[PATCH] D63793: Treat the range of representable values of floating-point types as [-inf, +inf] not as [-max, +max].

Mon Jul 8 02:52:18 PDT 2019

MaskRay added a comment.

In D63793#1572977 <https://reviews.llvm.org/D63793#1572977>, @ostannard wrote:

> There are a number of buildbot failures which look related to this, e.g.
>
> - http://lab.llvm.org:8011/builders/clang-cmake-armv7-full/builds/6730/steps/ninja%20check%201/logs/FAIL%3A%20UBSan-AddressSanitizer-armhf%3A%3Adiv-zero.cpp
> - http://lab.llvm.org:8011/builders/clang-cmake-armv7-full/builds/6730/steps/ninja%20check%201/logs/FAIL%3A%20UBSan-AddressSanitizer-armhf%3A%3Adiv-zero.cpp
>
>   I think that removing the float-divide-by-zero sanitizer from -fsanitize=undefined has also caused it to be marked as not supported, so it can't be enabled even explicitly.

Should be fixed by rCRT365307 <https://reviews.llvm.org/rCRT365307>.

Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D63793/new/

https://reviews.llvm.org/D63793