[PATCH] D25708: [WebAssembly] Fix for 0xc call_indirect changes

[Dan Gohman via llvm-commits]

sunfish added a comment.

Hrm, I see. My preference would be for the approach in this patch, rather than the other. It has the advantage that we handle it in one place, and don't require as much special-casing elsewhere.

As one minor nit, the CallIndirectFixup pass should be registered outside the getOptLevel() != CodeGenOpt::None test, since it's needed for correctness.


https://reviews.llvm.org/D25708