07:23 <shineworld> Morning all. I've got a bad issue with Rockchip linux kernel for Android 3.0.36+ (used in a CUBE_U25GT_PRO tablet). When I connect an accessory device (in this case a FTDI FT312D Android -> FT -> UART converter) just after sending command for accessory request (using libusb_control_transfer with 53) the kernel shutdown the endpoint before to reply to request and causing a (-USHUTDOWN)(-108). In follow link you can get libpcap logs and linux

07:23 <shineworld> program used to simulate an ADK device and get the error: https://www.dropbox.com/sh/4u0dkgih5rnuk3e/CxysS2jGo3

07:39 <shineworld> hi rellla

08:03 <shineworld> morning akaizen

08:25 <shineworld> Some android/kernel AOA expert here ?

09:02 <mosquito520> I m not expert...Orz

09:04 <shineworld> I've an issue with USB and android

09:06 <shineworld> I'm enough sure that is an issue on kernel level because I've compared AOSP with android sources and code is same

09:09 <shineworld> when android receive from host the command to enter in Android Accessory Mode the OS does that but 90 times on 100 not reply to request packet (timing questions ?) so the accessory device enter in ERROR

09:41 <mosquito520> shineworld: I have some question, does that FT312D UART TX RX worked?

09:42 <shineworld> YES

09:42 <mosquito520> cuz I don't know that "using libusb_control_transfer with 53" mean? Send another command to communitcation with FT312D?

09:43 <shineworld> they works perfectly in Nexus 4/7

09:43 <shineworld> and sometime in rockchip (10 time in 90)

09:44 <shineworld> Pratically 53 is ACCESSORY_START command which host (FTDI FT312D) send to Android to ask it to enter in Android Accessory Mode

09:44 <shineworld> Just a copy from my previous chat:

09:45 <shineworld> I've got a bad issue with Rockchip linux kernel for Android 3.0.36+ (used in a CUBE_U25GT_PRO_1.0.4 2013/05/18 tablet).

09:45 <shineworld> This happen 90 times on 100 but sometime the kernel is able to rebly to control transfer command before the USB will be restarted to enter in Android Open Accessory Mode.

09:45 <shineworld> When I connect an accessory device (in this case a FTDI FT312D to get Android -> FT312D -> UART interface -> PoS system RS232 port) just after sending command for accessory request (using libusb_control_transfer with ACCESSORY_START = 53) the kernel shutdown the endpoint before it reply to request and causing a USHUTDOWN (-108) error and invalidating connection with FTDI FT312D.

09:45 <shineworld> In follow link you can get libpcap logs and linux program that I wrote to simulate an ADK device and get the error:

09:45 <shineworld> https://www.dropbox.com/sh/4u0dkgih5rnuk3e/CxysS2jGo3

09:45 <shineworld> PS: PID and VID should be switched to VID_CUBE and related CUBE PID before to recompile sources.

09:48 <mosquito520> hold on second... Does that FT312D is simple USB to UART converter= =?

09:49 <shineworld> nope, isn't a simple USB to UART but is a Android READY USB to UART

09:50 <shineworld> Pratically is a host device "able" to accept and open a special communication modality (called Android Open Accessory) with Android OS.

09:51 <shineworld> At this point Android, using normal input/output streams can operate with UART ready devices like a PoS terminal

09:51 <naobsd> it doesn't gives /dev/ttyUSB* thing, there is no driver in kernel, it needs libusb, right?

09:51 <shineworld> Android Tablet -> USB cable -> FTDI FT312 chip -> PoS device with UART com

09:51 <mosquito520> uh, I m not realy understand what it mean... I'm wanna trying to figure 1 thing first, does that device has driving already?

09:52 <mosquito520> cuz that spec describe it's a UART converter="=

09:53 <mosquito520> the other thing, I found FTDI usb driver does not enable by default in kernel:)

09:54 <shineworld> A sec. adk-shine.c and binary IS for PC ... is an emulator that I've used to simulate the FTDI CHIP connected with an android tablet and use WireShark to get USB packets

09:55 <shineworld> 1) Android Open Accessory is a feature of Android used to change the normal function of USB port present in ALL systems >= 3.1

09:55 <shineworld> 2) FTDI CHIP is a little micro which implements the AOA protocol and get add a UART port in any android device

09:55 <naobsd> Android tablet act as host? or device? I'm confusing old one that android device act as device and external unit act as host

09:56 <shineworld> http://www.ftdichip.com/Support/Documents/ProgramGuides/FT31XD_Android_programmer_guide(FT_000532).pdf

09:57 <shineworld> in page 4 you can simply understand how it works

09:57 <shineworld> is very simple

09:58 <shineworld> when the android enter in "Android Open Accessory Mode" it close the USB device and open a new one with a different VID:PID

09:58 <naobsd> FT312 is host, hmm.

09:59 <shineworld> during this operation, unfortunately, rockchip kernel miss to reply to last USB command (used to ask android to switch mode)

09:59 <shineworld> this cause the error in FTDI connection...

09:59 <shineworld> seem only a reply timming question because on 100 times the problem disappear for 10

10:00 <shineworld> 1) FTDI (host) send command android to switch mode

10:00 <shineworld> 2) android reply to command and switch mode

10:00 <shineworld> 3) FTDI detect new device and work fine

10:00 <shineworld> VS

10:00 <shineworld> 1) FTDI (host) sendo command android to switch mode

10:00 <shineworld> 2) android switch mode (without reply to previous command)

10:01 <shineworld> 3) FDTI after 2 secs of timeout get in ERROR for missing reply due premature end-point closure

10:01 <shineworld> You can check that in libpcap files in my upon link

10:01 <shineworld> inquired packages are in filename

10:10 <mosquito520> He quit... =.=

10:11 <mosquito520> I've check with my contact window, He told me that He doesn't know this mode either...(laugh)

10:26 <naobsd> ?

10:27 <naobsd> I guess gadget drivers except mass storage/ptp/mtp and adb are not tested at all

11:07 <shineworld> <naobsd> I guess gadget drivers except mass storage/ptp/mtp and adb are not tested at all

11:07 <shineworld> naobsd, that is a bad thing ... for me :) I'm a newbie in kernel usb thigies so....

11:16 <naobsd> it's just my guess, but I never heard/saw someone do AOA with RK

11:17 <naobsd> shineworld: what's your goal, trying AOA? or controlling something via serial? both?

11:38 <shineworld> naobsd, I'm trying to connect an Rockchip based android tablet with a mine board ( https://www.dropbox.com/s/ui4k6te1mbffrcw/IMG_20131002_172731.jpg ) which have a lot of digital I/O and 4 motor controls.

11:38 <shineworld> near to usb cable (in board) there is the FTDI FT312 chip.

11:39 <shineworld> with Nexus products N4/N7 (2012/2013) and Samsung S3/S3 Mini all worked fine

11:40 <shineworld> The android application just connect to board and permit to program any settings, get history alarms, charts of memories, and so on

11:42 <shineworld> so my chain is : Android HMI program -> Android OS . usb in AOA mode -> usb cable -> FTDI FT312D chip -> UART signal converter -> My board main CPU

11:42 <naobsd> how about using USB host on RK Android tablet?

11:43 <shineworld> not all devices have USB Host capability (for example Nexus don't have) so use native USB feature is my "must"

11:44 <shineworld> However I wrote a PC simulator which permit a smart guy to debug WHAT'S happen in kernel ah find a fix

11:45 <shineworld> The simulator just do what FTDI FT312D chip does to get the comunication

11:45 <shineworld> with WireShark is then possible to capture the USB traffic an see the problem

11:45 <shineworld> I don't have rockchip kernel and android sources to do by myself (and overall skill in usb device things)

11:46 <shineworld> I need to find them somewhere

11:47 <shineworld> but my previous experiences with AllWinner product to get sources and build for a specific target (my CUBE tablet model) isn't so simple

13:27 <sjoerd> the function to actually setup the nand device never seem to be called

19:21 <mosquito520> sjoerd: cuz Rockchip using private nand driver...it should be release in the future in someone brand that released source