Treat these basically the same way as aliases.
Details
Diff Detail
Event Timeline
llvm/test/tools/llvm-reduce/remove-ifunc.ll | ||
---|---|---|
63 | mixing ptr and * will just end up using opaque pointers everywhere, so this part is pointless |
llvm/test/tools/llvm-reduce/remove-ifunc.ll | ||
---|---|---|
63 | I'm constantly wondering if I should be trying to write code that bothers trying to work with typed pointers | |
llvm/tools/llvm-reduce/deltas/ReduceAliases.cpp | ||
38 | I'm wondering if it would be better to lower this by emitting a call to the resolver, plus the indirect call of the result |
precommit bot is saying the added tests are failing
llvm/lib/Transforms/Utils/ModuleUtils.cpp | ||
---|---|---|
292 ↗ | (On Diff #478354) | this deserves a unittest, then I'd say we don't need as many llvm-reduce-specific tests |
llvm/test/tools/llvm-reduce/remove-ifunc-constantexpr.ll | ||
2 | not sure how useful an XFAIL test is if the verifier should fail on this anyway and I presume nobody's producing this sort of IR should add a FIXME verifier test separately |
not sure how useful an XFAIL test is if the verifier should fail on this anyway and I presume nobody's producing this sort of IR
should add a FIXME verifier test separately