<input>(0): Error: Signal Segmentation fault in phase PU_adjust_addr_flags -- processing abort

Accelerated Computing CUDA CUDA Programming and Performance
1

I am far from original cppIntegration source and seems I hit some bug:

tonu@quad:~/NVIDIA_CUDA_SDK/projects/cppIntegration$ make clean
tonu@quad:~/NVIDIA_CUDA_SDK/projects/cppIntegration$ make
Signal: Segmentation fault in PU_adjust_addr_flags phase.
(0): Error: Signal Segmentation fault in phase PU_adjust_addr_flags – processing aborted
*** Internal stack backtrace:
/usr/local/cuda/open64/lib//be [0x830487f]
/usr/local/cuda/open64/lib//be [0x8305403]
/usr/local/cuda/open64/lib//be [0x83057d8]
/usr/local/cuda/open64/lib//be [0x8304b19]
/usr/local/cuda/open64/lib//be [0x8305af0]
[0x4001a400]
/usr/local/cuda/open64/lib//be [0x81f84fd]
/usr/local/cuda/open64/lib//be [0x81f81e0]
/usr/local/cuda/open64/lib//be [0x81f92e2]
/usr/local/cuda/open64/lib//be [0x8059251]
/usr/local/cuda/open64/lib//be [0x805a23d]
/usr/local/cuda/open64/lib//be [0x805b3dc]
/usr/local/cuda/open64/lib//be [0x805c9eb]
/lib/tls/i686/cmov/libc.so.6(__libc_start_main+0xe5) [0x40167685]
/usr/local/cuda/open64/lib//be [0x8058a85]
nvopencc INTERNAL ERROR: /usr/local/cuda/open64/lib//be died due to signal 4
make: *** [obj/release/cppIntegration.cu.o] Error 255
tonu@quad:~/NVIDIA_CUDA_SDK/projects/cppIntegration$ /usr/local/cuda/open64/lib//be -V
(0): Warning: Unrecognized option (V) in “-V”
(0): Warning: No source files specified in command line
tonu@quad:~/NVIDIA_CUDA_SDK/projects/cppIntegration$ /usr/local/cuda/open64/lib//be -v
/usr/local/cuda/open64/lib//be -v
(0): Warning: No source files specified in command line
tonu@quad:~/NVIDIA_CUDA_SDK/projects/cppIntegration$ /usr/local/cuda/open64/lib//be --help
(0): Warning: Unrecognized option (-) in “–help”
(0): Warning: No source files specified in command line
tonu@quad:~/NVIDIA_CUDA_SDK/projects/cppIntegration$

As you see “be” knows nothing about his version. So don’t ask me. Source is available on request.

2

can you post source along with nvcc --version?

3

tonu@quad:~/NVIDIA_CUDA_SDK/projects/cppIntegration$ /usr/local/cuda/bin/nvcc --version

nvcc: NVIDIA ® Cuda compiler driver

Copyright © 2005-2007 NVIDIA Corporation

Built on Wed_Dec__3_18:29:25_PST_2008

Cuda compilation tools, release 2.1, V0.2.1221

tonu@quad:~/NVIDIA_CUDA_SDK/projects/cppIntegration$

Source is on http://no.spam.ee/~tonu/crash.tar.gz

4

In first step would be good to tell how to avoid this. I am trying to clean my code from all possibly buggy constructions but cannot yet make it compile.

5

ping?

6

NVIDIA doesn’t seem to be very helpful in this case. I really love Open Source in such cases as this bug would be fixed in 24 hours instead of waiting even bug confirmation forever.

Ayway if someone else hits same bug I got over it commenting out this line:

// memcpy (buffer, problem_ptr, problem_len);

Unsure if this helps in future too.

7

Sorry about the delayed response. I’ve filed bug 529154 to have this problem investigated further.