**** BEGIN LOGGING AT Mon Jan 25 02:59:57 2010 Jan 25 08:31:30 Morning Jan 25 08:54:32 http://gitorious.org/android-on-freerunner/platform_system_core/blobs/master/adb/SERVICES.TXT Jan 25 08:56:39 ok i still don't get why i could access it over wifi Jan 25 09:02:27 http://gitorious.org/android-on-freerunner/platform_system_core/commit/2f38b699713dc2587a771bd5d4c6a47329728f5e Jan 25 09:11:38 perhaps some one can do a git bicsect and work it out... Jan 25 09:38:27 d1b, intresting, not sure I interpit that code correct but could it be that the c code takes all connections and then some firewalling or routing magic fixes it on a "normal" android device? Jan 25 10:50:51 larlin: normal android droped netfilter Jan 25 10:50:54 iirc. Jan 25 10:53:03 have anyone tested this on a real android device? =) Jan 25 10:54:32 larlin: asked that in #android Jan 25 10:54:35 didn't get a real answer Jan 25 10:55:07 test when you get yours Jan 25 10:55:15 larlin: mine isn't a dev phone. Jan 25 10:55:21 but i will root it probably to get it Jan 25 10:55:53 I thought adb was available on all android phones? Jan 25 10:56:22 larlin: not sure. Jan 25 10:56:27 http://osdir.com/ml/android-porting/2009-01/msg00115.html Jan 25 10:56:31 says ADBHOST is old. Jan 25 10:56:57 it seems that otehrs have used adb over tcp/ip Jan 25 10:57:05 so this isn't unique to this port apparently... Jan 25 10:58:16 but what does this jibbrish mean? "the 'adbd' daemon that runs on a real device only listens to the USB communication channel" Jan 25 10:58:39 serial over USB? still network over USB? woodoo over USB? Jan 25 10:59:07 it is meant to work only over usb. apparently Jan 25 10:59:16 but there is also a tcp/ip extension. Jan 25 10:59:44 Android adb supports two kinds of connections: USB and TCP. For cupcake version, the adbd will first check whether /dev/android_usb or /dev/android exits or not. If existing, it will uses USB as the transportion. Otherwise, it will bind to TCP port 5555 by default. Jan 25 10:59:49 http://letsgoustc.spaces.live.com/Blog/cns!89AD27DFB5E249BA!1228.entry Jan 25 11:00:03 yeah but in what way.. Is it still a network over USB thing with good filtering to avoid it to spread to other interfaces or is it some other USB communication method Jan 25 11:00:06 there are other examples / people who did this. Jan 25 11:00:13 no idea Jan 25 11:01:01 intresting Jan 25 11:01:38 http://www.mail-archive.com/android-porting@googlegroups.com/msg01670.html Jan 25 11:04:05 if we switch adb over to using adb in a non network way that will close all possibilety of using the usb-network for ssh and things like that on the other hand the device will act more like a "real" device possibly making it easier to make it appear as a storage device for access to sdcard and such things Jan 25 11:04:48 mm. Jan 25 11:08:57 http://osdir.com/ml/android-porting/2009-01/msg00115.html Jan 25 11:12:27 I wounder what support is needed to make the daemon talk directly to the usb does it need some sort of driver or similar? Jan 25 11:12:56 no idea. Jan 25 11:13:13 trying to find this adbhost in the code atm Jan 25 11:13:25 yay for gitorious not having GREP Jan 25 11:13:26 etc. Jan 25 11:14:34 easiest to use repo and check out the whole repo (cupcake or eclair) and then grep in that locally Jan 25 11:15:01 too much bandwidth atm Jan 25 11:15:20 if you know where ADB is .. in a java file Jan 25 11:15:24 better for adbhost Jan 25 11:16:21 I can try a quick grep in one old version I got here Jan 25 11:16:26 k Jan 25 11:18:38 that would be good. Jan 25 11:18:46 adb start script looks like build/envsetup.sh here atleast Jan 25 11:19:04 i think there is a java file with adb in the name Jan 25 11:19:13 ls -laR|grep "adb*" perhaps? Jan 25 11:20:39 might be Adb Jan 25 11:21:42 my grep skills fails me tryed the one above and get's everything outputed Jan 25 11:22:17 m. Jan 25 11:22:32 try grep "adb*" * -R Jan 25 11:22:34 in the top dir Jan 25 11:22:44 try grep "adbhost" * -R (actually) Jan 25 11:25:23 it's unbelivable how much code there is in android =) Jan 25 11:27:09 (through USB for devices, through TCP for emulators) Jan 25 11:27:11 mm. Jan 25 11:27:19 perhaps it is just built in simulator mode? Jan 25 11:28:06 I think that for simlutaors it is so they didn't have to set up a fake usb device for adb Jan 25 11:28:19 yes probably. Jan 25 11:28:20 and adb reports the freerunner as a emulator not a device Jan 25 11:28:29 so it should be somewhere there it is controlled Jan 25 11:28:31 mm i dont' remember. Jan 25 11:28:38 does it say emulator and not device? Jan 25 11:28:42 are you sure .. Jan 25 11:28:46 yea Jan 25 11:29:01 well that would be easy to fix ... lets hope. Jan 25 11:29:08 I can check one sec Jan 25 11:30:48 larlin@hermes:~/Projects/openmoko$ ADBHOST=192.168.0.202 ./adb devices Jan 25 11:30:48 * daemon not running. starting it now * Jan 25 11:30:48 * daemon started successfully * Jan 25 11:30:48 List of devices attached Jan 25 11:30:48 emulator-5554 device Jan 25 11:31:20 right. Jan 25 11:32:02 problem is if we switch to non emulator it should require some sort of hardware code to manage where the usb port is and how it is accessed question is if that exists Jan 25 11:32:13 are you sure this is actually the case? Jan 25 11:32:45 sure no, I consider it likely Jan 25 11:32:48 i don't know what it says if adb is on the device. Jan 25 11:32:51 worth a test absolutly Jan 25 11:34:54 I'm pretty sure that the moko runs adb in this mode for the same reason the emulaters does it's easier to set up a network interface on usb then fixing it for the special adb part but if we can find the flag that changes this it would be intresting to test Jan 25 11:35:51 m. Jan 25 11:35:56 try adb devices Jan 25 11:36:24 larlin@hermes:~/Projects/openmoko$ ./adb devices Jan 25 11:36:24 List of devices attached Jan 25 11:36:24 emulator-5554 device Jan 25 11:36:24 ?? That was what I did above? Jan 25 11:36:40 yah same as above. Jan 25 11:43:20 List of devices attached Jan 25 11:43:20 HT842GZ39778 device Jan 25 11:43:28 is what others have like apparently Jan 25 11:43:31 with real phones Jan 25 11:44:42 yeah and I'm thinking that the emulator name is not a think because someone never changed the name it means that adb runs in "emulator" mode but I have no real proof for this Jan 25 11:46:49 m. Jan 25 11:57:06 I need to run away and fix some food could try to see if I can build a version with adb in a device mode and see if it fails spectacular =) Jan 25 12:00:09 you mean eat food :P Jan 25 12:13:05 d1b, no I need to buy it, make it and then eat it =) Jan 25 12:13:18 and first I needed to shower witch I have done now =) Jan 25 12:18:34 hi, I'm running android-on-freerunner eclair from 2009-53 Jan 25 12:18:47 I want to connect to the internet with usb Jan 25 12:18:57 I added default route Jan 25 12:18:59 ping works Jan 25 12:19:20 but the browser states it's offline Jan 25 12:19:20 have you verified that dns works? Jan 25 12:19:57 larlin AFAIK android does not use standard /etc/resolv.conf, so how do I check this? Jan 25 12:22:15 humm I have done it before but now I compleatly forgott how that works Jan 25 12:22:37 in the browser can you try a ip to some webserver? Jan 25 12:23:17 setprop net.dns1 Jan 25 12:23:40 aa, it works :) Jan 25 12:23:50 yes, when I use IP in browser it works :) Jan 25 12:24:06 then do as d1b suggested above then it should work Jan 25 12:24:10 setprop net.dns1 will be handful as well, thanks :) Jan 25 12:24:26 well food now!! cya later all! Jan 25 12:25:39 larlin: thanks again! Jan 25 13:22:42 you gotta be kidding have they messed up so I can't have java6 to make freerunner gah Jan 25 13:22:50 *android Jan 25 13:23:19 larlin: ? Jan 25 13:23:35 doesn't openjdk work? Jan 25 13:23:55 "Java 6 is not supported, because of incompatibilities with @Override." Jan 25 13:24:18 what lol? Jan 25 13:24:29 so what are you meant to use lol? Jan 25 13:24:35 java 5 Jan 25 13:24:41 woot... Jan 25 13:24:43 fail Jan 25 13:26:30 yeah very much so I would say Jan 25 13:29:28 so I had to pull inte repo from jaunty to install java-5 and change defaults that will come back and hunt me if I forget to reset it to java-6... Well building now Jan 25 15:30:11 hi Jan 25 17:01:31 hello again Jan 25 17:01:39 I'm trying to benchmark Android Browser Jan 25 17:02:05 specifically JavaScript, but it seems some sort of "stop all scripts after some time" rule is applied Jan 25 17:02:29 scripts start to run, but they never complete. Any way to overcome this? Jan 25 17:06:00 not that I know of there is no way to split the benchmark up in several parts? Jan 25 17:11:14 larlin: yes, now I try to do this :( Jan 25 17:11:27 I tried to find some setting about thsi Jan 25 17:11:49 "block JavaScript after X seconds" or something, but I found nothing :( Jan 25 17:12:29 I'll need to compare partial results Jan 25 17:19:23 hm Jan 25 17:34:21 Gabrys: You might have better luck asking your question on the regular Android mailing lists or IRC channels. Jan 25 17:34:45 AFAIK, we aren't doing anything freerunner-specific in that area Jan 25 17:36:47 larlin: Are you available? Jan 25 17:36:56 yeah Jan 25 17:37:23 trying to wrap my head around adbd usb =) Jan 25 17:37:24 in http://code.google.com/p/android-on-freerunner/issues/detail?id=108#c8 you mentioned research you've been doing on the adb issu Jan 25 17:37:27 confusing =) Jan 25 17:37:50 Do you have links to what you've found? Jan 25 17:38:17 humm that would be a idea =) To bad at that sorry, have been reading here and there Jan 25 17:38:26 will see if I can find that again Jan 25 17:38:49 Looking at the code, I see it do a TCP bind to localhost, but not to 0.0.0.0:5555, which seems to be the issue Jan 25 17:40:51 I have also read abou the local USB interface method, but I can seem to find a switch for it Jan 25 17:43:41 if I have understand everything correct(that would be the day) the emulator runs adb over network to avoid setting up a virtuall usb port. Jan 25 17:43:41 Normal devices uses a usb protocoll and there is a Linux usb gadget driver for this. Not sure how easy it would to set this up but should be possible Jan 25 17:47:23 here is the first bit about adb over IP Jan 25 17:47:23 http://osdir.com/ml/android-porting/2009-01/msg00165.html Jan 25 17:47:57 we need to make it backwards =) Jan 25 17:48:00 using adb over other devices is a great opportunity. Jan 25 17:50:41 That implies that just enabling the gadget driver in the kernel might do it Jan 25 17:51:17 I could try that and see if android explodes Jan 25 17:53:32 Looks like it is enabled Jan 25 17:54:00 I got the master brancher I only saw it as module? Is the module there also? Jan 25 17:55:43 Nothing on the master branch except wifi is built as a module. Where are you looking? Jan 25 17:56:12 if this is not in by default it probably needs to be somewhere also: Jan 25 17:56:12 http://android.git.kernel.org/?p=kernel/common.git;a=commit;h=fc20ff5191502a6c0bb9e112cea56feea8d12f4e Jan 25 17:56:41 cd kernel Jan 25 17:56:41 make menuconfig Jan 25 17:56:41 Or are I'm doing something silly? =) Jan 25 17:57:09 I'm talking about our kernel Jan 25 17:57:51 I thought that was in the master tree also?? Jan 25 17:59:04 Your looking at the Google kernel. That driver doesn't seem to be in ours. Jan 25 17:59:28 I have to go to a meeting. I'll try to catch you later. Jan 25 17:59:46 k cya Jan 25 18:00:13 Serdar1, you know where "our" kernel code is? Jan 25 18:00:34 finstaller for example Jan 25 18:00:59 git clone git://gitorious.org/android-on-freerunner/kernel_openmoko.git Jan 25 18:01:21 okay so it is in it own tree? Jan 25 18:01:31 own branch Jan 25 18:01:34 too Jan 25 18:01:46 it is mostly linked to the git.openmoko.org stuff. Jan 25 18:02:06 ahh Jan 25 18:02:58 so how manny kernels are there floating around 3 4? Jan 25 18:05:17 there? Jan 25 18:06:20 we got the google kernel in the git repo(master/cupcake/eclair), "our kernel", "installer kernel"<-- is this one different from "our kernel"? Jan 25 18:11:18 we don't use the google kernel Jan 25 18:11:34 we are useing the openmoko kernel Jan 25 18:12:05 http://git.openmoko.org/?p=kernel.git;a=summary Jan 25 18:13:39 just trying to get a overview of what code we got where Jan 25 18:15:02 if you want to get the kernel, than sync the installer branch Jan 25 18:15:25 btw. are you going to FOSDEM? Jan 25 18:23:39 Serdar1, yes I'm going there Jan 25 18:24:03 so it's only three of us Jan 25 18:24:20 no niels must comming too. Jan 25 18:42:24 asm/plat-s3c24xx/neo1973.h: No such file or directory Jan 25 18:54:58 hm.. crashing a lot while playing around with apps Jan 25 19:06:18 i can't seem to connect to my freerunner via usb with adb, is there a way to connect via wifi? Jan 25 19:08:13 mac? Jan 25 19:08:34 we find out, that it works with wifi too. Jan 25 19:08:37 nope, linux Jan 25 19:08:45 how can i tell adb to start? Jan 25 19:09:28 connect with wifi. Jan 25 19:09:34 eh, to start and connect to an ip adress, not to usb? Jan 25 19:09:54 huh? Jan 25 19:10:14 adb devices doesn't say anything, even though dmesg shows that the device is created Jan 25 19:14:06 do you want connect adb via wifi or not? Jan 25 19:14:56 yes Jan 25 19:15:05 but how do i tell adb that it should connect via wifi, not via usb? Jan 25 19:15:12 okay is your device connected to wifi? Jan 25 19:18:39 simple question. Jan 25 19:18:45 where's the answer? Jan 25 19:19:05 hi juan_arandaalvar Jan 25 19:20:44 yes, i can browse the net Jan 25 19:20:51 okay. Jan 25 19:21:10 than ADBHOST=ip_of_openmoko ./adb kill-server Jan 25 19:21:26 and than ADBHOST=ip_of_openmoko ./adb devices Jan 25 19:22:40 great, that was what i was looking for :) Jan 25 19:23:13 yes. Gabrys found it. Jan 25 19:24:24 great, this lets me actually connect to the freerunner :) Jan 25 19:24:55 it is a great opportunity, problem is, this is open to all. Jan 25 19:25:38 need to investigated, but it should be able for people to connect your phone too. Jan 25 19:25:45 why doesn't adb work over usb? Jan 25 19:25:52 we need a handshake. Jan 25 19:26:00 it works. Jan 25 19:26:21 ecraven, query? Jan 25 19:26:37 query? Jan 25 19:26:43 /query Jan 25 19:49:15 larlin: I tried cherry picking the commit you found, then another that that one depended on but that caused more issues Jan 25 19:49:47 I should probably contact Panicking about the state of the kernel Jan 25 19:50:08 yeah sounds like a good idea Jan 25 19:51:10 I was not able to compile the kernel in the kernel repo. so I haven't tried with the usb gadget module Jan 25 19:51:48 why? Jan 25 19:52:03 please try this: Jan 25 19:52:08 kernel/printk.c:39:38: error: asm/plat-s3c24xx/neo1973.h: No such file or directory Jan 25 19:52:08 make[1]: *** [kernel/printk.o] Error 1 Jan 25 19:52:08 make: *** [kernel] Error 2 Jan 25 19:53:13 I was able to compile the kernel Jan 25 19:53:16 but only the kernel Jan 25 19:53:32 are you useing the installer branch? Jan 25 19:53:51 the git repo you posted above Jan 25 19:53:54 He shouldn't be using the installer branch Jan 25 19:54:10 He should be on cupcake or master Jan 25 19:54:26 I'm on master Jan 25 19:54:30 the kernel isn't in cupcake or master. Jan 25 19:54:50 What do you mean? Jan 25 19:55:13 good that someone more then me is confused =) Jan 25 19:55:21 I am talking about kernel. Jan 25 19:55:25 like larlin does. Jan 25 19:56:10 http://gitorious.org/android-on-freerunner/kernel_openmoko/trees/master Jan 25 19:57:01 Larlin: I never build the kernel standalone. I always do it as part of the build Jan 25 19:57:18 What commands are you trying? Jan 25 19:58:12 all I need is where the kernel code is for the whole build so I can run menuconfig on it Jan 25 19:58:23 why? Jan 25 19:58:28 copy the .config Jan 25 19:58:44 arch/arm/configs/ Jan 25 19:58:50 and make oldconfig Jan 25 19:59:11 you got 4 configs there Jan 25 20:00:00 larlin: git clone git://gitorious.org/android-on-freerunner/kernel_openmoko.git Jan 25 20:00:17 then checkout the branch you want Jan 25 20:00:57 yeah that gives the error above Jan 25 20:02:36 trying again Jan 25 20:02:40 building kernel works for me. Jan 25 20:02:55 but I am useing the installer branch. Jan 25 20:03:30 The kernel in the cupcake and master branches is integrated with the Android build Jan 25 20:03:54 but is it the kernel that is used? Jan 25 20:04:02 Yes Jan 25 20:04:25 The installer kernel is only used in the installer. Jan 25 20:04:58 and that kernel is in the kernel dir in the master branch? Jan 25 20:05:19 right Jan 25 20:05:31 But you will need a toolchain Jan 25 20:05:36 Do you have one? Jan 25 20:05:51 I can build the whole system atleast I don't get any errors Jan 25 20:06:01 (from master branch) Jan 25 20:06:06 Then you are already building the kernel Jan 25 20:06:48 so then make menuconfig i kernel dir should change the kernel conf used and then I can make the whole system and try it? Jan 25 20:07:26 Let me try something Jan 25 20:07:30 Just a sec Jan 25 20:07:49 Why do you want to do make menuconfig? Jan 25 20:08:31 can you try this? Jan 25 20:08:32 http://pastebin.com/m24f76dea Jan 25 20:08:40 http://pastebin.com/m3b7f5d5c Jan 25 20:08:49 because it's the only way I got a snowballs chance in hell to find the and check the status of the Linux usb gadget driver Jan 25 20:08:49 http://pastebin.com/m3609a574 Jan 25 20:09:20 and than make -f Makefile.kernel Jan 25 20:09:25 should work. Jan 25 20:09:37 it works here. Jan 25 20:10:29 if it is the right code in the tree and get's build with the rest of the system I don't need to build the kernel seperate Jan 25 20:10:51 you can always build everything. Jan 25 20:11:49 larlin: The gadget driver code is not in our kernel--there is no drivers/usb/ gadget/ android_adb.c Jan 25 20:12:24 It's not a matter of changing the config Jan 25 20:13:26 If you want to play with the config, edit gta02_android_defconfig Jan 25 20:13:43 The kernel make may prompt you for additional changes Jan 25 20:14:06 there is to gadget driver, the Linux usb gadget driver and the android usb gadget driver right now I'm looking at the former. How hard would it be to add that file did it depend on alot of other code? Jan 25 20:14:52 the actual confi ends up in out/target/product/fr/obj/KERNEL/.config Jan 25 20:15:46 That's what I tried to do by cherry picking the commit, but it didn't work Jan 25 20:25:13 so is it the freerunner kernel that is to old or what is the problem? Jan 25 20:41:43 is there an easy patch for the mixer settings? Jan 25 20:43:13 hi Aitch_ Jan 25 20:43:48 Hi Serdar, just reading the log. It's been busy today Jan 25 21:33:04 larlin: Right, I think the problem is that no one has been consistently merging Android kernel updates into our kernel Jan 25 21:34:18 k I'm trying with a uggly manual patch apply just to check if it may work Jan 25 21:34:48 Michael and some other people have been working on moving to a new kernel version. I don't know where that stands Jan 25 21:35:37 Larlin: I tried a bit more and made some more progress. I'll let you know if I get it working Jan 25 21:38:40 huh? Jan 25 21:39:04 iirc we never supposed to merge with android kernel. Jan 25 21:47:39 well going to bed soon will try more tomorrow Jan 25 22:26:56 Hello Jan 25 22:49:49 hi juan_arandaalvar Jan 25 23:14:45 I'm back, Downloaded from the project page Jan 25 23:14:50 And build it Jan 25 23:15:26 but the fs image is in a directory called out/target/product/generic Jan 25 23:15:35 Did I forget to change something? Jan 25 23:16:10 What should TARGET_PRODUCT be changed to and where? Jan 25 23:16:26 is it in build/core/product_config? Jan 25 23:16:35 where can I learn to configure this stuff? Jan 25 23:16:51 sorry, I told you I had a bunch of questions Jan 25 23:36:32 how do you build it? Jan 25 23:36:59 make TARGET_PRODUCT=freerunner for cupcake Jan 25 23:37:11 make TARGET_PRODUCT=fr for master/eclair Jan 25 23:37:31 but afair it's on the wiki Jan 25 23:38:04 but before clean everything Jan 25 23:38:06 make clean Jan 25 23:38:08 make clobber Jan 26 00:47:01 Which wiki page? Jan 26 00:56:22 our Jan 26 00:56:27 at google codes Jan 26 00:57:00 http://code.google.com/p/android-on-freerunner/wiki/BuildIt Jan 26 02:02:50 hi all Jan 26 02:10:58 hi Jan 26 02:11:02 and good night **** ENDING LOGGING AT Tue Jan 26 02:59:56 2010