Root Overclocking + CM7 on Triumph

[sevansup]

Hi, I'm a little confused about everything I've read and I'm wondering if someone can help clear it up for me.

I have the newest version of tickerguy's CM7 build (vB.04) and I'm looking for a way to overclock that isn't going to break things within CM7. I've read about Isaac's kernel, but I'm still not sure if and how well it works. And what about the umph kernel? I can't seem to find anywhere if that will work or not for CM7.

Or is there another kernel that I'm not aware of that will do the trick? I understand overclocking is not always safe, but I was only planning on going to about 1.3 gHz. Thanks in advance for the helpful replies.

 

[Chairshot215]

Hi, I'm a little confused about everything I've read and I'm wondering if someone can help clear it up for me.

I have the newest version of tickerguy's CM7 build (vB.04) and I'm looking for a way to overclock that isn't going to break things within CM7. I've read about Isaac's kernel, but I'm still not sure if and how well it works. And what about the umph kernel? I can't seem to find anywhere if that will work or not for CM7.

Or is there another kernel that I'm not aware of that will do the trick? I understand overclocking is not always safe, but I was only planning on going to about 1.3 gHz. Thanks in advance for the helpful replies.

All I can help you out with is that the umph kernel is a no go. I believe I read that Isaac's kernel is a no go with the new beta releases of cm7 but not 100% on that but would still not do it untill reading otherwise. For the amount your looking to overclock I wouldn't spend that much time looking as the Rom outperforms stock ramped up to 1.9 without going above 1.2.

 

[agentc13]

The latest CM7 updates (since Isaacs OC kernel) have had some kernel level changes. It would work but will break things like FFC and probably the new wifi changes. Umph wont work and neither wil b_randons. They are stock kernels. I think there is someone working on a 2.3 (cm7 and miui) kernel that will allow OC, but I sould wait till those ROM's aren't having as many changes so it wouldn't break fixes.

 

[tickerguy]

If you're really hellbent to do it you can pull my kernel git off Github and stick the frequency tables in there you want, then recompile it. If you track upstream you'll be able to stay up with what I'm doing with it as well.

I'm not going to support it in my CM7 build as the risks are in my view just not worth it.

 

[ziggy46]

If you're really hellbent to do it you can pull my kernel git off Github and stick the frequency tables in there you want, then recompile it. If you track upstream you'll be able to stay up with what I'm doing with it as well.

I'm not going to support it in my CM7 build as the risks are in my view just not worth it.

@tickerguy- can you give me some info on building a kernel from your repo? i have just succesfully added govoners and frequencies to b_randons souce,compiled,and flashed. it all went perfectly! so i went to look at your and it looks slightly different. is there anything that i should specifically look for when attempting to build from your repo?

 

[tickerguy]

It's just the source directory. Pull it down, stick it in the appropriate place (the directory called "kernel" ;-)) set up the environment (architecture and eabi path) and type "make."

That's it.

 

[ziggy46]

i get these errors when trying to build with nothing modified...

scripts/kconfig/conf -s arch/arm/Kconfig
drivers/char/ftm_smd/Kconfig:4:warning: leading whitespace ignored
init/FIH_Kconfig:184:warning: choice value used outside its choice group
init/FIH_Kconfig:186:warning: choice value used outside its choice group
drivers/usb/gadget/Kconfig:805:warning: defaults for choice values not supported
  CHK     include/linux/version.h
make[1]: `include/asm-arm/mach-types.h' is up to date.
  CHK     include/linux/utsrelease.h
  UPD     include/linux/utsrelease.h
  SYMLINK include/asm -> include/asm-arm
  CC      kernel/bounds.s
cc1: error: unrecognized command line option "-mlittle-endian"
cc1: error: unrecognized command line option "-mapcs"
cc1: error: unrecognized command line option "-mno-sched-prolog"
cc1: error: unrecognized command line option "-mabi=apcs-gnu"
kernel/bounds.c:1: error: bad value (armv5t) for -march= switch
kernel/bounds.c:1: error: bad value (strongarm) for -mtune= switch
make[1]: *** [kernel/bounds.s] Error 1
make: *** [prepare0] Error 2

 

[tickerguy]

What toolchain are you using?

 

[ziggy46]

What toolchain are you using?

the newest linaro... which are you using?

 

[tickerguy]

I pulled the Android SDK on Ubuntu and use that - the specific "setup" commands I use to init the toolchain before typing "make" are:

#!/bin/sh
export ARCH=arm
export CROSS_COMPILE=~/myandroid/prebuilt/linux-x86/toolchain/arm-eabi-4.4.0/bi
n/arm-eabi-

Note that "make clean" doesn't work on Motorola's kernel tree (and mine as a result); they totally hosed it up (isn't that nice?) so you need to do a "find . -name '*.o' -exec rm {} \;" for a complete clean build. One of these days I'll take the time to run through the makefiles and fix that.

 

[isaacj87]

the newest linaro... which are you using?

If you have the CM source pulled, you have the toolchain both TG and I are using to build kernels.

 

[ziggy46]

i didnt pull the entire cm source... i just pulled the kernel source.

 

[BigEarsDavid]

I have rooted and installed gingerbread 2.3.2 on my HTC Eris and it will not go past the android screen, what do i need to do to fix. I have restored it to my last backup

 

[ziggy46]

I have rooted and installed gingerbread 2.3.2 on my HTC Eris and it will not go past the android screen, what do i need to do to fix. I have restored it to my last backup

wrong section... and im not familiar with htc phones, sorry

 

[tickerguy]

i didnt pull the entire cm source... i just pulled the kernel source.

You need the Android SDK - either by pulling CM7 or by pulling the SDK from Google. The cross-compilation environment has to be on your machine or it won't work - you're compiling for an ARM processor, not an Intel one.

 

[mantera]

I pulled the Android SDK on Ubuntu and use that - the specific "setup" commands I use to init the toolchain before typing "make" are:

#!/bin/sh
export ARCH=arm
export CROSS_COMPILE=~/myandroid/prebuilt/linux-x86/toolchain/arm-eabi-4.4.0/bi
n/arm-eabi-

Note that "make clean" doesn't work on Motorola's kernel tree (and mine as a result); they totally hosed it up (isn't that nice?) so you need to do a "find . -name '*.o' -exec rm {} \;" for a complete clean build. One of these days I'll take the time to run through the makefiles and fix that.

So, comparing your kernel folder/files to Isaac's, I see that yours is missing the arch/um and Documentation folders which contains arch/um/scripts/Makefile.rules and Documentation/Makefile which apparently the "make clean" commands uses? If I copy those folders in to your kernel files, the "make clean" command works.

Also, I'm no linux expert but for some reason, when I try running those export commands in the shell script as part of the script, they don't seem to take. But when I manually run them in the command prompt, they work.

Although after building your files that I pulled this morning, the kernel doesn't boot for me. I guess I must be doing something wrong.

 

[b_randon14]

I use make mrproper on mine everyday. And I believe make clean works on my source. Also you do not have to have the sdk if your only building a kernel. You just need toolchain(I use linaro) and mine and my brothers sources has an easy to usr build script. Its not for cm ron's though!

 

[b_randon14]

The reason why the export commands won't work is because only the child processes started by that script will have access to those variables. Wherss if uoi export this in the terminal then anything executed from the terminal until you close it will have access. That why in mine and my brothers sources our build script export arm and our toolchain then starts up as many instances of make as it can so that make will have access to the exported variables and plus the kernel will build faster because we run multiple instances of make at one time.

 

[mantera]

The reason why the export commands won't work is because only the child processes started by that script will have access to those variables. Wherss if uoi export this in the terminal then anything executed from the terminal until you close it will have access. That why in mine and my brothers sources our build script export arm and our toolchain then starts up as many instances of make as it can so that make will have access to the exported variables and plus the kernel will build faster because we run multiple instances of make at one time.

That makes sense. I should have thought of that. Doh! Thanks!

 

[tickerguy]

That's a script called "setup" that I have; I run it with:

. setup

(this runs it in the present shell context)

Then I do a "make -j4"

Which tells the system to parallel up to 4 particular pieces at once. In general using however many cores you have helps, beyond that you're likely doing no good and may do harm in terms of speed.

 

[b_randon14]

yeah thats how our script does it as well. i must have overlooked that script in your source!

 

[tickerguy]

I don't have the script in the source tree ;-)

 

[b_randon14]

Ah I see. That makes sense lol!!

 

[isaacj87]

i didnt pull the entire cm source... i just pulled the kernel source.

If you just need the right toolchain, you can find it here: https://github.com/CyanogenMod/android_prebuilt

You can just pull that if you don't feel like grabbing the entire CM source. Make sure you have git installed:

git clone git://github.com/CyanogenMod/android_prebuilt.git