**** BEGIN LOGGING AT Mon Aug 16 02:59:56 2010 Aug 16 03:10:59 denkenz: wtf does oFono have to do with GPS? :/ Aug 16 03:20:06 luke-jr: Please read the TODO entry related to GPS, it will answer your question Aug 16 03:20:23 I did.. Aug 16 03:20:44 Then what exactly are you asking? Aug 16 03:21:01 what does oFono have to do with GPS? Aug 16 03:21:24 oFono owns the modem ttys Aug 16 03:21:45 If the modem has a GPS unit, then enabling / disabling and otherwise controlling the GPS tty is up to oFono Aug 16 03:22:07 If the GPS unit is a discrete unit, then oFono has nothing to do with it Aug 16 03:42:40 modem ttys? Aug 16 03:42:50 :p Aug 16 03:42:56 phonet0 isn't a tty ☺ Aug 16 03:49:47 luke-jr: Please stop trolling, there are other drivers than phonet/isi Aug 16 09:45:57 Call.LineIdentification == callerId (on incoming calls)? But it's always empty (isimodem/n900modem). is it ok? any workaround? Aug 16 12:39:36 e-yes: a tcpdump -s 1500 -i phonet0 would help... Aug 16 12:52:30 pessi, http://slil.ru/29568299 http://pastebin.com/AGPrx9zA -- i see callee number in logs, but it's always empty in voicecall.LineIdentification property Aug 16 13:26:57 hello Aug 16 13:27:28 some1 alive here ? have question about probing 3g modems ... Aug 16 13:29:47 depending on driver Im using (ie. option.ko hso.ko ) I have different devices (/dev/ttyHSO0,1,2 or /dev/ttyUSB0,1,2,3 etc.... ). How to determine which device should I use for communication etc ? Aug 16 13:33:41 e-yes: the address presentation bits say "presentation restricted" Aug 16 13:45:48 denkenz: So I pushed initial work for the ZTE modems. Aug 16 13:46:27 pessi, what's "presentation restricted"? smth like "hide my number"? Aug 16 13:46:44 denkenz: USSD, CBS, Phonebook etc. do work. SMS works, but might have the same issues with incoming and delivery reports. Netreg is missing signal strength. Aug 16 13:46:57 i call from the same number and settings to maemo and it shows me number Aug 16 13:47:13 denkenz: Missing pieces are GPRS and the special RAT modem driver. Aug 16 13:48:25 e-yes: I suspect there is something fishy in Aki's decoding functions in arm.. Aug 16 14:30:24 denkenz: Around? Aug 16 14:48:07 holtmann: Am now Aug 16 14:48:19 I am playing with USSD a bit. Aug 16 14:48:39 The network operators that don't do STK are mostly emulating the menu choices via USSD. Aug 16 14:48:50 Maybe we should think about an USSD agent. Aug 16 14:49:36 I thought about it, however the USSD interaction is less structured Aug 16 14:50:04 I know, but in general it can have string or key input. Aug 16 14:50:21 The menu of course is just free from text. Aug 16 14:51:00 At minium we need to track the caller of a transaction and cancel if it goes of the bus. Aug 16 14:51:22 Can we have multiple USSD sessions at the same time? Aug 16 14:51:33 Trouble is, not all modems support cancel Aug 16 14:51:43 And 1 ussd session active only Aug 16 14:52:06 The modem might not support it, but we need to reset our state if the caller dies. Aug 16 14:52:42 If the caller dies we're still going to wait for the ussd response Aug 16 14:53:36 Yes. And we should not do that. If the caller dies, then we should just try to cancel. Aug 16 14:54:00 Anyhow, I am currently think that an agent concept (even really a lot simpler than STK agent) would be good here. Aug 16 14:54:32 More modems don't support cancel than do right now Aug 16 14:54:59 so its feasible, but making an agent with 2 functions is overkill Aug 16 14:55:46 We have 3 at least. Notification, Request, Response. Aug 16 14:56:22 Anyhow, I am just toying with ideas here. Aug 16 14:57:12 If the networks have an active req-rsp model in their network, then the agent makes a lot of sense to model it the same for the users. Aug 16 14:57:50 Currently the client/application needs to track a lot details with the State change etc. Aug 16 14:58:07 And manually calling Cancel to reset it. Aug 16 14:59:30 denkenz: And strings with umlauts are broken ;) Aug 16 14:59:52 ofonod[19299]: Aux:> AT+CUSD=1,"*100#",15\r Aug 16 14:59:52 ofonod[19299]: Aux:< \r\nOK\r\n Aug 16 14:59:52 ofonod[19299]: Modem:< \r\n+CUSD: 1,"Prepaid-Men\777\r\n1 Guthabenkonto\r\n2 Guthaben-Verf\777gbarkeit\r\n3 Aufladung Guthaben/Pack-to-Go\r\n4 Pack Manager\r\n7 Tarifinfo\r\n8 Hilfe",15\r\n Aug 16 14:59:52 ofonod[19299]: Aux:< \r\n+CUSD: 1,"Prepaid-Men\777\r\n1 Guthabenkonto\r\n2 Guthaben-Verf\777gbarkeit\r\n3 Aufladung Guthaben/Pack-to-Go\r\n4 Pack Manager\r\n7 Tarifinfo\r\n8 Hilfe",15\r\n Aug 16 14:59:54 ofonod[19299]: Modem:< \r\n+CME ERROR: 100\r\n Aug 16 14:59:54 ofonod[19299]: Aux:< \r\n+CME ERROR: 100\r\n Aug 16 15:00:06 The network also times out on ZTE cards with an error :( Aug 16 15:00:29 Request wouldn't be an Agent function, you'd pass an agent to it Aug 16 15:01:49 So you'd really have RequestResponse, Notify and Release Aug 16 15:02:06 For CUSD, the stupid modems all do it differently Aug 16 15:02:34 CUSD depends on the +CSCS setting, and unless the modem supports UTF8, they all get it wrong Aug 16 15:03:12 Not to mention if the string has embedded \r or \n Aug 16 15:05:19 Can we switch the charset first for every command? Aug 16 15:05:29 Otherwise non of the German operators will ever work. Aug 16 15:05:34 Doesn't help if CUSD is unsolicited Aug 16 15:06:03 Why not? Aug 16 15:06:34 Because we actually switch the charset in some situations (like phonebook import) Aug 16 15:06:42 Anyway, in your case its the parser screwing up Aug 16 15:06:51 We don't take '" "' into account Aug 16 15:07:51 Do we need a different parser for this as well. It should already be using permissive. Aug 16 15:09:05 We probably just need to update the STATE_RESPONSE to go into another STATE_RESPONSE_STRING when we see " Aug 16 15:09:16 and go back to STATE_RESPONSE from STATE_RESPONSE_STRING when we see another " Aug 16 15:09:30 Then hope to god there are no embedded " Aug 16 15:09:46 I am lost now. Aug 16 15:10:10 holtmann: ofonod[19299]: Aux:< \r\n+CUSD: 1,"Prepaid-Men\777\r\n1 Guthabenkonto\r\n2 Guthaben-Verf\777gbarkeit\r\n3 Aufladung Guthaben/Pack-to-Go\r\n4 Pack Manager\r\n7 Tarifinfo\r\n8 Hilfe",15\r\n Aug 16 15:10:20 See how there's a \r\n after Prepaid-Men? Aug 16 15:10:23 You mean in GAtSyntax? Aug 16 15:10:43 both parsers treat that as an end of line Aug 16 15:10:48 The \777 is u-umlaut btw. Aug 16 15:10:50 Even though its part of the string Aug 16 15:11:27 Yeah, the octal escape is another problem :( Aug 16 15:11:29 So that needs actually fixing. Don't we have the same crap with MBM and STK and v1 parser. Aug 16 15:11:44 All parsers are setup like this Aug 16 15:11:48 Actually that is the debug code in GAtChat doing that. Aug 16 15:12:46 What character set do you have set on that modem? Aug 16 15:13:19 GSM doesn't have \777 as valid Aug 16 15:13:23 No idea. Aug 16 15:14:02 I did the PB testing in between, but it goes back to this: Aug 16 15:14:03 ofonod[19299]: Aux:> AT+CSCS="IRA"\r Aug 16 15:14:03 ofonod[19299]: Aux:< \r\nOK\r\n Aug 16 15:14:34 But as I said, the GAtChat debug function does its own magic here. Aug 16 15:14:54 The \777 is most likely not what you see on the modem. Aug 16 15:15:10 \777 makes no sense Aug 16 15:15:33 Something is wrong there Aug 16 15:16:13 Check the debug char conversation code. Aug 16 15:16:15 ofonod[19299]: Aux:> AT+CUSD=1,"*104#",15\r Aug 16 15:16:15 ofonod[19299]: Aux:< \r\nOK\r\n Aug 16 15:16:15 ofonod[19299]: Modem:< \r\n+CUSD: 1,"Packmanager\r\n\r\n2 Pack buchen \r\n",15\r\n Aug 16 15:16:15 ofonod[19299]: Aux:< \r\n+CUSD: 1,"Packmanager\r\n\r\n2 Pack buchen \r\n",15\r\n Aug 16 15:16:18 ofonod[19299]: Aux:> AT+CUSD=2\r Aug 16 15:16:18 ofonod[19299]: Aux:< \r\nOK\r\n Aug 16 15:16:33 So cancel actually works for me. In case I call it before it network side times out. Aug 16 15:17:05 That is on ZTE hardware. I can test the others as well. I think the problem is that we never get told when it timed out. Aug 16 15:17:18 We are supposed to Aug 16 15:17:28 The modem always should tell us Aug 16 15:17:45 escaped_str[escaped++] = '\\'; Aug 16 15:17:45 escaped_str[escaped++] = '0' + ((c >> 6) & 07); Aug 16 15:17:45 escaped_str[escaped++] = '0' + ((c >> 3) & 07); Aug 16 15:17:46 escaped_str[escaped++] = '0' + (c & 07); Aug 16 15:17:50 Ok, that seems wrong... Aug 16 15:19:17 Crap and now I am stuck in a USSD transaction where the network never answers. Aug 16 15:19:29 We can not cancel it from our side. It is in active state. Aug 16 15:19:38 pessi, I found several places in isimodem/voicecall.c with presentation checks. what places must be checked first? Aug 16 15:20:24 denkenz: In that case USSD is screwed up and only reboot will help. Why can't we cancel USSD in active state? Aug 16 15:21:11 We do it something like: Aug 16 15:21:22 CUSD=foo <-- Request Aug 16 15:21:32 if the request returns, then we assume we can cancel Aug 16 15:21:52 otherwise we assume this is a non-cancelable transaction Aug 16 15:21:52 ofonod[26739]: Aux:> AT+CUSD=1,"*106#",15\r Aug 16 15:21:53 ofonod[26739]: Aux:< \r\nOK\r\n Aug 16 15:21:56 And then nothing. Aug 16 15:22:05 So cancel should work Aug 16 15:22:20 Cancel returns with: Aug 16 15:22:21 dbus.exceptions.DBusException: org.ofono.Error.InProgress: Operation already in progress Aug 16 15:22:49 {SupplementaryServices} [/zte1] State = active Aug 16 15:22:56 So the state transitions to active. Aug 16 15:23:49 ofonod[26824]: Aux:> AT+CSCS=?\r Aug 16 15:23:50 ofonod[26824]: Aux:< \r\n+CSCS: ("IRA","GSM","UCS2")\r\n\r\nOK\r\n Aug 16 15:23:55 Btw. in case you wanna know. Aug 16 15:25:18 Ok that's a bug Aug 16 15:29:59 Easy fixable? Aug 16 15:33:34 gimme 5 min Aug 16 15:42:16 holtmann: Try the fix I pushed Aug 16 15:46:32 ofonod[27625]: Aux:> AT+CUSD=1,"*106#",15\r Aug 16 15:46:32 ofonod[27625]: Aux:< \r\nOK\r\n Aug 16 15:46:32 ofonod[27625]: Modem:< \r\n+CUSD: 5\r\n Aug 16 15:46:32 ofonod[27625]: Aux:< \r\n+CUSD: 5\r\n Aug 16 15:47:16 And second time I get this: Aug 16 15:47:17 ofonod[27625]: Aux:> AT+CUSD=1,"*106#",15\r Aug 16 15:47:17 ofonod[27625]: Aux:< \r\nOK\r\n Aug 16 15:47:17 ofonod[27625]: Modem:< \r\n+CUSD: 0,"Lieber o2 Prepaid-Kunde, diese Abfrage steht Ihnen nur im Tarif o2 LOOP S/M/L zur Verf\777gung. Tarifwechsel k\777nnen Sie \777ber die Rufnummer 5667 vornehmen. (E06)",15\r\n Aug 16 15:47:17 ofonod[27625]: Aux:< \r\n+CUSD: 0,"Lieber o2 Prepaid-Kunde, diese Abfrage steht Ihnen nur im Tarif o2 LOOP S/M/L zur Verf\777gung. Tarifwechsel k\777nnen Sie \777ber die Rufnummer 5667 vornehmen. (E06)",15\r\n Aug 16 15:47:26 Seems the o2 network has no idea what it is doing. Aug 16 15:47:34 At least for undefined USSD codes. Aug 16 15:48:18 ofonod[27625]: Aux:> AT+CUSD=1,"*106#",15\r Aug 16 15:48:18 ofonod[27625]: Aux:< \r\nOK\r\n Aug 16 15:48:18 ofonod[27625]: Aux:> AT+CUSD=2\r Aug 16 15:48:18 ofonod[27625]: Aux:< \r\nOK\r\n Aug 16 15:48:20 ofonod[27625]: Modem:< \r\n+CUSD: 2\r\n Aug 16 15:48:21 ofonod[27625]: Aux:< \r\n+CUSD: 2\r\n Aug 16 15:48:29 Seems to be fine now. Pretty hard to get into this state. Aug 16 15:48:41 dbus.exceptions.DBusException: org.ofono.Error.Canceled: Operation has been canceled Aug 16 15:48:59 That error message is funny. Wouldn't expect an error when calling Cancel(). Aug 16 15:49:18 No, the Canceled should be for the Initiate Aug 16 15:49:26 double check that for me Aug 16 15:49:56 I did cancel. Aug 16 15:50:06 ofonod[27625]: Aux:> AT+CUSD=1,"*111#",15\r Aug 16 15:50:06 ofonod[27625]: Aux:< \r\nOK\r\n Aug 16 15:50:06 ofonod[27625]: Modem:< \r\n+CME ERROR: 100\r\n Aug 16 15:50:06 ofonod[27625]: Aux:< \r\n+CME ERROR: 100\r\n Aug 16 15:50:17 At least for other codes it returns an error. Aug 16 15:50:40 + if (ussd->pending) { Aug 16 15:50:40 + reply = __ofono_error_canceled(ussd->pending); Aug 16 15:50:41 + __ofono_dbus_pending_reply(&ussd->pending, reply); Aug 16 15:50:41 + } Aug 16 15:50:50 So Canceled should go to the Initiate, not the Cancel Aug 16 15:51:02 ofonod[27625]: Aux:> AT+CUSD=1,"*111#",15\r Aug 16 15:51:03 ofonod[27625]: Aux:< \r\nOK\r\n Aug 16 15:51:03 ofonod[27625]: Modem:< \r\n+CME ERROR: 100\r\n Aug 16 15:51:03 ofonod[27625]: Aux:< \r\n+CME ERROR: 100\r\n Aug 16 15:51:03 ofonod[27625]: Aux:> AT+CUSD=2\r Aug 16 15:51:03 ofonod[27625]: Aux:< \r\nERROR\r\n Aug 16 15:51:06 ofonod[27625]: Aux:> AT+CUSD=2\r Aug 16 15:51:07 ofonod[27625]: Aux:< \r\nERROR\r\n Aug 16 15:51:09 ofonod[27625]: Aux:> AT+CUSD=2\r Aug 16 15:51:13 ofonod[27625]: Aux:< \r\nERROR\r\n Aug 16 15:51:15 ofonod[27625]: Aux:> AT+CUSD=2\r Aug 16 15:51:17 ofonod[27625]: Aux:< \r\nERROR\r\n Aug 16 15:51:19 And I can't cancel these ones. They keep in active state. Aug 16 15:51:39 This is nasty carp. The error or timeout of USSD arrives via +CME notification. Aug 16 15:52:13 denkenz: The Cancel() call needs to reset the state even if CUSD=2 fails. Aug 16 15:53:08 I am stuck in an active USSD now. At least according to oFono, but the modem is already idle again. Aug 16 15:53:16 So only restart of ofonod can fix this. Aug 16 15:53:32 Ok, take that modem, and throw it out of the Window ;) Aug 16 15:54:18 Sure. On to the pile of the other 100. Aug 16 15:54:36 Problem is just that you told me that even for me internal MBM modem ;) Aug 16 15:55:04 hehe Aug 16 15:55:26 Your mbm is a saint compared to this Aug 16 15:55:44 And I don't think Cancel should reset the state Aug 16 15:55:59 Because on another modem it won't work this way Aug 16 15:56:41 Then we have to quirk it. That would be fine with me. However we can't have oFono being blocking the API and rendering USSD unusable only because of a user error. Aug 16 15:57:21 All Qualcomm modems seem to be broken this way. Aug 16 15:57:56 Yeah, so all Qualcomm modems can have USSD disabled Aug 16 15:58:34 And there's a race condition now in my fix... :( Aug 16 15:58:42 Can I hide under a rock somewhere? :) Aug 16 16:03:07 denkenz: Nope. We need USSD enabled for Qualcomm modems. Aug 16 16:03:24 And you can hide under a rock as long as you take your laptop with you. Aug 16 16:03:37 That's a pretty large rock then Aug 16 16:04:11 And laptop power only lasts for 2 hours or so, next you'll be telling me there should be an AC outlet there too Aug 16 16:04:57 While you are at. Maybe a car battery is not a bad idea either ;) Aug 16 16:05:29 But then, AC enabled rocks could be a nice business model. Aug 16 16:05:50 Anyway, I think I fixed the Cancel race conditions Aug 16 16:06:04 However, I've no idea what can be done about your silly CME ERROR 100 Aug 16 16:06:30 I know. That is pretty nasty handling. Aug 16 16:06:54 Since we have no idea what an unsolicited +CME responds to. Aug 16 16:08:39 ofonod[28019]: Aux:> AT+CUSD=1,"*111#",15\r Aug 16 16:08:39 ofonod[28019]: Aux:< \r\nOK\r\n Aug 16 16:08:39 ofonod[28019]: Modem:< \r\n+CME ERROR: 100\r\n Aug 16 16:08:39 ofonod[28019]: Aux:< \r\n+CME ERROR: 100\r\n Aug 16 16:08:39 ofonod[28019]: Aux:> AT+CUSD=2\r Aug 16 16:08:42 ofonod[28019]: Aux:< \r\nERROR\r\n Aug 16 16:08:43 ofonod[28019]: src/ussd.c:ussd_cancel_callback() ussd cancel failed with error: Unknown error type Aug 16 16:08:45 And I am stuck in active state. Aug 16 16:09:08 We maybe just need to use the QUALCOMM_MSM quirk to reset the sate on Cancel() method call. Aug 16 16:09:42 Also that it only says ERROR instead of giving a proper reason is pretty funny. Aug 16 16:10:03 Firmware made in China, what do you expect? Aug 16 16:10:27 [ org.ofono.SupplementaryServices ] Aug 16 16:10:27 State = active Aug 16 16:10:28 :( Aug 16 16:11:10 We would need a proper certification test suite for AT command specification. Aug 16 16:12:38 Anyhow, we need that quirk. Otherwise we dead-lock Qualcomm based cards. And not enabling USSD is not an option. Aug 16 16:15:44 Lol, why do you say that? Aug 16 16:15:53 We're probably the first ones to try USSD on these cards Aug 16 16:16:44 Operators want to use USSD. Especially the pre-paid modem cards are the only way to top up this way. You have no dialer. Aug 16 16:17:15 You might be right. We are the first ones to try this. Aug 16 16:17:24 At on Linux we are for sure the first ones every tried this. Aug 16 16:26:50 holtmann: Ok try your umlaut again Aug 16 16:30:16 USSD session state is idle Aug 16 16:30:16 (dbus.String(u'USSD'), dbus.String(u'', variant_level=1)) Aug 16 16:30:16 USSD session state is active Aug 16 16:30:16 USSD session state is idle Aug 16 16:30:20 Nothing. Aug 16 16:40:24 holtmann: I meant on the log Aug 16 16:40:30 it should show up as 377 now Aug 16 16:40:39 or something else reasonable Aug 16 16:43:36 ofonod[32073]: Modem:< \r\n+CUSD: 0,"Ihr Prepaid-Guthaben betr\377gt 2.35 EUR. Jetzt auch Ihr Guthaben aufladen: einfach *103*Aufladenummer# und die H\377rertaste eingeben.",15\r\n Aug 16 16:43:41 It does. Aug 16 16:43:58 Btw. this is the same O2 card that you have. Aug 16 16:44:23 And since I gave you a ZTE modem, you have the same broken hardware now ;) Aug 16 16:44:27 My AT&T signal is shit, I'd spend more time on cell reselections Aug 16 16:44:43 Make it roam to T-Mobile. Aug 16 16:46:06 I have enough info to do some investigations anyway Aug 16 16:46:23 Like fixing the silly embedded \r\ns Aug 16 16:46:52 If that works we should get somewhere and then a quirk for state reset on cancel. And this should be good enough. Aug 16 16:47:09 Do your umlauts show up in the reported string? Aug 16 16:47:20 That previous USSD should display just fine Aug 16 16:47:41 I never get the string reported anywhere else than in the debug AT log. Aug 16 16:48:35 That's strange Aug 16 16:49:13 ofonod[32073]: Aux:> AT+CUSD=1,"*101#",15\r Aug 16 16:49:14 ofonod[32073]: Aux:< \r\nOK\r\n Aug 16 16:49:14 ofonod[32073]: Aux:> AT+CUSD=2\r Aug 16 16:49:45 This is the worst. The card is blocking further AT commands while waiting for the +CSUD or +CME :( Aug 16 16:53:48 Yikes Aug 16 16:53:59 You seriously need to give up on USSD with that one Aug 16 16:58:02 Most of the times it just works fine. Aug 16 16:58:42 yeah until it stalls the aux channel and you're stuck Aug 16 17:04:26 [ org.ofono.SupplementaryServices ] Aug 16 17:04:27 State = user-response Aug 16 17:04:34 ofonod[32366]: Aux:> AT+CUSD=1,"*100#",15\r Aug 16 17:04:34 ofonod[32366]: Aux:< \r\nOK\r\n Aug 16 17:04:35 ofonod[32366]: Modem:< \r\n+CUSD: 1,"Prepaid-Men\377\r\n1 Guthabenkonto\r\n2 Guthaben-Verf\377gbarkeit\r\n3 Aufladung Guthaben/Pack-to-Go\r\n4 Pack Manager\r\n7 Tarifinfo\r\n8 Hilfe",15\r\n Aug 16 17:04:35 ofonod[32366]: Aux:< \r\n+CUSD: 1,"Prepaid-Men\377\r\n1 Guthabenkonto\r\n2 Guthaben-Verf\377gbarkeit\r\n3 Aufladung Guthaben/Pack-to-Go\r\n4 Pack Manager\r\n7 Tarifinfo\r\n8 Hilfe",15\r\n Aug 16 17:04:35 ofonod[32366]: Aux:> AT+CUSD=1,"1",15\r Aug 16 17:05:04 And it stalls :( Aug 16 17:05:06 Not nice. Aug 16 17:05:28 I think we need to fix for the "\r\n". Aug 16 17:16:53 denkenz: I pushed initiate-ussd test script. That should also work for user-response scenarios. At least the simple one. Aug 16 17:17:53 ofonod[32540]: Aux:> AT+CUSD=1,"*100#",15\r Aug 16 17:17:53 ofonod[32540]: Aux:< \r\nOK\r\n Aug 16 17:17:53 ofonod[32540]: Modem:< \r\n+CUSD: 1,"Prepaid-Men\377\r\n1 Guthabenkonto\r\n2 Guthaben-Verf\377gbarkeit\r\n3 Aufladung Guthaben/Pack-to-Go\r\n4 Pack Manager\r\n7 Tarifinfo\r\n8 Hilfe",15\r\n Aug 16 17:17:53 ofonod[32540]: Aux:< \r\n+CUSD: 1,"Prepaid-Men\377\r\n1 Guthabenkonto\r\n2 Guthaben-Verf\377gbarkeit\r\n3 Aufladung Guthaben/Pack-to-Go\r\n4 Pack Manager\r\n7 Tarifinfo\r\n8 Hilfe",15\r\n Aug 16 17:17:55 ofonod[32540]: Aux:> AT+CUSD=1,"1",15\r Aug 16 17:17:56 ofonod[32540]: Aux:< \r\nOK\r\n Aug 16 17:17:58 ofonod[32540]: Modem:< \r\n+CUSD: 0,"Ihr Prepaid-Guthaben betr\377gt 2.35 EUR. Jetzt auch Ihre Guthabenverf\377gbarkeit abrufen: einfach *102# und die H\377rertaste eingeben.",15\r\n Aug 16 17:18:01 ofonod[32540]: Aux:< \r\n+CUSD: 0,"Ihr Prepaid-Guthaben betr\377gt 2.35 EUR. Jetzt auch Ihre Guthabenverf\377gbarkeit abrufen: einfach *102# und die H\377rertaste eingeben.",15\r\n Aug 16 17:18:13 If I would get the umlaut strings, then this would be working fine. Aug 16 17:18:26 State: idle Aug 16 17:18:27 USSD: Aug 16 17:18:27 State: user-response Aug 16 17:18:27 Enter response: 1 Aug 16 17:18:31 But reported strings are empty. Aug 16 17:19:11 If we don't wanna do the agent style, then we should at least return the state to the Initiate() and Respond() method calls. Aug 16 17:19:26 So that the application doesn't have to monitor State property as well. Aug 16 17:19:44 And not stupidly keeps calling GetProperties like my test script does. Aug 16 17:23:53 Err, it doesn't really need to call state Aug 16 17:24:13 But I'm confused why the simple case doesn't emit the signals Aug 16 17:24:21 Or rather emits empty strings Aug 16 17:26:31 Can you set the CSCS to "GSM" ? Aug 16 17:26:50 Want me to hack that manually into it. Aug 16 17:27:17 yeah, just send it in the modem driver Aug 16 17:27:30 It seems our USSD driver does a GSM -> UTF8 conversion Aug 16 17:27:41 so maybe that umlaut is confusing it Aug 16 17:27:52 This area of the spec is fucked btw Aug 16 17:28:07 I'd rather get the USSD PDU from the modem somehow Aug 16 17:28:25 Letting the modem do the conversion is utterly braindead Aug 16 17:32:12 ofonod[5925]: Aux:> AT+CSCS="GSM"\r Aug 16 17:32:12 ofonod[5925]: Aux:< \r\nOK\r\n Aug 16 17:32:12 ofonod[5925]: Aux:> AT+CUSD=1,"*100#",15\r Aug 16 17:32:12 ofonod[5925]: Aux:< \r\nOK\r\n Aug 16 17:32:15 This breaks. Aug 16 17:32:20 No responses. Aug 16 17:33:26 So CUSD doesn't have a PDU mode? Aug 16 17:33:48 nope Aug 16 17:33:55 funny Aug 16 17:34:11 so you need to add a quirk to skip the gsm to utf8 conversion Aug 16 17:34:33 ofonod[5999]: Aux:> AT+CSCS="IRA"\r Aug 16 17:34:33 ofonod[5999]: Aux:< \r\nOK\r\n Aug 16 17:34:33 ofonod[5999]: Aux:> AT+CUSD=1,"*100#",15\r Aug 16 17:34:33 ofonod[5999]: Aux:< \r\nOK\r\n Aug 16 17:34:36 ofonod[5999]: Modem:< \r\n+CUSD: 1,"Prepaid-Men\377\r\n1 Guthabenkonto\r\n2 Guthaben-Verf\377gbarkeit\r\n3 Aufladung Guthaben/Pack-to-Go\r\n4 Pack Manager\r\n7 Tarifinfo\r\n8 Hilfe",15\r\n Aug 16 17:34:37 That works then. Aug 16 17:34:58 try UCS2 Aug 16 17:35:03 does it work on that one? Aug 16 17:36:30 The funny part is that IRA only supports 7 bit characters Aug 16 17:36:52 so the fact that you're getting umlaut as 255, means that the firmware guys had no fucking clue what they were doing Aug 16 17:39:07 ofonod[6094]: Aux:> AT+CSCS="UCS2"\r Aug 16 17:39:07 ofonod[6094]: Aux:< \r\nOK\r\n Aug 16 17:39:07 ofonod[6094]: Aux:> AT+CUSD=1,"*100#",15\r Aug 16 17:39:07 ofonod[6094]: Aux:< \r\nERROR\r\n Aug 16 17:41:50 ofonod[6177]: Aux:> AT+CSCS="GSM"\r Aug 16 17:41:50 ofonod[6177]: Aux:< \r\nOK\r\n Aug 16 17:41:51 ofonod[6177]: Aux:> AT+CUSD=1,"*100#",15\r Aug 16 17:41:51 ofonod[6177]: Aux:< \r\nOK\r\n Aug 16 17:41:52 ofonod[6177]: Modem:< \r\n+CUSD: 1,"Prepaid-Men~\r\n1 Guthabenkonto\r\n2 Guthaben-Verf~gbarkeit\r\n3 Aufladung Guthaben/Pack-to-Go\r\n4 Pack Manager\r\n7 Ta",15\r\n Aug 16 17:41:52 ofonod[6177]: Aux:< \r\n+CUSD: 1,"Prepaid-Men~\r\n1 Guthabenkonto\r\n2 Guthaben-Verf~gbarkeit\r\n3 Aufladung Guthaben/Pack-to-Go\r\n4 Pack Manager\r\n7 Ta",15\r\n Aug 16 17:41:55 ofonod[6177]: Aux:> AT+CSCS="GSM"\r Aug 16 17:41:57 ofonod[6177]: Aux:< \r\nOK\r\n Aug 16 17:41:59 ofonod[6177]: Aux:> AT+CUSD=1,"1",15\r Aug 16 17:42:01 ofonod[6177]: Aux:< \r\nOK\r\n Aug 16 17:42:03 ofonod[6177]: Modem:< \r\n+CUSD: 0,"Ihr Prepaid-Guthaben betr{gt 2.35 EUR. Jetzt auch Ihre Guthabenverf~gbarkeit abrufen: einfach *102# und die H|re",15\r\n Aug 16 17:42:06 ofonod[6177]: Aux:< \r\n+CUSD: 0,"Ihr Prepaid-Guthaben betr{gt 2.35 EUR. Jetzt auch Ihre Guthabenverf~gbarkeit abrufen: einfach *102# und die H|re",15\r\n Aug 16 17:42:09 ofonod[6177]: drivers/atmodem/ussd.c:cusd_parse() udhi 0 compressed 0 iso639 0 Aug 16 17:42:11 Almost ;) Aug 16 17:42:15 State: idle Aug 16 17:42:17 USSD: Aug 16 17:42:19 State: user-response Aug 16 17:42:21 Enter response: 1 Aug 16 17:42:23 Ihr Prepaid-Guthaben beträgt 2.35 EUR. Jetzt auch Ihre Guthabenverfügbarkeit abrufen: einfach *102# und die Höre Aug 16 17:43:16 Yeah, see so that's better Aug 16 17:43:34 You even get your umlauts Aug 16 17:43:47 Still the initial response has no message with it. Aug 16 17:43:57 USSD: %s this is Aug 16 17:44:13 This was a two stage USSD. I fail to see the menu. Aug 16 17:44:14 that's because of the embedded \r\n, let me push a fix Aug 16 17:45:31 holtmann: Try now Aug 16 17:49:30 Nope. Aug 16 17:49:54 ofonod[6697]: Aux:> AT+CSCS="GSM"\r Aug 16 17:49:54 ofonod[6697]: Aux:< \r\nOK\r\n Aug 16 17:49:54 ofonod[6697]: Aux:> AT+CUSD=1,"*101#",15\r Aug 16 17:49:54 ofonod[6697]: Aux:< \r\nOK\r\n Aug 16 17:49:54 ofonod[6697]: Modem:< \r\n+CUSD: 0,"Ihr Prepaid-Guthaben betr{gt 2.35 EUR. Jetzt auch Ihr Guthaben aufladen: einfach *103*Aufladenummer# und die H|re",15\r\n Aug 16 17:49:56 ofonod[6697]: Aux:< \r\n+CUSD: 0,"Ihr Prepaid-Guthaben betr{gt 2.35 EUR. Jetzt auch Ihr Guthaben aufladen: einfach *103*Aufladenummer# und die H|re",15\r\n Aug 16 17:49:58 ofonod[6697]: drivers/atmodem/ussd.c:cusd_parse() udhi 0 compressed 0 iso639 0 Aug 16 17:50:04 State: idle Aug 16 17:50:04 USSD: Ihr Prepaid-Guthaben beträgt 2.35 EUR. Jetzt auch Ihr Guthaben aufladen: einfach *103*Aufladenummer# und die Höre Aug 16 17:50:06 Without it works fine. Aug 16 17:50:22 Also it is cut short btw. Aug 16 17:51:01 It is not proper German sentence. It should finish with "Hoere druecken" and not just "Hoer" Aug 16 17:51:34 hmm, were you getting the full sentence with IRA? Aug 16 17:52:15 And my patch should not affect the USSDs you're getting, strange Aug 16 17:53:15 Yes, I think I was. Aug 16 17:53:43 Btw. I am thinking the CBS interferes with the USSD in this network. Aug 16 17:54:06 ofonod[6787]: Aux:> AT+CSCS="IRA"\r Aug 16 17:54:06 ofonod[6787]: Aux:< \r\nOK\r\n Aug 16 17:54:06 ofonod[6787]: Aux:> AT+CUSD=1,"*102#",15\r Aug 16 17:54:06 ofonod[6787]: Aux:< \r\nOK\r\n Aug 16 17:54:07 ofonod[6787]: Modem:< \r\n+CUSD: 0,"Ihr Prepaid-Guthaben ist bis zum 26.11.10 verf\377gbar. ",15\r\n Aug 16 17:54:07 ofonod[6787]: Aux:< \r\n+CUSD: 0,"Ihr Prepaid-Guthaben ist bis zum 26.11.10 verf\377gbar. ",15\r\n Aug 16 17:54:08 ofonod[6787]: drivers/atmodem/ussd.c:cusd_parse() udhi 0 compressed 0 iso639 0 Aug 16 17:54:37 Comapred to GSM: Aug 16 17:54:38 ofonod[6697]: Aux:> AT+CSCS="GSM"\r Aug 16 17:54:38 ofonod[6697]: Aux:< \r\nOK\r\n Aug 16 17:54:38 ofonod[6697]: Aux:> AT+CUSD=1,"*102#",15\r Aug 16 17:54:38 ofonod[6697]: Aux:< \r\nOK\r\n Aug 16 17:54:41 ofonod[6697]: Modem:< \r\n+CUSD: 0,"Ihr Prepaid-Guthaben ist bis zum 26.11.10 verf~",15\r\n Aug 16 17:54:41 ofonod[6697]: Aux:< \r\n+CUSD: 0,"Ihr Prepaid-Guthaben ist bis zum 26.11.10 verf~",15\r\n Aug 16 17:54:43 ofonod[6697]: drivers/atmodem/ussd.c:cusd_parse() udhi 0 compressed 0 iso639 0 Aug 16 17:59:17 The only suggestion I have is to assume that IRA on this firmware means UTF8 Aug 16 17:59:31 And to simply skip the gsm->utf8 conversion via a quirk Aug 16 18:01:07 Otherwise my patch should not affect the behavior Aug 16 18:02:38 Give me a hint in the code to test this. Aug 16 18:03:04 if (charset == SMS_CHARSET_7BIT) ?? Aug 16 18:03:08 Yeap Aug 16 18:03:25 So I just strdup the original string? Aug 16 18:03:52 Basically cusd parse with your quirk would just ofono_ussd_notify and return Aug 16 18:04:09 Right before cbs_dcs_decode Aug 16 18:04:14 if (converted) Aug 16 18:04:14 g_free(converted); Aug 16 18:04:20 Dude, really ;) Aug 16 18:04:38 Hey, that code is ancient ;) Aug 16 18:04:51 That driver was among the first written Aug 16 18:05:51 So I really think that USSD and CBS conflicts in this firmware. Aug 16 18:07:05 ofonod[7138]: Modem:< \r\n+ZPSTM: 4,534D53205365727669636573;1,496E666F205365727669636573;2,577C7274657262756368;3,454D61696C;4,466178;\r\n Aug 16 18:07:05 ofonod[7138]: Aux:< \r\n+ZPSTM: 4,534D53205365727669636573;1,496E666F205365727669636573;2,577C7274657262756368;3,454D61696C;4,466178;\r\n Aug 16 18:07:09 What the f* is this. Aug 16 18:07:29 ofonod[7138]: Modem:< \r\n+ZEND\r\n Aug 16 18:07:29 ofonod[7138]: Aux:< \r\n+ZEND\r\n Aug 16 18:07:30 ofonod[7138]: Modem:< \r\n+ZEND\r\n Aug 16 18:07:30 ofonod[7138]: Aux:< \r\n+ZEND\r\n Aug 16 18:07:33 SIM Toolkit ??? Aug 16 18:09:52 Probably Aug 16 18:10:23