Re: libffi fails to build on powerpc64-linux

[Kyle Moffett <kyle at moffetthome dot net>]

On Thu, Mar 8, 2012 at 18:37, David Edelsohn <dje.gcc@gmail.com> wrote:
>> On Tue, 2012-03-06 at 09:34 -0600, Peter Bergner wrote:
>> > > This passes the testsuite on soft-floating-point PowerPC, and it builds
>> > > and passes the testsuite on PowerPC e500 systems which cannot even
>> > > assemble the regular floating-point instruction set.
>> >
>> > Both of these are non FP power systems, so I can see why he didn't
>> > see these build errors, since the broken code is inside of
>> > "#ifndef __NO_FPRS__" (double negative?).
>>
>> Looking at the bugzilla further, I noticed Kyle has a blog here that
>> mentions he has moved to Google, so that explains why his email address
>> was bouncing. ÂI'm also guessing he's no longer interested in PowerPC
>> issues now that he's no longer at Boeing.
>>
>> That said, looking farther down that debian bugzilla, I noticed someone
>> complained about his patch breaking the debian build. ÂHe attached a
>> patch:
>>
>> Âhttp://bugs.debian.org/cgi-bin/bugreport.cgi?msg=40;filename=libffi-powerpc-build-fix.patch;att=1;bug=644338
>>
>> If you ignore the debian specific changes and look at the src/powerpc/ffi.c
>> change, you'll see it's the same as my patch, minus the hunks of my patch
>> that are silencing warnings, so I think my patch is probably safe to take
>> as is. ÂTherefore, I'm officially submitting the patch below to fix the
>> build issues.
>>
>> That still leaves the 32-bit many.c testsuite breakage I am seeing, but I
>> think that should be fixed in a separate patch since it is not related to
>> these changes at all.
>
> Adding a different email address for Kyle to possibly entice him to
> participate in this email thread.

Ah, hello there!  I'm afraid that I don't have any e500 systems to
test with anymore, but I will certainly say that the second fixup
patch I submitted to Debian is currently part of their official stack
of patches, and I built and tested it on powerpc, powerpc64, and
powerpcspe after having that problem with the first one.  I'm very
sorry for not following up and forwarding the fixed patch upstream
properly.

I believe Jeff Runningen is the current eXMeritus point-of-contact for
the Debian e500 stuff, although he probably hasn't had to touch libffi
yet and most likely doesn't have the time unless it's terminally
broken. :-D  I've CCed him anyways in the hope that he will have this
background in the event that he runs into problems in the future.

My apologies for the trouble, good luck!

Cheers,
Kyle Moffett