[llvm] [offload][SYCL] Add SYCL Module splitting. (PR #131347)

[Alexey Bader via llvm-commits]

bader wrote:

There should be nothing SYCL specific about the patch. This PR adds a function that distributes content of one LLVM module into one or more LLVM modules. It's generic LLVM transformation. The logic how content is distributed is currently limited to two cases:

1. Put each "entry point" (you can think of it as "GPU kernel") into separate LLVM module.
2. Put all functions with specific function attribute to a separate LLVM module.

_There are plans to extend this list by future patches_

(1) is GPU specific, but it can be useful for non-SYCL GPU programming models as well. IIRC, @jdoerfert has been involved in prototyping the GPU code splitting logic in OpenMP offload compiler.

(2) the name of the attribute is 'sycl-module-id', but the meaning is the same as C++ compilation unit. If two functions have the same "module-id" values, they are produced from the same compilation unit.
**NOTE**: In SYCL case we get functions produced from different C++ compilation units (CU) by linking LLVM modules for each CU. _This scenario might be rare for programming models using thinLTO framework for linking the device code as thinLTO naturally keeps the code split by compilation unit._

The primary use of this functionality is to reduce the code generation time for GPU code. This is critical for JIT compiling. Another nice property is splitting LLVM module allows the compiler to skip the code generation at all. This is useful if LLVM mode has code for different targets and we need to avoid code generation for "unwanted" targets.

The code mentions SYCL because it's currently used only by SYCL compiler. @sarnex, have you thought about using this function for the OpenMP offload compilation?

We can commit this code as a SYCL specific library right now and make it more generic if we find use cases outside of SYCL compilation flow.

https://github.com/llvm/llvm-project/pull/131347