**** BEGIN LOGGING AT Sat Jul 13 02:59:58 2013 Jul 13 06:55:36 could be, but removing it etc. didn't seem to help Jul 13 08:13:24 vvu: what kind of stuff? Jul 13 08:14:24 i am trying to set up communication between usb g_serial and Android USB Host API Jul 13 08:16:00 if i try to read on the BBB like cat /dev/ttyGS0 and echo on the Android it works Jul 13 08:16:28 but if i try to send something from within java it does reads nothing Jul 13 08:18:27 so are you using the usb API or accesing the file as a serial device? Jul 13 08:19:22 I have not followed up much on you work in the last weeks but isn't it a good time to switch to using the ADK API? Jul 13 08:20:24 switching to ADK means i should exchange roles? android=device BBB=host ? Jul 13 08:20:33 there is also a good chance android will probe the serial device to see if it is a modem Jul 13 08:21:28 mhmh if you have 10 mins to spare i can set up my board and give you some pastebins Jul 13 08:22:05 vvu: indeed you need to switch modes and I don't know how easy this will be. Jul 13 08:22:29 i would try to stick to same mode BBB device Android host Jul 13 08:23:07 At what stage are your? can tyou boot a kernel + ramdisk? Jul 13 08:23:13 yep Jul 13 08:23:20 but with some hacks Jul 13 08:23:38 the android kernel does not like the RNDIS AM335x chip implements Jul 13 08:24:08 well that's like the first step completed. the next step (interacting between android and the BBB) doesn't need to be the same Jul 13 08:24:27 USB_OTG needs to be disabled and RNDIS_HOST_DRIVER needs to be activated in the android kernel Jul 13 08:24:38 yep that is why i am switching to g_serial Jul 13 08:24:43 to get rid of that nasty rndis Jul 13 08:25:23 k-way: something like a mini guide http://beagleboot.vdev.ro/?p=184 Jul 13 08:25:25 really really weird but not critical at this stage Jul 13 08:26:23 brb 5 minutes need to get some coffee Jul 13 08:27:43 did you publish the working code on linux or did you rewrite the code in java? Jul 13 08:27:50 not really happy how the project evolves with all these mini hacks to get things done Jul 13 08:29:02 these are real and existing issues (e.g. the hacks needed) but fixing then directly implies a reduced scope of your project. Jul 13 08:29:24 and that not very good news. Jul 13 08:30:06 Also we wanted usb-host to make things easyer. Jul 13 08:31:20 k-way: true about the scope of the project. also told av500 i want to speed a bit up the work and if i have time to investigate what is happening behind the curtain with the android kernel Jul 13 08:31:38 with that rndis, that driver does some magic Jul 13 08:31:53 so where are you in terms of code (is the code rewritten in java?) Jul 13 08:32:09 yep, have all java code now Jul 13 08:32:16 booting ramdisk all in java Jul 13 08:32:39 with files from sdcard/emulated sdcard for devices that do not have slot Jul 13 08:33:02 can you split conserns (e.g. a two staged approach) Jul 13 08:33:46 now main issue i want to solve is talking to the board via host api Jul 13 08:33:59 the others are in the queue for lateeer Jul 13 08:34:01 step1) use your current code to put a fastboot (or dfu) enabled u-boot on your device. Jul 13 08:34:16 second step the interacting part (downloading images and stuff like that) Jul 13 08:34:50 mhmh av500 was thinking to boot the ramdisk and from there to flash the eMMC. we don`t really want to store full 2gigs of image on the androi device Jul 13 08:34:56 write only the blocks that are needed Jul 13 08:35:33 I don't know how av500 feels about this Jul 13 08:35:44 step 1 is done already; am335x_evm_usbspl config supports dfu Jul 13 08:35:50 tartarus confirmed he tested Jul 13 08:37:10 also something curious Jul 13 08:37:26 usb g_serial has 2 endopoints for communication the bulk ones Jul 13 08:37:42 to me it sounds like role change would be a good next step Jul 13 08:37:57 if i send something on the OUT endpoint(host->device) and then try to read from the IN endooint(device->host) i can read the same data Jul 13 08:38:02 everything from host api Jul 13 08:40:34 but as last resort i can put up some JNI code that talks directly to the file descriptor Jul 13 08:45:21 perhaps this is the same kind of problem. do people have problems unbinding kernel drivers from Android? Jul 13 08:46:10 mhmh i don`t really need to unbind Jul 13 08:46:15 http://lwn.net/Articles/143397/ Jul 13 08:46:45 if some random person tries the project he needs to compile the kernel with those 2 configs set Jul 13 08:47:21 how does /dev/ttyGSxx get configured Jul 13 08:47:59 and why don't you need to unbind? Jul 13 08:48:14 i compiled directly the g_serial driver into the BBB kernel Jul 13 08:48:35 on the android it is seen as ttyACM0 Jul 13 08:49:20 so Android has the usb serial driver compiled in Jul 13 08:49:29 yep Jul 13 08:49:50 lemme set up the board and give some pastebins Jul 13 08:50:19 if you want to talk usb you need to unbind the driver Jul 13 08:50:45 in the java code i`m doing that Jul 13 08:51:05 ok Jul 13 08:53:59 http://pastebin.com/nbn85rH8 Jul 13 08:55:01 someone in #beagle suggested i should disable flow control on the BBB part Jul 13 08:57:30 by all means look at drivers/usb/serial/usb-serial.c ( GrekKH wrote parts of that) Jul 13 08:59:09 I have to go. I don't think I can add much to that Jul 13 08:59:18 okei i will see what i can do Jul 13 08:59:21 thx for the tips Jul 13 08:59:27 look at use device! Jul 13 09:30:47 * vvu is happy! someone forked his BeagleDroid git :) Jul 13 17:03:02 koen, the image which you gave me seems to close the ssh connection whenever I try to load BB-BONE-BACONE. dmesg: http://pastebin.com/qUbZQjbN Jul 13 22:39:52 anyone here ? Jul 14 02:45:21 Oh for the record, ssh issues were possibly due to my beagle running out of storage... deleted a 500mb log file and things were fine. **** ENDING LOGGING AT Sun Jul 14 03:00:00 2013