Re: [PATCH 6/6] angstrom: use GCC 4.3.3 for Geode devices

[Koen Kooi <k.kooi@student.utwente.nl>]

On 13-05-09 17:18, Tom Rini wrote:
> On Wed, May 13, 2009 at 04:56:58PM +0200, Koen Kooi wrote:
>> On 13-05-09 16:46, Marcin Juszkiewicz wrote:
>>> From: Marcin Juszkiewicz<hrw@openembedded.org>
>>>
>>>
>>> Signed-off-by: Marcin Juszkiewicz<hrw@openembedded.org>
>>> ---
>>>    conf/distro/angstrom-2008.1.conf |    1 +
>>>    1 files changed, 1 insertions(+), 0 deletions(-)
>>>
>>> diff --git a/conf/distro/angstrom-2008.1.conf b/conf/distro/angstrom-2008.1.conf
>>> index a466834..b5dbc48 100644
>>> --- a/conf/distro/angstrom-2008.1.conf
>>> +++ b/conf/distro/angstrom-2008.1.conf
>>> @@ -101,6 +101,7 @@ ANGSTROM_GCC_VERSION_bfin			= "4.1.2"
>>>
>>>    ANGSTROM_GCC_VERSION_486sx      = "4.3.3"
>>>    ANGSTROM_GCC_VERSION_armv7a     = "4.3.3"
>>> +ANGSTROM_GCC_VERSION_geode      = "4.3.3"
>>>
>>>    # Uncomment this if want need to build an armv7a kernel with CSL toolchain (<2.6.27 don't boot with mainline gcc)
>>>    #KERNEL_CCSUFFIX_armv7a= "-4.2.1+csl-arm-2007q3-53"
>>
>> NAK, this would interfere with an outstanding patch that Graeme and I
>> are testing:
>>
>> http://dominion.thruhere.net/koen/OE/angstrom-dev-toolchain.diff
>
> wrt that patch, is there a reason you aren't bumping ppc as well?  As an
> old PPC guy, that's always had me perplexed..

When adding PPC support I had loads of troubles compiling gcc and glibc 
due to sqrt and type size problems.
My efika is too busy serving stuff to serve as a testbed and my 
dht-walnut lacks a psu currently, so I'm a bit apprehensive about 
bumping toolchains without runtime testing them.
If console-image actually builds with gcc 4.3.3 + binutils 2.18, that 
would be a *huge* step forward.

regards,

Koen