[Zero7]

RandomB. Thanks for the great work. I have been waiting for a long time for someone like you to come and revive our loved Novel. Appreciate your efforts. I would like to go for one of your versions.

[DVD2955]

Jaw dropping News!! I can not wait till you feel that there is a release for the general/average user like myself!



While you have gotten the OS working what apps have you tested? If any?

The few that I use the most are: Amazon Apps, Google Market, Amazon Kindle, ComiX, B&N or Nook, Adobe PDF, Akdio(sp?) eBook Reader.
I keep trying to run Amazon versions of "Anger Birds" but the graphics keep failing on me with my current setup.

:David

[sainthooligan]

nice work!Now, break's over, let's try and get this finished up by tomorrow. ;p

[randomblame]

woot got quadrant to run through it's benchmarks - zillions of egl errors in logcat but it's working - that's what counts right?

looks like it's gonna be buggy software 3d for a while but it's better than no 3d lol I'll keep playing with it.

This post has been edited by randomblame: 15 October 2011 - 01:17 PM

[randomblame]

Today's mission is to recompile libhardware_legacy.so
I also need to figure out what's wrong with my modified ramdisk so I can get the new versions of netdaemon and vold working

This post has been edited by randomblame: 15 October 2011 - 02:44 PM

[randomblame]

Ok I need help figuring out the ramdisk issue. I make a ramdisk, sign it, rename it etc and it feeds me this crap:6. fuse ramdisk.imgabout to run program [/tmp/fuse] with 3 args
header crc: 0x0ac9549c
compute crc: 0x0ac9549c
magic: OK
version: ramdisk
image crc: 0x0acab24a
compute crc: 0x46363cdd
Err: image crc fail!
run_program: child exited with status 9

It's obviously seeing the header and saying that's fine but it sees that the entirety of the file doesn't match what it's expecting. Is fuse locking me out? Is it even possible to modify the ramdisk? Is there a version of fuse I can stick in here to fix this crap?

edit*

so it looks like I have to modify fuse: http://www.slatedroi...ge__hl__ramdisk

I'm not very good with a hex editor. I can't seem to find the offset for the crc check. It would be very helpful if someone could point me in the right direction.


I'm trying to find a hex editor that can compare two binaries but no luck I tried putting two different fuses together in a folder and running
hexdump -C fuse > fuse.txt
hexdump -C fuse2 > fuse2.txt

and generating a diff between them but it didn't work bah

This post has been edited by randomblame: 15 October 2011 - 07:26 PM

[randomblame]

new rom up - fixes include software 3d and proper button mapping etc etc still no wifi I wasted the day fighting fuse trying to get a ramdisk to load. Without the ramdisk I can't get the right version of netd working and I'm pretty sure it's required for wifi If you're good with a hex editor step up

well in the process of trying to fix audio I fixed the backlight control. accelerometer still broken - don't know what to do about that it was broken in gingerbread too.

This post has been edited by randomblame: 15 October 2011 - 09:20 PM

[terminander]

HOWTO: Alter or create the PDN ramdisk.img by Roebeet.

http://www.slatedroi...pdn-ramdiskimg/


I couldn't get his command line to work, even though I cut & pasted. But I did fix your ramdisk. Removed the first 64 bytes, opened in archive manager and got ramdisk. Then mounted Ramdisk, looked at files, then ejected drive. Regzipped and ran mkimage and it worked. But probably just openning and closing in archive manager, would have fix the checksum.problem.

This post has been edited by terminander: 16 October 2011 - 02:31 AM

[randomblame]

terminander, on 16 October 2011 - 02:29 AM, said:

HOWTO: Alter or create the PDN ramdisk.img by Roebeet.

http://www.slatedroi...pdn-ramdiskimg/


I couldn't get his command line to work, even though I cut & pasted. But I did fix your ramdisk. Removed the first 64 bytes, opened in archive manager and got ramdisk. Then mounted Ramdisk, looked at files, then ejected drive. Regzipped and ran mkimage and it worked. But probably just openning and closing in archive manager, would have fix the checksum.problem.

Wait what? Its a different size how can it match fuses hardcoded value? Post it so I can see please and a recovery log


The bottom entry in the recovery log is newest keep that in mind

How I understand it is fuse checks for two things before it writes a ramdisk:
the first 64 bytes - the signature
the overall size

Those values are hard coded into fuse which requires a hex edit to make a new version of fuse for a new ramdisk. I really hope I'm wrong but I think you probably read the top entry in your recovery log

This post has been edited by randomblame: 16 October 2011 - 03:28 PM

[terminander]

Well I flashed it with fuse Fri night and I think it worked and booted. Today I had you newer system.img in and when I flashed/fused the ramdisk.img, it doesn't finish init and I didn't get adb. So I just flashed cm6 again.

 ramdisk(copy).zip (160.75K)
Number of downloads: 9

Just change .zip to .img. There are posts on altering ramdisk.img and recovery.img so fuse may only hardcode bootloader.img.

Rotation/accelerometer is in sensors.smdk6410.so sensors.s3c6410.so, but I had no luck getting it working, causes a constant restart of services.

[randomblame]

terminander, on 16 October 2011 - 03:52 PM, said:

Well I flashed it with fuse Fri night and I think it worked and booted. Today I had you newer system.img in and when I flashed/fused the ramdisk.img, it doesn't finish init and I didn't get adb. So I just flashed cm6 again.

ramdisk(copy).zip

Just change .zip to .img. There are posts on altering ramdisk.img and recovery.img so fuse may only hardcode bootloader.img.

Rotation/accelerometer is in sensors.smdk6410.so sensors.s3c6410.so, but I had no luck getting it working, causes a constant restart of services.

tested that ramdisk and it was flashed but no boot no adb just flashing blue screen I probably messed up the init.rc somehow strange I'll see what I can learn about this - my mind is boggled hehe

*edit
Ok so I'm pretty sure that the format of this ramdisk is wrong. It's simply not working at all with any system.imgs and it should be. I think it shouldn't be in the format of an image file gzipped it should be just a directory structure within a cpio.gz I think. I noticed the signature is different on it - how did you find the right signature? Is that the trick to overriding the crc issue?

ha I re-read that guide to remaking ramdisks and figured it out. The first time I read that I just used it to extract the ramdisk then I read some other guide to put it back together. That did it. Thanks a zillion

I guess fuse derives the end crc from the initial 64byte signature YAY

This post has been edited by randomblame: 16 October 2011 - 05:06 PM

[randomblame]

Well I've got a properly modified init.rc in the ramdisk that allows the new version of netd to work. Things are shaping up, it's faster and more stable now yay.

[jmeister234]

root@android:/ # lsmod


rtnet3070sta 106782 0 - Live 0xbf0bb000

rt3070sta 717732 1 rtnet3070sta, Live 0xbf009000 (P)

rtutil3070sta 26168 2 rtnet3070sta,rt3070sta, Live 0xbf000000

root@android:/ #

It appears the drivers are loaded. But in logcat it says:

E/SurfaceFlinger( 1939): eglSwapBuffers: EGL error 0x300c (EGL_BAD_PARAMETER)
E/WifiService( 1939): Invoking mWifiStateMachine.setWifiEnabled
E/WifiStateMachine( 1939): Failed to load driver!
E/WifiStateMachine( 1939): DriverFailedState
E/SurfaceFlinger( 1939): eglSwapBuffers: EGL error 0x300c (EGL_BAD_PARAMETER)

I'm using the working drivers from the alpha gingerbread - but i don't think it matters. And the EGL errors are just to show I didn't leave anything out.

[randomblame]

jmeister234, on 16 October 2011 - 09:42 PM, said:

root@android:/ # lsmod


rtnet3070sta 106782 0 - Live 0xbf0bb000

rt3070sta 717732 1 rtnet3070sta, Live 0xbf009000 (P)

rtutil3070sta 26168 2 rtnet3070sta,rt3070sta, Live 0xbf000000

root@android:/ #

It appears the drivers are loaded. But in logcat it says:

E/SurfaceFlinger( 1939): eglSwapBuffers: EGL error 0x300c (EGL_BAD_PARAMETER)
E/WifiService( 1939): Invoking mWifiStateMachine.setWifiEnabled
E/WifiStateMachine( 1939): Failed to load driver!
E/WifiStateMachine( 1939): DriverFailedState
E/SurfaceFlinger( 1939): eglSwapBuffers: EGL error 0x300c (EGL_BAD_PARAMETER)

I'm using the working drivers from the alpha gingerbread - but i don't think it matters. And the EGL errors are just to show I didn't leave anything out.

Yeah for some nutty reason I have to have all those fimg files in there for soft3d to work and libeglfimg is the one throwing the errors. I know the wifi modules are loading but libhardware_legacy has more wifi stuff in it and needs to be recompiled I think or I'm missing something obvious

Think of the kernel and modules as firmware we need something in userland that knows how to talk to that firmware and make it go.

This post has been edited by randomblame: 16 October 2011 - 10:22 PM

[jmeister234]

This guy has some recompiled drivers for the rt3070sta.ko

http://markit.dyndns.../ac100/32/wifi/

I noticed we're still on v.29 for the kernel - so maybe that doesn't help us...

[randomblame]

jmeister234, on 16 October 2011 - 10:31 PM, said:

This guy has some recompiled drivers for the rt3070sta.ko

http://markit.dyndns.../ac100/32/wifi/

I noticed we're still on v.29 for the kernel - so maybe that doesn't help us...

Try them they should work you can test them with gingerbread they might help the disconnect issue at least maybe more. It looks like gb required a rewrite for initialization so maybe that's our problem and maybe these will fix it.

[randomblame]

LoadWiFi.sh in /system/bin may need to be edited to reflect those new filenames but I'm not sure if its even being called anymore

edit never mind the file names are the same ha different file sizes though I'm optimistic trying it out now


edit getting different behaviour from the new modules but in the end it's about the same:

E/WifiService( 1917): Invoking mWifiStateMachine.setWifiEnabled
E/WifiStateMachine( 1917): Failed to load driver!
E/WifiStateMachine( 1917): DriverFailedState

edit again - no it looks like they didn't even load lsmod shows nothing - the issue was they were compiled against a different kernel version so I wonder if there is a way to force it or if I'm using the wrong files - or maybe we need to recompile it with his changes against our kernel - hopefully it's possible to do that without kernel source - or hopefully I just tried the wrong files I don't know argh


moar edits-

I tried all the modules I could there and insmod file -f which got me no where, so how do we compile a kernel module without kernel source code? Is it possible?

This post has been edited by randomblame: 17 October 2011 - 12:58 AM

[randomblame]

Well I uploaded the latest version fixes backlight and adds custom ramdisk and the real honeycomb netdaemon at last. Gotta keep grinding away at it.

I had an idea about the modules - if we could decompile the kernel - a big if - we could recompile it with no changes (it's hard to understand without comments and human language) and use the decompiled source to compile modules. I know THAT is possible - but how much of a pain in the ass will it be remains to be seen. I'm having trouble finding information about how to decompile a linux kernel- probably because no one ever has to. Pandigital can suck my keyboard.

[terminander]

When you try to load a module that's not compiled to your kernel, you get a vermagic mismatch error. Check dmesg. This version is the kerel version it was compiled with. Using a hex editor you you can change the ver in the driver. This should work for same kernel, but if the kernel is lower you will probably get linker unresolved symbol errors. I have tried to load drivers from my zt180 2.2 and thats what I get.

The vermagic for the S20 kernel is 2.6.29 mod_unload ARMv6 and should be the same for S10.

If you have driver sources. Build a generic 2.6.29 kernel setting the version to above. Then build the modules. I have played a bit with this, but I'm too old to learn all this stuff. There may be sources and firmware updates for our wifi (linux), but I'm a long way from playing with that.


Drivers are why our wifi is flacky, but not why there not working, maybe??

This post has been edited by terminander: 17 October 2011 - 05:20 AM

[terminander]

Here is the kernel msg when loading wifi drivers in cm6, where they work. I remember seeing this in honeycomb too.

 rt3070sta: module license 'RALINK' taints kernel.
 usb1 init
 rtusb init --->
 usbcore: registered new interface driver rt2870
 usb 1-1: new full speed USB device using s3c2410-ohci and address 2


 
 === pAd = cd302000, size = 930864 ===
 
 <-- RTMPAllocAdapterBlock, Status=0
 ra0 (usb): not using net_device_ops yet

I tried using ifconfig to up the adapter and it took a while and worked?? But reported ip 0.0.0.0, which makes sense. I should have loaded a good wpa_supplicant.conf in data/misc/wifi. Without the WiFi interface there is no other way to set SSID etc.