Cache current thread ID in TLS (#24699)

While looking at the strace of `corerun` built with logging and
debugging information, I was amazed at the number of gettid() calls it
was making.  While system calls are cheap, they're still not free;
cache this number in the thread local storage area.  Adds a branch, but
it's just a comparison with 0, so it's fine in comparison.

1 files changed, 12 insertions, 5 deletions

diff --git a/src/pal/src/include/pal/thread.hpp b/src/pal/src/include/pal/thread.hpp
index dd48c23e49..6227b721ce 100644
--- a/src/pal/src/include/pal/thread.hpp
+++ b/src/pal/src/include/pal/thread.hpp
@@ -828,25 +828,32 @@ Abstract:
 
 --*/
 #if defined(__linux__)
-#define THREADSilentGetCurrentThreadId() (SIZE_T)syscall(SYS_gettid)
+#define PlatformGetCurrentThreadId() (SIZE_T)syscall(SYS_gettid)
 #elif defined(__APPLE__)
-inline SIZE_T THREADSilentGetCurrentThreadId() {
+inline SIZE_T PlatformGetCurrentThreadId() {
     uint64_t tid;
     pthread_threadid_np(pthread_self(), &tid);
     return (SIZE_T)tid;
 }
 #elif defined(__FreeBSD__)
 #include <sys/thr.h>
-inline SIZE_T THREADSilentGetCurrentThreadId() {
+inline SIZE_T PlatformGetCurrentThreadId() {
     long tid;
     thr_self(&tid);
     return (SIZE_T)tid;
 }
 #elif defined(__NetBSD__)
 #include <lwp.h>
-#define THREADSilentGetCurrentThreadId() (SIZE_T)_lwp_self()
+#define PlatformGetCurrentThreadId() (SIZE_T)_lwp_self()
 #else
-#define THREADSilentGetCurrentThreadId() (SIZE_T)pthread_self()
+#define PlatformGetCurrentThreadId() (SIZE_T)pthread_self()
 #endif
 
+inline SIZE_T THREADSilentGetCurrentThreadId() {
+    static __thread SIZE_T tid;
+    if (!tid)
+        tid = PlatformGetCurrentThreadId();
+    return tid;
+}
+
 #endif // _PAL_THREAD_HPP_