libosmo-sccp/src/sccp_scoc.c

1999 lines
66 KiB
C
Raw Normal View History

/* SCCP Connection Oriented (SCOC) according to ITU-T Q.713/Q.714 */
/* (C) 2015-2017 by Harald Welte <laforge@gnumonks.org>
* All Rights reserved
*
* SPDX-License-Identifier: GPL-2.0+
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*
*/
/* This code is a bit of a hybrid between the ITU-T Q.71x specifications
* for SCCP (particularly its connection-oriented part), and the IETF
* RFC 3868 (SUA). The idea here is to have one shared code base of the
* state machines for SCCP Connection Oriented, and use those both from
* SCCP and SUA.
*
* To do so, all SCCP messages are translated to SUA messages in the
* input side, and all generated SUA messages are translated to SCCP on
* the output side.
*
* The Choice of going for SUA messages as the "native" format was based
* on their easier parseability, and the fact that there are features in
* SUA which classic SCCP cannot handle (like IP addresses in GT).
* However, all SCCP features can be expressed in SUA.
*
* The code only supports Class 2. No support for Class 3 is intended,
* but patches are of course always welcome.
*
* Missing other features:
* * Segmentation/Reassembly support
* * T(guard) after (re)start
* * freezing of local references
* * parsing/encoding of IPv4/IPv6 addresses
* * use of multiple Routing Contexts in SUA case
*/
#include <errno.h>
#include <string.h>
#include <osmocom/core/msgb.h>
#include <osmocom/core/utils.h>
#include <osmocom/core/linuxlist.h>
#include <osmocom/core/linuxrbtree.h>
#include <osmocom/core/logging.h>
#include <osmocom/core/timer.h>
#include <osmocom/core/fsm.h>
#include <osmocom/sigtran/sccp_helpers.h>
#include <osmocom/sigtran/sccp_sap.h>
#include <osmocom/sigtran/protocol/sua.h>
#include <osmocom/sccp/sccp_types.h>
#include "xua_internal.h"
#include "sccp_internal.h"
#define S(x) (1 << (x))
#define SCU_MSGB_SIZE 1024
/***********************************************************************
* SCCP connection table
***********************************************************************/
/* a logical connection within the SCCP instance */
struct sccp_connection {
/* entry in (struct sccp_instance)->connections */
struct rb_node node;
/* which instance are we part of? */
struct osmo_sccp_instance *inst;
/* which user owns us? */
struct osmo_sccp_user *user;
/* remote point code */
uint32_t remote_pc;
/* local/remote addresses and identities */
struct osmo_sccp_addr calling_addr;
struct osmo_sccp_addr called_addr;
/* SCCP connection identifier. Only relevant across the SCCP User SAP,
* i.e. between the local application using the SCCP stack provided by
* libosmo-sccp. Never transmitted over the wire! */
uint32_t conn_id;
/* SCCP Remote Connection Reference. Allocated by the remote
* SCCP stack to uniquely identify a SCCP connection on its end.
* We don't interpret it, but simply cache it here so we can use
* it whenever sending data to the peer. Only relevant over the
* wire, not to be used across the SCCP user SAP */
uint32_t remote_ref;
uint32_t importance;
uint32_t sccp_class;
uint32_t release_cause; /* WAIT_CONN_CONF */
struct msgb *opt_data_cache;
/* incoming (true) or outgoing (false) */
bool incoming;
/* Osmo FSM Instance of sccp_scoc_fsm */
struct osmo_fsm_inst *fi;
/* Connect timer */
struct osmo_timer_list t_conn;
/* inactivity timers */
struct osmo_timer_list t_ias;
struct osmo_timer_list t_iar;
/* release timers */
struct osmo_timer_list t_rel;
struct osmo_timer_list t_int;
struct osmo_timer_list t_rep_rel;
};
/***********************************************************************
* various helper functions
***********************************************************************/
enum sccp_connection_state {
S_IDLE,
S_CONN_PEND_IN,
S_CONN_PEND_OUT,
S_ACTIVE,
S_DISCONN_PEND,
S_RESET_IN,
S_RESET_OUT,
S_BOTHWAY_RESET,
S_WAIT_CONN_CONF,
};
/* Events that this FSM can process */
enum sccp_scoc_event {
/* Primitives from SCCP-User */
SCOC_E_SCU_N_CONN_REQ,
SCOC_E_SCU_N_CONN_RESP,
SCOC_E_SCU_N_DISC_REQ,
SCOC_E_SCU_N_DATA_REQ,
SCOC_E_SCU_N_EXP_DATA_REQ,
/* Events from RCOC (Routing for Connection Oriented) */
SCOC_E_RCOC_CONN_IND,
SCOC_E_RCOC_ROUT_FAIL_IND,
SCOC_E_RCOC_RLSD_IND,
SCOC_E_RCOC_REL_COMPL_IND,
SCOC_E_RCOC_CREF_IND,
SCOC_E_RCOC_CC_IND,
SCOC_E_RCOC_DT1_IND,
SCOC_E_RCOC_DT2_IND,
SCOC_E_RCOC_IT_IND,
SCOC_E_RCOC_OTHER_NPDU,
SCOC_E_RCOC_ERROR_IND,
/* Timer Events */
SCOC_E_T_IAR_EXP,
SCOC_E_T_IAS_EXP,
SCOC_E_CONN_TMR_EXP,
SCOC_E_T_REL_EXP,
SCOC_E_T_INT_EXP,
SCOC_E_T_REP_REL_EXP,
};
static const struct value_string scoc_event_names[] = {
/* Primitives from SCCP-User */
{ SCOC_E_SCU_N_CONN_REQ, "N-CONNECT.req" },
{ SCOC_E_SCU_N_CONN_RESP, "N-CONNECT.resp" },
{ SCOC_E_SCU_N_DISC_REQ, "N-DISCONNECT.req" },
{ SCOC_E_SCU_N_DATA_REQ, "N-DATA.req" },
{ SCOC_E_SCU_N_EXP_DATA_REQ, "N-EXPEDITED_DATA.req" },
/* Events from RCOC (Routing for Connection Oriented) */
{ SCOC_E_RCOC_CONN_IND, "RCOC-CONNECT.ind" },
{ SCOC_E_RCOC_ROUT_FAIL_IND, "RCOC-ROUT_FAIL.ind" },
{ SCOC_E_RCOC_RLSD_IND, "RCOC-RELEASED.ind" },
{ SCOC_E_RCOC_REL_COMPL_IND, "RCOC-RELEASE_COMPLETE.ind" },
{ SCOC_E_RCOC_CREF_IND, "RCOC-CONNECT_REFUSED.ind" },
{ SCOC_E_RCOC_CC_IND, "RCOC-CONNECT_CONFIRM.ind" },
{ SCOC_E_RCOC_DT1_IND, "RCOC-DT1.ind" },
{ SCOC_E_RCOC_DT2_IND, "RCOC-DT2.ind" },
{ SCOC_E_RCOC_IT_IND, "RCOC-IT.ind" },
{ SCOC_E_RCOC_OTHER_NPDU, "RCOC-OTHER_NPDU.ind" },
{ SCOC_E_RCOC_ERROR_IND, "RCOC-ERROR.ind" },
{ SCOC_E_T_IAR_EXP, "T(iar)_expired" },
{ SCOC_E_T_IAS_EXP, "T(ias)_expired" },
{ SCOC_E_CONN_TMR_EXP, "T(conn)_expired" },
{ SCOC_E_T_REL_EXP, "T(rel)_expired" },
{ SCOC_E_T_INT_EXP, "T(int)_expired" },
{ SCOC_E_T_REP_REL_EXP, "T(rep_rel)_expired" },
{ 0, NULL }
};
/* how to map a SCCP CO message to an event */
static const struct xua_msg_event_map sua_scoc_event_map[] = {
{ SUA_MSGC_CO, SUA_CO_CORE, SCOC_E_RCOC_CONN_IND },
{ SUA_MSGC_CO, SUA_CO_RELRE, SCOC_E_RCOC_RLSD_IND },
{ SUA_MSGC_CO, SUA_CO_RELCO, SCOC_E_RCOC_REL_COMPL_IND },
{ SUA_MSGC_CO, SUA_CO_COREF, SCOC_E_RCOC_CREF_IND },
{ SUA_MSGC_CO, SUA_CO_COAK, SCOC_E_RCOC_CC_IND },
{ SUA_MSGC_CO, SUA_CO_CODT, SCOC_E_RCOC_DT1_IND },
{ SUA_MSGC_CO, SUA_CO_COIT, SCOC_E_RCOC_IT_IND },
{ SUA_MSGC_CO, SUA_CO_COERR, SCOC_E_RCOC_ERROR_IND },
};
/* map from SCU-primitives to SCOC FSM events */
static const struct osmo_prim_event_map scu_scoc_event_map[] = {
{ SCCP_SAP_USER, OSMO_SCU_PRIM_N_CONNECT, PRIM_OP_REQUEST,
SCOC_E_SCU_N_CONN_REQ },
{ SCCP_SAP_USER, OSMO_SCU_PRIM_N_CONNECT, PRIM_OP_RESPONSE,
SCOC_E_SCU_N_CONN_RESP },
{ SCCP_SAP_USER, OSMO_SCU_PRIM_N_DATA, PRIM_OP_REQUEST,
SCOC_E_SCU_N_DATA_REQ },
{ SCCP_SAP_USER, OSMO_SCU_PRIM_N_DISCONNECT, PRIM_OP_REQUEST,
SCOC_E_SCU_N_DISC_REQ },
{ SCCP_SAP_USER, OSMO_SCU_PRIM_N_EXPEDITED_DATA, PRIM_OP_REQUEST,
SCOC_E_SCU_N_EXP_DATA_REQ },
{ 0, 0, 0, OSMO_NO_EVENT }
};
/***********************************************************************
* Timer Handling
***********************************************************************/
/* Mostly pasted from Appendix C.4 of ITU-T Q.714 (05/2001) -- some of their descriptions are quite
* unintelligible out of context, for which we have our own description here. */
const struct osmo_tdef osmo_sccp_timer_defaults[OSMO_SCCP_TIMERS_LEN] = {
{ .T = OSMO_SCCP_TIMER_CONN_EST, .default_val = 1*60, .unit = OSMO_TDEF_S,
.desc = "Waiting for connection confirm message, 1 to 2 minutes" },
{ .T = OSMO_SCCP_TIMER_IAS, .default_val = 7*60, .unit = OSMO_TDEF_S,
.desc = "Send keep-alive: on an idle connection, delay before sending an Idle Timer message, 5 to 10 minutes" }, /* RFC 3868 Ch. 8. */
{ .T = OSMO_SCCP_TIMER_IAR, .default_val = 15*60, .unit = OSMO_TDEF_S,
.desc = "Receive keep-alive: on an idle connection, delay until considering a connection as stale, 11 to 21 minutes" }, /* RFC 3868 Ch. 8. */
{ .T = OSMO_SCCP_TIMER_REL, .default_val = 10, .unit = OSMO_TDEF_S,
.desc = "Waiting for release complete message, 10 to 20 seconds" },
{ .T = OSMO_SCCP_TIMER_REPEAT_REL, .default_val = 10, .unit = OSMO_TDEF_S,
.desc = "Waiting for release complete message; or to repeat sending released message after the initial expiry, 10 to 20 seconds" },
{ .T = OSMO_SCCP_TIMER_INT, .default_val = 1*60, .unit = OSMO_TDEF_S,
.desc = "Waiting for release complete message; or to release connection resources, freeze the LRN and "
"alert a maintenance function after the initial expiry, extending to 1 minute" },
{ .T = OSMO_SCCP_TIMER_GUARD, .default_val = 23*60, .unit = OSMO_TDEF_S,
.desc = "Waiting to resume normal procedure for temporary connection sections during the restart procedure, 23 to 25 minutes" },
{ .T = OSMO_SCCP_TIMER_RESET, .default_val = 10, .unit = OSMO_TDEF_S,
.desc = "Waiting to release temporary connection section or alert maintenance function after reset request message is sent, 10 to 20 seconds" },
{ .T = OSMO_SCCP_TIMER_REASSEMBLY, .default_val = 10, .unit = OSMO_TDEF_S,
.desc = "Waiting to receive all the segments of the remaining segments, single segmented message after receiving the first segment, 10 to 20 seconds" },
{}
};
make SCCP timers configurable The previous hardcoded SCCP timers may cause SCCP connection releases, if the peer is configured with far lower timers than libosmo-sccp. Testing with a specific SCCPlite MSC, I experienced an iar of just over three minutes, meaning that calls would be cut off by the MSC, since the osmo-bsc failed to send an Inactivity Timer message until seven minutes have passed. With this patch, SCCP timers are configurable by the user. Define constant global default timers, and variable user-configurable timers with each osmo_sccp_instance. Add VTY UI to configure the timers. Users must call osmo_sccp_vty_init() to get the sccp-timer config nodes under the 'cs7' node. Show the new UI in ss7_asp_test.vty. Note that even though this function is not new at all, until recently, all of our SCCP users (osmo-bsc, osmo-msc, osmo-sgsn, osmo-hnbgw) failed to call osmo_sccp_vty_init(), and thus also missed out on the various 'show' commands defined in sccp_vty.c. In other words, to benefit from the timer configurability, the patches to call osmo_sccp_vty_init() must first be merged to the corresponding master branches. If a 'sccp-timer' config command occurs, the cs7 instance must allocate an SCCP instance in order to store the timer config. Do that by calling the recently added osmo_ss7_ensure_sccp() function. Hence remove the limitation that the SCCP instance must not be populated from the "simple" setup function. If we want to configure SCCP timers beforehand, there must be an SCCP instance for that, and there is no hard reason to require a NULL SCCP instance, besides the desire to prevent this function from being invoked twice. Change-Id: I28a7362aa838e648ecc9b26ee53dbcade81a9d65
2018-09-26 15:12:23 +00:00
/* Appendix C.4 of ITU-T Q.714 */
const struct value_string osmo_sccp_timer_names[] = {
{ OSMO_SCCP_TIMER_CONN_EST, "conn_est" },
{ OSMO_SCCP_TIMER_IAS, "ias" },
{ OSMO_SCCP_TIMER_IAR, "iar" },
{ OSMO_SCCP_TIMER_REL, "rel" },
{ OSMO_SCCP_TIMER_REPEAT_REL, "repeat_rel" },
{ OSMO_SCCP_TIMER_INT, "int" },
{ OSMO_SCCP_TIMER_GUARD, "guard" },
{ OSMO_SCCP_TIMER_RESET, "reset" },
{ OSMO_SCCP_TIMER_REASSEMBLY, "reassembly" },
{}
};
osmo_static_assert(ARRAY_SIZE(osmo_sccp_timer_defaults) == (OSMO_SCCP_TIMERS_LEN) &&
ARRAY_SIZE(osmo_sccp_timer_names) == (OSMO_SCCP_TIMERS_LEN),
make SCCP timers configurable The previous hardcoded SCCP timers may cause SCCP connection releases, if the peer is configured with far lower timers than libosmo-sccp. Testing with a specific SCCPlite MSC, I experienced an iar of just over three minutes, meaning that calls would be cut off by the MSC, since the osmo-bsc failed to send an Inactivity Timer message until seven minutes have passed. With this patch, SCCP timers are configurable by the user. Define constant global default timers, and variable user-configurable timers with each osmo_sccp_instance. Add VTY UI to configure the timers. Users must call osmo_sccp_vty_init() to get the sccp-timer config nodes under the 'cs7' node. Show the new UI in ss7_asp_test.vty. Note that even though this function is not new at all, until recently, all of our SCCP users (osmo-bsc, osmo-msc, osmo-sgsn, osmo-hnbgw) failed to call osmo_sccp_vty_init(), and thus also missed out on the various 'show' commands defined in sccp_vty.c. In other words, to benefit from the timer configurability, the patches to call osmo_sccp_vty_init() must first be merged to the corresponding master branches. If a 'sccp-timer' config command occurs, the cs7 instance must allocate an SCCP instance in order to store the timer config. Do that by calling the recently added osmo_ss7_ensure_sccp() function. Hence remove the limitation that the SCCP instance must not be populated from the "simple" setup function. If we want to configure SCCP timers beforehand, there must be an SCCP instance for that, and there is no hard reason to require a NULL SCCP instance, besides the desire to prevent this function from being invoked twice. Change-Id: I28a7362aa838e648ecc9b26ee53dbcade81a9d65
2018-09-26 15:12:23 +00:00
assert_osmo_sccp_timers_count);
static void sccp_timer_schedule(const struct sccp_connection *conn,
struct osmo_timer_list *timer,
enum osmo_sccp_timer timer_name)
{
const unsigned long val_sec = osmo_tdef_get(conn->inst->tdefs, timer_name, OSMO_TDEF_S, -1);
osmo_timer_schedule(timer, val_sec, 0);
make SCCP timers configurable The previous hardcoded SCCP timers may cause SCCP connection releases, if the peer is configured with far lower timers than libosmo-sccp. Testing with a specific SCCPlite MSC, I experienced an iar of just over three minutes, meaning that calls would be cut off by the MSC, since the osmo-bsc failed to send an Inactivity Timer message until seven minutes have passed. With this patch, SCCP timers are configurable by the user. Define constant global default timers, and variable user-configurable timers with each osmo_sccp_instance. Add VTY UI to configure the timers. Users must call osmo_sccp_vty_init() to get the sccp-timer config nodes under the 'cs7' node. Show the new UI in ss7_asp_test.vty. Note that even though this function is not new at all, until recently, all of our SCCP users (osmo-bsc, osmo-msc, osmo-sgsn, osmo-hnbgw) failed to call osmo_sccp_vty_init(), and thus also missed out on the various 'show' commands defined in sccp_vty.c. In other words, to benefit from the timer configurability, the patches to call osmo_sccp_vty_init() must first be merged to the corresponding master branches. If a 'sccp-timer' config command occurs, the cs7 instance must allocate an SCCP instance in order to store the timer config. Do that by calling the recently added osmo_ss7_ensure_sccp() function. Hence remove the limitation that the SCCP instance must not be populated from the "simple" setup function. If we want to configure SCCP timers beforehand, there must be an SCCP instance for that, and there is no hard reason to require a NULL SCCP instance, besides the desire to prevent this function from being invoked twice. Change-Id: I28a7362aa838e648ecc9b26ee53dbcade81a9d65
2018-09-26 15:12:23 +00:00
}
/* T(ias) has expired, send a COIT message to the peer */
static void tx_inact_tmr_cb(void *data)
{
struct sccp_connection *conn = data;
osmo_fsm_inst_dispatch(conn->fi, SCOC_E_T_IAS_EXP, NULL);
}
/* T(iar) has expired, notify the FSM about it */
static void rx_inact_tmr_cb(void *data)
{
struct sccp_connection *conn = data;
osmo_fsm_inst_dispatch(conn->fi, SCOC_E_T_IAR_EXP, NULL);
}
/* T(rel) has expired, notify the FSM about it */
static void rel_tmr_cb(void *data)
{
struct sccp_connection *conn = data;
osmo_fsm_inst_dispatch(conn->fi, SCOC_E_T_REL_EXP, NULL);
}
/* T(int) has expired, notify the FSM about it */
static void int_tmr_cb(void *data)
{
struct sccp_connection *conn = data;
osmo_fsm_inst_dispatch(conn->fi, SCOC_E_T_INT_EXP, NULL);
}
/* T(repeat_rel) has expired, notify the FSM about it */
static void rep_rel_tmr_cb(void *data)
{
struct sccp_connection *conn = data;
osmo_fsm_inst_dispatch(conn->fi, SCOC_E_T_REP_REL_EXP, NULL);
}
/* T(conn) has expired, notify the FSM about it */
static void conn_tmr_cb(void *data)
{
struct sccp_connection *conn = data;
osmo_fsm_inst_dispatch(conn->fi, SCOC_E_CONN_TMR_EXP, NULL);
}
/* Re-start the Tx inactivity timer */
static void conn_restart_tx_inact_timer(struct sccp_connection *conn)
{
make SCCP timers configurable The previous hardcoded SCCP timers may cause SCCP connection releases, if the peer is configured with far lower timers than libosmo-sccp. Testing with a specific SCCPlite MSC, I experienced an iar of just over three minutes, meaning that calls would be cut off by the MSC, since the osmo-bsc failed to send an Inactivity Timer message until seven minutes have passed. With this patch, SCCP timers are configurable by the user. Define constant global default timers, and variable user-configurable timers with each osmo_sccp_instance. Add VTY UI to configure the timers. Users must call osmo_sccp_vty_init() to get the sccp-timer config nodes under the 'cs7' node. Show the new UI in ss7_asp_test.vty. Note that even though this function is not new at all, until recently, all of our SCCP users (osmo-bsc, osmo-msc, osmo-sgsn, osmo-hnbgw) failed to call osmo_sccp_vty_init(), and thus also missed out on the various 'show' commands defined in sccp_vty.c. In other words, to benefit from the timer configurability, the patches to call osmo_sccp_vty_init() must first be merged to the corresponding master branches. If a 'sccp-timer' config command occurs, the cs7 instance must allocate an SCCP instance in order to store the timer config. Do that by calling the recently added osmo_ss7_ensure_sccp() function. Hence remove the limitation that the SCCP instance must not be populated from the "simple" setup function. If we want to configure SCCP timers beforehand, there must be an SCCP instance for that, and there is no hard reason to require a NULL SCCP instance, besides the desire to prevent this function from being invoked twice. Change-Id: I28a7362aa838e648ecc9b26ee53dbcade81a9d65
2018-09-26 15:12:23 +00:00
sccp_timer_schedule(conn, &conn->t_ias, OSMO_SCCP_TIMER_IAS);
}
/* Re-start the Rx inactivity timer */
static void conn_restart_rx_inact_timer(struct sccp_connection *conn)
{
make SCCP timers configurable The previous hardcoded SCCP timers may cause SCCP connection releases, if the peer is configured with far lower timers than libosmo-sccp. Testing with a specific SCCPlite MSC, I experienced an iar of just over three minutes, meaning that calls would be cut off by the MSC, since the osmo-bsc failed to send an Inactivity Timer message until seven minutes have passed. With this patch, SCCP timers are configurable by the user. Define constant global default timers, and variable user-configurable timers with each osmo_sccp_instance. Add VTY UI to configure the timers. Users must call osmo_sccp_vty_init() to get the sccp-timer config nodes under the 'cs7' node. Show the new UI in ss7_asp_test.vty. Note that even though this function is not new at all, until recently, all of our SCCP users (osmo-bsc, osmo-msc, osmo-sgsn, osmo-hnbgw) failed to call osmo_sccp_vty_init(), and thus also missed out on the various 'show' commands defined in sccp_vty.c. In other words, to benefit from the timer configurability, the patches to call osmo_sccp_vty_init() must first be merged to the corresponding master branches. If a 'sccp-timer' config command occurs, the cs7 instance must allocate an SCCP instance in order to store the timer config. Do that by calling the recently added osmo_ss7_ensure_sccp() function. Hence remove the limitation that the SCCP instance must not be populated from the "simple" setup function. If we want to configure SCCP timers beforehand, there must be an SCCP instance for that, and there is no hard reason to require a NULL SCCP instance, besides the desire to prevent this function from being invoked twice. Change-Id: I28a7362aa838e648ecc9b26ee53dbcade81a9d65
2018-09-26 15:12:23 +00:00
sccp_timer_schedule(conn, &conn->t_iar, OSMO_SCCP_TIMER_IAR);
}
/* Re-start both Rx and Tx inactivity timers */
static void conn_start_inact_timers(struct sccp_connection *conn)
{
conn_restart_tx_inact_timer(conn);
conn_restart_rx_inact_timer(conn);
}
/* Stop both Rx and Tx inactivity timers */
static void conn_stop_inact_timers(struct sccp_connection *conn)
{
osmo_timer_del(&conn->t_ias);
osmo_timer_del(&conn->t_iar);
}
/* Start release timer T(rel) */
static void conn_start_rel_timer(struct sccp_connection *conn)
{
make SCCP timers configurable The previous hardcoded SCCP timers may cause SCCP connection releases, if the peer is configured with far lower timers than libosmo-sccp. Testing with a specific SCCPlite MSC, I experienced an iar of just over three minutes, meaning that calls would be cut off by the MSC, since the osmo-bsc failed to send an Inactivity Timer message until seven minutes have passed. With this patch, SCCP timers are configurable by the user. Define constant global default timers, and variable user-configurable timers with each osmo_sccp_instance. Add VTY UI to configure the timers. Users must call osmo_sccp_vty_init() to get the sccp-timer config nodes under the 'cs7' node. Show the new UI in ss7_asp_test.vty. Note that even though this function is not new at all, until recently, all of our SCCP users (osmo-bsc, osmo-msc, osmo-sgsn, osmo-hnbgw) failed to call osmo_sccp_vty_init(), and thus also missed out on the various 'show' commands defined in sccp_vty.c. In other words, to benefit from the timer configurability, the patches to call osmo_sccp_vty_init() must first be merged to the corresponding master branches. If a 'sccp-timer' config command occurs, the cs7 instance must allocate an SCCP instance in order to store the timer config. Do that by calling the recently added osmo_ss7_ensure_sccp() function. Hence remove the limitation that the SCCP instance must not be populated from the "simple" setup function. If we want to configure SCCP timers beforehand, there must be an SCCP instance for that, and there is no hard reason to require a NULL SCCP instance, besides the desire to prevent this function from being invoked twice. Change-Id: I28a7362aa838e648ecc9b26ee53dbcade81a9d65
2018-09-26 15:12:23 +00:00
sccp_timer_schedule(conn, &conn->t_rel, OSMO_SCCP_TIMER_REL);
}
/* Start repeat release timer T(rep_rel) */
static void conn_start_rep_rel_timer(struct sccp_connection *conn)
{
make SCCP timers configurable The previous hardcoded SCCP timers may cause SCCP connection releases, if the peer is configured with far lower timers than libosmo-sccp. Testing with a specific SCCPlite MSC, I experienced an iar of just over three minutes, meaning that calls would be cut off by the MSC, since the osmo-bsc failed to send an Inactivity Timer message until seven minutes have passed. With this patch, SCCP timers are configurable by the user. Define constant global default timers, and variable user-configurable timers with each osmo_sccp_instance. Add VTY UI to configure the timers. Users must call osmo_sccp_vty_init() to get the sccp-timer config nodes under the 'cs7' node. Show the new UI in ss7_asp_test.vty. Note that even though this function is not new at all, until recently, all of our SCCP users (osmo-bsc, osmo-msc, osmo-sgsn, osmo-hnbgw) failed to call osmo_sccp_vty_init(), and thus also missed out on the various 'show' commands defined in sccp_vty.c. In other words, to benefit from the timer configurability, the patches to call osmo_sccp_vty_init() must first be merged to the corresponding master branches. If a 'sccp-timer' config command occurs, the cs7 instance must allocate an SCCP instance in order to store the timer config. Do that by calling the recently added osmo_ss7_ensure_sccp() function. Hence remove the limitation that the SCCP instance must not be populated from the "simple" setup function. If we want to configure SCCP timers beforehand, there must be an SCCP instance for that, and there is no hard reason to require a NULL SCCP instance, besides the desire to prevent this function from being invoked twice. Change-Id: I28a7362aa838e648ecc9b26ee53dbcade81a9d65
2018-09-26 15:12:23 +00:00
sccp_timer_schedule(conn, &conn->t_rep_rel, OSMO_SCCP_TIMER_REPEAT_REL);
}
/* Start interval timer T(int) */
static void conn_start_int_timer(struct sccp_connection *conn)
{
make SCCP timers configurable The previous hardcoded SCCP timers may cause SCCP connection releases, if the peer is configured with far lower timers than libosmo-sccp. Testing with a specific SCCPlite MSC, I experienced an iar of just over three minutes, meaning that calls would be cut off by the MSC, since the osmo-bsc failed to send an Inactivity Timer message until seven minutes have passed. With this patch, SCCP timers are configurable by the user. Define constant global default timers, and variable user-configurable timers with each osmo_sccp_instance. Add VTY UI to configure the timers. Users must call osmo_sccp_vty_init() to get the sccp-timer config nodes under the 'cs7' node. Show the new UI in ss7_asp_test.vty. Note that even though this function is not new at all, until recently, all of our SCCP users (osmo-bsc, osmo-msc, osmo-sgsn, osmo-hnbgw) failed to call osmo_sccp_vty_init(), and thus also missed out on the various 'show' commands defined in sccp_vty.c. In other words, to benefit from the timer configurability, the patches to call osmo_sccp_vty_init() must first be merged to the corresponding master branches. If a 'sccp-timer' config command occurs, the cs7 instance must allocate an SCCP instance in order to store the timer config. Do that by calling the recently added osmo_ss7_ensure_sccp() function. Hence remove the limitation that the SCCP instance must not be populated from the "simple" setup function. If we want to configure SCCP timers beforehand, there must be an SCCP instance for that, and there is no hard reason to require a NULL SCCP instance, besides the desire to prevent this function from being invoked twice. Change-Id: I28a7362aa838e648ecc9b26ee53dbcade81a9d65
2018-09-26 15:12:23 +00:00
sccp_timer_schedule(conn, &conn->t_int, OSMO_SCCP_TIMER_INT);
}
/* Stop all release related timers: T(rel), T(int) and T(rep_rel) */
static void conn_stop_release_timers(struct sccp_connection *conn)
{
osmo_timer_del(&conn->t_rel);
osmo_timer_del(&conn->t_int);
osmo_timer_del(&conn->t_rep_rel);
}
/* Start connect timer T(conn) */
static void conn_start_connect_timer(struct sccp_connection *conn)
{
make SCCP timers configurable The previous hardcoded SCCP timers may cause SCCP connection releases, if the peer is configured with far lower timers than libosmo-sccp. Testing with a specific SCCPlite MSC, I experienced an iar of just over three minutes, meaning that calls would be cut off by the MSC, since the osmo-bsc failed to send an Inactivity Timer message until seven minutes have passed. With this patch, SCCP timers are configurable by the user. Define constant global default timers, and variable user-configurable timers with each osmo_sccp_instance. Add VTY UI to configure the timers. Users must call osmo_sccp_vty_init() to get the sccp-timer config nodes under the 'cs7' node. Show the new UI in ss7_asp_test.vty. Note that even though this function is not new at all, until recently, all of our SCCP users (osmo-bsc, osmo-msc, osmo-sgsn, osmo-hnbgw) failed to call osmo_sccp_vty_init(), and thus also missed out on the various 'show' commands defined in sccp_vty.c. In other words, to benefit from the timer configurability, the patches to call osmo_sccp_vty_init() must first be merged to the corresponding master branches. If a 'sccp-timer' config command occurs, the cs7 instance must allocate an SCCP instance in order to store the timer config. Do that by calling the recently added osmo_ss7_ensure_sccp() function. Hence remove the limitation that the SCCP instance must not be populated from the "simple" setup function. If we want to configure SCCP timers beforehand, there must be an SCCP instance for that, and there is no hard reason to require a NULL SCCP instance, besides the desire to prevent this function from being invoked twice. Change-Id: I28a7362aa838e648ecc9b26ee53dbcade81a9d65
2018-09-26 15:12:23 +00:00
sccp_timer_schedule(conn, &conn->t_conn, OSMO_SCCP_TIMER_CONN_EST);
}
/* Stop connect timer T(conn) */
static void conn_stop_connect_timer(struct sccp_connection *conn)
{
osmo_timer_del(&conn->t_conn);
}
/***********************************************************************
* SUA Instance and Connection handling
***********************************************************************/
static void conn_destroy(struct sccp_connection *conn);
static struct sccp_connection *conn_find_by_id(const struct osmo_sccp_instance *inst, uint32_t id)
{
struct sccp_connection *conn;
const struct rb_node *node = inst->connections.rb_node;
while (node) {
conn = container_of(node, struct sccp_connection, node);
if (id < conn->conn_id)
node = node->rb_left;
else if (id > conn->conn_id)
node = node->rb_right;
else
return conn;
}
return NULL;
}
static int conn_add_node(struct osmo_sccp_instance *inst, struct sccp_connection *conn)
{
struct rb_node **n = &(inst->connections.rb_node);
struct rb_node *parent = NULL;
while (*n) {
struct sccp_connection *it;
it = container_of(*n, struct sccp_connection, node);
parent = *n;
if (conn->conn_id < it->conn_id) {
n = &((*n)->rb_left);
} else if (conn->conn_id > it->conn_id) {
n = &((*n)->rb_right);
} else {
LOGP(DLSCCP, LOGL_ERROR,
"Trying to reserve already reserved conn_id %u\n", conn->conn_id);
return -EEXIST;
}
}
rb_link_node(&conn->node, parent, n);
rb_insert_color(&conn->node, &inst->connections);
return 0;
}
bool osmo_sccp_conn_id_exists(const struct osmo_sccp_instance *inst, uint32_t id)
{
return conn_find_by_id(inst, id) ? true : false;
}
#define INIT_TIMER(x, fn, priv) do { (x)->cb = fn; (x)->data = priv; } while (0)
/* allocate + init a SCCP Connection with given ID */
static struct sccp_connection *conn_create_id(struct osmo_sccp_user *user, uint32_t conn_id)
{
struct sccp_connection *conn = talloc_zero(user->inst, struct sccp_connection);
char name[16];
conn->conn_id = conn_id;
conn->inst = user->inst;
conn->user = user;
if (conn_add_node(user->inst, conn) < 0) {
talloc_free(conn);
return NULL;
}
INIT_TIMER(&conn->t_conn, conn_tmr_cb, conn);
INIT_TIMER(&conn->t_ias, tx_inact_tmr_cb, conn);
INIT_TIMER(&conn->t_iar, rx_inact_tmr_cb, conn);
INIT_TIMER(&conn->t_rel, rel_tmr_cb, conn);
INIT_TIMER(&conn->t_int, int_tmr_cb, conn);
INIT_TIMER(&conn->t_rep_rel, rep_rel_tmr_cb, conn);
/* this might change at runtime, as it is not a constant :/ */
sccp_scoc_fsm.log_subsys = DLSCCP;
/* we simply use the connection ID as FSM instance name */
snprintf(name, sizeof(name), "%u", conn->conn_id);
conn->fi = osmo_fsm_inst_alloc(&sccp_scoc_fsm, conn, conn,
LOGL_DEBUG, name);
if (!conn->fi) {
rb_erase(&conn->node, &user->inst->connections);
talloc_free(conn);
return NULL;
}
return conn;
}
/* Return an unused SCCP connection ID.
* Callers should check the returned value: on negative return value, there are no unused IDs available.
* \param[in] sccp The SCCP instance to determine a new connection ID for.
* \return unused ID on success (range [0x0, 0x00fffffe]) or negative on elapsed max_attempts without an unused id (<0).
*/
int osmo_sccp_instance_next_conn_id(struct osmo_sccp_instance *sccp)
{
int max_attempts = 0x00FFFFFE;
/* SUA: RFC3868 sec 3.10.4:
* The source reference number is a 4 octet long integer.
* This is allocated by the source SUA instance.
* M3UA/SCCP: ITU-T Q.713 sec 3.3:
* The "source local reference" parameter field is a three-octet field containing a
* reference number which is generated and used by the local node to identify the
* connection section after the connection section is set up.
* The coding "all ones" is reserved for future use.
* Hence, as we currently use the connection ID also as local reference,
* let's simply use 24 bit ids to fit all link types (excluding 0x00ffffff).
*/
while (OSMO_LIKELY((max_attempts--) > 0)) {
/* Optimized modulo operation (% 0x00FFFFFE) using bitwise AND plus CMP: */
sccp->next_id = (sccp->next_id + 1) & 0x00FFFFFF;
if (OSMO_UNLIKELY(sccp->next_id == 0x00FFFFFF))
sccp->next_id = 0;
if (!conn_find_by_id(sccp, sccp->next_id))
return sccp->next_id;
}
return -1;
}
/* Search for next free connection ID and allocate conn */
static struct sccp_connection *conn_create(struct osmo_sccp_user *user)
{
int conn_id = osmo_sccp_instance_next_conn_id(user->inst);
if (conn_id < 0)
return NULL;
return conn_create_id(user, conn_id);
}
static void conn_opt_data_clear_cache(struct sccp_connection *conn)
{
if (conn->opt_data_cache) {
msgb_free(conn->opt_data_cache);
conn->opt_data_cache = NULL;
}
}
/* destroy a SCCP connection state, releasing all timers, terminating
* FSM and releasing associated memory */
static void conn_destroy(struct sccp_connection *conn)
{
conn_opt_data_clear_cache(conn);
conn_stop_connect_timer(conn);
conn_stop_inact_timers(conn);
conn_stop_release_timers(conn);
rb_erase(&conn->node, &conn->inst->connections);
osmo_fsm_inst_term(conn->fi, OSMO_FSM_TERM_REQUEST, NULL);
talloc_free(conn);
}
/* allocate a message buffer for an SCCP User Primitive */
static struct msgb *scu_msgb_alloc(void)
{
return msgb_alloc(SCU_MSGB_SIZE, "SCCP User Primitive");
}
/* generate a RELRE (release request) xua_msg for given conn */
static struct xua_msg *xua_gen_relre(struct sccp_connection *conn,
uint32_t cause,
struct osmo_scu_prim *prim)
{
struct xua_msg *xua = xua_msg_alloc();
if (!xua)
return NULL;
xua->hdr = XUA_HDR(SUA_MSGC_CO, SUA_CO_RELRE);
xua_msg_add_u32(xua, SUA_IEI_ROUTE_CTX, conn->inst->route_ctx);
xua_msg_add_u32(xua, SUA_IEI_DEST_REF, conn->remote_ref);
xua_msg_add_u32(xua, SUA_IEI_SRC_REF, conn->conn_id);
xua_msg_add_u32(xua, SUA_IEI_CAUSE, SUA_CAUSE_T_RELEASE | cause);
/* optional: importance */
if (prim && msgb_l2(prim->oph.msg))
xua_msg_add_data(xua, SUA_IEI_DATA, msgb_l2len(prim->oph.msg),
msgb_l2(prim->oph.msg));
return xua;
}
/* generate xua_msg, encode it and send it to SCRC */
static int xua_gen_relre_and_send(struct sccp_connection *conn, uint32_t cause,
struct osmo_scu_prim *prim)
{
struct xua_msg *xua;
xua = xua_gen_relre(conn, cause, prim);
if (!xua)
return -1;
/* amend this with point code information; The SUA RELRE
* includes neither called nor calling party address! */
xua->mtp.dpc = conn->remote_pc;
sccp_scrc_rx_scoc_conn_msg(conn->inst, xua);
xua_msg_free(xua);
return 0;
}
/* Send cached optional data (if any) from expected message type and clear cache */
static void xua_opt_data_send_cache(struct sccp_connection *conn, int exp_type, uint8_t msg_class)
{
const struct xua_dialect *dialect = &xua_dialect_sua;
const struct xua_msg_class *xmc = dialect->class[msg_class];
if (!conn->opt_data_cache)
return;
if (conn->opt_data_cache->cb[0] != exp_type) {
/* Caller (from the FSM) knows what was the source of Optional Data we're sending.
* Compare this information with source of Optional Data recorded while caching
* to make sure we're on the same page.
*/
LOGP(DLSCCP, LOGL_ERROR, "unexpected message type %s != cache source %s\n",
xua_class_msg_name(xmc, exp_type), xua_class_msg_name(xmc, conn->opt_data_cache->cb[0]));
} else {
osmo_sccp_tx_data(conn->user, conn->conn_id, msgb_data(conn->opt_data_cache), msgb_length(conn->opt_data_cache));
}
conn_opt_data_clear_cache(conn);
}
/* Check if optional data should be dropped, log given error message if so */
static bool xua_drop_data_check_drop(const struct osmo_scu_prim *prim, unsigned lim, const char *message)
{
if (msgb_l2len(prim->oph.msg) > lim) {
LOGP(DLSCCP, LOGL_ERROR,
"%s: dropping optional data with length %u > %u - %s\n",
osmo_scu_prim_name(&prim->oph), msgb_l2len(prim->oph.msg), lim, message);
return true;
}
return false;
}
/* Cache the optional data (if necessary)
* returns true if Optional Data should be kept while encoding the message */
static bool xua_opt_data_cache_keep(struct sccp_connection *conn, const struct osmo_scu_prim *prim, int msg_type)
{
uint8_t *buf;
uint32_t max_optional_data = conn->inst->max_optional_data;
if (xua_drop_data_check_drop(prim, SCCP_MAX_DATA, "cache overrun"))
return false;
if (msgb_l2len(prim->oph.msg) > max_optional_data) {
if (conn->opt_data_cache) {
/* Caching optional data, but there already is optional data occupying the cache: */
LOGP(DLSCCP, LOGL_ERROR, "replacing unsent %u bytes of optional data cache with %s optional data\n",
msgb_length(conn->opt_data_cache), osmo_scu_prim_name(&prim->oph));
msgb_trim(conn->opt_data_cache, 0);
} else {
conn->opt_data_cache = msgb_alloc_c(conn, SCCP_MAX_DATA, "SCCP optional data cache for CR/CC/RLSD");
}
buf = msgb_put(conn->opt_data_cache, msgb_l2len(prim->oph.msg));
memcpy(buf, msgb_l2(prim->oph.msg), msgb_l2len(prim->oph.msg));
conn->opt_data_cache->cb[0] = msg_type;
return false;
}
return true;
}
/* Check optional Data size limit, cache if necessary, return indication whether original opt data should be sent */
static bool xua_opt_data_length_lim(struct sccp_connection *conn, const struct osmo_scu_prim *prim, int msg_type)
{
uint32_t max_optional_data = conn->inst->max_optional_data;
if (!(prim && msgb_l2(prim->oph.msg) && msgb_l2len(prim->oph.msg)))
return false;
switch (msg_type) {
case SUA_CO_CORE: /* §4.2 Connection request (CR) */
case SUA_CO_COAK: /* §4.3 Connection confirm (CC) */
return xua_opt_data_cache_keep(conn, prim, msg_type);
case SUA_CO_COREF: /* §4.4 Connection refused (CREF) */
if (xua_drop_data_check_drop(prim, max_optional_data, "over ITU-T Rec. Q.713 §4.4 limit")) {
/* From the state diagrams in ITU-T Rec Q.714, there's no way to send DT1 neither before nor after CREF
* at this point, so the only option we have is to drop optional data:
* see Figure C.3 / Q.714 (sheet 2 of 6) */
return false;
}
break;
case SUA_CO_RELRE: /* §4.5 Released (RLSD) */
if (msgb_l2len(prim->oph.msg) > max_optional_data) {
if (xua_drop_data_check_drop(prim, SCCP_MAX_DATA, "protocol error"))
return false;
/* There's no need to cache the optional data since the connection is still active at this point:
* Send the Optional Data in a DT1 ahead of the RLSD, because it is too large to be sent in one message.
*/
osmo_sccp_tx_data(conn->user, conn->conn_id, msgb_l2(prim->oph.msg), msgb_l2len(prim->oph.msg));
return false;
}
break;
default:
return true;
}
return true;
}
/* generate a 'struct xua_msg' of requested type from connection +
* primitive data */
static struct xua_msg *xua_gen_msg_co(struct sccp_connection *conn, uint32_t event,
const struct osmo_scu_prim *prim, int msg_type)
{
bool encode_opt_data = xua_opt_data_length_lim(conn, prim, msg_type);
struct xua_msg *xua = xua_msg_alloc();
if (!xua)
return NULL;
switch (msg_type) {
case SUA_CO_CORE: /* Connect Request == SCCP CR */
xua->hdr = XUA_HDR(SUA_MSGC_CO, SUA_CO_CORE);
xua_msg_add_u32(xua, SUA_IEI_ROUTE_CTX, conn->inst->route_ctx);
xua_msg_add_u32(xua, SUA_IEI_PROTO_CLASS, conn->sccp_class);
xua_msg_add_u32(xua, SUA_IEI_SRC_REF, conn->conn_id);
xua_msg_add_sccp_addr(xua, SUA_IEI_DEST_ADDR, &conn->called_addr);
xua_msg_add_u32(xua, SUA_IEI_SEQ_CTRL, 0); /* TODO */
/* optional: sequence number (class 3 only) */
if (conn->calling_addr.presence)
xua_msg_add_sccp_addr(xua, SUA_IEI_SRC_ADDR, &conn->calling_addr);
/* optional: data */
if (encode_opt_data)
xua_msg_add_data(xua, SUA_IEI_DATA, msgb_l2len(prim->oph.msg), msgb_l2(prim->oph.msg));
/* optional: hop count */
/* optional: importance */
break;
case SUA_CO_COAK: /* Connect Acknowledge == SCCP CC */
xua->hdr = XUA_HDR(SUA_MSGC_CO, SUA_CO_COAK);
xua_msg_add_u32(xua, SUA_IEI_ROUTE_CTX, conn->inst->route_ctx);
xua_msg_add_u32(xua, SUA_IEI_PROTO_CLASS, conn->sccp_class);
xua_msg_add_u32(xua, SUA_IEI_DEST_REF, conn->remote_ref);
xua_msg_add_u32(xua, SUA_IEI_SRC_REF, conn->conn_id);
xua_msg_add_u32(xua, SUA_IEI_SEQ_CTRL, 0); /* TODO */
/* optional: sequence number (class 3 only) */
if (conn->called_addr.presence)
xua_msg_add_sccp_addr(xua, SUA_IEI_SRC_ADDR, &conn->called_addr);
/* optional: hop count; importance; priority */
/* FIXME: destination address will [only] be present in
* case the CORE message conveys the source address
* parameter */
if (conn->calling_addr.presence)
xua_msg_add_sccp_addr(xua, SUA_IEI_DEST_ADDR, &conn->calling_addr);
/* optional: data */
if (encode_opt_data)
xua_msg_add_data(xua, SUA_IEI_DATA, msgb_l2len(prim->oph.msg), msgb_l2(prim->oph.msg));
/* optional: importance */
break;
case SUA_CO_RELRE: /* Release Request == SCCP RLSD */
if (!prim)
goto prim_needed;
xua->hdr = XUA_HDR(SUA_MSGC_CO, SUA_CO_RELRE);
xua_msg_add_u32(xua, SUA_IEI_ROUTE_CTX, conn->inst->route_ctx);
xua_msg_add_u32(xua, SUA_IEI_DEST_REF, conn->remote_ref);
xua_msg_add_u32(xua, SUA_IEI_SRC_REF, conn->conn_id);
xua_msg_add_u32(xua, SUA_IEI_CAUSE, SUA_CAUSE_T_RELEASE | prim->u.disconnect.cause);
/* optional: data */
if (encode_opt_data)
xua_msg_add_data(xua, SUA_IEI_DATA, msgb_l2len(prim->oph.msg), msgb_l2(prim->oph.msg));
/* optional: importance */
break;
case SUA_CO_RELCO: /* Release Confirm == SCCP RLC */
xua->hdr = XUA_HDR(SUA_MSGC_CO, SUA_CO_RELCO);
xua_msg_add_u32(xua, SUA_IEI_ROUTE_CTX, conn->inst->route_ctx);
xua_msg_add_u32(xua, SUA_IEI_DEST_REF, conn->remote_ref);
xua_msg_add_u32(xua, SUA_IEI_SRC_REF, conn->conn_id);
break;
case SUA_CO_CODT: /* Connection Oriented Data Transfer == SCCP DT1 */
if (!prim)
goto prim_needed;
xua->hdr = XUA_HDR(SUA_MSGC_CO, SUA_CO_CODT);
xua_msg_add_u32(xua, SUA_IEI_ROUTE_CTX, conn->inst->route_ctx);
/* Sequence number only in expedited data */
xua_msg_add_u32(xua, SUA_IEI_DEST_REF, conn->remote_ref);
/* optional: priority; correlation id */
xua_msg_add_data(xua, SUA_IEI_DATA, msgb_l2len(prim->oph.msg),
msgb_l2(prim->oph.msg));
break;
case SUA_CO_COIT: /* Connection Oriented Interval Timer == SCCP IT */
xua->hdr = XUA_HDR(SUA_MSGC_CO, SUA_CO_COIT);
xua_msg_add_u32(xua, SUA_IEI_ROUTE_CTX, conn->inst->route_ctx);
xua_msg_add_u32(xua, SUA_IEI_PROTO_CLASS, conn->sccp_class);
xua_msg_add_u32(xua, SUA_IEI_SRC_REF, conn->conn_id);
xua_msg_add_u32(xua, SUA_IEI_DEST_REF, conn->remote_ref);
/* optional: sequence number; credit (both class 3 only) */
break;
case SUA_CO_COREF: /* Connect Refuse == SCCP CREF */
xua->hdr = XUA_HDR(SUA_MSGC_CO, SUA_CO_COREF);
xua_msg_add_u32(xua, SUA_IEI_ROUTE_CTX, conn->inst->route_ctx);
xua_msg_add_u32(xua, SUA_IEI_DEST_REF, conn->remote_ref);
//xua_msg_add_u32(xua, SUA_IEI_CAUSE, SUA_CAUSE_T_REFUSAL | prim->u.disconnect.cause);
xua_msg_add_u32(xua, SUA_IEI_CAUSE, SUA_CAUSE_T_REFUSAL | SCCP_REFUSAL_UNEQUIPPED_USER);
/* optional: source addr */
if (conn->called_addr.presence)
xua_msg_add_sccp_addr(xua, SUA_IEI_SRC_ADDR, &conn->called_addr);
/* conditional: dest addr */
if (conn->calling_addr.presence)
xua_msg_add_sccp_addr(xua, SUA_IEI_DEST_ADDR, &conn->calling_addr);
/* optional: data */
if (encode_opt_data)
xua_msg_add_data(xua, SUA_IEI_DATA, msgb_l2len(prim->oph.msg), msgb_l2(prim->oph.msg));
/* optional: importance */
break;
/* FIXME */
default:
LOGP(DLSCCP, LOGL_ERROR, "Don't know how to encode msg_type %u\n", msg_type);
xua_msg_free(xua);
return NULL;
}
return xua;
prim_needed:
xua_msg_free(xua);
LOGP(DLSCCP, LOGL_ERROR, "%s must be called with valid 'prim' "
"pointer for msg_type=%u\n", __func__, msg_type);
return NULL;
}
/* generate xua_msg, encode it and send it to SCRC
* returns 0 on success, negative on error
*/
static int xua_gen_encode_and_send(struct sccp_connection *conn, uint32_t event,
const struct osmo_scu_prim *prim, int msg_type)
{
struct xua_msg *xua;
xua = xua_gen_msg_co(conn, event, prim, msg_type);
if (!xua)
return -ENOMEM;
/* amend this with point code information; Many CO msgs
* includes neither called nor calling party address! */
xua->mtp.dpc = conn->remote_pc;
sccp_scrc_rx_scoc_conn_msg(conn->inst, xua);
xua_msg_free(xua);
return 0;
}
/* allocate a SCU primitive to be sent to the user */
static struct osmo_scu_prim *scu_prim_alloc(unsigned int primitive, enum osmo_prim_operation operation)
{
struct msgb *upmsg = scu_msgb_alloc();
struct osmo_scu_prim *prim;
prim = (struct osmo_scu_prim *) msgb_put(upmsg, sizeof(*prim));
osmo_prim_init(&prim->oph, SCCP_SAP_USER,
primitive, operation, upmsg);
upmsg->l2h = upmsg->tail;
return prim;
}
/* high-level function to generate a SCCP User primitive of requested
* type based on the connection and currently processed XUA message */
static void scu_gen_encode_and_send(struct sccp_connection *conn, uint32_t event,
struct xua_msg *xua, unsigned int primitive,
enum osmo_prim_operation operation)
{
struct osmo_scu_prim *scu_prim;
struct osmo_scu_disconn_param *udisp;
struct osmo_scu_connect_param *uconp;
struct osmo_scu_data_param *udatp;
struct xua_msg_part *data_ie;
scu_prim = scu_prim_alloc(primitive, operation);
switch (OSMO_PRIM_HDR(&scu_prim->oph)) {
case OSMO_PRIM(OSMO_SCU_PRIM_N_DISCONNECT, PRIM_OP_INDICATION):
udisp = &scu_prim->u.disconnect;
udisp->conn_id = conn->conn_id;
udisp->responding_addr = conn->called_addr;
udisp->importance = conn->importance;
udisp->originator = OSMO_SCCP_ORIG_UNDEFINED;
//udisp->in_sequence_control;
if (xua) {
udisp->cause = xua_msg_get_u32(xua, SUA_IEI_CAUSE);
if (xua_msg_find_tag(xua, SUA_IEI_SRC_ADDR))
sua_addr_parse(&udisp->responding_addr, xua, SUA_IEI_SRC_ADDR);
data_ie = xua_msg_find_tag(xua, SUA_IEI_DATA);
udisp->importance = xua_msg_get_u32(xua, SUA_IEI_IMPORTANCE);
if (data_ie) {
struct msgb *upmsg = scu_prim->oph.msg;
upmsg->l2h = msgb_put(upmsg, data_ie->len);
memcpy(upmsg->l2h, data_ie->dat, data_ie->len);
}
}
break;
case OSMO_PRIM(OSMO_SCU_PRIM_N_CONNECT, PRIM_OP_INDICATION):
uconp = &scu_prim->u.connect;
uconp->conn_id = conn->conn_id;
uconp->called_addr = conn->called_addr;
uconp->calling_addr = conn->calling_addr;
uconp->sccp_class = conn->sccp_class;
uconp->importance = conn->importance;
if (xua) {
data_ie = xua_msg_find_tag(xua, SUA_IEI_DATA);
if (data_ie) {
struct msgb *upmsg = scu_prim->oph.msg;
upmsg->l2h = msgb_put(upmsg, data_ie->len);
memcpy(upmsg->l2h, data_ie->dat, data_ie->len);
}
}
break;
case OSMO_PRIM(OSMO_SCU_PRIM_N_CONNECT, PRIM_OP_CONFIRM):
uconp = &scu_prim->u.connect;
uconp->conn_id = conn->conn_id;
uconp->called_addr = conn->called_addr;
uconp->calling_addr = conn->calling_addr;
//scu_prim->u.connect.in_sequence_control
uconp->sccp_class = xua_msg_get_u32(xua, SUA_IEI_PROTO_CLASS) & 3;
uconp->importance = xua_msg_get_u32(xua, SUA_IEI_IMPORTANCE);
data_ie = xua_msg_find_tag(xua, SUA_IEI_DATA);
if (data_ie) {
struct msgb *upmsg = scu_prim->oph.msg;
upmsg->l2h = msgb_put(upmsg, data_ie->len);
memcpy(upmsg->l2h, data_ie->dat, data_ie->len);
}
break;
case OSMO_PRIM(OSMO_SCU_PRIM_N_DATA, PRIM_OP_INDICATION):
udatp = &scu_prim->u.data;
udatp->conn_id = conn->conn_id;
udatp->importance = conn->importance;
data_ie = xua_msg_find_tag(xua, SUA_IEI_DATA);
if (data_ie) {
struct msgb *upmsg = scu_prim->oph.msg;
upmsg->l2h = msgb_put(upmsg, data_ie->len);
memcpy(upmsg->l2h, data_ie->dat, data_ie->len);
}
break;
default:
LOGPFSML(conn->fi, LOGL_ERROR, "Unsupported primitive %u:%u\n",
scu_prim->oph.primitive, scu_prim->oph.operation);
talloc_free(scu_prim->oph.msg);
return;
}
sccp_user_prim_up(conn->user, scu_prim);
}
/***********************************************************************
* Actual SCCP Connection Oriented Control (SCOC) Finite State Machine
***********************************************************************/
/* Figure C.2/Q.714 (sheet 1 of 7) and C.3/Q.714 (sheet 1 of 6) */
static void scoc_fsm_idle(struct osmo_fsm_inst *fi, uint32_t event, void *data)
{
struct sccp_connection *conn = fi->priv;
struct osmo_scu_prim *prim = NULL;
struct osmo_scu_connect_param *uconp;
struct xua_msg *xua = NULL;
int rc;
switch (event) {
case SCOC_E_SCU_N_CONN_REQ:
prim = data;
uconp = &prim->u.connect;
/* copy relevant parameters from prim to conn */
conn->called_addr = uconp->called_addr;
conn->calling_addr = uconp->calling_addr;
conn->sccp_class = uconp->sccp_class;
/* generate + send CR PDU to SCRC */
rc = xua_gen_encode_and_send(conn, event, prim, SUA_CO_CORE);
if (rc < 0)
LOGPFSML(fi, LOGL_ERROR, "Failed to initiate connection: %s\n", strerror(-rc));
else {
/* start connection timer */
conn_start_connect_timer(conn);
osmo_fsm_inst_state_chg(fi, S_CONN_PEND_OUT, 0, 0);
}
break;
#if 0
case SCOC_E_SCU_N_TYPE1_REQ:
/* ?!? */
break;
#endif
case SCOC_E_RCOC_RLSD_IND:
/* send release complete to SCRC */
xua_gen_encode_and_send(conn, event, NULL, SUA_CO_RELCO);
break;
case SCOC_E_RCOC_REL_COMPL_IND:
/* do nothing */
break;
case SCOC_E_RCOC_OTHER_NPDU:
#if 0
if (src_ref) {
/* FIXME: send ERROR to SCRC */
}
#endif
break;
/* destination node / incoming connection */
/* Figure C.3 / Q.714 (sheet 1 of 6) */
case SCOC_E_RCOC_CONN_IND:
xua = data;
/* copy relevant parameters from xua to conn */
sua_addr_parse(&conn->calling_addr, xua, SUA_IEI_SRC_ADDR);
sua_addr_parse(&conn->called_addr, xua, SUA_IEI_DEST_ADDR);
conn->remote_ref = xua_msg_get_u32(xua, SUA_IEI_SRC_REF);
conn->sccp_class = xua_msg_get_u32(xua, SUA_IEI_PROTO_CLASS) & 3;
conn->importance = xua_msg_get_u32(xua, SUA_IEI_IMPORTANCE);
/* 3.1.6.1 The originating node of the CR message
* (identified by the OPC in the calling party address
* or by default by the OPC in the MTP label, [and the
* MTP-SAP instance]) is associated with the incoming
* connection section. */
if (conn->calling_addr.presence & OSMO_SCCP_ADDR_T_PC)
conn->remote_pc = conn->calling_addr.pc;
else {
/* Hack to get the MTP label here ?!? */
conn->remote_pc = xua->mtp.opc;
}
osmo_fsm_inst_state_chg(fi, S_CONN_PEND_IN, 0, 0);
/* N-CONNECT.ind to User */
scu_gen_encode_and_send(conn, event, xua, OSMO_SCU_PRIM_N_CONNECT,
PRIM_OP_INDICATION);
break;
}
}
static void scoc_fsm_idle_onenter(struct osmo_fsm_inst *fi, uint32_t old_state)
{
conn_destroy(fi->priv);
}
/* Figure C.3 / Q.714 (sheet 2 of 6) */
static void scoc_fsm_conn_pend_in(struct osmo_fsm_inst *fi, uint32_t event, void *data)
{
struct sccp_connection *conn = fi->priv;
struct osmo_scu_prim *prim = NULL;
switch (event) {
case SCOC_E_SCU_N_CONN_RESP:
prim = data;
/* FIXME: assign local reference (only now?) */
/* FIXME: assign sls, protocol class and credit */
xua_gen_encode_and_send(conn, event, prim, SUA_CO_COAK);
/* start inactivity timers */
conn_start_inact_timers(conn);
osmo_fsm_inst_state_chg(fi, S_ACTIVE, 0, 0);
xua_opt_data_send_cache(conn, SUA_CO_COAK, SUA_MSGC_CO);
break;
case SCOC_E_SCU_N_DISC_REQ:
prim = data;
/* release resources: implicit */
xua_gen_encode_and_send(conn, event, prim, SUA_CO_COREF);
/* N. B: we've ignored CREF sending errors as there's no recovery option anyway */
osmo_fsm_inst_state_chg(fi, S_IDLE, 0, 0);
break;
}
}
/* Figure C.2/Q.714 (sheet 2 of 7) */
static void scoc_fsm_conn_pend_out(struct osmo_fsm_inst *fi, uint32_t event, void *data)
{
struct sccp_connection *conn = fi->priv;
struct osmo_scu_prim *prim = NULL;
struct xua_msg *xua = NULL;
switch (event) {
case SCOC_E_SCU_N_DISC_REQ:
prim = data;
conn->release_cause = prim->u.disconnect.cause;
osmo_fsm_inst_state_chg(fi, S_WAIT_CONN_CONF, 0, 0);
/* keep conn timer running(!) */
break;
case SCOC_E_CONN_TMR_EXP:
/* N-DISCONNECT.ind to user */
scu_gen_encode_and_send(conn, event, NULL, OSMO_SCU_PRIM_N_DISCONNECT,
PRIM_OP_INDICATION);
/* below implicitly releases resources + local ref */
osmo_fsm_inst_state_chg(fi, S_IDLE, 0, 0);
break;
case SCOC_E_RCOC_ROUT_FAIL_IND:
case SCOC_E_RCOC_CREF_IND:
xua = data;
/* stop conn timer */
conn_stop_connect_timer(conn);
/* release local res + ref (implicit by going to idle) */
/* N-DISCONNECT.ind to user */
scu_gen_encode_and_send(conn, event, xua, OSMO_SCU_PRIM_N_DISCONNECT,
PRIM_OP_INDICATION);
/* below implicitly releases resources + local ref */
osmo_fsm_inst_state_chg(fi, S_IDLE, 0, 0);
break;
case SCOC_E_RCOC_RLSD_IND:
xua = data;
/* RLC to SCRC */
xua_gen_encode_and_send(conn, event, NULL, SUA_CO_RELCO);
/* stop conn timer */
conn_stop_connect_timer(conn);
/* release local res + ref (implicit) */
/* N-DISCONNECT.ind to user */
scu_gen_encode_and_send(conn, event, xua, OSMO_SCU_PRIM_N_DISCONNECT,
PRIM_OP_INDICATION);
osmo_fsm_inst_state_chg(fi, S_IDLE, 0, 0);
break;
case SCOC_E_RCOC_OTHER_NPDU:
xua = data;
conn_start_connect_timer(conn);
/* release local res + ref (implicit) */
/* N-DISCONNECT.ind to user */
scu_gen_encode_and_send(conn, event, xua, OSMO_SCU_PRIM_N_DISCONNECT,
PRIM_OP_INDICATION);
osmo_fsm_inst_state_chg(fi, S_IDLE, 0, 0);
break;
case SCOC_E_RCOC_CC_IND:
xua = data;
/* stop conn timer */
conn_stop_connect_timer(conn);
/* start inactivity timers */
conn_start_inact_timers(conn);
/* TODO: assign PCU and credit */
/* associate remote ref to conn */
conn->remote_ref = xua_msg_get_u32(xua, SUA_IEI_SRC_REF);
/* 3.1.4.2 The node sending the CC message (identified
* by the parameter OPC contained in the
* MTP-TRANSFER.indication primitive which conveyed the
* CC message [plus the MTP-SAP instance]) is associated
* with the connection section. */
conn->remote_pc = xua->mtp.opc;
osmo_fsm_inst_state_chg(fi, S_ACTIVE, 0, 0);
/* If CR which was used to initiate this connection had excessive Optional Data which we had to cache,
* now is the time to send it: the connection is already active but we hadn't notified upper layers about it
* so we have the connection all to ourselves and can use it to transmit "leftover" data via DT1 */
xua_opt_data_send_cache(conn, SUA_CO_CORE, xua->hdr.msg_class);
/* N-CONNECT.conf to user */
scu_gen_encode_and_send(conn, event, xua, OSMO_SCU_PRIM_N_CONNECT,
PRIM_OP_CONFIRM);
break;
}
}
/* Figure C.2/Q.714 (sheet 3 of 7) */
static void scoc_fsm_wait_conn_conf(struct osmo_fsm_inst *fi, uint32_t event, void *data)
{
struct sccp_connection *conn = fi->priv;
struct xua_msg *xua = NULL;
switch (event) {
case SCOC_E_RCOC_RLSD_IND:
xua = data;
/* release complete to SCRC */
xua_gen_encode_and_send(conn, event, NULL, SUA_CO_RELCO);
/* stop conn timer */
conn_stop_connect_timer(conn);
/* release local res + ref (implicit) */
osmo_fsm_inst_state_chg(fi, S_IDLE, 0, 0);
break;
case SCOC_E_RCOC_CC_IND:
xua = data;
/* stop conn timer */
conn_stop_connect_timer(conn);
/* associate rem ref to conn */
conn->remote_ref = xua_msg_get_u32(xua, SUA_IEI_SRC_REF);
sccp_scoc: Fix remote PC not assigned preventing RSLD to be sent in st WAIT_CONN_CONF Scenario: RUA Connect triggers SCCP CR towards peer, and we move to CONN_PEND_OUT state where we expect to receive CC. However, if some timer (like X31) times out before we receive CC (eg because CC takes a long time to come), we end up in state WAIT_CONN_CONF. In that state, according to Figure C.2/Q.714 (sheet 2 of 7), among other possibilite we wait for CC, and if it arrives, we send an RLSD to the peer to inform him that we released the conn, and wait for the peer to ack the RLSD, then release all state. Given the scenario above, scoc_fsm_wait_conn_conf() was not assigning the received OPC from the CC to the conn->remote_pc (unlike scoc_fsm_conn_pend_out() which does it properly). As a result, when trying to send teh RLSD it would fail and never transmit RLSD, taking then a long time to release through T(rel) (10-20 seconds), and probably longer in the peer (T(iar) or similar?). Rua Connect triggers tx of SCCP CC: Received SCCP User Primitive (N-CONNECT.request) XUA_AS(as-clnt-msc-0)[0x55f11c7df980]{AS_ACTIVE}: Received Event AS-TRANSFER.req //Tx CC SCCP-SCOC(929)[0x55f11c909c90]{IDLE}: State change to CONN_PEND_OUT (no timeout) ... X31 timeout triggers state change: map_sccp(...-SCCP-929)[0x55f11c909820]{wait_cc}: Timeout of X31 SCCP-SCOC(929)[0x55f11c909c90]{CONN_PEND_OUT}: Received Event N-DISCONNECT.req SCCP-SCOC(929)[0x55f11c909c90]{CONN_PEND_OUT}: State change to WAIT_CONN_CONF (no timeout) ... CC arrives, but conn_remote_pc is not properly assigned and tx of RLSD fails: SCCP-SCOC(929)[0x55f11c909c90]{WAIT_CONN_CONF}: Received Event RCOC-CONNECT_CONFIRM.ind MTP-TRANSFER.req from SCCP without DPC?!? called=RI=0 // PROBLEM HERE!!!! SCCP-SCOC(929)[0x55f11c909c90]{WAIT_CONN_CONF}: State change to DISCONN_PEND (no timeout) ... SCCP-SCOC(929)[0x55f11c909c90]{DISCONN_PEND}: Received Event T(rel)_expired Related: SYS#6616 Change-Id: I9f9f78c92dd95f38af7b03037e60a1c993d7e5b0
2023-10-30 16:21:54 +00:00
/* 3.1.4.2 The node sending the CC message (identified
* by the parameter OPC contained in the
* MTP-TRANSFER.indication primitive which conveyed the
* CC message [plus the MTP-SAP instance]) is associated
* with the connection section. */
conn->remote_pc = xua->mtp.opc;
/* released to SCRC */
xua_gen_relre_and_send(conn, conn->release_cause, NULL);
/* start rel timer */
conn_start_rel_timer(conn);
osmo_fsm_inst_state_chg(fi, S_DISCONN_PEND, 0, 0);
break;
case SCOC_E_RCOC_OTHER_NPDU:
case SCOC_E_RCOC_CREF_IND:
case SCOC_E_RCOC_ROUT_FAIL_IND:
xua = data;
/* stop conn timer */
conn_stop_connect_timer(conn);
/* release local res + ref */
osmo_fsm_inst_state_chg(fi, S_IDLE, 0, 0);
break;
case SCOC_E_CONN_TMR_EXP:
/* release local res + ref */
osmo_fsm_inst_state_chg(fi, S_IDLE, 0, 0);
break;
}
}
/* C.2/Q.714 (sheet 4+5 of 7) and C.3/Q714 (sheet 3+4 of 6) */
static void scoc_fsm_active(struct osmo_fsm_inst *fi, uint32_t event, void *data)
{
struct xua_msg *xua = data;
struct sccp_connection *conn = fi->priv;
struct osmo_scu_prim *prim = NULL;
switch (event) {
#pragma message ("TODO: internal disco: send N-DISCONNECT.ind to user")
/* send N-DISCONNECT.ind to user */
/*scu_gen_encode_and_send(conn, event, xua, OSMO_SCU_PRIM_N_DISCONNECT,
PRIM_OP_INDICATION);*/
/* fall-through */
case SCOC_E_SCU_N_DISC_REQ:
prim = data;
/* stop inact timers */
conn_stop_inact_timers(conn);
/* send RLSD to SCRC */
xua_gen_encode_and_send(conn, event, prim, SUA_CO_RELRE);
/* start rel timer */
conn_start_rel_timer(conn);
osmo_fsm_inst_state_chg(fi, S_DISCONN_PEND, 0, 0);
break;
case SCOC_E_RCOC_CREF_IND:
case SCOC_E_RCOC_CC_IND:
case SCOC_E_RCOC_REL_COMPL_IND:
/* do nothing */
break;
case SCOC_E_RCOC_RLSD_IND:
/* send N-DISCONNECT.ind to user */
scu_gen_encode_and_send(conn, event, xua, OSMO_SCU_PRIM_N_DISCONNECT,
PRIM_OP_INDICATION);
/* release res + local ref (implicit) */
/* stop inact timers */
conn_stop_inact_timers(conn);
/* RLC to SCRC */
xua_gen_encode_and_send(conn, event, NULL, SUA_CO_RELCO);
osmo_fsm_inst_state_chg(fi, S_IDLE, 0, 0);
break;
case SCOC_E_RCOC_ERROR_IND:
xua = data;
/* FIXME: check for cause service_class_mismatch */
/* release res + local ref (implicit) */
/* send N-DISCONNECT.ind to user */
scu_gen_encode_and_send(conn, event, xua, OSMO_SCU_PRIM_N_DISCONNECT,
PRIM_OP_INDICATION);
/* stop inact timers */
conn_stop_inact_timers(conn);
osmo_fsm_inst_state_chg(fi, S_IDLE, 0, 0);
break;
case SCOC_E_T_IAR_EXP:
/* stop inact timers */
conn_stop_inact_timers(conn);
xua = xua_msg_alloc();
xua_msg_add_u32(xua, SUA_IEI_CAUSE,
SUA_CAUSE_T_RELEASE | SCCP_RELEASE_CAUSE_EXPIRATION_INACTIVE);
xua_msg_add_u32(xua, SUA_IEI_IMPORTANCE, conn->importance);
/* Send N-DISCONNECT.ind to local user */
scu_gen_encode_and_send(conn, event, xua, OSMO_SCU_PRIM_N_DISCONNECT,
PRIM_OP_INDICATION);
talloc_free(xua);
/* Send RLSD to peer */
xua_gen_relre_and_send(conn, SCCP_RELEASE_CAUSE_EXPIRATION_INACTIVE, NULL);
/* start release timer */
conn_start_rel_timer(conn);
osmo_fsm_inst_state_chg(fi, S_DISCONN_PEND, 0, 0);
break;
case SCOC_E_RCOC_ROUT_FAIL_IND:
/* send N-DISCONNECT.ind to user */
scu_gen_encode_and_send(conn, event, NULL, OSMO_SCU_PRIM_N_DISCONNECT,
PRIM_OP_INDICATION);
/* stop inact timers */
conn_stop_inact_timers(conn);
/* start release timer */
conn_start_rel_timer(conn);
osmo_fsm_inst_state_chg(fi, S_DISCONN_PEND, 0, 0);
break;
/* Figure C.4/Q.714 */
case SCOC_E_SCU_N_DATA_REQ:
case SCOC_E_SCU_N_EXP_DATA_REQ:
prim = data;
xua_gen_encode_and_send(conn, event, prim, SUA_CO_CODT);
conn_restart_tx_inact_timer(conn);
break;
case SCOC_E_RCOC_DT1_IND:
/* restart receive inactivity timer */
conn_restart_rx_inact_timer(conn);
/* TODO: M-bit */
scu_gen_encode_and_send(conn, event, xua, OSMO_SCU_PRIM_N_DATA,
PRIM_OP_INDICATION);
break;
/* Figure C.4/Q.714 (sheet 4 of 4) */
case SCOC_E_RCOC_IT_IND:
xua = data;
/* check if remote reference is what we expect */
/* check class is what we expect */
if (xua_msg_get_u32(xua, SUA_IEI_SRC_REF) != conn->remote_ref ||
xua_msg_get_u32(xua, SUA_IEI_PROTO_CLASS) != conn->sccp_class) {
/* Release connection */
/* Stop inactivity Timers */
conn_stop_inact_timers(conn);
xua = xua_msg_alloc();
xua_msg_add_u32(xua, SUA_IEI_CAUSE,
SUA_CAUSE_T_RELEASE | SCCP_RELEASE_CAUSE_INCONSISTENT_CONN_DATA);
xua_msg_add_u32(xua, SUA_IEI_IMPORTANCE, conn->importance);
/* send N-DISCONNECT.ind to user */
scu_gen_encode_and_send(conn, event, xua,
OSMO_SCU_PRIM_N_DISCONNECT,
PRIM_OP_INDICATION);
talloc_free(xua);
/* Send RLSD to SCRC */
xua_gen_relre_and_send(conn, SCCP_RELEASE_CAUSE_INCONSISTENT_CONN_DATA, NULL);
talloc_free(xua);
/* Start release timer */
conn_start_rel_timer(conn);
osmo_fsm_inst_state_chg(fi, S_DISCONN_PEND, 0, 0);
}
conn_restart_rx_inact_timer(conn);
break;
case SCOC_E_T_IAS_EXP:
/* Send IT to peer */
xua_gen_encode_and_send(conn, event, NULL, SUA_CO_COIT);
conn_restart_tx_inact_timer(conn);
break;
}
}
/* C.2/Q.714 (sheet 6+7 of 7) and C.3/Q.714 (sheet 5+6 of 6) */
static void scoc_fsm_disconn_pend(struct osmo_fsm_inst *fi, uint32_t event, void *data)
{
struct sccp_connection *conn = fi->priv;
switch (event) {
case SCOC_E_RCOC_REL_COMPL_IND:
case SCOC_E_RCOC_RLSD_IND:
/* release res + local ref (implicit) */
/* freeze local ref */
/* stop release + interval timers */
conn_stop_release_timers(conn);
osmo_fsm_inst_state_chg(fi, S_IDLE, 0, 0);
break;
case SCOC_E_RCOC_ROUT_FAIL_IND:
case SCOC_E_RCOC_OTHER_NPDU:
/* do nothing */
break;
case SCOC_E_T_REL_EXP: /* release timer exp */
/* send RLSD */
xua_gen_relre_and_send(conn, SCCP_RELEASE_CAUSE_UNQUALIFIED, NULL);
/* start interval timer */
conn_start_int_timer(conn);
/* start repeat release timer */
conn_start_rep_rel_timer(conn);
break;
case SCOC_E_T_INT_EXP: /* interval timer exp */
/* TODO: Inform maintenance */
/* stop release and interval timers */
conn_stop_release_timers(conn);
osmo_fsm_inst_state_chg(fi, S_IDLE, 0, 0);
break;
case SCOC_E_T_REP_REL_EXP: /* repeat release timer exp */
/* send RLSD */
xua_gen_relre_and_send(conn, SCCP_RELEASE_CAUSE_UNQUALIFIED, NULL);
/* re-start repeat release timer */
conn_start_rep_rel_timer(conn);
break;
}
}
static const struct osmo_fsm_state sccp_scoc_states[] = {
[S_IDLE] = {
.name = "IDLE",
.action = scoc_fsm_idle,
.onenter= scoc_fsm_idle_onenter,
.in_event_mask = S(SCOC_E_SCU_N_CONN_REQ) |
//S(SCOC_E_SCU_N_TYPE1_REQ) |
S(SCOC_E_RCOC_CONN_IND) |
S(SCOC_E_RCOC_RLSD_IND) |
S(SCOC_E_RCOC_REL_COMPL_IND) |
S(SCOC_E_RCOC_OTHER_NPDU),
.out_state_mask = S(S_CONN_PEND_OUT) |
S(S_CONN_PEND_IN),
},
[S_CONN_PEND_IN] = {
.name = "CONN_PEND_IN",
.action = scoc_fsm_conn_pend_in,
.in_event_mask = S(SCOC_E_SCU_N_CONN_RESP) |
S(SCOC_E_SCU_N_DISC_REQ),
.out_state_mask = S(S_IDLE) |
S(S_ACTIVE),
},
[S_CONN_PEND_OUT] = {
.name = "CONN_PEND_OUT",
.action = scoc_fsm_conn_pend_out,
.in_event_mask = S(SCOC_E_SCU_N_DISC_REQ) |
S(SCOC_E_CONN_TMR_EXP) |
S(SCOC_E_RCOC_ROUT_FAIL_IND) |
S(SCOC_E_RCOC_RLSD_IND) |
S(SCOC_E_RCOC_OTHER_NPDU) |
S(SCOC_E_RCOC_CREF_IND) |
S(SCOC_E_RCOC_CC_IND),
.out_state_mask = S(S_IDLE) |
S(S_ACTIVE) |
S(S_WAIT_CONN_CONF),
},
[S_ACTIVE] = {
.name = "ACTIVE",
.action = scoc_fsm_active,
.in_event_mask = S(SCOC_E_SCU_N_DISC_REQ) |
/* internal disconnect */
S(SCOC_E_RCOC_CREF_IND) |
S(SCOC_E_RCOC_REL_COMPL_IND) |
S(SCOC_E_RCOC_RLSD_IND) |
S(SCOC_E_RCOC_ERROR_IND) |
S(SCOC_E_T_IAR_EXP) |
S(SCOC_E_T_IAS_EXP) |
S(SCOC_E_RCOC_ROUT_FAIL_IND) |
S(SCOC_E_SCU_N_DATA_REQ) |
S(SCOC_E_SCU_N_EXP_DATA_REQ) |
S(SCOC_E_RCOC_DT1_IND) |
S(SCOC_E_RCOC_IT_IND),
.out_state_mask = S(S_IDLE) |
S(S_DISCONN_PEND),
},
[S_DISCONN_PEND] = {
.name = "DISCONN_PEND",
.action = scoc_fsm_disconn_pend,
.in_event_mask = S(SCOC_E_RCOC_REL_COMPL_IND) |
S(SCOC_E_RCOC_RLSD_IND) |
S(SCOC_E_RCOC_ROUT_FAIL_IND) |
S(SCOC_E_RCOC_OTHER_NPDU) |
S(SCOC_E_T_REL_EXP) |
S(SCOC_E_T_INT_EXP) |
S(SCOC_E_T_REP_REL_EXP),
.out_state_mask = S(S_IDLE),
},
[S_RESET_IN] = {
.name = "RESET_IN",
},
[S_RESET_OUT] = {
.name = "RESET_OUT",
},
[S_BOTHWAY_RESET] = {
.name = "BOTHWAY_RESET",
},
[S_WAIT_CONN_CONF] = {
.name = "WAIT_CONN_CONF",
.action = scoc_fsm_wait_conn_conf,
.in_event_mask = S(SCOC_E_RCOC_RLSD_IND) |
S(SCOC_E_RCOC_CC_IND) |
S(SCOC_E_RCOC_OTHER_NPDU) |
S(SCOC_E_CONN_TMR_EXP) |
S(SCOC_E_RCOC_CREF_IND) |
S(SCOC_E_RCOC_ROUT_FAIL_IND),
.out_state_mask = S(S_IDLE) |
S(S_DISCONN_PEND),
},
};
struct osmo_fsm sccp_scoc_fsm = {
.name = "SCCP-SCOC",
.states = sccp_scoc_states,
.num_states = ARRAY_SIZE(sccp_scoc_states),
/* ".log_subsys = DLSCCP" doesn't work as DLSCCP is not a constant */
.event_names = scoc_event_names,
};
/* map from SCCP return cause to SCCP Refusal cause */
static const uint8_t cause_map_cref[] = {
[SCCP_RETURN_CAUSE_SUBSYSTEM_CONGESTION] =
SCCP_REFUSAL_SUBSYTEM_CONGESTION,
[SCCP_RETURN_CAUSE_SUBSYSTEM_FAILURE] =
SCCP_REFUSAL_SUBSYSTEM_FAILURE,
[SCCP_RETURN_CAUSE_UNEQUIPPED_USER] =
SCCP_REFUSAL_UNEQUIPPED_USER,
[SCCP_RETURN_CAUSE_UNQUALIFIED] =
SCCP_REFUSAL_UNQUALIFIED,
[SCCP_RETURN_CAUSE_SCCP_FAILURE] =
SCCP_REFUSAL_SCCP_FAILURE,
[SCCP_RETURN_CAUSE_HOP_COUNTER_VIOLATION] =
SCCP_REFUSAL_HOP_COUNTER_VIOLATION,
};
static uint8_t get_cref_cause_for_ret(uint8_t ret_cause)
{
if (ret_cause < ARRAY_SIZE(cause_map_cref))
return cause_map_cref[ret_cause];
else
return SCCP_REFUSAL_UNQUALIFIED;
}
/* Generate a COREF message purely based on an incoming SUA message,
* without the use of any local connection state */
static struct xua_msg *gen_coref_without_conn(struct osmo_sccp_instance *inst,
struct xua_msg *xua_in,
uint32_t ref_cause)
{
struct xua_msg *xua;
xua = xua_msg_alloc();
xua->hdr = XUA_HDR(SUA_MSGC_CO, SUA_CO_COREF);
xua_msg_add_u32(xua, SUA_IEI_ROUTE_CTX, inst->route_ctx);
xua_msg_copy_part(xua, SUA_IEI_DEST_REF, xua_in, SUA_IEI_SRC_REF);
xua_msg_add_u32(xua, SUA_IEI_CAUSE, SUA_CAUSE_T_REFUSAL | ref_cause);
/* optional: source addr */
xua_msg_copy_part(xua, SUA_IEI_SRC_ADDR, xua_in, SUA_IEI_DEST_ADDR);
/* conditional: dest addr */
xua_msg_copy_part(xua, SUA_IEI_DEST_ADDR, xua_in, SUA_IEI_SRC_ADDR);
/* optional: importance */
xua_msg_copy_part(xua, SUA_IEI_IMPORTANCE, xua_in, SUA_IEI_IMPORTANCE);
/* optional: data */
xua_msg_copy_part(xua, SUA_IEI_DATA, xua_in, SUA_IEI_DATA);
return xua;
}
/* Find a SCCP user for given SUA message (based on SUA_IEI_DEST_ADDR */
static struct osmo_sccp_user *sccp_find_user(struct osmo_sccp_instance *inst,
struct xua_msg *xua)
{
int rc;
struct osmo_sccp_addr called_addr;
rc = sua_addr_parse(&called_addr, xua, SUA_IEI_DEST_ADDR);
if (rc < 0) {
LOGP(DLSCCP, LOGL_ERROR, "Cannot find SCCP User for XUA "
"Message %s without valid DEST_ADDR\n",
xua_hdr_dump(xua, &xua_dialect_sua));
return NULL;
}
if (!(called_addr.presence & OSMO_SCCP_ADDR_T_SSN)) {
LOGP(DLSCCP, LOGL_ERROR, "Cannot resolve SCCP User for "
"XUA Message %s without SSN in CalledAddr\n",
xua_hdr_dump(xua, &xua_dialect_sua));
return NULL;
}
return sccp_user_find(inst, called_addr.ssn, called_addr.pc);
}
/*! \brief SCOC: Receive SCRC Routing Failure
* \param[in] inst SCCP Instance on which we operate
* \param[in] xua SUA message that was failed to route
* \param[in] return_cause Reason (cause) for routing failure */
void sccp_scoc_rx_scrc_rout_fail(struct osmo_sccp_instance *inst,
struct xua_msg *xua, uint32_t return_cause)
{
uint32_t conn_id;
struct sccp_connection *conn;
LOGP(DLSCCP, LOGL_NOTICE, "SCRC Routing Failure for message %s\n",
xua_hdr_dump(xua, &xua_dialect_sua));
/* try to dispatch to connection FSM (if any) */
conn_id = xua_msg_get_u32(xua, SUA_IEI_DEST_REF);
conn = conn_find_by_id(inst, conn_id);
if (conn) {
osmo_fsm_inst_dispatch(conn->fi,
SCOC_E_RCOC_ROUT_FAIL_IND, xua);
} else {
/* generate + send CREF directly */
struct xua_msg *cref;
uint8_t cref_cause = get_cref_cause_for_ret(return_cause);
cref = gen_coref_without_conn(inst, xua, cref_cause);
sccp_scrc_rx_scoc_conn_msg(inst, cref);
xua_msg_free(cref);
}
}
/* Generate a COERR based in input arguments */
static struct xua_msg *gen_coerr(uint32_t route_ctx, uint32_t dest_ref,
uint32_t err_cause)
{
struct xua_msg *xua = xua_msg_alloc();
xua->hdr = XUA_HDR(SUA_MSGC_CO, SUA_CO_COERR);
xua_msg_add_u32(xua, SUA_IEI_ROUTE_CTX, route_ctx);
xua_msg_add_u32(xua, SUA_IEI_DEST_REF, dest_ref);
xua_msg_add_u32(xua, SUA_IEI_CAUSE, SUA_CAUSE_T_ERROR | err_cause);
return xua;
}
/* generate COERR from incoming XUA and send it */
static void tx_coerr_from_xua(struct osmo_sccp_instance *inst,
struct xua_msg *in, uint32_t err_cause)
{
struct xua_msg *xua;
uint32_t route_ctx, dest_ref;
route_ctx = xua_msg_get_u32(in, SUA_IEI_ROUTE_CTX);
/* get *source* reference and use as destination ref */
dest_ref = xua_msg_get_u32(in, SUA_IEI_SRC_REF);
xua = gen_coerr(route_ctx, dest_ref, err_cause);
/* copy over the MTP parameters */
xua->mtp.dpc = in->mtp.opc;
xua->mtp.opc = in->mtp.dpc;
xua->mtp.sio = in->mtp.sio;
/* sent to SCRC for transmission */
sccp_scrc_rx_scoc_conn_msg(inst, xua);
xua_msg_free(xua);
}
/* Generate a RELCO based in input arguments */
static struct xua_msg *gen_relco(uint32_t route_ctx, uint32_t dest_ref,
uint32_t src_ref)
{
struct xua_msg *xua = xua_msg_alloc();
xua->hdr = XUA_HDR(SUA_MSGC_CO, SUA_CO_RELCO);
xua_msg_add_u32(xua, SUA_IEI_ROUTE_CTX, route_ctx);
xua_msg_add_u32(xua, SUA_IEI_DEST_REF, dest_ref);
xua_msg_add_u32(xua, SUA_IEI_SRC_REF, src_ref);
return xua;
}
/* generate RELCO from incoming XUA and send it */
static void tx_relco_from_xua(struct osmo_sccp_instance *inst,
struct xua_msg *in)
{
struct xua_msg *xua;
uint32_t route_ctx, dest_ref, src_ref;
route_ctx = xua_msg_get_u32(in, SUA_IEI_ROUTE_CTX);
/* get *source* reference and use as destination ref */
dest_ref = xua_msg_get_u32(in, SUA_IEI_SRC_REF);
/* get *dest* reference and use as source ref */
src_ref = xua_msg_get_u32(in, SUA_IEI_DEST_REF);
xua = gen_relco(route_ctx, dest_ref, src_ref);
/* copy over the MTP parameters */
xua->mtp.dpc = in->mtp.opc;
xua->mtp.opc = in->mtp.dpc;
xua->mtp.sio = in->mtp.sio;
/* send to SCRC for transmission */
sccp_scrc_rx_scoc_conn_msg(inst, xua);
xua_msg_free(xua);
}
/* Generate a RLSD based in input arguments */
static struct xua_msg *gen_rlsd(uint32_t route_ctx, uint32_t dest_ref,
uint32_t src_ref)
{
struct xua_msg *xua = xua_msg_alloc();
xua->hdr = XUA_HDR(SUA_MSGC_CO, SUA_CO_RELRE);
xua_msg_add_u32(xua, SUA_IEI_ROUTE_CTX, route_ctx);
xua_msg_add_u32(xua, SUA_IEI_DEST_REF, dest_ref);
xua_msg_add_u32(xua, SUA_IEI_SRC_REF, src_ref);
return xua;
}
/* Generate a RLSD to both the remote side and the local conn */
static void tx_rlsd_from_xua_twoway(struct sccp_connection *conn,
struct xua_msg *in)
{
struct xua_msg *xua;
uint32_t route_ctx, dest_ref, src_ref;
route_ctx = xua_msg_get_u32(in, SUA_IEI_ROUTE_CTX);
/* get *source* reference and use as destination ref */
dest_ref = xua_msg_get_u32(in, SUA_IEI_SRC_REF);
/* get *source* reference and use as destination ref */
src_ref = xua_msg_get_u32(in, SUA_IEI_DEST_REF);
/* Generate RLSD towards remote peer */
xua = gen_rlsd(route_ctx, dest_ref, src_ref);
/* copy over the MTP parameters */
xua->mtp.dpc = in->mtp.opc;
xua->mtp.opc = in->mtp.dpc;
xua->mtp.sio = in->mtp.sio;
/* send to SCRC for transmission */
sccp_scrc_rx_scoc_conn_msg(conn->inst, xua);
xua_msg_free(xua);
/* Generate RLSD towards local peer */
xua = gen_rlsd(conn->inst->route_ctx, conn->conn_id, conn->remote_ref);
xua->mtp.dpc = in->mtp.dpc;
xua->mtp.opc = conn->remote_pc;
xua->mtp.sio = in->mtp.sio;
osmo_fsm_inst_dispatch(conn->fi, SCOC_E_RCOC_RLSD_IND, xua);
xua_msg_free(xua);
}
/* process received message for unassigned local reference */
static void sccp_scoc_rx_unass_local_ref(struct osmo_sccp_instance *inst,
struct xua_msg *xua)
{
/* we have received a message with unassigned destination local
* reference and thus apply the action indicated in Table
* B.2/Q.714 */
switch (xua->hdr.msg_type) {
case SUA_CO_COAK: /* CC */
case SUA_CO_COIT: /* IT */
case SUA_CO_RESRE: /* RSR */
case SUA_CO_RESCO: /* RSC */
/* Send COERR */
tx_coerr_from_xua(inst, xua, SCCP_ERROR_LRN_MISMATCH_UNASSIGNED);
break;
case SUA_CO_COREF: /* CREF */
case SUA_CO_RELCO: /* RLC */
case SUA_CO_CODT: /* DT1 */
case SUA_CO_CODA: /* AK */
case SUA_CO_COERR: /* ERR */
/* DISCARD */
break;
case SUA_CO_RELRE: /* RLSD */
/* Send RLC */
tx_relco_from_xua(inst, xua);
break;
default:
LOGP(DLSCCP, LOGL_NOTICE, "Unhandled %s\n",
xua_hdr_dump(xua, &xua_dialect_sua));
break;
}
}
/* process received message for invalid source local reference */
static void sccp_scoc_rx_inval_src_ref(struct sccp_connection *conn,
struct xua_msg *xua,
uint32_t inval_src_ref)
{
LOGP(DLSCCP, LOGL_NOTICE,
"Received message for source ref %u on conn with mismatching remote ref %u\n",
inval_src_ref, conn->remote_ref);
/* we have received a message with invalid source local
* reference and thus apply the action indicated in Table
* B.2/Q.714 */
switch (xua->hdr.msg_type) {
case SUA_CO_RELRE: /* RLSD */
case SUA_CO_RESRE: /* RSR */
case SUA_CO_RESCO: /* RSC */
/* Send ERR */
tx_coerr_from_xua(conn->inst, xua, SCCP_ERROR_LRN_MISMATCH_INCONSISTENT);
break;
case SUA_CO_COIT: /* IT */
/* FIXME: RLSD to both sides */
tx_rlsd_from_xua_twoway(conn, xua);
break;
case SUA_CO_RELCO: /* RLC */
/* DISCARD */
break;
default:
LOGP(DLSCCP, LOGL_NOTICE, "Unhandled %s\n",
xua_hdr_dump(xua, &xua_dialect_sua));
break;
}
}
/* process received message for invalid origin point code */
static void sccp_scoc_rx_inval_opc(struct sccp_connection *conn,
struct xua_msg *xua)
{
LOGP(DLSCCP, LOGL_NOTICE,
"Received message %s for opc=%u=%s on conn with mismatching remote pc=%u=%s\n",
xua_hdr_dump(xua, &xua_dialect_sua),
xua->mtp.opc, osmo_ss7_pointcode_print(conn->inst->ss7, xua->mtp.opc),
conn->remote_pc, osmo_ss7_pointcode_print2(conn->inst->ss7, conn->remote_pc));
/* we have received a message with invalid origin PC and thus
* apply the action indicated in Table B.2/Q.714 */
switch (xua->hdr.msg_type) {
case SUA_CO_RELRE: /* RLSD */
case SUA_CO_RESRE: /* RSR */
case SUA_CO_RESCO: /* RSC */
/* Send ERR */
tx_coerr_from_xua(conn->inst, xua, SCCP_ERROR_POINT_CODE_MISMATCH);
break;
case SUA_CO_RELCO: /* RLC */
case SUA_CO_CODT: /* DT1 */
case SUA_CO_CODA: /* AK */
case SUA_CO_COERR: /* ERR */
/* DISCARD */
break;
default:
LOGP(DLSCCP, LOGL_NOTICE, "Unhandled %s\n",
xua_hdr_dump(xua, &xua_dialect_sua));
break;
}
}
/*! \brief Main entrance function for primitives from the SCRC (Routing Control)
* \param[in] inst SCCP Instance in which we operate
* \param[in] xua SUA message in xua_msg format */
void sccp_scoc_rx_from_scrc(struct osmo_sccp_instance *inst,
struct xua_msg *xua)
{
struct sccp_connection *conn;
struct osmo_sccp_user *scu;
uint32_t src_loc_ref;
int event;
/* we basically try to convert the SUA message into an event,
* and then dispatch the event to the connection-specific FSM.
* If it is a CORE (Connect REquest), we create the connection
* (and implicitly its FSM) first */
if (xua->hdr.msg_type == SUA_CO_CORE) {
scu = sccp_find_user(inst, xua);
if (!scu) {
/* this shouldn't happen, as the caller should
* have already verified that a local user is
* equipped for this SSN */
LOGP(DLSCCP, LOGL_ERROR, "Cannot find user for "
"CORE ?!?\n");
return;
}
/* Allocate new connection */
conn = conn_create(scu);
conn->incoming = true;
} else {
uint32_t conn_id;
/* Resolve existing connection */
conn_id = xua_msg_get_u32(xua, SUA_IEI_DEST_REF);
conn = conn_find_by_id(inst, conn_id);
if (!conn) {
LOGP(DLSCCP, LOGL_NOTICE, "Received %s: Cannot find connection for "
"local reference %u\n", xua_hdr_dump(xua, &xua_dialect_sua), conn_id);
sccp_scoc_rx_unass_local_ref(inst, xua);
return;
}
}
OSMO_ASSERT(conn);
OSMO_ASSERT(conn->fi);
DEBUGP(DLSCCP, "Received %s for local reference %u\n",
xua_hdr_dump(xua, &xua_dialect_sua), conn->conn_id);
if (xua->hdr.msg_type != SUA_CO_CORE &&
xua->hdr.msg_type != SUA_CO_COAK &&
xua->hdr.msg_type != SUA_CO_COREF) {
if (xua_msg_find_tag(xua, SUA_IEI_SRC_REF)) {
/* Check if received source local reference !=
* the one we saved in local state */
src_loc_ref = xua_msg_get_u32(xua, SUA_IEI_SRC_REF);
if (src_loc_ref != conn->remote_ref) {
sccp_scoc_rx_inval_src_ref(conn, xua, src_loc_ref);
return;
}
}
/* Check if received OPC != the remote_pc we stored locally */
if (xua->mtp.opc != conn->remote_pc) {
sccp_scoc_rx_inval_opc(conn, xua);
return;
}
}
/* Map from XUA message to event */
event = xua_msg_event_map(xua, sua_scoc_event_map, ARRAY_SIZE(sua_scoc_event_map));
if (event < 0) {
LOGP(DLSCCP, LOGL_ERROR, "Cannot map SCRC msg %s to event\n",
xua_hdr_dump(xua, &xua_dialect_sua));
/* Table B.1/Q714 states DISCARD for any message with
* unknown type */
return;
}
/* Dispatch event to existing connection */
osmo_fsm_inst_dispatch(conn->fi, event, xua);
}
/* get the Connection ID of the given SCU primitive */
static uint32_t scu_prim_conn_id(const struct osmo_scu_prim *prim)
{
switch (prim->oph.primitive) {
case OSMO_SCU_PRIM_N_CONNECT:
return prim->u.connect.conn_id;
case OSMO_SCU_PRIM_N_DATA:
return prim->u.data.conn_id;
case OSMO_SCU_PRIM_N_DISCONNECT:
return prim->u.disconnect.conn_id;
case OSMO_SCU_PRIM_N_RESET:
return prim->u.reset.conn_id;
default:
return 0;
}
}
add caller-owns-msgb variant osmo_sccp_user_sap_down_nofree() Add osmo_sccp_user_sap_down_nofree(), which is identical to osmo_sccp_user_sap_down(), but doesn't imply a msgb_free(). To implement that, sccp_sclc_user_sap_down_nofree() with the same msgb semantics is required. Rationale: Avoiding msgb leaks is easiest if the caller retains ownership of the msgb. Take this hypothetical chain where leaks are obviously avoided: void send() { msg = msgb_alloc(); dispatch(msg); msgb_free(msg); } void dispatch(msg) { osmo_fsm_inst_dispatch(fi, msg); } void fi_on_event(fi, data) { if (socket_is_ok) socket_write((struct msgb*)data); } void socket_write(msgb) { if (!ok1) return; if (ok2) { if (!ok3) return; write(sock, msg->data); } } However, if the caller passes ownership down to the msgb consumer, things become nightmarishly complex: void send() { msg = msgb_alloc(); rc = dispatch(msg); /* dispatching event failed? */ if (rc) msgb_free(msg); } int dispatch(msg) { if (osmo_fsm_inst_dispatch(fi, msg)) return -1; if (something_else()) return -1; // <-- double free! } void fi_on_event(fi, data) { if (socket_is_ok) { socket_write((struct msgb*)data); else /* socket didn't consume? */ msgb_free(data); } int socket_write(msgb) { if (!ok1) return -1; // <-- leak! if (ok2) { if (!ok3) goto out; write(sock, msg->data); } out: msgb_free(msg); return -2; } If any link in this call chain fails to be aware of the importance to return a failed RC or to free a msgb if the chain is broken, or to not return a failed RC if the msgb is consumed, we have a hidden msgb leak or double free. This is the case with osmo_sccp_user_sap_down(). In new osmo-msc, passing data through various FSM instances, there is high potential for leak/double-free bugs. A very large brain is required to track down every msgb path. osmo_sccp_user_sap_down_nofree() makes this problem trivial to solve even for humans. Change-Id: Ic818efa78b90f727e1a94c18b60d9a306644f340
2019-03-10 03:41:27 +00:00
/*! Main entrance function for primitives from SCCP User.
* The caller is required to free oph->msg, otherwise the same as osmo_sccp_user_sap_down().
* \param[in] scu SCCP User sending us the primitive
* \param[in] oph Osmocom primitive sent by the user
* \returns 0 on success; negative on error */
add caller-owns-msgb variant osmo_sccp_user_sap_down_nofree() Add osmo_sccp_user_sap_down_nofree(), which is identical to osmo_sccp_user_sap_down(), but doesn't imply a msgb_free(). To implement that, sccp_sclc_user_sap_down_nofree() with the same msgb semantics is required. Rationale: Avoiding msgb leaks is easiest if the caller retains ownership of the msgb. Take this hypothetical chain where leaks are obviously avoided: void send() { msg = msgb_alloc(); dispatch(msg); msgb_free(msg); } void dispatch(msg) { osmo_fsm_inst_dispatch(fi, msg); } void fi_on_event(fi, data) { if (socket_is_ok) socket_write((struct msgb*)data); } void socket_write(msgb) { if (!ok1) return; if (ok2) { if (!ok3) return; write(sock, msg->data); } } However, if the caller passes ownership down to the msgb consumer, things become nightmarishly complex: void send() { msg = msgb_alloc(); rc = dispatch(msg); /* dispatching event failed? */ if (rc) msgb_free(msg); } int dispatch(msg) { if (osmo_fsm_inst_dispatch(fi, msg)) return -1; if (something_else()) return -1; // <-- double free! } void fi_on_event(fi, data) { if (socket_is_ok) { socket_write((struct msgb*)data); else /* socket didn't consume? */ msgb_free(data); } int socket_write(msgb) { if (!ok1) return -1; // <-- leak! if (ok2) { if (!ok3) goto out; write(sock, msg->data); } out: msgb_free(msg); return -2; } If any link in this call chain fails to be aware of the importance to return a failed RC or to free a msgb if the chain is broken, or to not return a failed RC if the msgb is consumed, we have a hidden msgb leak or double free. This is the case with osmo_sccp_user_sap_down(). In new osmo-msc, passing data through various FSM instances, there is high potential for leak/double-free bugs. A very large brain is required to track down every msgb path. osmo_sccp_user_sap_down_nofree() makes this problem trivial to solve even for humans. Change-Id: Ic818efa78b90f727e1a94c18b60d9a306644f340
2019-03-10 03:41:27 +00:00
int osmo_sccp_user_sap_down_nofree(struct osmo_sccp_user *scu, struct osmo_prim_hdr *oph)
{
struct osmo_scu_prim *prim = (struct osmo_scu_prim *) oph;
struct osmo_sccp_instance *inst = scu->inst;
struct sccp_connection *conn;
int rc = 0;
int event;
LOGP(DLSCCP, LOGL_DEBUG, "Received SCCP User Primitive (%s)\n",
osmo_scu_prim_name(&prim->oph));
switch (OSMO_PRIM_HDR(&prim->oph)) {
case OSMO_PRIM(OSMO_SCU_PRIM_N_UNITDATA, PRIM_OP_REQUEST):
/* other CL primitives? */
/* Connectionless by-passes this altogether */
add caller-owns-msgb variant osmo_sccp_user_sap_down_nofree() Add osmo_sccp_user_sap_down_nofree(), which is identical to osmo_sccp_user_sap_down(), but doesn't imply a msgb_free(). To implement that, sccp_sclc_user_sap_down_nofree() with the same msgb semantics is required. Rationale: Avoiding msgb leaks is easiest if the caller retains ownership of the msgb. Take this hypothetical chain where leaks are obviously avoided: void send() { msg = msgb_alloc(); dispatch(msg); msgb_free(msg); } void dispatch(msg) { osmo_fsm_inst_dispatch(fi, msg); } void fi_on_event(fi, data) { if (socket_is_ok) socket_write((struct msgb*)data); } void socket_write(msgb) { if (!ok1) return; if (ok2) { if (!ok3) return; write(sock, msg->data); } } However, if the caller passes ownership down to the msgb consumer, things become nightmarishly complex: void send() { msg = msgb_alloc(); rc = dispatch(msg); /* dispatching event failed? */ if (rc) msgb_free(msg); } int dispatch(msg) { if (osmo_fsm_inst_dispatch(fi, msg)) return -1; if (something_else()) return -1; // <-- double free! } void fi_on_event(fi, data) { if (socket_is_ok) { socket_write((struct msgb*)data); else /* socket didn't consume? */ msgb_free(data); } int socket_write(msgb) { if (!ok1) return -1; // <-- leak! if (ok2) { if (!ok3) goto out; write(sock, msg->data); } out: msgb_free(msg); return -2; } If any link in this call chain fails to be aware of the importance to return a failed RC or to free a msgb if the chain is broken, or to not return a failed RC if the msgb is consumed, we have a hidden msgb leak or double free. This is the case with osmo_sccp_user_sap_down(). In new osmo-msc, passing data through various FSM instances, there is high potential for leak/double-free bugs. A very large brain is required to track down every msgb path. osmo_sccp_user_sap_down_nofree() makes this problem trivial to solve even for humans. Change-Id: Ic818efa78b90f727e1a94c18b60d9a306644f340
2019-03-10 03:41:27 +00:00
return sccp_sclc_user_sap_down_nofree(scu, oph);
case OSMO_PRIM(OSMO_SCU_PRIM_N_CONNECT, PRIM_OP_REQUEST):
/* Allocate new connection structure */
conn = conn_create_id(scu, prim->u.connect.conn_id);
if (!conn) {
/* FIXME: inform SCCP user with proper reply */
LOGP(DLSCCP, LOGL_ERROR, "Cannot create conn-id for primitive %s\n",
osmo_scu_prim_name(&prim->oph));
add caller-owns-msgb variant osmo_sccp_user_sap_down_nofree() Add osmo_sccp_user_sap_down_nofree(), which is identical to osmo_sccp_user_sap_down(), but doesn't imply a msgb_free(). To implement that, sccp_sclc_user_sap_down_nofree() with the same msgb semantics is required. Rationale: Avoiding msgb leaks is easiest if the caller retains ownership of the msgb. Take this hypothetical chain where leaks are obviously avoided: void send() { msg = msgb_alloc(); dispatch(msg); msgb_free(msg); } void dispatch(msg) { osmo_fsm_inst_dispatch(fi, msg); } void fi_on_event(fi, data) { if (socket_is_ok) socket_write((struct msgb*)data); } void socket_write(msgb) { if (!ok1) return; if (ok2) { if (!ok3) return; write(sock, msg->data); } } However, if the caller passes ownership down to the msgb consumer, things become nightmarishly complex: void send() { msg = msgb_alloc(); rc = dispatch(msg); /* dispatching event failed? */ if (rc) msgb_free(msg); } int dispatch(msg) { if (osmo_fsm_inst_dispatch(fi, msg)) return -1; if (something_else()) return -1; // <-- double free! } void fi_on_event(fi, data) { if (socket_is_ok) { socket_write((struct msgb*)data); else /* socket didn't consume? */ msgb_free(data); } int socket_write(msgb) { if (!ok1) return -1; // <-- leak! if (ok2) { if (!ok3) goto out; write(sock, msg->data); } out: msgb_free(msg); return -2; } If any link in this call chain fails to be aware of the importance to return a failed RC or to free a msgb if the chain is broken, or to not return a failed RC if the msgb is consumed, we have a hidden msgb leak or double free. This is the case with osmo_sccp_user_sap_down(). In new osmo-msc, passing data through various FSM instances, there is high potential for leak/double-free bugs. A very large brain is required to track down every msgb path. osmo_sccp_user_sap_down_nofree() makes this problem trivial to solve even for humans. Change-Id: Ic818efa78b90f727e1a94c18b60d9a306644f340
2019-03-10 03:41:27 +00:00
return rc;
}
break;
case OSMO_PRIM(OSMO_SCU_PRIM_N_CONNECT, PRIM_OP_RESPONSE):
case OSMO_PRIM(OSMO_SCU_PRIM_N_DATA, PRIM_OP_REQUEST):
case OSMO_PRIM(OSMO_SCU_PRIM_N_DISCONNECT, PRIM_OP_REQUEST):
case OSMO_PRIM(OSMO_SCU_PRIM_N_RESET, PRIM_OP_REQUEST):
/* Resolve existing connection structure */
conn = conn_find_by_id(inst, scu_prim_conn_id(prim));
if (!conn) {
/* FIXME: inform SCCP user with proper reply */
LOGP(DLSCCP, LOGL_ERROR, "Received unknown conn-id %u for primitive %s\n",
scu_prim_conn_id(prim), osmo_scu_prim_name(&prim->oph));
add caller-owns-msgb variant osmo_sccp_user_sap_down_nofree() Add osmo_sccp_user_sap_down_nofree(), which is identical to osmo_sccp_user_sap_down(), but doesn't imply a msgb_free(). To implement that, sccp_sclc_user_sap_down_nofree() with the same msgb semantics is required. Rationale: Avoiding msgb leaks is easiest if the caller retains ownership of the msgb. Take this hypothetical chain where leaks are obviously avoided: void send() { msg = msgb_alloc(); dispatch(msg); msgb_free(msg); } void dispatch(msg) { osmo_fsm_inst_dispatch(fi, msg); } void fi_on_event(fi, data) { if (socket_is_ok) socket_write((struct msgb*)data); } void socket_write(msgb) { if (!ok1) return; if (ok2) { if (!ok3) return; write(sock, msg->data); } } However, if the caller passes ownership down to the msgb consumer, things become nightmarishly complex: void send() { msg = msgb_alloc(); rc = dispatch(msg); /* dispatching event failed? */ if (rc) msgb_free(msg); } int dispatch(msg) { if (osmo_fsm_inst_dispatch(fi, msg)) return -1; if (something_else()) return -1; // <-- double free! } void fi_on_event(fi, data) { if (socket_is_ok) { socket_write((struct msgb*)data); else /* socket didn't consume? */ msgb_free(data); } int socket_write(msgb) { if (!ok1) return -1; // <-- leak! if (ok2) { if (!ok3) goto out; write(sock, msg->data); } out: msgb_free(msg); return -2; } If any link in this call chain fails to be aware of the importance to return a failed RC or to free a msgb if the chain is broken, or to not return a failed RC if the msgb is consumed, we have a hidden msgb leak or double free. This is the case with osmo_sccp_user_sap_down(). In new osmo-msc, passing data through various FSM instances, there is high potential for leak/double-free bugs. A very large brain is required to track down every msgb path. osmo_sccp_user_sap_down_nofree() makes this problem trivial to solve even for humans. Change-Id: Ic818efa78b90f727e1a94c18b60d9a306644f340
2019-03-10 03:41:27 +00:00
return rc;
}
break;
default:
LOGP(DLSCCP, LOGL_ERROR, "Received unknown primitive %s\n",
osmo_scu_prim_name(&prim->oph));
add caller-owns-msgb variant osmo_sccp_user_sap_down_nofree() Add osmo_sccp_user_sap_down_nofree(), which is identical to osmo_sccp_user_sap_down(), but doesn't imply a msgb_free(). To implement that, sccp_sclc_user_sap_down_nofree() with the same msgb semantics is required. Rationale: Avoiding msgb leaks is easiest if the caller retains ownership of the msgb. Take this hypothetical chain where leaks are obviously avoided: void send() { msg = msgb_alloc(); dispatch(msg); msgb_free(msg); } void dispatch(msg) { osmo_fsm_inst_dispatch(fi, msg); } void fi_on_event(fi, data) { if (socket_is_ok) socket_write((struct msgb*)data); } void socket_write(msgb) { if (!ok1) return; if (ok2) { if (!ok3) return; write(sock, msg->data); } } However, if the caller passes ownership down to the msgb consumer, things become nightmarishly complex: void send() { msg = msgb_alloc(); rc = dispatch(msg); /* dispatching event failed? */ if (rc) msgb_free(msg); } int dispatch(msg) { if (osmo_fsm_inst_dispatch(fi, msg)) return -1; if (something_else()) return -1; // <-- double free! } void fi_on_event(fi, data) { if (socket_is_ok) { socket_write((struct msgb*)data); else /* socket didn't consume? */ msgb_free(data); } int socket_write(msgb) { if (!ok1) return -1; // <-- leak! if (ok2) { if (!ok3) goto out; write(sock, msg->data); } out: msgb_free(msg); return -2; } If any link in this call chain fails to be aware of the importance to return a failed RC or to free a msgb if the chain is broken, or to not return a failed RC if the msgb is consumed, we have a hidden msgb leak or double free. This is the case with osmo_sccp_user_sap_down(). In new osmo-msc, passing data through various FSM instances, there is high potential for leak/double-free bugs. A very large brain is required to track down every msgb path. osmo_sccp_user_sap_down_nofree() makes this problem trivial to solve even for humans. Change-Id: Ic818efa78b90f727e1a94c18b60d9a306644f340
2019-03-10 03:41:27 +00:00
return -1;
}
/* Map from primitive to event */
event = osmo_event_for_prim(oph, scu_scoc_event_map);
/* Dispatch event into connection */
add caller-owns-msgb variant osmo_sccp_user_sap_down_nofree() Add osmo_sccp_user_sap_down_nofree(), which is identical to osmo_sccp_user_sap_down(), but doesn't imply a msgb_free(). To implement that, sccp_sclc_user_sap_down_nofree() with the same msgb semantics is required. Rationale: Avoiding msgb leaks is easiest if the caller retains ownership of the msgb. Take this hypothetical chain where leaks are obviously avoided: void send() { msg = msgb_alloc(); dispatch(msg); msgb_free(msg); } void dispatch(msg) { osmo_fsm_inst_dispatch(fi, msg); } void fi_on_event(fi, data) { if (socket_is_ok) socket_write((struct msgb*)data); } void socket_write(msgb) { if (!ok1) return; if (ok2) { if (!ok3) return; write(sock, msg->data); } } However, if the caller passes ownership down to the msgb consumer, things become nightmarishly complex: void send() { msg = msgb_alloc(); rc = dispatch(msg); /* dispatching event failed? */ if (rc) msgb_free(msg); } int dispatch(msg) { if (osmo_fsm_inst_dispatch(fi, msg)) return -1; if (something_else()) return -1; // <-- double free! } void fi_on_event(fi, data) { if (socket_is_ok) { socket_write((struct msgb*)data); else /* socket didn't consume? */ msgb_free(data); } int socket_write(msgb) { if (!ok1) return -1; // <-- leak! if (ok2) { if (!ok3) goto out; write(sock, msg->data); } out: msgb_free(msg); return -2; } If any link in this call chain fails to be aware of the importance to return a failed RC or to free a msgb if the chain is broken, or to not return a failed RC if the msgb is consumed, we have a hidden msgb leak or double free. This is the case with osmo_sccp_user_sap_down(). In new osmo-msc, passing data through various FSM instances, there is high potential for leak/double-free bugs. A very large brain is required to track down every msgb path. osmo_sccp_user_sap_down_nofree() makes this problem trivial to solve even for humans. Change-Id: Ic818efa78b90f727e1a94c18b60d9a306644f340
2019-03-10 03:41:27 +00:00
return osmo_fsm_inst_dispatch(conn->fi, event, prim);
}
add caller-owns-msgb variant osmo_sccp_user_sap_down_nofree() Add osmo_sccp_user_sap_down_nofree(), which is identical to osmo_sccp_user_sap_down(), but doesn't imply a msgb_free(). To implement that, sccp_sclc_user_sap_down_nofree() with the same msgb semantics is required. Rationale: Avoiding msgb leaks is easiest if the caller retains ownership of the msgb. Take this hypothetical chain where leaks are obviously avoided: void send() { msg = msgb_alloc(); dispatch(msg); msgb_free(msg); } void dispatch(msg) { osmo_fsm_inst_dispatch(fi, msg); } void fi_on_event(fi, data) { if (socket_is_ok) socket_write((struct msgb*)data); } void socket_write(msgb) { if (!ok1) return; if (ok2) { if (!ok3) return; write(sock, msg->data); } } However, if the caller passes ownership down to the msgb consumer, things become nightmarishly complex: void send() { msg = msgb_alloc(); rc = dispatch(msg); /* dispatching event failed? */ if (rc) msgb_free(msg); } int dispatch(msg) { if (osmo_fsm_inst_dispatch(fi, msg)) return -1; if (something_else()) return -1; // <-- double free! } void fi_on_event(fi, data) { if (socket_is_ok) { socket_write((struct msgb*)data); else /* socket didn't consume? */ msgb_free(data); } int socket_write(msgb) { if (!ok1) return -1; // <-- leak! if (ok2) { if (!ok3) goto out; write(sock, msg->data); } out: msgb_free(msg); return -2; } If any link in this call chain fails to be aware of the importance to return a failed RC or to free a msgb if the chain is broken, or to not return a failed RC if the msgb is consumed, we have a hidden msgb leak or double free. This is the case with osmo_sccp_user_sap_down(). In new osmo-msc, passing data through various FSM instances, there is high potential for leak/double-free bugs. A very large brain is required to track down every msgb path. osmo_sccp_user_sap_down_nofree() makes this problem trivial to solve even for humans. Change-Id: Ic818efa78b90f727e1a94c18b60d9a306644f340
2019-03-10 03:41:27 +00:00
/*! Main entrance function for primitives from SCCP User.
* Implies a msgb_free(oph->msg), otherwise the same as osmo_sccp_user_sap().
* \param[in] scu SCCP User sending us the primitive
* \param[in] oph Osmocom primitive sent by the user
* \returns 0 on success; negative on error */
int osmo_sccp_user_sap_down(struct osmo_sccp_user *scu, struct osmo_prim_hdr *oph)
{
struct osmo_scu_prim *prim = (struct osmo_scu_prim *) oph;
struct msgb *msg = prim->oph.msg;
int rc = osmo_sccp_user_sap_down_nofree(scu, oph);
msgb_free(msg);
return rc;
}
void sccp_scoc_flush_connections(struct osmo_sccp_instance *inst)
{
struct rb_node *node;
while ((node = rb_first(&inst->connections))) {
struct sccp_connection *conn = container_of(node, struct sccp_connection, node);
conn_destroy(conn);
}
}
#include <osmocom/vty/vty.h>
static void vty_show_connection(struct vty *vty, struct sccp_connection *conn)
{
struct osmo_ss7_instance *s7i = conn->inst->ss7;
struct osmo_sccp_addr *remote_addr;
uint32_t local_pc = OSMO_SS7_PC_INVALID;
if (osmo_ss7_pc_is_valid(conn->user->pc))
local_pc = conn->user->pc;
else if (osmo_ss7_pc_is_valid(s7i->cfg.primary_pc))
local_pc = s7i->cfg.primary_pc;
if (conn->incoming)
remote_addr = &conn->calling_addr;
else
remote_addr = &conn->called_addr;
vty_out(vty, "%c %06x %3u %7s ", conn->incoming ? 'I' : 'O',
conn->conn_id, conn->user->ssn,
osmo_ss7_pointcode_print(s7i, local_pc));
vty_out(vty, "%16s %06x %3u %7s%s",
osmo_fsm_inst_state_name(conn->fi), conn->remote_ref, remote_addr->ssn,
osmo_ss7_pointcode_print(s7i, conn->remote_pc),
VTY_NEWLINE);
}
void sccp_scoc_show_connections(struct vty *vty, struct osmo_sccp_instance *inst)
{
struct sccp_connection *conn;
struct rb_node *node;
vty_out(vty, "I Local Conn. Remote %s", VTY_NEWLINE);
vty_out(vty, "O Ref SSN PC State Ref SSN PC %s", VTY_NEWLINE);
vty_out(vty, "- ------ --- ------- ---------------- ------ --- -------%s", VTY_NEWLINE);
for (node = rb_first(&inst->connections); node; node = rb_next(node)) {
conn = container_of(node, struct sccp_connection, node);
vty_show_connection(vty, conn);
}
}