stdlib: Add arc4random, arc4random_buf, and arc4random_uniform (BZ #4417)

The implementation is based on scalar Chacha20 with per-thread cache.
It uses getrandom or /dev/urandom as fallback to get the initial entropy,
and reseeds the internal state on every 16MB of consumed buffer.

To improve performance and lower memory consumption the per-thread cache
is allocated lazily on first arc4random functions call, and if the
memory allocation fails getentropy or /dev/urandom is used as fallback.
The cache is also cleared on thread exit iff it was initialized (so if
arc4random is not called it is not touched).

Although it is lock-free, arc4random is still not async-signal-safe
(the per thread state is not updated atomically).

The ChaCha20 implementation is based on RFC8439 [1], omitting the final
XOR of the keystream with the plaintext because the plaintext is a
stream of zeros.  This strategy is similar to what OpenBSD arc4random
does.

The arc4random_uniform is based on previous work by Florian Weimer,
where the algorithm is based on Jérémie Lumbroso paper Optimal Discrete
Uniform Generation from Coin Flips, and Applications (2013) [2], who
credits Donald E. Knuth and Andrew C. Yao, The complexity of nonuniform
random number generation (1976), for solving the general case.

The main advantage of this method is the that the unit of randomness is not
the uniform random variable (uint32_t), but a random bit.  It optimizes the
internal buffer sampling by initially consuming a 32-bit random variable
and then sampling byte per byte.  Depending of the upper bound requested,
it might lead to better CPU utilization.

Checked on x86_64-linux-gnu, aarch64-linux, and powerpc64le-linux-gnu.

Co-authored-by: Florian Weimer <fweimer@redhat.com>
Reviewed-by: Yann Droneaud <ydroneaud@opteya.com>

[1] https://datatracker.ietf.org/doc/html/rfc8439
[2] https://arxiv.org/pdf/1304.1916.pdf

1 files changed, 187 insertions, 0 deletions

diff --git a/stdlib/chacha20.c b/stdlib/chacha20.c
new file mode 100644
index 0000000000..c47b8418f2
--- /dev/null
+++ b/stdlib/chacha20.c
@@ -0,0 +1,187 @@
+/* Generic ChaCha20 implementation (used on arc4random).
+   Copyright (C) 2022 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <https://www.gnu.org/licenses/>.  */
+
+#include <array_length.h>
+#include <endian.h>
+#include <stddef.h>
+#include <stdint.h>
+#include <string.h>
+
+/* 32-bit stream position, then 96-bit nonce.  */
+#define CHACHA20_IV_SIZE	16
+#define CHACHA20_KEY_SIZE	32
+
+#define CHACHA20_STATE_LEN	16
+
+/* The ChaCha20 implementation is based on RFC8439 [1], omitting the final
+   XOR of the keystream with the plaintext because the plaintext is a
+   stream of zeros.  */
+
+enum chacha20_constants
+{
+  CHACHA20_CONSTANT_EXPA = 0x61707865U,
+  CHACHA20_CONSTANT_ND_3 = 0x3320646eU,
+  CHACHA20_CONSTANT_2_BY = 0x79622d32U,
+  CHACHA20_CONSTANT_TE_K = 0x6b206574U
+};
+
+static inline uint32_t
+read_unaligned_32 (const uint8_t *p)
+{
+  uint32_t r;
+  memcpy (&r, p, sizeof (r));
+  return r;
+}
+
+static inline void
+write_unaligned_32 (uint8_t *p, uint32_t v)
+{
+  memcpy (p, &v, sizeof (v));
+}
+
+#if __BYTE_ORDER == __BIG_ENDIAN
+# define read_unaligned_le32(p) __builtin_bswap32 (read_unaligned_32 (p))
+# define set_state(v)		__builtin_bswap32 ((v))
+#else
+# define read_unaligned_le32(p) read_unaligned_32 ((p))
+# define set_state(v)		(v)
+#endif
+
+static inline void
+chacha20_init (uint32_t *state, const uint8_t *key, const uint8_t *iv)
+{
+  state[0]  = CHACHA20_CONSTANT_EXPA;
+  state[1]  = CHACHA20_CONSTANT_ND_3;
+  state[2]  = CHACHA20_CONSTANT_2_BY;
+  state[3]  = CHACHA20_CONSTANT_TE_K;
+
+  state[4]  = read_unaligned_le32 (key + 0 * sizeof (uint32_t));
+  state[5]  = read_unaligned_le32 (key + 1 * sizeof (uint32_t));
+  state[6]  = read_unaligned_le32 (key + 2 * sizeof (uint32_t));
+  state[7]  = read_unaligned_le32 (key + 3 * sizeof (uint32_t));
+  state[8]  = read_unaligned_le32 (key + 4 * sizeof (uint32_t));
+  state[9]  = read_unaligned_le32 (key + 5 * sizeof (uint32_t));
+  state[10] = read_unaligned_le32 (key + 6 * sizeof (uint32_t));
+  state[11] = read_unaligned_le32 (key + 7 * sizeof (uint32_t));
+
+  state[12] = read_unaligned_le32 (iv + 0 * sizeof (uint32_t));
+  state[13] = read_unaligned_le32 (iv + 1 * sizeof (uint32_t));
+  state[14] = read_unaligned_le32 (iv + 2 * sizeof (uint32_t));
+  state[15] = read_unaligned_le32 (iv + 3 * sizeof (uint32_t));
+}
+
+static inline uint32_t
+rotl32 (unsigned int shift, uint32_t word)
+{
+  return (word << (shift & 31)) | (word >> ((-shift) & 31));
+}
+
+static void
+state_final (const uint8_t *src, uint8_t *dst, uint32_t v)
+{
+#ifdef CHACHA20_XOR_FINAL
+  v ^= read_unaligned_32 (src);
+#endif
+  write_unaligned_32 (dst, v);
+}
+
+static inline void
+chacha20_block (uint32_t *state, uint8_t *dst, const uint8_t *src)
+{
+  uint32_t x0, x1, x2, x3, x4, x5, x6, x7;
+  uint32_t x8, x9, x10, x11, x12, x13, x14, x15;
+
+  x0 = state[0];
+  x1 = state[1];
+  x2 = state[2];
+  x3 = state[3];
+  x4 = state[4];
+  x5 = state[5];
+  x6 = state[6];
+  x7 = state[7];
+  x8 = state[8];
+  x9 = state[9];
+  x10 = state[10];
+  x11 = state[11];
+  x12 = state[12];
+  x13 = state[13];
+  x14 = state[14];
+  x15 = state[15];
+
+  for (int i = 0; i < 20; i += 2)
+    {
+#define QROUND(_x0, _x1, _x2, _x3) 			\
+  do {							\
+   _x0 = _x0 + _x1; _x3 = rotl32 (16, (_x0 ^ _x3)); 	\
+   _x2 = _x2 + _x3; _x1 = rotl32 (12, (_x1 ^ _x2)); 	\
+   _x0 = _x0 + _x1; _x3 = rotl32 (8,  (_x0 ^ _x3));	\
+   _x2 = _x2 + _x3; _x1 = rotl32 (7,  (_x1 ^ _x2));	\
+  } while(0)
+
+      QROUND (x0, x4, x8,  x12);
+      QROUND (x1, x5, x9,  x13);
+      QROUND (x2, x6, x10, x14);
+      QROUND (x3, x7, x11, x15);
+
+      QROUND (x0, x5, x10, x15);
+      QROUND (x1, x6, x11, x12);
+      QROUND (x2, x7, x8,  x13);
+      QROUND (x3, x4, x9,  x14);
+    }
+
+  state_final (&src[0], &dst[0], set_state (x0 + state[0]));
+  state_final (&src[4], &dst[4], set_state (x1 + state[1]));
+  state_final (&src[8], &dst[8], set_state (x2 + state[2]));
+  state_final (&src[12], &dst[12], set_state (x3 + state[3]));
+  state_final (&src[16], &dst[16], set_state (x4 + state[4]));
+  state_final (&src[20], &dst[20], set_state (x5 + state[5]));
+  state_final (&src[24], &dst[24], set_state (x6 + state[6]));
+  state_final (&src[28], &dst[28], set_state (x7 + state[7]));
+  state_final (&src[32], &dst[32], set_state (x8 + state[8]));
+  state_final (&src[36], &dst[36], set_state (x9 + state[9]));
+  state_final (&src[40], &dst[40], set_state (x10 + state[10]));
+  state_final (&src[44], &dst[44], set_state (x11 + state[11]));
+  state_final (&src[48], &dst[48], set_state (x12 + state[12]));
+  state_final (&src[52], &dst[52], set_state (x13 + state[13]));
+  state_final (&src[56], &dst[56], set_state (x14 + state[14]));
+  state_final (&src[60], &dst[60], set_state (x15 + state[15]));
+
+  state[12]++;
+}
+
+static void
+chacha20_crypt (uint32_t *state, uint8_t *dst, const uint8_t *src,
+		size_t bytes)
+{
+  while (bytes >= CHACHA20_BLOCK_SIZE)
+    {
+      chacha20_block (state, dst, src);
+
+      bytes -= CHACHA20_BLOCK_SIZE;
+      dst += CHACHA20_BLOCK_SIZE;
+      src += CHACHA20_BLOCK_SIZE;
+    }
+
+  if (__glibc_unlikely (bytes != 0))
+    {
+      uint8_t stream[CHACHA20_BLOCK_SIZE];
+      chacha20_block (state, stream, src);
+      memcpy (dst, stream, bytes);
+      explicit_bzero (stream, sizeof stream);
+    }
+}