use ChaCha8-based CSPRNG

Makefile

@@ -2,14 +2,15 @@ CPPFLAGS := -D_GNU_SOURCE
 CFLAGS := -std=c11 -Wall -Wextra -O2 -flto -fPIC -fvisibility=hidden -pedantic
 LDFLAGS := -Wl,--as-needed
 LDLIBS := -lpthread
-OBJECTS := malloc.o random.o util.o
+OBJECTS := chacha.o malloc.o random.o util.o
 
 hardened_malloc.so: $(OBJECTS)
 	$(CC) $(CFLAGS) $(LDFLAGS) -shared $^ $(LDLIBS) -o $@
 
 malloc.o: malloc.c malloc.h random.h util.h
-random.o: random.c random.h util.h
+random.o: random.c random.h chacha.h util.h
 util.o: util.c util.h
+chacha.o: chacha.c chacha.h
 
 clean:
 	rm -f hardened_malloc.so $(OBJECTS)

chacha.c

@@ -0,0 +1,188 @@
+/*
+Based on chacha-merged.c version 20080118
+D. J. Bernstein
+Public domain.
+*/
+
+#include "chacha.h"
+
+#define U8C(v) (v##U)
+#define U32C(v) (v##U)
+
+#define U8V(v) ((u8)(v) & U8C(0xFF))
+#define U32V(v) ((u32)(v) & U32C(0xFFFFFFFF))
+
+#define ROTL32(v, n) \
+  (U32V((v) << (n)) | ((v) >> (32 - (n))))
+
+#define U8TO32_LITTLE(p) \
+  (((u32)((p)[0])      ) | \
+   ((u32)((p)[1]) <<  8) | \
+   ((u32)((p)[2]) << 16) | \
+   ((u32)((p)[3]) << 24))
+
+#define U32TO8_LITTLE(p, v) \
+  do { \
+    (p)[0] = U8V((v)      ); \
+    (p)[1] = U8V((v) >>  8); \
+    (p)[2] = U8V((v) >> 16); \
+    (p)[3] = U8V((v) >> 24); \
+  } while (0)
+
+#define ROTATE(v,c) (ROTL32(v,c))
+#define XOR(v,w) ((v) ^ (w))
+#define PLUS(v,w) (U32V((v) + (w)))
+#define PLUSONE(v) (PLUS((v),1))
+
+#define QUARTERROUND(a,b,c,d) \
+  a = PLUS(a,b); d = ROTATE(XOR(d,a),16); \
+  c = PLUS(c,d); b = ROTATE(XOR(b,c),12); \
+  a = PLUS(a,b); d = ROTATE(XOR(d,a), 8); \
+  c = PLUS(c,d); b = ROTATE(XOR(b,c), 7);
+
+static const char sigma[16] = "expand 32-byte k";
+static const char tau[16] = "expand 16-byte k";
+
+void chacha_keysetup(chacha_ctx *x,const u8 *k,u32 kbits)
+{
+  const char *constants;
+
+  x->input[4] = U8TO32_LITTLE(k + 0);
+  x->input[5] = U8TO32_LITTLE(k + 4);
+  x->input[6] = U8TO32_LITTLE(k + 8);
+  x->input[7] = U8TO32_LITTLE(k + 12);
+  if (kbits == 256) { /* recommended */
+    k += 16;
+    constants = sigma;
+  } else { /* kbits == 128 */
+    constants = tau;
+  }
+  x->input[8] = U8TO32_LITTLE(k + 0);
+  x->input[9] = U8TO32_LITTLE(k + 4);
+  x->input[10] = U8TO32_LITTLE(k + 8);
+  x->input[11] = U8TO32_LITTLE(k + 12);
+  x->input[0] = U8TO32_LITTLE(constants + 0);
+  x->input[1] = U8TO32_LITTLE(constants + 4);
+  x->input[2] = U8TO32_LITTLE(constants + 8);
+  x->input[3] = U8TO32_LITTLE(constants + 12);
+}
+
+void chacha_ivsetup(chacha_ctx *x,const u8 *iv)
+{
+  x->input[12] = 0;
+  x->input[13] = 0;
+  x->input[14] = U8TO32_LITTLE(iv + 0);
+  x->input[15] = U8TO32_LITTLE(iv + 4);
+}
+
+void
+chacha_keystream_bytes(chacha_ctx *x,u8 *c,u32 bytes)
+{
+  u32 x0, x1, x2, x3, x4, x5, x6, x7, x8, x9, x10, x11, x12, x13, x14, x15;
+  u32 j0, j1, j2, j3, j4, j5, j6, j7, j8, j9, j10, j11, j12, j13, j14, j15;
+  u8 *ctarget;
+  u8 tmp[64];
+  unsigned i;
+
+  if (!bytes) return;
+
+  j0 = x->input[0];
+  j1 = x->input[1];
+  j2 = x->input[2];
+  j3 = x->input[3];
+  j4 = x->input[4];
+  j5 = x->input[5];
+  j6 = x->input[6];
+  j7 = x->input[7];
+  j8 = x->input[8];
+  j9 = x->input[9];
+  j10 = x->input[10];
+  j11 = x->input[11];
+  j12 = x->input[12];
+  j13 = x->input[13];
+  j14 = x->input[14];
+  j15 = x->input[15];
+
+  for (;;) {
+    if (bytes < 64) {
+      ctarget = c;
+      c = tmp;
+    }
+    x0 = j0;
+    x1 = j1;
+    x2 = j2;
+    x3 = j3;
+    x4 = j4;
+    x5 = j5;
+    x6 = j6;
+    x7 = j7;
+    x8 = j8;
+    x9 = j9;
+    x10 = j10;
+    x11 = j11;
+    x12 = j12;
+    x13 = j13;
+    x14 = j14;
+    x15 = j15;
+    for (i = 8;i > 0;i -= 2) {
+      QUARTERROUND( x0, x4, x8,x12)
+      QUARTERROUND( x1, x5, x9,x13)
+      QUARTERROUND( x2, x6,x10,x14)
+      QUARTERROUND( x3, x7,x11,x15)
+      QUARTERROUND( x0, x5,x10,x15)
+      QUARTERROUND( x1, x6,x11,x12)
+      QUARTERROUND( x2, x7, x8,x13)
+      QUARTERROUND( x3, x4, x9,x14)
+    }
+    x0 = PLUS(x0,j0);
+    x1 = PLUS(x1,j1);
+    x2 = PLUS(x2,j2);
+    x3 = PLUS(x3,j3);
+    x4 = PLUS(x4,j4);
+    x5 = PLUS(x5,j5);
+    x6 = PLUS(x6,j6);
+    x7 = PLUS(x7,j7);
+    x8 = PLUS(x8,j8);
+    x9 = PLUS(x9,j9);
+    x10 = PLUS(x10,j10);
+    x11 = PLUS(x11,j11);
+    x12 = PLUS(x12,j12);
+    x13 = PLUS(x13,j13);
+    x14 = PLUS(x14,j14);
+    x15 = PLUS(x15,j15);
+
+    j12 = PLUSONE(j12);
+    if (!j12) {
+      j13 = PLUSONE(j13);
+      /* stopping at 2^70 bytes per nonce is user's responsibility */
+    }
+
+    U32TO8_LITTLE(c + 0,x0);
+    U32TO8_LITTLE(c + 4,x1);
+    U32TO8_LITTLE(c + 8,x2);
+    U32TO8_LITTLE(c + 12,x3);
+    U32TO8_LITTLE(c + 16,x4);
+    U32TO8_LITTLE(c + 20,x5);
+    U32TO8_LITTLE(c + 24,x6);
+    U32TO8_LITTLE(c + 28,x7);
+    U32TO8_LITTLE(c + 32,x8);
+    U32TO8_LITTLE(c + 36,x9);
+    U32TO8_LITTLE(c + 40,x10);
+    U32TO8_LITTLE(c + 44,x11);
+    U32TO8_LITTLE(c + 48,x12);
+    U32TO8_LITTLE(c + 52,x13);
+    U32TO8_LITTLE(c + 56,x14);
+    U32TO8_LITTLE(c + 60,x15);
+
+    if (bytes <= 64) {
+      if (bytes < 64) {
+        for (i = 0;i < bytes;++i) ctarget[i] = c[i];
+      }
+      x->input[12] = j12;
+      x->input[13] = j13;
+      return;
+    }
+    bytes -= 64;
+    c += 64;
+  }
+}

chacha.h

@@ -0,0 +1,18 @@
+#ifndef CHACHA_H
+#define CHACHA_H
+
+#include <stdint.h>
+
+typedef uint8_t u8;
+typedef uint32_t u32;
+
+typedef struct
+{
+  u32 input[16];
+} chacha_ctx;
+
+void chacha_keysetup(chacha_ctx *x,const u8 *k,u32 kbits);
+void chacha_ivsetup(chacha_ctx *x,const u8 *iv);
+void chacha_keystream_bytes(chacha_ctx *x,u8 *c,u32 bytes);
+
+#endif

random.c

@@ -16,6 +16,8 @@ static ssize_t getrandom(void *buf, size_t buflen, unsigned int flags) {
 }
 #endif
 
+#include "chacha.h"
+
 static void get_random_seed(void *buf, size_t size) {
     while (size > 0) {
         ssize_t r;
@@ -33,20 +35,34 @@ static void get_random_seed(void *buf, size_t size) {
     }
 }
 
+#define KEY_SIZE 32
+#define IV_SIZE 8
+
 void random_state_init(struct random_state *state) {
-    state->index = RANDOM_CACHE_SIZE;
+    uint8_t rnd[KEY_SIZE + IV_SIZE];
+    get_random_seed(rnd, sizeof(rnd));
+    chacha_keysetup(&state->ctx, rnd, KEY_SIZE * 8);
+    chacha_ivsetup(&state->ctx, rnd + KEY_SIZE);
+    chacha_keystream_bytes(&state->ctx, state->cache, RANDOM_CACHE_SIZE);
+    state->index = 0;
+    state->reseed = 0;
 }
 
 void get_random_bytes(struct random_state *state, void *buf, size_t size) {
     if (size > RANDOM_CACHE_SIZE / 2) {
-        get_random_seed(buf, size);
+        chacha_keystream_bytes(&state->ctx, state->cache, RANDOM_CACHE_SIZE);
         return;
     }
 
     while (size) {
         if (state->index == RANDOM_CACHE_SIZE) {
-            state->index = 0;
-            get_random_seed(state->cache, RANDOM_CACHE_SIZE);
+            if (state->reseed >= RANDOM_RESEED_SIZE) {
+                random_state_init(state);
+            } else {
+                chacha_keystream_bytes(&state->ctx, state->cache, RANDOM_CACHE_SIZE);
+                state->index = 0;
+                state->reseed += RANDOM_CACHE_SIZE;
+            }
         }
         size_t remaining = RANDOM_CACHE_SIZE - state->index;
         size_t copy_size = size < remaining ? size : remaining;

random.h

@@ -3,10 +3,15 @@
 
 #include <stdint.h>
 
-#define RANDOM_CACHE_SIZE 4096
+#include "chacha.h"
+
+#define RANDOM_CACHE_SIZE 256ULL
+#define RANDOM_RESEED_SIZE 256ULL * 1024
 
 struct random_state {
     size_t index;
+    size_t reseed;
+    chacha_ctx ctx;
     uint8_t cache[RANDOM_CACHE_SIZE];
 };