[mlir] [llvm] [OpenMPIRBuilder] Do not call host runtime for GPU teams codegen (PR #79984)

[Sergio Afonso via llvm-commits]

================
@@ -0,0 +1,25 @@
+// RUN: mlir-translate -mlir-to-llvmir %s | FileCheck %s
+
+// The aim of the test is to check the LLVM IR codegen for the device
+// for omp teams construct
+
+module attributes {dlti.dl_spec = #dlti.dl_spec<#dlti.dl_entry<"dlti.alloca_memory_space", 5 : ui32>>, llvm.data_layout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7:8:9", llvm.target_triple = "amdgcn-amd-amdhsa", omp.is_gpu = true, omp.is_target_device = true, omp.target = #omp.target<target_cpu = "gfx90a", target_features = "">} {
----------------
skatrak wrote:

Can this be tested by just having the `omp.is_target_device` attribute set?

https://github.com/llvm/llvm-project/pull/79984