Make more threading improvements

- ASAN memory morgue is now lockless
- Make C11 atomics header more portable
- Rewrote pthread keys support to be lockless
- Simplify Python's unicode table unpacking code
- Make crash report write(2) closer to being atomic
- Make it possible to strace/ftrace a single thread
- ASAN now checks nul-terminated strings fast and properly
- Windows fork() now restores TLS memory of calling thread

libc/sock/recvmsg.c

@@ -93,7 +93,7 @@ ssize_t recvmsg(int fd, struct msghdr *msg, int flags) {
   }
 
 #if defined(SYSDEBUG) && _DATATRACE
-  if (__strace > 0) {
+  if (__strace > 0 && strace_enabled(0) > 0) {
     if (!msg || (rc == -1 && errno == EFAULT)) {
       DATATRACE("recvmsg(%d, %p, %#x) → %'ld% m", fd, msg, flags, rc);
     } else {

libc/sock/sendmsg.c

@@ -80,7 +80,7 @@ ssize_t sendmsg(int fd, const struct msghdr *msg, int flags) {
   }
 
 #if defined(SYSDEBUG) && _DATATRACE
-  if (__strace > 0) {
+  if (__strace > 0 && strace_enabled(0) > 0) {
     kprintf(STRACE_PROLOGUE "sendmsg(");
     if ((!IsAsan() && kisdangerous(msg)) ||
         (IsAsan() && !__asan_is_valid(msg, sizeof(*msg)))) {

libc/sock/struct/pollfd.internal.h

@@ -2,6 +2,7 @@
 #define COSMOPOLITAN_LIBC_SOCK_STRUCT_POLLFD_INTERNAL_H_
 #include "libc/calls/struct/sigset.h"
 #include "libc/calls/struct/timespec.h"
+#include "libc/mem/alloca.h"
 #include "libc/sock/struct/pollfd.h"
 #if !(__ASSEMBLER__ + __LINKER__ + 0)
 COSMOPOLITAN_C_START_
@@ -12,6 +13,9 @@ int sys_ppoll(struct pollfd *, size_t, const struct timespec *,
 int sys_poll_metal(struct pollfd *, size_t, unsigned);
 int sys_poll_nt(struct pollfd *, uint64_t, uint64_t *, const sigset_t *) hidden;
 
+const char *DescribePollFds(char[300], ssize_t, struct pollfd *, size_t);
+#define DescribePollFds(x, y, z) DescribePollFds(alloca(300), x, y, z)
+
 COSMOPOLITAN_C_END_
 #endif /* !(__ASSEMBLER__ + __LINKER__ + 0) */
 #endif /* COSMOPOLITAN_LIBC_SOCK_STRUCT_POLLFD_INTERNAL_H_ */