ppc linux libffi fix plus necessary testcase

[Kevin B. Hendricks]

Hi,

Is there some reason the ffi.c fix for passing float args did not make it 
into the gcc-3.2.2 official source tarball?

A later fix by Franz for the ppc_closure.S file made it in but these 
changes did not and as a result the bug still exists when passing lots of 
floating point args on ppc.

Can we add it to the gcc-3.2 branch just to make sure it does not get 
dropped or forgotten?

Thanks,

Kevin

On January 13, 2003 03:37, Andreas Tobler wrote:
> Hi,
>
> Kevin Hendricks has a libffi fix in the pipe for ppc linux. With my test
> case extension I could verify that the fix was/is working.
> The patch for libffi can be applied from 3.2.1 and up I guess. Tested on
> trunk.
> The ffitest.diff is against the trunk ffitest. I sent this several
> times, got also an ack from Anthony. But I can't ci since I don't have
> access. Could anyone of you take care of?
>
> Thanks in advance.
>
> Regards,
> Andreas
>
> 2003-01-13 Andreas Tobler <a.tobler@schweiz.ch>
>
> 	* libffi/src/ffitest.c
>          add closure testcases
>
> 2003-01-13 Kevin B. Hendricks <khendricks@ivey.uwo.ca>
>
> 	* libffi/src/powerpc/ffi.c
>          fix alignment bug for float (4 byte aligned iso 8 byte)