bwendling wrote: > > If I return the LValueToRValue cast and run EmitLValue on it > > I'm surprised EmitLValue didn't error out on that. The result of an > LValueToRValue cast is an rvalue. Calling EmitPointerWithAlignment on it > should work, I think. > > > I understand, but there are ~30 different types of implicit casts and it's > > hard to know which can be ignored and which shouldn't. > > The complexity here is, unfortunately, mostly unavoidable... maybe worth > looking into some simplified classification of casts for common usage. > > > Convert a __builtin_dynamic_object_size with a FAM with __counted_by into > > the appropriate calculations in Sema. > > This might be reasonable. > > > because we lose the fact that the original expression was a > > __builtin_dynamic_object_size call > > If we went this route, probably we'd wrap the calculation in a dedicated node > that contains both the original AST, and the lowered computation, or > something like that.
I went ahead and used the `EmitPointerWithAlignment` for all (it worked without needing special casing). I'll look into doing this in Sema as a potential cleanup. https://github.com/llvm/llvm-project/pull/110487 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
