osmo-msc/src/libmsc/ran_conn.c

913 lines
27 KiB
C
Raw Normal View History

/* MSC RAN connection implementation */
/*
* (C) 2016-2018 by sysmocom s.m.f.c. <info@sysmocom.de>
* All Rights Reserved
*
* Author: Neels Hofmeyr
*
* 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.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU Affero General Public License for more details.
*
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*
*/
#include <osmocom/core/logging.h>
#include <osmocom/core/fsm.h>
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
#include <osmocom/core/signal.h>
#include <osmocom/msc/ran_conn.h>
#include <osmocom/msc/vlr.h>
#include <osmocom/msc/debug.h>
#include <osmocom/msc/transaction.h>
#include <osmocom/msc/signal.h>
#include <osmocom/msc/a_iface.h>
#include <osmocom/msc/sgs_iface.h>
#include <osmocom/msc/iucs.h>
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
#include "../../bscconfig.h"
#ifdef BUILD_IU
#include <osmocom/ranap/iu_client.h>
#else
#include <osmocom/msc/iu_dummy.h>
#endif
#define RAN_CONN_TIMEOUT 5 /* seconds */
static const struct value_string ran_conn_fsm_event_names[] = {
OSMO_VALUE_STRING(RAN_CONN_E_COMPLETE_LAYER_3),
OSMO_VALUE_STRING(RAN_CONN_E_CLASSMARK_UPDATE),
OSMO_VALUE_STRING(RAN_CONN_E_ACCEPTED),
OSMO_VALUE_STRING(RAN_CONN_E_COMMUNICATING),
OSMO_VALUE_STRING(RAN_CONN_E_RELEASE_WHEN_UNUSED),
OSMO_VALUE_STRING(RAN_CONN_E_MO_CLOSE),
OSMO_VALUE_STRING(RAN_CONN_E_CN_CLOSE),
OSMO_VALUE_STRING(RAN_CONN_E_UNUSED),
{ 0, NULL }
};
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
static void update_counters(struct osmo_fsm_inst *fi, bool conn_accepted)
{
struct ran_conn *conn = fi->priv;
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
switch (conn->complete_layer3_type) {
case COMPLETE_LAYER3_LU:
rate_ctr_inc(&conn->network->msc_ctrs->ctr[
conn_accepted ? MSC_CTR_LOC_UPDATE_COMPLETED
: MSC_CTR_LOC_UPDATE_FAILED]);
break;
case COMPLETE_LAYER3_CM_SERVICE_REQ:
rate_ctr_inc(&conn->network->msc_ctrs->ctr[
conn_accepted ? MSC_CTR_CM_SERVICE_REQUEST_ACCEPTED
: MSC_CTR_CM_SERVICE_REQUEST_REJECTED]);
break;
case COMPLETE_LAYER3_PAGING_RESP:
rate_ctr_inc(&conn->network->msc_ctrs->ctr[
conn_accepted ? MSC_CTR_PAGING_RESP_ACCEPTED
: MSC_CTR_PAGING_RESP_REJECTED]);
break;
default:
break;
}
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
}
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
static void evaluate_acceptance_outcome(struct osmo_fsm_inst *fi, bool conn_accepted)
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
{
struct ran_conn *conn = fi->priv;
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
update_counters(fi, conn_accepted);
/* Trigger transactions that we paged for */
if (conn->complete_layer3_type == COMPLETE_LAYER3_PAGING_RESP) {
subscr_paging_dispatch(GSM_HOOK_RR_PAGING,
conn_accepted ? GSM_PAGING_SUCCEEDED : GSM_PAGING_EXPIRED,
NULL, conn, conn->vsub);
}
if (conn->complete_layer3_type == COMPLETE_LAYER3_CM_SERVICE_REQ
&& conn_accepted) {
conn->received_cm_service_request = true;
ran_conn_get(conn, RAN_CONN_USE_CM_SERVICE);
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
}
if (conn_accepted)
osmo_signal_dispatch(SS_SUBSCR, S_SUBSCR_ATTACHED, conn->vsub);
}
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
static void log_close_event(struct osmo_fsm_inst *fi, uint32_t event, void *data)
{
enum gsm48_reject_value *cause = data;
/* The close event itself is logged by the FSM. We can only add the cause value, if present. */
if (!cause || !*cause)
return;
LOGPFSML(fi, LOGL_NOTICE, "Close event, cause: %s\n", gsm48_reject_value_name(*cause));
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
}
static void ran_conn_fsm_new(struct osmo_fsm_inst *fi, uint32_t event, void *data)
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
{
switch (event) {
case RAN_CONN_E_COMPLETE_LAYER_3:
osmo_fsm_inst_state_chg(fi, RAN_CONN_S_AUTH_CIPH, RAN_CONN_TIMEOUT, 0);
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
return;
case RAN_CONN_E_ACCEPTED:
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
evaluate_acceptance_outcome(fi, true);
osmo_fsm_inst_state_chg(fi, RAN_CONN_S_ACCEPTED, RAN_CONN_TIMEOUT, 0);
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
return;
case RAN_CONN_E_MO_CLOSE:
case RAN_CONN_E_CN_CLOSE:
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
log_close_event(fi, event, data);
evaluate_acceptance_outcome(fi, false);
/* fall through */
case RAN_CONN_E_UNUSED:
osmo_fsm_inst_state_chg(fi, RAN_CONN_S_RELEASING, RAN_CONN_TIMEOUT, 0);
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
return;
default:
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
OSMO_ASSERT(false);
}
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
}
static void ran_conn_fsm_auth_ciph(struct osmo_fsm_inst *fi, uint32_t event, void *data)
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
{
/* If accepted, transition the state, all other cases mean failure. */
switch (event) {
case RAN_CONN_E_ACCEPTED:
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
evaluate_acceptance_outcome(fi, true);
osmo_fsm_inst_state_chg(fi, RAN_CONN_S_ACCEPTED, RAN_CONN_TIMEOUT, 0);
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
return;
case RAN_CONN_E_UNUSED:
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
LOGPFSML(fi, LOGL_DEBUG, "Awaiting results for Auth+Ciph, overruling event %s\n",
osmo_fsm_event_name(fi->fsm, event));
return;
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
case RAN_CONN_E_MO_CLOSE:
case RAN_CONN_E_CN_CLOSE:
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
log_close_event(fi, event, data);
evaluate_acceptance_outcome(fi, false);
osmo_fsm_inst_state_chg(fi, RAN_CONN_S_RELEASING, RAN_CONN_TIMEOUT, 0);
return;
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
default:
OSMO_ASSERT(false);
}
}
int ran_conn_classmark_request_then_cipher_mode_cmd(struct ran_conn *conn, bool umts_aka,
A5/n Ciph: request Classmark Update if missing When the VLR requests a Ciphering Mode with vlr_ops.set_ciph_mode(), and if we need a ciph algo flag from a Classmark information that is not yet known (usually CM 2 during LU), send a BSSMAP Classmark Request to get it. To manage the intermission of the Classmark Request, add - msc_classmark_request_then_cipher_mode_cmd(), - state SUBSCR_CONN_S_WAIT_CLASSMARK_UPDATE, - event SUBSCR_CONN_E_CLASSMARK_UPDATE. From state AUTH_CIPH, switch to state WAIT_CLASSMARK_UPDATE. Once the BSSMAP Classmark Response, is received, switch back to SUBSCR_CONN_S_AUTH_CIPH and re-initiate Ciphering Mode. To be able to re-enter the Ciphering Mode algo decision, factor it out into msc_geran_set_cipher_mode(). Rationale: In the following commit, essentially we stopped supporting A5/3 ciphering: commit 71330720b6efdda2fcfd3e9c0cb45f89e32e5670 "MSC: Intersect configured A5 algorithms with MS-supported ones" Change-Id: Id124923ee52a357cb7d3e04d33f585214774f3a3 A5/3 was no longer supported because from that commit on, we strictly checked the MS-supported ciphers, but we did not have Classmark 2 available during Location Updating. This patch changes that: when Classmark 2 is missing, actively request it by a BSSMAP Classmark Request; continue Ciphering only after the Response. Always request missing Classmark, even if a lesser cipher were configured available. If the Classmark Update response fails to come in, cause an attach failure. Instead, we could attempt to use a lesser cipher that is also enabled. That is left as a future feature, should that become relevant. I think it's unlikely. Technically, we could now end up requesting a Classmark Updating both during LU (vlr_lu_fsm) and CM Service/Paging Response (proc_arq_fsm), but in practice the only time we lack a Classmark is: during Location Updating with A5/3 enabled. A5/1 support is indicated in CM1 which is always available, and A5/3 support is indicated in CM2, which is always available during CM Service Request as well as Paging Response. So this patch has practical relevance only for Location Updating. For networks that permit only A5/3, this patch fixes Location Updating. For networks that support A5/3 and A5/1, so far we always used A5/1 during LU, and after this patch we request CM2 and likely use A5/3 instead. In msc_vlr_test_gsm_ciph, verify that requesting Classmark 2 for A5/3 works during LU. Also verify that the lack of a Classmark Response results in attach failure. In msc_vlr_test_gsm_ciph, a hacky unit test fakes a situation where a CM2 is missing during proc_arq_fsm and proves that that code path works, even though the practical relevance is currently zero. It would only become interesting if ciphering algorithms A5/4 and higher became relevant, because support of those would be indicated in Classmark 3, which would always require a Classmark Request. Related: OS#3043 Depends: I4a2e1d3923e33912579c4180aa1ff8e8f5abb7e7 (libosmocore) Change-Id: I73c7cb6a86624695bd9c0f59abb72e2fdc655131
2018-09-13 01:23:07 +00:00
bool retrieve_imeisv)
{
int rc;
conn->geran_set_cipher_mode.umts_aka = umts_aka;
conn->geran_set_cipher_mode.retrieve_imeisv = retrieve_imeisv;
rc = a_iface_tx_classmark_request(conn);
if (rc) {
LOGP(DMM, LOGL_ERROR, "%s: cannot send BSSMAP Classmark Request\n",
vlr_subscr_name(conn->vsub));
return -EIO;
}
osmo_fsm_inst_state_chg(conn->fi, RAN_CONN_S_WAIT_CLASSMARK_UPDATE, RAN_CONN_TIMEOUT, 0);
A5/n Ciph: request Classmark Update if missing When the VLR requests a Ciphering Mode with vlr_ops.set_ciph_mode(), and if we need a ciph algo flag from a Classmark information that is not yet known (usually CM 2 during LU), send a BSSMAP Classmark Request to get it. To manage the intermission of the Classmark Request, add - msc_classmark_request_then_cipher_mode_cmd(), - state SUBSCR_CONN_S_WAIT_CLASSMARK_UPDATE, - event SUBSCR_CONN_E_CLASSMARK_UPDATE. From state AUTH_CIPH, switch to state WAIT_CLASSMARK_UPDATE. Once the BSSMAP Classmark Response, is received, switch back to SUBSCR_CONN_S_AUTH_CIPH and re-initiate Ciphering Mode. To be able to re-enter the Ciphering Mode algo decision, factor it out into msc_geran_set_cipher_mode(). Rationale: In the following commit, essentially we stopped supporting A5/3 ciphering: commit 71330720b6efdda2fcfd3e9c0cb45f89e32e5670 "MSC: Intersect configured A5 algorithms with MS-supported ones" Change-Id: Id124923ee52a357cb7d3e04d33f585214774f3a3 A5/3 was no longer supported because from that commit on, we strictly checked the MS-supported ciphers, but we did not have Classmark 2 available during Location Updating. This patch changes that: when Classmark 2 is missing, actively request it by a BSSMAP Classmark Request; continue Ciphering only after the Response. Always request missing Classmark, even if a lesser cipher were configured available. If the Classmark Update response fails to come in, cause an attach failure. Instead, we could attempt to use a lesser cipher that is also enabled. That is left as a future feature, should that become relevant. I think it's unlikely. Technically, we could now end up requesting a Classmark Updating both during LU (vlr_lu_fsm) and CM Service/Paging Response (proc_arq_fsm), but in practice the only time we lack a Classmark is: during Location Updating with A5/3 enabled. A5/1 support is indicated in CM1 which is always available, and A5/3 support is indicated in CM2, which is always available during CM Service Request as well as Paging Response. So this patch has practical relevance only for Location Updating. For networks that permit only A5/3, this patch fixes Location Updating. For networks that support A5/3 and A5/1, so far we always used A5/1 during LU, and after this patch we request CM2 and likely use A5/3 instead. In msc_vlr_test_gsm_ciph, verify that requesting Classmark 2 for A5/3 works during LU. Also verify that the lack of a Classmark Response results in attach failure. In msc_vlr_test_gsm_ciph, a hacky unit test fakes a situation where a CM2 is missing during proc_arq_fsm and proves that that code path works, even though the practical relevance is currently zero. It would only become interesting if ciphering algorithms A5/4 and higher became relevant, because support of those would be indicated in Classmark 3, which would always require a Classmark Request. Related: OS#3043 Depends: I4a2e1d3923e33912579c4180aa1ff8e8f5abb7e7 (libosmocore) Change-Id: I73c7cb6a86624695bd9c0f59abb72e2fdc655131
2018-09-13 01:23:07 +00:00
return 0;
}
static void ran_conn_fsm_wait_classmark_update(struct osmo_fsm_inst *fi, uint32_t event, void *data)
A5/n Ciph: request Classmark Update if missing When the VLR requests a Ciphering Mode with vlr_ops.set_ciph_mode(), and if we need a ciph algo flag from a Classmark information that is not yet known (usually CM 2 during LU), send a BSSMAP Classmark Request to get it. To manage the intermission of the Classmark Request, add - msc_classmark_request_then_cipher_mode_cmd(), - state SUBSCR_CONN_S_WAIT_CLASSMARK_UPDATE, - event SUBSCR_CONN_E_CLASSMARK_UPDATE. From state AUTH_CIPH, switch to state WAIT_CLASSMARK_UPDATE. Once the BSSMAP Classmark Response, is received, switch back to SUBSCR_CONN_S_AUTH_CIPH and re-initiate Ciphering Mode. To be able to re-enter the Ciphering Mode algo decision, factor it out into msc_geran_set_cipher_mode(). Rationale: In the following commit, essentially we stopped supporting A5/3 ciphering: commit 71330720b6efdda2fcfd3e9c0cb45f89e32e5670 "MSC: Intersect configured A5 algorithms with MS-supported ones" Change-Id: Id124923ee52a357cb7d3e04d33f585214774f3a3 A5/3 was no longer supported because from that commit on, we strictly checked the MS-supported ciphers, but we did not have Classmark 2 available during Location Updating. This patch changes that: when Classmark 2 is missing, actively request it by a BSSMAP Classmark Request; continue Ciphering only after the Response. Always request missing Classmark, even if a lesser cipher were configured available. If the Classmark Update response fails to come in, cause an attach failure. Instead, we could attempt to use a lesser cipher that is also enabled. That is left as a future feature, should that become relevant. I think it's unlikely. Technically, we could now end up requesting a Classmark Updating both during LU (vlr_lu_fsm) and CM Service/Paging Response (proc_arq_fsm), but in practice the only time we lack a Classmark is: during Location Updating with A5/3 enabled. A5/1 support is indicated in CM1 which is always available, and A5/3 support is indicated in CM2, which is always available during CM Service Request as well as Paging Response. So this patch has practical relevance only for Location Updating. For networks that permit only A5/3, this patch fixes Location Updating. For networks that support A5/3 and A5/1, so far we always used A5/1 during LU, and after this patch we request CM2 and likely use A5/3 instead. In msc_vlr_test_gsm_ciph, verify that requesting Classmark 2 for A5/3 works during LU. Also verify that the lack of a Classmark Response results in attach failure. In msc_vlr_test_gsm_ciph, a hacky unit test fakes a situation where a CM2 is missing during proc_arq_fsm and proves that that code path works, even though the practical relevance is currently zero. It would only become interesting if ciphering algorithms A5/4 and higher became relevant, because support of those would be indicated in Classmark 3, which would always require a Classmark Request. Related: OS#3043 Depends: I4a2e1d3923e33912579c4180aa1ff8e8f5abb7e7 (libosmocore) Change-Id: I73c7cb6a86624695bd9c0f59abb72e2fdc655131
2018-09-13 01:23:07 +00:00
{
struct ran_conn *conn = fi->priv;
A5/n Ciph: request Classmark Update if missing When the VLR requests a Ciphering Mode with vlr_ops.set_ciph_mode(), and if we need a ciph algo flag from a Classmark information that is not yet known (usually CM 2 during LU), send a BSSMAP Classmark Request to get it. To manage the intermission of the Classmark Request, add - msc_classmark_request_then_cipher_mode_cmd(), - state SUBSCR_CONN_S_WAIT_CLASSMARK_UPDATE, - event SUBSCR_CONN_E_CLASSMARK_UPDATE. From state AUTH_CIPH, switch to state WAIT_CLASSMARK_UPDATE. Once the BSSMAP Classmark Response, is received, switch back to SUBSCR_CONN_S_AUTH_CIPH and re-initiate Ciphering Mode. To be able to re-enter the Ciphering Mode algo decision, factor it out into msc_geran_set_cipher_mode(). Rationale: In the following commit, essentially we stopped supporting A5/3 ciphering: commit 71330720b6efdda2fcfd3e9c0cb45f89e32e5670 "MSC: Intersect configured A5 algorithms with MS-supported ones" Change-Id: Id124923ee52a357cb7d3e04d33f585214774f3a3 A5/3 was no longer supported because from that commit on, we strictly checked the MS-supported ciphers, but we did not have Classmark 2 available during Location Updating. This patch changes that: when Classmark 2 is missing, actively request it by a BSSMAP Classmark Request; continue Ciphering only after the Response. Always request missing Classmark, even if a lesser cipher were configured available. If the Classmark Update response fails to come in, cause an attach failure. Instead, we could attempt to use a lesser cipher that is also enabled. That is left as a future feature, should that become relevant. I think it's unlikely. Technically, we could now end up requesting a Classmark Updating both during LU (vlr_lu_fsm) and CM Service/Paging Response (proc_arq_fsm), but in practice the only time we lack a Classmark is: during Location Updating with A5/3 enabled. A5/1 support is indicated in CM1 which is always available, and A5/3 support is indicated in CM2, which is always available during CM Service Request as well as Paging Response. So this patch has practical relevance only for Location Updating. For networks that permit only A5/3, this patch fixes Location Updating. For networks that support A5/3 and A5/1, so far we always used A5/1 during LU, and after this patch we request CM2 and likely use A5/3 instead. In msc_vlr_test_gsm_ciph, verify that requesting Classmark 2 for A5/3 works during LU. Also verify that the lack of a Classmark Response results in attach failure. In msc_vlr_test_gsm_ciph, a hacky unit test fakes a situation where a CM2 is missing during proc_arq_fsm and proves that that code path works, even though the practical relevance is currently zero. It would only become interesting if ciphering algorithms A5/4 and higher became relevant, because support of those would be indicated in Classmark 3, which would always require a Classmark Request. Related: OS#3043 Depends: I4a2e1d3923e33912579c4180aa1ff8e8f5abb7e7 (libosmocore) Change-Id: I73c7cb6a86624695bd9c0f59abb72e2fdc655131
2018-09-13 01:23:07 +00:00
switch (event) {
case RAN_CONN_E_CLASSMARK_UPDATE:
A5/n Ciph: request Classmark Update if missing When the VLR requests a Ciphering Mode with vlr_ops.set_ciph_mode(), and if we need a ciph algo flag from a Classmark information that is not yet known (usually CM 2 during LU), send a BSSMAP Classmark Request to get it. To manage the intermission of the Classmark Request, add - msc_classmark_request_then_cipher_mode_cmd(), - state SUBSCR_CONN_S_WAIT_CLASSMARK_UPDATE, - event SUBSCR_CONN_E_CLASSMARK_UPDATE. From state AUTH_CIPH, switch to state WAIT_CLASSMARK_UPDATE. Once the BSSMAP Classmark Response, is received, switch back to SUBSCR_CONN_S_AUTH_CIPH and re-initiate Ciphering Mode. To be able to re-enter the Ciphering Mode algo decision, factor it out into msc_geran_set_cipher_mode(). Rationale: In the following commit, essentially we stopped supporting A5/3 ciphering: commit 71330720b6efdda2fcfd3e9c0cb45f89e32e5670 "MSC: Intersect configured A5 algorithms with MS-supported ones" Change-Id: Id124923ee52a357cb7d3e04d33f585214774f3a3 A5/3 was no longer supported because from that commit on, we strictly checked the MS-supported ciphers, but we did not have Classmark 2 available during Location Updating. This patch changes that: when Classmark 2 is missing, actively request it by a BSSMAP Classmark Request; continue Ciphering only after the Response. Always request missing Classmark, even if a lesser cipher were configured available. If the Classmark Update response fails to come in, cause an attach failure. Instead, we could attempt to use a lesser cipher that is also enabled. That is left as a future feature, should that become relevant. I think it's unlikely. Technically, we could now end up requesting a Classmark Updating both during LU (vlr_lu_fsm) and CM Service/Paging Response (proc_arq_fsm), but in practice the only time we lack a Classmark is: during Location Updating with A5/3 enabled. A5/1 support is indicated in CM1 which is always available, and A5/3 support is indicated in CM2, which is always available during CM Service Request as well as Paging Response. So this patch has practical relevance only for Location Updating. For networks that permit only A5/3, this patch fixes Location Updating. For networks that support A5/3 and A5/1, so far we always used A5/1 during LU, and after this patch we request CM2 and likely use A5/3 instead. In msc_vlr_test_gsm_ciph, verify that requesting Classmark 2 for A5/3 works during LU. Also verify that the lack of a Classmark Response results in attach failure. In msc_vlr_test_gsm_ciph, a hacky unit test fakes a situation where a CM2 is missing during proc_arq_fsm and proves that that code path works, even though the practical relevance is currently zero. It would only become interesting if ciphering algorithms A5/4 and higher became relevant, because support of those would be indicated in Classmark 3, which would always require a Classmark Request. Related: OS#3043 Depends: I4a2e1d3923e33912579c4180aa1ff8e8f5abb7e7 (libosmocore) Change-Id: I73c7cb6a86624695bd9c0f59abb72e2fdc655131
2018-09-13 01:23:07 +00:00
/* Theoretically, this event can be used for requesting Classmark in various situations.
* So far though, the only time we send a Classmark Request is during Ciphering. As soon
* as more such situations arise, we need to add state to indicate what action should
* follow after a Classmark Update is received (e.g.
* ran_conn_classmark_request_then_cipher_mode_cmd() sets an enum value to indicate that
A5/n Ciph: request Classmark Update if missing When the VLR requests a Ciphering Mode with vlr_ops.set_ciph_mode(), and if we need a ciph algo flag from a Classmark information that is not yet known (usually CM 2 during LU), send a BSSMAP Classmark Request to get it. To manage the intermission of the Classmark Request, add - msc_classmark_request_then_cipher_mode_cmd(), - state SUBSCR_CONN_S_WAIT_CLASSMARK_UPDATE, - event SUBSCR_CONN_E_CLASSMARK_UPDATE. From state AUTH_CIPH, switch to state WAIT_CLASSMARK_UPDATE. Once the BSSMAP Classmark Response, is received, switch back to SUBSCR_CONN_S_AUTH_CIPH and re-initiate Ciphering Mode. To be able to re-enter the Ciphering Mode algo decision, factor it out into msc_geran_set_cipher_mode(). Rationale: In the following commit, essentially we stopped supporting A5/3 ciphering: commit 71330720b6efdda2fcfd3e9c0cb45f89e32e5670 "MSC: Intersect configured A5 algorithms with MS-supported ones" Change-Id: Id124923ee52a357cb7d3e04d33f585214774f3a3 A5/3 was no longer supported because from that commit on, we strictly checked the MS-supported ciphers, but we did not have Classmark 2 available during Location Updating. This patch changes that: when Classmark 2 is missing, actively request it by a BSSMAP Classmark Request; continue Ciphering only after the Response. Always request missing Classmark, even if a lesser cipher were configured available. If the Classmark Update response fails to come in, cause an attach failure. Instead, we could attempt to use a lesser cipher that is also enabled. That is left as a future feature, should that become relevant. I think it's unlikely. Technically, we could now end up requesting a Classmark Updating both during LU (vlr_lu_fsm) and CM Service/Paging Response (proc_arq_fsm), but in practice the only time we lack a Classmark is: during Location Updating with A5/3 enabled. A5/1 support is indicated in CM1 which is always available, and A5/3 support is indicated in CM2, which is always available during CM Service Request as well as Paging Response. So this patch has practical relevance only for Location Updating. For networks that permit only A5/3, this patch fixes Location Updating. For networks that support A5/3 and A5/1, so far we always used A5/1 during LU, and after this patch we request CM2 and likely use A5/3 instead. In msc_vlr_test_gsm_ciph, verify that requesting Classmark 2 for A5/3 works during LU. Also verify that the lack of a Classmark Response results in attach failure. In msc_vlr_test_gsm_ciph, a hacky unit test fakes a situation where a CM2 is missing during proc_arq_fsm and proves that that code path works, even though the practical relevance is currently zero. It would only become interesting if ciphering algorithms A5/4 and higher became relevant, because support of those would be indicated in Classmark 3, which would always require a Classmark Request. Related: OS#3043 Depends: I4a2e1d3923e33912579c4180aa1ff8e8f5abb7e7 (libosmocore) Change-Id: I73c7cb6a86624695bd9c0f59abb72e2fdc655131
2018-09-13 01:23:07 +00:00
* Ciphering should continue afterwards). But right now, it is accurate to always
* continue with Ciphering: */
/* During Ciphering, we needed Classmark information. The Classmark Update has come in,
* go back into the Set Ciphering Command procedure. */
osmo_fsm_inst_state_chg(fi, RAN_CONN_S_AUTH_CIPH, RAN_CONN_TIMEOUT, 0);
if (ran_conn_geran_set_cipher_mode(conn, conn->geran_set_cipher_mode.umts_aka,
A5/n Ciph: request Classmark Update if missing When the VLR requests a Ciphering Mode with vlr_ops.set_ciph_mode(), and if we need a ciph algo flag from a Classmark information that is not yet known (usually CM 2 during LU), send a BSSMAP Classmark Request to get it. To manage the intermission of the Classmark Request, add - msc_classmark_request_then_cipher_mode_cmd(), - state SUBSCR_CONN_S_WAIT_CLASSMARK_UPDATE, - event SUBSCR_CONN_E_CLASSMARK_UPDATE. From state AUTH_CIPH, switch to state WAIT_CLASSMARK_UPDATE. Once the BSSMAP Classmark Response, is received, switch back to SUBSCR_CONN_S_AUTH_CIPH and re-initiate Ciphering Mode. To be able to re-enter the Ciphering Mode algo decision, factor it out into msc_geran_set_cipher_mode(). Rationale: In the following commit, essentially we stopped supporting A5/3 ciphering: commit 71330720b6efdda2fcfd3e9c0cb45f89e32e5670 "MSC: Intersect configured A5 algorithms with MS-supported ones" Change-Id: Id124923ee52a357cb7d3e04d33f585214774f3a3 A5/3 was no longer supported because from that commit on, we strictly checked the MS-supported ciphers, but we did not have Classmark 2 available during Location Updating. This patch changes that: when Classmark 2 is missing, actively request it by a BSSMAP Classmark Request; continue Ciphering only after the Response. Always request missing Classmark, even if a lesser cipher were configured available. If the Classmark Update response fails to come in, cause an attach failure. Instead, we could attempt to use a lesser cipher that is also enabled. That is left as a future feature, should that become relevant. I think it's unlikely. Technically, we could now end up requesting a Classmark Updating both during LU (vlr_lu_fsm) and CM Service/Paging Response (proc_arq_fsm), but in practice the only time we lack a Classmark is: during Location Updating with A5/3 enabled. A5/1 support is indicated in CM1 which is always available, and A5/3 support is indicated in CM2, which is always available during CM Service Request as well as Paging Response. So this patch has practical relevance only for Location Updating. For networks that permit only A5/3, this patch fixes Location Updating. For networks that support A5/3 and A5/1, so far we always used A5/1 during LU, and after this patch we request CM2 and likely use A5/3 instead. In msc_vlr_test_gsm_ciph, verify that requesting Classmark 2 for A5/3 works during LU. Also verify that the lack of a Classmark Response results in attach failure. In msc_vlr_test_gsm_ciph, a hacky unit test fakes a situation where a CM2 is missing during proc_arq_fsm and proves that that code path works, even though the practical relevance is currently zero. It would only become interesting if ciphering algorithms A5/4 and higher became relevant, because support of those would be indicated in Classmark 3, which would always require a Classmark Request. Related: OS#3043 Depends: I4a2e1d3923e33912579c4180aa1ff8e8f5abb7e7 (libosmocore) Change-Id: I73c7cb6a86624695bd9c0f59abb72e2fdc655131
2018-09-13 01:23:07 +00:00
conn->geran_set_cipher_mode.retrieve_imeisv)) {
LOGPFSML(fi, LOGL_ERROR,
"Sending Cipher Mode Command failed, aborting attach\n");
vlr_subscr_cancel_attach_fsm(conn->vsub, OSMO_FSM_TERM_ERROR,
GSM48_REJECT_NETWORK_FAILURE);
}
return;
case RAN_CONN_E_UNUSED:
A5/n Ciph: request Classmark Update if missing When the VLR requests a Ciphering Mode with vlr_ops.set_ciph_mode(), and if we need a ciph algo flag from a Classmark information that is not yet known (usually CM 2 during LU), send a BSSMAP Classmark Request to get it. To manage the intermission of the Classmark Request, add - msc_classmark_request_then_cipher_mode_cmd(), - state SUBSCR_CONN_S_WAIT_CLASSMARK_UPDATE, - event SUBSCR_CONN_E_CLASSMARK_UPDATE. From state AUTH_CIPH, switch to state WAIT_CLASSMARK_UPDATE. Once the BSSMAP Classmark Response, is received, switch back to SUBSCR_CONN_S_AUTH_CIPH and re-initiate Ciphering Mode. To be able to re-enter the Ciphering Mode algo decision, factor it out into msc_geran_set_cipher_mode(). Rationale: In the following commit, essentially we stopped supporting A5/3 ciphering: commit 71330720b6efdda2fcfd3e9c0cb45f89e32e5670 "MSC: Intersect configured A5 algorithms with MS-supported ones" Change-Id: Id124923ee52a357cb7d3e04d33f585214774f3a3 A5/3 was no longer supported because from that commit on, we strictly checked the MS-supported ciphers, but we did not have Classmark 2 available during Location Updating. This patch changes that: when Classmark 2 is missing, actively request it by a BSSMAP Classmark Request; continue Ciphering only after the Response. Always request missing Classmark, even if a lesser cipher were configured available. If the Classmark Update response fails to come in, cause an attach failure. Instead, we could attempt to use a lesser cipher that is also enabled. That is left as a future feature, should that become relevant. I think it's unlikely. Technically, we could now end up requesting a Classmark Updating both during LU (vlr_lu_fsm) and CM Service/Paging Response (proc_arq_fsm), but in practice the only time we lack a Classmark is: during Location Updating with A5/3 enabled. A5/1 support is indicated in CM1 which is always available, and A5/3 support is indicated in CM2, which is always available during CM Service Request as well as Paging Response. So this patch has practical relevance only for Location Updating. For networks that permit only A5/3, this patch fixes Location Updating. For networks that support A5/3 and A5/1, so far we always used A5/1 during LU, and after this patch we request CM2 and likely use A5/3 instead. In msc_vlr_test_gsm_ciph, verify that requesting Classmark 2 for A5/3 works during LU. Also verify that the lack of a Classmark Response results in attach failure. In msc_vlr_test_gsm_ciph, a hacky unit test fakes a situation where a CM2 is missing during proc_arq_fsm and proves that that code path works, even though the practical relevance is currently zero. It would only become interesting if ciphering algorithms A5/4 and higher became relevant, because support of those would be indicated in Classmark 3, which would always require a Classmark Request. Related: OS#3043 Depends: I4a2e1d3923e33912579c4180aa1ff8e8f5abb7e7 (libosmocore) Change-Id: I73c7cb6a86624695bd9c0f59abb72e2fdc655131
2018-09-13 01:23:07 +00:00
LOGPFSML(fi, LOGL_DEBUG, "Awaiting results for Auth+Ciph, overruling event %s\n",
osmo_fsm_event_name(fi->fsm, event));
return;
case RAN_CONN_E_MO_CLOSE:
case RAN_CONN_E_CN_CLOSE:
A5/n Ciph: request Classmark Update if missing When the VLR requests a Ciphering Mode with vlr_ops.set_ciph_mode(), and if we need a ciph algo flag from a Classmark information that is not yet known (usually CM 2 during LU), send a BSSMAP Classmark Request to get it. To manage the intermission of the Classmark Request, add - msc_classmark_request_then_cipher_mode_cmd(), - state SUBSCR_CONN_S_WAIT_CLASSMARK_UPDATE, - event SUBSCR_CONN_E_CLASSMARK_UPDATE. From state AUTH_CIPH, switch to state WAIT_CLASSMARK_UPDATE. Once the BSSMAP Classmark Response, is received, switch back to SUBSCR_CONN_S_AUTH_CIPH and re-initiate Ciphering Mode. To be able to re-enter the Ciphering Mode algo decision, factor it out into msc_geran_set_cipher_mode(). Rationale: In the following commit, essentially we stopped supporting A5/3 ciphering: commit 71330720b6efdda2fcfd3e9c0cb45f89e32e5670 "MSC: Intersect configured A5 algorithms with MS-supported ones" Change-Id: Id124923ee52a357cb7d3e04d33f585214774f3a3 A5/3 was no longer supported because from that commit on, we strictly checked the MS-supported ciphers, but we did not have Classmark 2 available during Location Updating. This patch changes that: when Classmark 2 is missing, actively request it by a BSSMAP Classmark Request; continue Ciphering only after the Response. Always request missing Classmark, even if a lesser cipher were configured available. If the Classmark Update response fails to come in, cause an attach failure. Instead, we could attempt to use a lesser cipher that is also enabled. That is left as a future feature, should that become relevant. I think it's unlikely. Technically, we could now end up requesting a Classmark Updating both during LU (vlr_lu_fsm) and CM Service/Paging Response (proc_arq_fsm), but in practice the only time we lack a Classmark is: during Location Updating with A5/3 enabled. A5/1 support is indicated in CM1 which is always available, and A5/3 support is indicated in CM2, which is always available during CM Service Request as well as Paging Response. So this patch has practical relevance only for Location Updating. For networks that permit only A5/3, this patch fixes Location Updating. For networks that support A5/3 and A5/1, so far we always used A5/1 during LU, and after this patch we request CM2 and likely use A5/3 instead. In msc_vlr_test_gsm_ciph, verify that requesting Classmark 2 for A5/3 works during LU. Also verify that the lack of a Classmark Response results in attach failure. In msc_vlr_test_gsm_ciph, a hacky unit test fakes a situation where a CM2 is missing during proc_arq_fsm and proves that that code path works, even though the practical relevance is currently zero. It would only become interesting if ciphering algorithms A5/4 and higher became relevant, because support of those would be indicated in Classmark 3, which would always require a Classmark Request. Related: OS#3043 Depends: I4a2e1d3923e33912579c4180aa1ff8e8f5abb7e7 (libosmocore) Change-Id: I73c7cb6a86624695bd9c0f59abb72e2fdc655131
2018-09-13 01:23:07 +00:00
log_close_event(fi, event, data);
evaluate_acceptance_outcome(fi, false);
osmo_fsm_inst_state_chg(fi, RAN_CONN_S_RELEASING, RAN_CONN_TIMEOUT, 0);
A5/n Ciph: request Classmark Update if missing When the VLR requests a Ciphering Mode with vlr_ops.set_ciph_mode(), and if we need a ciph algo flag from a Classmark information that is not yet known (usually CM 2 during LU), send a BSSMAP Classmark Request to get it. To manage the intermission of the Classmark Request, add - msc_classmark_request_then_cipher_mode_cmd(), - state SUBSCR_CONN_S_WAIT_CLASSMARK_UPDATE, - event SUBSCR_CONN_E_CLASSMARK_UPDATE. From state AUTH_CIPH, switch to state WAIT_CLASSMARK_UPDATE. Once the BSSMAP Classmark Response, is received, switch back to SUBSCR_CONN_S_AUTH_CIPH and re-initiate Ciphering Mode. To be able to re-enter the Ciphering Mode algo decision, factor it out into msc_geran_set_cipher_mode(). Rationale: In the following commit, essentially we stopped supporting A5/3 ciphering: commit 71330720b6efdda2fcfd3e9c0cb45f89e32e5670 "MSC: Intersect configured A5 algorithms with MS-supported ones" Change-Id: Id124923ee52a357cb7d3e04d33f585214774f3a3 A5/3 was no longer supported because from that commit on, we strictly checked the MS-supported ciphers, but we did not have Classmark 2 available during Location Updating. This patch changes that: when Classmark 2 is missing, actively request it by a BSSMAP Classmark Request; continue Ciphering only after the Response. Always request missing Classmark, even if a lesser cipher were configured available. If the Classmark Update response fails to come in, cause an attach failure. Instead, we could attempt to use a lesser cipher that is also enabled. That is left as a future feature, should that become relevant. I think it's unlikely. Technically, we could now end up requesting a Classmark Updating both during LU (vlr_lu_fsm) and CM Service/Paging Response (proc_arq_fsm), but in practice the only time we lack a Classmark is: during Location Updating with A5/3 enabled. A5/1 support is indicated in CM1 which is always available, and A5/3 support is indicated in CM2, which is always available during CM Service Request as well as Paging Response. So this patch has practical relevance only for Location Updating. For networks that permit only A5/3, this patch fixes Location Updating. For networks that support A5/3 and A5/1, so far we always used A5/1 during LU, and after this patch we request CM2 and likely use A5/3 instead. In msc_vlr_test_gsm_ciph, verify that requesting Classmark 2 for A5/3 works during LU. Also verify that the lack of a Classmark Response results in attach failure. In msc_vlr_test_gsm_ciph, a hacky unit test fakes a situation where a CM2 is missing during proc_arq_fsm and proves that that code path works, even though the practical relevance is currently zero. It would only become interesting if ciphering algorithms A5/4 and higher became relevant, because support of those would be indicated in Classmark 3, which would always require a Classmark Request. Related: OS#3043 Depends: I4a2e1d3923e33912579c4180aa1ff8e8f5abb7e7 (libosmocore) Change-Id: I73c7cb6a86624695bd9c0f59abb72e2fdc655131
2018-09-13 01:23:07 +00:00
return;
default:
OSMO_ASSERT(false);
}
}
static bool ran_conn_fsm_has_active_transactions(struct osmo_fsm_inst *fi)
{
struct ran_conn *conn = fi->priv;
struct gsm_trans *trans;
if (conn->silent_call) {
LOGPFSML(fi, LOGL_DEBUG, "%s: silent call still active\n", __func__);
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
return true;
}
if (conn->received_cm_service_request) {
LOGPFSML(fi, LOGL_DEBUG, "%s: still awaiting first request after a CM Service Request\n",
__func__);
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
return true;
}
if (conn->vsub && !llist_empty(&conn->vsub->cs.requests)) {
struct subscr_request *sr;
if (!log_check_level(fi->fsm->log_subsys, LOGL_DEBUG)) {
llist_for_each_entry(sr, &conn->vsub->cs.requests, entry) {
LOGPFSML(fi, LOGL_DEBUG, "%s: still active: %s\n",
__func__, sr->label);
}
}
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
return true;
}
if ((trans = trans_has_conn(conn))) {
LOGPFSML(fi, LOGL_DEBUG,
"%s: connection still has active transaction: %s\n",
__func__, gsm48_pdisc_name(trans->protocol));
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
return true;
}
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
return false;
}
static void ran_conn_fsm_accepted_enter(struct osmo_fsm_inst *fi, uint32_t prev_state)
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
{
struct ran_conn *conn = fi->priv;
/* Stop Location Update expiry for this subscriber. While the subscriber
* has an open connection the LU expiry timer must remain disabled.
* Otherwise we would kick the subscriber off the network when the timer
* expires e.g. during a long phone call.
* The LU expiry timer will restart once the connection is closed. */
conn->vsub->expire_lu = VLR_SUBSCRIBER_NO_EXPIRATION;
if (!ran_conn_fsm_has_active_transactions(fi))
osmo_fsm_inst_dispatch(fi, RAN_CONN_E_UNUSED, NULL);
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
}
static void ran_conn_fsm_accepted(struct osmo_fsm_inst *fi, uint32_t event, void *data)
{
switch (event) {
case RAN_CONN_E_COMPLETE_LAYER_3:
/* When Authentication is off, we may already be in the Accepted state when the code
* evaluates the Compl L3. Simply ignore. This just cosmetically mutes the error log
* about the useless event. */
return;
case RAN_CONN_E_COMMUNICATING:
osmo_fsm_inst_state_chg(fi, RAN_CONN_S_COMMUNICATING, 0, 0);
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
return;
case RAN_CONN_E_MO_CLOSE:
case RAN_CONN_E_CN_CLOSE:
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
log_close_event(fi, event, data);
/* fall through */
case RAN_CONN_E_UNUSED:
osmo_fsm_inst_state_chg(fi, RAN_CONN_S_RELEASING, RAN_CONN_TIMEOUT, 0);
return;
default:
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
OSMO_ASSERT(false);
}
}
static void ran_conn_fsm_communicating(struct osmo_fsm_inst *fi, uint32_t event, void *data)
{
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
switch (event) {
case RAN_CONN_E_COMMUNICATING:
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
/* no-op */
return;
case RAN_CONN_E_MO_CLOSE:
case RAN_CONN_E_CN_CLOSE:
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
log_close_event(fi, event, data);
/* fall through */
case RAN_CONN_E_UNUSED:
osmo_fsm_inst_state_chg(fi, RAN_CONN_S_RELEASING, RAN_CONN_TIMEOUT, 0);
return;
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
default:
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
OSMO_ASSERT(false);
}
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
}
static int ran_conn_fsm_timeout(struct osmo_fsm_inst *fi)
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
{
struct ran_conn *conn = fi->priv;
if (ran_conn_in_release(conn)) {
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
LOGPFSML(fi, LOGL_ERROR, "Timeout while releasing, discarding right now\n");
osmo_fsm_inst_term(fi, OSMO_FSM_TERM_TIMEOUT, NULL);
} else {
enum gsm48_reject_value cause = GSM48_REJECT_CONGESTION;
osmo_fsm_inst_dispatch(fi, RAN_CONN_E_CN_CLOSE, &cause);
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
}
return 0;
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
}
static void ran_conn_fsm_releasing_onenter(struct osmo_fsm_inst *fi, uint32_t prev_state)
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
{
struct ran_conn *conn = fi->priv;
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
/* The SGs interface needs to access vsub struct members to send the
* release message, however the following release procedures will
* remove conn->vsub, so we need to send the release right now. */
if (conn->via_ran == OSMO_RAT_EUTRAN_SGS) {
sgs_iface_tx_release(conn);
}
/* Use count for either conn->a.waiting_for_clear_complete or
* conn->iu.waiting_for_release_complete. 'get' it early, so we don't deallocate after tearing
* down active transactions. Safeguard against double-get (though it shouldn't happen). */
if (!ran_conn_used_by(conn, RAN_CONN_USE_RELEASE))
ran_conn_get(conn, RAN_CONN_USE_RELEASE);
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
/* Cancel pending CM Service Requests */
if (conn->received_cm_service_request) {
conn->received_cm_service_request = false;
ran_conn_put(conn, RAN_CONN_USE_CM_SERVICE);
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
}
/* Cancel all VLR FSMs, if any */
vlr_subscr_cancel_attach_fsm(conn->vsub, OSMO_FSM_TERM_ERROR, GSM48_REJECT_CONGESTION);
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
if (conn->vsub) {
/* The subscriber has no active connection anymore.
* Restart the periodic Location Update expiry timer for this subscriber. */
vlr_subscr_enable_expire_lu(conn->vsub);
}
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
/* If we're closing in a middle of a trans, we need to clean up */
trans_conn_closed(conn);
switch (conn->via_ran) {
case OSMO_RAT_GERAN_A:
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
a_iface_tx_clear_cmd(conn);
if (conn->a.waiting_for_clear_complete) {
LOGPFSML(fi, LOGL_ERROR,
"Unexpected: conn is already waiting for BSSMAP Clear Complete\n");
break;
}
conn->a.waiting_for_clear_complete = true;
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
break;
case OSMO_RAT_UTRAN_IU:
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
ranap_iu_tx_release(conn->iu.ue_ctx, NULL);
if (conn->iu.waiting_for_release_complete) {
LOGPFSML(fi, LOGL_ERROR,
"Unexpected: conn is already waiting for Iu Release Complete\n");
break;
}
conn->iu.waiting_for_release_complete = true;
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
break;
case OSMO_RAT_EUTRAN_SGS:
/* Release message is already sent at the beginning of this
* functions (see above), but we still need to notify the
* conn that a release has been sent / is in progress. */
ran_conn_sgs_release_sent(conn);
break;
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
default:
LOGP(DMM, LOGL_ERROR, "%s: Unknown RAN type, cannot tx release/clear\n",
vlr_subscr_name(conn->vsub));
break;
}
}
static void ran_conn_fsm_releasing(struct osmo_fsm_inst *fi, uint32_t event, void *data)
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
{
OSMO_ASSERT(event == RAN_CONN_E_UNUSED);
osmo_fsm_inst_state_chg(fi, RAN_CONN_S_RELEASED, 0, 0);
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
}
static void ran_conn_fsm_released(struct osmo_fsm_inst *fi, uint32_t prev_state)
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
{
/* Terminate, deallocate and also deallocate the ran_conn, which is allocated as
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
* a talloc child of fi. Also calls the cleanup function. */
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
osmo_fsm_inst_term(fi, OSMO_FSM_TERM_REGULAR, NULL);
}
#define S(x) (1 << (x))
static const struct osmo_fsm_state ran_conn_fsm_states[] = {
[RAN_CONN_S_NEW] = {
.name = OSMO_STRINGIFY(RAN_CONN_S_NEW),
.in_event_mask = S(RAN_CONN_E_COMPLETE_LAYER_3) |
S(RAN_CONN_E_ACCEPTED) |
S(RAN_CONN_E_MO_CLOSE) |
S(RAN_CONN_E_CN_CLOSE) |
S(RAN_CONN_E_UNUSED),
.out_state_mask = S(RAN_CONN_S_AUTH_CIPH) |
S(RAN_CONN_S_ACCEPTED) |
S(RAN_CONN_S_RELEASING),
.action = ran_conn_fsm_new,
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
},
[RAN_CONN_S_AUTH_CIPH] = {
.name = OSMO_STRINGIFY(RAN_CONN_S_AUTH_CIPH),
.in_event_mask = S(RAN_CONN_E_ACCEPTED) |
S(RAN_CONN_E_MO_CLOSE) |
S(RAN_CONN_E_CN_CLOSE) |
S(RAN_CONN_E_UNUSED),
.out_state_mask = S(RAN_CONN_S_WAIT_CLASSMARK_UPDATE) |
S(RAN_CONN_S_ACCEPTED) |
S(RAN_CONN_S_RELEASING),
.action = ran_conn_fsm_auth_ciph,
},
[RAN_CONN_S_WAIT_CLASSMARK_UPDATE] = {
.name = OSMO_STRINGIFY(RAN_CONN_S_WAIT_CLASSMARK_UPDATE),
.in_event_mask = S(RAN_CONN_E_CLASSMARK_UPDATE) |
S(RAN_CONN_E_MO_CLOSE) |
S(RAN_CONN_E_CN_CLOSE) |
S(RAN_CONN_E_UNUSED),
.out_state_mask = S(RAN_CONN_S_AUTH_CIPH) |
S(RAN_CONN_S_RELEASING),
.action = ran_conn_fsm_wait_classmark_update,
A5/n Ciph: request Classmark Update if missing When the VLR requests a Ciphering Mode with vlr_ops.set_ciph_mode(), and if we need a ciph algo flag from a Classmark information that is not yet known (usually CM 2 during LU), send a BSSMAP Classmark Request to get it. To manage the intermission of the Classmark Request, add - msc_classmark_request_then_cipher_mode_cmd(), - state SUBSCR_CONN_S_WAIT_CLASSMARK_UPDATE, - event SUBSCR_CONN_E_CLASSMARK_UPDATE. From state AUTH_CIPH, switch to state WAIT_CLASSMARK_UPDATE. Once the BSSMAP Classmark Response, is received, switch back to SUBSCR_CONN_S_AUTH_CIPH and re-initiate Ciphering Mode. To be able to re-enter the Ciphering Mode algo decision, factor it out into msc_geran_set_cipher_mode(). Rationale: In the following commit, essentially we stopped supporting A5/3 ciphering: commit 71330720b6efdda2fcfd3e9c0cb45f89e32e5670 "MSC: Intersect configured A5 algorithms with MS-supported ones" Change-Id: Id124923ee52a357cb7d3e04d33f585214774f3a3 A5/3 was no longer supported because from that commit on, we strictly checked the MS-supported ciphers, but we did not have Classmark 2 available during Location Updating. This patch changes that: when Classmark 2 is missing, actively request it by a BSSMAP Classmark Request; continue Ciphering only after the Response. Always request missing Classmark, even if a lesser cipher were configured available. If the Classmark Update response fails to come in, cause an attach failure. Instead, we could attempt to use a lesser cipher that is also enabled. That is left as a future feature, should that become relevant. I think it's unlikely. Technically, we could now end up requesting a Classmark Updating both during LU (vlr_lu_fsm) and CM Service/Paging Response (proc_arq_fsm), but in practice the only time we lack a Classmark is: during Location Updating with A5/3 enabled. A5/1 support is indicated in CM1 which is always available, and A5/3 support is indicated in CM2, which is always available during CM Service Request as well as Paging Response. So this patch has practical relevance only for Location Updating. For networks that permit only A5/3, this patch fixes Location Updating. For networks that support A5/3 and A5/1, so far we always used A5/1 during LU, and after this patch we request CM2 and likely use A5/3 instead. In msc_vlr_test_gsm_ciph, verify that requesting Classmark 2 for A5/3 works during LU. Also verify that the lack of a Classmark Response results in attach failure. In msc_vlr_test_gsm_ciph, a hacky unit test fakes a situation where a CM2 is missing during proc_arq_fsm and proves that that code path works, even though the practical relevance is currently zero. It would only become interesting if ciphering algorithms A5/4 and higher became relevant, because support of those would be indicated in Classmark 3, which would always require a Classmark Request. Related: OS#3043 Depends: I4a2e1d3923e33912579c4180aa1ff8e8f5abb7e7 (libosmocore) Change-Id: I73c7cb6a86624695bd9c0f59abb72e2fdc655131
2018-09-13 01:23:07 +00:00
},
[RAN_CONN_S_ACCEPTED] = {
.name = OSMO_STRINGIFY(RAN_CONN_S_ACCEPTED),
/* allow everything to release for any odd behavior */
.in_event_mask = S(RAN_CONN_E_COMPLETE_LAYER_3) |
S(RAN_CONN_E_COMMUNICATING) |
S(RAN_CONN_E_RELEASE_WHEN_UNUSED) |
S(RAN_CONN_E_ACCEPTED) |
S(RAN_CONN_E_MO_CLOSE) |
S(RAN_CONN_E_CN_CLOSE) |
S(RAN_CONN_E_UNUSED),
.out_state_mask = S(RAN_CONN_S_RELEASING) |
S(RAN_CONN_S_COMMUNICATING),
.onenter = ran_conn_fsm_accepted_enter,
.action = ran_conn_fsm_accepted,
},
[RAN_CONN_S_COMMUNICATING] = {
.name = OSMO_STRINGIFY(RAN_CONN_S_COMMUNICATING),
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
/* allow everything to release for any odd behavior */
.in_event_mask = S(RAN_CONN_E_RELEASE_WHEN_UNUSED) |
S(RAN_CONN_E_ACCEPTED) |
S(RAN_CONN_E_COMMUNICATING) |
S(RAN_CONN_E_MO_CLOSE) |
S(RAN_CONN_E_CN_CLOSE) |
S(RAN_CONN_E_UNUSED),
.out_state_mask = S(RAN_CONN_S_RELEASING),
.action = ran_conn_fsm_communicating,
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
},
[RAN_CONN_S_RELEASING] = {
.name = OSMO_STRINGIFY(RAN_CONN_S_RELEASING),
.in_event_mask = S(RAN_CONN_E_UNUSED),
.out_state_mask = S(RAN_CONN_S_RELEASED),
.onenter = ran_conn_fsm_releasing_onenter,
.action = ran_conn_fsm_releasing,
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
},
[RAN_CONN_S_RELEASED] = {
.name = OSMO_STRINGIFY(RAN_CONN_S_RELEASED),
.onenter = ran_conn_fsm_released,
},
};
static void ran_conn_fsm_cleanup(struct osmo_fsm_inst *fi, enum osmo_fsm_term_cause cause);
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
static struct osmo_fsm ran_conn_fsm = {
.name = "RAN_conn",
.states = ran_conn_fsm_states,
.num_states = ARRAY_SIZE(ran_conn_fsm_states),
.allstate_event_mask = 0,
.allstate_action = NULL,
.log_subsys = DMM,
.event_names = ran_conn_fsm_event_names,
.cleanup = ran_conn_fsm_cleanup,
.timer_cb = ran_conn_fsm_timeout,
};
refactor log ctx for vlr_subscr and ran_conn ran_conn_get_conn_id(): instead of a talloc allocated string, return a static buffer in ran_conn_get_conn_id(). So far this function had no callers. Refactor ran_conn_update_id() API: during early L3-Complete, when no subscriber is associated yet, update the FSM Id by the MI type seen in the L3 Complete message: ran_conn_update_id_from_mi(). Later on set the vsub and re-update. Call vlr.ops->subscr_update when the TMSI is updated, so that log context includes the TMSI from then on. Enrich context for vlr_subscr_name and ran_conn fi name. Include all available information in vlr_subscr_name(); instead of either IMSI or MSISDN or TMSI, print all of them when present. Instead of a short log, rather have more valuable context. A context info would now look like: Process_Access_Request_VLR(IMSI-901700000014706:MSISDN-2023:TMSI-0x08BDE4EC:GERAN-A-3:PAGING_RESP) It does get quite long, but ensures easy correlation of any BSSAP / IuCS messages with log output, especially if multiple subscribers are busy at the same time. Print TMSI and TMSInew in uppercase hexadecimal, which is the typical representation in the telecom world. When showing the RAN conn id GERAN_A-00000017 becomes GERAN-A-23 - We usually write the conn_id in decimal. - Leading zeros are clutter and might suggest hexadecimal format. - 'GERAN-A' and 'UTRAN-Iu' are the strings defined by osmo_rat_type_name(). Depends: I7798c3ef983c2e333b2b9cbffef6f366f370bd81 (libosmocore) Depends: Ica25919758ef6cba8348da199b0ae7e0ba628798 (libosmocore) Change-Id: I66a68ce2eb8957a35855a3743d91a86299900834
2019-01-03 01:32:14 +00:00
/* Return statically allocated string of the ran_conn RAT type and id. */
const char *ran_conn_get_conn_id(struct ran_conn *conn)
{
refactor log ctx for vlr_subscr and ran_conn ran_conn_get_conn_id(): instead of a talloc allocated string, return a static buffer in ran_conn_get_conn_id(). So far this function had no callers. Refactor ran_conn_update_id() API: during early L3-Complete, when no subscriber is associated yet, update the FSM Id by the MI type seen in the L3 Complete message: ran_conn_update_id_from_mi(). Later on set the vsub and re-update. Call vlr.ops->subscr_update when the TMSI is updated, so that log context includes the TMSI from then on. Enrich context for vlr_subscr_name and ran_conn fi name. Include all available information in vlr_subscr_name(); instead of either IMSI or MSISDN or TMSI, print all of them when present. Instead of a short log, rather have more valuable context. A context info would now look like: Process_Access_Request_VLR(IMSI-901700000014706:MSISDN-2023:TMSI-0x08BDE4EC:GERAN-A-3:PAGING_RESP) It does get quite long, but ensures easy correlation of any BSSAP / IuCS messages with log output, especially if multiple subscribers are busy at the same time. Print TMSI and TMSInew in uppercase hexadecimal, which is the typical representation in the telecom world. When showing the RAN conn id GERAN_A-00000017 becomes GERAN-A-23 - We usually write the conn_id in decimal. - Leading zeros are clutter and might suggest hexadecimal format. - 'GERAN-A' and 'UTRAN-Iu' are the strings defined by osmo_rat_type_name(). Depends: I7798c3ef983c2e333b2b9cbffef6f366f370bd81 (libosmocore) Depends: Ica25919758ef6cba8348da199b0ae7e0ba628798 (libosmocore) Change-Id: I66a68ce2eb8957a35855a3743d91a86299900834
2019-01-03 01:32:14 +00:00
static char id[42];
int rc;
uint32_t conn_id;
switch (conn->via_ran) {
case OSMO_RAT_GERAN_A:
refactor log ctx for vlr_subscr and ran_conn ran_conn_get_conn_id(): instead of a talloc allocated string, return a static buffer in ran_conn_get_conn_id(). So far this function had no callers. Refactor ran_conn_update_id() API: during early L3-Complete, when no subscriber is associated yet, update the FSM Id by the MI type seen in the L3 Complete message: ran_conn_update_id_from_mi(). Later on set the vsub and re-update. Call vlr.ops->subscr_update when the TMSI is updated, so that log context includes the TMSI from then on. Enrich context for vlr_subscr_name and ran_conn fi name. Include all available information in vlr_subscr_name(); instead of either IMSI or MSISDN or TMSI, print all of them when present. Instead of a short log, rather have more valuable context. A context info would now look like: Process_Access_Request_VLR(IMSI-901700000014706:MSISDN-2023:TMSI-0x08BDE4EC:GERAN-A-3:PAGING_RESP) It does get quite long, but ensures easy correlation of any BSSAP / IuCS messages with log output, especially if multiple subscribers are busy at the same time. Print TMSI and TMSInew in uppercase hexadecimal, which is the typical representation in the telecom world. When showing the RAN conn id GERAN_A-00000017 becomes GERAN-A-23 - We usually write the conn_id in decimal. - Leading zeros are clutter and might suggest hexadecimal format. - 'GERAN-A' and 'UTRAN-Iu' are the strings defined by osmo_rat_type_name(). Depends: I7798c3ef983c2e333b2b9cbffef6f366f370bd81 (libosmocore) Depends: Ica25919758ef6cba8348da199b0ae7e0ba628798 (libosmocore) Change-Id: I66a68ce2eb8957a35855a3743d91a86299900834
2019-01-03 01:32:14 +00:00
conn_id = conn->a.conn_id;
break;
case OSMO_RAT_UTRAN_IU:
refactor log ctx for vlr_subscr and ran_conn ran_conn_get_conn_id(): instead of a talloc allocated string, return a static buffer in ran_conn_get_conn_id(). So far this function had no callers. Refactor ran_conn_update_id() API: during early L3-Complete, when no subscriber is associated yet, update the FSM Id by the MI type seen in the L3 Complete message: ran_conn_update_id_from_mi(). Later on set the vsub and re-update. Call vlr.ops->subscr_update when the TMSI is updated, so that log context includes the TMSI from then on. Enrich context for vlr_subscr_name and ran_conn fi name. Include all available information in vlr_subscr_name(); instead of either IMSI or MSISDN or TMSI, print all of them when present. Instead of a short log, rather have more valuable context. A context info would now look like: Process_Access_Request_VLR(IMSI-901700000014706:MSISDN-2023:TMSI-0x08BDE4EC:GERAN-A-3:PAGING_RESP) It does get quite long, but ensures easy correlation of any BSSAP / IuCS messages with log output, especially if multiple subscribers are busy at the same time. Print TMSI and TMSInew in uppercase hexadecimal, which is the typical representation in the telecom world. When showing the RAN conn id GERAN_A-00000017 becomes GERAN-A-23 - We usually write the conn_id in decimal. - Leading zeros are clutter and might suggest hexadecimal format. - 'GERAN-A' and 'UTRAN-Iu' are the strings defined by osmo_rat_type_name(). Depends: I7798c3ef983c2e333b2b9cbffef6f366f370bd81 (libosmocore) Depends: Ica25919758ef6cba8348da199b0ae7e0ba628798 (libosmocore) Change-Id: I66a68ce2eb8957a35855a3743d91a86299900834
2019-01-03 01:32:14 +00:00
conn_id = iu_get_conn_id(conn->iu.ue_ctx);
break;
default:
refactor log ctx for vlr_subscr and ran_conn ran_conn_get_conn_id(): instead of a talloc allocated string, return a static buffer in ran_conn_get_conn_id(). So far this function had no callers. Refactor ran_conn_update_id() API: during early L3-Complete, when no subscriber is associated yet, update the FSM Id by the MI type seen in the L3 Complete message: ran_conn_update_id_from_mi(). Later on set the vsub and re-update. Call vlr.ops->subscr_update when the TMSI is updated, so that log context includes the TMSI from then on. Enrich context for vlr_subscr_name and ran_conn fi name. Include all available information in vlr_subscr_name(); instead of either IMSI or MSISDN or TMSI, print all of them when present. Instead of a short log, rather have more valuable context. A context info would now look like: Process_Access_Request_VLR(IMSI-901700000014706:MSISDN-2023:TMSI-0x08BDE4EC:GERAN-A-3:PAGING_RESP) It does get quite long, but ensures easy correlation of any BSSAP / IuCS messages with log output, especially if multiple subscribers are busy at the same time. Print TMSI and TMSInew in uppercase hexadecimal, which is the typical representation in the telecom world. When showing the RAN conn id GERAN_A-00000017 becomes GERAN-A-23 - We usually write the conn_id in decimal. - Leading zeros are clutter and might suggest hexadecimal format. - 'GERAN-A' and 'UTRAN-Iu' are the strings defined by osmo_rat_type_name(). Depends: I7798c3ef983c2e333b2b9cbffef6f366f370bd81 (libosmocore) Depends: Ica25919758ef6cba8348da199b0ae7e0ba628798 (libosmocore) Change-Id: I66a68ce2eb8957a35855a3743d91a86299900834
2019-01-03 01:32:14 +00:00
return "ran-unknown";
}
refactor log ctx for vlr_subscr and ran_conn ran_conn_get_conn_id(): instead of a talloc allocated string, return a static buffer in ran_conn_get_conn_id(). So far this function had no callers. Refactor ran_conn_update_id() API: during early L3-Complete, when no subscriber is associated yet, update the FSM Id by the MI type seen in the L3 Complete message: ran_conn_update_id_from_mi(). Later on set the vsub and re-update. Call vlr.ops->subscr_update when the TMSI is updated, so that log context includes the TMSI from then on. Enrich context for vlr_subscr_name and ran_conn fi name. Include all available information in vlr_subscr_name(); instead of either IMSI or MSISDN or TMSI, print all of them when present. Instead of a short log, rather have more valuable context. A context info would now look like: Process_Access_Request_VLR(IMSI-901700000014706:MSISDN-2023:TMSI-0x08BDE4EC:GERAN-A-3:PAGING_RESP) It does get quite long, but ensures easy correlation of any BSSAP / IuCS messages with log output, especially if multiple subscribers are busy at the same time. Print TMSI and TMSInew in uppercase hexadecimal, which is the typical representation in the telecom world. When showing the RAN conn id GERAN_A-00000017 becomes GERAN-A-23 - We usually write the conn_id in decimal. - Leading zeros are clutter and might suggest hexadecimal format. - 'GERAN-A' and 'UTRAN-Iu' are the strings defined by osmo_rat_type_name(). Depends: I7798c3ef983c2e333b2b9cbffef6f366f370bd81 (libosmocore) Depends: Ica25919758ef6cba8348da199b0ae7e0ba628798 (libosmocore) Change-Id: I66a68ce2eb8957a35855a3743d91a86299900834
2019-01-03 01:32:14 +00:00
rc = snprintf(id, sizeof(id), "%s-%u", osmo_rat_type_name(conn->via_ran), conn_id);
/* < 0 is error, == 0 is empty, >= size means truncation. Not really expecting this to catch on in any practical
* situation. */
if (rc <= 0 || rc >= sizeof(id)) {
LOGP(DMM, LOGL_ERROR, "Error with conn id; rc=%d\n", rc);
return "conn-id-error";
}
return id;
}
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
/* Tidy up before the FSM deallocates */
static void ran_conn_fsm_cleanup(struct osmo_fsm_inst *fi, enum osmo_fsm_term_cause cause)
{
struct ran_conn *conn = fi->priv;
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
if (ran_conn_fsm_has_active_transactions(fi)) {
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
LOGPFSML(fi, LOGL_ERROR, "Deallocating despite active transactions\n");
trans_conn_closed(conn);
}
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
if (!conn) {
LOGP(DRLL, LOGL_ERROR, "Freeing NULL RAN connection\n");
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
return;
}
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
if (conn->vsub) {
DEBUGP(DRLL, "%s: Freeing RAN connection\n", vlr_subscr_name(conn->vsub));
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
conn->vsub->lu_fsm = NULL;
conn->vsub->msc_conn_ref = NULL;
vlr_subscr_put(conn->vsub, VSUB_USE_CONN);
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
conn->vsub = NULL;
} else
DEBUGP(DRLL, "Freeing RAN connection with NULL subscriber\n");
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
llist_del(&conn->entry);
}
/* Signal success of Complete Layer 3. Allow to keep the conn open for Auth and Ciph. */
void ran_conn_complete_layer_3(struct ran_conn *conn)
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
{
if (!conn)
return;
osmo_fsm_inst_dispatch(conn->fi, RAN_CONN_E_COMPLETE_LAYER_3, NULL);
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
}
void ran_conn_release_when_unused(struct ran_conn *conn)
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
{
if (!conn)
return;
if (ran_conn_in_release(conn)) {
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
DEBUGP(DMM, "%s: %s: conn already in release (%s)\n",
vlr_subscr_name(conn->vsub), __func__,
osmo_fsm_inst_state_name(conn->fi));
return;
}
if (conn->fi->state == RAN_CONN_S_NEW) {
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
DEBUGP(DMM, "%s: %s: conn still being established (%s)\n",
vlr_subscr_name(conn->vsub), __func__,
osmo_fsm_inst_state_name(conn->fi));
return;
}
osmo_fsm_inst_dispatch(conn->fi, RAN_CONN_E_RELEASE_WHEN_UNUSED, NULL);
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
}
static void conn_close(struct ran_conn *conn, uint32_t cause, uint32_t event)
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
{
if (!conn) {
LOGP(DMM, LOGL_ERROR, "Cannot release NULL connection\n");
return;
}
if (ran_conn_in_release(conn)) {
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
DEBUGP(DMM, "%s(vsub=%s, cause=%u): already in release, ignore.\n",
__func__, vlr_subscr_name(conn->vsub), cause);
return;
}
osmo_fsm_inst_dispatch(conn->fi, event, &cause);
}
void ran_conn_close(struct ran_conn *conn, uint32_t cause)
{
return conn_close(conn, cause, RAN_CONN_E_CN_CLOSE);
}
void ran_conn_mo_close(struct ran_conn *conn, uint32_t cause)
{
return conn_close(conn, cause, RAN_CONN_E_MO_CLOSE);
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
}
bool ran_conn_in_release(struct ran_conn *conn)
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
{
if (!conn || !conn->fi)
return true;
if (conn->fi->state == RAN_CONN_S_RELEASING)
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
return true;
if (conn->fi->state == RAN_CONN_S_RELEASED)
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
return true;
return false;
}
bool ran_conn_is_accepted(const struct ran_conn *conn)
{
if (!conn)
return false;
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
if (!conn->vsub)
return false;
if (!(conn->fi->state == RAN_CONN_S_ACCEPTED
|| conn->fi->state == RAN_CONN_S_COMMUNICATING))
return false;
return true;
}
/* Indicate that *some* communication is happening with the phone, so that the conn FSM no longer times
* out to release within a few seconds. */
void ran_conn_communicating(struct ran_conn *conn)
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
{
osmo_fsm_inst_dispatch(conn->fi, RAN_CONN_E_COMMUNICATING, NULL);
Use libvlr in libmsc (large refactoring) Original libvlr code is by Harald Welte <laforge@gnumonks.org>, polished and tweaked by Neels Hofmeyr <nhofmeyr@sysmocom.de>. This is a long series of trial-and-error development collapsed in one patch. This may be split in smaller commits if reviewers prefer that. If we can keep it as one, we have saved ourselves the additional separation work. SMS: The SQL based lookup of SMS for attached subscribers no longer works since the SQL database no longer has the subscriber data. Replace with a round-robin on the SMS recipient MSISDNs paired with a VLR subscriber RAM lookup whether the subscriber is currently attached. If there are many SMS for not-attached subscribers in the SMS database, this will become inefficient: a DB hit returns a pending SMS, the RAM lookup will reveal that the subscriber is not attached, after which the DB is hit for the next SMS. It would become more efficient e.g. by having an MSISDN based hash list for the VLR subscribers and by marking non-attached SMS recipients in the SMS database so that they can be excluded with the SQL query already. There is a sanity limit to do at most 100 db hits per attempt to find a pending SMS. So if there are more than 100 stored SMS waiting for their recipients to actually attach to the MSC, it may take more than one SMS queue trigger to deliver SMS for subscribers that are actually attached. This is not very beautiful, but is merely intended to carry us over to a time when we have a proper separate SMSC entity. Introduce gsm_subscriber_connection ref-counting in libmsc. Remove/Disable VTY and CTRL commands to create subscribers, which is now a task of the OsmoHLR. Adjust the python tests accordingly. Remove VTY cmd subscriber-keep-in-ram. Use OSMO_GSUP_PORT = 4222 instead of 2222. See I4222e21686c823985be8ff1f16b1182be8ad6175. So far use the LAC from conn->bts, will be replaced by conn->lac in Id3705236350d5f69e447046b0a764bbabc3d493c. Related: OS#1592 OS#1974 Change-Id: I639544a6cdda77a3aafc4e3446a55393f60e4050
2016-06-19 16:06:02 +00:00
}
void ran_conn_init(void)
{
osmo_fsm_register(&ran_conn_fsm);
}
/* Allocate a new RAN conn and FSM.
* Deallocation is by ran_conn_put(): when the use count reaches zero, the
* RAN_CONN_E_RELEASE_COMPLETE event is dispatched, the FSM terminates and deallocates both FSM and
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
* conn. As long as the FSM is waiting for responses from the subscriber, it will itself hold a use count
* on the conn. */
struct ran_conn *ran_conn_alloc(struct gsm_network *network,
enum osmo_rat_type via_ran, uint16_t lac)
{
struct ran_conn *conn;
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
struct osmo_fsm_inst *fi;
fi = osmo_fsm_inst_alloc(&ran_conn_fsm, network, NULL, LOGL_DEBUG, NULL);
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
if (!fi) {
LOGP(DMM, LOGL_ERROR, "Failed to allocate conn FSM\n");
return NULL;
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
}
conn = talloc_zero(fi, struct ran_conn);
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
if (!conn) {
osmo_fsm_inst_free(fi);
return NULL;
}
*conn = (struct ran_conn){
.network = network,
.via_ran = via_ran,
.lac = lac,
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
.fi = fi,
};
switch (via_ran) {
case OSMO_RAT_GERAN_A:
conn->log_subsys = DBSSAP;
break;
case OSMO_RAT_UTRAN_IU:
conn->log_subsys = DRANAP;
break;
case OSMO_RAT_EUTRAN_SGS:
conn->log_subsys = DSGS;
break;
default:
conn->log_subsys = DMSC;
break;
}
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
fi->priv = conn;
llist_add_tail(&conn->entry, &network->ran_conns);
return conn;
}
bool ran_conn_is_establishing_auth_ciph(const struct ran_conn *conn)
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
{
if (!conn)
return false;
return conn->fi->state == RAN_CONN_S_AUTH_CIPH;
refactor subscr_conn and subscr_conn_fsm de-/alloc Refactor: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. 2. Add separate AUTH_CIPH state to the FSM. 3. Use conn->use_count to trigger conn release. 4. Add separate RELEASING state to the FSM. 5. Add rate counters for each of the three Complete Layer 3 types. Details: 1. Glue the gsm_subscriber_connection alloc to the subscr_conn_fsm. Historically, a gsm_subscriber_connection was allocated in libbsc land, and only upon Complete Layer 3 did libmsc add the fsm instance. After splitting openbsc.git into a separate osmo-msc, this is no longer necessary, hence: Closely tie gsm_subscriber_connection allocation to the subscr_conn_fsm instance: talloc the conn as a child of the FSM instance, and discard the conn as soon as the FSM terminates. 2. Add separate AUTH_CIPH state to the FSM. Decoding the Complete Layer 3 message is distinctly separate from waiting for the VLR FSMs to conclude. Use the NEW state as "we don't know if this is a valid message yet", and the AUTH_CIPH state as "evaluating, don't release". A profound effect of this: should we for any odd reason fail to leave the FSM's NEW state, the conn will be released right at the end of msc_compl_l3(), without needing to trigger release in each code path. 3. Use conn->use_count to trigger conn release. Before, the FSM itself would hold a use count on the conn, and hence we would need to ask it whether it is ready to release the conn yet by dispatching events, to achieve a use_count decrement. Instead, unite the FSM instance and conn, and do not hold a use count by the FSM. Hence, trigger an FSM "UNUSED" event only when the use_count reaches zero. As long as use counts are done correctly, the FSM will terminate correctly. These exceptions: - The new AUTH_CIPH state explicitly ignores UNUSED events, since we expect the use count to reach zero while evaluating Authentication and Ciphering. (I experimented with holding a use count by AUTH_CIPH onenter() and releasing by onleave(), but the use count and thus the conn are released before the next state can initiate transactions that would increment the use count again. Same thing for the VLR FSMs holding a use count, they should be done before we advance to the next state. The easiest is to simply expect zero use count during the AUTH_CIPH state.) - A CM Service Request means that even though the MSC would be through with all it wants to do, we shall still wait for a request to follow from the MS. Hence the FSM holds a use count on itself while a CM Service is pending. - While waiting for a Release/Clear Complete, the FSM holds a use count on itself. 4. Add separate RELEASING state to the FSM. If we decide to release for other reasons than a use count reaching zero, we still need to be able to wait for the msc_dtap() use count on the conn to release. (An upcoming patch will further use the RELEASING state to properly wait for Clear Complete / Release Complete messages.) 5. Add rate counters for each of the three Complete Layer 3 types. Besides LU, also count CM Service Request and Paging Response acceptance/rejections. Without these counters, only very few of the auth+ciph outcomes actually show in the counters. Related: OS#3122 Change-Id: I55feb379e176a96a831e105b86202b17a0ffe889
2018-03-30 22:02:14 +00:00
}
const struct value_string complete_layer3_type_names[] = {
{ COMPLETE_LAYER3_NONE, "NONE" },
{ COMPLETE_LAYER3_LU, "LU" },
{ COMPLETE_LAYER3_CM_SERVICE_REQ, "CM_SERVICE_REQ" },
{ COMPLETE_LAYER3_PAGING_RESP, "PAGING_RESP" },
{ 0, NULL }
};
refactor log ctx for vlr_subscr and ran_conn ran_conn_get_conn_id(): instead of a talloc allocated string, return a static buffer in ran_conn_get_conn_id(). So far this function had no callers. Refactor ran_conn_update_id() API: during early L3-Complete, when no subscriber is associated yet, update the FSM Id by the MI type seen in the L3 Complete message: ran_conn_update_id_from_mi(). Later on set the vsub and re-update. Call vlr.ops->subscr_update when the TMSI is updated, so that log context includes the TMSI from then on. Enrich context for vlr_subscr_name and ran_conn fi name. Include all available information in vlr_subscr_name(); instead of either IMSI or MSISDN or TMSI, print all of them when present. Instead of a short log, rather have more valuable context. A context info would now look like: Process_Access_Request_VLR(IMSI-901700000014706:MSISDN-2023:TMSI-0x08BDE4EC:GERAN-A-3:PAGING_RESP) It does get quite long, but ensures easy correlation of any BSSAP / IuCS messages with log output, especially if multiple subscribers are busy at the same time. Print TMSI and TMSInew in uppercase hexadecimal, which is the typical representation in the telecom world. When showing the RAN conn id GERAN_A-00000017 becomes GERAN-A-23 - We usually write the conn_id in decimal. - Leading zeros are clutter and might suggest hexadecimal format. - 'GERAN-A' and 'UTRAN-Iu' are the strings defined by osmo_rat_type_name(). Depends: I7798c3ef983c2e333b2b9cbffef6f366f370bd81 (libosmocore) Depends: Ica25919758ef6cba8348da199b0ae7e0ba628798 (libosmocore) Change-Id: I66a68ce2eb8957a35855a3743d91a86299900834
2019-01-03 01:32:14 +00:00
static void _ran_conn_update_id(struct ran_conn *conn, const char *subscr_identity)
{
refactor log ctx for vlr_subscr and ran_conn ran_conn_get_conn_id(): instead of a talloc allocated string, return a static buffer in ran_conn_get_conn_id(). So far this function had no callers. Refactor ran_conn_update_id() API: during early L3-Complete, when no subscriber is associated yet, update the FSM Id by the MI type seen in the L3 Complete message: ran_conn_update_id_from_mi(). Later on set the vsub and re-update. Call vlr.ops->subscr_update when the TMSI is updated, so that log context includes the TMSI from then on. Enrich context for vlr_subscr_name and ran_conn fi name. Include all available information in vlr_subscr_name(); instead of either IMSI or MSISDN or TMSI, print all of them when present. Instead of a short log, rather have more valuable context. A context info would now look like: Process_Access_Request_VLR(IMSI-901700000014706:MSISDN-2023:TMSI-0x08BDE4EC:GERAN-A-3:PAGING_RESP) It does get quite long, but ensures easy correlation of any BSSAP / IuCS messages with log output, especially if multiple subscribers are busy at the same time. Print TMSI and TMSInew in uppercase hexadecimal, which is the typical representation in the telecom world. When showing the RAN conn id GERAN_A-00000017 becomes GERAN-A-23 - We usually write the conn_id in decimal. - Leading zeros are clutter and might suggest hexadecimal format. - 'GERAN-A' and 'UTRAN-Iu' are the strings defined by osmo_rat_type_name(). Depends: I7798c3ef983c2e333b2b9cbffef6f366f370bd81 (libosmocore) Depends: Ica25919758ef6cba8348da199b0ae7e0ba628798 (libosmocore) Change-Id: I66a68ce2eb8957a35855a3743d91a86299900834
2019-01-03 01:32:14 +00:00
struct vlr_subscr *vsub = conn->vsub;
if (osmo_fsm_inst_update_id_f(conn->fi, "%s:%s:%s",
subscr_identity,
ran_conn_get_conn_id(conn),
complete_layer3_type_name(conn->complete_layer3_type))
!= 0)
return; /* osmo_fsm_inst_update_id_f() will log an error. */
if (vsub) {
if (vsub->lu_fsm)
osmo_fsm_inst_update_id(vsub->lu_fsm, conn->fi->id);
if (vsub->auth_fsm)
osmo_fsm_inst_update_id(vsub->auth_fsm, conn->fi->id);
if (vsub->proc_arq_fsm)
osmo_fsm_inst_update_id(vsub->proc_arq_fsm, conn->fi->id);
}
LOGPFSML(conn->fi, LOGL_DEBUG, "Updated ID\n");
}
/* Compose an ID almost like gsm48_mi_to_string(), but print the MI type along, and print a TMSI as hex. */
void ran_conn_update_id_from_mi(struct ran_conn *conn, const uint8_t *mi, uint8_t mi_len)
{
_ran_conn_update_id(conn, osmo_mi_name(mi, mi_len));
}
/* Update ran_conn->fi id string from current conn->vsub and conn->complete_layer3_type. */
void ran_conn_update_id(struct ran_conn *conn)
{
_ran_conn_update_id(conn, vlr_subscr_name(conn->vsub));
}
/* Iterate all ran_conn instances that are relevant for this subscriber, and update FSM ID strings for all of the FSM
* instances. */
void ran_conn_update_id_for_vsub(struct vlr_subscr *for_vsub)
{
struct gsm_network *network;
struct ran_conn *conn;
if (!for_vsub)
return;
network = for_vsub->vlr->user_ctx;
OSMO_ASSERT(network);
llist_for_each_entry(conn, &network->ran_conns, entry) {
if (conn->vsub == for_vsub)
ran_conn_update_id(conn);
}
}
static void rx_close_complete(struct ran_conn *conn, const char *label, bool *flag)
{
if (!conn)
return;
if (!ran_conn_in_release(conn)) {
LOGPFSML(conn->fi, LOGL_ERROR, "Received unexpected %s, discarding right now\n",
label);
trans_conn_closed(conn);
osmo_fsm_inst_term(conn->fi, OSMO_FSM_TERM_ERROR, NULL);
return;
}
if (*flag) {
*flag = false;
ran_conn_put(conn, RAN_CONN_USE_RELEASE);
}
}
void ran_conn_rx_bssmap_clear_complete(struct ran_conn *conn)
{
rx_close_complete(conn, "BSSMAP Clear Complete", &conn->a.waiting_for_clear_complete);
}
void ran_conn_rx_iu_release_complete(struct ran_conn *conn)
{
rx_close_complete(conn, "Iu Release Complete", &conn->iu.waiting_for_release_complete);
}
void ran_conn_sgs_release_sent(struct ran_conn *conn)
{
bool dummy_waiting_for_release_complete = true;
/* Note: In SGsAP there is no confirmation of a release. */
rx_close_complete(conn, "SGs Release Complete", &dummy_waiting_for_release_complete);
}
const struct value_string ran_conn_use_names[] = {
{ RAN_CONN_USE_UNTRACKED, "UNTRACKED" },
{ RAN_CONN_USE_COMPL_L3, "compl_l3" },
{ RAN_CONN_USE_DTAP, "dtap" },
{ RAN_CONN_USE_AUTH_CIPH, "auth+ciph" },
{ RAN_CONN_USE_CM_SERVICE, "cm_service" },
{ RAN_CONN_USE_TRANS_CC, "trans_cc" },
{ RAN_CONN_USE_TRANS_SMS, "trans_sms" },
{ RAN_CONN_USE_TRANS_NC_SS, "trans_nc_ss" },
{ RAN_CONN_USE_SILENT_CALL, "silent_call" },
{ RAN_CONN_USE_RELEASE, "release" },
{ 0, NULL }
};
static const char *used_ref_counts_str(struct ran_conn *conn)
{
static char buf[256];
int bit_nr;
char *pos = buf;
*pos = '\0';
if (conn->use_tokens < 0)
return "invalid";
#define APPEND_STR(fmt, args...) do { \
int remain = sizeof(buf) - (pos - buf) - 1; \
int l = -1; \
if (remain > 0) \
l = snprintf(pos, remain, "%s" fmt, (pos == buf? "" : ","), ##args); \
if (l < 0 || l > remain) { \
buf[sizeof(buf) - 1] = '\0'; \
return buf; \
} \
pos += l; \
} while(0)
for (bit_nr = 0; (1 << bit_nr) <= conn->use_tokens; bit_nr++) {
if (conn->use_tokens & (1 << bit_nr)) {
APPEND_STR("%s", get_value_string(ran_conn_use_names, bit_nr));
}
}
return buf;
#undef APPEND_STR
}
/* increment the ref-count. Needs to be called by every user */
struct ran_conn *_ran_conn_get(struct ran_conn *conn, enum ran_conn_use balance_token,
const char *file, int line)
{
OSMO_ASSERT(conn);
if (balance_token != RAN_CONN_USE_UNTRACKED) {
uint32_t flag = 1 << balance_token;
OSMO_ASSERT(balance_token < 32);
if (conn->use_tokens & flag)
LOGPSRC(DREF, LOGL_ERROR, file, line,
"%s: MSC conn use error: using an already used token: %s\n",
vlr_subscr_name(conn->vsub),
ran_conn_use_name(balance_token));
conn->use_tokens |= flag;
}
conn->use_count++;
LOGPSRC(DREF, LOGL_DEBUG, file, line,
"%s: MSC conn use + %s == %u (0x%x: %s)\n",
vlr_subscr_name(conn->vsub), ran_conn_use_name(balance_token),
conn->use_count, conn->use_tokens, used_ref_counts_str(conn));
return conn;
}
/* decrement the ref-count. Once it reaches zero, we release */
void _ran_conn_put(struct ran_conn *conn, enum ran_conn_use balance_token,
const char *file, int line)
{
OSMO_ASSERT(conn);
if (balance_token != RAN_CONN_USE_UNTRACKED) {
uint32_t flag = 1 << balance_token;
OSMO_ASSERT(balance_token < 32);
if (!(conn->use_tokens & flag))
LOGPSRC(DREF, LOGL_ERROR, file, line,
"%s: MSC conn use error: freeing an unused token: %s\n",
vlr_subscr_name(conn->vsub),
ran_conn_use_name(balance_token));
conn->use_tokens &= ~flag;
}
if (conn->use_count == 0) {
LOGPSRC(DREF, LOGL_ERROR, file, line,
"%s: MSC conn use - %s failed: is already 0\n",
vlr_subscr_name(conn->vsub),
ran_conn_use_name(balance_token));
return;
}
conn->use_count--;
LOGPSRC(DREF, LOGL_DEBUG, file, line,
"%s: MSC conn use - %s == %u (0x%x: %s)\n",
vlr_subscr_name(conn->vsub), ran_conn_use_name(balance_token),
conn->use_count, conn->use_tokens, used_ref_counts_str(conn));
if (conn->use_count == 0)
osmo_fsm_inst_dispatch(conn->fi, RAN_CONN_E_UNUSED, NULL);
}
bool ran_conn_used_by(struct ran_conn *conn, enum ran_conn_use token)
{
return conn && (conn->use_tokens & (1 << token));
}