**** BEGIN LOGGING AT Wed Jun 15 02:59:57 2011 Jun 15 07:34:23 heyho Jun 15 07:34:31 mickey|office: ping Jun 15 10:42:15 hi mickey|daddy mickey|office Jun 15 10:43:41 morning GNUtoo, how are things? Jun 15 10:44:24 you know fsonetworkd, right? Jun 15 10:45:55 here's a route on htcdream because of connmand: Jun 15 10:45:56 http://www.pastie.org/private/xu1gtz1lkxnpbdxkwmma Jun 15 10:46:07 it breaks iliwi I guess Jun 15 10:46:15 because it breaks udhcpc -i wlan0 Jun 15 10:46:28 (too much default routes) Jun 15 10:52:38 GNUtoo: i don't know the state of morphis' fsogsmd/connmand integration, but if this would be working, we could disable (perhaps adding a configuration entry) the custom route creation done by fsogsmd's ppp plugin. Jun 15 10:52:54 ok Jun 15 10:53:02 I should ask morphis then Jun 15 10:53:22 I don't know why is connmand used and what it is used for Jun 15 10:53:46 maybe if I disable pdp would stop working or something like that Jun 15 10:54:14 I'll ask morphis, thanks a lot Jun 15 12:13:55 heyho Jun 15 12:22:52 mickey|office: ping Jun 15 12:22:55 GNUtoo: ping Jun 15 12:23:45 morphis: pong Jun 15 12:24:49 mickey|office: fine, you are available :) Jun 15 12:24:56 yo Jun 15 12:25:06 should we talk in private about the presentation? Jun 15 12:26:39 ok Jun 15 12:26:44 or here :) Jun 15 12:33:53 morphis, pong Jun 15 12:34:14 morphis, I'll ping you again later Jun 15 12:38:49 GNUtoo: ok Jun 15 12:48:07 morphis, for htcdream I've that: http://www.pastie.org/private/xu1gtz1lkxnpbdxkwmma Jun 15 12:51:08 it blocks udhcpc -i wlan0 Jun 15 12:54:46 GNUtoo: whats that? Jun 15 12:56:14 morphis, route command Jun 15 12:56:24 basically connmand causes that Jun 15 12:56:27 yes, but what should it tell me? Jun 15 12:56:54 the question is: how do I ignore some interfaces? Jun 15 12:57:01 I could do it in the init script Jun 15 12:57:04 GNUtoo: you should take a look at this: http://git.freesmartphone.org/?p=linux-2.6.git;a=blob;f=arch/arm/mach-msm/smem_log.c;h=e85e37a5b0945fe97d624552cf209a9c5852df93;hb=refs/heads/palmpixi/master Jun 15 12:57:08 but I'm not sure it's the good solution Jun 15 12:57:18 why you need to ignore them? Jun 15 12:58:40 basically I've that long route table Jun 15 12:58:45 with many default Jun 15 12:58:54 the many default are gone if I kill connmand Jun 15 12:59:10 without killing connmand udhcpc -i wlan0 blocks Jun 15 12:59:18 let me wait for the error message Jun 15 12:59:28 thanks a lot for the log driver Jun 15 13:00:36 I've to wait some minutes Jun 15 13:00:41 like 5 min maybe Jun 15 13:02:04 and it also blocks normal networking trough usbnet Jun 15 13:02:11 I cannot download from the feeds Jun 15 13:02:14 but I can ssh in it Jun 15 13:06:57 morphis, what's the use of fsonetworkd and connmand plugin? Jun 15 13:07:21 handle wifi scanning etc....? Jun 15 13:07:29 and ppp? Jun 15 13:07:37 (the htcdream uses rmnet) Jun 15 13:07:49 (instead of ppp) Jun 15 13:10:10 still not failed(udhcpc...still waiting) Jun 15 13:15:17 GNUtoo: about blocking the interface I have currently no solution Jun 15 13:15:29 killall connmand unblocks it Jun 15 13:15:32 GNUtoo: there is currently one one usecase for fsonetworkd Jun 15 13:15:40 or maybe two Jun 15 13:15:46 ppp and wifi? Jun 15 13:15:47 1. Offering a default route for ppp Jun 15 13:16:01 2. And forwarding a wireless connection to the pc Jun 15 13:16:02 I think we got one with the qmi plugin Jun 15 13:16:19 the fsogsmd ppp plugin provides the route to fsonetworkd Jun 15 13:16:22 which sets it then Jun 15 13:16:39 but if we want to use connman for that we need to rethink the whole process Jun 15 13:16:47 the fsogsmd connman plugin is unfinished Jun 15 13:16:53 ah ? Jun 15 13:16:56 I started working on it but never finished it Jun 15 13:17:00 ok Jun 15 13:17:06 can it be disabled then? Jun 15 13:17:10 at least for htcdream Jun 15 13:17:13 and nexusone Jun 15 13:17:50 it can Jun 15 13:18:06 ok, how because I do not see that: Jun 15 13:18:13 [fsonetwork.connman] Jun 15 13:18:33 in conf/default/fsonetwork.conf Jun 15 13:18:48 maybe it's connman alone that does this mess Jun 15 13:19:05 that is to say, the connman init script Jun 15 13:19:47 ha fsonetworkd wasn't even launched Jun 15 13:20:35 so I guess I've to modify the init script Jun 15 13:20:40 to ignore rmnet* Jun 15 13:20:44 adn dummy* Jun 15 13:21:49 thanks a lot Jun 15 13:22:04 btw I guess you don't want the geeksphone one Jun 15 13:22:23 because I forgot to ask you Jun 15 13:22:30 who should get it between me and you Jun 15 13:23:06 hm Jun 15 13:23:17 I did some more research this night Jun 15 13:23:21 ok Jun 15 13:23:41 and there is a relevant difference between the protocol used in the geeksphone and the version used by Palm Jun 15 13:23:58 ah? Jun 15 13:24:00 the geeksphone uses the rpc calls for dealing with the modem and it's services Jun 15 13:24:17 and Palm uses one service to fake a serial channel to the modem Jun 15 13:24:37 which they use then to submit the messages we know from the Palm Pre Jun 15 13:24:55 (there are other and different message but in general it's the same) Jun 15 13:25:09 as the message also have the same format Jun 15 13:25:15 ok Jun 15 13:25:20 they still have a subsystem id and a group id Jun 15 13:25:29 which is something the geeksphone doesn't use Jun 15 13:25:34 so the main difference is the transport Jun 15 13:25:41 and the nexus s seems to use AT Jun 15 13:25:45 no Jun 15 13:25:51 not the transport Jun 15 13:25:52 nexus S uses AT!!!!!!! Jun 15 13:26:08 the difference is all above the plain rpc stuff Jun 15 13:26:12 GNUtoo: ok Jun 15 13:26:21 then I got something wrong :) Jun 15 13:26:53 geeksphone is using the rpc stuff as you know it from the gps implementation for the geeksphone Jun 15 13:27:07 ij Jun 15 13:27:13 *ok Jun 15 13:27:36 I know there is also audio that needs rpc Jun 15 13:28:56 morphis, for the nexus S, their ril uses AT? Jun 15 13:29:50 ah I remember Jun 15 13:29:57 the palm pre has 2 layers Jun 15 13:30:04 under the msmcomm protocol Jun 15 13:30:10 so here one layer is rpc Jun 15 13:30:17 and there is a higher level Jun 15 13:30:22 and then msmcomm Jun 15 13:31:47 the -I for ignoring interface didn't work Jun 15 13:31:53 it still blocks with that: Jun 15 13:32:05 default * 0.0.0.0 U 0 0 0 usb0 Jun 15 13:32:05 default 192.168.0.200 0.0.0.0 UG 8 0 0 usb0 Jun 15 13:38:32 GNUtoo: google provides the binaries for the telephony part of the nexus s one some site Jun 15 13:38:43 GNUtoo: and that one does not have any reference to the rpc stuff Jun 15 13:38:51 it's a addition to the reference ril Jun 15 13:39:24 ok Jun 15 13:39:31 but I didn't see some AT strings in it Jun 15 13:39:50 only that: Jun 15 13:39:51 Sending AT in ASCII Jun 15 13:40:17 and some small things like Jun 15 13:40:18 ATE0 Jun 15 13:40:27 and Jun 15 13:40:33 ATD#777 Jun 15 13:44:01 thats because it's an addition to the reference ril Jun 15 13:44:04 which uses AT Jun 15 13:46:23 GNUtoo: about who should receive the geeksphone Jun 15 13:46:27 GNUtoo: I don't know Jun 15 13:46:38 it's leviathan's phone, right? Jun 15 13:47:03 yes Jun 15 13:47:07 he got it for free Jun 15 13:47:10 because I had to get it Jun 15 13:47:23 but I tought he would do more work than me Jun 15 13:47:35 if you plan to do more work than me on it Jun 15 13:47:38 you should get it Jun 15 13:48:21 hm Jun 15 13:48:32 I am only interested in it because of the modem protocol Jun 15 13:48:33 nothing else Jun 15 13:48:52 as for more I don't have the time to do any work Jun 15 13:49:03 me too, mainly because of that, but I planed to do a quick port like for the nexus one Jun 15 13:49:10 ok Jun 15 13:49:35 I am even planing to get an old palm pixi to explore their way of using the shared memory interface of the msm Jun 15 13:49:42 how old is the geeksphone? Jun 15 13:50:09 it's like an htcdream, with the same wifi card than the freerunner Jun 15 13:50:19 the screen resolution is inferior tough Jun 15 13:50:31 and it's a pain to do any kernel development with it Jun 15 13:53:01 ok Jun 15 13:53:07 why is it a pain? Jun 15 13:53:11 cause of flashing? Jun 15 13:53:15 yes Jun 15 13:53:22 you have to put the kernel on the microsd Jun 15 13:53:26 and then to flash Jun 15 13:53:28 and to reboot Jun 15 13:53:32 if I understood well Jun 15 13:53:47 baaah Jun 15 13:53:56 sounds not as a simple way Jun 15 13:54:38 GNUtoo: you have to decide, if I can get the device I will only use it to do some research on the modem protocol in implement it in the msmcomm project Jun 15 13:54:48 if I am done with it I will give it to someone else Jun 15 13:55:02 ok Jun 15 13:55:06 but I can't do anything before beginning of august Jun 15 13:55:26 if I get it now, how many help will I get from you? Jun 15 13:55:34 do you think it's doable? Jun 15 13:55:43 (if I get it) Jun 15 13:57:52 the main problem is, you have to dive very deep into the internals and assembler bits Jun 15 13:58:03 nothing is obvious and it's painful Jun 15 13:58:10 then you should get it Jun 15 13:58:17 I don't want to do taht Jun 15 13:58:20 leviathan, ping Jun 15 13:58:24 as the payload is different from what I know from the pre phones Jun 15 13:58:43 so the sounds like the same but in details it's 100% different Jun 15 13:59:28 ah ok Jun 15 13:59:47 the phone is rooted by default? Jun 15 13:59:51 yes Jun 15 14:00:18 as what should be nice is that we have basic support for the plain-rpc variant of the protocol in msmcomm Jun 15 14:00:29 we need to make it suitable for every device anyway Jun 15 14:00:37 yes Jun 15 14:00:39 as the details seems to differ between the devices itself Jun 15 14:00:55 but I hope we can get it Jun 15 14:01:11 that what I even want to kick off as a separate project within FSO at the end of this year Jun 15 14:01:19 as the rpc stuff we need for the future Jun 15 14:01:25 as more and more devices has it Jun 15 14:01:27 yes for newer palm devices Jun 15 14:01:33 at least Jun 15 14:01:38 and more and more devices has a qualcomm chip in it Jun 15 14:01:55 so they will all use the share memory Jun 15 14:01:59 indeed Jun 15 14:02:01 s/share/shared/ Jun 15 14:02:01 morphis meant: so they will all use the shared memory Jun 15 14:02:27 so it should be very nice if we have a solid base to support only the basic feature of every msm phone in the future Jun 15 14:02:39 thats the plan I have currently for msmcomm Jun 15 14:02:50 it needs more documentation Jun 15 14:02:52 etc. Jun 15 14:03:31 and maybe qualcomm will open source everything in the end :) Jun 15 14:05:15 yes possible Jun 15 14:05:39 how do I remove that route: Jun 15 14:05:49 default * 0.0.0.0 U 0 0 0 usb0 Jun 15 14:06:40 (Destination Gateway Genmask Flags Metric Ref Use Iface) Jun 15 14:07:00 I tried many commands Jun 15 14:07:02 route del default Jun 15 14:07:05 didn't work Jun 15 14:07:12 neither most of the thing I tried Jun 15 14:08:41 hm, route del is what I would suggest Jun 15 14:08:52 try to set another route as default Jun 15 14:08:59 and then remove the one Jun 15 14:09:31 ok Jun 15 14:10:26 didn't work Jun 15 14:10:33 this connman thing is kind of hard Jun 15 14:12:12 aha connman re-creates it Jun 15 14:13:50 route del -net 0.0.0.0 no? Jun 15 14:14:57 jeepingben, I think the route is re-created by connmand Jun 15 14:15:15 but I'll try Jun 15 14:23:25 sigh connman Jun 15 14:23:26 sigh Jun 15 14:23:57 I think it's not even related to the routes Jun 15 14:24:12 ah maybe Jun 15 15:06:18 GRRR when I udhcpc -i wlan0 connman tries to do that: Jun 15 15:06:31 dummy0 {add} route 0.0.0.0 gw 0.0.0.0 scope 253 Jun 15 15:06:39 dummy0 {del} route 0.0.0.0 gw 0.0.0.0 scope 253 Jun 15 15:06:42 and re-add Jun 15 15:06:45 and re-del Jun 15 15:06:46 etc... Jun 15 15:06:56 and I even made a python script Jun 15 15:07:06 for disabling the handle of the interfaces Jun 15 15:07:31 http://pastie.org/2072671 Jun 15 15:27:20 as a quickfix, i'd just start it with the appropriate -I options Jun 15 15:27:32 or -i Jun 15 15:29:59 mickey|daddy, I was doing -I on the wrongs interfaces Jun 15 15:30:02 I forgott wlan0 Jun 15 15:30:08 so I've a big question Jun 15 15:30:22 what should we do with reguards to connman because I was told that: Jun 15 15:30:35 If that interface is under ConnMan control, then you are out of luck with running anything manuallay. Jun 15 15:30:43 However connmand -I wlan0 will make it stay away from that interface. Jun 15 15:36:11 so iliwi uses udhcpc -i wlan0, right? Jun 15 15:36:12 so we should rewrite,replace,or adapt iliwi? Jun 15 15:36:49 because ignoring all interfaces is not a good options Jun 15 15:36:53 it makes connman useless Jun 15 15:37:05 maybe I should try ignoring just usb0 and wifi Jun 15 15:37:20 anyway rmnet doesn't need connmand Jun 15 15:37:30 and we don't even activate fsonetworkd that is unfinished Jun 15 15:37:41 so we should get rid of connmad, right? Jun 15 15:39:18 it depends on whether you want a quickfix or a midtime solution Jun 15 15:39:32 we do want connman to handle our interfaces sooner or later Jun 15 15:40:10 i think it's the best option we have, esp. wrt. to its integration with bluez and ofono/fsogsmd Jun 15 15:40:21 but obviously there are not many UIs yet for it Jun 15 15:41:36 I'm ok with the quick fix Jun 15 15:41:43 else we have to replace the UIs Jun 15 15:41:46 who would do that? Jun 15 15:41:56 there was a connman based UI long time ago Jun 15 15:42:02 what was its name? Jun 15 15:42:08 then the quick fix is to make it ignore the interfaces we have other tools for Jun 15 15:42:32 hmm, i don't recall Jun 15 15:43:05 no that doesn't work either Jun 15 15:43:13 we need to migrate everything to connman Jun 15 15:43:26 I made it ignore all the interfaces Jun 15 15:43:35 but still it blocks Jun 15 15:43:38 with the default route Jun 15 15:43:44 you cannot touch the route Jun 15 15:43:46 yes, if we really want to leverage its capabilities, that's the way to go Jun 15 15:43:46 that's why morphis did the plugin Jun 15 15:43:47 it re-creates it Jun 15 15:43:49 etc... Jun 15 15:44:02 morphis didn't finish his work Jun 15 15:44:05 well, if you make it ignore all interfaces, you might as well not start it ;) Jun 15 15:44:27 yes that means not shipping it Jun 15 15:44:32 since there is an init script Jun 15 15:44:36 fsonetworkd is not used Jun 15 15:44:43 the init script is used to start it Jun 15 15:44:54 get rid of the script Jun 15 15:45:03 those who want connman can start it via dbus Jun 15 15:45:26 nijibbect Jun 15 15:45:29 oops Jun 15 15:45:32 mokonnect Jun 15 15:45:38 right, that was it Jun 15 15:47:16 so I've to work on that instead of doing more important things Jun 15 15:47:31 I tought to quickely fix the htcdream's wifi problem Jun 15 17:22:23 mickey|daddy, mokoconnect is not great at all Jun 15 17:22:44 it doesn't work well on dream, it kills connmand and restarts it and the GUI is not great Jun 15 17:22:52 I'll try it on the n900 Jun 15 17:23:02 so we should rewrite something I guess Jun 15 17:23:09 but I really don't want to do it Jun 15 17:23:32 I should concentrate on *important* things Jun 15 17:23:40 not on writting a connection GUI Jun 15 17:24:19 for instance: Jun 15 17:24:32 *better audio quality within the n900 calls Jun 15 17:24:38 *the isi modem Jun 15 17:24:44 for instance make 3g work Jun 15 17:24:48 or make sms work Jun 15 17:26:56 yep Jun 15 17:26:56 i agree Jun 15 17:27:09 when we provide the functionality, the UIs will come along Jun 15 17:27:16 it doesn#t work the other way round :) Jun 15 17:37:27 basically it works(iliwi) on n900 because Jun 15 17:37:36 *usb0 got a problem and didn't come up Jun 15 17:37:50 *phonet0 is not ethernet Jun 15 17:37:59 so no other interface than wlan0 Jun 15 17:38:01 and it works Jun 15 17:38:31 on om-gta02 the wlan0 appear Jun 15 17:39:17 and I was clearly said that you cannot manage the interface and use it manually Jun 15 17:42:23 GNUtoo: you already talked to leviathan about the geeksphone? Jun 15 17:42:35 I keep pinging him Jun 15 17:42:40 and he doesn't respond Jun 15 17:42:42 ping leviathan Jun 15 17:46:41 ah ok Jun 15 17:47:12 I off for today Jun 15 17:47:14 bye Jun 15 17:53:03 mickey|daddy, I finally solved the connman mistery: Jun 15 17:53:18 DAEMON="/usr/sbin/connmand -I usb0,wlan0,dummy0,rmnet0,rmnet1,rmnet2 -r" Jun 15 17:53:23 before I did: Jun 15 17:53:29 -I usb0 -I wlan0 etc... Jun 15 17:54:11 the other devices helped me to see my error Jun 15 17:54:29 now mrmoku|italy and JaMa are not here Jun 15 17:54:33 what whould I do? Jun 15 17:54:35 commit that Jun 15 17:54:45 ? Jun 15 17:55:10 ping leviathan Jun 15 18:00:21 GNUtoo: ooh, the syntax. didn't know that either Jun 15 18:00:56 GNUtoo: if it fixes the current situation, go ahead and commit it for shr Jun 15 18:01:46 yes but I wonder what's best between: Jun 15 18:01:56 *add a fixed init script Jun 15 18:02:00 *removing connmand Jun 15 18:02:01 *both Jun 15 19:17:24 http://www.youtube.com/watch?v=ATne4aBeDE0 Jun 15 19:17:44 you will soon will be able to change icons position on the desktop Jun 15 19:17:48 with elfe :) Jun 15 19:34:59 nice Jun 15 20:08:55 ping leviathan Jun 15 22:01:51 <[Rui]> ping! Jun 15 22:02:05 <[Rui]> mickey|daddy: GRATS! Jun 15 22:03:00 <[Rui]> mickey|daddy: address for "cigar" sending? :) **** ENDING LOGGING AT Thu Jun 16 02:59:57 2011