Hi Kiko,

Yes, as Bero said, the busybox compiling will be failed with the old tool chain if we don't add those "-I" options manually.

But thank you to alert me that I have to correct a thing which is if you want to compile your platform source code with our latest RC tool chain: https://android-build.linaro.org/builds/~linaro-android/toolchain-4.6-2011.08/#build=6, please use branch linaro_android_2.3.4 of my repository, and it's already in the manifest, you don't need to modify a thing. For a previous build of tool chain: https://android-build.linaro.org/builds/~linaro-android/toolchain-4.6-2011.08/#build=5, you can use the branch linaro_android_2.3.5.

In the build=6, the compiler prefix has been restored to arm-eabi, and it will cause compiling error as Bero said. I'm trying to solve this with Bero. But in the build=5, the prefix is arm-linux-androideabi, so everything is fine.

Thanks.


BR
Botao Sun

On Mon, Aug 22, 2011 at 9:30 PM, Bernhard Rosenkranzer <bernhard.rosenkranzer@linaro.org> wrote:
On 22 August 2011 05:56, Christian Robottom Reis <kiko@linaro.org> wrote:
> On Fri, Aug 19, 2011 at 11:45:47AM -0700, Zach Pfeffer wrote:
>> On 19 August 2011 09:36, Botao Sun <botao.sun@linaro.org> wrote:
>> > Currently, we have 2 different situations according to our 2 different tool
>> > chains - old and new.
>
> What do old and new mean here?

old: Toolchain targeting generic arm-eabi
new: Toolchain targeting arm-linux-androideabi

The difference is that the "old" toolchain doesn't have any
understanding of Android internals (stuff like crtbegin.o vs. crt0.o)
while the "new" toolchain does -- so "gcc -o busybox *.o" does the
right thing with the "new" toolchain, but complains about missing
crt0.o and friends with the "old" toolchain.

The fix is to add the -nostdlib /path/to/crtbegin.o /path/to/libc ...
mess Android's build system adds everywhere while we're using the
arm-eabi-* toolchain.

ttyl
bero

_______________________________________________
linaro-dev mailing list
linaro-dev@lists.linaro.org
http://lists.linaro.org/mailman/listinfo/linaro-dev