Project

General

Profile

Actions

Bug #3253

closed

osmo-bts-trx doesn't enable A5/3

Added by laforge almost 6 years ago. Updated almost 6 years ago.

Status:
Resolved
Priority:
High
Assignee:
Category:
osmo-bts-trx
Target version:
-
Start date:
05/09/2018
Due date:
% Done:

100%

Spec Reference:

Description

I've implemented a set of encryption related test cases in BTS_Tests.ttcn, see the laforge/bts-encryption branch at http://git.osmocom.org/osmo-ttcn3-hacks/commit/?h=laforge/bts-encryptionx%x%

TC_chan_act_a51 and TC_chan_act_a52 pass without trouble, while TC_chan_act_a53 fails: The RSL/RLL message sent on Abis to the BTS never is received on L1CTL.

trxcon seems to correctly detect that A5/3 is selected, but then appears to be unable to make sense of the bursts it receives:

<0001> l1ctl.c:731 L1CTL_CRYPTO_REQ (algo=A5/3, key_len=8)
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=185 (83/102) for SDCCH/4(2)
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=195 (93/102) for SACCH/4(2)
<0005> sched_clck.c:140 Clock indication: fn=204
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=236 (32/102) for SDCCH/4(2)
<0005> sched_clck.c:140 Clock indication: fn=255
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=287 (83/102) for SDCCH/4(2)
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=297 (93/102) for SACCH/4(2)
<0005> sched_clck.c:140 Clock indication: fn=306
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=338 (32/102) for SDCCH/4(2)
<0005> sched_clck.c:140 Clock indication: fn=357
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=389 (83/102) for SDCCH/4(2)
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=399 (93/102) for SACCH/4(2)
<0005> sched_clck.c:140 Clock indication: fn=408
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=440 (32/102) for SDCCH/4(2)
<0005> sched_clck.c:140 Clock indication: fn=459
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=491 (83/102) for SDCCH/4(2)
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=501 (93/102) for SACCH/4(2)
<0005> sched_clck.c:140 Clock indication: fn=510
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=542 (32/102) for SDCCH/4(2)
<0005> sched_clck.c:140 Clock indication: fn=561
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=593 (83/102) for SDCCH/4(2)
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=603 (93/102) for SACCH/4(2)
<0005> sched_clck.c:140 Clock indication: fn=612
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=644 (32/102) for SDCCH/4(2)
<0005> sched_clck.c:140 Clock indication: fn=663
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=695 (83/102) for SDCCH/4(2)
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=705 (93/102) for SACCH/4(2)
<0005> sched_clck.c:140 Clock indication: fn=714
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=746 (32/102) for SDCCH/4(2)
<0005> sched_clck.c:140 Clock indication: fn=765
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=797 (83/102) for SDCCH/4(2)
<0006> sched_lchan_xcch.c:106 Received bad data frame at fn=807 (93/102) for SACCH/4(2)
<0005> sched_clck.c:140 Clock indication: fn=816

Files

20180509-osmobts-a53.pcapng 20180509-osmobts-a53.pcapng 13.1 KB laforge, 05/09/2018 05:39 PM

Related issues

Related to OsmoBTS - Bug #3254: OsmoBTS doesn't reject RSL CHAN ACT for unsupported ciphersResolvedlaforge05/09/2018

Actions
Actions #1

Updated by laforge almost 6 years ago

now it looks actually more like a osmo-bts problem:

when running the A5/1 testcase, I get:

DRSL <0000> rsl.c:2613 (bts=0,trx=0,ts=0,ss=2) Rx RSL CHAN_ACTIV
DRSL <0000> rsl.c:1043  chan_nr=0x30 type=0x00 mode=0x00
DL1C <0006> l1sap.c:1406 activating channel chan_nr=0x30 trx=0
DL1C <0006> scheduler.c:615 Activating SDCCH/4(2) on trx=0 ts=0
DL1C <0006> scheduler.c:615 Activating SACCH/4(2) on trx=0 ts=0
DL1C <0006> scheduler.c:664 Set mode 3, 0, handover 0 on SDCCH/4(2) of trx=0 ts=0
DTRX <000b> trx_if.c:240 Enqueuing TRX control command 'CMD NOHANDOVER 0 2'
DL1C <0006> scheduler.c:729 Set a5/1 uplink for SDCCH/4(2) on trx=0 ts=0
DL1C <0006> scheduler.c:729 Set a5/1 uplink for SACCH/4(2) on trx=0 ts=0
DL1C <0006> scheduler.c:729 Set a5/1 downlink for SDCCH/4(2) on trx=0 ts=0
DL1C <0006> scheduler.c:729 Set a5/1 downlink for SACCH/4(2) on trx=0 ts=0
DL1C <0006> l1sap.c:578 activate confirm chan_nr=0x30 trx=0
DRSL <0000> rsl.c:662 (bts=0,trx=0,ts=0,ss=2) Tx CHAN ACT ACK

but when running the A5/3 testcase, I see:

DRSL <0000> rsl.c:2613 (bts=0,trx=0,ts=0,ss=2) Rx RSL CHAN_ACTIV
DRSL <0000> rsl.c:1043  chan_nr=0x30 type=0x00 mode=0x00
DL1C <0006> l1sap.c:1406 activating channel chan_nr=0x30 trx=0
DL1C <0006> scheduler.c:615 Activating SDCCH/4(2) on trx=0 ts=0
DL1C <0006> scheduler.c:615 Activating SACCH/4(2) on trx=0 ts=0
DL1C <0006> scheduler.c:664 Set mode 3, 0, handover 0 on SDCCH/4(2) of trx=0 ts=0
DTRX <000b> trx_if.c:240 Enqueuing TRX control command 'CMD NOHANDOVER 0 2'
DL1C <0006> scheduler.c:729 Set a5/0 uplink for SDCCH/4(2) on trx=0 ts=0
DL1C <0006> scheduler.c:729 Set a5/0 uplink for SACCH/4(2) on trx=0 ts=0
DL1C <0006> scheduler.c:729 Set a5/0 downlink for SDCCH/4(2) on trx=0 ts=0
DL1C <0006> scheduler.c:729 Set a5/0 downlink for SACCH/4(2) on trx=0 ts=0
DL1C <0006> l1sap.c:578 activate confirm chan_nr=0x30 trx=0
DRSL <0000> rsl.c:662 (bts=0,trx=0,ts=0,ss=2) Tx CHAN ACT ACK

and this despite RSL CHAN ACT clearly stating A5/3 as algorithm (see attached pcap). odd.

Actions #2

Updated by laforge almost 6 years ago

  • Related to Bug #3254: OsmoBTS doesn't reject RSL CHAN ACT for unsupported ciphers added
Actions #3

Updated by laforge almost 6 years ago

  • Project changed from OsmocomBB to OsmoBTS
  • Subject changed from trxcon doesn't interoperate with OsmoBTS on A5/3 to osmo-bts-trx doesn't enable A5/3
  • Category changed from OsmocomBB Layer 0 (DSP) to osmo-bts-trx
  • Priority changed from Normal to High
  • % Done changed from 20 to 80

In fact, the most embarrassing part of this bug is that osmo-bts-trx/main.c simply doesn't enable A5/3. Not sure why that would be, given that the software implementation in libosmcore already in 2015.

Actions #4

Updated by laforge almost 6 years ago

Actions #5

Updated by laforge almost 6 years ago

  • Status changed from In Progress to Resolved
  • % Done changed from 80 to 100

patch merged.

Actions

Also available in: Atom PDF

Add picture from clipboard (Maximum size: 48.8 MB)