[PATCH] D37054: Require address space to be specified when creating functions (2/3)

[Dylan McKay via Phabricator via llvm-commits]

dylanmckay added a comment.

> There are no testcases included here with ll-files including calls/bitcasts etc with non-zero addrspace, so it is a little bit tricky to understand how far I'm supposed to get by only using your set of patches.



> I've not analysed this any more since I discovered the problems in LLParser in December. I'm still kind of waiting for some response to understand more about the status of this patch (is it tested? is it supposed to work? do you have some test cases?).

I will work on adding some more test cases @bjope, sorry for being so slow with this!

@arichardson

Thanks a lot for looking into and submitting https://reviews.llvm.org/D43645! I will happily review it

@bjope

> One problem I've seen is related to globals with forward reference of a function pointer (although I've not tested it with a clean top-of-tree yet)
>  ..
>  Maybe there should be an addrspace attribute on the declare statement? At the moment it does not help, because LLParser does not parse it.

Your writeup and suggested change SGTM, I will fix the patch up and resubmit


https://reviews.llvm.org/D37054