2002-11-25 Robert Collins <rbtcollins@hotmail.com>

* readme: Document running portions of the test suite (Thanks Egor!).
        * winsup.api/pthread/mainthreadexits.c: New file, derived from
        Thomas Pfaff's test cases.
        * winsup.api/pthread/threadidafterfork.c: Ditto.

winsup/testsuite/winsup.api/pthread/mainthreadexits.c

@@ -0,0 +1,47 @@
+#include <stdio.h>
+#include <unistd.h>
+#include <pthread.h>
+
+static void * Thread (void *);
+
+static pthread_t main_thread;
+static pthread_t secondThread;
+static int result = 2;
+
+int main(void)
+{
+  main_thread = pthread_self ();
+
+  if (pthread_create (&secondThread, NULL, Thread, NULL))
+      exit (1);
+  sleep (5);
+  pthread_exit (&result);
+  /* If pthread_exit doesm't (which would be a bug) then we do */
+  return 1;
+}
+
+static void * Thread (void *not_used)
+{
+  void *myresult;
+  /* We should be able to join this */
+  if (pthread_join (main_thread, &myresult))
+      exit (1);
+
+  if (*(int *)myresult != 2)
+      exit (1);
+
+  exit (0);
+}
+/*
+This valid code doesn't work at all. The mainthread object in MTinterface
+is not properly initialized, the cancel_event is NULL and the win32_obj_id
+is NULL because myself->hProcess is NULL when MTinterface is initialized
+(and i don't think that a process handle can be used as thread handle).
+Even if the handles would be valid the pthread_join call would try to
+delete a thread object that is created static which would result in a
+corrupted heap.
+
+Concept test Contributed by Thomas Pfaff <tpfaff@gmx.net>
+Scriptable test by Robert Collins <rbtcollins@hotmail.com>
+
+*/