cuda-gdb 3.0 - breaking in kernel

I’m having issues getting cuda-gdb to step into my kernel. It will break at the proper line number but performing a “list” gives me the code in the calling code’s main function.

I’ve attached my CUDA code and pasted my makefile. Any help would be much appreciated.

ubuntu = Ubuntu 10.04, g++, MPICH, FFTW, CUDA 3.0

SHELL = /bin/sh
#.IGNORE:

System-specific settings

NVCC = nvcc
NVCCFLAGS = -g -G
CC = g+±4.2
CCFLAGS = -g -O -I/usr/include/mpich2 -DFFT_FFTW -DLAMMPS_GZIP -I/usr/local/cuda/include
DEPFLAGS = -M
LINK = g+±4.2
LINKFLAGS = -g -O -L/usr/include/mpich2 -L/usr/local/cuda/lib64
USRLIB = -lfftw -lmpich -lcuda -lcudart
SYSLIB =
ARCHIVE = ar
ARFLAGS = -rc
SIZE = size

Link target

(EXE): (OBJ) (CU_OBJ) (LINK) (LINKFLAGS) (OBJ) (CU_OBJ) (USRLIB) (SYSLIB) -o (EXE)
(SIZE) (EXE)

Library target

lib: (OBJ) (ARCHIVE) (ARFLAGS) (EXE) $(OBJ)

Compilation rules

%.o:%.cu
(NVCC) (NVCCFLAGS) -c $<

%.o:%.cpp
(CC) (CCFLAGS) -c $<

%.d:%.cpp
(CC) (CCFLAGS) (DEPFLAGS) < > $@

Individual dependencies

DEPENDS = (OBJ:.o=.d) include (DEPENDS)

kernel.cu (9.26 KB)