Environment:
    $ sw_vers 
    ProductName:    Mac OS X
    ProductVersion: 10.11.6
    BuildVersion:   15G1108
    
    GPU: GeForce GTX 680
    CUDA Driver Version: 8.0.51
    GPU Driver Version: 10.11.14 (346.03.15f04)

Steps to reproduce:

1) compose a test kernel taking a texture argument
    $ cat test.clkern 
        kernel void tst(image2d_t img) {
            (void)img;
        }

2) compile to SPIR
    $ /System/Library/Frameworks/OpenCL.framework/Libraries/openclc -c -x cl -arch gpu_64 -emit-llvm test.clkern -o test.spir

3) attempt to load and build output from command above with clCreateProgramWithBinary and clBuildProgram 
    $ ./loadbin 
    Using: GeForce GTX 680
    [CL_DEVICE_NOT_AVAILABLE] : OpenCL Error : Error: Build Program driver returned (10007)
    Break on OpenCLErrorBreak to debug.
    OpenCL Warning : clBuildProgram failed: could not build program for 0x1022700 (GeForce GTX 680) (err:-2)
    Break on OpenCLWarningBreak to debug.
    [CL_BUILD_ERROR] : OpenCL Build Error : Compiler build log:
    ptxas application ptx input, line 18; error   : Variable 'tst_param_0' cannot be allocated in .param state space
    ptxas application ptx input, line 18; error   : Texture/surface variable must be at module or entry parameter scope
    ptxas fatal   : Ptx assembly aborted due to errors

    Break on OpenCLErrorBreak to debug.
    Binary build log:
    ptxas application ptx input, line 18; error   : Variable 'tst_param_0' cannot be allocated in .param state space
    ptxas application ptx input, line 18; error   : Texture/surface variable must be at module or entry parameter scope
    ptxas fatal   : Ptx assembly aborted due to errors