Change in osmo-bsc[master]: doc: add lchan-release.msc

[Harald Welte]

Harald Welte has submitted this change and it was merged. ( https://gerrit.osmocom.org/9487 )

Change subject: doc: add lchan-release.msc
......................................................................

doc: add lchan-release.msc

Change-Id: Ibcf2326cbc5a5f3f249ae15564825c83b5a063c8
---
M doc/Makefile.am
A doc/lchan-release.msc
2 files changed, 150 insertions(+), 0 deletions(-)

Approvals:
  Harald Welte: Looks good to me, approved
  Jenkins Builder: Verified



diff --git a/doc/Makefile.am b/doc/Makefile.am
index cc2e277..3c8c2e0 100644
--- a/doc/Makefile.am
+++ b/doc/Makefile.am
@@ -5,6 +5,7 @@
 msc: \
 	$(builddir)/handover.png \
 	$(builddir)/assignment.png \
+	$(builddir)/lchan-release.png \
 	$(NULL)
 
 $(builddir)/%.png: $(srcdir)/%.msc
diff --git a/doc/lchan-release.msc b/doc/lchan-release.msc
new file mode 100644
index 0000000..f75b559
--- /dev/null
+++ b/doc/lchan-release.msc
@@ -0,0 +1,149 @@
+msc {
+	hscale=2;
+	ms [label="MS"], bts [label="BTS"], bsc[label="BSC"], bsc_gscon[label="BSC conn FSM"];
+
+	ms note bsc_gscon [label="various lchan release scenarios"];
+	
+
+	ms rbox bsc_gscon [label="IF BSC releases, from BSSMAP Clear Request"];
+	bsc note bsc [label="lchan_release() may be called with sacch_deact=true or false.
+		Currently, the only time lchan_release(sacch_deact=true) is invoked is upon BSSMAP Clear
+		Command, i.e. when the MSC instructs to stop using an active lchan.
+		Some error handling code paths however directly invoke
+		rsl_rf_chan_release(error=1, SACCH_DEACTIVATE)."];
+
+	---;
+	bsc_gscon note bsc_gscon [label="Rx: BSSMAP Clear Request from MSC"];
+	bsc <- bsc_gscon [label="gsm0808_clear()"];
+	bsc box bsc [label="lchan_release(sacch_deact=1)"];
+	bsc box bsc [label="lchan->state = LCHAN_S_REL_REQ"];
+	bsc box bsc [label="_lchan_handle_release(sacch_deact=1)"];
+	bsc box bsc [label="rsl_release_sapis_from(start=1)"];
+	bts <= bsc [label="RSL Release Request (Local End)..."];
+	bts <= bsc [label="...for each SAPI, except link_id=0"];
+	ms <= bsc [label="RR Channel Release"];
+	ms note bsc [label="There is no ACK for RR Channel Release"];
+	bsc box bsc [label="rsl_deact_sacch()"];
+	bts <= bsc [label="RSL Deactivate SACCH"];
+	bsc abox bsc [label="Start T3109 (net->T3109, t3109_expired())"];
+	...;
+	--- [label="If T3109 expires"];
+	bsc box bsc [label="t3109_expired()"];
+	bsc box bsc [label="rsl_rf_chan_release(error=1)"];
+	bts <= bsc [label="RSL Release Request (Local End)..."];
+	bts <= bsc [label="...for each SAPI, except link_id=0"];
+	bsc box bsc [label="lchan->state = LCHAN_S_REL_REQ"];
+	bts <= bsc [label="RSL RF Channel Release"];
+	---;
+	...;
+	bsc rbox bsc [label="continue in the 'Common' part"];
+	--- [label="END: 'BSSMAP Clear Request'"];
+	...;
+	...;
+
+	ms rbox bsc_gscon [label="IF BSC releases, from implicitly unused lchan"];
+	bsc note bsc [label="The BSC may release old unused lchans after Handover, or release lchans
+		after some error condition."];
+	bsc note bsc [label="BSC decides to release an unused lchan"];
+	bsc box bsc [label="lchan_release(sacch_deact=0)"];
+	bsc box bsc [label="lchan->state = LCHAN_S_REL_REQ"];
+	bsc box bsc [label="_lchan_handle_release(sacch_deact=0)"];
+	bts <= bsc [label="RSL Release Request (Local End)..."];
+	bts <= bsc [label="...for all SAPIs"];
+	...;
+	bts note bsc_gscon [linecolor="red",
+	  label="There seems to be no timer watching over RSL Release Request!"];
+	...;
+	bts => bsc [label="RSL Release Confirm..."];
+	bts => bsc [label="...for all SAPIs"];
+	bsc rbox bsc [label="continue in the 'Common' part"];
+	---;
+	...;
+	...;
+
+	ms rbox bsc_gscon [label="IF MS releases"];
+	ms => bts [label="DISC"];
+	bts => bsc [label="RLL Release Ind..."];
+	bts => bsc [label="...for each SAPI"];
+	bsc rbox bsc [label="continue in the 'Common' part"];
+	---;
+	...;
+	...;
+	ms rbox bsc_gscon [label="Common"];
+	--- [label="for each SAPI (?)"];
+	bts => bsc [label="RLL Release Confirm / RLL Release Ind"];
+	bsc box bsc [label="abis_rsl_rx_rll()"];
+	bsc box bsc [label="mark lchan->sapis[link_id] = LCHAN_SAPI_UNUSED"];
+	bsc box bsc [label="rll_indication()"];
+	bsc box bsc [label="for each bsc_rll_req matching this link_id:
+		disable timer, call cb(BSC_RLLR_IND_REL_IND)"];
+	bsc box bsc [label="rsl_handle_release()"];
+	--- [label="IF all SAPIs are unused"];
+	bsc box bsc [label="Stop T3109"];
+	bsc note bsc [label="T3109 was started if the MSC requested the release"];
+	bsc abox bsc [label="Start T3111 (net->T3111 value, t3111_expired())"];
+	--- [label="END: all SAPIs are unused"];
+	bsc -> bsc_gscon [label="GSCON_EV_RLL_REL_IND (only if RLL Release Ind)"];
+	--- [label="END: for each SAPI"];
+
+	...;
+	bsc box bsc [label="T3111 expires"];
+	bsc box bsc [label="rsl_rf_chan_release()"];
+	bsc box bsc [label="Stop T3109"];
+	bsc note bsc [label="[If lchan->state is LCHAN_S_REL_ERR, don't do anything]"];
+	bsc abox bsc [label="Start lchan->act_timer (4s, lchan_deact_tmr_cb())"];
+	bts <= bsc [label="RSL RF Channel Release"];
+
+	...;
+	--- [label="IF lchan->act_timer expires"];
+	bsc box bsc [label="lchan_deact_tmr_cb()"];
+	bsc box bsc [label="rsl_lchan_mark_broken(): lchan->state = LCHAN_S_BROKEN"];
+	bsc box bsc [label="lchan_free() (see below)"];
+	--- [label="END: 'lchan->act_timer expires'"];
+	...;
+
+	bts => bsc [label="RSL RF Channel Release ACK"];
+	bsc box bsc [label="rsl_rx_rf_chan_rel_ack()"];
+	bsc box bsc [label="Stop lchan->act_timer"];
+	bsc box bsc [label="Stop T3111"];
+
+	--- [label="IF lchan->state == LCHAN_S_BROKEN"];
+	bsc note bsc [label="If an ACK comes in late, for specific BTS models, we may choose to 'repair'
+		the lchan so that it is usable again, by calling do_lchan_free() directly (see below)."];
+	bsc box bsc [label="rsl_rx_rf_chan_rel_ack() exits here and none of below actions happen.
+		The lchan remains LCHAN_S_BROKEN indefinitely."];
+	--- [label="END: lchan->state == LCHAN_S_BROKEN"];
+	bsc box bsc [label="do_lchan_free()"];
+	--- [label="IF lchan->state == LCHAN_S_REL_ERR"];
+	bsc note bsc [label="If release failed, we take the lchan back into operation after due
+		timeout"];
+	bsc abox bsc [label="Start lchan->error_timer (T3111+2, error_timeout_cb())"];
+	bsc note bsc [label="do_lchan_free() continues, async:"];
+	...;
+	bsc box bsc [label="error_timeout_cb()"];
+	bsc box bsc [label="lchan->state = LCHAN_S_NONE"];
+	bsc box bsc [label="dyn TS: activate PDCH..."];
+	--- [label="ELSE"];
+	bsc box bsc [label="lchan->state = LCHAN_S_NONE"];
+	--- [label="END: lchan->state == LCHAN_S_REL_ERR"];
+	bsc box bsc [label="lchan_free()"];
+	--- [label="IF conn is still associated (and not dyn TS in switchover)"];
+	bsc -> bsc [label="S_LCHAN_UNEXPECTED_RELEASE"];
+	bsc box bsc [label="handle_release()"];
+	bsc box bsc [label="Stop T10"];
+	bsc note bsc [linecolor=orange,label="conn->T10 is actually dead code, it is never started.
+		Instead, the conn FSM starts ST_WAIT_ASS_COMPL with a T10 value."];
+	bsc -> bsc_gscon [label="GSCON_EV_RR_ASS_FAIL"];
+	bsc -> bsc_gscon [label="GSCON_EV_TX_SCCP: BSSMAP Clear Request"];
+	bsc box bsc [label="bsc_clear_handover()"];
+	bsc box bsc [label="Stop T3103"];
+	bsc box bsc [label="free handover struct"];
+	bsc box bsc [label="lchan->conn = NULL"];
+	--- [label="END: 'conn is still associated'"];
+	bsc box bsc [label="Stop T3101"];
+	bsc note bsc [label="T3101 is started when sending an RR Immediate Assignment"];
+	bsc -> bsc [label="S_CHALLOC_FREED"];
+	bsc -> bsc [label="rll_lchan_signal()"];
+	bsc box bsc [label="for each bsc_rll_req matching this lchan:
+		disable timer, call cb(BSC_RLLR_IND_REL_IND)"];
+}

-- 
To view, visit https://gerrit.osmocom.org/9487
To unsubscribe, or for help writing mail filters, visit https://gerrit.osmocom.org/settings

Gerrit-Project: osmo-bsc
Gerrit-Branch: master
Gerrit-MessageType: merged
Gerrit-Change-Id: Ibcf2326cbc5a5f3f249ae15564825c83b5a063c8
Gerrit-Change-Number: 9487
Gerrit-PatchSet: 7
Gerrit-Owner: Neels Hofmeyr <nhofmeyr at sysmocom.de>
Gerrit-Reviewer: Harald Welte <laforge at gnumonks.org>
Gerrit-Reviewer: Jenkins Builder
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osmocom.org/pipermail/gerrit-log/attachments/20180608/d22e1c7d/attachment.htm>