**** BEGIN LOGGING AT Wed May 26 02:59:57 2010 May 26 04:25:17 zhenhua: ppp server needs to give out the IP address to the client May 26 04:25:37 zhenhua: When GPRS context is activated the IP is negotiated via PPP/LCP May 26 04:26:27 zhenhua: Secondly we need to support this even when the modem does not talk ppp natively May 26 04:26:41 zhenhua: e.g. ISI or high-speed interfaces like HSO/MBM May 26 04:27:47 yang_office: Yes, TP-UDL cannot exceed 140 bytes in the case of 8-bit May 26 04:28:14 yang_office: Most likely the case has incorrect dcs May 26 04:31:48 denkenz: it makes the interface be more complicated. who is going to implement ppp server part? May 26 04:32:23 denkenz: so far, i have a simple ppp command forwarding to modem only May 26 04:32:51 zhenhua: Someone has to, ppp forwarding is not an option May 26 04:33:08 ppp forwarding is quite simple. May 26 04:33:28 it can be used for the modem that talks with ppp May 26 04:33:35 yes, but does not help us May 26 04:33:53 The core does not know that a modem talks ppp or ip over pigeons May 26 04:34:12 yes, that is clear May 26 04:35:27 denkenz: i refactor my emulator part, so now it's more like a call history implementation May 26 04:35:49 different emulator register driver statically under plugins dir May 26 04:36:08 i think it makes more sense May 26 04:36:33 it depends on plugins/bluetooth.c to detect bluetooth activity. May 26 04:45:01 yang_office: Ah, I understand now why they have UDL of 160 May 26 04:45:12 a short message to be sent to the network in an SMS-SUBMIT message where the text needs to be packed by the ME. May 26 04:45:43 We actually need to pack it ourselves May 26 04:47:49 denkenz: So we need to change the parser May 26 04:48:26 yang_office: Yes, probably to handle the specific case of the command qualifier being 'packing required' May 26 04:48:43 Basically the guys who thought this up need to be shot May 26 04:50:13 But in some ber-tlv, the data is in format of septet May 26 04:50:32 yeah, the command qualifier tells you whether packing is required or not May 26 04:50:41 Thus I see. May 26 04:50:50 31.111 Section 6.4.10 May 26 04:52:11 ok, thanks. May 26 04:52:24 I will read it and submit the patch. May 26 04:53:03 zhenhua: I fixed the style issues and resent the patch set. May 26 04:53:44 yang_office: Can you recheck all the unit tests that you've had a comment on for incorrectness of the spec? May 26 04:53:58 yang_office: Perhaps there's an errata? May 26 04:54:25 Where is the errata? May 26 04:54:48 some specs have it, dig around on 3gpp site May 26 04:54:49 Actually all the rest are inconsistency between the data and description. May 26 04:55:03 So I think they are just typo. May 26 04:55:20 But I will have a double check. May 26 04:55:24 You're probably right, but double check just in case May 26 04:56:21 I have another question on extract_bcd_number() May 26 04:56:39 sure May 26 04:56:46 In our implementation, 12 means 'a' May 26 04:57:01 But it actually means a pause, right? May 26 04:57:19 static const char digit_lut[] = "0123456789*#abc\0"; May 26 04:57:58 Gah, its been a while May 26 04:58:59 In the test spec, there is some example such as '1' p '2'. Then I check it with "1a2", am I correct? May 26 04:59:56 I think its all context specific May 26 05:00:26 for phone numbers it might well be that 'a' is meant as a pause May 26 05:01:05 ok, I think I am fine with the check. May 26 05:01:40 Just wonder why we interpret 12 to 'a'. May 26 05:01:41 It made sense to me about 6 months ago, now I need to recheck the specs ;) May 26 05:02:03 Which spec says 12 needs to be 'a'. May 26 05:02:43 I ever saw the spec related to bcd number. But it doesn't tell me that. I wonder if i miss some other spec. May 26 05:03:14 compare sim_parse_mcc_mnc and extract_bcd_number May 26 05:03:25 The luts are actually different May 26 05:03:45 That's because the BCD encoding is actually context specific May 26 05:04:47 Do you imply maybe for stk, we need another specific table? May 26 05:05:05 table means luts May 26 05:05:29 btw, that's the full name of lut? May 26 05:05:58 s/that/what May 26 05:06:32 look-up-table May 26 05:07:08 i see May 26 05:08:06 Thus do we need another table for stk? May 26 05:08:09 padovan: just saw that, thanks. May 26 05:08:17 For example, send dtmf 1.1.1 May 26 05:08:30 DTMF String: "1" pause "2" May 26 05:08:58 pause is 'C' May 26 05:09:04 yang_office: potentially May 26 05:09:33 23.040 (SMS) defines BCD as follows: May 26 05:09:34 If a mobile receives an address field containing non‑integer information in the semi‑octets other than "1111" (e.g. 1110) it shall display the semi‑octet as the representation given in GSM 44.008 [12] under "called BCD number", viz 1010="*", 1011="#", 1100="a", 1101="b", 1110="c". In the event of a discrepancy between the values quoted here and the values specified in GSM 44.008[12] then GSM 44.008 [12] shall take precedence. If a May 26 05:09:36 mobile receives "1111" in a position prior to the last semi‑octet then processing shall commence with the next semi‑octet and the intervening semi‑octet shall be ignored. May 26 05:11:17 So we are safe to map 'C' May 26 05:11:30 map 'C' to "a" May 26 05:11:51 ok, thanks May 26 05:12:00 nod, I'm off to sleep now :) May 26 05:12:25 Good night, and you will see some patch tomorrow morning;) May 26 05:15:42 holtmann: I still can't see your mail :( May 26 07:57:24 holtmann: weird, I can't reproduce the udev related ofono crash anymore May 26 08:21:29 kvalo: Good since I didn't see anything obvious wrong. May 26 09:18:12 ok, if I could get ofono + phonesim working, I would be glad May 26 09:25:06 holtmann: strangely enough I received one email from you today, but still nothing from yesterday May 26 09:29:38 kvalo: Stupid mailserver. Assume they are lost :( May 26 09:29:58 lpotter: No idea what's so hard with that. May 26 09:50:00 it's not. but I've always had a love-hate relationship with phonesim :) May 26 09:53:25 lpotter: cann't u use phonesim with ofono? May 26 09:53:39 yup May 26 09:53:39 i think you can get answer from mailing list. :-) May 26 09:53:47 what's your problem May 26 09:54:12 no problem. :) May 26 09:54:37 okay May 26 10:32:35 * balrog-k1n notes the whole BCD notation in gsm is not actual BCD (because BCD is only for numbers) May 26 13:35:24 balrog-k1n: They're always referred to as BCD in the specs, but they're also context specific May 26 13:39:13 denkenz: i know, my point is D in BCD is for decimal and real BCD is independent of context May 26 13:40:24 if your number can contain *s and #s and A, B, C, and Ds, it's hexadecimal :) May 26 13:42:33 I think its binary coded digit May 26 13:43:28 either way, its abused :) May 26 13:43:33 oh, never heard binary coded digit, but may be May 26 14:07:13 denkenz: around? May 26 14:10:21 akiniemi: yep May 26 14:13:49 denkenz: your thoughts on Pekka's online patch? May 26 14:13:56 denkenz: ready to be pushed? May 26 14:15:19 akiniemi: I honestly haven't done a thorough review, at a glance it looks good May 26 14:20:31 I'm trying to go through my backlog, so I'll try to review them today / tomorrow May 26 14:24:02 akiniemi: i'll fix the isimodem and the set_online() call without callback May 26 14:24:21 ...the reason why I did not use cbd May 26 14:25:21 oohaah. I got the mini pci adaptors from Taiwan, but still no modems. But they are coming Lauttasaari. May 26 14:27:07 pessi: lol May 26 14:28:09 pessi: what's the state of n900 support over usb in ofono? I tried it quickly, but didn't get far May 26 14:28:39 two different "Elisa" services were visible in connman but they didn't work May 26 14:29:38 kvalo: voicecalls should work May 26 14:30:00 nah, I don't talk with people. I want data! ;) May 26 14:33:02 there was the problem of using SOS address when PC address was supposed to be used May 26 14:39:05 seems to me that head works fine May 26 14:39:16 kvalo: If you get cdc_phonet working, then GPRS should work May 26 14:39:52 {PrimaryDataContext} [/isimodem0/primarycontext1] Settings = dbus.Dictionary({db May 26 14:39:52 us.String(u'Interface'): dbus.String(u'gprs0', variant_level=1), dbus.String(u'M May 26 14:39:52 ethod'): dbus.String(u'dhcp', variant_level=1), dbus.String(u'DomainNameServers' May 26 14:39:52 ): dbus.Array([dbus.String(u'193.229.0.40'), dbus.String(u'193.229.0.42')], sign May 26 14:39:52 ature=dbus.Signature('s'), variant_level=1), dbus.String(u'Address'): dbus.Strin May 26 14:39:54 g(u'85.156.7.11', variant_level=1)}, signature=dbus.Signature('sv'), variant_lev May 26 14:39:57 el=1) May 26 14:40:18 pessi: way cool, thanks May 26 14:40:53 akiniemi: I'll check that I have cdc_phonet. on my laptop, right? May 26 14:42:12 kvalo: yup May 26 14:43:36 kiitti May 26 14:44:39 kvalo: eipä kestä :) May 26 14:45:07 kvalo: stock ubuntu kernel AFAIK doesn't have cdc_phonet built May 26 14:46:15 that explains, I was using the default kernel from 10.04 May 26 14:46:19 at least karmic has May 26 14:46:41 actually, mine has also May 26 14:47:01 but I'll debug it later May 26 14:50:08 Oh. I stand corrected. ;) May 26 14:59:15 akiniemi&denkenz: I sent the new isimodem driver and appropriately modified core pathc May 26 15:01:20 yang_office: I just added sim versions of extract_bcd_number / encode_bcd_number so use those for your DTMF stuff May 26 15:02:23 akiniemi: For your mms patch, I'm thinking we still might want some form of a timer May 26 15:02:47 akiniemi: Just to avoid sending CMMS n-1 times on a n-fragment sms message May 26 15:03:35 kvalo: Are you sending new versions of the fix for your crash? May 26 15:03:39 my timer code was just plain wrong though May 26 15:03:49 I really wanna get a release done today. May 26 15:04:13 Or latest tomorrow. May 26 15:04:21 holtmann: I was planning to send them tomorrow, but if you are waiting for them. I need 30 mins May 26 15:04:30 I would also like to have new release of ofono :) May 26 15:04:35 with my patches May 26 15:06:54 denkenz: perhaps that should be handled in the AT driver, though May 26 15:07:46 akiniemi: Does your MMS flag work differently from AT one? May 26 15:08:18 akiniemi: Also, we should take care of multiple single fragment smses being mms-ed too May 26 15:09:22 denkenz: yes, in ISI, we set mms as a flag but otherwise the req is identical May 26 15:10:14 denkenz: true, we should be checking if tx queue length > 1 May 26 15:10:24 akiniemi: So why don't we do the timer in the core May 26 15:10:57 denkenz: what did you want to avoid with CMMS? May 26 15:11:19 akiniemi: just sending CMMS a bunch of times May 26 15:11:33 e.g. I send a 200 fragment message, with your patch we send CMMS 199 times May 26 15:12:03 When in reality we probably need to once-twice May 26 15:12:15 denkenz: that's then different from ISI May 26 15:13:02 akiniemi: Ok, then lets at least do the mms flag for multiple pending smses May 26 15:13:10 akiniemi: I take care of doing the timer in the driver May 26 15:14:27 denkenz: I'll fix the patch and resend May 26 15:14:41 akiniemi: thanks May 26 16:18:04 unit/test-stkutil.c:4801: error: initialized field overwritten May 26 16:18:50 that's weird May 26 16:18:57 balrog-k1n: ^^^^^ May 26 16:19:03 pessi: some gcc versions don't understand the initialisers in test-stkutil.c May 26 16:19:18 pessi: i have a patch for that (badly formatter, hence not submitted) May 26 16:19:23 formatted* May 26 16:19:34 pessi: What gcc are you using? May 26 16:19:59 i'm using 4.3.2 and test-stkutil.c won't build here May 26 16:20:01 without the patch May 26 16:20:14 hmm, 4.4.2 works fine here May 26 16:20:32 I have 4.4.3 and it is also just fine. May 26 16:20:49 it's the 4.3 then, that's what i was thinking May 26 16:21:38 what's the proposed solution? May 26 16:21:48 * balrog-k1n tries to pastebin the file.. May 26 16:22:29 scratchbox one May 26 16:22:31 sbox-arm-none-linux-gnueabi-gcc (cs2009q1) 4.3.3 May 26 16:23:15 pessi: Can you compile with -fdiagnostics-show-option May 26 16:23:33 holtmann: kvalo does want that, he needs to reset the state to IDLE May 26 16:23:56 denkenz: I figured. Just double checking since the previous code wasn't doing that. May 26 16:24:16 And in his previous patches he has keeping the return; May 26 16:25:01 denkenz: So after Kalle's patches, what do you have pending? May 26 16:25:07 Yeah, his patches are broken up wrong though May 26 16:25:18 http://openstreetmap.pl/balrog/gcc4.3.patch anyway, i can't find a single pastebin that accepts files May 26 16:25:20 The second chunk of 1_2 is not really relevant May 26 16:25:30 pastebin.ca did May 26 16:25:40 pastebin.ca it times out for me May 26 16:25:48 s/it// May 26 16:25:58 holtmann: Or at least the commit message should be changed May 26 16:26:10 ftr i've never seen .a[n] = 5; style initialisers before May 26 16:26:57 The whole [0] seems useless May 26 16:27:10 can't we just make it into a regular array/struct initializer? May 26 16:27:30 yeah, it may be useless, i just ran a sed on that file May 26 16:27:51 .items = { { .id = 2, .text = "Item 2" } } May 26 16:28:01 probably save some code too May 26 16:28:53 denkenz: Commit message" May 26 16:29:37 holtmann: Yeah, its not remove cgact, maybe 'refactor context deactivate' May 26 16:29:37 holtmann: xyzzy. it won't tell anything May 26 16:30:14 pessi: Too bad. Sometimes this magic tells what the compiler really thinks. May 26 16:30:35 denkenz: last i checked this syntax didn't not allow fewer elements than the size of the array May 26 16:30:58 doing the [0] = ...; [1] = ...; it just zeroed the other elements May 26 16:31:02 balrog-k1n: I always thought it did May 26 16:31:09 balrog-k1n: One way to find out :) May 26 16:31:15 it may have changed May 26 16:32:26 - .items[0] = { May 26 16:32:28 - .id = 0x11, May 26 16:32:29 - .text = "One" May 26 16:32:31 - }, May 26 16:32:32 - .items[1] = { May 26 16:32:34 - .id = 0x12, May 26 16:32:35 - .text = "Two" May 26 16:32:37 + .items = { May 26 16:32:38 + { .id = 0x11, .text = "One" }, May 26 16:32:40 + { .id = 0x12, .text = "Two" }, May 26 16:32:41 Works for me May 26 16:35:00 yeah, works with 4.3 too May 26 16:35:08 holtmann: I've nothing pending, but we should probably fix this stkutil problem May 26 16:35:16 I agree. May 26 16:38:27 balrog-k1n: So are you writing some sed-fu to do this? May 26 16:39:45 denkenz: do we want it upstream? May 26 16:40:01 balrog-k1n: Yeah, 4.3 is a fairly recent compiler May 26 16:40:08 balrog-k1n: So lets make it work May 26 16:40:13 ok May 26 16:40:48 in the meantime I review Kalle's patches so we can do this release May 26 16:41:07 i just need to check all of the tests use consecutive indices May 26 16:41:18 nod May 26 17:08:17 balrog-k1n: Can you rebase your patch? May 26 17:08:37 balrog-k1n: Ah nevermind May 26 17:37:02 denkenz: true, second chunk was not needed in my patch 1 May 26 17:38:14 kvalo: No worries, I just edited the commit message May 26 17:38:24 denkenz: thanks May 26 17:38:47 holtmann: Ok I think we're good to go May 26 17:46:58 denkenz: Did you push everything? May 26 17:47:20 aw crap, forgot to hit enter May 26 17:47:46 holtmann: now pushed May 26 17:48:36 I run a few internal build test and then this gets released. May 26 18:14:16 denkenz: Last chance. Build tests are done and I am composing the changelog now. May 26 18:15:56 holtmann: Not that we tested any of this, but go for it May 26 19:14:04 So ofono-0.21.tar.gz is uploading right now. **** ENDING LOGGING AT Thu May 27 02:59:57 2010