This patch changes the index argument of lvxl?/lve[bhw]x/stvxl?/stve[bhw]x builtins from int to long. Because on 64-bit subtargets, an extra extsw will always been generated, which is incorrect.
Details
Details
- Reviewers
jsji nemanjai shchenz anhtuyen - Group Reviewers
Restricted Project - Commits
- rG240dde948252: [PowerPC] Change altivec indexed load/store builtins argument type
Diff Detail
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
Comment Actions
LGTM. Thanks for fixing this.
I am just curious, was this revealed from a real world test that used an offset that is so large that it doesn't fit into 32 bits?