diff options
| author | Koakuma <koachan@protonmail.com> | 2022-02-15 20:22:01 +0700 |
|---|---|---|
| committer | Veikka Tuominen <git@vexu.eu> | 2022-02-17 14:46:04 +0200 |
| commit | 755d116ecf2d221e2ad8b572c58a7d4f99163ff9 (patch) | |
| tree | 91cd4f2689e323750c0f539bfb60777afcd3e3c4 /test/behavior/floatop.zig | |
| parent | 02902cc099d9736d589ac4f0e0ff671d7dfa9557 (diff) | |
| download | zig-755d116ecf2d221e2ad8b572c58a7d4f99163ff9.tar.gz zig-755d116ecf2d221e2ad8b572c58a7d4f99163ff9.zip | |
stage1: use u16 for __truncxfhf2/__extendhfxf2 on non-Arm CPUs
Non-Arm CPUs use u16 as the parameter to __extendhfxf2 and the return value of
__truncxfhf2, so insert appropriate bitcasts in gen_soft_f80_widen_or_shorten.
Otherwise, LLVM might crash because the functions are called in a different way
than its compiler-rt definition.
This fixes stage1 build on SPARCv9, and possibly other non-x86, non-Arm CPUs.
Diffstat (limited to 'test/behavior/floatop.zig')
0 files changed, 0 insertions, 0 deletions