First commit

hancm · May 13, 2015 · 0b97f28 · 0b97f28
commit 0b97f28
Showing 49 changed files with 12,158 additions and 0 deletions.
diff --git a/FERMI_DEVICES b/FERMI_DEVICES
@@ -0,0 +1,50 @@
+GeForce GT 415M	0x0dee
+GeForce GT 420	0x0de2
+GeForce GT 420M	0x0df1
+GeForce GT 420M	0x0df3
+GeForce GT 425M	0x0df0
+GeForce GT 430	0x0de1
+GeForce GT 435M	0x0df2
+GeForce GT 440	0x0dc0
+GeForce GT 440	0x0de0
+GeForce GT 445M	0x0dd2
+GeForce GT 520	0x1040
+GeForce GT 520M	0x0df7
+GeForce GT 520M	0x1050
+GeForce GT 525M	0x0dec
+GeForce GT 525M	0x0df5
+GeForce GT 530	0x0de5
+GeForce GT 550M	0x0dd6
+GeForce GT 550M	0x0df6
+GeForce GT 555M	0x0dcd
+GeForce GTS 450	0x0dc4
+GeForce GTS 450	0x1245
+GeForce GTX 460	0x0e22
+GeForce GTX 460	0x0e24
+GeForce GTX 460 SE	0x0e23
+GeForce GTX 460M	0x0dd1
+GeForce GTX 465		0x06c4
+GeForce GTX 470		0x06cd
+GeForce GTX 470M	0x0e30
+GeForce GTX 480		0x06c0
+GeForce GTX 480M	0x06ca
+GeForce GTX 485M	0x0e31
+GeForce GTX 550 Ti	0x1244
+GeForce GTX 560 Ti	0x1082
+GeForce GTX 560 Ti	0x1200
+GeForce GTX 570	0x1081
+GeForce GTX 580	0x1080
+GeForce GTX 590	0x1088
+Tesla C2050 / C2070	0x06d1
+Tesla C2070	0x0772
+Tesla C2070	0x0772
+Tesla C2075	0x1096
+Tesla M2050	0x06de
+Tesla M2050	0x077a
+Tesla M2050	0x082f
+Tesla M2050	0x0843
+Tesla M2050	0x0846
+Tesla M2070	0x06d2
+Tesla M2070-Q	0x06df
+Tesla M2075	0x1094
+Tesla M2090	0x1091
diff --git a/LICENSE b/LICENSE
@@ -0,0 +1,45 @@
+SSLShader has been developed at the Computer Science and
+Electrical Engineering Departments at KAIST by 
+Keon Jang <keonjang@an.kaist.ac.kr>, Sangjin Han <sangjin@an.kaist.ac.kr>,
+Seungyeop Han <syhan@cs.washington.edu>, Sue Moon<sbmoon@kaist.edu>,
+and KyoungSoo Park<kyoungsoo@ee.kaist.ac.kr>.
+
+
+Copyright (c) 2011, KAIST All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+1. Redistributions of source code must retain the above copyright
+   notice, this list of conditions and the following disclaimer.
+2. Redistributions in binary form must reproduce the above copyright
+   notice, this list of conditions and the following disclaimer in the
+   documentation and/or other materials provided with the distribution.
+3. All advertising materials mentioning features or use of this software
+   must display the following acknowledgement:
+   This product includes software developed by the KAIST.
+4. Neither the name of the KAIST nor the
+   names of its contributors may be used to endorse or promote products
+   derived from this software without specific prior written permission.
+
+ THIS SOFTWARE IS PROVIDED BY KAIST ''AS IS'' AND ANY
+ EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ DISCLAIMED. IN NO EVENT SHALL KAIST BE LIABLE FOR ANY
+ DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+ ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+Unless otherwise noted, all codes are licensed under the above License.
+ you may not use this file except in compliance with the License.
+ you may obtain copy of the License at
+
+     http://shader.kaist.edu/sslshader/LICENSE
+
+ Unless required by applicable law or agreed to in writing,
+ software distributed under the License is distributed on as "AS IS" BASIS,
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ See the License for the specific language govnerning permissions and
+ limitations under the License.
diff --git a/Makefile b/Makefile
@@ -0,0 +1,56 @@
+GPUCRYPTO_DIR = ./
+
+include Makefile.in
+
+OBJS_DIR = objs
+TARGET_DIR = lib
+TARGET_FILE = libgpucrypto.a
+TARGET = $(addprefix $(TARGET_DIR)/, $(TARGET_FILE))
+
+.SUFFIXES : .cu .cc .o
+
+CU_SRC_FILES = $(wildcard *.cu)
+CC_SRC_FILES = $(wildcard *.cc)
+HEADER_FILES = $(wildcard *.hh) $(wildcard *.h)
+
+SRC_FILES = $(CU_SRC_FILES) $(CC_SRC_FILES)
+OBJS_FILE = $(CU_SRC_FILES:.cu=.o) $(CC_SRC_FILES:.cc=.o)
+
+OBJS = $(addprefix $(OBJS_DIR)/, $(OBJS_FILE))
+DEPS = Makefile.dep
+
+all: $(TARGET) test
+
+$(TARGET): $(DEPS) $(OBJS) | $(TARGET_DIR) $(OBJS_DIR)
+	ar rcs $@ $(OBJS)
+
+$(TARGET_DIR):
+	mkdir $(TARGET_DIR)
+
+$(OBJS_DIR):
+	mkdir $(OBJS_DIR)
+
+$(DEPS): $(SRC_FILES) $(HEADER_FILES)
+	$(CC) -MM -MP -x c++ $(CU_SRC_FILES) $(CC_SRC_FILES) | sed 's![^:]*.o:!objs/&!g' > Makefile.dep
+
+$(OBJS_DIR)/%.o : %.cc
+	$(CC) $(CFLAGS) $(INC) $(NVINC) -c $< -o $@
+
+$(OBJS_DIR)/%.o : %.cu
+	$(NVCC) $(NVCFLAGS) $(NVINC) -c $< -o $@
+
+.PHONY : test clean doc
+
+test: $(TARGET)
+	make -C ./test
+
+clean:
+	rm -f $(TARGET) $(OBJS) $(DEPS)
+	make clean -C ./test
+
+doc: $(SRC_FILES) doxygen.config
+	doxygen doxygen.config
+
+ifneq ($(MAKECMDGOALS), clean)
+-include $(DEPS)
+endif
diff --git a/Makefile.dep b/Makefile.dep
@@ -0,0 +1,13 @@
+objs/rsa_context.o: rsa_context.cc rsa_context.hh
+
+rsa_context.hh:
+objs/rsa_context_rns.o: rsa_context_rns.cc rsa_context_rns.hh rsa_context.hh \
+  rsa_cuda.h bn_kn.h
+
+rsa_context_rns.hh:
+
+rsa_context.hh:
+
+rsa_cuda.h:
+
+bn_kn.h:
diff --git a/Makefile.in b/Makefile.in
@@ -0,0 +1,42 @@
+#modify directory configuration below to match your configuration \
+ if you want use system default openssl-dev installation then comment out the OPENSSL_DIR line
+OPENSSL_DIR = /usr/local/ssl
+CUDA_TOOLKIT_DIR = /usr/local/cuda
+CUDA_SDK_DIR = /home/$(USER)/NVIDIA_GPU_Computing_SDK
+
+#openssl config
+ifneq ($(OPENSSL_DIR), )
+	OPENSSL_LIBDIR = -L$(OPENSSL_DIR)/lib64
+	OPENSSL_INCDIR = -I$(OPENSSL_DIR)/include
+endif
+OPENSSL_LIB = -lcrypto
+
+#generic config
+CC = g++
+CFLAGS = -O2 -g -Wall
+INC = $(OPENSSL_INCDIR)
+
+#CUDA related config
+NVCC = nvcc
+NVCFLAGS= --ptxas-options=-v -O2 --use_fast_math
+NVINC = -I/usr/local/cuda/include -I$(CUDA_SDK_DIR)/C/common/inc
+
+CUDA_LIBDIR = -L$(CUDA_TOOLKIT_DIR)/lib64  -L$(CUDA_SDK_DIR)/C/lib
+CUDA_LIBS = -lcudart
+
+#below lines automatically detects the fermi devices using lspci command
+#if any CUDA device that supports architecture 2.x is detected,
+#then it enables use of 64-bit integer for RSA and generate compute_20 code.
+#you can manually set IS_FERMI value below if it does not automatically detect your fermi device.
+#read FERMI_DEVICES file for supported cards list.
+CUDA_DEVICE_IDS = $(shell cat $(GPUCRYPTO_DIR)FERMI_DEVICES | sed "s/^.*0x//g" | awk '{print "10de:"$$1}')
+IS_FERMI = $(foreach DEVID, $(CUDA_DEVICE_IDS), $(shell lspci -nn | grep -q $(DEVID) && echo "1"))
+
+ifneq (,$(findstring 1,$(IS_FERMI)))
+	NVCFLAGS += -DMP_USE_64BIT=1 -gencode arch=compute_20,code=sm_20
+	CFLAGS += -DMP_USE_64BIT=1
+else
+	NVCFLAGS += -DMP_USE_64BIT=0 -gencode arch=compute_12,code=sm_12
+	CFLAGS += -DMP_USE_64BIT=0
+endif
+
diff --git a/README b/README
@@ -0,0 +1,117 @@
+GENERAL
+=======
+libgpucrypto is subset of SSLShader software that implements
+few cryptographic algorithms: AES, SHA1, RSA using CUDA.
+This code is distributed under BSD-style license.
+Read LICENSE for more details.
+
+It requires CUDA 1.2 or above capable devices.
+We have tested our code on GTX285, GTX480, and GTX580.
+
+Below is the machine specification we used for the test.
+
+CPU: Intel X5650 2.66Ghz x 2
+M/B: Super Micro X8DAH
+RAM: DDR3 4GB x 6
+
+libgpucrypto is developed under Linux environment,
+and it is dependent on CUDA and OpenSSL.
+We have tesed our code on CUDA 3.2 and 4.0
+along with OpenSSL 1.0.0 (Mar 29, 2010).
+Below, we list versions of software that we have tested.
+
+* CUDA 4.0
+CUDA driver  : 270.41.19
+CUDA toolkit : 4.0.17
+CUDA SDK     : 4.0.17
+
+* CUDA 3.2
+CUDA driver  : 260.19.26
+CUDA toolkit : 3.2.16
+CUDA SDK     : 3.2.16
+
+* O/S
+Ubuntu 10.04 LTS 64bit
+
+We never tested our code on other Linux distributions nor 32 bit systems.
+Please do not ask for help on distribution specific issues.
+
+
+SETUP
+=====
+
+1) install required libraries
+ you can download CUDA stuff at
+ http://developer.nvidia.com/cuda-toolkit-40
+
+2) install OpenSSL libraries and headers
+ you can download OpenSSL at
+ http://openssl.org/source/
+
+3) configure following variables in Makefile.in
+ OPENSSL_DIR
+ CUDA_TOOLKIT_DIR
+ CUDA_SDK_DIR
+
+if you're using system default opeenssl development library,
+then you can leave it as blank.
+
+4) build libgpucrypto
+ make
+
+5) try running test code
+
+examples)
+#./bin/aes_test -m ENC
+#./bin/rsa_test -m MP
+#./bin/sha_test
+
+you can see more detailed usage by running program w/o arguments or w/ incorrect one :).
+
+
+HOW TO USE
+==========
+Please see test folder for example codes that uses libgpucrypto.
+Using this code requires some prior knowledge on CUDA programming and
+GPU codes are written in a way to optimize performance rather than usability,
+and the data structure it's API receives is quite complex to document.
+Please see below functions and codes for how to use GPU library.
+
+RSA: rsa_test.cc: test_latency_rsa
+AES: aes_test.cc: test_latency_ase_cbc_encrypt, test_latency_aes_cbc_decrypt
+SHA1: sha_test.cc: test_latency_hmac_sha1
+
+AES and SHA1 test codes have a function to transform
+somewhat human friendly data structure to GPU code's structure.
+See aes_cbc_encrypt_prepare, aes_cbc_decrypt_prepare, and hmac_sha1_prepare
+to better understand the exact data structure used for GPU code.
+
+
+NOTE
+====
+* Support for multi-threaded applications
+Sharing a device context between threads does not work
+CUDA 3.2 or earlier versions.
+CUDA 4.0 supports sharing GPU context among multiple threads,
+but we have not tested this capability with our code.
+
+* 64-bit native integer arithmetic for RSA
+RSA codes exploits 64-bit native integer support in CUDA 2.x devices.
+Use of 64-bit native integer or not is decided during build process.
+If you have CUDA 2.x device, and our codes does not utilize it correctly,
+then you may fix it by modifying IS_FERMI variable in Makefile.in file.
+
+* Alignment issue
+AES and SHA1 GPU codes can handle non multiple of 16 bytes data,
+however input data should be aligned before passed into GPU.
+The current test code that we provide do not deal with it,
+and will not work correctly if you try non-multiple of 16 bytes.
+
+
+CONTACTS
+========
+More information about the project is available at 
+http://shader.kaist.edu/sslshader
+
+If you have trouble using the library
+please contact us at tengig@an.kaist.ac.kr.
diff --git a/README.md b/README.md
@@ -0,0 +1,7 @@
+# libgpucrypto
+
+I did not write this, nor do I maintain a fork of this code. 
+
+It took me a long time to find this code, as the original authors no longer share the source-code, however the license explicitly allows redistribution.
+
+As a result of the above, I ham placing this code back in the public domain, in hopes that others can benefit.