[llvm-bugs] [Bug 34379] New: CUDA compilation fails due to unknown type name '__device__' separate compilation and linking solves the problem

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=34379

            Bug ID: 34379
           Summary: CUDA compilation fails due to unknown type name
                    '__device__' separate compilation and linking solves
                    the problem
           Product: clang
           Version: 5.0
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: CUDA
          Assignee: unassignedclangbugs at nondot.org
          Reporter: lorenz.braun at ziti.uni-heidelberg.de
                CC: llvm-bugs at lists.llvm.org

Created attachment 19060
  --> https://bugs.llvm.org/attachment.cgi?id=19060&action=edit
Streamcluster application with makefile to reproduce the error + log of the
compilation with error

The streamcluster application of the rodinia benchmark [0] cannot be compiled
with clang++. Compiling the source files in separate objects and linking them
afterwards works just fine. 

Changing the order included headers might help. 

Steps to reproduce:
1) extract tar
2) change path to cuda (i used cuda 8.0)
3) run make

Result: compilation fails

Expected Result: compilation is successful
(you can set NVCC to nvcc, then it works to)

Build-Date & Hardware:
2017-30-08
Linux octane 4.2.0-27-generic #32~14.04.1-Ubuntu SMP Fri Jan 22 15:32:26 UTC
2016 x86_64 x86_64 x86_64 GNU/Linux
CUDA 8.0
LLVM 5.0
not yet tested on trunk


[0] https://www.cs.virginia.edu/~skadron/wiki/rodinia/index.php/Main_Page

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20170830/595075d4/attachment.html>