Make numerous improvements

- Python static hello world now 1.8mb
- Python static fully loaded now 10mb
- Python HTTPS client now uses MbedTLS
- Python REPL now completes import stmts
- Increase stack size for Python for now
- Begin synthesizing posixpath and ntpath
- Restore Python \N{UNICODE NAME} support
- Restore Python NFKD symbol normalization
- Add optimized code path for Intel SHA-NI
- Get more Python unit tests passing faster
- Get Python help() pagination working on NT
- Python hashlib now supports MbedTLS PBKDF2
- Make memcpy/memmove/memcmp/bcmp/etc. faster
- Add Mersenne Twister and Vigna to LIBC_RAND
- Provide privileged __printf() for error code
- Fix zipos opendir() so that it reports ENOTDIR
- Add basic chmod() implementation for Windows NT
- Add Cosmo's best functions to Python cosmo module
- Pin function trace indent depth to that of caller
- Show memory diagram on invalid access in MODE=dbg
- Differentiate stack overflow on crash in MODE=dbg
- Add stb_truetype and tools for analyzing font files
- Upgrade to UNICODE 13 and reduce its binary footprint
- COMPILE.COM now logs resource usage of build commands
- Start implementing basic poll() support on bare metal
- Set getauxval(AT_EXECFN) to GetModuleFileName() on NT
- Add descriptions to strerror() in non-TINY build modes
- Add COUNTBRANCH() macro to help with micro-optimizations
- Make error / backtrace / asan / memory code more unbreakable
- Add fast perfect C implementation of μ-Law and a-Law audio codecs
- Make strtol() functions consistent with other libc implementations
- Improve Linenoise implementation (see also github.com/jart/bestline)
- COMPILE.COM now suppresses stdout/stderr of successful build commands

libc/rand/rngset.c

@@ -16,40 +16,75 @@
 │ TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR             │
 │ PERFORMANCE OF THIS SOFTWARE.                                                │
 ╚─────────────────────────────────────────────────────────────────────────────*/
+#include "libc/calls/calls.h"
+#include "libc/intrin/asan.internal.h"
 #include "libc/rand/rand.h"
-#include "libc/rand/xorshift.h"
-#include "libc/str/str.h"
+#include "libc/stdio/stdio.h"
 
 /**
  * Fills memory with random bytes, e.g.
  *
- *     char buf[1024];
- *     rngset(buf, sizeof(buf), rand64, -1);
+ *     char buf[512];
+ *     rngset(buf, sizeof(buf), 0, 0);
+ *
+ * If reseed is zero then the internal PRNG is disabled and bytes are
+ * simply copied in little-endian order from the seed function. If seed
+ * is NULL then the reseed parameter is used as the seed value for the
+ * internal PRNG. If seed!=NULL and reseed>8 then reseed is the number
+ * of bytes after which the seed() function should be called again, to
+ * freshen up the PRNG.
+ *
+ * The main advantage of this generator is that it produces data at 13
+ * gigabytes per second since Vigna's Algorithm vectorizes better than
+ * alternatives, going even faster than xorshift.
  *
- * @param seed can be rand64() and is always called at least once
- * @param reseed is bytes between seed() calls and -1 disables it
  * @return original buf
  */
-void *rngset(void *buf, size_t size, uint64_t seed(void), size_t reseed) {
-  unsigned char *p;
-  uint64_t i, x, state;
-  p = buf;
-  state = seed();
-  for (i = 0; size - i >= sizeof(x); i += sizeof(x)) {
-    x = MarsagliaXorshift64(&state);
-    memcpy(p + i, &x, sizeof(x));
-    if (i >= reseed) {
-      state = seed();
-      p += i;
-      size -= i;
-      i = 0;
+void *rngset(void *b, size_t n, uint64_t seed(void), size_t reseed) {
+  size_t m;
+  uint64_t i, x, t = 0;
+  unsigned char *p = b;
+  if (!seed) {
+    t = reseed;
+    reseed = -1;
+  } else if (reseed < 8) {
+    reseed = 8;
+  }
+  while (n) {
+    if (seed) t = seed();
+    if (!seed || reseed > 8) {
+      n -= (m = reseed < n ? reseed : n);
+      while (m >= 8) {
+        x = (t += 0x9e3779b97f4a7c15);
+        x = (x ^ (x >> 30)) * 0xbf58476d1ce4e5b9;
+        x = (x ^ (x >> 27)) * 0x94d049bb133111eb;
+        x = (x ^ (x >> 31));
+        __builtin_memcpy(p, &x, 8);
+        p += 8;
+        m -= 8;
+      }
+      while (m--) {
+        *p++ = t;
+        t >>= 8;
+      }
+    } else if (n >= 8) {
+      p[0] = (0x00000000000000FF & t) >> 000;
+      p[1] = (0x000000000000FF00 & t) >> 010;
+      p[2] = (0x0000000000FF0000 & t) >> 020;
+      p[3] = (0x00000000FF000000 & t) >> 030;
+      p[4] = (0x000000FF00000000 & t) >> 040;
+      p[5] = (0x0000FF0000000000 & t) >> 050;
+      p[6] = (0x00FF000000000000 & t) >> 060;
+      p[7] = (0xFF00000000000000 & t) >> 070;
+      p += 8;
+      n -= 8;
+    } else {
+      while (n) {
+        *p++ = t;
+        t >>= 8;
+        --n;
+      }
     }
   }
-  if (i < size) {
-    x = MarsagliaXorshift64(&state);
-    for (; i < size; ++i, x >>= 8) {
-      p[i] = x & 0xff;
-    }
-  }
-  return buf;
+  return b;
 }