Details
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
llvm/include/llvm/FuzzMutate/OpDescriptor.h | ||
---|---|---|
95–106 | This is just Type::getScalarType | |
104 | No need for make | |
125–133 | This is just Ty->isIntOrIntVectorTy(1) | |
144 | Ty->isFPOrFPVectorTy() | |
194 | no else after return | |
209 | Pretty sure those don't exist. Closest would be vectors in arrays, which I don't think support arithmetic operations | |
llvm/unittests/FuzzMutate/StrategiesTest.cpp | ||
37–38 | The odd sizes, particularly 1 and 3, are most likely to break something |
llvm/include/llvm/FuzzMutate/OpDescriptor.h | ||
---|---|---|
125–133 | With the API you mentioned, (really helpful btw), I think I'll remove anyXorVecX for now, they won't come back until I finish bitcast instructions, which is so complicated I have to put it in a stand alone patch. |
@arsenm Hey Matt, does these changes look good to you? If so we can merge it, I have some more patches depending on this one. I am hoping we can get these patches out so to better help fuzzing in the community. :)
llvm/include/llvm/FuzzMutate/OpDescriptor.h | ||
---|---|---|
190–191 | Separate declaration statements with dyn_cast? | |
192 | Usual style is to not include the explicit nullptr comparisons throughout here | |
207 | Do you mean VectorType::isValidElementType? | |
llvm/unittests/FuzzMutate/OperationsTest.cpp | ||
189 | Add some tests with scalable vectors, and vectors of pointers |
No need for make