Crash invoke()'ing Method with "double" as return type

Wed Jun 28 19:17:00 GMT 2006

> Martin> And that did not change anything - it still fails invoking
> Martin> "ffi_call_SYSV". Maybe there is a similar buffer overflow on
> Martin> PPC32? Can anybody else verify/try this on PPC32? Maybe I'm
> Martin> just not doing it right...
>> One thing you could try is extract libffi from svn trunk, then build
> it for your platform. This isn't totally easy due to the configure
> situation, but you can make it work "somehow". Then, run the libffi
> test suite from trunk. If that works better, you can backport the
> new libffi to your branch.
>> If that doesn't work then you've found a bug. You'd probably need a
> MIPS expert to look at the assembly in this case.

But I already did (at least I think I did) fetch libffi from svn trunk:
http://gcc.gnu.org/viewcvs/trunk/libffi/
I even double fetched ffi.c and sysv.S and diff'ed them against my 
source (just to make sure).
Next, I printf-filled ffi.c in order to see that it indeed did get into 
my libgcj. Finally, I made a deliberate error in sysv.S, as well, to see 
that things wouldn't compile.
(Had to fetch "http://gcc.gnu.org/viewcvs/trunk/libffi/depcomp", as 
well, to make things work, but that seem to be it - along with running 
config.status)
So I'm rather convinced that I tried libffi from svn trunk?!
// Martin