This is merely a historical archive of years 2008-2021, before the migration to mailman3.
A maintained and still updated list archive can be found at https://lists.osmocom.org/hyperkitty/list/OpenBSC@lists.osmocom.org/.
Dennis Wehrle openbsc at wehrle.itHi Today i had a closer look at the abis pcap file (which i sent on Friday). For me one thing is "strange" but i don't know if this is a normal behavior. Is it OK that sometimes a packet contains 2 sub-packets? For instance: 01. IPv4 02. TCP --------------- 03. IPA 04. RSL 05. CP-DATA 06. RP-DATA 07. SMS-DELIVER --------------- 08. IPA 09. RSL 10. CP-DATA 11. RP-DATA 12. SMS-DELIVER ---------------- As you can see No. 3-7 and 8-12 are both on the same frame. Each sub-packet contains a sms-part (but not for the same sms). Maybe this could be the problem, but i don't know. One (potentially) related issue is that the TIO (transaction identifier) is changing wrongly. The messages are sent as follows (from abis pcap): 01. SMS_1 part 1/5: TIO: 0 02. SMS_2 part 1/4: TIO: 0 03. SMS_1 part 2/5: TIO: 1 04. SMS_2 part 2/4: TIO: 2 (SMS_1) + TIO: 1 (SMS_2) 05. SMS_1 part 3/5: TIO: 2 (SMS_2) + TIO: 3 (SMS_1) 06. SMS_2 part 3/4: TIO: 4 (SMS_1) + TIO: 3 (SMS_2) 07. SMS_2 part 3/4: TIO: 4 (SMS_2) 08. SMS_1 part 4/5: TIO: 5 (SMS_2) + TIO: 5 (SMS_1) 09. SMS_2 part 4/4: TIO: 6 (SMS_1) + TIO: 6 (SMS_2) 10. SMS_2 part 4/4: TIO: 0 (SMS_2) 11. SMS_1 part 5/5: TIO: 1 (SMS_2) + TIO: 0 (SMS_1) 12. SMS_2 part 4/4: TIO: 2 (SMS_2) 13. SMS_1 part 5/5: TIO: 1 (SMS_1) After messages 06, 09 and 13 i get a CP-ERROR message with cause 98: "Message not compatible with short message protocol state" And after messages 07, 08, and 12 i get a CP-ERROR message with cause 81: "Invalid Transaction Identifier value" I hope i'm on the right path because the usage of valgrind was useless (for me). Best regards Dennis Wehrle Am 24.11.2011 15:45, schrieb Harald Welte: > Hi Dennis, > > thanks for your bug report, I'd be happy to investigate it - but the > first question is of course: Do you have the pcap log files of the Abis > interface? > > Also, regarding your suspicion of memory related issues: One idea might > be to try the same test with openbsc under valgrind. > > Regards, > Harald >