RFA: Add support for DWARF-based fixed point types

[Simon Marchi]

On 2020-11-08 1:30 a.m., Joel Brobecker wrote:
> Hello,
>
> This patch series adds support for fixed point types described
> using a pure DWARF approach. While the main trigger was the desire
> to deprecate the GNAT encodings used in the case of Ada fixed point
> types, the implementation is done in a way that support is generic.
> Other languages which have support for such kinds of types should
> have little work to do to support them as well.
>
> The implementation is based on top of GMP: We use GMP to store and
> operate on rational values, with the added bonus that there is no
> size limit on either the numerator nor denominator. As discussed
> previously, the dependency on GMP is made mandatory
> (https://sourceware.org/pipermail/gdb-patches/2018-March/147373.html).
>
> One part that this patch series does not include is function calls
> and returns. Since those changes are all arch-specific, I will
> submit them in a separate series of its own, to avoid those other
> patches getting lost in this patch series.
>
> Each patch of this series has been tested on x86_64-linux using
> our testsuite. In addition, these changes have been tested using
> AdaCore's testing, on numerous architectures and OSes.
>
> Looking forward to everyone's feedback!

Hi Joel,

I gave some comments that are more on the coding style side of things,
since I am that that familiar with fixed point types.  In general it
seemed to make sense.  I'm fine with you merging this series, addressing
the comments as you see fit.

Simon