**** BEGIN LOGGING AT Mon Oct 04 02:59:57 2010 Oct 04 06:34:55 pessi: I need few clarifications on the latest modifications done on isimodem. Oct 04 06:49:31 I'm not able to use the N900 in PC Suite mode over USB with ofono running in PC side after the patch isigen: generic modem driver plugin for ISI Oct 04 06:49:49 Has the procedure changed? Oct 04 06:51:09 Jeevaka: Did you copy the new udev rules? Oct 04 07:06:01 Jeevaka: It works fine for me btw. Oct 04 07:14:30 Whats the new udev rules? Oct 04 07:16:14 holtmann: n900 works for me as well. but dns is still broken. Oct 04 07:16:39 holtmann: any chances to apply pessi's patch which fixes it? Oct 04 07:20:02 ok, I got it. Oct 04 07:53:35 holtmann: need your opinion on something. while the terminology differs cdma1x seems to be compatible with gprs as far as the ofono api is concerned, this is important because the connectionmanager service only (and explicitly?) only supports gprs. the one issue I seem to be having is that the only way to see what data contexts are active on a cdma1x device is +CLCC and look for #777 (usually) with a high cid, I think this means I nee Oct 04 07:56:04 so anyway, I've begun to implement a gprs driver for the cdma1xdata "modem" driver, and a plugin to use it on my phone, I'm hoping you can tell me the cleanest way to hook +CLCC without copy and pasting atmodem.c Oct 04 07:58:11 tmzt_, just so you know one of your messages got cut off at the 300 character limit Oct 04 07:58:25 oops Oct 04 07:58:27 tmzt_, " with a high cid, I think this means I ne" Oct 04 07:58:41 I think this means I need to hook the atmodem CLCC callback in a vendor-specific way Oct 04 07:58:44 07:54 < tmzt_> so anyway, I've begun to implement a gprs driver for the cdma1xdata "modem" driver, and a plugin to use it on my phone, I'm hoping you can tell me the cleanest way to hook +CLCC without copy and pasting atmodem.c Oct 04 08:15:53 tmzt_: Why do you need +CLCC for it. Oct 04 08:17:59 it may be possible to use CONNECT, but only for the initial connection Oct 04 08:19:40 root@rhobuntu:~# echo -e 'AT+CLCC\r' > /dev/smd0 Oct 04 08:19:40 root@rhobuntu:~# AT+CLCC Oct 04 08:19:40 +CLCC: 254,0,0,1,0,"#777",129 Oct 04 08:19:40 OK Oct 04 08:19:56 we also get this, but it's HTC cdma specific Oct 04 08:19:59 OK Oct 04 08:20:00 [MSM] Call End, curr_mode 2, call_type 2, end_status 25 Oct 04 08:20:00 $HTC_3GIND: 2 Oct 04 08:20:00 $HTC_3GIND: 1 Oct 04 08:20:54 holtmann: I mean there are no 3gpp gprs context commands to show the active context Oct 04 08:21:31 we also have to use at+chld to switch from data to voice or back (drop or hold the voice call) because CDMA doesn't support simulataenous voice and data Oct 04 08:21:51 That is tricky and nasty. Oct 04 08:22:06 You might need to have a chat with Denis when he is back from vacation next week. Oct 04 08:22:10 I think, from reading specs and ofono source, that gsm does this with AT+CGDCONT and can bring up and down or suspend contexts Oct 04 08:22:14 ok Oct 04 08:22:32 we also have to use hayes commands to dial in the first place :) Oct 04 08:22:38 atd#777 Oct 04 08:22:42 A phone call will suspend the GPRS connection with GSM. At least on most equipment. Oct 04 08:22:53 There are some newer ones that can do call and GPRS at the same time. Oct 04 08:22:57 how is it restored ifthe call isn't dropped? Oct 04 08:23:04 With UMTS is should be always possible to have calls and GPRS at the same time. Oct 04 08:23:15 However we have the GPRS suspended property for this. Oct 04 08:23:33 Since essentially it will just suspend your IP connection. And not fully drop it. Oct 04 08:23:38 You just don't get any data over it. Oct 04 08:23:59 I haven't checked the +CLCC output when the data is suspened for a voice call Oct 04 08:24:22 I know it stops, probably stops the hldc completely?? Oct 04 08:24:46 to get the data back you can end the call with at+chld=12 Oct 04 08:25:13 I haven't tried placing voice on hold yet and seeing if data returns Oct 04 08:25:16 I will have to try that Oct 04 08:25:36 Play with it. CDMA is nasty actually. That specification is worse than GSM. Oct 04 08:25:48 okay, I'll see what I can do Oct 04 08:25:51 Other problems will be text mode only SMS. That is just insane. Oct 04 08:26:02 there's a gatchat for dialing data (csd, etc.) calls? Oct 04 08:26:14 that's true, I'm trying to get the +CMT messages decoded Oct 04 08:26:17 gsmdial is an example tool. Oct 04 08:26:18 as well Oct 04 08:26:41 Forget about SMS text mode. That is something doomed for disaster. Oct 04 08:27:04 And oFono core will not allow you to give us text SMS. You need to give oFono PDUs. Oct 04 08:27:09 I don't have much choice :) I"m hoping to switch my phone to linux full time Oct 04 08:27:27 Play with it. Oct 04 08:27:29 what is the cdma tpdu stuff? Oct 04 08:27:38 No idea. Do you get access to it? Oct 04 08:28:06 If you are not talking the secret Qualcomm binary protocol, you are lost most likely. Their AT commands are limited on CDMA. Oct 04 08:28:09 I'm not sure yet, I simple get a +CMT a number and a long hex string in the cat /dev/smd0& output Oct 04 08:28:24 no, since I'm on a windows mobile device that doesn't support qmi Oct 04 08:28:38 oh, I see Oct 04 08:28:40 Okay. Maybe then you get lucky. Oct 04 08:28:42 that's needed for text? Oct 04 08:29:03 If you get a hex string then it is either some weird character set or PDUs. Oct 04 08:29:05 it would be if the text was stored in NVFS on the modem, but I don't think it is on this one Oct 04 08:30:17 it looks like 214 hex digits, I think that's constant Oct 04 08:30:31 or it could be because they are all from my voice mail service (callwave) Oct 04 08:31:39 this one is 236 Oct 04 08:33:19 00000210020204011891400601FC..then maybe a sequence Oct 04 08:44:11 is it possible this is 6bit encoded? Oct 04 08:50:49 here's AAAA sent to myself from my google voice account Oct 04 08:50:50 +CMT: ,40 Oct 04 08:50:51 00000210020207028CE95DCC65800601FC08150003168D3001061024183060800306101004044847 Oct 04 08:50:58 much shorter Oct 04 09:04:29 okay, our android porting people already have it working :) Oct 04 09:05:29 holtmann: http://gitorious.org/linux-on-qualcomm-s-msm/android-htc-ril/blobs/029f9d08a90e9403f2bd4ef8f4aa2c4c0d373d42/sms.c Oct 04 09:05:32 it's gpl Oct 04 09:07:21 If is is PDU mode, then oFono should just work. Oct 04 09:07:47 Just a few different offsets here and there. Oct 04 09:08:12 +CMT: is standard? Oct 04 09:08:24 Feel free to create some example data of CDMA SMS PDUs and stuff them into our unit test directory. Oct 04 09:08:39 sure, i just need to figure out how to sanitize them Oct 04 09:08:55 We are supporting +CMT already. Oct 04 09:09:33 ok, the 40 means 40 bytes? Oct 04 09:09:34 oFono supports +CMT and +CMTI. Oct 04 09:09:55 I'll just watch on dbus and see if it's decoding them Oct 04 09:10:19 CDMA PDUs are bit different if I recall. Not much actually in the end. Oct 04 09:13:32 tmzt_: you need to provide decoding of CDMA SMS PDUs.More information on the CDMA SMS can be found in the 3GPP2 C.S0015-B specification Oct 04 09:15:16 ok Oct 04 09:15:18 thank you Oct 04 09:15:39 any idea on the other issue, if 1xdata call management is standarized there somewhere? Oct 04 09:19:51 May be you have already gone through the specifications in the following link: http://www.3gpp2.org/public_html/specs/alltsgscfm.cfm Oct 04 09:27:22 tmzt_: Hopefully, above link will have information on what you are looking for Oct 04 10:09:49 tmzt_: CDMA does not have an official AT command spec Oct 04 10:10:18 tmzt_: From the (very) few modems I've seen, they simply re-use the GSM 27.005 SMS AT command spec, hence +CMT notifications Oct 04 10:10:37 However, some do not support PDU mode at all Oct 04 10:11:36 denkenz: did you have a chance to look at my sms watch patch? Oct 04 10:11:48 I was more trying to figure out what that hex string is Oct 04 10:11:56 akiniemi: Not really, I'm on vacation ;) Oct 04 10:12:09 it looks like dzo's code will work but I'm too tired to adapt it as a standalone Oct 04 10:12:31 denkenz: hehe, I call that a "vacation" ;) Oct 04 10:12:33 holtmann said ofono's mode might already support it if the offsets are changed Oct 04 10:12:44 denkenz: but no hurry, they can wait. Oct 04 10:13:04 denkenz: mind if I fill up smart-messaging.c? Oct 04 10:13:09 /bin/bash: include/ofono: No such file or directory Oct 04 10:13:09 make[1]: *** [include/ofono/log.h] Error 127 Oct 04 10:13:14 tmzt_: Not likely, the CDMA PDU is different Oct 04 10:13:31 I thought i did something, but this is git with ./bootstrap-configure run Oct 04 10:14:16 akiniemi: Remember, I am a day ahead ;) Oct 04 10:14:22 denkenz: http://gitorious.org/linux-on-qualcomm-s-msm/android-htc-ril/blobs/029f9d08a90e9403f2bd4ef8f4aa2c4c0d373d42/sms.c Oct 04 10:14:34 akiniemi: But yeah, feel free to fill in smart-messaging.c Oct 04 10:22:40 why could it possibly be tring to *run* include/ofono Oct 04 10:23:08 There's a script that creates include/*.h Oct 04 10:23:19 rather include/ofono/*.h Oct 04 10:23:27 Maybe the magic doesn't work on your shell Oct 04 10:23:32 holtmann would know more Oct 04 10:23:46 and when I create the directory it complains it's a directory Oct 04 10:23:52 let me check my shell Oct 04 10:24:06 -bash Oct 04 10:24:30 /bin/bash Oct 04 10:24:30 make Oct 04 10:24:32 same thing Oct 04 10:24:57 this is a clean clone from git Oct 04 10:25:23 holtmann wrote the build system, so he knows more Oct 04 10:25:35 ok Oct 04 10:25:37 I don't grok autofoo or shell :) Oct 04 10:25:50 no, I don't really eitehr Oct 04 10:58:48 Are the default_en_list_no_sim contents supposed to be removed/not shown, when SIM is present? Oct 04 10:59:14 Seems to behave badly with the isigen driver Oct 04 10:59:14 no, those numbers are guaranteed to always work Oct 04 10:59:46 Since isigen doesn't support reading ens from SIM Oct 04 11:00:11 It means 118 is shown as an emergency number, although it's directory service here Oct 04 11:01:56 The spec says that certain numbers are always available Oct 04 11:02:09 Perhaps we got the list wrong, or Finland is simply weird Oct 04 11:03:03 Certain numbers, yes, but not the no_sim list if there is a SIM? Oct 04 11:04:36 Let me try to find you the quote Oct 04 11:09:05 Ok, I see whats happening Oct 04 11:09:21 Your don't support ECC at all, so the new list is never set Oct 04 11:12:06 We might need to add ECC support as a reverse-abstraction hack to the sim driver :/ Oct 04 11:12:40 Although, it's not a huge issue, inconvenience at most -- it's not like dialing 118 will actually make an E911 call her Oct 04 11:12:46 s/her/here Oct 04 11:14:22 akiniemi: The problem is that the EFecc has the same id but different file format between 31.102 and 51.011 Oct 04 11:14:45 denkenz: oh? Oct 04 11:14:58 akiniemi: Yes, transparent vs linear fixed Oct 04 11:15:14 hence we try both Oct 04 11:15:55 So the best way to solve this is to somehow detect that both failed Oct 04 11:16:08 If so, then set the default list, namely 911 and 112 Oct 04 11:31:37 akiniemi: emergency calls do not have numbers.. but of course, the modem decides when to dial an emergency call Oct 04 11:53:27 pessi: right, that's why it doesn't really hurt to list 118 as emergency number Oct 04 11:53:46 pessi: it still won't route to the emergency center, even when dialed through oFono Oct 04 11:55:36 except that the emergency number list is the only indication we have that it is indeed an emergency number Oct 04 11:55:55 Otherwise we can invoke emergency call handling procedures in error Oct 04 11:56:38 denkenz: that it right Oct 04 11:56:59 So maybe worth implementing EFecc support for isimodem, then Oct 04 11:57:46 We need to fix this in the core, but having isimodem synthesize EFecc might be a good idea regardless Oct 04 11:57:54 nod Oct 04 12:30:31 akiniemi: do you have any idea why this SUBSYSTEMNS=="usb" does not work in Lucid? Oct 04 12:59:36 pessi: no idea Oct 04 13:00:05 pessi: to work around it, I have /etc/ofono/modem.conf Oct 04 13:00:33 pessi: Works for me on Fedora 13. Oct 04 13:01:00 akiniemi: I am pretty close in ditching modemconf plugin right now. Oct 04 13:01:21 Especially since we don't need it for IFX. I just have to convert Calypso for Freerunner and we are done with it. Oct 04 13:01:28 Then we get a phonesim.conf and that is it. Oct 04 13:02:27 holtmann: can you keep it around until udev unbreaks itself in Ubuntu, perhaps with meerkat? Oct 04 13:02:55 I need another month until this is all cleaned up and test, but then it goes away. Just a heads up here. Oct 04 13:03:43 A month might suffice. Someone needs to test this stuff on meerkat, though. Oct 04 13:03:54 pessi: didn't you have some work around in udev rules, too? Oct 04 13:04:50 sure Oct 04 13:05:01 still, i don't understand how they work ;) Oct 04 13:12:39 akiniemi: I sent the three patches against udev Oct 04 13:13:24 it seems to me that enumerating uses the results from rules that were in effect when the device was originally inserted Oct 04 13:14:02 but I don't understand why the isigen stuff gets lost in Lucid Oct 04 13:16:01 but it does not if the rules use vendah id instead of subsystems Oct 04 13:17:08 susbsytems-based driver gets lost even if 1) usb subsystem is included in enumeration 2) enumeration is not filtered by subsystems at all Oct 04 13:35:08 akiniemi: what do you want to test in maverick? Oct 04 13:40:45 kvalo: isigen udev rules Oct 04 13:41:34 if you first connect n900 and then start ofono, does it get detected by ofono 0.31? Oct 04 13:41:57 pessi: ok, I can test that later today Oct 04 13:42:15 kvalo: thanks Oct 04 13:48:21 denkenz: do you have any preferences in online/offline watch/notification? expose modem_state, or only online/offline? Oct 04 13:57:26 akiniemi, pessi: ok, I'm testing with maverick right now Oct 04 13:59:47 this is a daily build from yesterday Oct 04 14:00:23 modem.conf has everything commented out Oct 04 14:02:08 pessi: so this is without pc suite mode? ie. I just plug in the cable and don't do anything else on n900? Oct 04 14:03:04 should be Oct 04 14:04:00 works for me, tested twice Oct 04 14:04:13 oh yeah, but this isn't the maverick kernel Oct 04 14:04:49 when I reboot I'll try with the default maverick kernel Oct 04 14:06:27 pessi: odd, this way the connection is much faster Oct 04 14:07:31 pessi: do you remember the 20s stalls I had with n900 on pc suite mode. I don't see them now Oct 04 14:07:50 earlier apt-get update was impossible, now there are no problems Oct 04 14:12:42 20 sec? Oct 04 14:12:47 crap, nevermind. I saw the stall again Oct 04 14:13:19 Mika has added the GPRS suspend and resume code Oct 04 14:13:45 when we discussed last time, the theory was that it was due to roaming between 2g and 3g Oct 04 14:15:05 pessi: I see it like this: http://paste.ubuntu.com/505761/ Oct 04 16:48:12 is there any ofono client already available? i mean, some kind of testing GUI? Oct 04 19:05:55 holtmann: any clue why your build system would try to *execute* a directory? Oct 04 19:06:09 root@rhobuntu:~/src/ofono# make Oct 04 19:06:09 make --no-print-directory all-am Oct 04 19:06:09 include/ofono Oct 04 19:06:09 /bin/bash: include/ofono: No such file or directory Oct 04 19:06:10 make[1]: *** [include/ofono/log.h] Error 127 Oct 04 19:06:19 this is clean git head Oct 04 19:08:43 I'm cloning out again, but I don't think that's the issue Oct 04 19:14:39 kvalo: Does the Technology property change when you see such stalls? Oct 04 19:16:51 akiniemi: I didn't check that. but the icon in n900 is always 3g Oct 04 19:17:20 akiniemi: I'll follow that property the next time I try n900 Oct 04 19:29:02 kvalo: ok, I'll try to reproduce tomorrow at work Oct 04 19:29:50 kvalo: might be that the N900 userspace is doing something weird Oct 04 19:30:01 kvalo: was the device otherwise idle? Oct 04 19:36:10 akiniemi: yes. but I have sshd and other goodies running on it. Oct 04 19:36:53 akiniemi: I think the location depends here. at the city I didn't see any problems. nor when I have forced it to 2G Oct 04 19:37:06 from n900 ui, I mean **** ENDING LOGGING AT Tue Oct 05 02:59:57 2010