Re: [Ltt-dev] [PATCH 00/09] atomic.h : standardizing atomic primitives

Andrew Morton wrote:
> On Sat, 27 Jan 2007 21:09:11 +0100
> Willy Tarreau <[email protected]> wrote:
> 
>> On Sat, Jan 27, 2007 at 12:05:12PM -0800, Andrew Morton wrote:
>>> On Sat, 27 Jan 2007 13:11:16 -0500
>>> Mathieu Desnoyers <[email protected]> wrote:
>>>
>>>> I am currently trying crosstool by Dan Kegel, it looks promising.
>>>> http://www.kegel.com/crosstool/
>>> Yeah, I spent a frustrating two days with crosstool, managed to eke a
>>> number of cross-compilers out of it, but it took a *lot* of experimentation
>>> with gcc, glibc and binutils versions to get combinations which actually
>>> work.  Good luck ;)
>>>
>>> There used to be someone who had a full suite, and who regularly published
>>> cross-compile results, but he stopped 6-12 months ago and I forget who that
>>> clever person was?
>> Wasn't it buildroot from Erik Andersen ?
>>
>>    http://buildroot.uclibc.org/
>>
> 
> No, it was http://l4x.org/k/ It still appears to be operating, with
> scary-looking results.
> 
> Jan, is there any way in which you can help us publish a full suite of
> cross-compiler binaries?

Probably not. I could publish a qemu i386 image with all cross compilers
though. But some are not build from source but are obtained from more or
less obscure sources (m32r, sh64). Currently this

 CHK     include/linux/utsrelease.h
"2.6.20-rc6cat:include/config/kernel.release:Nosuchfileordirectory" exceeds 64 characters
make[1]: *** [include/linux/utsrelease.h] Error 1
make: *** [_all] Error 2

bug, which I reported weeks ago, makes the result invalid for most
archs. But as I get nearly zero feedback about the results and I've
lots of other obligations currently, my motivation to work on that is
pretty much nil.

Jan