USB_ModeSwitch

Thanks Josh for the Link..

I tried to find out the version of compiler with which my device kernel is compiled. I found this:

u0_a221@android:/ # cat /proc/version
Linux version 3.0.31-471788 (se.infra@SEP-131) (gcc version 4.4.1 (Sourcery G++ Lite 2010q1-202) ) #1 SMP PREEMPT Mon Dec 3 20:49:31 KST 2012

So it's the exactly the one which you mentioned in previous mail. I'm going to compile this new compiler, lets hope this time it start working. Fingers Crossed.

Hey, the new SWA JB release from Samsung seems indeed to contain a new config file.

I'll try it too - don't think they are using a different kernel configuration for India ...

Hmm, I don't know - the new package they provided results in modules with this:

0xb5a27644 module_layout

That's the checksum I got for the working 3110 modules.


But for the 3100 I extracted the checksum from a GT-P3100 firmware and got this:

0xb482dc01 module_layout

Can you check what the stock modules on your device are showing? You may want to revisit my previous post:
http://www.draisberghof.de/usb_modeswit ... 8357#p8357

My device module_layout is something else only

0x143474f1 module_layout

What I managed to get is this
0xf9b2d754 module_layout
which is no where.

I'm again going to compile lets see what I get this time..

EDIT: Also I manage to get this module layout: 0x1efb671a
But still no success..
I'm attaching my config file. have a look..

I suggest to first build the kernel exactly as indicated, with no changes to the original configuration. Then check "scsi_wait_scan.ko".

I checked again with a module from GT-P3100 that someone uploaded for me and I also got

0x143474f1 module_layout

So you must get this result. If you have the CodeSourcery toolchain running and you don't get the exact same module layout, there is no way that the configuration provided by Samsung is correct.

In this case, write to them again and provide the differing module layout side by side. This is what I have done for the GT-P3110.

Hahaa !

I was thinking about the checksum 0x143474f1 for a while. I had seen it before.

Then I checked and it came up after the first try from Samsung to provide a fix for the GT-P3110. It did not work there, but - lets try that configuration on the GT-P3100.

It might actually work!

Eureka.. it worked..

Thanks alot Josh.. Can you send me the config file for this.. I'll to make a comparision with the current one that I'm using..
You can post this module in your Module section with a remark saying works with 0x143474f1 module layout..

I tried ppp widget.. everything went like butter..

Thanks once again Josh..

I suggest to first build the kernel exactly as indicated, with no changes to the original configuration. Then check "scsi_wait_scan.ko".

Yesterday I tried around 15 times in all possible way. But I could not manage to get the required module_layout..

In this case, write to them again and provide the differing module layout side by side. This is what I have done for the GT-P3110

I sent a mail yesterday but still waiting for the response from samsung..

However, yours module worked for me.. Please send me the config file or tell me which source file you dpwnloaded from samsung site..

Here you go ... I find it incomprehensible that Samsung fails to attach any form of version information to the configurations or to the kernel source.

Anyway, as indicated this configuration was included as the default configuration in one attempt to get the config right for the GT-P3100. The original name is
android_espresso_omap4430_r04_user_defconfig.

I figure you can give it any name and put it into "arch/arm/config", then run
"make ARCH=arm <your-custom-config-name>".

BTW, here is a script that I am running sourced (with a "." command) before compiling Android kernels: I call it "envk" and run it in the shell before compiling with ". envk" - no more need to edit the Makefile and I can just run "make" without giving the ARCH variable.

The compiler path has to be adjusted, obviously.

Hi Josh

Im amazed by your efforts in helping people with modules and want to thank you for your great app.
I was trying to get modules for my Galaxy S3 - official JB 4.1.2 (I9300XXELKC) 3.0.31-566833 without success
Maybe you can help ?
One more question. If I compile whole kernel with necessary modules and flash to device will it work ?
Im asking because as i understand it whole problem is when i try to compile only modules and must know exact kernel configuration to get right "MAGIC NUMBER" ?
Will compilation of whole kernel bypass this problem ?

gregorez wrote:Im asking because as i understand it whole problem is when i try to compile only modules and must know exact kernel configuration to get right "MAGIC NUMBER" ?

That's exactly the problem.

For the S3, there is already update no. 7 available, but I'm not sure if it matches the kernel config on the device.

If you would like to help, follow the steps in this post (in a new thread for the I9300):
http://www.draisberghof.de/usb_modeswit ... 8496#p8496

gregorez wrote:Will compilation of whole kernel bypass this problem ?

It does help - but usually you will lose some hardware support because of closed source drivers that are missing from the kernel source package.

Many have tried before ...

But, as stated before, the correct configuration file has to be provided under the rules of the GPL, and Samsung has just started to comply.
If we find that the "magic numbers" from your module do not match the ones in the latest I9300 kernel (I've compiled it two days ago), then a complaint at the Samsung site will get things moving.

Thanks Josh, http://forum.xda-developers.com/showthr ... 515&page=2

Hi Josh,

I just updated to the following Android version
4.1.2 and

Kernel version
3.0.31-692503
se.infra@SEP - 111 #1
SMP PREEMPT Wed Dec 12 21:17:45
KST 2012.

I updated from the prevous set up which was as follows :-

4.1.1 and

Kernel version
3.0.31-471788
se.infra@SEP - 131 #1
SMP PREEMPT Mon Dec 3 20:49:31
KST 2012.


I tried connecting to the net with my usb datacard and it worked!! What a surprise!! A software upgrade and this still works. Samsung saved you some work Josh. BTW, now it works with the wifi on (not sure if I had to turn off wifi or turn on wifi to connect through my datacard earlier).

I suppose that is a bonus of the MODVERSIONS feature - once the checksum is matching, the kernel name is not checked anymore ...