Otherwise we set it always to zero, which is not correct, and we assert inside alignTo (Assertion failed: Align != 0u && "Align can't be 0."). Makes lli slightly better on Windows, at least for me =)
Details
Details
Diff Detail
Diff Detail
- Repository
- rL LLVM
Event Timeline
Comment Actions
lgtm
test/ExecutionEngine/MCJIT/coff-alignment.ll | ||
---|---|---|
5–8 ↗ | (On Diff #76504) | Want to return the i32 from main so that this isn't fragile to codegen changes? |
test/ExecutionEngine/MCJIT/coff-alignment.ll | ||
---|---|---|
5–8 ↗ | (On Diff #76504) | Sure. Thanks for your review, Reid. |
Comment Actions
We probably need some logic to handle when the aligncomm directive is present but this is still a good improvement :)