**** BEGIN LOGGING AT Fri May 21 02:59:57 2010 May 21 05:43:34 kvalo: If you can't send commands then it most likely HUPed May 21 05:44:36 kvalo: Hack into read_watcher_destroy_notify in GAtIO May 21 05:48:55 denkenz: ok, thanks May 21 12:09:58 denkenz: yeah, huawei also hangsup for me :( May 21 13:12:31 kvalo: Then we have to figure out how to reopen the device May 21 13:15:43 denkenz: yeah, I was reading the logs from may 1st when you and holtmann discussed about that. not an easy task :( May 21 13:18:16 especially when we still haven't agreed May 21 13:19:01 On the Huawei E160G it is a bit easier, since it has 3 ttys and one can be allocated exclusively to ppp May 21 13:19:09 I understand that your device only has 2? May 21 13:19:25 mine has three also May 21 13:19:42 I have been using thirst and third only May 21 13:19:53 *first May 21 13:20:32 I have no idea what the second tty does May 21 13:22:32 does the 2nd tty accept AT commands? if not its probably the Diagnostics port May 21 13:25:31 the 2nd tty does not accept any commands. I tried atz, for example May 21 13:28:00 yeah, sounds like a QCDM port May 21 13:29:20 Can we check whether the HUP is faked on that port? e.g. if we ignore it for a second can we start writing to it again? May 21 13:29:51 Re-initing all atom drivers' gatchats is not going to be fun otherwise May 21 13:29:54 sure, I'll try that May 21 13:30:55 running low on battery, be back in a bit May 21 13:31:00 ok May 21 13:36:09 denkenz: I will use gsmdial to test for the fake hup, easier for me. is that ok? May 21 13:36:18 that's fine May 21 13:36:27 cool May 21 13:36:30 I was planning to do it myself, but never enough time :( May 21 13:36:38 I know the feeling :) May 21 13:41:04 I know that if you ignore the HUP and try to read from the fd anyway it returns EBUSY May 21 13:41:22 so in theory the socket is still valid.. May 21 13:56:40 denkenz: so how should I ignore HUPs? I didn't quickly figure out how to do it. the destroy functions got called anyway May 21 13:57:12 denkenz: and if I just comment out the code in gatio.c destroy functions, gsmdial fails May 21 13:58:36 inside giochannel watch, return TRUE May 21 13:58:45 in case of HUP/error it returns FALSE May 21 14:01:54 ok, found it now May 21 14:07:09 hmm, I don't know if I did this right, but I don't get a reply May 21 14:07:16 but I need to test more May 21 14:07:33 and I'll send a patch to you so you know what I really did May 21 14:42:05 denkenz: I'll call it a day now. weather is too good for working friday evenings :) I'll continue again on monday May 21 15:10:17 kvalo: Test with the test-qcdm test tool from the ofono source ;) May 21 15:32:10 holtmann: oh, that's cool. I'll test it next week **** ENDING LOGGING AT Sat May 22 02:59:57 2010