projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: b556726) | patch
[flang][runtime] Fixed identity value for REAL(16) == __float128.
authorSlava Zakharin <szakharin@nvidia.com>
Thu, 22 Sep 2022 22:45:10 +0000 (15:45 -0700)
committerSlava Zakharin <szakharin@nvidia.com>
Wed, 28 Sep 2022 17:11:37 +0000 (10:11 -0700)
commit8985cfd93980a1d7d92699bfefce34e7bf5887fb
tree83f9bd4bf6d299412846f882b705bf0ede4c612ctree | snapshot
parentb556726ccc5670637e84f1b26ef7e998c94f1d42commit | diff
[flang][runtime] Fixed identity value for REAL(16) == __float128.

std::numeric_limits<__float128>::max/lowest return 0.0, so recreate
value of FLT128_MAX ourselves to avoid using quadmath.h's FLT128_MAX
that is currently causes warnings with GCC -Wpedantic.

Differential Revision: https://reviews.llvm.org/D134496
flang/runtime/extrema.cpp diff | blob | history
flang/unittests/Runtime/Reduction.cpp diff | blob | history
Domain: System / Toolchain;