2018-03-30 21:04:04 +00:00
|
|
|
full talloc report on 'msgb' (total 0 bytes in 1 blocks)
|
|
|
|
talloc_total_blocks(tall_bsc_ctx) == 12
|
|
|
|
|
2018-03-01 23:40:58 +00:00
|
|
|
===== test_early_stage
|
2017-01-25 14:04:16 +00:00
|
|
|
- NULL conn
|
|
|
|
msc_subscr_conn_is_accepted() == false
|
|
|
|
- freshly allocated 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
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_NEW}: Allocated
|
2017-01-25 14:04:16 +00:00
|
|
|
msc_subscr_conn_is_accepted() == false
|
|
|
|
- conn_fsm present, in state NEW
|
|
|
|
msc_subscr_conn_is_accepted() == false
|
|
|
|
requests shall be thwarted
|
2017-03-10 01:15:20 +00:00
|
|
|
DRLL Dispatching 04.08 message GSM48_MT_CC_SETUP (0x3:0x5)
|
|
|
|
DRLL subscr unknown: Message not permitted for initial conn: GSM48_MT_CC_SETUP
|
|
|
|
DRLL Dispatching 04.08 message unknown 0x33 (0x5:0x33)
|
|
|
|
DRLL subscr unknown: Message not permitted for initial conn: unknown 0x33
|
|
|
|
DRLL Dispatching 04.08 message GSM48_MT_RR_SYSINFO_1 (0x6:0x19)
|
|
|
|
DRLL subscr unknown: Message not permitted for initial conn: GSM48_MT_RR_SYSINFO_1
|
2018-01-24 21:53:00 +00:00
|
|
|
DRLL Dispatching 04.08 message SMS:0x01 (0x9:0x1)
|
|
|
|
DRLL subscr unknown: Message not permitted for initial conn: SMS:0x01
|
2017-01-25 14:04:16 +00:00
|
|
|
- fake: acceptance
|
|
|
|
DREF VLR subscr unknown usage increases to: 1
|
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
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_NEW}: state_chg to SUBSCR_CONN_S_ACCEPTED
|
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_ACCEPTED}: subscr_conn_fsm_has_active_transactions: silent call still active
|
2017-01-25 14:04:16 +00:00
|
|
|
llist_count(&net->subscr_conns) == 1
|
|
|
|
msc_subscr_conn_is_accepted() == true
|
|
|
|
- CLOSE event marks conn_fsm as released and frees the 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
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_ACCEPTED}: Received Event SUBSCR_CONN_E_CN_CLOSE
|
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_ACCEPTED}: state_chg to SUBSCR_CONN_S_RELEASING
|
|
|
|
DREF unknown: MSC conn use + release == 1 (0x100)
|
|
|
|
DREF VLR subscr unknown usage increases to: 2
|
|
|
|
DREF VLR subscr unknown usage decreases to: 1
|
2017-04-09 10:32:51 +00:00
|
|
|
- BSSAP Clear --RAN_GERAN_A--> MS
|
2018-04-01 18:55:54 +00:00
|
|
|
bssap_clear_sent == 1
|
|
|
|
- BSS sends BSSMAP Clear Complete
|
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
|
|
|
DREF unknown: MSC conn use - release == 0 (0x0)
|
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_RELEASING}: Received Event SUBSCR_CONN_E_UNUSED
|
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_RELEASING}: state_chg to SUBSCR_CONN_S_RELEASED
|
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_RELEASED}: Terminating (cause = OSMO_FSM_TERM_REGULAR)
|
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_RELEASED}: subscr_conn_fsm_has_active_transactions: silent call still active
|
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_RELEASED}: Deallocating despite active transactions
|
|
|
|
DRLL unknown: Freeing subscriber connection
|
2017-01-25 14:04:16 +00:00
|
|
|
DREF VLR subscr unknown usage decreases to: 0
|
|
|
|
DREF freeing VLR subscr unknown
|
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
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_RELEASED}: Freeing instance
|
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_RELEASED}: Deallocated
|
2017-01-25 14:04:16 +00:00
|
|
|
llist_count(&net->subscr_conns) == 0
|
2018-03-01 23:40:58 +00:00
|
|
|
===== test_early_stage: SUCCESS
|
2017-01-25 14:04:16 +00:00
|
|
|
|
|
|
|
full talloc report on 'msgb' (total 0 bytes in 1 blocks)
|
2018-03-30 21:04:04 +00:00
|
|
|
talloc_total_blocks(tall_bsc_ctx) == 12
|
2017-01-25 14:04:16 +00:00
|
|
|
|
2018-03-01 23:40:58 +00:00
|
|
|
===== test_cm_service_without_lu
|
2017-01-25 14:04:16 +00:00
|
|
|
- CM Service Request without a prior Location Updating
|
2017-03-10 01:15:20 +00:00
|
|
|
MSC <--RAN_GERAN_A-- MS: GSM48_MT_MM_CM_SERV_REQ
|
2017-01-25 14:04:16 +00:00
|
|
|
new 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
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_NEW}: Allocated
|
2017-11-22 13:33:12 +00:00
|
|
|
DREF unknown: MSC conn use + compl_l3 == 1 (0x1)
|
2017-03-10 01:15:20 +00:00
|
|
|
DRLL Dispatching 04.08 message GSM48_MT_MM_CM_SERV_REQ (0x5:0x24)
|
2017-01-25 14:04:16 +00:00
|
|
|
DMM <- CM SERVICE REQUEST serv_type=0x08 MI(IMSI)=901700000004620
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(CM_SERVICE_REQ:901700000004620){SUBSCR_CONN_S_NEW}: Updated ID
|
|
|
|
DVLR Process_Access_Request_VLR(CM_SERVICE_REQ:901700000004620){PR_ARQ_S_INIT}: Allocated
|
|
|
|
DVLR Process_Access_Request_VLR(CM_SERVICE_REQ:901700000004620){PR_ARQ_S_INIT}: is child of Subscr_Conn(CM_SERVICE_REQ:901700000004620)
|
|
|
|
DVLR Process_Access_Request_VLR(CM_SERVICE_REQ:901700000004620){PR_ARQ_S_INIT}: rev=GSM net=GERAN (no Auth)
|
|
|
|
DVLR Process_Access_Request_VLR(CM_SERVICE_REQ:901700000004620){PR_ARQ_S_INIT}: Received Event PR_ARQ_E_START
|
2018-04-06 00:57:51 +00:00
|
|
|
DVLR Process_Access_Request_VLR(CM_SERVICE_REQ:901700000004620){PR_ARQ_S_INIT}: proc_arq_fsm_done(MS_IDENTITY_NOT_DERVIVABLE)
|
2018-04-02 21:17:50 +00:00
|
|
|
DVLR Process_Access_Request_VLR(CM_SERVICE_REQ:901700000004620){PR_ARQ_S_INIT}: state_chg to PR_ARQ_S_DONE
|
2018-04-06 00:57:51 +00:00
|
|
|
DVLR Process_Access_Request_VLR(CM_SERVICE_REQ:901700000004620){PR_ARQ_S_DONE}: Process Access Request result: MS_IDENTITY_NOT_DERVIVABLE
|
|
|
|
- sending CM Service Reject for unknown, cause: MS_IDENTITY_NOT_DERVIVABLE
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(CM_SERVICE_REQ:901700000004620){SUBSCR_CONN_S_NEW}: Received Event SUBSCR_CONN_E_CN_CLOSE
|
|
|
|
DMM Subscr_Conn(CM_SERVICE_REQ:901700000004620){SUBSCR_CONN_S_NEW}: state_chg to SUBSCR_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
|
|
|
DREF unknown: MSC conn use + release == 2 (0x101)
|
|
|
|
- BSSAP Clear --RAN_GERAN_A--> MS
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(CM_SERVICE_REQ:901700000004620){SUBSCR_CONN_S_RELEASING}: Received Event SUBSCR_CONN_E_COMPLETE_LAYER_3
|
|
|
|
DMM Subscr_Conn(CM_SERVICE_REQ:901700000004620){SUBSCR_CONN_S_RELEASING}: Event SUBSCR_CONN_E_COMPLETE_LAYER_3 not permitted
|
2018-04-01 18:55:54 +00:00
|
|
|
DREF unknown: MSC conn use - compl_l3 == 1 (0x100)
|
|
|
|
bssap_clear_sent == 1
|
|
|
|
- conn was released
|
|
|
|
- BSS sends BSSMAP Clear Complete
|
|
|
|
DREF unknown: MSC conn use - release == 0 (0x0)
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(CM_SERVICE_REQ:901700000004620){SUBSCR_CONN_S_RELEASING}: Received Event SUBSCR_CONN_E_UNUSED
|
|
|
|
DMM Subscr_Conn(CM_SERVICE_REQ:901700000004620){SUBSCR_CONN_S_RELEASING}: state_chg to SUBSCR_CONN_S_RELEASED
|
|
|
|
DMM Subscr_Conn(CM_SERVICE_REQ:901700000004620){SUBSCR_CONN_S_RELEASED}: Terminating (cause = OSMO_FSM_TERM_REGULAR)
|
|
|
|
DVLR Process_Access_Request_VLR(CM_SERVICE_REQ:901700000004620){PR_ARQ_S_DONE}: Terminating (cause = OSMO_FSM_TERM_PARENT)
|
|
|
|
DVLR Process_Access_Request_VLR(CM_SERVICE_REQ:901700000004620){PR_ARQ_S_DONE}: Removing from parent Subscr_Conn(CM_SERVICE_REQ:901700000004620)
|
|
|
|
DVLR Process_Access_Request_VLR(CM_SERVICE_REQ:901700000004620){PR_ARQ_S_DONE}: Freeing instance
|
|
|
|
DVLR Process_Access_Request_VLR(CM_SERVICE_REQ:901700000004620){PR_ARQ_S_DONE}: Deallocated
|
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
|
|
|
DRLL Freeing subscriber connection with NULL subscriber
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(CM_SERVICE_REQ:901700000004620){SUBSCR_CONN_S_RELEASED}: Freeing instance
|
|
|
|
DMM Subscr_Conn(CM_SERVICE_REQ:901700000004620){SUBSCR_CONN_S_RELEASED}: Deallocated
|
2017-01-25 14:04:16 +00:00
|
|
|
llist_count(&net->subscr_conns) == 0
|
2018-03-01 23:40:58 +00:00
|
|
|
===== test_cm_service_without_lu: SUCCESS
|
2017-01-25 14:04:16 +00:00
|
|
|
|
|
|
|
full talloc report on 'msgb' (total 0 bytes in 1 blocks)
|
2018-03-30 21:04:04 +00:00
|
|
|
talloc_total_blocks(tall_bsc_ctx) == 12
|
2017-01-25 14:04:16 +00:00
|
|
|
|
2018-03-01 23:40:58 +00:00
|
|
|
===== test_two_lu
|
2017-01-25 14:04:16 +00:00
|
|
|
- Location Update request causes a GSUP LU request to HLR
|
2017-03-10 01:15:20 +00:00
|
|
|
MSC <--RAN_GERAN_A-- MS: GSM48_MT_MM_LOC_UPD_REQUEST
|
2017-01-25 14:04:16 +00:00
|
|
|
new 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
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_NEW}: Allocated
|
2017-11-22 13:33:12 +00:00
|
|
|
DREF unknown: MSC conn use + compl_l3 == 1 (0x1)
|
2017-03-10 01:15:20 +00:00
|
|
|
DRLL Dispatching 04.08 message GSM48_MT_MM_LOC_UPD_REQUEST (0x5:0x8)
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_NEW}: Updated ID
|
2017-01-25 14:04:16 +00:00
|
|
|
DMM LOCATION UPDATING REQUEST: MI(IMSI)=901700000004620 type=IMSI ATTACH
|
2017-11-22 01:55:52 +00:00
|
|
|
DMM LU/new-LAC: 1/23
|
2018-04-02 21:17:50 +00:00
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_IDLE}: Allocated
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_IDLE}: is child of Subscr_Conn(LU:901700000004620)
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_IDLE}: rev=GSM net=GERAN (no Auth)
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_IDLE}: Received Event VLR_ULA_E_UPDATE_LA
|
2017-01-25 14:04:16 +00:00
|
|
|
DREF VLR subscr unknown usage increases to: 1
|
|
|
|
DVLR set IMSI on subscriber; IMSI=901700000004620 id=901700000004620
|
|
|
|
DVLR New subscr, IMSI: 901700000004620
|
|
|
|
DREF VLR subscr IMSI:901700000004620 usage increases to: 2
|
2017-07-18 13:39:27 +00:00
|
|
|
DREF VLR subscr IMSI:901700000004620 usage decreases to: 1
|
2018-04-02 21:17:50 +00:00
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_IDLE}: vlr_loc_upd_node1()
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_IDLE}: vlr_loc_upd_post_auth()
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_IDLE}: vlr_loc_upd_post_ciph()
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_IDLE}: vlr_loc_upd_node_4()
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_IDLE}: state_chg to VLR_ULA_S_WAIT_HLR_UPD
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_INIT}: Allocated
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_INIT}: is child of vlr_lu_fsm(LU:901700000004620)
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_INIT}: Received Event UPD_HLR_VLR_E_START
|
2017-01-25 14:04:16 +00:00
|
|
|
DVLR GSUP tx: 04010809710000004026f0
|
|
|
|
GSUP --> HLR: OSMO_GSUP_MSGT_UPDATE_LOCATION_REQUEST: 04010809710000004026f0
|
2018-04-02 21:17:50 +00:00
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_INIT}: state_chg to UPD_HLR_VLR_S_WAIT_FOR_DATA
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_NEW}: Received Event SUBSCR_CONN_E_COMPLETE_LAYER_3
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_NEW}: state_chg to SUBSCR_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
|
|
|
DREF IMSI:901700000004620: MSC conn use - compl_l3 == 0 (0x0)
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_AUTH_CIPH}: Received Event SUBSCR_CONN_E_UNUSED
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_AUTH_CIPH}: Awaiting results for Auth+Ciph, overruling event SUBSCR_CONN_E_UNUSED
|
2017-01-25 14:04:16 +00:00
|
|
|
lu_result_sent == 0
|
|
|
|
- HLR sends _INSERT_DATA_REQUEST, VLR responds with _INSERT_DATA_RESULT
|
|
|
|
<-- GSUP rx OSMO_GSUP_MSGT_INSERT_DATA_REQUEST: 10010809710000004026f00804036470f1
|
|
|
|
DVLR GSUP rx 17: 10010809710000004026f00804036470f1
|
|
|
|
DREF VLR subscr IMSI:901700000004620 usage increases to: 2
|
|
|
|
DVLR IMSI:901700000004620 has MSISDN:46071
|
2017-12-27 12:23:44 +00:00
|
|
|
DVLR SUBSCR(MSISDN:46071) VLR: update for IMSI=901700000004620 (MSISDN=46071, used=2)
|
2017-01-25 14:04:16 +00:00
|
|
|
DVLR GSUP tx: 12010809710000004026f0
|
|
|
|
GSUP --> HLR: OSMO_GSUP_MSGT_INSERT_DATA_RESULT: 12010809710000004026f0
|
|
|
|
DREF VLR subscr MSISDN:46071 usage decreases to: 1
|
|
|
|
<-- GSUP rx OSMO_GSUP_MSGT_INSERT_DATA_REQUEST: vlr_gsupc_read_cb() returns 0
|
|
|
|
lu_result_sent == 0
|
|
|
|
- having received subscriber data does not mean acceptance
|
|
|
|
msc_subscr_conn_is_accepted() == false
|
|
|
|
requests shall be thwarted
|
2017-03-10 01:15:20 +00:00
|
|
|
DRLL Dispatching 04.08 message GSM48_MT_CC_SETUP (0x3:0x5)
|
|
|
|
DRLL subscr MSISDN:46071: Message not permitted for initial conn: GSM48_MT_CC_SETUP
|
|
|
|
DRLL Dispatching 04.08 message unknown 0x33 (0x5:0x33)
|
|
|
|
DRLL subscr MSISDN:46071: Message not permitted for initial conn: unknown 0x33
|
|
|
|
DRLL Dispatching 04.08 message GSM48_MT_RR_SYSINFO_1 (0x6:0x19)
|
|
|
|
DRLL subscr MSISDN:46071: Message not permitted for initial conn: GSM48_MT_RR_SYSINFO_1
|
2018-01-24 21:53:00 +00:00
|
|
|
DRLL Dispatching 04.08 message SMS:0x01 (0x9:0x1)
|
|
|
|
DRLL subscr MSISDN:46071: Message not permitted for initial conn: SMS:0x01
|
2017-01-25 14:04:16 +00:00
|
|
|
lu_result_sent == 0
|
|
|
|
- HLR also sends GSUP _UPDATE_LOCATION_RESULT
|
|
|
|
<-- GSUP rx OSMO_GSUP_MSGT_UPDATE_LOCATION_RESULT: 06010809710000004026f0
|
|
|
|
DVLR GSUP rx 11: 06010809710000004026f0
|
|
|
|
DREF VLR subscr MSISDN:46071 usage increases to: 2
|
2018-04-02 21:17:50 +00:00
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_WAIT_HLR_UPD}: Received Event VLR_ULA_E_HLR_LU_RES
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_WAIT_FOR_DATA}: Received Event UPD_HLR_VLR_E_UPD_LOC_ACK
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_WAIT_FOR_DATA}: state_chg to UPD_HLR_VLR_S_DONE
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_DONE}: Terminating (cause = OSMO_FSM_TERM_REGULAR)
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_DONE}: Removing from parent vlr_lu_fsm(LU:901700000004620)
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_DONE}: Freeing instance
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_DONE}: Deallocated
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_WAIT_HLR_UPD}: Received Event VLR_ULA_E_UPD_HLR_COMPL
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_WAIT_HLR_UPD}: state_chg to VLR_ULA_S_WAIT_LU_COMPL
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_INIT}: Allocated
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_INIT}: is child of vlr_lu_fsm(LU:901700000004620)
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_INIT}: Received Event LU_COMPL_VLR_E_START
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_INIT}: state_chg to LU_COMPL_VLR_S_WAIT_SUB_PRES
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:901700000004620){SUB_PRES_VLR_S_INIT}: Allocated
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:901700000004620){SUB_PRES_VLR_S_INIT}: is child of lu_compl_vlr_fsm(LU:901700000004620)
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:901700000004620){SUB_PRES_VLR_S_INIT}: Received Event SUB_PRES_VLR_E_START
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:901700000004620){SUB_PRES_VLR_S_INIT}: state_chg to SUB_PRES_VLR_S_DONE
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:901700000004620){SUB_PRES_VLR_S_DONE}: Terminating (cause = OSMO_FSM_TERM_REGULAR)
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:901700000004620){SUB_PRES_VLR_S_DONE}: Removing from parent lu_compl_vlr_fsm(LU:901700000004620)
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:901700000004620){SUB_PRES_VLR_S_DONE}: Freeing instance
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:901700000004620){SUB_PRES_VLR_S_DONE}: Deallocated
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_WAIT_SUB_PRES}: Received Event LU_COMPL_VLR_E_SUB_PRES_COMPL
|
2017-01-25 14:04:16 +00:00
|
|
|
- sending LU Accept for MSISDN:46071
|
|
|
|
DREF VLR subscr MSISDN:46071 usage increases to: 3
|
2018-04-02 21:17:50 +00:00
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_WAIT_SUB_PRES}: state_chg to LU_COMPL_VLR_S_DONE
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_WAIT_LU_COMPL}: Received Event VLR_ULA_E_LU_COMPL_SUCCESS
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_DONE}: Terminating (cause = OSMO_FSM_TERM_PARENT)
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_DONE}: Removing from parent vlr_lu_fsm(LU:901700000004620)
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_DONE}: Freeing instance
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_DONE}: Deallocated
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_WAIT_LU_COMPL}: state_chg to VLR_ULA_S_DONE
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_AUTH_CIPH}: Received Event SUBSCR_CONN_E_ACCEPTED
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_AUTH_CIPH}: state_chg to SUBSCR_CONN_S_ACCEPTED
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_ACCEPTED}: Received Event SUBSCR_CONN_E_UNUSED
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_ACCEPTED}: state_chg to SUBSCR_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
|
|
|
DREF MSISDN:46071: MSC conn use + release == 1 (0x100)
|
|
|
|
DREF VLR subscr MSISDN:46071 usage increases to: 4
|
|
|
|
DREF VLR subscr MSISDN:46071 usage decreases to: 3
|
|
|
|
- BSSAP Clear --RAN_GERAN_A--> MS
|
2018-04-01 18:55:54 +00:00
|
|
|
DREF VLR subscr MSISDN:46071 usage decreases to: 2
|
|
|
|
<-- GSUP rx OSMO_GSUP_MSGT_UPDATE_LOCATION_RESULT: vlr_gsupc_read_cb() returns 0
|
|
|
|
bssap_clear_sent == 1
|
|
|
|
- LU was successful, and the conn has already been closed
|
|
|
|
lu_result_sent == 1
|
|
|
|
- BSS sends BSSMAP Clear Complete
|
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
|
|
|
DREF MSISDN:46071: MSC conn use - release == 0 (0x0)
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_RELEASING}: Received Event SUBSCR_CONN_E_UNUSED
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_RELEASING}: state_chg to SUBSCR_CONN_S_RELEASED
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_RELEASED}: Terminating (cause = OSMO_FSM_TERM_REGULAR)
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_DONE}: Terminating (cause = OSMO_FSM_TERM_PARENT)
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_DONE}: Removing from parent Subscr_Conn(LU:901700000004620)
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_DONE}: fsm_lu_cleanup called with cause OSMO_FSM_TERM_PARENT
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_DONE}: Freeing instance
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_DONE}: Deallocated
|
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
|
|
|
DRLL MSISDN:46071: Freeing subscriber connection
|
2018-04-01 18:55:54 +00:00
|
|
|
DREF VLR subscr MSISDN:46071 usage decreases to: 1
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_RELEASED}: Freeing instance
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_RELEASED}: Deallocated
|
2017-01-25 14:04:16 +00:00
|
|
|
llist_count(&net->subscr_conns) == 0
|
|
|
|
---
|
|
|
|
- verify that the MS can send another LU request
|
|
|
|
- Location Update request causes a GSUP LU request to HLR
|
2017-03-10 01:15:20 +00:00
|
|
|
MSC <--RAN_GERAN_A-- MS: GSM48_MT_MM_LOC_UPD_REQUEST
|
2017-01-25 14:04:16 +00:00
|
|
|
new 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
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_NEW}: Allocated
|
2017-11-22 13:33:12 +00:00
|
|
|
DREF unknown: MSC conn use + compl_l3 == 1 (0x1)
|
2017-03-10 01:15:20 +00:00
|
|
|
DRLL Dispatching 04.08 message GSM48_MT_MM_LOC_UPD_REQUEST (0x5:0x8)
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_NEW}: Updated ID
|
2017-01-25 14:04:16 +00:00
|
|
|
DMM LOCATION UPDATING REQUEST: MI(IMSI)=901700000004620 type=IMSI ATTACH
|
2017-11-22 01:55:52 +00:00
|
|
|
DMM LU/new-LAC: 1/23
|
2018-04-02 21:17:50 +00:00
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_IDLE}: Allocated
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_IDLE}: is child of Subscr_Conn(LU:901700000004620)
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_IDLE}: rev=GSM net=GERAN (no Auth)
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_IDLE}: Received Event VLR_ULA_E_UPDATE_LA
|
2017-01-25 14:04:16 +00:00
|
|
|
DREF VLR subscr MSISDN:46071 usage increases to: 2
|
|
|
|
DREF VLR subscr MSISDN:46071 usage increases to: 3
|
2017-07-18 13:39:27 +00:00
|
|
|
DREF VLR subscr MSISDN:46071 usage decreases to: 2
|
2018-04-02 21:17:50 +00:00
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_IDLE}: vlr_loc_upd_node1()
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_IDLE}: vlr_loc_upd_post_auth()
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_IDLE}: vlr_loc_upd_post_ciph()
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_IDLE}: vlr_loc_upd_node_4()
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_IDLE}: state_chg to VLR_ULA_S_WAIT_HLR_UPD
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_INIT}: Allocated
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_INIT}: is child of vlr_lu_fsm(LU:901700000004620)
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_INIT}: Received Event UPD_HLR_VLR_E_START
|
2017-01-25 14:04:16 +00:00
|
|
|
DVLR GSUP tx: 04010809710000004026f0
|
|
|
|
GSUP --> HLR: OSMO_GSUP_MSGT_UPDATE_LOCATION_REQUEST: 04010809710000004026f0
|
2018-04-02 21:17:50 +00:00
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_INIT}: state_chg to UPD_HLR_VLR_S_WAIT_FOR_DATA
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_NEW}: Received Event SUBSCR_CONN_E_COMPLETE_LAYER_3
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_NEW}: state_chg to SUBSCR_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
|
|
|
DREF MSISDN:46071: MSC conn use - compl_l3 == 0 (0x0)
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_AUTH_CIPH}: Received Event SUBSCR_CONN_E_UNUSED
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_AUTH_CIPH}: Awaiting results for Auth+Ciph, overruling event SUBSCR_CONN_E_UNUSED
|
2017-01-25 14:04:16 +00:00
|
|
|
lu_result_sent == 0
|
|
|
|
- HLR sends _INSERT_DATA_REQUEST, VLR responds with _INSERT_DATA_RESULT
|
|
|
|
<-- GSUP rx OSMO_GSUP_MSGT_INSERT_DATA_REQUEST: 10010809710000004026f00804036470f1
|
|
|
|
DVLR GSUP rx 17: 10010809710000004026f00804036470f1
|
|
|
|
DREF VLR subscr MSISDN:46071 usage increases to: 3
|
|
|
|
DVLR IMSI:901700000004620 has MSISDN:46071
|
2017-12-27 12:23:44 +00:00
|
|
|
DVLR SUBSCR(MSISDN:46071) VLR: update for IMSI=901700000004620 (MSISDN=46071, used=3)
|
2017-01-25 14:04:16 +00:00
|
|
|
DVLR GSUP tx: 12010809710000004026f0
|
|
|
|
GSUP --> HLR: OSMO_GSUP_MSGT_INSERT_DATA_RESULT: 12010809710000004026f0
|
|
|
|
DREF VLR subscr MSISDN:46071 usage decreases to: 2
|
|
|
|
<-- GSUP rx OSMO_GSUP_MSGT_INSERT_DATA_REQUEST: vlr_gsupc_read_cb() returns 0
|
|
|
|
lu_result_sent == 0
|
|
|
|
- having received subscriber data does not mean acceptance
|
|
|
|
msc_subscr_conn_is_accepted() == false
|
|
|
|
requests shall be thwarted
|
2017-03-10 01:15:20 +00:00
|
|
|
DRLL Dispatching 04.08 message GSM48_MT_CC_SETUP (0x3:0x5)
|
|
|
|
DRLL subscr MSISDN:46071: Message not permitted for initial conn: GSM48_MT_CC_SETUP
|
|
|
|
DRLL Dispatching 04.08 message unknown 0x33 (0x5:0x33)
|
|
|
|
DRLL subscr MSISDN:46071: Message not permitted for initial conn: unknown 0x33
|
|
|
|
DRLL Dispatching 04.08 message GSM48_MT_RR_SYSINFO_1 (0x6:0x19)
|
|
|
|
DRLL subscr MSISDN:46071: Message not permitted for initial conn: GSM48_MT_RR_SYSINFO_1
|
2018-01-24 21:53:00 +00:00
|
|
|
DRLL Dispatching 04.08 message SMS:0x01 (0x9:0x1)
|
|
|
|
DRLL subscr MSISDN:46071: Message not permitted for initial conn: SMS:0x01
|
2017-01-25 14:04:16 +00:00
|
|
|
lu_result_sent == 0
|
|
|
|
- HLR also sends GSUP _UPDATE_LOCATION_RESULT
|
|
|
|
<-- GSUP rx OSMO_GSUP_MSGT_UPDATE_LOCATION_RESULT: 06010809710000004026f0
|
|
|
|
DVLR GSUP rx 11: 06010809710000004026f0
|
|
|
|
DREF VLR subscr MSISDN:46071 usage increases to: 3
|
2018-04-02 21:17:50 +00:00
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_WAIT_HLR_UPD}: Received Event VLR_ULA_E_HLR_LU_RES
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_WAIT_FOR_DATA}: Received Event UPD_HLR_VLR_E_UPD_LOC_ACK
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_WAIT_FOR_DATA}: state_chg to UPD_HLR_VLR_S_DONE
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_DONE}: Terminating (cause = OSMO_FSM_TERM_REGULAR)
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_DONE}: Removing from parent vlr_lu_fsm(LU:901700000004620)
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_DONE}: Freeing instance
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:901700000004620){UPD_HLR_VLR_S_DONE}: Deallocated
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_WAIT_HLR_UPD}: Received Event VLR_ULA_E_UPD_HLR_COMPL
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_WAIT_HLR_UPD}: state_chg to VLR_ULA_S_WAIT_LU_COMPL
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_INIT}: Allocated
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_INIT}: is child of vlr_lu_fsm(LU:901700000004620)
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_INIT}: Received Event LU_COMPL_VLR_E_START
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_INIT}: state_chg to LU_COMPL_VLR_S_WAIT_SUB_PRES
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:901700000004620){SUB_PRES_VLR_S_INIT}: Allocated
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:901700000004620){SUB_PRES_VLR_S_INIT}: is child of lu_compl_vlr_fsm(LU:901700000004620)
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:901700000004620){SUB_PRES_VLR_S_INIT}: Received Event SUB_PRES_VLR_E_START
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:901700000004620){SUB_PRES_VLR_S_INIT}: state_chg to SUB_PRES_VLR_S_DONE
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:901700000004620){SUB_PRES_VLR_S_DONE}: Terminating (cause = OSMO_FSM_TERM_REGULAR)
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:901700000004620){SUB_PRES_VLR_S_DONE}: Removing from parent lu_compl_vlr_fsm(LU:901700000004620)
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:901700000004620){SUB_PRES_VLR_S_DONE}: Freeing instance
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:901700000004620){SUB_PRES_VLR_S_DONE}: Deallocated
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_WAIT_SUB_PRES}: Received Event LU_COMPL_VLR_E_SUB_PRES_COMPL
|
2017-01-25 14:04:16 +00:00
|
|
|
- sending LU Accept for MSISDN:46071
|
2018-04-02 21:17:50 +00:00
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_WAIT_SUB_PRES}: state_chg to LU_COMPL_VLR_S_DONE
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_WAIT_LU_COMPL}: Received Event VLR_ULA_E_LU_COMPL_SUCCESS
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_DONE}: Terminating (cause = OSMO_FSM_TERM_PARENT)
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_DONE}: Removing from parent vlr_lu_fsm(LU:901700000004620)
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_DONE}: Freeing instance
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:901700000004620){LU_COMPL_VLR_S_DONE}: Deallocated
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_WAIT_LU_COMPL}: state_chg to VLR_ULA_S_DONE
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_AUTH_CIPH}: Received Event SUBSCR_CONN_E_ACCEPTED
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_AUTH_CIPH}: state_chg to SUBSCR_CONN_S_ACCEPTED
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_ACCEPTED}: Received Event SUBSCR_CONN_E_UNUSED
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_ACCEPTED}: state_chg to SUBSCR_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
|
|
|
DREF MSISDN:46071: MSC conn use + release == 1 (0x100)
|
|
|
|
DREF VLR subscr MSISDN:46071 usage increases to: 4
|
|
|
|
DREF VLR subscr MSISDN:46071 usage decreases to: 3
|
|
|
|
- BSSAP Clear --RAN_GERAN_A--> MS
|
2018-04-01 18:55:54 +00:00
|
|
|
DREF VLR subscr MSISDN:46071 usage decreases to: 2
|
|
|
|
<-- GSUP rx OSMO_GSUP_MSGT_UPDATE_LOCATION_RESULT: vlr_gsupc_read_cb() returns 0
|
|
|
|
bssap_clear_sent == 1
|
|
|
|
- LU was successful, and the conn has already been closed
|
|
|
|
lu_result_sent == 1
|
|
|
|
- BSS sends BSSMAP Clear Complete
|
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
|
|
|
DREF MSISDN:46071: MSC conn use - release == 0 (0x0)
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_RELEASING}: Received Event SUBSCR_CONN_E_UNUSED
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_RELEASING}: state_chg to SUBSCR_CONN_S_RELEASED
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_RELEASED}: Terminating (cause = OSMO_FSM_TERM_REGULAR)
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_DONE}: Terminating (cause = OSMO_FSM_TERM_PARENT)
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_DONE}: Removing from parent Subscr_Conn(LU:901700000004620)
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_DONE}: fsm_lu_cleanup called with cause OSMO_FSM_TERM_PARENT
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_DONE}: Freeing instance
|
|
|
|
DVLR vlr_lu_fsm(LU:901700000004620){VLR_ULA_S_DONE}: Deallocated
|
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
|
|
|
DRLL MSISDN:46071: Freeing subscriber connection
|
2018-04-01 18:55:54 +00:00
|
|
|
DREF VLR subscr MSISDN:46071 usage decreases to: 1
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_RELEASED}: Freeing instance
|
|
|
|
DMM Subscr_Conn(LU:901700000004620){SUBSCR_CONN_S_RELEASED}: Deallocated
|
2017-01-25 14:04:16 +00:00
|
|
|
llist_count(&net->subscr_conns) == 0
|
|
|
|
---
|
|
|
|
- subscriber detaches
|
2017-03-10 01:15:20 +00:00
|
|
|
MSC <--RAN_GERAN_A-- MS: GSM48_MT_MM_IMSI_DETACH_IND
|
2017-01-25 14:04:16 +00:00
|
|
|
new 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
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_NEW}: Allocated
|
2017-11-22 13:33:12 +00:00
|
|
|
DREF unknown: MSC conn use + compl_l3 == 1 (0x1)
|
2017-03-10 01:15:20 +00:00
|
|
|
DRLL Dispatching 04.08 message GSM48_MT_MM_IMSI_DETACH_IND (0x5:0x1)
|
2017-01-25 14:04:16 +00:00
|
|
|
DMM IMSI DETACH INDICATION: MI(IMSI)=901700000004620
|
|
|
|
DREF VLR subscr MSISDN:46071 usage increases to: 2
|
|
|
|
DMM IMSI DETACH for MSISDN:46071
|
2018-04-06 00:57:51 +00:00
|
|
|
DREF VLR subscr MSISDN:46071 usage increases to: 3
|
|
|
|
DREF VLR subscr MSISDN:46071 usage decreases to: 2
|
2017-01-25 14:04:16 +00:00
|
|
|
DREF VLR subscr MSISDN:46071 usage decreases to: 1
|
|
|
|
DREF VLR subscr MSISDN:46071 usage decreases to: 0
|
|
|
|
DREF freeing VLR subscr MSISDN:46071
|
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
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_NEW}: Received Event SUBSCR_CONN_E_CN_CLOSE
|
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_NEW}: state_chg to SUBSCR_CONN_S_RELEASING
|
|
|
|
DREF unknown: MSC conn use + release == 2 (0x101)
|
2017-04-09 10:32:51 +00:00
|
|
|
- BSSAP Clear --RAN_GERAN_A--> MS
|
2018-04-01 18:55:54 +00:00
|
|
|
DREF unknown: MSC conn use - compl_l3 == 1 (0x100)
|
|
|
|
bssap_clear_sent == 1
|
|
|
|
- BSS sends BSSMAP Clear Complete
|
|
|
|
DREF unknown: MSC conn use - release == 0 (0x0)
|
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
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_RELEASING}: Received Event SUBSCR_CONN_E_UNUSED
|
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_RELEASING}: state_chg to SUBSCR_CONN_S_RELEASED
|
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_RELEASED}: Terminating (cause = OSMO_FSM_TERM_REGULAR)
|
2017-01-25 14:04:16 +00:00
|
|
|
DRLL Freeing subscriber connection with NULL subscriber
|
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
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_RELEASED}: Freeing instance
|
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_RELEASED}: Deallocated
|
2017-01-25 14:04:16 +00:00
|
|
|
llist_count(&net->subscr_conns) == 0
|
2018-03-01 23:40:58 +00:00
|
|
|
===== test_two_lu: SUCCESS
|
2017-01-25 14:04:16 +00:00
|
|
|
|
|
|
|
full talloc report on 'msgb' (total 0 bytes in 1 blocks)
|
2018-03-30 21:04:04 +00:00
|
|
|
talloc_total_blocks(tall_bsc_ctx) == 12
|
2017-01-25 14:04:16 +00:00
|
|
|
|
2018-03-01 23:40:58 +00:00
|
|
|
===== test_lu_unknown_tmsi
|
2017-01-25 14:04:16 +00:00
|
|
|
- Location Update request with unknown TMSI sends ID Request for IMSI
|
2017-03-10 01:15:20 +00:00
|
|
|
MSC <--RAN_GERAN_A-- MS: GSM48_MT_MM_LOC_UPD_REQUEST
|
2017-01-25 14:04:16 +00:00
|
|
|
new 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
|
|
|
DMM Subscr_Conn{SUBSCR_CONN_S_NEW}: Allocated
|
2017-11-22 13:33:12 +00:00
|
|
|
DREF unknown: MSC conn use + compl_l3 == 1 (0x1)
|
2017-03-10 01:15:20 +00:00
|
|
|
DRLL Dispatching 04.08 message GSM48_MT_MM_LOC_UPD_REQUEST (0x5:0x8)
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(LU:591536962){SUBSCR_CONN_S_NEW}: Updated ID
|
2017-01-25 14:04:16 +00:00
|
|
|
DMM LOCATION UPDATING REQUEST: MI(TMSI)=591536962 type=IMSI ATTACH
|
2017-11-22 01:55:52 +00:00
|
|
|
DMM LU/new-LAC: 1/23
|
2018-04-02 21:17:50 +00:00
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_IDLE}: Allocated
|
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_IDLE}: is child of Subscr_Conn(LU:591536962)
|
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_IDLE}: rev=GSM net=GERAN (no Auth)
|
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_IDLE}: Received Event VLR_ULA_E_UPDATE_LA
|
2017-01-25 14:04:16 +00:00
|
|
|
DREF VLR subscr unknown usage increases to: 1
|
|
|
|
DVLR New subscr, TMSI: 0x23422342
|
|
|
|
DREF VLR subscr TMSI:0x23422342 usage increases to: 2
|
2017-07-18 13:39:27 +00:00
|
|
|
DREF VLR subscr TMSI:0x23422342 usage decreases to: 1
|
2018-04-02 21:17:50 +00:00
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_IDLE}: vlr_loc_upd_want_imsi()
|
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_IDLE}: state_chg to VLR_ULA_S_WAIT_IMSI
|
2016-05-20 19:59:55 +00:00
|
|
|
DMSC msc_tx 3 bytes to TMSI:0x23422342 via RAN_GERAN_A
|
2017-03-10 01:15:20 +00:00
|
|
|
- DTAP --RAN_GERAN_A--> MS: GSM48_MT_MM_ID_REQ: 051801
|
2017-01-25 14:04:16 +00:00
|
|
|
- DTAP matches expected message
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(LU:591536962){SUBSCR_CONN_S_NEW}: Received Event SUBSCR_CONN_E_COMPLETE_LAYER_3
|
|
|
|
DMM Subscr_Conn(LU:591536962){SUBSCR_CONN_S_NEW}: state_chg to SUBSCR_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
|
|
|
DREF TMSI:0x23422342: MSC conn use - compl_l3 == 0 (0x0)
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(LU:591536962){SUBSCR_CONN_S_AUTH_CIPH}: Received Event SUBSCR_CONN_E_UNUSED
|
|
|
|
DMM Subscr_Conn(LU:591536962){SUBSCR_CONN_S_AUTH_CIPH}: Awaiting results for Auth+Ciph, overruling event SUBSCR_CONN_E_UNUSED
|
2017-01-25 14:04:16 +00:00
|
|
|
lu_result_sent == 0
|
|
|
|
msc_subscr_conn_is_accepted() == false
|
|
|
|
requests shall be thwarted
|
2017-03-10 01:15:20 +00:00
|
|
|
DRLL Dispatching 04.08 message GSM48_MT_CC_SETUP (0x3:0x5)
|
|
|
|
DRLL subscr TMSI:0x23422342: Message not permitted for initial conn: GSM48_MT_CC_SETUP
|
|
|
|
DRLL Dispatching 04.08 message unknown 0x33 (0x5:0x33)
|
|
|
|
DRLL subscr TMSI:0x23422342: Message not permitted for initial conn: unknown 0x33
|
|
|
|
DRLL Dispatching 04.08 message GSM48_MT_RR_SYSINFO_1 (0x6:0x19)
|
|
|
|
DRLL subscr TMSI:0x23422342: Message not permitted for initial conn: GSM48_MT_RR_SYSINFO_1
|
2018-01-24 21:53:00 +00:00
|
|
|
DRLL Dispatching 04.08 message SMS:0x01 (0x9:0x1)
|
|
|
|
DRLL subscr TMSI:0x23422342: Message not permitted for initial conn: SMS:0x01
|
2017-01-25 14:04:16 +00:00
|
|
|
- MS tells us the IMSI, causes a GSUP LU request to HLR
|
2017-03-10 01:15:20 +00:00
|
|
|
MSC <--RAN_GERAN_A-- MS: GSM48_MT_MM_ID_RESP
|
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
|
|
|
DREF TMSI:0x23422342: MSC conn use + dtap == 1 (0x2)
|
2017-03-10 01:15:20 +00:00
|
|
|
DRLL Dispatching 04.08 message GSM48_MT_MM_ID_RESP (0x5:0x19)
|
2017-01-25 14:04:16 +00:00
|
|
|
DMM IDENTITY RESPONSE: MI(IMSI)=901700000004620
|
|
|
|
DVLR set IMSI on subscriber; IMSI=901700000004620 id=901700000004620
|
2018-04-02 21:17:50 +00:00
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_WAIT_IMSI}: Received Event VLR_ULA_E_ID_IMSI
|
2017-01-25 14:04:16 +00:00
|
|
|
DVLR set IMSI on subscriber; IMSI=901700000004620 id=901700000004620
|
2018-04-02 21:17:50 +00:00
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_WAIT_IMSI}: vlr_loc_upd_node1()
|
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_WAIT_IMSI}: vlr_loc_upd_post_auth()
|
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_WAIT_IMSI}: vlr_loc_upd_post_ciph()
|
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_WAIT_IMSI}: vlr_loc_upd_node_4()
|
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_WAIT_IMSI}: state_chg to VLR_ULA_S_WAIT_HLR_UPD
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:591536962){UPD_HLR_VLR_S_INIT}: Allocated
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:591536962){UPD_HLR_VLR_S_INIT}: is child of vlr_lu_fsm(LU:591536962)
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:591536962){UPD_HLR_VLR_S_INIT}: Received Event UPD_HLR_VLR_E_START
|
2017-01-25 14:04:16 +00:00
|
|
|
DVLR GSUP tx: 04010809710000004026f0
|
|
|
|
GSUP --> HLR: OSMO_GSUP_MSGT_UPDATE_LOCATION_REQUEST: 04010809710000004026f0
|
2018-04-02 21:17:50 +00:00
|
|
|
DVLR upd_hlr_vlr_fsm(LU:591536962){UPD_HLR_VLR_S_INIT}: state_chg to UPD_HLR_VLR_S_WAIT_FOR_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
|
|
|
DREF IMSI:901700000004620: MSC conn use - dtap == 0 (0x0)
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(LU:591536962){SUBSCR_CONN_S_AUTH_CIPH}: Received Event SUBSCR_CONN_E_UNUSED
|
|
|
|
DMM Subscr_Conn(LU:591536962){SUBSCR_CONN_S_AUTH_CIPH}: Awaiting results for Auth+Ciph, overruling event SUBSCR_CONN_E_UNUSED
|
2017-01-25 14:04:16 +00:00
|
|
|
lu_result_sent == 0
|
|
|
|
- HLR sends _INSERT_DATA_REQUEST, VLR responds with _INSERT_DATA_RESULT
|
|
|
|
<-- GSUP rx OSMO_GSUP_MSGT_INSERT_DATA_REQUEST: 10010809710000004026f00804036470f1
|
|
|
|
DVLR GSUP rx 17: 10010809710000004026f00804036470f1
|
|
|
|
DREF VLR subscr IMSI:901700000004620 usage increases to: 2
|
|
|
|
DVLR IMSI:901700000004620 has MSISDN:46071
|
2017-12-27 12:23:44 +00:00
|
|
|
DVLR SUBSCR(MSISDN:46071) VLR: update for IMSI=901700000004620 (MSISDN=46071, used=2)
|
2017-01-25 14:04:16 +00:00
|
|
|
DVLR GSUP tx: 12010809710000004026f0
|
|
|
|
GSUP --> HLR: OSMO_GSUP_MSGT_INSERT_DATA_RESULT: 12010809710000004026f0
|
|
|
|
DREF VLR subscr MSISDN:46071 usage decreases to: 1
|
|
|
|
<-- GSUP rx OSMO_GSUP_MSGT_INSERT_DATA_REQUEST: vlr_gsupc_read_cb() returns 0
|
|
|
|
lu_result_sent == 0
|
|
|
|
- having received subscriber data does not mean acceptance
|
|
|
|
msc_subscr_conn_is_accepted() == false
|
|
|
|
requests shall be thwarted
|
2017-03-10 01:15:20 +00:00
|
|
|
DRLL Dispatching 04.08 message GSM48_MT_CC_SETUP (0x3:0x5)
|
|
|
|
DRLL subscr MSISDN:46071: Message not permitted for initial conn: GSM48_MT_CC_SETUP
|
|
|
|
DRLL Dispatching 04.08 message unknown 0x33 (0x5:0x33)
|
|
|
|
DRLL subscr MSISDN:46071: Message not permitted for initial conn: unknown 0x33
|
|
|
|
DRLL Dispatching 04.08 message GSM48_MT_RR_SYSINFO_1 (0x6:0x19)
|
|
|
|
DRLL subscr MSISDN:46071: Message not permitted for initial conn: GSM48_MT_RR_SYSINFO_1
|
2018-01-24 21:53:00 +00:00
|
|
|
DRLL Dispatching 04.08 message SMS:0x01 (0x9:0x1)
|
|
|
|
DRLL subscr MSISDN:46071: Message not permitted for initial conn: SMS:0x01
|
2017-01-25 14:04:16 +00:00
|
|
|
lu_result_sent == 0
|
|
|
|
- HLR also sends GSUP _UPDATE_LOCATION_RESULT
|
|
|
|
<-- GSUP rx OSMO_GSUP_MSGT_UPDATE_LOCATION_RESULT: 06010809710000004026f0
|
|
|
|
DVLR GSUP rx 11: 06010809710000004026f0
|
|
|
|
DREF VLR subscr MSISDN:46071 usage increases to: 2
|
2018-04-02 21:17:50 +00:00
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_WAIT_HLR_UPD}: Received Event VLR_ULA_E_HLR_LU_RES
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:591536962){UPD_HLR_VLR_S_WAIT_FOR_DATA}: Received Event UPD_HLR_VLR_E_UPD_LOC_ACK
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:591536962){UPD_HLR_VLR_S_WAIT_FOR_DATA}: state_chg to UPD_HLR_VLR_S_DONE
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:591536962){UPD_HLR_VLR_S_DONE}: Terminating (cause = OSMO_FSM_TERM_REGULAR)
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:591536962){UPD_HLR_VLR_S_DONE}: Removing from parent vlr_lu_fsm(LU:591536962)
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:591536962){UPD_HLR_VLR_S_DONE}: Freeing instance
|
|
|
|
DVLR upd_hlr_vlr_fsm(LU:591536962){UPD_HLR_VLR_S_DONE}: Deallocated
|
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_WAIT_HLR_UPD}: Received Event VLR_ULA_E_UPD_HLR_COMPL
|
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_WAIT_HLR_UPD}: state_chg to VLR_ULA_S_WAIT_LU_COMPL
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:591536962){LU_COMPL_VLR_S_INIT}: Allocated
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:591536962){LU_COMPL_VLR_S_INIT}: is child of vlr_lu_fsm(LU:591536962)
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:591536962){LU_COMPL_VLR_S_INIT}: Received Event LU_COMPL_VLR_E_START
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:591536962){LU_COMPL_VLR_S_INIT}: state_chg to LU_COMPL_VLR_S_WAIT_SUB_PRES
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:591536962){SUB_PRES_VLR_S_INIT}: Allocated
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:591536962){SUB_PRES_VLR_S_INIT}: is child of lu_compl_vlr_fsm(LU:591536962)
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:591536962){SUB_PRES_VLR_S_INIT}: Received Event SUB_PRES_VLR_E_START
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:591536962){SUB_PRES_VLR_S_INIT}: state_chg to SUB_PRES_VLR_S_DONE
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:591536962){SUB_PRES_VLR_S_DONE}: Terminating (cause = OSMO_FSM_TERM_REGULAR)
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:591536962){SUB_PRES_VLR_S_DONE}: Removing from parent lu_compl_vlr_fsm(LU:591536962)
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:591536962){SUB_PRES_VLR_S_DONE}: Freeing instance
|
|
|
|
DVLR sub_pres_vlr_fsm(LU:591536962){SUB_PRES_VLR_S_DONE}: Deallocated
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:591536962){LU_COMPL_VLR_S_WAIT_SUB_PRES}: Received Event LU_COMPL_VLR_E_SUB_PRES_COMPL
|
2017-01-25 14:04:16 +00:00
|
|
|
- sending LU Accept for MSISDN:46071
|
|
|
|
DREF VLR subscr MSISDN:46071 usage increases to: 3
|
2018-04-02 21:17:50 +00:00
|
|
|
DVLR lu_compl_vlr_fsm(LU:591536962){LU_COMPL_VLR_S_WAIT_SUB_PRES}: state_chg to LU_COMPL_VLR_S_DONE
|
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_WAIT_LU_COMPL}: Received Event VLR_ULA_E_LU_COMPL_SUCCESS
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:591536962){LU_COMPL_VLR_S_DONE}: Terminating (cause = OSMO_FSM_TERM_PARENT)
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:591536962){LU_COMPL_VLR_S_DONE}: Removing from parent vlr_lu_fsm(LU:591536962)
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:591536962){LU_COMPL_VLR_S_DONE}: Freeing instance
|
|
|
|
DVLR lu_compl_vlr_fsm(LU:591536962){LU_COMPL_VLR_S_DONE}: Deallocated
|
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_WAIT_LU_COMPL}: state_chg to VLR_ULA_S_DONE
|
|
|
|
DMM Subscr_Conn(LU:591536962){SUBSCR_CONN_S_AUTH_CIPH}: Received Event SUBSCR_CONN_E_ACCEPTED
|
|
|
|
DMM Subscr_Conn(LU:591536962){SUBSCR_CONN_S_AUTH_CIPH}: state_chg to SUBSCR_CONN_S_ACCEPTED
|
|
|
|
DMM Subscr_Conn(LU:591536962){SUBSCR_CONN_S_ACCEPTED}: Received Event SUBSCR_CONN_E_UNUSED
|
|
|
|
DMM Subscr_Conn(LU:591536962){SUBSCR_CONN_S_ACCEPTED}: state_chg to SUBSCR_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
|
|
|
DREF MSISDN:46071: MSC conn use + release == 1 (0x100)
|
|
|
|
DREF VLR subscr MSISDN:46071 usage increases to: 4
|
|
|
|
DREF VLR subscr MSISDN:46071 usage decreases to: 3
|
|
|
|
- BSSAP Clear --RAN_GERAN_A--> MS
|
2018-04-01 18:55:54 +00:00
|
|
|
DREF VLR subscr MSISDN:46071 usage decreases to: 2
|
|
|
|
<-- GSUP rx OSMO_GSUP_MSGT_UPDATE_LOCATION_RESULT: vlr_gsupc_read_cb() returns 0
|
|
|
|
bssap_clear_sent == 1
|
|
|
|
- LU was successful, and the conn has already been closed
|
|
|
|
lu_result_sent == 1
|
|
|
|
- BSS sends BSSMAP Clear Complete
|
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
|
|
|
DREF MSISDN:46071: MSC conn use - release == 0 (0x0)
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(LU:591536962){SUBSCR_CONN_S_RELEASING}: Received Event SUBSCR_CONN_E_UNUSED
|
|
|
|
DMM Subscr_Conn(LU:591536962){SUBSCR_CONN_S_RELEASING}: state_chg to SUBSCR_CONN_S_RELEASED
|
|
|
|
DMM Subscr_Conn(LU:591536962){SUBSCR_CONN_S_RELEASED}: Terminating (cause = OSMO_FSM_TERM_REGULAR)
|
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_DONE}: Terminating (cause = OSMO_FSM_TERM_PARENT)
|
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_DONE}: Removing from parent Subscr_Conn(LU:591536962)
|
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_DONE}: fsm_lu_cleanup called with cause OSMO_FSM_TERM_PARENT
|
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_DONE}: Freeing instance
|
|
|
|
DVLR vlr_lu_fsm(LU:591536962){VLR_ULA_S_DONE}: Deallocated
|
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
|
|
|
DRLL MSISDN:46071: Freeing subscriber connection
|
2018-04-01 18:55:54 +00:00
|
|
|
DREF VLR subscr MSISDN:46071 usage decreases to: 1
|
2018-04-02 21:17:50 +00:00
|
|
|
DMM Subscr_Conn(LU:591536962){SUBSCR_CONN_S_RELEASED}: Freeing instance
|
|
|
|
DMM Subscr_Conn(LU:591536962){SUBSCR_CONN_S_RELEASED}: Deallocated
|
2017-01-25 14:04:16 +00:00
|
|
|
llist_count(&net->subscr_conns) == 0
|
|
|
|
DREF freeing VLR subscr MSISDN:46071
|
2018-03-01 23:40:58 +00:00
|
|
|
===== test_lu_unknown_tmsi: SUCCESS
|
2017-01-25 14:04:16 +00:00
|
|
|
|
|
|
|
full talloc report on 'msgb' (total 0 bytes in 1 blocks)
|
2018-03-30 21:04:04 +00:00
|
|
|
talloc_total_blocks(tall_bsc_ctx) == 12
|
2017-01-25 14:04:16 +00:00
|
|
|
|
|
|
|
full talloc report on 'msgb' (total 0 bytes in 1 blocks)
|
2018-03-30 21:04:04 +00:00
|
|
|
talloc_total_blocks(tall_bsc_ctx) == 12
|
2017-01-25 14:04:16 +00:00
|
|
|
|