**** BEGIN LOGGING AT Wed Nov 18 03:00:03 2009 Nov 18 03:04:04 hehe, nod Nov 18 03:31:24 denkenz, AT*ENAP should be mbm specific command? I can not find the spec to describe it? Can you show me? Nov 18 03:31:46 http://www.natisbad.org/E4300/Dell_Wireless_5530_AT_cmd_ref.html Nov 18 03:32:57 also http://cgit.freedesktop.org/ModemManager/ModemManager/ Nov 18 03:33:28 thanks Nov 18 10:54:00 martin_lab: Still there? Nov 18 10:55:15 The Active property doesn't work correctly. Even after I activated the context successfully, it still shows Active=0. If the signal that gets send out is wrong. Nov 18 10:55:22 Also the signal for the Settings is never send. Nov 18 10:55:55 denkenz: My F3607GW doesn't crash on GATT. Nov 18 10:56:32 However sometimes COPS returns an ERROR that we might not handle correctly. And even GATT can error. Mainly that has something to do with a broken antenna. Trying that card without antenna is bad idea. Nov 18 14:55:42 holtmann: Hmm, I might have broken Active being set properly Nov 18 14:56:24 I didn't look into so code yet. So please fix it. However besides that, my card seems to work. Nov 18 15:01:04 holtmann: I just got my MD300 Nov 18 15:01:22 And my whole USB EHCI controller just died on me. Nov 18 15:02:12 So it seems the card worked once and now it is crashing all the time. Nov 18 15:02:27 And taking the USB bus with it. Nov 18 15:03:05 heh Nov 18 15:03:23 Mine takes the kernel down Nov 18 15:03:29 Good thing I'm doing it in a VM Nov 18 15:13:45 AT+CFUN=1 Nov 18 15:13:45 OK Nov 18 15:13:45 AT+COPS=0 Nov 18 15:13:45 OK Nov 18 15:13:45 cu: Got hangup signal Nov 18 15:13:50 And now crash. Nov 18 15:14:08 It doesn't crash the kernel, but that is a different bug. That got fixed. Nov 18 15:15:07 Yah, I get the same when running configuration 3 Nov 18 15:15:14 It crashes right after COPS Nov 18 15:15:15 denkenz: I am not kidding here. I had this working perfectly fine. And then out of a sudden after 3 hours or so. I check back on it and crash. And since then always. Nov 18 15:15:41 holtmann: I think you just got lucky Nov 18 15:15:49 I managed to make mine work once too Nov 18 15:15:52 :) Nov 18 15:15:52 I only have two USB configurations on mine. Nov 18 15:16:15 pessi: Vendor=0bdb ProdID=1906 Rev= 0.00 Nov 18 15:16:15 Sage: Manufacturer=Û Nov 18 15:16:15 F3607gw Nov 18 15:16:15 Sage: Product=F3607gw Nov 18 15:16:20 And this odd ball of course ;) Nov 18 15:16:30 Nevermind, configuration=2 Nov 18 15:16:46 Forgot the 3607 doesn't have the USB memory Stick configuration Nov 18 15:17:33 I am using configuration 1. Not sure what is really the difference. Nov 18 15:17:45 Maybe this is an antenna thing after all. Nov 18 15:19:42 AT+CFUN=1 Nov 18 15:19:42 OK Nov 18 15:19:42 +CGREG: 2 Nov 18 15:19:42 +CREG: 2 Nov 18 15:19:47 And then crash. Nov 18 15:20:26 yep, my symptoms exactly Nov 18 15:21:16 This needs a firmware update or so. This is fucked. Nov 18 15:22:24 Does the something 300 USB stick work any better? Nov 18 15:23:58 trying now Nov 18 15:32:06 sigh, that one requires changes to udev.c Nov 18 15:34:38 Ok, pushed a fix so the mbm driver actually detects the md300 Nov 18 15:35:03 I'm still curious why MD300 reports 4 ACM ports on lsusb but only 2 show up Nov 18 15:52:08 sigh, MBM never sends a CGREG Nov 18 15:53:15 It does for me on the Ericsson cards, but the SE one seems to be pretty old. Nov 18 15:53:24 Or maybe you have to use a different TTY again. Nov 18 15:54:05 Nah, I think it just simply never sends it Nov 18 15:55:24 I have to put in cgreg query after sending the attach/detach Nov 18 15:55:53 Also, the MD300 doesn't support E2NAP Nov 18 15:56:07 You actually have to query ENAP every second or so Nov 18 15:58:06 Lol, it also sends +CREG: 4,"cellid","location" Nov 18 15:58:20 So registration state: Uknown, and btw here's a ci and lac too Nov 18 16:05:45 Crap. That is just stupid. Nov 18 16:06:42 Seems the only half-proper card is the one that I have in my X200. Who actually does QA for this firmware? These bugs are obvious :( Nov 18 16:12:15 The Option 401 is actually not half bad Nov 18 16:12:56 I meant MBM based cards. Nov 18 16:28:11 holtmann: btw, we still need to figure out what to do with _disable Nov 18 16:28:28 holtmann: Right now all of the drivers do something like chat_send(cfun=0) Nov 18 16:28:35 then shutdown and unref it right afterward Nov 18 16:29:19 None of them actually bring the device down Nov 18 16:36:32 holtmann: Pushed out the latest changes Nov 18 16:36:44 I can now attach / activate / receive settings properly Nov 18 16:50:23 ofonod[29353]: > AT+CGREG=2\r Nov 18 16:50:23 ofonod[29353]: < \r\nOK\r\n Nov 18 16:50:23 ofonod[29353]: > AT+CGAUTO=0\r Nov 18 16:50:23 ofonod[29353]: < \r\nERROR\r\n\r\n*EMWI: 1,0\r\n Nov 18 16:50:23 ofonod[29353]: > AT+CGEREP=2,1\r Nov 18 16:50:23 ofonod[29353]: < \r\nERROR\r\n Nov 18 16:50:26 A lot of errors. Nov 18 16:50:49 dbus.exceptions.DBusException: org.ofono.Error.InProgress: Operation already in progress Nov 18 16:50:57 When trying to activate context. Nov 18 16:52:00 CGAUTO is almost never supported Nov 18 16:52:04 feel free to ignore Nov 18 16:52:16 CGEREP is not supported on MBM devices Nov 18 16:52:35 You have to fix this Active stuff. Since that is broken. Nov 18 16:53:09 ofonod[29353]: < \r\n*ESTKSMS: ,0,0\r\n Nov 18 16:53:09 ofonod[29353]: < \r\n*ESTKDISP: 1,1,0,Ihre Nr 0171-xxxxxxx Nov 18 16:53:13 How'd you trigger that Nov 18 16:53:23 Funny stuff. Nov 18 16:53:27 are you sure you're not trying to activate twice? Nov 18 16:53:29 Seems to be some network related crap. Nov 18 16:54:05 ESTK is SimToolkit Nov 18 16:54:49 Aha. Funny stuff. Nov 18 16:56:20 holtmann: Activate just works for me Nov 18 16:57:30 I've done it three times now :P Nov 18 17:00:19 Now it works, but there are some race conditions somewhere. Nov 18 17:00:37 run dbus-monitor in the background Nov 18 17:00:45 Sometimes if you're running d-feet it sends the event twice Nov 18 17:00:54 So the second one will get a busy error Nov 18 17:01:25 I am not using d-feet. Nov 18 17:01:51 Also you need to query with AT*ENAP? what the current state of it is. Nov 18 17:02:21 After a crash or something the GRPS context might be still connected. Nov 18 17:02:30 yeah, I know Nov 18 17:02:43 First we need to make power=false work Nov 18 17:03:44 And ideally we should actually CFUN=4 then CFUN=1 the modems Nov 18 17:04:42 And the AT+COPS=? needs to be reduced or put on a second TTY. It causes problems. We need to be smarter here. Nov 18 17:04:52 I'm willing to re-sync the state for voicecalls, since that is a good idea Nov 18 17:05:04 But once we have multiple contexts, forget about syncing those Nov 18 17:05:14 I'm getting rid of COPS scanning Nov 18 17:05:26 I'll introduce ProposeScan method Nov 18 17:07:01 Okay. Nov 18 17:07:18 Did you find a method that would retrieve the current IP settings for that context and avoid the useless DHCP stuff. Nov 18 17:07:26 Not on MBM Nov 18 17:07:34 This works on Option though Nov 18 17:07:47 Also I think we have to make oFono responsible for bringing the interface up when Powered=true. Nov 18 17:08:11 what interface? Nov 18 17:08:16 usb0. Nov 18 17:08:27 Or in future kernels it will be wwan0. Nov 18 17:08:28 We don't get the interface until Active=True Nov 18 17:08:54 Okay, then we have to bring it up before sending Settings information. Nov 18 17:09:12 That should be fine Nov 18 17:09:20 Care to write the necessary magic? Nov 18 17:09:40 There is an inet_ifup and inet_ifdown in ConnMan that you can copy. Nov 18 17:10:52 and we should also bring it down on activated=False Nov 18 17:10:56 correct? Nov 18 17:11:16 Though ideally it should be the plugin doing it Nov 18 17:11:24 Since something like ppp might not need this Nov 18 17:13:39 Yes. The plugin needs to do that. Nov 18 17:14:34 Have to be careful though Nov 18 17:14:54 Actually not really. We always get a network interface. We can always just bring it up. Nov 18 17:15:10 This is standard network device stuff. So that is just fine. Nov 18 17:15:16 If you want to bring it down though Nov 18 17:15:25 There's a few paths to do so Nov 18 17:15:33 You lose netreg for instance, or user tells you to Nov 18 17:15:58 Anyway, I figured out why you got a busy on setting 'Active' Nov 18 17:16:15 if the modem is in the process of attaching, we'll tell you to bugger off Nov 18 17:16:17 You send out Settings = {} and then bring the interface down. That is just fine. No need to push this into every single plugin. Nov 18 17:16:36 We need better error codes. Nov 18 17:17:12 Well, operation is in progress actually Nov 18 17:17:36 But I can introduce a new error return like AttachInProgress Nov 18 17:18:18 Sounds better. Nov 18 17:18:37 And we need the up/down stuff since LOWER_UP is broken with this card. Crap. Nov 18 17:19:58 Just add the necessary magic to pri_reset_context_settings & pri_update_context_settings Nov 18 17:20:37 I'll make sure the reset is called if we lose netreg Nov 18 17:24:37 holtmann: I propose we change all foo_disable to send a CFUN=0, return -EINPROGRESS and unref chat in the callback Nov 18 17:24:55 We'll not cleanup properly on a Ctrl-C, but at least SetProperty(Powered,False) will work Nov 18 17:25:34 The proper fix is to run a nested event loop somewhere Nov 18 17:29:36 CFUN=0 is risky. For my MBM hardware in my laptop it is similar to RFKILL and takes the device off the bus. Nov 18 17:30:10 But in general unrefing the chat in the callback seems fine to me. Nov 18 17:36:36 Well, CFUN=4 for devices that need it Nov 18 17:36:47 Just right now SetPowered basically doesn't work, it times out Nov 18 17:36:56 Since we unref chat right after sending the CFUN Nov 18 17:37:13 So neither SetPowered nor Ctrl-C actually work as desired Nov 18 19:17:26 holtmann: Do you wanna have interface specific errors? Nov 18 19:17:35 e.g. org.ofono.DataConnectionManager.NotAttach Nov 18 19:17:41 or .AttachInProgress Nov 18 19:17:46 No. org.ofono.Error Nov 18 19:18:02 Otherwise you go mad with the bindings. Nov 18 19:18:12 nod Nov 18 19:18:14 Same as we do with ConnMan and BlueZ. Nov 18 19:26:01 holtmann: Ok, I pushed out the improved error handling Nov 18 19:26:05 feel free to comment Nov 18 19:27:58 Looks fine. Nov 18 19:28:56 So theoretically we we should now have a stable context activation Nov 18 19:29:03 I've removed operator scanning Nov 18 19:29:20 Someone just needs to test it with a real GPRS plan Nov 18 19:31:04 FYI: http://www.sonyericsson.com/cws/download/1/356/327/1207040438/PC300_MD300_GSG_AT_Command_R2a_EN.pdf Nov 18 19:31:11 That's the AT command manual for the MD300 Nov 18 19:38:47 What do you mean by real GPRS plan? Nov 18 19:42:18 My GPRS is some weird WAP bs Nov 18 19:42:58 Still haven't gotten around to getting a real plan, since it requires a contract Nov 18 19:43:21 will do it after I come back Nov 18 20:14:30 I have a real plan. Let me see how far I get. I can also test the HSO stuff once you put it in. Nov 18 20:14:46 And Martin can work on the ConnMan plugin now. Nov 18 20:20:32 nod, the HSO stuff might get tricky since the manual says that unsolicited notifications only arrive on the Control port Nov 18 20:20:37 and we need the OWANDATA one Nov 18 20:20:40 But we see Nov 18 20:20:46 Should have it by end of day today Nov 18 20:22:32 I just love GAtChat though Nov 18 20:22:37 It makes things so simple Nov 18 20:23:16 Tell me about it. Good that I don't have to fiddle around with regex. Nov 18 20:23:45 Whoever thought that would be a good idea needs to go back to programming 101 lecture. Nov 18 21:21:19 holtmann: Option does not report the netmask Nov 18 21:21:37 holtmann: Is there a sane value you can calulcate or a sane default? Nov 18 21:21:39 It is point to point anyway. So who cares. Nov 18 21:21:50 holtmann: So I leave it blank? Nov 18 21:21:56 Yes. Just leave it out. Nov 18 21:22:02 ok Nov 18 21:41:28 same issue with cfun on cdma modems. by the I'm getting a touch pro 2 and won't have this touch pro cdma much longer. the tp2 I'm getting is cdma/gsm model Nov 18 21:47:09 denkenz: you might need a custom driver for that md300, cdc-acm basically guesses what's really a serial port Nov 18 21:50:29 tmzt: Seems to work for me ;) Nov 18 21:51:25 you didn't need the other two ports then? Nov 18 21:52:32 No, the MBM can run on a single port Nov 18 21:52:44 Besides, MD300 actually has 3 USB configurations Nov 18 21:52:51 1 has a single ACM port, the other 2 Nov 18 21:53:16 The 3rd is the weird USB Stick mode Nov 18 21:54:20 for drivers? Nov 18 21:54:33 sure that doesn't require ejecting? Nov 18 21:54:53 nope Nov 18 21:55:01 is it cd or writable, or is it for microsd reader Nov 18 21:55:04 ah Nov 18 21:55:11 Dunno, don't care Nov 18 21:55:21 I think it comes with the software for windows crap Nov 18 21:55:25 right Nov 18 21:55:38 Just switch usb configuration to 3 and you got 4 wdm ports and 2 acm ports Nov 18 21:55:51 wdm? Nov 18 21:56:05 Wireless Device Management Nov 18 21:56:11 They seem to be just another serial port Nov 18 21:56:25 I see Nov 18 22:06:57 holtmann: I've /dev/HS3, but that one's not showing up in udev Nov 18 22:07:23 holtmann: Any idea what could cause that? Nov 18 22:11:38 what is that? Nov 18 22:13:27 ttyHS3, Option ttys Nov 18 22:14:32 ttyHS3 then? Nov 18 22:16:04 yeah, shorthand ;) Nov 18 22:47:03 sigh, Option just needs to be shot Nov 18 22:47:22 Again, their proprietory commands don't format according to AT command standards Nov 18 23:49:09 holtmann: Another stupidity, HSO reports gateway as 0.0.0.0 Nov 19 00:25:24 denkenz: have u received my mail about ciev_callheld=1? Nov 19 00:33:45 zhenhua1: Yes, I haven't thought too much about it since been busy with GPRS Nov 19 00:34:20 zhenhua1: I'm actually fine with polling in that case Nov 19 00:34:35 zhenhua1: However, we do need to have an in_mpty_call variable or something Nov 19 00:34:46 Where we just poll every 2 seconds or so Nov 19 00:36:46 And your point in 1b is actually valid, so it might be better to just poll Nov 19 00:36:52 See how that works out Nov 19 00:37:14 yeah. so i suggest to poll. it's safer than guess Nov 19 00:37:37 but poll it every 2 seconds is really bad. Nov 19 00:37:52 we can check if mpty_call > 2, then poll. Nov 19 00:37:58 otherwise, no need to poll Nov 19 00:38:00 That's what I mean Nov 19 00:38:17 If we go from callheld=1 to callheld=0 or callheld=1 Nov 19 00:38:23 It is likely that we did a merge Nov 19 00:38:38 So we poll and if we have more than 1 held or more than 1 active, we're in mpty Nov 19 00:38:52 At that point just start a regular poll Nov 19 00:39:26 why not add check in CHLD=3 Nov 19 00:39:36 rather callheld=1 -> callheld=0 Nov 19 00:39:43 Because the merge could be done at the gateway Nov 19 00:39:59 yeah. you're right Nov 19 00:41:20 callheld=1->callheld=0 may not safe. Nov 19 00:41:33 too many cases will cause callheld=0. Nov 19 00:42:06 i'd rather count the total call numbers in poll_clcc_cb Nov 19 00:42:19 and then start poll Nov 19 00:42:31 again, that's what I meant ;) Nov 19 00:42:36 like what we did in atmodem. poll_again Nov 19 00:42:53 ok. then we're done. :) Nov 19 00:43:05 nod, its ugly, but the best we can do Nov 19 00:43:16 Unless some manufacturer came up with their own extensions Nov 19 00:44:23 holtmann: Ok, my Option 401 is now working flawlessly ;) Nov 19 00:56:09 Pushed out, if anyone has Option or MBM hardware feel free to test it and report how well it works :) **** ENDING LOGGING AT Thu Nov 19 02:59:56 2009