| commit | caa16eabb4f319b8710e2379e138696b1f1d0232 | [log] [tgz] |
|---|---|---|
| author | Eli Friedman <efriedma@codeaurora.org> | Thu Jun 14 22:58:48 2018 +0000 |
| committer | Eli Friedman <efriedma@codeaurora.org> | Thu Jun 14 22:58:48 2018 +0000 |
| tree | f6910bcffe7219aa4b1fa2f5061ca9e0eb2919c3 | |
| parent | a94df7d526ce45abe6b10f04b4a73a3aa3bef0a4 [diff] |
Make uitofp and sitofp defined on overflow. IEEE 754 defines the expected result on overflow. As far as I know, hardware implementations (of f16), and compiler-rt (__floatuntisf) correctly return +-Inf on overflow. And I can't think of any useful transform that would take advantage of overflow being undefined here. Differential Revision: https://reviews.llvm.org/D47807 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@334777 91177308-0d34-0410-b5e6-96231b3b80d8