cuTexObjectCreate return error code

Question

I am working with Cuda driver api, now i have a problem about Cuda texture object creation, it always return CUDA_ERROR_INVALUD_VALUE.

I found function cuTexObjectCreate is wrapped by macro :

#if __CUDA_API_VERSION >= 5000
...function decl...
#endif

And my Cuda context created by cuCtxCreate_v2 always return api version 3020. Does that means I should create a high version context? and which function should I call? Or it's something else?

That's the sample code:

struct CudaDriverTest
{
    CUdevice m_device = 0;
    CUcontext m_primaryContext = nullptr;
    CUcontext m_context = nullptr;
    CUarray m_array = nullptr;
    CUtexObject m_texture = 0;
    CUdeviceptr m_output = 0;

    CudaDriverTest(size_t w, size_t h, float* data) :
        m_image(w, h, QImage::Format_Grayscale8)
    {
        // begin cuda driver api staff
        HANDLE_ERROR(cuInit(0));
        int deviceCount = 0;
        HANDLE_ERROR(cuDeviceGetCount(&deviceCount));
        assert(deviceCount == 1);
        HANDLE_ERROR(cuDeviceGet(&m_device, 0));
        char name[256];
        HANDLE_ERROR(cuDeviceGetName(name, 256, m_device));
        std::cout << "device name:" << name << std::endl;
        int major = 0;
        int minor = 0;
        HANDLE_ERROR(cuDeviceComputeCapability(&major, &minor, m_device));
        std::cout << "major compute capability : " << major << ", minor compute capability : " << minor << std::endl;
        HANDLE_ERROR(cuCtxCreate(&m_context, CU_CTX_SCHED_AUTO, m_device));
        std::uint32_t version;
        HANDLE_ERROR(cuCtxGetApiVersion(m_context, &version));
        std::cout << "context api version : " << version << std::endl;
        // array
        CUDA_ARRAY_DESCRIPTOR arrDesc;
        memset(&arrDesc, 0, sizeof(arrDesc));
        arrDesc.Format = CUarray_format::CU_AD_FORMAT_FLOAT;
        arrDesc.NumChannels = 1;
        arrDesc.Width = w;
        arrDesc.Height = h;
        HANDLE_ERROR(cuArrayCreate(&m_array, &arrDesc));
        CUDA_MEMCPY2D cpy2d;
        memset(&cpy2d, 0, sizeof(cpy2d));
        cpy2d.srcMemoryType = CUmemorytype::CU_MEMORYTYPE_HOST;
        cpy2d.srcHost = data;
        cpy2d.srcPitch = w * sizeof(float);
        cpy2d.dstMemoryType = CUmemorytype::CU_MEMORYTYPE_ARRAY;
        cpy2d.dstArray = m_array;
        cpy2d.dstPitch = w * sizeof(float);
        cpy2d.WidthInBytes = w * sizeof(float);
        cpy2d.Height = h;
        HANDLE_ERROR(cuMemcpy2D(&cpy2d));
        // texture object
        CUDA_RESOURCE_DESC resDesc;
        memset(&resDesc, 0, sizeof(resDesc));
        resDesc.resType = CUresourcetype::CU_RESOURCE_TYPE_ARRAY;
        resDesc.res.array.hArray = m_array;
        CUDA_TEXTURE_DESC texDesc;
        memset(&texDesc, 0, sizeof(texDesc));
        texDesc.addressMode[0] = CUaddress_mode::CU_TR_ADDRESS_MODE_WRAP;
        texDesc.addressMode[1] = CUaddress_mode::CU_TR_ADDRESS_MODE_WRAP;
        texDesc.addressMode[2] = CUaddress_mode::CU_TR_ADDRESS_MODE_WRAP;
        texDesc.filterMode = CUfilter_mode::CU_TR_FILTER_MODE_LINEAR;
        texDesc.flags = CU_TRSF_READ_AS_INTEGER;
        CUDA_RESOURCE_VIEW_DESC viewDesc;
        memset(&viewDesc, 0, sizeof(viewDesc));
        viewDesc.width = w;
        viewDesc.height = h;
        viewDesc.format = CUresourceViewFormat::CU_RES_VIEW_FORMAT_FLOAT_1X32;
        HANDLE_ERROR(cuTexObjectCreate(&m_texture, &resDesc, &texDesc, &viewDesc));
        // output
        HANDLE_ERROR(cuMemAlloc(&m_output, w * h * sizeof(float)));
    }
}

PS: working environment: GTX960;Cuda7.5;windows7-x64;visual studio 2013;

Answer 1

It's my fault, I did not correctly ZeroMemory the the parameters. PS. Thanks @talonmies for editing the question.