Project

General

Profile

Actions

Bug #1156

closed

Nvidia CUDA/Suricata 2.0 Compile Problem

Added by Shant Kassardjian about 10 years ago. Updated about 8 years ago.

Status:
Closed
Priority:
Normal
Assignee:
-
Target version:
-
Affected Versions:
Effort:
Difficulty:
Label:

Description

Info:
Dell R720 - 32 hyper-threaded cores
GPU: Nvidia Tesla K20
Ram: 128GB
NIC: Intel x520 (attempting pf_ring + cuda)
OS: Debian Wheezy, linux kernel 3.2.0

Nvidia packages installed:
nvidia-cuda-toolkit 5.0.35-8~bpo70+1 (installed from wheezy-backports as suggested by Victor Julien)
nvidia-kernel-amd64 (module to recognize Nvidia GPU)

Attempted to compile suricata-2.0 by running the following command:

./configure --enable-gccprotect --enable-profiling --enable-cuda --with-cuda-includes=/usr/local/cuda/include/ --with-cuda-libraries=/usr/local/cuda/lib64/ --enable-pfring --with-libpfring-libraries=/usr/local/pfring/lib --with-libpfring-includes=/usr/local/pfring/include --with-libpcap-libraries=/usr/local/pfring/lib/ --with-libpcap-includes=/usr/local/pfring/include LD_RUN_PATH="/usr/local/pfring/lib:/usr/lib:/usr/local/lib" --prefix=/usr/local/pfring/

Compile error occurs during the 'make' phase which results in with the following error message (Suricata 1.4.7 compiles with no error):

/usr/bin/nvcc -O2 -o util-mpm-ac-cuda-kernel.ptx_sm_35 -arch=sm_35 -ptx util-mpm-ac-cuda-kernel.cu
/usr/bin/python ptxdump.py cuda-ptxdump util-mpm-ac-cuda-kernel.ptx_sm_10 util-mpm-ac-cuda-kernel.ptx_sm_11 util-mpm-ac-cuda-kernel.ptx_sm_12 util-mpm-ac-cuda-kernel.ptx_sm_13 util-mpm-ac-cuda-kernel.ptx_sm_20 util-mpm-ac-cuda-kernel.ptx_sm_21 util-mpm-ac-cuda-kernel.ptx_sm_30 util-mpm-ac-cuda-kernel.ptx_sm_35
ptxdump.py: CUmodule util_mpm_ac_cuda_kernel_sm_10 packed successfully
ptxdump.py: CUmodule util_mpm_ac_cuda_kernel_sm_11 packed successfully
ptxdump.py: CUmodule util_mpm_ac_cuda_kernel_sm_12 packed successfully
ptxdump.py: CUmodule util_mpm_ac_cuda_kernel_sm_13 packed successfully
ptxdump.py: CUmodule util_mpm_ac_cuda_kernel_sm_20 packed successfully
ptxdump.py: CUmodule util_mpm_ac_cuda_kernel_sm_21 packed successfully
ptxdump.py: CUmodule util_mpm_ac_cuda_kernel_sm_30 packed successfully
ptxdump.py: CUmodule util_mpm_ac_cuda_kernel_sm_35 packed successfully
ptxdump.py: cuda-ptxdump.h written successfully
make all-am
make3: Entering directory `/suricata-2.0/src'
gcc -DHAVE_CONFIG_H -I. -I.. -I./../libhtp/ -I/usr/local/pfring/include -I/usr/local/pfring/include -I/usr/local/cuda/include -DLOCAL_STATE_DIR=\"/usr/local/pfring/var\" -g -O2 -Wextra -Werror-implicit-function-declaration -fno-tree-pre -Wall -Wno-unused-parameter -std=gnu99 -fstack-protector -D_FORTIFY_SOURCE=2 -Wformat -Wformat-security -march=native -DPROFILING -DHAVE_LIBNET11 -D_BSD_SOURCE -D__BSD_SOURCE -D__FAVOR_BSD -DHAVE_NET_ETHERNET_H -DHAVE_PFRING -I/usr/include -DLIBPCAP_VERSION_MAJOR=1 -DHAVE_PCAP_SET_BUFF -D__SC_CUDA_SUPPORT__ -DHAVE_LIBCAP_NG -MT alert-debuglog.o -MD -MP -MF .deps/alert-debuglog.Tpo -c -o alert-debuglog.o alert-debuglog.c
In file included from util-cuda-handlers.h:28:0,
from decode.h:617,
from detect-engine-alert.h:28,
from suricata-common.h:321,
from alert-debuglog.c:24:
util-cuda.h:150:29: error: unknown type name 'CUipcEventHandle'
util-cuda.h:151:27: error: unknown type name 'CUipcMemHandle'
util-cuda.h:152:49: error: unknown type name 'CUipcEventHandle'
util-cuda.h:153:48: error: unknown type name 'CUipcMemHandle'
make3: * [alert-debuglog.o] Error 1
make3: Leaving directory `/suricata-2.0/src'
make2:
[all] Error 2
make2: Leaving directory `/suricata-2.0/src'
make1:
[all-recursive] Error 1
make1: Leaving directory `/suricata-2.0'
make: *
[all] Error 2

Please let me know if further information is required.

Actions #1

Updated by Andreas Herz about 8 years ago

  • Status changed from New to Closed

CUDA support is broken, see #1699 so we need to work on that in general.

Actions

Also available in: Atom PDF