Crypto++  8.8
Free C++ class library of cryptographic schemes
cham.cpp
1 // cham.cpp - written and placed in the public domain by Kim Sung Hee and Jeffrey Walton
2 // Based on "CHAM: A Family of Lightweight Block Ciphers for
3 // Resource-Constrained Devices" by Bonwook Koo, Dongyoung Roh,
4 // Hyeonjin Kim, Younghoon Jung, Dong-Geon Lee, and Daesung Kwon
5 
6 #include "pch.h"
7 #include "config.h"
8 
9 #include "cham.h"
10 #include "misc.h"
11 #include "cpu.h"
12 
13 // CHAM table of parameters
14 // +-------------------------------------------------
15 // +cipher n k r w k/w
16 // +-------------------------------------------------
17 // +CHAM-64/128 64 128 80 16 8
18 // +CHAM-128/128 128 128 80 32 4
19 // +CHAM-128/256 128 256 96 32 8
20 // +-------------------------------------------------
21 
22 ANONYMOUS_NAMESPACE_BEGIN
23 
26 
27 /// \brief CHAM encryption round
28 /// \tparam RR the round number residue
29 /// \tparam KW the number of key words
30 /// \tparam T words type
31 /// \param x the state array
32 /// \param k the subkey table
33 /// \param i the round number
34 /// \details CHAM_EncRound applies the encryption round to the plain text.
35 /// RR is the "round residue" and it is used modulo 4. ProcessAndXorBlock
36 /// may provide a fully unrolled encryption transformation, or provide
37 /// a transformation that loops using multiples of 4 encryption rounds.
38 /// \details CHAM_EncRound calculates indexes into the x[] array based
39 /// on the round number residue. There is no need for the assignments
40 /// that shift values in preparations for the next round.
41 /// \details CHAM_EncRound depends on the round number. The actual round
42 /// being executed is passed through the parameter <tt>i</tt>. If
43 /// ProcessAndXorBlock fully unrolled the loop then the parameter
44 /// <tt>i</tt> would be unnecessary.
45 template <unsigned int RR, unsigned int KW, class T>
46 inline void CHAM_EncRound(T x[4], const T k[KW], unsigned int i)
47 {
48  CRYPTOPP_CONSTANT(IDX0 = (RR+0) % 4);
49  CRYPTOPP_CONSTANT(IDX1 = (RR+1) % 4);
50  CRYPTOPP_CONSTANT(IDX3 = (RR+3+1) % 4);
51  CRYPTOPP_CONSTANT(R1 = (RR % 2 == 0) ? 1 : 8);
52  CRYPTOPP_CONSTANT(R2 = (RR % 2 == 0) ? 8 : 1);
53 
54  // Follows conventions in the ref impl
55  const T kk = k[i % KW];
56  const T aa = x[IDX0] ^ static_cast<T>(i);
57  const T bb = rotlConstant<R1>(x[IDX1]) ^ kk;
58  x[IDX3] = rotlConstant<R2>(static_cast<T>(aa + bb));
59 }
60 
61 /// \brief CHAM decryption round
62 /// \tparam RR the round number residue
63 /// \tparam KW the number of key words
64 /// \tparam T words type
65 /// \param x the state array
66 /// \param k the subkey table
67 /// \param i the round number
68 /// \details CHAM_DecRound applies the decryption round to the cipher text.
69 /// RR is the "round residue" and it is used modulo 4. ProcessAndXorBlock
70 /// may provide a fully unrolled decryption transformation, or provide
71 /// a transformation that loops using multiples of 4 decryption rounds.
72 /// \details CHAM_DecRound calculates indexes into the x[] array based
73 /// on the round number residue. There is no need for the assignments
74 /// that shift values in preparations for the next round.
75 /// \details CHAM_DecRound depends on the round number. The actual round
76 /// being executed is passed through the parameter <tt>i</tt>. If
77 /// ProcessAndXorBlock fully unrolled the loop then the parameter
78 /// <tt>i</tt> would be unnecessary.
79 template <unsigned int RR, unsigned int KW, class T>
80 inline void CHAM_DecRound(T x[4], const T k[KW], unsigned int i)
81 {
82  CRYPTOPP_CONSTANT(IDX0 = (RR+0) % 4);
83  CRYPTOPP_CONSTANT(IDX1 = (RR+1) % 4);
84  CRYPTOPP_CONSTANT(IDX3 = (RR+3+1) % 4);
85  CRYPTOPP_CONSTANT(R1 = (RR % 2 == 0) ? 8 : 1);
86  CRYPTOPP_CONSTANT(R2 = (RR % 2 == 0) ? 1 : 8);
87 
88  // Follows conventions in the ref impl
89  const T kk = k[i % KW];
90  const T aa = rotrConstant<R1>(x[IDX3]);
91  const T bb = rotlConstant<R2>(x[IDX1]) ^ kk;
92  x[IDX0] = static_cast<T>(aa - bb) ^ static_cast<T>(i);
93 }
94 
95 ANONYMOUS_NAMESPACE_END
96 
97 NAMESPACE_BEGIN(CryptoPP)
98 
99 #if CRYPTOPP_CHAM128_ADVANCED_PROCESS_BLOCKS
100 # if (CRYPTOPP_SSSE3_AVAILABLE)
101 extern size_t CHAM64_Enc_AdvancedProcessBlocks_SSSE3(const word16* subKeys, size_t rounds,
102  const byte *inBlocks, const byte *xorBlocks, byte *outBlocks, size_t length, word32 flags);
103 
104 extern size_t CHAM64_Dec_AdvancedProcessBlocks_SSSE3(const word16* subKeys, size_t rounds,
105  const byte *inBlocks, const byte *xorBlocks, byte *outBlocks, size_t length, word32 flags);
106 
107 extern size_t CHAM128_Enc_AdvancedProcessBlocks_SSSE3(const word32* subKeys, size_t rounds,
108  const byte *inBlocks, const byte *xorBlocks, byte *outBlocks, size_t length, word32 flags);
109 
110 extern size_t CHAM128_Dec_AdvancedProcessBlocks_SSSE3(const word32* subKeys, size_t rounds,
111  const byte *inBlocks, const byte *xorBlocks, byte *outBlocks, size_t length, word32 flags);
112 # endif // CRYPTOPP_SSSE3_AVAILABLE
113 #endif // CRYPTOPP_CHAM128_ADVANCED_PROCESS_BLOCKS
114 
115 void CHAM64::Base::UncheckedSetKey(const byte *userKey, unsigned int keyLength, const NameValuePairs &params)
116 {
117  CRYPTOPP_UNUSED(params);
118  m_kw = keyLength/sizeof(word16);
119  m_rk.New(2*m_kw);
120 
121  for (size_t i = 0; i < m_kw; userKey += sizeof(word32))
122  {
123  // Do not cast the buffer. It will SIGBUS on some ARM and SPARC.
124  const word32 rk = GetWord<word32>(false, BIG_ENDIAN_ORDER, userKey);
125 
126  const word16 rk1 = static_cast<word16>(rk >> 16);
127  m_rk[i] = rk1 ^ rotlConstant<1>(rk1) ^ rotlConstant<8>(rk1);
128  m_rk[(i + m_kw) ^ 1] = rk1 ^ rotlConstant<1>(rk1) ^ rotlConstant<11>(rk1);
129  i++;
130 
131  const word16 rk2 = static_cast<word16>(rk & 0xffff);
132  m_rk[i] = rk2 ^ rotlConstant<1>(rk2) ^ rotlConstant<8>(rk2);
133  m_rk[(i + m_kw) ^ 1] = rk2 ^ rotlConstant<1>(rk2) ^ rotlConstant<11>(rk2);
134  i++;
135  }
136 }
137 
138 void CHAM64::Enc::ProcessAndXorBlock(const byte *inBlock, const byte *xorBlock, byte *outBlock) const
139 {
140  // Do not cast the buffer. It will SIGBUS on some ARM and SPARC.
141  GetBlock<word16, BigEndian> iblock(inBlock);
142  iblock(m_x[0])(m_x[1])(m_x[2])(m_x[3]);
143 
144  const int R = 80;
145  for (int i = 0; i < R; i+=16)
146  {
147  CHAM_EncRound< 0, 16>(m_x.begin(), m_rk.begin(), i+0);
148  CHAM_EncRound< 1, 16>(m_x.begin(), m_rk.begin(), i+1);
149  CHAM_EncRound< 2, 16>(m_x.begin(), m_rk.begin(), i+2);
150  CHAM_EncRound< 3, 16>(m_x.begin(), m_rk.begin(), i+3);
151  CHAM_EncRound< 4, 16>(m_x.begin(), m_rk.begin(), i+4);
152  CHAM_EncRound< 5, 16>(m_x.begin(), m_rk.begin(), i+5);
153  CHAM_EncRound< 6, 16>(m_x.begin(), m_rk.begin(), i+6);
154  CHAM_EncRound< 7, 16>(m_x.begin(), m_rk.begin(), i+7);
155  CHAM_EncRound< 8, 16>(m_x.begin(), m_rk.begin(), i+8);
156  CHAM_EncRound< 9, 16>(m_x.begin(), m_rk.begin(), i+9);
157  CHAM_EncRound<10, 16>(m_x.begin(), m_rk.begin(), i+10);
158  CHAM_EncRound<11, 16>(m_x.begin(), m_rk.begin(), i+11);
159  CHAM_EncRound<12, 16>(m_x.begin(), m_rk.begin(), i+12);
160  CHAM_EncRound<13, 16>(m_x.begin(), m_rk.begin(), i+13);
161  CHAM_EncRound<14, 16>(m_x.begin(), m_rk.begin(), i+14);
162  CHAM_EncRound<15, 16>(m_x.begin(), m_rk.begin(), i+15);
163  }
164 
165  PutBlock<word16, BigEndian> oblock(xorBlock, outBlock);
166  oblock(m_x[0])(m_x[1])(m_x[2])(m_x[3]);
167 }
168 
169 void CHAM64::Dec::ProcessAndXorBlock(const byte *inBlock, const byte *xorBlock, byte *outBlock) const
170 {
171  // Do not cast the buffer. It will SIGBUS on some ARM and SPARC.
172  GetBlock<word16, BigEndian> iblock(inBlock);
173  iblock(m_x[0])(m_x[1])(m_x[2])(m_x[3]);
174 
175  const int R = 80;
176  for (int i = R-1; i >=0 ; i-=16)
177  {
178  CHAM_DecRound<15, 16>(m_x.begin(), m_rk.begin(), i-0);
179  CHAM_DecRound<14, 16>(m_x.begin(), m_rk.begin(), i-1);
180  CHAM_DecRound<13, 16>(m_x.begin(), m_rk.begin(), i-2);
181  CHAM_DecRound<12, 16>(m_x.begin(), m_rk.begin(), i-3);
182  CHAM_DecRound<11, 16>(m_x.begin(), m_rk.begin(), i-4);
183  CHAM_DecRound<10, 16>(m_x.begin(), m_rk.begin(), i-5);
184  CHAM_DecRound< 9, 16>(m_x.begin(), m_rk.begin(), i-6);
185  CHAM_DecRound< 8, 16>(m_x.begin(), m_rk.begin(), i-7);
186  CHAM_DecRound< 7, 16>(m_x.begin(), m_rk.begin(), i-8);
187  CHAM_DecRound< 6, 16>(m_x.begin(), m_rk.begin(), i-9);
188  CHAM_DecRound< 5, 16>(m_x.begin(), m_rk.begin(), i-10);
189  CHAM_DecRound< 4, 16>(m_x.begin(), m_rk.begin(), i-11);
190  CHAM_DecRound< 3, 16>(m_x.begin(), m_rk.begin(), i-12);
191  CHAM_DecRound< 2, 16>(m_x.begin(), m_rk.begin(), i-13);
192  CHAM_DecRound< 1, 16>(m_x.begin(), m_rk.begin(), i-14);
193  CHAM_DecRound< 0, 16>(m_x.begin(), m_rk.begin(), i-15);
194  }
195 
196  PutBlock<word16, BigEndian> oblock(xorBlock, outBlock);
197  oblock(m_x[0])(m_x[1])(m_x[2])(m_x[3]);
198 }
199 
200 std::string CHAM128::Base::AlgorithmProvider() const
201 {
202 #if defined(CRYPTOPP_SSSE3_AVAILABLE)
203  if (HasSSSE3())
204  return "SSSE3";
205 #endif
206  return "C++";
207 }
208 
209 void CHAM128::Base::UncheckedSetKey(const byte *userKey, unsigned int keyLength, const NameValuePairs &params)
210 {
211  CRYPTOPP_UNUSED(params);
212  m_kw = keyLength/sizeof(word32);
213  m_rk.New(2*m_kw);
214 
215  for (size_t i = 0; i < m_kw; userKey += sizeof(word32))
216  {
217  // Do not cast the buffer. It will SIGBUS on some ARM and SPARC.
218  const word32 rk = GetWord<word32>(false, BIG_ENDIAN_ORDER, userKey);
219  m_rk[i] = rk ^ rotlConstant<1>(rk) ^ rotlConstant<8>(rk);
220  m_rk[(i + m_kw) ^ 1] = rk ^ rotlConstant<1>(rk) ^ rotlConstant<11>(rk);
221  i++;
222  }
223 }
224 
225 void CHAM128::Enc::ProcessAndXorBlock(const byte *inBlock, const byte *xorBlock, byte *outBlock) const
226 {
227  // Do not cast the buffer. It will SIGBUS on some ARM and SPARC.
228  GetBlock<word32, BigEndian> iblock(inBlock);
229  iblock(m_x[0])(m_x[1])(m_x[2])(m_x[3]);
230 
231  switch (m_kw)
232  {
233  case 4: // 128-bit key
234  {
235  const int R = 80;
236  for (int i = 0; i < R; i+=8)
237  {
238  CHAM_EncRound<0, 8>(m_x.begin(), m_rk.begin(), i+0);
239  CHAM_EncRound<1, 8>(m_x.begin(), m_rk.begin(), i+1);
240  CHAM_EncRound<2, 8>(m_x.begin(), m_rk.begin(), i+2);
241  CHAM_EncRound<3, 8>(m_x.begin(), m_rk.begin(), i+3);
242  CHAM_EncRound<4, 8>(m_x.begin(), m_rk.begin(), i+4);
243  CHAM_EncRound<5, 8>(m_x.begin(), m_rk.begin(), i+5);
244  CHAM_EncRound<6, 8>(m_x.begin(), m_rk.begin(), i+6);
245  CHAM_EncRound<7, 8>(m_x.begin(), m_rk.begin(), i+7);
246  }
247  break;
248  }
249  case 8: // 256-bit key
250  {
251  const int R = 96;
252  for (int i = 0; i < R; i+=16)
253  {
254  CHAM_EncRound< 0, 16>(m_x.begin(), m_rk.begin(), i+0);
255  CHAM_EncRound< 1, 16>(m_x.begin(), m_rk.begin(), i+1);
256  CHAM_EncRound< 2, 16>(m_x.begin(), m_rk.begin(), i+2);
257  CHAM_EncRound< 3, 16>(m_x.begin(), m_rk.begin(), i+3);
258  CHAM_EncRound< 4, 16>(m_x.begin(), m_rk.begin(), i+4);
259  CHAM_EncRound< 5, 16>(m_x.begin(), m_rk.begin(), i+5);
260  CHAM_EncRound< 6, 16>(m_x.begin(), m_rk.begin(), i+6);
261  CHAM_EncRound< 7, 16>(m_x.begin(), m_rk.begin(), i+7);
262  CHAM_EncRound< 8, 16>(m_x.begin(), m_rk.begin(), i+8);
263  CHAM_EncRound< 9, 16>(m_x.begin(), m_rk.begin(), i+9);
264  CHAM_EncRound<10, 16>(m_x.begin(), m_rk.begin(), i+10);
265  CHAM_EncRound<11, 16>(m_x.begin(), m_rk.begin(), i+11);
266  CHAM_EncRound<12, 16>(m_x.begin(), m_rk.begin(), i+12);
267  CHAM_EncRound<13, 16>(m_x.begin(), m_rk.begin(), i+13);
268  CHAM_EncRound<14, 16>(m_x.begin(), m_rk.begin(), i+14);
269  CHAM_EncRound<15, 16>(m_x.begin(), m_rk.begin(), i+15);
270  }
271  break;
272  }
273  default:
274  CRYPTOPP_ASSERT(0);
275  }
276 
277  PutBlock<word32, BigEndian> oblock(xorBlock, outBlock);
278  oblock(m_x[0])(m_x[1])(m_x[2])(m_x[3]);
279 }
280 
281 void CHAM128::Dec::ProcessAndXorBlock(const byte *inBlock, const byte *xorBlock, byte *outBlock) const
282 {
283  // Do not cast the buffer. It will SIGBUS on some ARM and SPARC.
284  GetBlock<word32, BigEndian> iblock(inBlock);
285  iblock(m_x[0])(m_x[1])(m_x[2])(m_x[3]);
286 
287  switch (m_kw)
288  {
289  case 4: // 128-bit key
290  {
291  const int R = 80;
292  for (int i = R-1; i >= 0; i-=8)
293  {
294  CHAM_DecRound<7, 8>(m_x.begin(), m_rk.begin(), i-0);
295  CHAM_DecRound<6, 8>(m_x.begin(), m_rk.begin(), i-1);
296  CHAM_DecRound<5, 8>(m_x.begin(), m_rk.begin(), i-2);
297  CHAM_DecRound<4, 8>(m_x.begin(), m_rk.begin(), i-3);
298  CHAM_DecRound<3, 8>(m_x.begin(), m_rk.begin(), i-4);
299  CHAM_DecRound<2, 8>(m_x.begin(), m_rk.begin(), i-5);
300  CHAM_DecRound<1, 8>(m_x.begin(), m_rk.begin(), i-6);
301  CHAM_DecRound<0, 8>(m_x.begin(), m_rk.begin(), i-7);
302  }
303  break;
304  }
305  case 8: // 256-bit key
306  {
307  const int R = 96;
308  for (int i = R-1; i >= 0; i-=16)
309  {
310  CHAM_DecRound<15, 16>(m_x.begin(), m_rk.begin(), i-0);
311  CHAM_DecRound<14, 16>(m_x.begin(), m_rk.begin(), i-1);
312  CHAM_DecRound<13, 16>(m_x.begin(), m_rk.begin(), i-2);
313  CHAM_DecRound<12, 16>(m_x.begin(), m_rk.begin(), i-3);
314  CHAM_DecRound<11, 16>(m_x.begin(), m_rk.begin(), i-4);
315  CHAM_DecRound<10, 16>(m_x.begin(), m_rk.begin(), i-5);
316  CHAM_DecRound< 9, 16>(m_x.begin(), m_rk.begin(), i-6);
317  CHAM_DecRound< 8, 16>(m_x.begin(), m_rk.begin(), i-7);
318  CHAM_DecRound< 7, 16>(m_x.begin(), m_rk.begin(), i-8);
319  CHAM_DecRound< 6, 16>(m_x.begin(), m_rk.begin(), i-9);
320  CHAM_DecRound< 5, 16>(m_x.begin(), m_rk.begin(), i-10);
321  CHAM_DecRound< 4, 16>(m_x.begin(), m_rk.begin(), i-11);
322  CHAM_DecRound< 3, 16>(m_x.begin(), m_rk.begin(), i-12);
323  CHAM_DecRound< 2, 16>(m_x.begin(), m_rk.begin(), i-13);
324  CHAM_DecRound< 1, 16>(m_x.begin(), m_rk.begin(), i-14);
325  CHAM_DecRound< 0, 16>(m_x.begin(), m_rk.begin(), i-15);
326  }
327  break;
328  }
329  default:
330  CRYPTOPP_ASSERT(0);
331  }
332 
333  PutBlock<word32, BigEndian> oblock(xorBlock, outBlock);
334  oblock(m_x[0])(m_x[1])(m_x[2])(m_x[3]);
335 }
336 
337 #if CRYPTOPP_CHAM128_ADVANCED_PROCESS_BLOCKS
338 size_t CHAM128::Enc::AdvancedProcessBlocks(const byte *inBlocks, const byte *xorBlocks,
339  byte *outBlocks, size_t length, word32 flags) const
340 {
341 # if (CRYPTOPP_SSSE3_AVAILABLE)
342  if (HasSSSE3()) {
343  const size_t rounds = (m_kw == 4 ? 80 : 96);
344  return CHAM128_Enc_AdvancedProcessBlocks_SSSE3(m_rk, rounds,
345  inBlocks, xorBlocks, outBlocks, length, flags);
346  }
347 # endif // CRYPTOPP_SSSE3_AVAILABLE
348  return BlockTransformation::AdvancedProcessBlocks(inBlocks, xorBlocks, outBlocks, length, flags);
349 }
350 
351 size_t CHAM128::Dec::AdvancedProcessBlocks(const byte *inBlocks, const byte *xorBlocks,
352  byte *outBlocks, size_t length, word32 flags) const
353 {
354 # if (CRYPTOPP_SSSE3_AVAILABLE)
355  if (HasSSSE3()) {
356  const size_t rounds = (m_kw == 4 ? 80 : 96);
357  return CHAM128_Dec_AdvancedProcessBlocks_SSSE3(m_rk, rounds,
358  inBlocks, xorBlocks, outBlocks, length, flags);
359  }
360 # endif // CRYPTOPP_SSSE3_AVAILABLE
361  return BlockTransformation::AdvancedProcessBlocks(inBlocks, xorBlocks, outBlocks, length, flags);
362 }
363 #endif // CRYPTOPP_CHAM128_ADVANCED_PROCESS_BLOCKS
364 
365 NAMESPACE_END
Classes for the CHAM block cipher.
virtual size_t AdvancedProcessBlocks(const byte *inBlocks, const byte *xorBlocks, byte *outBlocks, size_t length, word32 flags) const
Encrypt and xor multiple blocks using additional flags.
Access a block of memory.
Definition: misc.h:2975
Interface for retrieving values given their names.
Definition: cryptlib.h:327
Access a block of memory.
Definition: misc.h:3016
void New(size_type newSize)
Change size without preserving contents.
Definition: secblock.h:1126
Library configuration file.
unsigned int word32
32-bit unsigned datatype
Definition: config_int.h:72
unsigned short word16
16-bit unsigned datatype
Definition: config_int.h:69
Functions for CPU features and intrinsics.
@ BIG_ENDIAN_ORDER
byte order is big-endian
Definition: cryptlib.h:152
Utility functions for the Crypto++ library.
T rotlConstant(T x)
Performs a left rotate.
Definition: misc.h:1757
T rotrConstant(T x)
Performs a right rotate.
Definition: misc.h:1783
Crypto++ library namespace.
Precompiled header file.
#define CRYPTOPP_ASSERT(exp)
Debugging and diagnostic assertion.
Definition: trap.h:68