hipStreamAddCallback compile error about hip

[sunway513]

Hi, could you try to use a newer HIP release? Such as the following one:
https://github.com/GPUOpen-ProfessionalCompute-Tools/HIP/releases/tag/release_1.0.16435
I just tried the square sample, it works fine on my config on NVidia machine.
If you still have a problem after a newer HIP, please provide your hipconfig information here.

[dinosaur8312]

Hi,

I have upgraded the version, but still got the same error.

I don't think including the typename when calling a function as in the /opt/rocm/hip/include/hip/nvcc_detail/hip_runtime_api.h is the right way.

inline static hipError_t hipStreamAddCallback(hipStream_t stream, hipStreamCallback_t callback, void *userData, unsigned int flags)
{
return hipCUDAErrorTohipError(cudaStreamAddCallback(cudaStream_t stream,
cudaStreamCallback_t callback, void *userData, unsigned int flags));
}

I posted the hip version info and hipconfig info here:

opt/rocm/hip/bin/hipcc --version
HIP version: 1.0.16453
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2015 NVIDIA Corporation
Built on Tue_Aug_11_14:27:32_CDT_2015
Cuda compilation tools, release 7.5, V7.5.17

/opt/rocm/hip/bin/hipconfig --full
HIP version : 1.0.16453

== hipconfig
HIP_PATH : /opt/rocm/hip
HIP_PLATFORM : nvcc
CPP_CONFIG : -D__HIP_PLATFORM_NVCC__= -I/opt/rocm/hip/include -I/cgv/Compilers/nvidia/CUDA-7.5.18/include

== nvcc
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2013 NVIDIA Corporation
Built on Wed_Jul_17_18:36:13_PDT_2013
Cuda compilation tools, release 5.5, V5.5.0

=== Environment Variables
PATH=/usr/local/cuda/bin:.:/usr/local/bin:/bin:/usr/bin:/usr/sbin:/etc:/usr/bin/X11:/cgv/Compilers/intel/compiler-14.0.1/bin:/tad/local/lam-7.1.4/bin
LD_LIBRARY_PATH=/cgv/Compilers/nvidia/CUDA/lib64:/cgv/Compilers/nvidia/CUDA-3.2/lib64:/cgv/Compilers/nvidia/CUDA-5.0.35_db6/lib64:/cgv/Compilers/nvidia/CUDA-5.5.22_db6/lib64:/cgv/Compilers/mpi/openmpi/lib64:/cgv/Compilers/intel/compiler-14.0.1/composer_xe_2013_sp1.1.106/compiler/lib/intel64:/cgv/Compilers/intel/compiler-14.0.1/composer_xe_2013_sp1.1.106/mpirt/lib/intel64:/cgv/Compilers/intel/compiler-14.0.1/composer_xe_2013_sp1.1.106/ipp/../compiler/lib/intel64:/cgv/Compilers/intel/compiler-14.0.1/composer_xe_2013_sp1.1.106/ipp/lib/intel64:/cgv/Compilers/intel/compiler-14.0.1/composer_xe_2013_sp1.1.106/mkl/lib/intel64:/cgv/Compilers/intel/compiler-14.0.1/composer_xe_2013_sp1.1.106/tbb/lib/intel64/gcc4.4:/TANGO/3/rel/lib:/vega/1/rel/Oracle/lib:
CUDA_CACHE_DISABLE=1
CUDA=/cgv/Compilers/nvidia/CUDA-7.5.18
CUDA_PATH=/cgv/Compilers/nvidia/CUDA-7.5.18

== Linux Kernel
Hostname : hougpu002
Linux hougpu002 3.14.19-grml64w-01 #1 SMP Fri Sep 19 15:34:40 CDT 2014 x86_64 GNU/Linux
No LSB modules are available.
Distributor ID: Debian
Description: Debian GNU/Linux 7.11 (wheezy)
Release: 7.11
Codename: wheezy

[sunway513]

Hi,
Thank you for the additional information and pointing out this issue.
Did you modify the square sample? If so, could you share it here?
There are errors in this API translation, please let me know if the attached patch could solve it for you.
callback.patch.zip

[dinosaur8312]

Hi,

The issue is gone if I manually modify the header file by add brackets to each argument as follow:

inline static hipError_t hipStreamAddCallback(hipStream_t stream, hipStreamCallback_t callback, void *userData, unsigned int flags)
{
return hipCUDAErrorTohipError(cudaStreamAddCallback((cudaStream_t) stream,
(cudaStreamCallback_t) callback, (void *)userData, (unsigned int) flags));
}

I haven't try the patch yet. But I think it should work if it give similar fix.

Thank you!