[llvm-commits] [patch] fp128 sint_to_fp folding fix

[Renato Golin]

On 7 January 2013 17:25, Duncan Sands <baldrick at free.fr> wrote:

> I don't understand the question: there is no choice, there is no need for a
> default.  The type (ppc/ieee) is in the node type already, ppcf128 or f128
> when talking about codegen value types, PPC_FP128Ty or FP128Ty at the IR
> level.
> It's only the front-end (clang) that has to decide what type to output to
> the
> IR.  After that that everything is completely specified.
>

The current code defaults to creating a PPC long double in the constructor,
which is probably there for historical reasons.

In this case, a new value is created (a NULL) which has to have a type that
is compatible with the platform in which it's running. Maybe each platform
could have an option on which kind of long double they support, that you
use to create new APFloats?

--renato
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130107/8c7753ef/attachment.html>