Improve locks and signals

- Introduce fast spinlock API
- Double rand64() perf w/ spinlock
- Improve raise() on New Technology
- Support gettid() across platforms
- Implement SA_NODEFER on New Technology
- Move the lock intrinsics into LIBC_INTRIN
- Make SIGTRAP recoverable on New Technology
- Block SIGCHLD in wait4() on New Technology
- Add threading prototypes for XNU and FreeBSD
- Rewrite abort() fixing its minor bugs on XNU/NT
- Shave down a lot of the content in libc/bits/bits.h
- Let signal handlers modify CPU registers on New Technology

third_party/make/main.c

@@ -781,7 +781,7 @@ handle_runtime_exceptions (struct _EXCEPTION_POINTERS *exinfo)
 
   /* turn this on if we want to put stuff in the event log too */
 #ifdef USE_EVENT_LOG
-  hEventSource = RegisterEventSource (NULL, "GNU Make");
+  hEventSource = RegisterEventSource (NULL, u"GNU Make");
   lpszStrings[0] = errmsg;
 
   if (hEventSource != NULL)

third_party/python/Modules/expat/xmlparse.c

@@ -1,5 +1,4 @@
 #include "libc/assert.h"
-#include "libc/bits/bits.h"
 #include "libc/calls/calls.h"
 #include "libc/calls/struct/timeval.h"
 #include "libc/errno.h"
@@ -596,29 +595,11 @@ static unsigned long get_hash_secret_salt(XML_Parser parser) {
   return parser->m_hash_secret_salt;
 }
 
-static uint64_t getsome(void) {
-  int i;
-  char cf;
-  uint64_t x;
-  if (X86_HAVE(RDRND)) {
-    for (i = 0; i < 10; ++i) {
-      asm volatile(CFLAG_ASM("rdrand\t%1")
-                   : CFLAG_CONSTRAINT(cf), "=r"(x)
-                   : /* no inputs */
-                   : "cc");
-      if (cf) return x;
-      asm volatile("pause");
-    }
-  }
-  if (getrandom(&x, 8, 0) != 8) abort();
-  return x;
-}
-
 static XML_Bool /* only valid for root parser */
 startParsing(XML_Parser parser) {
   /* hash functions must be initialized before setContext() is called */
   if (!parser->m_hash_secret_salt) {
-    parser->m_hash_secret_salt = getsome();
+    parser->m_hash_secret_salt = rdrand();
   }
   if (parser->m_ns) {
     /* implicit context only set for root parser, since child

third_party/python/Modules/getpath.c

@@ -4,10 +4,10 @@
 │ Python 3                                                                     │
 │ https://docs.python.org/3/license.html                                       │
 ╚─────────────────────────────────────────────────────────────────────────────*/
-#include "libc/bits/bits.h"
 #include "libc/calls/calls.h"
 #include "libc/calls/struct/stat.h"
 #include "libc/errno.h"
+#include "libc/intrin/cmpxchg.h"
 #include "libc/log/log.h"
 #include "libc/mem/alloca.h"
 #include "libc/mem/mem.h"
@@ -662,7 +662,7 @@ wchar_t *
 Py_GetProgramFullPath(void)
 {
     static bool once;
-    if (cmpxchg(&once, false, true)) {
+    if (_cmpxchg(&once, false, true)) {
         progpath = utf8toutf32(program_executable_name, -1, 0);
         __cxa_atexit(free, progpath, 0);
     }

third_party/python/Python/getcopyright.c

@@ -4,8 +4,8 @@
 │ Python 3                                                                     │
 │ https://docs.python.org/3/license.html                                       │
 ╚─────────────────────────────────────────────────────────────────────────────*/
-#include "libc/bits/bits.h"
 #include "libc/bits/weaken.h"
+#include "libc/intrin/cmpxchg.h"
 #include "libc/stdio/append.internal.h"
 #include "libc/str/str.h"
 #include "third_party/python/Include/pylifecycle.h"
@@ -30,7 +30,7 @@ Py_GetCopyright(void)
     const char *p;
     static bool once;
     static char *res;
-    if (cmpxchg(&once, 0, 1)) {
+    if (_cmpxchg(&once, 0, 1)) {
         appends(&res, "");
         for (p = *weaken(kLegalNotices); *p; p += strlen(p) + 1) {
             appends(&res, p);

third_party/python/Python/random.c

@@ -5,7 +5,6 @@
 │ https://docs.python.org/3/license.html                                       │
 ╚─────────────────────────────────────────────────────────────────────────────*/
 #include "libc/assert.h"
-#include "libc/bits/bits.h"
 #include "libc/calls/calls.h"
 #include "libc/calls/weirdtypes.h"
 #include "libc/errno.h"
@@ -389,35 +388,10 @@ _PyOS_URandomNonblock(void *buffer, Py_ssize_t size)
     return pyurandom(buffer, size, 0, 1);
 }
 
-static uint64_t
-getsome(void)
-{
-    int i;
-    char cf;
-    uint64_t x;
-    for (i = 0; i < 10; ++i) {
-        asm volatile(CFLAG_ASM("rdrand\t%1")
-                     : CFLAG_CONSTRAINT(cf), "=r"(x)
-                     : /* no inputs */
-                     : "cc");
-        if (cf) return x;
-        asm volatile("pause");
-    }
-    while (getrandom(&x, sizeof(x), GRND_NONBLOCK) != sizeof(x)) {
-        if (errno != EINTR) {
-            x ^= rdtsc();
-            x += getpid();
-            break;
-        }
-    }
-    return x;
-}
-
 void
 _PyRandom_Init(void)
 {
     char *env;
-    uint64_t x;
     const unsigned char *auxrng;
     unsigned char *secret = (unsigned char *)&_Py_HashSecret.uc;
     Py_ssize_t secret_size = sizeof(_Py_HashSecret_t);
@@ -456,7 +430,8 @@ _PyRandom_Init(void)
         }
     }
     else {
-        int res;
+        uint64_t x;
+        int res, i, j;
         /* _PyRandom_Init() is called very early in the Python initialization
            and so exceptions cannot be used (use raise=0).
            _PyRandom_Init() must not block Python initialization: call
@@ -465,15 +440,14 @@ _PyRandom_Init(void)
         /* 
          * [jart] modified to be more efficient
          */
-        x = getsome();
-        memcpy(secret, &x, 8);
-        if ((auxrng = (const unsigned char *)getauxval(AT_RANDOM))) {
-            memcpy(secret + 8, auxrng, 16);
-        } else {
-            x = getsome();
-            memcpy(secret + 8, &x, 8);
-            x = getsome();
-            memcpy(secret + 16, &x, 8);
+        for (i = 0; i < secret_size;) {
+            x = rdrand(); // will failover to getrandom() etc.
+            for (j = 0; j < 8; ++j) {
+                if (i < secret_size) {
+                    secret[i++] = x;
+                    x >>= 8;
+                }
+            }
         }
 #else
         res = pyurandom(secret, secret_size, 0, 0);