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

[Bjorn Pettersson via Phabricator via llvm-commits]

bjope added a comment.

In https://reviews.llvm.org/D37054#956959, @bjope wrote:

> I've applied your patches to my downstream repo, trying to figure out how much of our own patches for having different address space for functions that can be removed and still getting existing test cases to work when only having your patches.
>  Lots of new things happens now when there are addrspace on all function definitions. Our old implementation kind of added the addrspace information when fetching function pointers from the symbol table instead of adding addrspace when creating the functions. So I still have a couple of issues to analyse. I think that I at least tracked down one problem with calls to a function given a local function pointer variable (see my inline comment in LLParser.cpp).
>
> 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.
>  Are you testing this with some other out-of-tree target?


ping!


https://reviews.llvm.org/D37054