osmo-mgw/src/libosmo-mgcp/mgcp_osmux.c

774 lines
25 KiB
C
Raw Normal View History

/*
* (C) 2012-2013 by Pablo Neira Ayuso <pablo@gnumonks.org>
* (C) 2012-2013 by On Waves ehf <http://www.on-waves.com>
* All rights not specifically granted under this license are reserved.
*
* This program is free software; you can redistribute it and/or modify it
* under the terms of the GNU Affero General Public License as published by the
* Free Software Foundation; either version 3 of the License, or (at your
* option) any later version.
*/
#include <stdio.h> /* for printf */
#include <string.h> /* for memcpy */
#include <stdlib.h> /* for abs */
#include <inttypes.h> /* for PRIu64 */
#include <netinet/in.h>
#include <osmocom/core/msgb.h>
#include <osmocom/core/socket.h>
#include <osmocom/core/talloc.h>
#include <osmocom/netif/osmux.h>
#include <osmocom/netif/rtp.h>
#include <osmocom/netif/amr.h>
#include <osmocom/mgcp/debug.h>
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
#include <osmocom/mgcp/mgcp.h>
#include <osmocom/mgcp/mgcp_protocol.h>
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
#include <osmocom/mgcp/osmux.h>
#include <osmocom/mgcp/mgcp_conn.h>
#include <osmocom/mgcp/mgcp_endp.h>
#include <osmocom/mgcp/mgcp_trunk.h>
static struct osmo_fd osmux_fd_v4;
static struct osmo_fd osmux_fd_v6;
static LLIST_HEAD(osmux_handle_list);
struct osmux_handle {
struct llist_head head;
struct osmux_in_handle *in;
struct osmo_sockaddr rem_addr;
int refcnt;
};
const struct value_string osmux_state_strs[] = {
{ OSMUX_STATE_DISABLED, "disabled" },
{ OSMUX_STATE_ACTIVATING, "activating" },
{ OSMUX_STATE_ENABLED, "enabled" },
{ 0, NULL }
};
static const struct rate_ctr_group_desc rate_ctr_group_osmux_desc = {
.group_name_prefix = "conn_osmux",
.group_description = "Osmux connection statistics",
.class_id = 1,
.num_ctr = ARRAY_SIZE(mgcp_conn_osmux_rate_ctr_desc),
.ctr_desc = mgcp_conn_osmux_rate_ctr_desc
};
static void rtpconn_osmux_rate_ctr_add(struct mgcp_conn_rtp *conn_rtp, int id, int inc)
{
struct rate_ctr_group *conn_osmux_stats = conn_rtp->osmux.ctrg;
struct rate_ctr_group *trunk_osmux_stats = conn_rtp->conn->endp->trunk->ratectr.all_osmux_conn_stats;
/* add to both the per-connection and the global stats */
rate_ctr_add(rate_ctr_group_get_ctr(conn_osmux_stats, id), inc);
rate_ctr_add(rate_ctr_group_get_ctr(trunk_osmux_stats, id), inc);
}
static void rtpconn_osmux_rate_ctr_inc(struct mgcp_conn_rtp *conn_rtp, int id)
{
rtpconn_osmux_rate_ctr_add(conn_rtp, id, 1);
}
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
/* Deliver OSMUX batch to the remote end */
static void osmux_deliver_cb(struct msgb *batch_msg, void *data)
{
struct osmux_handle *handle = data;
socklen_t dest_len;
int rc, fd;
struct mgcp_trunk *trunk = (struct mgcp_trunk *)osmux_fd_v4.data;
struct rate_ctr_group *all_osmux_stats = trunk->ratectr.all_osmux_conn_stats;
switch (handle->rem_addr.u.sa.sa_family) {
case AF_INET6:
dest_len = sizeof(handle->rem_addr.u.sin6);
fd = osmux_fd_v6.fd;
break;
case AF_INET:
default:
dest_len = sizeof(handle->rem_addr.u.sin);
fd = osmux_fd_v4.fd;
break;
}
rc = sendto(fd, batch_msg->data, batch_msg->len, 0,
(struct sockaddr *)&handle->rem_addr.u.sa, dest_len);
if (rc < 0) {
char errbuf[129];
strerror_r(errno, errbuf, sizeof(errbuf));
LOGP(DOSMUX, LOGL_NOTICE, "osmux sendto(%s) failed: %s\n",
osmo_sockaddr_to_str(&handle->rem_addr), errbuf);
rate_ctr_inc(rate_ctr_group_get_ctr(all_osmux_stats, OSMUX_DROPPED_PACKETS_CTR));
} else {
rate_ctr_inc(rate_ctr_group_get_ctr(all_osmux_stats, OSMUX_PACKETS_TX_CTR));
}
msgb_free(batch_msg);
}
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
/* Lookup existing OSMUX handle for specified destination address. */
static struct osmux_handle *
osmux_handle_find_get(const struct osmo_sockaddr *rem_addr)
{
struct osmux_handle *h;
llist_for_each_entry(h, &osmux_handle_list, head) {
if (osmo_sockaddr_cmp(&h->rem_addr, rem_addr) == 0) {
h->refcnt++;
LOGP(DOSMUX, LOGL_DEBUG,
"Using existing OSMUX handle for addr=%s (rfcnt=%u)\n",
osmo_sockaddr_to_str(rem_addr), h->refcnt);
return h;
}
}
return NULL;
}
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
/* Put down no longer needed OSMUX handle */
static void osmux_handle_put(struct osmux_in_handle *in)
{
struct osmux_handle *h;
llist_for_each_entry(h, &osmux_handle_list, head) {
if (h->in == in) {
LOGP(DOSMUX, LOGL_DEBUG,
"Putting existing OSMUX handle for addr=%s (rfcnt=%u)\n",
osmo_sockaddr_to_str(&h->rem_addr), h->refcnt);
if (--h->refcnt == 0) {
LOGP(DOSMUX, LOGL_INFO,
"Releasing unused osmux handle for %s\n",
osmo_sockaddr_to_str(&h->rem_addr));
llist_del(&h->head);
TALLOC_FREE(h->in);
talloc_free(h);
}
return;
}
}
LOGP(DOSMUX, LOGL_ERROR, "Cannot find Osmux input handle %p\n", in);
}
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
/* Allocate free OSMUX handle */
static struct osmux_handle *
osmux_handle_alloc(const struct mgcp_trunk *trunk, const struct osmo_sockaddr *rem_addr)
{
struct osmux_handle *h;
const struct mgcp_config *cfg = trunk->cfg;
h = talloc_zero(trunk, struct osmux_handle);
if (!h)
return NULL;
h->rem_addr = *rem_addr;
h->refcnt++;
h->in = osmux_xfrm_input_alloc(h);
if (!h->in) {
talloc_free(h);
return NULL;
}
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
/* sequence number to start OSMUX message from */
osmux_xfrm_input_set_initial_seqnum(h->in, 0);
osmux_xfrm_input_set_batch_factor(h->in, cfg->osmux.batch_factor);
/* If batch size is zero, the library defaults to 1472 bytes. */
osmux_xfrm_input_set_batch_size(h->in, cfg->osmux.batch_size);
osmux_xfrm_input_set_deliver_cb(h->in, osmux_deliver_cb, h);
llist_add(&h->head, &osmux_handle_list);
LOGP(DOSMUX, LOGL_DEBUG, "Created new OSMUX handle for rem_addr=%s\n",
osmo_sockaddr_to_str(rem_addr));
return h;
}
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
/* Lookup existing handle for a specified address, if the handle can not be
* found, the function will automatically allocate one */
static struct osmux_in_handle *
osmux_handle_find_or_create(const struct mgcp_trunk *trunk, const struct osmo_sockaddr *rem_addr)
{
struct osmux_handle *h;
h = osmux_handle_find_get(rem_addr);
if (h != NULL)
return h->in;
h = osmux_handle_alloc(trunk, rem_addr);
if (h == NULL)
return NULL;
return h->in;
}
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
/*! send RTP packet through OSMUX connection.
* \param[in] conn associated RTP connection
* \param[in] msg msgb containing an RTP AMR packet
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
* \returns 0 on success, -1 on ERROR */
int conn_osmux_send_rtp(struct mgcp_conn_rtp *conn, struct msgb *msg)
{
int ret;
struct msgb *msg2;
if (!conn->end.output_enabled) {
rtpconn_osmux_rate_ctr_inc(conn, OSMUX_RTP_PACKETS_TX_DROPPED_CTR);
return -1;
}
if (conn->osmux.state != OSMUX_STATE_ENABLED) {
LOGPCONN(conn->conn, DOSMUX, LOGL_INFO, "forwarding RTP to Osmux conn not yet enabled, dropping (cid=%d)\n",
conn->osmux.remote_cid);
rtpconn_osmux_rate_ctr_inc(conn, OSMUX_RTP_PACKETS_TX_DROPPED_CTR);
return -1;
}
/* msg is not owned by us and will be freed by the caller stack upon return: */
msg2 = msgb_copy_c(conn->conn, msg, "osmux-rtp-send");
if (!msg2)
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
return -1;
while ((ret = osmux_xfrm_input(conn->osmux.in, msg2, conn->osmux.remote_cid)) > 0) {
/* batch full, build and deliver it */
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
osmux_xfrm_input_deliver(conn->osmux.in);
}
if (ret < 0) {
rtpconn_osmux_rate_ctr_inc(conn, OSMUX_RTP_PACKETS_TX_DROPPED_CTR);
} else {
rtpconn_osmux_rate_ctr_inc(conn, OSMUX_RTP_PACKETS_TX_CTR);
rtpconn_osmux_rate_ctr_add(conn, OSMUX_AMR_OCTETS_TX_CTR, msgb_length(msg2) - sizeof(struct rtp_hdr));
}
return 0;
}
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
/* Lookup the endpoint that corresponds to the specified address (port) */
static struct mgcp_conn_rtp*
osmux_conn_lookup(const struct mgcp_trunk *trunk, uint8_t local_cid, const struct osmo_sockaddr *rem_addr)
{
struct mgcp_endpoint *endp;
struct mgcp_conn *conn = NULL;
struct mgcp_conn_rtp *conn_rtp;
struct osmux_handle *h;
int i;
for (i = 0; i < trunk->number_endpoints; i++) {
endp = trunk->endpoints[i];
llist_for_each_entry(conn, &endp->conns, entry) {
if (conn->type != MGCP_CONN_TYPE_RTP)
continue;
conn_rtp = &conn->u.rtp;
if (!mgcp_conn_rtp_is_osmux(conn_rtp))
continue;
/* If Osmux peer is behind NAT:
* + state OSMUX_STATE_ACTIVATING: we cannot validate the remote address & port for
* the conn since we actually set conn's remote IP address from the info we received
* from this source address. NOTE: This means if Osmux peer is behind NAT we cannot
* have more than 1 osmux trunk since it's not possible to differentiate based on
* remote address.
* + state OSMUX_STATE_ENABLED: the conn is fully established (remote address is known),
* it is now posssible to validate the remote address doesn't change.
* If Osmux peer is not behind NAT:
* We can always validate the remote IP address is matching the one we received in
* CRCX/MDCX, we can support several parallel Osmux trunks since we can differentiate same CID
* based on remote IP addr+port. However, if Osmux peer is not behind NAT it means it will go
* into OSMUX_STATE_ENABLED as soon as the remote address is known, meaning if the conn is not
* in that state we cannot (nor want) validate the IP address, we want to skip it until it is
* ready (to avoid 3rd party data injections).
* TLDR: When in OSMUX_STATE_ENABLED we can always validate remote address+port.
* When in OSMUX_STATE_ACTIVATING, in case peer behind NAT we select it,
* in case peer NOT behind NAT we skip it.
*/
if (conn_rtp->osmux.state == OSMUX_STATE_ENABLED) {
h = osmux_xfrm_input_get_deliver_cb_data(conn_rtp->osmux.in);
if (osmo_sockaddr_cmp(&h->rem_addr, rem_addr) != 0)
continue;
} else if (!trunk->cfg->osmux.peer_behind_nat) {
LOGPCONN(conn, DOSMUX, LOGL_DEBUG, "osmux_conn_lookup(rem_addr=%s local_cid=%d): Skipping because not (yet) ENABLED\n",
osmo_sockaddr_to_str(rem_addr), local_cid);
continue; /* skip, read above */
} /* else: continue CID validation below: */
if (conn_rtp->osmux.local_cid == local_cid)
return conn_rtp;
}
}
LOGP(DOSMUX, LOGL_DEBUG, "Cannot find osmux conn with rem_addr=%s local_cid=%d\n",
osmo_sockaddr_to_str(rem_addr), local_cid);
return NULL;
}
static void scheduled_from_osmux_tx_rtp_cb(struct msgb *msg, void *data)
{
struct mgcp_conn_rtp *conn = data;
const struct mgcp_endpoint *endp = conn->conn->endp;
struct osmux_handle *h = osmux_xfrm_input_get_deliver_cb_data(conn->osmux.in);
struct osmo_rtp_msg_ctx *mc = OSMO_RTP_MSG_CTX(msg);
*mc = (struct osmo_rtp_msg_ctx){
.proto = MGCP_PROTO_RTP,
.conn_src = conn,
.from_addr = &h->rem_addr,
};
endp->type->dispatch_rtp_cb(msg);
msgb_free(msg);
}
static struct msgb *osmux_recv(struct osmo_fd *ofd, struct osmo_sockaddr *addr)
{
struct msgb *msg;
socklen_t slen = sizeof(addr->u.sas);
int ret;
msg = msgb_alloc(4096, "OSMUX");
if (!msg) {
LOGP(DOSMUX, LOGL_ERROR, "cannot allocate message\n");
return NULL;
}
ret = recvfrom(ofd->fd, msg->data, msg->data_len, 0, &addr->u.sa, &slen);
if (ret <= 0) {
msgb_free(msg);
LOGP(DOSMUX, LOGL_ERROR, "cannot receive message\n");
return NULL;
}
msgb_put(msg, ret);
return msg;
}
/* To be called every time some AMR data is received on a connection
* returns: 0 if conn can process data, negative if an error ocurred and data should not be further processed */
static int conn_osmux_event_data_received(struct mgcp_conn_rtp *conn, const struct osmo_sockaddr *rem_addr)
{
const struct mgcp_config *cfg;
switch(conn->osmux.state) {
case OSMUX_STATE_ACTIVATING:
/* If peer is not behind NAT, transition to OSMUX_STATE_ENABLED is done through
* conn_osmux_rx_mdcx() whenever a CRCX/MDCX with the remote address is received.
*/
cfg = conn->conn->endp->trunk->cfg;
if (!cfg->osmux.peer_behind_nat) {
/* osmux_conn_lookup() should never provide us with an
* ACTIVATING conn without NAT in first place. This should never happen. */
LOGPCONN(conn->conn, DOSMUX, LOGL_ERROR,
"Unexpected rx osmux data for conn in ACTIVATING state without NAT\n");
return -1;
}
/* We have to wait until we received the remote CID from CRCX/MDCX. */
if (!conn->osmux.remote_cid_present)
return -1;
/* Update remote address with the src address of the package we received */
conn->end.addr = *rem_addr;
/* mgcp_rtp_end_remote_addr_available() is now true and we can enable the conn: */
if (conn_osmux_enable(conn) < 0) {
LOGPCONN(conn->conn, DOSMUX, LOGL_ERROR,
"Could not enable osmux conn: %s\n",
mgcp_conn_dump(conn->conn));
return -1;
}
return 0;
case OSMUX_STATE_ENABLED:
return 0;
default:
return -1;
}
}
/* To be called every time an CRCX/MDCX is received.
* returns: 0 if conn can continue, negative if an error ocurred during setup */
int conn_osmux_event_rx_crcx_mdcx(struct mgcp_conn_rtp *conn)
{
struct mgcp_config *cfg;
switch (conn->osmux.state) {
case OSMUX_STATE_ACTIVATING:
/* If peer is behind NAT, we have to wait until 1st osmux frame is received
* to discover peer's real remote address */
cfg = conn->conn->endp->trunk->cfg;
if (cfg->osmux.peer_behind_nat)
return 0;
/* Keep waiting to receive remote CID through CRCX/MDCX */
if (!conn->osmux.remote_cid_present)
return 0;
/* keep waiting to receive remote address through CRCX/MDCX */
if (!mgcp_rtp_end_remote_addr_available(&conn->end))
return 0;
/* Since the peer is not behind NAT, we can enable the conn right away since the proper
* remote address is now known: */
if (conn_osmux_enable(conn) < 0)
return -1;
/* We have already transitioned to OSMUX_STATE_ENABLED here */
return 0;
case OSMUX_STATE_ENABLED:
return 0;
default:
OSMO_ASSERT(NULL);
}
return 0;
}
/* Old versions of osmux used to send dummy packets [0x23 0x<CID>] to punch the
* hole in the NAT. Let's handle them speficially. */
static int osmux_handle_legacy_dummy(const struct mgcp_trunk *trunk, const struct osmo_sockaddr *rem_addr,
struct msgb *msg)
{
uint8_t osmux_cid = msg->data[1];
struct mgcp_conn_rtp *conn;
conn = osmux_conn_lookup(trunk, osmux_cid, rem_addr);
if (!conn) {
LOGP(DOSMUX, LOGL_DEBUG,
"Cannot find conn for Osmux CID %d\n", osmux_cid);
goto out;
}
conn_osmux_event_data_received(conn, rem_addr);
/* Only needed to punch hole in firewall, it can be dropped */
out:
msgb_free(msg);
return 0;
}
#define osmux_chunk_length(msg, rem) ((rem) - (msg)->len)
static int osmux_read_fd_cb(struct osmo_fd *ofd, unsigned int what)
{
struct msgb *msg;
struct osmux_hdr *osmuxh;
struct osmo_sockaddr rem_addr;
uint32_t rem;
struct mgcp_trunk *trunk = ofd->data;
struct rate_ctr_group *all_rtp_stats = trunk->ratectr.all_osmux_conn_stats;
char addr_str[64];
msg = osmux_recv(ofd, &rem_addr);
if (!msg)
return -1;
rate_ctr_inc(rate_ctr_group_get_ctr(all_rtp_stats, OSMUX_PACKETS_RX_CTR));
osmo_sockaddr_to_str_buf(addr_str, sizeof(addr_str), &rem_addr);
if (trunk->cfg->osmux.usage == OSMUX_USAGE_OFF) {
LOGP(DOSMUX, LOGL_ERROR,
"Peer %s wants to use Osmux but MGCP Client did not request it\n",
addr_str);
goto out;
}
/* Catch legacy dummy message and process them separately: */
if (msg->len == 2 && msg->data[0] == MGCP_DUMMY_LOAD)
return osmux_handle_legacy_dummy(trunk, &rem_addr, msg);
rem = msg->len;
while((osmuxh = osmux_xfrm_output_pull(msg)) != NULL) {
struct mgcp_conn_rtp *conn_src;
conn_src = osmux_conn_lookup(trunk, osmuxh->circuit_id,
&rem_addr);
if (!conn_src) {
LOGP(DOSMUX, LOGL_DEBUG,
"Cannot find a src conn for %s CID=%d\n",
addr_str, osmuxh->circuit_id);
goto next;
}
if (conn_osmux_event_data_received(conn_src, &rem_addr) < 0)
goto next;
mgcp_conn_watchdog_kick(conn_src->conn);
rtpconn_osmux_rate_ctr_inc(conn_src, OSMUX_CHUNKS_RX_CTR);
rtpconn_osmux_rate_ctr_add(conn_src, OSMUX_OCTETS_RX_CTR,
osmux_chunk_length(msg, rem));
osmux_xfrm_output_sched(conn_src->osmux.out, osmuxh);
next:
rem = msg->len;
}
out:
msgb_free(msg);
return 0;
}
int osmux_init(struct mgcp_trunk *trunk)
{
int ret;
struct mgcp_config *cfg = trunk->cfg;
/* So far we only support running on one trunk: */
OSMO_ASSERT(trunk == mgcp_trunk_by_num(cfg, MGCP_TRUNK_VIRTUAL, MGCP_VIRT_TRUNK_ID));
osmo_fd_setup(&osmux_fd_v4, -1, OSMO_FD_READ, osmux_read_fd_cb, trunk, 0);
osmo_fd_setup(&osmux_fd_v6, -1, OSMO_FD_READ, osmux_read_fd_cb, trunk, 0);
if (cfg->osmux.local_addr_v4) {
ret = mgcp_create_bind(cfg->osmux.local_addr_v4, &osmux_fd_v4, cfg->osmux.local_port,
cfg->endp_dscp, cfg->endp_priority);
if (ret < 0) {
LOGP(DOSMUX, LOGL_ERROR, "Cannot bind OSMUX IPv4 socket to %s:%u\n",
cfg->osmux.local_addr_v4, cfg->osmux.local_port);
return ret;
}
ret = osmo_fd_register(&osmux_fd_v4);
if (ret < 0) {
LOGP(DOSMUX, LOGL_ERROR, "Cannot register OSMUX IPv4 socket %s\n",
osmo_sock_get_name2(osmux_fd_v4.fd));
return ret;
}
LOGP(DOSMUX, LOGL_INFO, "OSMUX IPv4 socket listening on %s\n",
osmo_sock_get_name2(osmux_fd_v4.fd));
}
if (cfg->osmux.local_addr_v6) {
ret = mgcp_create_bind(cfg->osmux.local_addr_v6, &osmux_fd_v6, cfg->osmux.local_port,
cfg->endp_dscp, cfg->endp_priority);
if (ret < 0) {
LOGP(DOSMUX, LOGL_ERROR, "Cannot bind OSMUX IPv6 socket to [%s]:%u\n",
cfg->osmux.local_addr_v6, cfg->osmux.local_port);
return ret;
}
ret = osmo_fd_register(&osmux_fd_v6);
if (ret < 0) {
LOGP(DOSMUX, LOGL_ERROR, "Cannot register OSMUX IPv6 socket %s\n",
osmo_sock_get_name2(osmux_fd_v6.fd));
return ret;
}
LOGP(DOSMUX, LOGL_INFO, "OSMUX IPv6 socket listening on %s\n",
osmo_sock_get_name2(osmux_fd_v6.fd));
}
cfg->osmux.initialized = true;
return 0;
}
/*! relase OSXMUX cid, that had been allocated to this connection.
* \param[in] conn connection with OSMUX cid to release */
static void conn_osmux_release_local_cid(struct mgcp_conn_rtp *conn)
{
if (conn->osmux.local_cid_allocated)
osmux_cid_pool_put(conn->osmux.local_cid);
conn->osmux.local_cid = 0;
conn->osmux.local_cid_allocated = false;
}
/*! allocate local OSMUX cid to connection.
* \param[in] conn connection for which we allocate the local OSMUX cid
* \returns Allocated OSMUX cid, -1 on error (no free CIDs avail).
*/
static int conn_osmux_allocate_local_cid(struct mgcp_conn_rtp *conn)
{
OSMO_ASSERT(conn->osmux.local_cid_allocated == false);
int osmux_cid = osmux_cid_pool_get_next();
if (osmux_cid == -1) {
LOGPCONN(conn->conn, DOSMUX, LOGL_INFO,
"no available local Osmux CID to allocate!\n");
return -1;
}
conn->osmux.local_cid = (uint8_t) osmux_cid;
conn->osmux.local_cid_allocated = true;
return osmux_cid;
}
/*! Initialize Osmux bits of a conn.
* \param[in] conn Osmux connection to initialize
* \returns 0 on success, negative on ERROR */
int osmux_init_conn(struct mgcp_conn_rtp *conn)
{
if (conn_osmux_allocate_local_cid(conn) == -1)
return -1;
conn->osmux.ctrg = rate_ctr_group_alloc(conn->conn, &rate_ctr_group_osmux_desc, conn->ctrg->idx);
conn->type = MGCP_RTP_OSMUX;
/* Annotate Osmux circuit ID and set it to negotiating state until this
* is fully set up from the dummy load. */
conn->osmux.state = OSMUX_STATE_ACTIVATING;
return 0;
}
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
/*! enable OSXMUX circuit for a specified connection.
* \param[in] conn connection to enable
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
* \returns 0 on success, -1 on ERROR */
int conn_osmux_enable(struct mgcp_conn_rtp *conn)
{
OSMO_ASSERT(conn->osmux.state == OSMUX_STATE_ACTIVATING);
OSMO_ASSERT(conn->osmux.remote_cid_present == true);
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
/*! If osmux is enabled, initialize the output handler. This handler is
* used to reconstruct the RTP flow from osmux. The RTP SSRC is
* allocated based on the circuit ID (conn_net->osmux.cid), which is unique
* in the local scope to the BSC/BSC-NAT. We use it to divide the RTP
* SSRC space (2^32) by the OSMUX_CID_MAX + 1 possible circuit IDs, then randomly
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
* select one value from that window. Thus, we have no chance to have
* overlapping RTP SSRC traveling to the BTSes behind the BSC,
* similarly, for flows traveling to the MSC.
*/
const struct mgcp_trunk *trunk = conn->conn->endp->trunk;
static const uint32_t rtp_ssrc_winlen = UINT32_MAX / (OSMUX_CID_MAX + 1);
bool osmux_dummy = trunk->cfg->osmux.dummy_padding;
/* Wait until we have the remote connection information, be it from MDCX (peer not behind NAT)
* or later learned from first received remote osmux packet (peer behind NAT) */
if (!mgcp_rtp_end_remote_addr_available(&conn->end)) {
LOGPCONN(conn->conn, DOSMUX, LOGL_INFO,
"Osmux remote address/port still unknown\n");
return -1;
}
conn->osmux.in = osmux_handle_find_or_create(trunk, &conn->end.addr);
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
if (!conn->osmux.in) {
LOGPCONN(conn->conn, DOSMUX, LOGL_ERROR,
"Cannot allocate input osmux handle for conn:%s\n",
mgcp_conn_dump(conn->conn));
return -1;
}
if (osmux_xfrm_input_open_circuit(conn->osmux.in, conn->osmux.remote_cid, osmux_dummy) < 0) {
LOGPCONN(conn->conn, DOSMUX, LOGL_ERROR,
"Cannot open osmux circuit %u for conn:%s\n",
conn->osmux.remote_cid, mgcp_conn_dump(conn->conn));
return -1;
}
conn->osmux.out = osmux_xfrm_output_alloc(conn->conn);
osmux_xfrm_output_set_rtp_ssrc(conn->osmux.out,
(conn->osmux.remote_cid * rtp_ssrc_winlen) +
(random() % rtp_ssrc_winlen));
osmux_xfrm_output_set_rtp_pl_type(conn->osmux.out, conn->end.codec->payload_type);
osmux_xfrm_output_set_tx_cb(conn->osmux.out,
scheduled_from_osmux_tx_rtp_cb, conn);
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
conn->osmux.state = OSMUX_STATE_ENABLED;
LOGPCONN(conn->conn, DOSMUX, LOGL_INFO,
"Osmux CID %u towards %s is now enabled\n",
conn->osmux.remote_cid,
osmo_sockaddr_to_str(&conn->end.addr));
return 0;
}
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
/*! disable OSXMUX circuit for a specified connection.
* \param[in] conn connection to disable */
void conn_osmux_disable(struct mgcp_conn_rtp *conn)
{
OSMO_ASSERT(conn->osmux.state != OSMUX_STATE_DISABLED);
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
LOGPCONN(conn->conn, DOSMUX, LOGL_INFO,
"Releasing connection using local Osmux CID %u\n", conn->osmux.local_cid);
struct rate_ctr_group *all_osmux_stats = conn->conn->endp->trunk->ratectr.all_osmux_conn_stats;
rate_ctr_inc(rate_ctr_group_get_ctr(all_osmux_stats, OSMUX_NUM_CONNECTIONS));
if (conn->osmux.state == OSMUX_STATE_ENABLED) {
/* We are closing, we don't need pending RTP packets to be transmitted */
osmux_xfrm_output_set_tx_cb(conn->osmux.out, NULL, NULL);
TALLOC_FREE(conn->osmux.out);
osmux_xfrm_input_close_circuit(conn->osmux.in, conn->osmux.remote_cid);
conn->osmux.state = OSMUX_STATE_DISABLED;
osmux_handle_put(conn->osmux.in);
conn->osmux.remote_cid = 0;
conn->osmux.remote_cid_present = false;
}
conn_osmux_release_local_cid(conn);
rate_ctr_group_free(conn->osmux.ctrg);
conn->osmux.ctrg = NULL;
}
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
/*! send RTP dummy packet to OSMUX connection port.
* \param[in] conn associated RTP connection
* \returns bytes sent, -1 on error */
int osmux_send_dummy(struct mgcp_conn_rtp *conn)
{
char ipbuf[INET6_ADDRSTRLEN];
struct osmux_hdr *osmuxh;
int buf_len;
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
/*! The dummy packet will not be sent via the actual OSMUX connection,
* instead it is sent out of band to port where the remote OSMUX
* multplexer is listening. The goal is to ensure that the connection
* is kept open */
/*! We don't need to send the dummy load for osmux so often as another
* endpoint may have already punched the hole in the firewall. This
* approach is simple though. */
buf_len = sizeof(struct osmux_hdr) + osmo_amr_bytes(AMR_FT_0);
osmuxh = (struct osmux_hdr *) alloca(buf_len);
memset(osmuxh, 0, buf_len);
osmuxh->ft = OSMUX_FT_DUMMY;
osmuxh->amr_ft = AMR_FT_0;
osmuxh->circuit_id = conn->osmux.remote_cid;
LOGPCONN(conn->conn, DOSMUX, LOGL_DEBUG,
"sending OSMUX dummy load to %s:%u CID %u\n",
osmo_sockaddr_ntop(&conn->end.addr.u.sa, ipbuf),
osmo_sockaddr_port(&conn->end.addr.u.sa), conn->osmux.remote_cid);
return mgcp_udp_send(osmux_fd_v4.fd, &conn->end.addr, (char *)osmuxh, buf_len);
}
/* Keeps track of locally allocated Osmux circuit ID. +7 to round up to 8 bit boundary. */
static uint8_t osmux_cid_bitmap[(OSMUX_CID_MAX + 1 + 7) / 8];
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
/*! count the number of taken OSMUX cids.
* \returns number of OSMUX cids in use */
int osmux_cid_pool_count_used(void)
{
int i, j, used = 0;
for (i = 0; i < sizeof(osmux_cid_bitmap); i++) {
for (j = 0; j < 8; j++) {
if (osmux_cid_bitmap[i] & (1 << j))
used += 1;
}
}
return used;
}
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
/*! take a free OSMUX cid.
* \returns OSMUX cid */
int osmux_cid_pool_get_next(void)
{
int i, j;
for (i = 0; i < sizeof(osmux_cid_bitmap); i++) {
for (j = 0; j < 8; j++) {
if (osmux_cid_bitmap[i] & (1 << j))
continue;
osmux_cid_bitmap[i] |= (1 << j);
LOGP(DOSMUX, LOGL_DEBUG,
"Allocating Osmux CID %u from pool\n", (i * 8) + j);
return (i * 8) + j;
}
}
LOGP(DOSMUX, LOGL_ERROR, "All Osmux circuits are in use!\n");
return -1;
}
/*! take a specific OSMUX cid.
* \param[in] osmux_cid OSMUX cid */
void osmux_cid_pool_get(uint8_t osmux_cid)
{
LOGP(DOSMUX, LOGL_DEBUG, "Allocating Osmux CID %u from pool\n", osmux_cid);
osmux_cid_bitmap[osmux_cid / 8] |= (1 << (osmux_cid % 8));
}
Initially implement the new osmo-mgw and libosmo-mgcp Leave the old osmo-bsc_mgcp and libosmo-legacy-mgcp as it is; on a copy thereof (added by a previous commit), apply changes to initially implement the new osmo-mgw. Adjust build system and debian packaging to accomodate the new libosmo-mgcp and osmo-mgw. The main differences: *) use a list to manage rtp connections. Aggregate all rtp related information inside a single struct. Use a linked list to manage the both connections (net and bts). The idea behind using a list is that we might support conference calls at some later point. Store the linked list in struct mgcp_endpoint, have a private linked list for each endpoint. The list contains connection items which are implemented in struct mgcp_conn. A connection is allocated and freed using the functions in mgcp_conn.c. A connection is allocated on the reception of a CRCX command and freed with the reception of a DLCX command. *) remove external transcoder feature Fortunatelly the external transcoder feature is not needed anymore. This patch removes the related code. *) vty: get rid of CONN_BTS and CONN_NET Since the new connection model does not make a difference between BTS and NET connections the VTY should not use the fixed CONN_BTS and CONN_NET constants. - Handle the conns list inside the endpoint directly - introduce function to dump basic rtp connection info - introduce human readable names for connections Parts of the code adjusted to use generalized connections instead of explicit BTS/NET ones: - teach mgcp_send_dummy() to send dummy packets to any RTP connection - network: generalize mgcp_bind_net/bts_rtp_port() - network: generalize mgcp_send() - tap: generalize call tapping feature - stat: generalize statistics - Replace rtp_data_net() and rtp_data_bts() with generalized rtp_data_rx() *) mgcp_protocol.c fixes: - check ci string before it is converted: In case of missing ci, a nullpointer is delivered to strtoul(). Add a function that takes ci, checks it and converts it to an uint32_t. Use the return code to react on missing ci. - output error message on missing CI. - when parsing the mode, print log message when mode is missing. - use mode_orig when mode is missing. - fix ptime formatstring to use %u rather than %d. - cosmetic: log when connection is deleted on DLCX. - change loglevels of CRCX, MDCX, DLCX events from DEBUG to NOTICE. *) mgcp_test - apply rename of strline_r() to mgcp_strline(). - MGCP command macros: - Add 'I: 1' parameters. - Use proper port numbers: from m=audio 0 RTP/AVP 126 to m=audio 16002 RTP/AVP 128 - Change ptime to 'a=ptime:40' because this is what the MGW currently returns. CRCX generally feed a ptime:40 and this is expected to be returned. - struct mgcp_test: Use only one ptype, there are no explicit BTS and NET endpoints anymore. Hence remove one column from tests[]. - test_messages(): - Enable: remove '#if 0' - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. - remove endpoint init, now done internally. - add false asserts in error cases. - test_retransmission(): - remove endpoint init, now done internally. - add false asserts in error cases. - test_packet_error_detection(): - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - remove endpoint init, now done internally. - add false assert in error case. - Assert that a conn really vanishes on DLCX, previously the conn would remain and just be unused, now it is actually discarded. - test_no_cycle() - Remove concept of BTS and NET endpoints: test only one conn, as they are now interchangeable anyway. Use arbitrary conn ids (e.g. 4711). - test_no_name() - Enable: remove '#if 0'. - remove endpoint init, now done internally. - add false assert in error case. - mgcp_test.ok: adjust expected results to status quo: - We now see two dummy packets instead of one, now sent to both sides because we don't know of BTS or NET side. (maybe drop dummy packets later...) - packet duration, conn mode: now sane defaults show instead of unset. - various whitespace and formatting changes from lindent. Change-Id: Ie008599136c7ed8a0dfbb0cf803188975a499fc5
2017-08-22 14:35:41 +00:00
/*! put back a no longer used OSMUX cid.
* \param[in] osmux_cid OSMUX cid */
void osmux_cid_pool_put(uint8_t osmux_cid)
{
LOGP(DOSMUX, LOGL_DEBUG, "Osmux CID %u is back to the pool\n", osmux_cid);
osmux_cid_bitmap[osmux_cid / 8] &= ~(1 << (osmux_cid % 8));
}
/*! check if OSMUX cid is already taken */
bool osmux_cid_pool_allocated(uint8_t osmux_cid)
{
return !!(osmux_cid_bitmap[osmux_cid / 8] & (1 << (osmux_cid % 8)));
}