FAIL: Clang :: CodeGenOpenCL/builtins-r600.cl (2981 of 20487)

[robinchrist]

Freshly compiled with clang-5.0, running checks via make check-all
Ubuntu 16.04

******************** TEST 'Clang :: CodeGenOpenCL/builtins-r600.cl' FAILED ********************
Script:
--
/home/robin/Development/SPIRVExperimental/llvm/build/./bin/clang -cc1 -internal-isystem /home/robin/Development/SPIRVExperimental/llvm/build/bin/../lib/clang/3.6.1/include -nostdsysteminc -triple r600-unknown-unknown -S -emit-llvm -o - /home/robin/Development/SPIRVExperimental/llvm/tools/clang/test/CodeGenOpenCL/builtins-r600.cl | /home/robin/Development/SPIRVExperimental/llvm/build/./bin/FileCheck /home/robin/Development/SPIRVExperimental/llvm/tools/clang/test/CodeGenOpenCL/builtins-r600.cl
/home/robin/Development/SPIRVExperimental/llvm/build/./bin/clang -cc1 -internal-isystem /home/robin/Development/SPIRVExperimental/llvm/build/bin/../lib/clang/3.6.1/include -nostdsysteminc -triple amdgcn-unknown-unknown -S -emit-llvm -o - /home/robin/Development/SPIRVExperimental/llvm/tools/clang/test/CodeGenOpenCL/builtins-r600.cl | /home/robin/Development/SPIRVExperimental/llvm/build/./bin/FileCheck /home/robin/Development/SPIRVExperimental/llvm/tools/clang/test/CodeGenOpenCL/builtins-r600.cl
--
Exit Code: 1

Command Output (stderr):
--
/home/robin/Development/SPIRVExperimental/llvm/tools/clang/test/CodeGenOpenCL/builtins-r600.cl:8:11: error: expected string not found in input
// CHECK: call { double, i1 } @llvm.AMDGPU.div.scale.f64(double %a, double %b, i1 true)
          ^
<stdin>:6:32: note: scanning from here
define void @test_div_scale_f64(double addrspace(1)* %out, i32 addrspace(1)* %flagout, double %a, double %b) #0 {
                               ^
<stdin>:19:7: note: possible intended match here
 %2 = call { double, i1 } @llvm.AMDGPU.div.scale.f64(double %0, double %1, i1 true)
      ^
/home/robin/Development/SPIRVExperimental/llvm/tools/clang/test/CodeGenOpenCL/builtins-r600.cl:21:11: error: expected string not found in input
// CHECK: call { float, i1 } @llvm.AMDGPU.div.scale.f32(float %a, float %b, i1 true)
          ^
<stdin>:38:32: note: scanning from here
define void @test_div_scale_f32(float addrspace(1)* %out, i32 addrspace(1)* %flagout, float %a, float %b) #0 {
                               ^
<stdin>:51:7: note: possible intended match here
 %2 = call { float, i1 } @llvm.AMDGPU.div.scale.f32(float %0, float %1, i1 true)
      ^

--

********************

[bader]

It looks a bug in the test. Variable names do not translated to LLVM IR in release mode, so these checks shouldn't not rely on them.
Since this test was taken from the LLVM community code I expect it's already fixed in the ToT. I guess it was done here: llvm-mirror/clang@ca56346.