Using FindFirstFileA() instead of CreateFileA() to get the file size

Since we already call FindFirstFileA() and it returns the size there is
no need to open/close every file to get it's size

platform_windows.c

@@ -1,3 +1,4 @@
+#include "arena.h"
 #include "platform.h"
 
 // ----------------------------- Globals ------------------------------------
@@ -201,16 +202,7 @@ void scan_folder_windows_parallel(const char *base, DirQueue *q) {
 
       platform_get_file_owner(full, fe->owner, sizeof(fe->owner));
 
-      LARGE_INTEGER size;
-      HANDLE hf =
-          CreateFileA(full, GENERIC_READ, FILE_SHARE_READ | FILE_SHARE_WRITE,
-                      NULL, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL);
-
-      if (hf != INVALID_HANDLE_VALUE) {
-        if (GetFileSizeEx(hf, &size))
-          fe->size_bytes = (uint64_t)size.QuadPart;
-        CloseHandle(hf);
-      }
+      fe->size_bytes = ((uint64_t)fd.nFileSizeHigh << 32) | fd.nFileSizeLow;
 
       mpmc_push(&g_file_queue, fe);
     }
@@ -474,7 +466,7 @@ int main(int argc, char **argv) {
   arena_free(&gp_arena, (u8 **)&buf, len);
 
   // Add some extra threads to overlap I/O more aggressively
-  u8 num_threads = hw_threads * 2;
+  size_t num_threads = hw_threads * 2;
   if (num_threads < 2)
     num_threads = 2;
 
@@ -526,7 +518,9 @@ int main(int argc, char **argv) {
 
   WaitForMultipleObjects((DWORD)scan_threads, scan_tids, TRUE, INFINITE);
 
-  mpmc_producers_finished(&g_file_queue, num_threads);
+  for (size_t i = 0; i < num_threads; i++) {
+    mpmc_push(&g_file_queue, NULL);
+  }
 
   atomic_store(&g_scan_done, 1);
 
@@ -597,7 +591,7 @@ int main(int argc, char **argv) {
   double total_mb = (double)total_bytes / (1024.0 * 1024.0);
   double avg_mbps = total_mb / total_seconds;
   printf("Total: %.2f MB, Average: %.2f MB/s\n", total_mb, avg_mbps);
-  printf("  Total time  : %.2f seconds\n", total_seconds);
+  printf("  Total time  : %.2f seconds\n\n", total_seconds);
 
   return 0;
 }