diff options
| author | wanglei <wanglei@loongson.cn> | 2022-11-03 11:10:41 +0800 |
|---|---|---|
| committer | Weining Lu <luweining@loongson.cn> | 2022-11-03 11:40:50 +0800 |
| commit | 5adb090795914a4122492117f3af8dacaf69301f (patch) | |
| tree | 1ad2eafeaae1340980f4728f2532fd2cbe7fc3b6 /llvm/lib/Support/CommandLine.cpp | |
| parent | 1024497fffd18bba92485852cfedebb829499953 (diff) | |
| download | llvm-5adb090795914a4122492117f3af8dacaf69301f.zip llvm-5adb090795914a4122492117f3af8dacaf69301f.tar.gz llvm-5adb090795914a4122492117f3af8dacaf69301f.tar.bz2 | |
[LoongArch] Fix codegen for [su]itofp instructions
This patch fixes codegen for `[su]itofp` instructions.
In LoongArch, a legal int-to-float conversion is done in two steps:
1. Move the data from `GPR` to `FPR`. (FRLen >= GRLen)
2. Conversion in `FPR`. (the data in `FPR` is treated as a signed value)
Based on the above features, when the type's BitWidth meets the
requirements, all `SINT_TO_FP` are legal, all `UINT_TO_FP` are expand
and lowered to libcall when appropriate.
The only special case is, LoongArch64 with `+f,-d` features. At this
point, custom processing is required for `[SU]INT_TO_FP`. Of course, we
can also ignore it and use libcall directly.
Differential Revision: https://reviews.llvm.org/D136916
Diffstat (limited to 'llvm/lib/Support/CommandLine.cpp')
0 files changed, 0 insertions, 0 deletions