALLOC_FREE(state->child->monitor_fde);
        close(state->child->sock);
        state->child->sock = -1;
}

static void child_socket_readable(struct tevent_context *ev,
                                  struct tevent_fd *fde,
                                  uint16_t flags,
                                  void *private_data)
{
        struct winbindd_child *child = private_data;

        if ((flags & TEVENT_FD_READ) == 0) {
                return;
        }

        TALLOC_FREE(child->monitor_fde);

        /*
         * We're only active when there is no outstanding child
         * request. Arriving here means the child closed its socket,
         * it died. Do the same here.
         */

        SMB_ASSERT(child->sock != -1);

        close(child->sock);
        child->sock = -1;
}

static struct winbindd_child *choose_domain_child(struct winbindd_domain *domain)
{
        struct winbindd_child *shortest = &domain->children[0];
        struct winbindd_child *current;
        int i;

        for (i=0; i<talloc_array_length(domain->children); i++) {
                size_t shortest_len, current_len;

                current = &domain->children[i];
                current_len = tevent_queue_length(current->queue);

                if (current_len == 0) {
                        /* idle child */
                        return current;
                }

                shortest_len = tevent_queue_length(shortest->queue);

                if (current_len < shortest_len) {
                        shortest = current;
                }
        }

        return shortest;
}

struct dcerpc_binding_handle *dom_child_handle(struct winbindd_domain *domain)
{
        return domain->binding_handle;
}

struct wb_domain_request_state {
        struct tevent_context *ev;
        struct tevent_queue_entry *queue_entry;
        struct winbindd_domain *domain;
        struct winbindd_child *child;
        struct winbindd_request *request;
        struct winbindd_request *init_req;
        struct winbindd_response *response;
        struct tevent_req *pending_subreq;
        struct wbint_InitConnection r;
};

static void wb_domain_request_cleanup(struct tevent_req *req,
                                      enum tevent_req_state req_state)
{
        struct wb_domain_request_state *state = tevent_req_data(
                req, struct wb_domain_request_state);

        /*
         * If we're completely done or got a failure.
         * we should remove ourself from the domain queue,
         * after removing the child subreq from the child queue
         * and give the next one in the queue the chance
         * to check for an idle child.
         */
        TALLOC_FREE(state->pending_subreq);
        TALLOC_FREE(state->queue_entry);
        tevent_queue_start(state->domain->queue);
}

static void wb_domain_request_trigger(struct tevent_req *req,
                                      void *private_data);
static void wb_domain_request_gotdc(struct tevent_req *subreq);
static void wb_domain_request_initialized(struct tevent_req *subreq);
static void wb_domain_request_done(struct tevent_req *subreq);

struct tevent_req *wb_domain_request_send(TALLOC_CTX *mem_ctx,
                                          struct tevent_context *ev,
                                          struct winbindd_domain *domain,
                                          struct winbindd_request *request)
{
        struct tevent_req *req;
        struct wb_domain_request_state *state;

        req = tevent_req_create(mem_ctx, &state,
                                struct wb_domain_request_state);
        if (req == NULL) {
                return NULL;
        }

        state->domain = domain;
        state->ev = ev;
        state->request = request;

        tevent_req_set_cleanup_fn(req, wb_domain_request_cleanup);

        state->queue_entry = tevent_queue_add_entry(
                        domain->queue, state->ev, req,
                        wb_domain_request_trigger, NULL);
        if (tevent_req_nomem(state->queue_entry, req)) {
                return tevent_req_post(req, ev);
        }

        return req;
}

static void wb_domain_request_trigger(struct tevent_req *req,
                                      void *private_data)
{
        struct wb_domain_request_state *state = tevent_req_data(
                req, struct wb_domain_request_state);
        struct winbindd_domain *domain = state->domain;
        struct tevent_req *subreq = NULL;
        size_t shortest_queue_length;

        state->child = choose_domain_child(domain);
        shortest_queue_length = tevent_queue_length(state->child->queue);
        if (shortest_queue_length > 0) {
                /*
                 * All children are busy, we need to stop
                 * the queue and untrigger our own queue
                 * entry. Once a pending request
                 * is done it calls tevent_queue_start
                 * and we get retriggered.
                 */
                state->child = NULL;
                tevent_queue_stop(state->domain->queue);
                tevent_queue_entry_untrigger(state->queue_entry);
                return;
        }

        if (domain->initialized) {
                subreq = wb_child_request_send(state, state->ev, state->child,
                                               state->request);
                if (tevent_req_nomem(subreq, req)) {
                        return;
                }
                tevent_req_set_callback(subreq, wb_domain_request_done, req);
                state->pending_subreq = subreq;

                /*
                 * Once the domain is initialized and
                 * once we placed our real request into the child queue,
                 * we can remove ourself from the domain queue
                 * and give the next one in the queue the chance
                 * to check for an idle child.
                 */
                TALLOC_FREE(state->queue_entry);
                return;
        }

        state->init_req = talloc_zero(state, struct winbindd_request);
        if (tevent_req_nomem(state->init_req, req)) {
                return;
        }

        if (IS_DC || domain->primary || domain->internal) {
                /* The primary domain has to find the DC name itself */
                state->r.in.dcname = talloc_strdup(state, "");
                if (tevent_req_nomem(state->r.in.dcname, req)) {
                        return;
                }

                subreq = dcerpc_wbint_InitConnection_r_send(state,
                                                state->ev,
                                                state->child->binding_handle,
                                                &state->r);
                if (tevent_req_nomem(subreq, req)) {
                        return;
                }
                tevent_req_set_callback(subreq, wb_domain_request_initialized,
                                        req);
                state->pending_subreq = subreq;
                return;
        }

        /*
         * This is *not* the primary domain,
         * let's ask our DC about a DC name.
         *
         * We prefer getting a dns name in dc_unc,
         * which is indicated by DS_RETURN_DNS_NAME.
         * For NT4 domains we still get the netbios name.
         */
        subreq = wb_dsgetdcname_send(state, state->ev,
                                     state->domain->name,
                                     NULL, /* domain_guid */
                                     NULL, /* site_name */
                                     DS_RETURN_DNS_NAME); /* flags */
        if (tevent_req_nomem(subreq, req)) {
                return;
        }
        tevent_req_set_callback(subreq, wb_domain_request_gotdc, req);
        state->pending_subreq = subreq;
        return;
}

static void wb_domain_request_gotdc(struct tevent_req *subreq)
{
        struct tevent_req *req = tevent_req_callback_data(
                subreq, struct tevent_req);
        struct wb_domain_request_state *state = tevent_req_data(
                req, struct wb_domain_request_state);
        struct netr_DsRGetDCNameInfo *dcinfo = NULL;
        NTSTATUS status;
        const char *dcname = NULL;

        state->pending_subreq = NULL;

        status = wb_dsgetdcname_recv(subreq, state, &dcinfo);
        TALLOC_FREE(subreq);
        if (tevent_req_nterror(req, status)) {
                return;
        }
        dcname = dcinfo->dc_unc;
        while (dcname != NULL && *dcname == '\\') {
                dcname++;
        }

        state->r.in.dcname = talloc_strdup(state, dcname);
        if (tevent_req_nomem(state->r.in.dcname, req)) {
                return;
        }

        TALLOC_FREE(dcinfo);

        subreq = dcerpc_wbint_InitConnection_r_send(state,
                                                state->ev,
                                                state->child->binding_handle,
                                                &state->r);
        if (tevent_req_nomem(subreq, req)) {
                return;
        }
        tevent_req_set_callback(subreq, wb_domain_request_initialized, req);
        state->pending_subreq = subreq;
}

static void wb_domain_request_initialized(struct tevent_req *subreq)
{
        struct tevent_req *req = tevent_req_callback_data(
                subreq, struct tevent_req);
        struct wb_domain_request_state *state = tevent_req_data(
                req, struct wb_domain_request_state);
        NTSTATUS status;

        state->pending_subreq = NULL;

        status = dcerpc_wbint_InitConnection_r_recv(subreq, state);
        TALLOC_FREE(subreq);
        if (NT_STATUS_IS_ERR(status)) {
                tevent_req_error(req, map_errno_from_nt_status(status));
                return;
        }

        status = state->r.out.result;
        if (NT_STATUS_IS_ERR(status)) {
                tevent_req_error(req, map_errno_from_nt_status(status));
                return;
        }

        state->domain->sid = *state->r.out.sid;

        talloc_free(state->domain->name);
        state->domain->name = talloc_strdup(state->domain, *state->r.out.name);
        if (state->domain->name == NULL) {
                tevent_req_error(req, ENOMEM);
                return;
        }

        if (*state->r.out.alt_name != NULL &&
            strlen(*state->r.out.alt_name) > 0) {
                talloc_free(state->domain->alt_name);

                state->domain->alt_name = talloc_strdup(state->domain,
                                                        *state->r.out.alt_name);
                if (state->domain->alt_name == NULL) {
                        tevent_req_error(req, ENOMEM);
                        return;
                }
        }

        state->domain->native_mode =
                        (*state->r.out.flags & WB_DOMINFO_DOMAIN_NATIVE);
        state->domain->active_directory =
                        (*state->r.out.flags & WB_DOMINFO_DOMAIN_AD);
        state->domain->initialized = true;

        subreq = wb_child_request_send(state, state->ev, state->child,
                                       state->request);
        if (tevent_req_nomem(subreq, req)) {
                return;
        }
        tevent_req_set_callback(subreq, wb_domain_request_done, req);
        state->pending_subreq = subreq;

        /*
         * Once the domain is initialized and
         * once we placed our real request into the child queue,
         * we can remove ourself from the domain queue
         * and give the next one in the queue the chance
         * to check for an idle child.
         */
        TALLOC_FREE(state->queue_entry);
}

static void wb_domain_request_done(struct tevent_req *subreq)
{
        struct tevent_req *req = tevent_req_callback_data(
                subreq, struct tevent_req);
        struct wb_domain_request_state *state = tevent_req_data(
                req, struct wb_domain_request_state);
        int ret, err;

        state->pending_subreq = NULL;

        ret = wb_child_request_recv(subreq, talloc_tos(), &state->response,
                                    &err);
        TALLOC_FREE(subreq);
        if (ret == -1) {
                tevent_req_error(req, err);
                return;
        }
        tevent_req_done(req);
}

int wb_domain_request_recv(struct tevent_req *req, TALLOC_CTX *mem_ctx,
                           struct winbindd_response **presponse, int *err)
{
        struct wb_domain_request_state *state = tevent_req_data(
                req, struct wb_domain_request_state);

        if (tevent_req_is_unix_error(req, err)) {
                return -1;
        }
        *presponse = talloc_move(mem_ctx, &state->response);
        return 0;
}

static void child_process_request(struct winbindd_child *child,
                                  struct winbindd_cli_state *state)
{
        struct winbindd_domain *domain = child->domain;

        /* Free response data - we may be interrupted and receive another
           command before being able to send this data off. */

        state->response->result = WINBINDD_ERROR;
        state->response->length = sizeof(struct winbindd_response);

        /* as all requests in the child are sync, we can use talloc_tos() */
        state->mem_ctx = talloc_tos();

        /* Process command */
        state->response->result = winbindd_dual_ndrcmd(domain, state);
}

void setup_child(struct winbindd_domain *domain, struct winbindd_child *child,
                 const char *logprefix,
                 const char *logname)
{
        const struct loadparm_substitution *lp_sub =
                loadparm_s3_global_substitution();

        if (logprefix && logname) {
                char *logbase = NULL;

                if (*lp_logfile(talloc_tos(), lp_sub)) {
                        char *end = NULL;

                        if (asprintf(&logbase, "%s", lp_logfile(talloc_tos(), lp_sub)) < 0) {
                                smb_panic("Internal error: asprintf failed");
                        }

                        if ((end = strrchr_m(logbase, '/'))) {
                                *end = '\0';
                        }
                } else {
                        if (asprintf(&logbase, "%s", get_dyn_LOGFILEBASE()) < 0) {
                                smb_panic("Internal error: asprintf failed");
                        }
                }

                if (asprintf(&child->logfilename, "%s/%s-%s",
                             logbase, logprefix, logname) < 0) {
                        SAFE_FREE(logbase);
                        smb_panic("Internal error: asprintf failed");
                }

                SAFE_FREE(logbase);
        } else {
                smb_panic("Internal error: logprefix == NULL && "
                          "logname == NULL");
        }

        child->pid = 0;
        child->sock = -1;
        child->domain = domain;
        child->queue = tevent_queue_create(NULL, "winbind_child");
        SMB_ASSERT(child->queue != NULL);

        child->binding_handle = wbint_binding_handle(NULL, NULL, child);
        SMB_ASSERT(child->binding_handle != NULL);
}

struct winbind_child_died_state {
        pid_t pid;
        struct winbindd_child *child;
};

static bool winbind_child_died_fn(struct winbindd_child *child,
                                  void *private_data)
{
        struct winbind_child_died_state *state = private_data;

        if (child->pid == state->pid) {
                state->child = child;
                return false;
        }
        return true;
}

void winbind_child_died(pid_t pid)
{
        struct winbind_child_died_state state = { .pid = pid };

        forall_children(winbind_child_died_fn, &state);

        if (state.child == NULL) {
                DEBUG(5, ("Already reaped child %u died\n", (unsigned int)pid));
                return;
        }

        state.child->pid = 0;
}

/* Ensure any negative cache entries with the netbios or realm names are removed. */

void winbindd_flush_negative_conn_cache(struct winbindd_domain *domain)
{
        flush_negative_conn_cache_for_domain(domain->name);
        if (domain->alt_name != NULL) {
                flush_negative_conn_cache_for_domain(domain->alt_name);
        }
}

/*
 * Parent winbindd process sets its own debug level first and then
 * sends a message to all the winbindd children to adjust their debug
 * level to that of parents.
 */

struct winbind_msg_relay_state {
        struct messaging_context *msg_ctx;
        uint32_t msg_type;
        DATA_BLOB *data;
};

static bool winbind_msg_relay_fn(struct winbindd_child *child,
                                 void *private_data)
{
        struct winbind_msg_relay_state *state = private_data;

        DBG_DEBUG("sending message to pid %u.\n",
                  (unsigned int)child->pid);

        messaging_send(state->msg_ctx, pid_to_procid(child->pid),
                       state->msg_type, state->data);
        return true;
}

void winbind_msg_debug(struct messaging_context *msg_ctx,
                         void *private_data,
                         uint32_t msg_type,
                         struct server_id server_id,
                         DATA_BLOB *data)
{
        struct winbind_msg_relay_state state = {
                .msg_ctx = msg_ctx, .msg_type = msg_type, .data = data
        };

        DEBUG(10,("winbind_msg_debug: got debug message.\n"));

        debug_message(msg_ctx, private_data, MSG_DEBUG, server_id, data);

        forall_children(winbind_msg_relay_fn, &state);
}

void winbind_disconnect_dc_parent(struct messaging_context *msg_ctx,
                                  void *private_data,
                                  uint32_t msg_type,
                                  struct server_id server_id,
                                  DATA_BLOB *data)
{
        struct winbind_msg_relay_state state = {
                .msg_ctx = msg_ctx, .msg_type = msg_type, .data = data
        };

        DBG_DEBUG("Got disconnect_dc message\n");

        forall_children(winbind_msg_relay_fn, &state);
}

static bool winbindd_child_msg_filter(struct messaging_rec *rec,
                                      void *private_data)
{
        struct winbindd_child *child = talloc_get_type_abort(private_data,
                        struct winbindd_child);

        if (rec->msg_type == MSG_SMB_CONF_UPDATED) {
                DBG_DEBUG("Got reload-config message\n");
                winbindd_reload_services_file(child->logfilename);
        }

        return false;
}

/* React on 'smbcontrol winbindd reload-config' in the same way as on SIGHUP*/
void winbindd_msg_reload_services_parent(struct messaging_context *msg,
                                         void *private_data,
                                         uint32_t msg_type,
                                         struct server_id server_id,
                                         DATA_BLOB *data)
{
        struct winbind_msg_relay_state state = {
                .msg_ctx = msg,
                .msg_type = msg_type,
                .data = data,
        };
        bool ok;

        DBG_DEBUG("Got reload-config message\n");

        /* Flush various caches */
        winbindd_flush_caches();

        winbindd_reload_services_file((const char *)private_data);

        /* Set tevent_thread_call_depth_set_callback according to debug level */
        if (lp_winbind_debug_traceid() && debuglevel_get() > 1) {
                tevent_thread_call_depth_set_callback(winbind_call_flow, NULL);
        } else {
                tevent_thread_call_depth_set_callback(NULL, NULL);
        }

        ok = add_trusted_domains_dc();
        if (!ok) {
                DBG_ERR("add_trusted_domains_dc() failed\n");
        }

        forall_children(winbind_msg_relay_fn, &state);
}

/* Set our domains as offline and forward the offline message to our children. */

struct winbind_msg_on_offline_state {
        struct messaging_context *msg_ctx;
        uint32_t msg_type;
};

static bool winbind_msg_on_offline_fn(struct winbindd_child *child,
                                      void *private_data)
{
        struct winbind_msg_on_offline_state *state = private_data;

        if (child->domain->internal) {
                return true;
        }

        /*
         * Each winbindd child should only process requests for one
         * domain - make sure we only set it online / offline for that
         * domain.
         */
        DBG_DEBUG("sending message to pid %u for domain %s.\n",
                  (unsigned int)child->pid, child->domain->name);

        messaging_send_buf(state->msg_ctx,
                           pid_to_procid(child->pid),
                           state->msg_type,
                           (const uint8_t *)child->domain->name,
                           strlen(child->domain->name)+1);

        return true;
}

void winbind_msg_offline(struct messaging_context *msg_ctx,
                         void *private_data,
                         uint32_t msg_type,
                         struct server_id server_id,
                         DATA_BLOB *data)
{
        struct winbind_msg_on_offline_state state = {
                .msg_ctx = msg_ctx,
                .msg_type = MSG_WINBIND_OFFLINE,
        };
        struct winbindd_domain *domain;

        DEBUG(10,("winbind_msg_offline: got offline message.\n"));

        if (!lp_winbind_offline_logon()) {
                DEBUG(10,("winbind_msg_offline: rejecting offline message.\n"));
                return;
        }

        /* Set our global state as offline. */
        if (!set_global_winbindd_state_offline()) {
                DEBUG(10,("winbind_msg_offline: offline request failed.\n"));
                return;
        }

        /* Set all our domains as offline. */
        for (domain = domain_list(); domain; domain = domain->next) {
                if (domain->internal) {
                        continue;
                }
                DEBUG(5,("winbind_msg_offline: marking %s offline.\n", domain->name));
                domain->online = false;
        }

        forall_domain_children(winbind_msg_on_offline_fn, &state);
}

/* Set our domains as online and forward the online message to our children. */

void winbind_msg_online(struct messaging_context *msg_ctx,
                        void *private_data,
                        uint32_t msg_type,
                        struct server_id server_id,
                        DATA_BLOB *data)
{
        struct winbind_msg_on_offline_state state = {
                .msg_ctx = msg_ctx,
                .msg_type = MSG_WINBIND_ONLINE,
        };

        DEBUG(10,("winbind_msg_online: got online message.\n"));

        if (!lp_winbind_offline_logon()) {
                DEBUG(10,("winbind_msg_online: rejecting online message.\n"));
                return;
        }

        /* Set our global state as online. */
        set_global_winbindd_state_online();

        smb_nscd_flush_user_cache();
        smb_nscd_flush_group_cache();

        /* Tell all our child domains to go online online. */
        forall_domain_children(winbind_msg_on_offline_fn, &state);
}

static const char *collect_onlinestatus(TALLOC_CTX *mem_ctx)
{
        struct winbindd_domain *domain;
        char *buf = NULL;

        if ((buf = talloc_asprintf(mem_ctx, "global:%s ",
                                   get_global_winbindd_state_offline() ?
                                   "Offline":"Online")) == NULL) {
                return NULL;
        }

        for (domain = domain_list(); domain; domain = domain->next) {
                if ((buf = talloc_asprintf_append_buffer(buf, "%s:%s ",
                                                  domain->name,
                                                  domain->online ?
                                                  "Online":"Offline")) == NULL) {
                        return NULL;
                }
        }

        buf = talloc_asprintf_append_buffer(buf, "\n");

        DEBUG(5,("collect_onlinestatus: %s", buf));

        return buf;
}

void winbind_msg_onlinestatus(struct messaging_context *msg_ctx,
                              void *private_data,
                              uint32_t msg_type,
                              struct server_id server_id,
                              DATA_BLOB *data)
{
        TALLOC_CTX *mem_ctx;
        const char *message;

        DEBUG(5,("winbind_msg_onlinestatus received.\n"));

        mem_ctx = talloc_init("winbind_msg_onlinestatus");
        if (mem_ctx == NULL) {
                return;
        }

        message = collect_onlinestatus(mem_ctx);
        if (message == NULL) {
                talloc_destroy(mem_ctx);
                return;
        }

        messaging_send_buf(msg_ctx, server_id, MSG_WINBIND_ONLINESTATUS,
                           (const uint8_t *)message, strlen(message) + 1);

        talloc_destroy(mem_ctx);
}

void winbind_msg_dump_domain_list(struct messaging_context *msg_ctx,
                                  void *private_data,
                                  uint32_t msg_type,
                                  struct server_id server_id,
                                  DATA_BLOB *data)
{
        TALLOC_CTX *mem_ctx;
        const char *message = NULL;
        const char *domain = NULL;
        char *s = NULL;
        NTSTATUS status;
        struct winbindd_domain *dom = NULL;

        DEBUG(5,("winbind_msg_dump_domain_list received.\n"));

        mem_ctx = talloc_init("winbind_msg_dump_domain_list");
        if (!mem_ctx) {
                return;
        }

        if (data->length > 0) {
                domain = (const char *)data->data;
        }

        if (domain) {

                DEBUG(5,("winbind_msg_dump_domain_list for domain: %s\n",
                        domain));

                message = NDR_PRINT_STRUCT_STRING(mem_ctx, winbindd_domain,
                                                  find_domain_from_name_noinit(domain));
                if (!message) {
                        talloc_destroy(mem_ctx);
                        return;
                }

                messaging_send_buf(msg_ctx, server_id,
                                   MSG_WINBIND_DUMP_DOMAIN_LIST,
                                   (const uint8_t *)message, strlen(message) + 1);

                talloc_destroy(mem_ctx);

                return;
        }

        DEBUG(5,("winbind_msg_dump_domain_list all domains\n"));

        for (dom = domain_list(); dom; dom=dom->next) {
                message = NDR_PRINT_STRUCT_STRING(mem_ctx, winbindd_domain, dom);
                if (!message) {
                        talloc_destroy(mem_ctx);
                        return;
                }

                s = talloc_asprintf_append(s, "%s\n", message);
                if (!s) {
                        talloc_destroy(mem_ctx);
                        return;
                }
        }

        status = messaging_send_buf(msg_ctx, server_id,
                                    MSG_WINBIND_DUMP_DOMAIN_LIST,
                                    (uint8_t *)s, strlen(s) + 1);
        if (!NT_STATUS_IS_OK(status)) {
                DEBUG(0,("failed to send message: %s\n",
                nt_errstr(status)));
        }

        talloc_destroy(mem_ctx);
}

static void account_lockout_policy_handler(struct tevent_context *ctx,
                                           struct tevent_timer *te,
                                           struct timeval now,
                                           void *private_data)
{
        struct winbindd_child *child =
                (struct winbindd_child *)private_data;
        TALLOC_CTX *mem_ctx = NULL;
        struct samr_DomInfo12 lockout_policy;
        NTSTATUS result;

        DEBUG(10,("account_lockout_policy_handler called\n"));

        TALLOC_FREE(child->lockout_policy_event);

        if ( !winbindd_can_contact_domain( child->domain ) ) {
                DEBUG(10,("account_lockout_policy_handler: Removing myself since I "
                          "do not have an incoming trust to domain %s\n",
                          child->domain->name));

                return;
        }

        mem_ctx = talloc_init("account_lockout_policy_handler ctx");
        if (!mem_ctx) {
                result = NT_STATUS_NO_MEMORY;
        } else {
                result = wb_cache_lockout_policy(child->domain, mem_ctx,
                                                 &lockout_policy);
        }
        TALLOC_FREE(mem_ctx);

        if (!NT_STATUS_IS_OK(result)) {
                DEBUG(10,("account_lockout_policy_handler: lockout_policy failed error %s\n",
                         nt_errstr(result)));
        }

        child->lockout_policy_event = tevent_add_timer(global_event_context(), NULL,
                                                      timeval_current_ofs(3600, 0),
                                                      account_lockout_policy_handler,
                                                      child);
}

static time_t get_machine_password_timeout(void)
{
        /* until we have gpo support use lp setting */
        return lp_machine_password_timeout();
}

static bool calculate_next_machine_pwd_change(const char *domain,
                                              struct timeval *t)
{
        time_t pass_last_set_time;
        time_t timeout;
        time_t next_change;
        struct timeval tv;
        char *pw;

        pw = secrets_fetch_machine_password(domain,
                                            &pass_last_set_time,
                                            NULL);

        if (pw == NULL) {
                DEBUG(0,("cannot fetch own machine password ????\n"));
                return false;
        }

        SAFE_FREE(pw);

        timeout = get_machine_password_timeout();
        if (timeout == 0) {
                DEBUG(10,("machine password never expires\n"));
                return false;
        }

        tv.tv_sec = pass_last_set_time;
        DEBUG(10, ("password last changed %s\n",
                   timeval_string(talloc_tos(), &tv, false)));
        tv.tv_sec += timeout;
        DEBUGADD(10, ("password valid until %s\n",
                      timeval_string(talloc_tos(), &tv, false)));

        if (time(NULL) < (pass_last_set_time + timeout)) {
                next_change = pass_last_set_time + timeout;
                DEBUG(10,("machine password still valid until: %s\n",
                        http_timestring(talloc_tos(), next_change)));
                *t = tevent_timeval_set(next_change, 0);

                if (lp_clustering()) {
                        uint8_t randbuf;
                        /*
                         * When having a cluster, we have several
                         * winbinds racing for the password change. In
                         * the machine_password_change_handler()
                         * function we check if someone else was
                         * faster when the event triggers. We add a
                         * 255-second random delay here, so that we
                         * don't run to change the password at the
                         * exact same moment.
                         */
                        generate_random_buffer(&randbuf, sizeof(randbuf));
                        DEBUG(10, ("adding %d seconds randomness\n",
                                   (int)randbuf));
                        t->tv_sec += randbuf;
                }
                return true;
        }

        DEBUG(10,("machine password expired, needs immediate change\n"));

        *t = timeval_zero();

        return true;
}

static void machine_password_change_handler(struct tevent_context *ctx,
                                            struct tevent_timer *te,
                                            struct timeval now,
                                            void *private_data)
{
        struct messaging_context *msg_ctx = global_messaging_context();
        struct winbindd_child *child =
                (struct winbindd_child *)private_data;
        struct rpc_pipe_client *netlogon_pipe = NULL;
        struct netlogon_creds_cli_context *netlogon_creds_ctx = NULL;
        NTSTATUS result;
        struct timeval next_change;

        DEBUG(10,("machine_password_change_handler called\n"));

        TALLOC_FREE(child->machine_password_change_event);

        if (!calculate_next_machine_pwd_change(child->domain->name,
                                               &next_change)) {
                DEBUG(10, ("calculate_next_machine_pwd_change failed\n"));
                return;
        }

        DEBUG(10, ("calculate_next_machine_pwd_change returned %s\n",
                   timeval_string(talloc_tos(), &next_change, false)));

        if (!timeval_expired(&next_change)) {
                DEBUG(10, ("Someone else has already changed the pw\n"));
                goto done;
        }

        if (!winbindd_can_contact_domain(child->domain)) {
                DEBUG(10,("machine_password_change_handler: Removing myself since I "
                          "do not have an incoming trust to domain %s\n",
                          child->domain->name));
                return;
        }

        result = cm_connect_netlogon_secure(child->domain,
                                            &netlogon_pipe,
                                            &netlogon_creds_ctx);
        if (!NT_STATUS_IS_OK(result)) {
                DEBUG(10,("machine_password_change_handler: "
                        "failed to connect netlogon pipe: %s\n",
                         nt_errstr(result)));
                return;
        }

        result = trust_pw_change(netlogon_creds_ctx,
                                 msg_ctx,
                                 netlogon_pipe->binding_handle,
                                 child->domain->name,
                                 child->domain->dcname,
                                 false); /* force */

        DEBUG(10, ("machine_password_change_handler: "
                   "trust_pw_change returned %s\n",
                   nt_errstr(result)));

        if (NT_STATUS_EQUAL(result, NT_STATUS_ACCESS_DENIED) ) {
                DEBUG(3,("machine_password_change_handler: password set returned "
                         "ACCESS_DENIED.  Maybe the trust account "
                         "password was changed and we didn't know it. "
                         "Killing connections to domain %s\n",
                         child->domain->name));
                invalidate_cm_connection(child->domain);
        }

        if (!calculate_next_machine_pwd_change(child->domain->name,
                                               &next_change)) {
                DEBUG(10, ("calculate_next_machine_pwd_change failed\n"));
                return;
        }

        DEBUG(10, ("calculate_next_machine_pwd_change returned %s\n",
                   timeval_string(talloc_tos(), &next_change, false)));

        if (!NT_STATUS_IS_OK(result)) {
                struct timeval tmp;
                /*
                 * In case of failure, give the DC a minute to recover
                 */
                tmp = timeval_current_ofs(60, 0);
                next_change = timeval_max(&next_change, &tmp);
        }

done:
        child->machine_password_change_event = tevent_add_timer(global_event_context(), NULL,
                                                              next_change,
                                                              machine_password_change_handler,
                                                              child);
}

/* Deal with a request to go offline. */

static void child_msg_offline(struct messaging_context *msg,
                              void *private_data,
                              uint32_t msg_type,
                              struct server_id server_id,
                              DATA_BLOB *data)
{
        struct winbindd_domain *domain;
        struct winbindd_domain *primary_domain = NULL;
        const char *domainname = (const char *)data->data;

        if (data->data == NULL || data->length == 0) {
                return;
        }

        DEBUG(5,("child_msg_offline received for domain %s.\n", domainname));

        if (!lp_winbind_offline_logon()) {
                DEBUG(10,("child_msg_offline: rejecting offline message.\n"));
                return;
        }

        primary_domain = find_our_domain();

        /* Mark the requested domain offline. */

        for (domain = domain_list(); domain; domain = domain->next) {
                if (domain->internal) {
                        continue;
                }
                if (strequal(domain->name, domainname)) {
                        DEBUG(5,("child_msg_offline: marking %s offline.\n", domain->name));
                        set_domain_offline(domain);
                        /* we are in the trusted domain, set the primary domain
                         * offline too */
                        if (domain != primary_domain) {
                                set_domain_offline(primary_domain);
                        }
                }
        }
}

/* Deal with a request to go online. */

static void child_msg_online(struct messaging_context *msg,
                             void *private_data,
                             uint32_t msg_type,
                             struct server_id server_id,
                             DATA_BLOB *data)
{
        struct winbindd_domain *domain;
        struct winbindd_domain *primary_domain = NULL;
        const char *domainname = (const char *)data->data;

        if (data->data == NULL || data->length == 0) {
                return;
        }

        DEBUG(5,("child_msg_online received for domain %s.\n", domainname));

        if (!lp_winbind_offline_logon()) {
                DEBUG(10,("child_msg_online: rejecting online message.\n"));
                return;
        }

        primary_domain = find_our_domain();

        /* Set our global state as online. */
        set_global_winbindd_state_online();

        /* Try and mark everything online - delete any negative cache entries
           to force a reconnect now. */

        for (domain = domain_list(); domain; domain = domain->next) {
                if (domain->internal) {
                        continue;
                }
                if (strequal(domain->name, domainname)) {
                        DEBUG(5,("child_msg_online: requesting %s to go online.\n", domain->name));
                        winbindd_flush_negative_conn_cache(domain);
                        set_domain_online_request(domain);

                        /* we can be in trusted domain, which will contact primary domain
                         * we have to bring primary domain online in trusted domain process
                         * see, winbindd_dual_pam_auth() --> winbindd_dual_pam_auth_samlogon()
                         * --> contact_domain = find_our_domain()
                         * */
                        if (domain != primary_domain) {
                                winbindd_flush_negative_conn_cache(primary_domain);
                                set_domain_online_request(primary_domain);
                        }
                }
        }
}

struct winbindd_reinit_after_fork_state {
        const struct winbindd_child *myself;
};

static bool winbindd_reinit_after_fork_fn(struct winbindd_child *child,
                                          void *private_data)
{
        struct winbindd_reinit_after_fork_state *state = private_data;

        if (child == state->myself) {
                return true;
        }

        /* Destroy all possible events in child list. */
        TALLOC_FREE(child->lockout_policy_event);
        TALLOC_FREE(child->machine_password_change_event);

        /*
         * Children should never be able to send each other messages,
         * all messages must go through the parent.
         */
        child->pid = (pid_t)0;

        /*
         * Close service sockets to all other children
         */
        if (child->sock != -1) {
                close(child->sock);
                child->sock = -1;
        }

        return true;
}

NTSTATUS winbindd_reinit_after_fork(const struct winbindd_child *myself,
                                    const char *logfilename)
{
        struct winbindd_reinit_after_fork_state state = { .myself = myself };
        struct winbindd_domain *domain;
        NTSTATUS status;

        status = reinit_after_fork(
                global_messaging_context(),
                global_event_context(),
                true);
        if (!NT_STATUS_IS_OK(status)) {
                DEBUG(0,("reinit_after_fork() failed\n"));
                return status;
        }
        initialize_password_db(true, global_event_context());

        close_conns_after_fork();

        if (logfilename != NULL) {
                lp_set_logfile(logfilename);
                reopen_logs();
        }

        if (!winbindd_setup_sig_term_handler(false)) {
                return NT_STATUS_NO_MEMORY;
        }

        if (!winbindd_setup_sig_hup_handler(logfilename)) {
                return NT_STATUS_NO_MEMORY;
        }

        /* Stop zombies in children */
        CatchChild();

        /* Don't handle the same messages as our parent. */
        messaging_deregister(global_messaging_context(),
                             MSG_SMB_CONF_UPDATED, NULL);
        messaging_deregister(global_messaging_context(),
                             MSG_SHUTDOWN, NULL);
        messaging_deregister(global_messaging_context(),
                             MSG_WINBIND_OFFLINE, NULL);
        messaging_deregister(global_messaging_context(),
                             MSG_WINBIND_ONLINE, NULL);
        messaging_deregister(global_messaging_context(),
                             MSG_WINBIND_ONLINESTATUS, NULL);
        messaging_deregister(global_messaging_context(),
                             MSG_WINBIND_DUMP_DOMAIN_LIST, NULL);
        messaging_deregister(global_messaging_context(),
                             MSG_DEBUG, NULL);

        messaging_deregister(global_messaging_context(),
                             MSG_WINBIND_DOMAIN_OFFLINE, NULL);
        messaging_deregister(global_messaging_context(),
                             MSG_WINBIND_DOMAIN_ONLINE, NULL);

        /* We have destroyed all events in the winbindd_event_context
         * in reinit_after_fork(), so clean out all possible pending
         * event pointers. */

        /* Deal with check_online_events. */

        for (domain = domain_list(); domain; domain = domain->next) {
                TALLOC_FREE(domain->check_online_event);
        }

        /* Ensure we're not handling a credential cache event inherited
         * from our parent. */

        ccache_remove_all_after_fork();

        forall_children(winbindd_reinit_after_fork_fn, &state);

        return NT_STATUS_OK;
}

/*
 * In a child there will be only one domain, reference that here.
 */
static struct winbindd_domain *child_domain;

struct winbindd_domain *wb_child_domain(void)
{
        return child_domain;
}

struct child_handler_state {
        struct winbindd_child *child;
        struct winbindd_cli_state cli;
};

static void child_handler(struct tevent_context *ev, struct tevent_fd *fde,
                          uint16_t flags, void *private_data)
{
        struct child_handler_state *state =
                (struct child_handler_state *)private_data;
        NTSTATUS status;
        uint64_t parent_traceid;

        /* fetch a request from the main daemon */
        status = child_read_request(state->cli.sock, state->cli.request);

        if (!NT_STATUS_IS_OK(status)) {
                /* we lost contact with our parent */
                _exit(0);
        }

        /* read traceid from request */
        parent_traceid = state->cli.request->traceid;
        debug_traceid_set(parent_traceid);

        DEBUG(4,("child daemon request %d\n",
                 (int)state->cli.request->cmd));

        ZERO_STRUCTP(state->cli.response);
        state->cli.request->null_term = '\0';
        state->cli.mem_ctx = talloc_tos();
        child_process_request(state->child, &state->cli);

        DEBUG(4, ("Finished processing child request %d\n",
                  (int)state->cli.request->cmd));

        SAFE_FREE(state->cli.request->extra_data.data);

        status = child_write_response(state->cli.sock, state->cli.response);
        if (!NT_STATUS_IS_OK(status)) {
                exit(1);
        }
}

static bool fork_domain_child(struct winbindd_child *child)
{
        int fdpair[2];
        struct child_handler_state state;
        struct winbindd_request request;
        struct winbindd_response response;
        struct winbindd_domain *primary_domain = NULL;
        NTSTATUS status;
        ssize_t nwritten;
        struct tevent_fd *fde;
        struct tevent_req *req = NULL;

        if (child->domain) {
                DEBUG(10, ("fork_domain_child called for domain '%s'\n",
                           child->domain->name));
        } else {
                DEBUG(10, ("fork_domain_child called without domain.\n"));
        }

        if (socketpair(AF_UNIX, SOCK_STREAM, 0, fdpair) != 0) {
                DEBUG(0, ("Could not open child pipe: %s\n",
                          strerror(errno)));
                return False;
        }

        ZERO_STRUCT(state);
        state.child = child;
        state.cli.pid = getpid();
        state.cli.request = &request;
        state.cli.response = &response;

        child->pid = fork();

        if (child->pid == -1) {
                DEBUG(0, ("Could not fork: %s\n", strerror(errno)));
                close(fdpair[0]);
                close(fdpair[1]);
                return False;
        }

        if (child->pid != 0) {
                /* Parent */
                ssize_t nread;
                int rc;

                close(fdpair[0]);

                nread = sys_read(fdpair[1], &status, sizeof(status));
                if (nread != sizeof(status)) {
                        DEBUG(1, ("fork_domain_child: Could not read child status: "
                                  "nread=%d, error=%s\n", (int)nread,
                                  strerror(errno)));
                        close(fdpair[1]);
                        return false;
                }
                if (!NT_STATUS_IS_OK(status)) {
                        DEBUG(1, ("fork_domain_child: Child status is %s\n",
                                  nt_errstr(status)));
                        close(fdpair[1]);
                        return false;
                }

                child->monitor_fde = tevent_add_fd(global_event_context(),
                                                   global_event_context(),
                                                   fdpair[1],
                                                   TEVENT_FD_READ,
                                                   child_socket_readable,
                                                   child);
                if (child->monitor_fde == NULL) {
                        DBG_WARNING("tevent_add_fd failed\n");
                        close(fdpair[1]);
                        return false;
                }

                rc = set_blocking(fdpair[1], false);
                if (rc < 0) {
                        close(fdpair[1]);
                        return false;
                }

                child->sock = fdpair[1];

                return true;
        }

        /* Child */
        child_domain = child->domain;

        DEBUG(10, ("Child process %d\n", (int)getpid()));

        state.cli.sock = fdpair[0];
        close(fdpair[1]);

        /* Reset traceid and deactivate call_depth tracking */
        if (lp_winbind_debug_traceid()) {
                debug_traceid_set(1);
                tevent_thread_call_depth_set_callback(NULL, NULL);
        }

        status = winbindd_reinit_after_fork(child, child->logfilename);

        /* setup callbacks again, one of them is removed in reinit_after_fork */
        if (lp_winbind_debug_traceid()) {
                winbind_debug_traceid_setup(global_event_context());
        }

        nwritten = sys_write(state.cli.sock, &status, sizeof(status));
        if (nwritten != sizeof(status)) {
                DEBUG(1, ("fork_domain_child: Could not write status: "
                          "nwritten=%d, error=%s\n", (int)nwritten,
                          strerror(errno)));
                _exit(0);
        }
        if (!NT_STATUS_IS_OK(status)) {
                DEBUG(1, ("winbindd_reinit_after_fork failed: %s\n",
                          nt_errstr(status)));
                _exit(0);
        }

        if (child_domain != NULL) {
                process_set_title("wb[%s]", "domain child [%s]", child_domain->name);
        } else if (is_idmap_child(child)) {
                process_set_title("wb-idmap", "idmap child");
        }

        /* Handle online/offline messages. */
        messaging_register(global_messaging_context(), NULL,
                           MSG_WINBIND_OFFLINE, child_msg_offline);
        messaging_register(global_messaging_context(), NULL,
                           MSG_WINBIND_ONLINE, child_msg_online);
        messaging_register(global_messaging_context(), NULL,
                           MSG_DEBUG, debug_message);
        messaging_register(global_messaging_context(), NULL,
                           MSG_WINBIND_IP_DROPPED,
                           winbind_msg_ip_dropped);
        messaging_register(global_messaging_context(), NULL,
                           MSG_WINBIND_DISCONNECT_DC,
                           winbind_msg_disconnect_dc);

        req = messaging_filtered_read_send(global_event_context(),
                                           global_event_context(),
                                           global_messaging_context(),
                                           winbindd_child_msg_filter,
                                           child);
        if (req == NULL) {
                DBG_ERR("messaging_filtered_read_send failed\n");
                _exit(1);
        }

        primary_domain = find_our_domain();

        if (primary_domain == NULL) {
                smb_panic("no primary domain found");
        }

        /* It doesn't matter if we allow cache login,
         * try to bring domain online after fork. */
        if ( child->domain ) {
                child->domain->startup = True;
                child->domain->startup_time = time_mono(NULL);
                /* we can be in primary domain or in trusted domain
                 * If we are in trusted domain, set the primary domain
                 * in start-up mode */
                if (!(child->domain->internal)) {
                        set_domain_online_request(child->domain);
                        if (!(child->domain->primary)) {
                                primary_domain->startup = True;
                                primary_domain->startup_time = time_mono(NULL);
                                set_domain_online_request(primary_domain);
                        }
                }
        }

        /* We might be in the idmap child...*/
        if (child->domain && !(child->domain->internal) &&
            lp_winbind_offline_logon()) {

                set_domain_online_request(child->domain);

                if (primary_domain && (primary_domain != child->domain)) {
                        /* We need to talk to the primary
                         * domain as well as the trusted
                         * domain inside a trusted domain
                         * child.
                         * See the code in :
                         * set_dc_type_and_flags_trustinfo()
                         * for details.
                         */
                        set_domain_online_request(primary_domain);
                }

                child->lockout_policy_event = tevent_add_timer(
                        global_event_context(), NULL, timeval_zero(),
                        account_lockout_policy_handler,
                        child);
        }

        if (child->domain && child->domain->primary &&
            !USE_KERBEROS_KEYTAB &&
            lp_server_role() == ROLE_DOMAIN_MEMBER) {

                struct timeval next_change;

                if (calculate_next_machine_pwd_change(child->domain->name,
                                                       &next_change)) {
                        child->machine_password_change_event = tevent_add_timer(
                                global_event_context(), NULL, next_change,
                                machine_password_change_handler,
                                child);
                }
        }

        fde = tevent_add_fd(global_event_context(), NULL, state.cli.sock,
                            TEVENT_FD_READ, child_handler, &state);
        if (fde == NULL) {
                DEBUG(1, ("tevent_add_fd failed\n"));
                _exit(1);
        }

        while (1) {

                int ret;
                TALLOC_CTX *frame = talloc_stackframe();

                ret = tevent_loop_once(global_event_context());
                if (ret != 0) {
                        DEBUG(1, ("tevent_loop_once failed: %s\n",
                                  strerror(errno)));
                        _exit(1);
                }

                if (child->domain && child->domain->startup &&
                                (time_mono(NULL) > child->domain->startup_time + 30)) {
                        /* No longer in "startup" mode. */
                        DEBUG(10,("fork_domain_child: domain %s no longer in 'startup' mode.\n",
                                child->domain->name ));
                        child->domain->startup = False;
                }

                TALLOC_FREE(frame);
        }
}

void winbind_msg_ip_dropped_parent(struct messaging_context *msg_ctx,
                                   void *private_data,
                                   uint32_t msg_type,
                                   struct server_id server_id,
                                   DATA_BLOB *data)
{
        struct winbind_msg_relay_state state = {
                .msg_ctx = msg_ctx,
                .msg_type = msg_type,
                .data = data,
        };

        winbind_msg_ip_dropped(msg_ctx, private_data, msg_type,
                               server_id, data);

        forall_children(winbind_msg_relay_fn, &state);
}

void winbindd_terminate(bool is_parent)
{
        if (is_parent) {
                /* When parent goes away we should
                 * remove the socket file. Not so
                 * when children terminate.
                 */
                char *path = NULL;

                if (asprintf(&path, "%s/%s",
                        lp_winbindd_socket_directory(), WINBINDD_SOCKET_NAME) > 0) {
                        unlink(path);
                        SAFE_FREE(path);
                }
        }

        idmap_close();

        netlogon_creds_cli_close_global_db();

#if 0
        if (interactive) {
                TALLOC_CTX *mem_ctx = talloc_init("end_description");
                char *description = talloc_describe_all(mem_ctx);

                DEBUG(3, ("tallocs left:\n%s\n", description));
                talloc_destroy(mem_ctx);
        }
#endif

        if (is_parent) {
                pidfile_unlink(lp_pid_directory(), "winbindd");
        }

        exit(0);
}

static void winbindd_sig_term_handler(struct tevent_context *ev,
                                      struct tevent_signal *se,
                                      int signum,
                                      int count,
                                      void *siginfo,
                                      void *private_data)
{
        bool *p = talloc_get_type_abort(private_data, bool);
        bool is_parent = *p;

        TALLOC_FREE(p);

        DEBUG(0,("Got sig[%d] terminate (is_parent=%d)\n",
                 signum, is_parent));
        winbindd_terminate(is_parent);
}

bool winbindd_setup_sig_term_handler(bool parent)
{
        struct tevent_signal *se;
        bool *is_parent;

        is_parent = talloc(global_event_context(), bool);
        if (!is_parent) {
                return false;
        }

        *is_parent = parent;

        se = tevent_add_signal(global_event_context(),
                               is_parent,
                               SIGTERM, 0,
                               winbindd_sig_term_handler,
                               is_parent);
        if (!se) {
                DEBUG(0,("failed to setup SIGTERM handler\n"));
                talloc_free(is_parent);
                return false;
        }

        se = tevent_add_signal(global_event_context(),
                               is_parent,
                               SIGINT, 0,
                               winbindd_sig_term_handler,
                               is_parent);
        if (!se) {
                DEBUG(0,("failed to setup SIGINT handler\n"));
                talloc_free(is_parent);
                return false;
        }

        se = tevent_add_signal(global_event_context(),
                               is_parent,
                               SIGQUIT, 0,
                               winbindd_sig_term_handler,
                               is_parent);
        if (!se) {
                DEBUG(0,("failed to setup SIGINT handler\n"));
                talloc_free(is_parent);
                return false;
        }

        return true;
}

static void flush_caches_noinit(void)
{
        /*
         * We need to invalidate cached user list entries on a SIGHUP
         * otherwise cached access denied errors due to restrict anonymous
         * hang around until the sequence number changes.
         * NB
         * Skip uninitialized domains when flush cache.
         * If domain is not initialized, it means it is never
         * used or never become online. look, wcache_invalidate_cache()
         * -> get_cache() -> init_dc_connection(). It causes a lot of traffic
         * for unused domains and large traffic for primary domain's DC if there
         * are many domains..
         */

        if (!wcache_invalidate_cache_noinit()) {
                DEBUG(0, ("invalidating the cache failed; revalidate the cache\n"));
                if (!winbindd_cache_validate_and_initialize()) {
                        exit(1);
                }
        }
}

static void winbindd_sig_hup_handler(struct tevent_context *ev,
                                     struct tevent_signal *se,
                                     int signum,
                                     int count,
                                     void *siginfo,
                                     void *private_data)
{
        const char *file = (const char *)private_data;

        DEBUG(1,("Reloading services after SIGHUP\n"));
        flush_caches_noinit();
        winbindd_reload_services_file(file);
}

bool winbindd_setup_sig_hup_handler(const char *lfile)
{
        struct tevent_signal *se;
        char *file = NULL;

        if (lfile) {
                file = talloc_strdup(global_event_context(),
                                     lfile);
                if (!file) {
                        return false;
                }
        }

        se = tevent_add_signal(global_event_context(),
                               global_event_context(),
                               SIGHUP, 0,
                               winbindd_sig_hup_handler,
                               file);
        if (!se) {
                return false;
        }

        return true;
}