This is the mail archive of the
java@gcc.gnu.org
mailing list for the Java project.
Re: Fwd: Status of gcj/java on PPC Linux?
- To: khendricks at ivey dot uwo dot ca
- Subject: Re: Fwd: Status of gcj/java on PPC Linux?
- From: Tom Tromey <tromey at redhat dot com>
- Date: 08 Feb 2001 17:52:09 -0700
- Cc: Mark Wielaard <mark at klomp dot org>, java at gcc dot gnu dot org
- References: <200102081755.f18Htq923345@ashley.ivey.uwo.ca>
- Reply-To: tromey at redhat dot com
>>>>> "Kevin" == Kevin B Hendricks <khendricks@ivey.uwo.ca> writes:
Kevin> Okay one more thing. I don't want my patches to break anything
Kevin> for any other POWERPC port. Unfortunately the trampoline sizes
Kevin> and code and how parameters are passed are different for
Kevin> different AIX and Linux PPC.
Ok.
Kevin> I need to define some things in ffi.h.in to turn on closures
Kevin> and set the trampoline size and don't want to simply use the
Kevin> libffi defined(POWERPC) for fear it breaks AIX.
You could just enable it and then call FFI_ASSERT in the closure code
to force a failure on AIX. (I'm partially guessing here; I don't know
all that much about libffi.)
Another approach would be to add a new #define to ffi.h.in that says
something about the ABI. Or, you could introduce a new target like
"POWERPCLINUX" (or whatever is appropriate). This latter approach
might require more hacking though.
Kevin> Will defined(__linux__) && defined(__powerpc__) work or is
Kevin> there something like defined(LINUX) && defined(POWERPC) you
Kevin> would rather I used.
Is it just Linux or is it a whole class of systems with a different
name? In the POWERPC section I see two ABIs: FFI_SYSV and
FFI_GCC_SYSV. Which is the `Linux' one?
Tom