Fixes for Darwin PPC32

[evanmiller]

This PR consolidates some of @kernigh's work over in #211. It successfully compiles and runs the Fibonacci test, without a crash, on PPC32 hardware running Darwin 8:

$ ./a.out
0 1 1 2 3 5 8 13 21 34
main: done
$ uname -r
8.11.0

For make_fcontext, use the diff provided here:

#211 (comment)

For ontop_context, adapt the Linux PPC32 fixes from here:

df8fb6b

[barracuda156]

@evanmiller Do you think we should first ensure that fiber does not fail with these changes? Given that is was building before with not fully fixed context, it presumably should build with the properly fixed one.

[evanmiller]

@evanmiller Do you think we should first ensure that fiber does not fail with these changes? Given that is was building before with not fully fixed context, it presumably should build with the properly fixed one.

@barracuda156 Which compilers / Boost versions were successfully building Fiber on PPC? I encountered an issue with GCC7 / Boost 1.81 but I can't comment on any other combinations.

[barracuda156]

@evanmiller Do you think we should first ensure that fiber does not fail with these changes? Given that is was building before with not fully fixed context, it presumably should build with the properly fixed one.

@barracuda156 Which compilers / Boost versions were successfully building Fiber on PPC? I encountered an issue with GCC7 / Boost 1.81 but I can't comment on any other combinations.

Well, everything was building with my earlier minimal fixes to context (which are not sufficient, as we know now).

[evanmiller]

@barracuda156 If you were able to build everything before, why don't you try building Fiber with these proposed changes and report back?

[barracuda156]

@barracuda156 If you were able to build everything before, why don't you try building Fiber with these proposed changes and report back?

@evanmiller I will do it soonish.

[olk]

ty

[barracuda156]

Now we have ppc64 left to fix. And possibly fiber for both.

[barracuda156]

@evanmiller Could you remind, what prevented us from fixing ppc64 along?