[clang] [Clang][SYCL] Introduce clang-sycl-link-wrapper to link SYCL offloading device code (Part 1 of many) (PR #112245)

[Joseph Huber via cfe-commits]

jhuber6 wrote:

> I'm also concerned that there are no tests for the new tool. There are tests for the clang driver changes, but none for the new tool. The new tool seems to be just a wrapper around llvm-link, which does make me think this should probably just be folded into llvm-link. Understanding the direction here would be helpful.

There is one, but it should probably be expanded [clang-sycl-linker-test.cpp](https://github.com/llvm/llvm-project/pull/112245/files#diff-373ef22afa527adac2fdce560a26553ed7280e4bc0b75f01ff77e1d64140f491). The purpose is to behave like a Toolchain linker so that when we do `--target=spirv64-whatever` it will create a valid block for that target as is customary with `clang` cross-compiling.

I don't know exactly what SYCL needs that's different from `llvm-link`, maybe the others can chime in because I thought that it required the backend being run on it. I think the reason this doesn't use the LTO interface right now is because the SPIR-V target isn't done? I'm sure some of the SYCL team can explain what they're doing but I assume the goal is

foo.bc bar.bc -> image.spv

https://github.com/llvm/llvm-project/pull/112245