[KERNEL] Buzz Vega | .32.35 | CIFS | TUN | EXT4 | SLAB | NOOP | WiFi | NTFS | HOST/SLAVE | 0.7

[Guest nickwarn]

just wanted to let every one know it plays nice with the latest firmware from POV

[Guest MaxiP]

Just flashed and lost CIFS support :(

This is a must have for me as all my media on my NAS - I know you will have an update out soon but can I reverse what this ROM has done to CIFS for now ?

[Guest eriksp]

Is cifs manager pointing to the correct location of the module ?

Thanks for pointing this out. Enabled module loading via insmod in cifsmanager settings and everything works!

[Guest Buzing Bee]

Nice

Glad to here its all working :(

[Guest Fuzzysteve]

Wooo!

Thanks for adding the Tun Module.

VPN Connections is working well, for setting up a pure IPSEC vpn connection. (it has a minor screwup with dns when it disconnects, but that's in the bug list. Just need to cycle wifi to fix it)

http://code.google.com/p/get-a-robot-vpnc/

[Guest MaxiP]

Thanks for pointing this out. Enabled module loading via insmod in cifsmanager settings and everything works!

Enabled insmod and cifs share is now mounting correctly. Only tine will tell what improvements there are for me - so far, browsing with dolphin hd feels slower but I'll give it a few days.

Thanks

[Guest TerryH]

I am on Vega 1.09 + Original Ganjamans Addon for market and it works really good, but need the USB-Host working for the 3G-Dongle.

Just need the USB-Host - Anybody tested this already with the 1.09?

[Guest the_corvus]

Yesterday i had problem with bluetooth too... it conects correctly, but when i try to disconnect i have 3 scenarios:

1) Desconnect correctly, but imposible to reconect again with any device

2) Vega stalls and must restart with power button

3) Vega reboots

I wants to remember to everybody that you need to load the modules that you want to use... they are at /system/lib/modules/, so if you wants cifs, in a terminal (connectbot or similar) do: insmod /system/lib/modules/cifs.ko

Once done, you can use cifs, or ntfs (tested, but no automount, must be mounted at hand), or any module...

Corvus.

[Guest MaxiP]

An observation that may (or may not) be linked to the Buzz ROM

After re-building (1.09, MCR r8 & Buzz 0.3), the battery meter is very flakey and only updates after a reboot. I remember some time ago there was an issue with this and various builds so not 100% sure if Buzz is at fault. However, previous buiold (1.09 & MCR r8) didn't exhibit the same problem.

[Guest TerryH]

An observation that may (or may not) be linked to the Buzz ROM

After re-building (1.09, MCR r8 & Buzz 0.3), the battery meter is very flakey and only updates after a reboot. I remember some time ago there was an issue with this and various builds so not 100% sure if Buzz is at fault. However, previous buiold (1.09 & MCR r8) didn't exhibit the same problem.

MCR r8 is a full Rom 1.08 with the modifications from Paul. So if you flash 1.09 and then flash R8 you have 1.08...

http://android.modaco.com/content/advent-v...nline-kitchen/#

[Guest theatomicfart]

Any ETA for the BSOD fix?

[Guest olionair]

i installed this kernal then i booted normally and it booted fine but after an attempt to boot into host it will not boot at all

any ideas?

[Guest simonta]

Any ETA for the BSOD fix?

There's a nice simple fix over on XDA devs. Reformats the misc partition so recovers from BSOD without a flash.

[Guest Buzing Bee]

I'll have the BSOD of death prevention out today.

Just out of curiosity how are people booting to host mode :-

Holding power and home all the time.

Holding power until backlight comes on then release power and hold back

I ways use the second way and iv never got the black screen of death

[Guest Richard_Arkless]

I havent had the vega yet but im wondering if your thinking about overclocking the kernel, I think it will get upto 1.1-1.2ghz, I know its pointless as im sure it is very fast by default but still its great to get the most performance out of it

Edited March 12, 2011 by Richard_Arkless

[Guest theatomicfart]

There's a nice simple fix over on XDA devs. Reformats the misc partition so recovers from BSOD without a flash.

Found the thread with the fix. Many thanks for that, should save a bit of time if it happens again. All I want now is for some one with a considerably bigger brain than mine to post a way of booting straight into CWM..... :-)

[Guest Buzing Bee]

Main problem is that our boot-loader doesn't allow it ....

[Guest Zebrahead]

Isn't using the back button whilst booting for Host mode kinda dangerous? Seems rather close to putting it into nvFlash mode!

Wouldn't it be better off with vol down or something?

None the less, thanks ^^

Zeb

[Guest Buzing Bee]

Vol + or - causes some srange bugs (like other GPIO pins not working)

Really using back while booting is not bad as there is a very specific route to get into nvflash

:D

[Guest PASCHA]

I'll have the BSOD of death prevention out today.

Just out of curiosity how are people booting to host mode :-

Holding power and home all the time.

Holding power until backlight comes on then release power and hold back

I ways use the second way and iv never got the black screen of death

There some problems this kernel.

1. Wifi down and give me message "fault " and dont get back.

2. Ext2 partition dont work under usb. Host mode.

3. I cant. Use ext4 newer.!!!!! I dont see it with any file manager

[Guest craskman]

Hi buzz,

I have been playing around with your sources/kernel. IMHO is the best source we have for vega kernel atm.

I was specifically playing around on getting a little extra, so I have made few changes (you might have a try in case you think it helps)

1) In the [.config] I have set NO to almost all debug properties. As enduser we wont be looking into these messages anyway and we might speedup the things a bit.

2) When compiling the kernel, I have set the same optimisations as pershoot did on gtab kernel. On the main Makefile I have set the AFLAGS and CFLAGS with [-O3 -mtune=cortex-a9 -ftree-vectorize -ffast-math -fsingle-precision-constant]. I think pershoot is using -O1 and codesourcery cross compiler, I am using google's NDK crosscompiler. (I think -O3 is enough, all directives are included on O3 optimisations anyway)

3) In the config file I have raised the preempt HZ from 100 to 200 (as I found on Paul's kernel)

I am still not sure if NDK cross compiler ignores the optimisation levels or the cortex-a9 directive, but everything compiles successfully.

In the end, I didn't find diferences in quadrant, I am getting always between 2100 and 2400 (seems quadrant is not consistent).

Maybe is placebo but I feel the whole thing snappier, including heavy sites on dolphin, flash content and interaction/transitions on mostly of the apps.

But as I have said, maybe is placebo.

Have you try some optimisations? or a different cross compiler?

btw, you usb mode switch trick is spot on:)

CrK

[Guest Buzing Bee]

Cheers for the feedback ... iv been playing around with optimization and the extra build flags and haven't found anything conclusive yet. Like u said benchmarks dont change so you have the risk of getting a placebo "it seems faster but no way to prove" :(

Im now patching up to .32 so we will have a new version soon :D

----------------------------------------------------

PS

here is my BSOD prevention / fix *stops misc from mounting at boot

please test

http://dl.dropbox.com/u/15005030/BSOD_Prevent.zip

have fun ppl :(

[Guest craskman]

Nice one.

May I ask you to create a git for you ramdisk (like paul did). It's easier and quicker to check things out and diferences.

Just curiosity, have you tried kernel .36 (honey)?

I have but something wrong in the early stage of boot (before reaches the kernel I think). I don't know much about ARM architecture, but seems that the current .36 is not initialising 'vectors'/memory addresses(?) properly for vega (shuttle boards). I reckon that some knowledge on how ARM boot process works and looking to the ''shuttle' specific code (from current vega kernel) we might put it to boot. After that we will have some challenges with specific drivers for vega, but regarding NV* stuff, I 'hope' that we can use 'almost' as-is. Even the last .38-rc supports tegra-harmony oob, so I think our issue is just the initialisation of the kernel. I checked the main initialisation addresses between .32-vega and .36-tegra and they seem the same, but surely something missing on .36 to enable the booting on vega. (but as i said, my knowledge on ARM is very slim)

Another thing is, it might me possible to run honey+.32 kernel on vega. If so, doesn't worth to spend too much time on porting the .36 kernel for vega.

What is your view on this?

[Guest Buzing Bee]

well we will no more when 3.0 hits AOSP iv been looking at porting the vega to .36 to the vega and it is a huge job to do and as things stand we dont know where its failing to load. . . . if the vega is getting 3.0 in April then its not worth trying as well it would take longer than that to port.

3.0 may run on .32 kernel with some back porting of some stuff but like i said we will know more when 3.0 drops to AOSP

[Guest Buzing Bee]

0.4 is out

See main post for changelog :D