This is the mail archive of the
libffi-discuss@sourceware.org
mailing list for the libffi project.
Re: [RFC][PATCH 0/3] Complex type support (FFI_TYPE_COMPLEX)
- From: Dominik Vogt <vogt at linux dot vnet dot ibm dot com>
- To: libffi-discuss at sourceware dot org
- Date: Mon, 29 Sep 2014 11:24:42 +0100
- Subject: Re: [RFC][PATCH 0/3] Complex type support (FFI_TYPE_COMPLEX)
- Authentication-results: sourceware.org; auth=none
- References: <20140722092734 dot GA23937 at linux dot vnet dot ibm dot com> <87y4teh9c6 dot fsf at moxielogic dot com>
- Reply-to: vogt at linux dot vnet dot ibm dot com
On Sat, Sep 20, 2014 at 06:31:21AM -0400, Anthony Green wrote:
> I'm not all that familiar with how GCC handles complex types on other
> architectures, but hopefully they are as straight forward as on s390.
I think the real issue here is the Abi. On x86, complex types
seem to be treated exactly like structs with two elements, but on
s390[x] that is not the case (structs of size 8 are passed in a
register, but complex types are always passed on the stack).
Other than that, complex types are hopefully treated the same
inside gcc.
> It would be nice if the complex test cases turned into expected failures
> for platforms that don't have the back end support.
Sorry for the late reply, but I've had a cold. Have you already
resolved this issue in that patch here?
Only run the complex type tests on supported platforms.
If there's something left to do, I can look into it now.
Ciao
Dominik ^_^ ^_^
--
Dominik Vogt
IBM Germany