GCC 4.9.2 Build Fails at cc1

[Cyd Haselton]

On Fri, Dec 5, 2014 at 12:35 PM, Cyd Haselton <chaselton@gmail.com> wrote:
> On Fri, Dec 5, 2014 at 3:34 AM, Andrew Haley <aph@redhat.com> wrote:
>> On 05/12/14 00:08, Cyd Haselton wrote:
>>> On Thu, Dec 4, 2014 at 5:01 PM, Andrew Haley <aph@redhat.com> wrote:
>>>> On 12/04/2014 09:00 PM, Cyd Haselton wrote:
>>>>> Why is make complaining about those functions when they aren't being
>>>>> referenced? Are the functions supposed to be transformed in some way?
>>>>
>>>> The names are being munged, like this:
>>>>
>>>> #ifndef __MPN
>>>> #if __GMP_HAVE_TOKEN_PASTE
>>>> #define __MPN(x) __gmpn_##x
>>>> #else
>>>> #define __MPN(x) __gmpn_/**/x
>>>> #endif
>>>> #endif
>>>>
>>>> #define mpn_tdiv_qr __MPN(tdiv_qr)
>>>>
>>>> I have no idea why.  But it might have some relevance for your problem.
>>>
>>> Any chance this started in GCC 4.9.x?  If not, it's back to the code-parsing...
>>
>> I don't know.  Hint: compile with -save-temps.  Look at the preprocessed
>> source.
>>
>> Andrew.
>>
>>
> I'm still trying to find which Makefile to add -save-temps to in order
> to output the intermediate files (gmp/Makefile, mpfr/Makefile and
> gcc/Makefile yield nothing).

Update:  Ignore the following:

> However, while waiting for another run of make to fail, I did an ls
> -alt in the build directory and, according to the timestamps, it
> appears as though mpfr is being configured and built before gmp.
> Shouldn't this be the other way around, and could this possibly cause
> the errors I'm seeing?

running make distclean && make to start everything from scratch showed
gmp being built before mpfr.