nptl: extend test coverage for sched_yield

We add sched_yield() API testing to the existing thread affinity
test case because it allows us to test sched_yield() operation
in the following scenarios:

  * On a main thread.
  * On multiple threads simultaneously.
  * On every CPU the system reports simultaneously.

The ensures we exercise sched_yield() in as many scenarios as
we would exercise calls to the affinity functions.

Additionally, the test is improved by adding a semaphore to coordinate
all the threads running, so that an early starter thread won't consume
cpu resources that could be used to start the other threads.

Co-authored-by: DJ Delorie <dj@redhat.com>
Reviewed-by: Carlos O'Donell <carlos@redhat.com>

1 files changed, 3 insertions, 2 deletions

diff --git a/sysdeps/unix/sysv/linux/tst-skeleton-affinity.c b/sysdeps/unix/sysv/linux/tst-skeleton-affinity.c
index c2bb80d..cf97c3f 100644
--- a/sysdeps/unix/sysv/linux/tst-skeleton-affinity.c
+++ b/sysdeps/unix/sysv/linux/tst-skeleton-affinity.c
@@ -38,6 +38,7 @@
 #include <sched.h>
 #include <stdbool.h>
 #include <stdio.h>
+#include <support/test-driver.h>
 
 /* CPU set configuration determined.  Can be used from early_test.  */
 struct conf
@@ -253,12 +254,12 @@ do_test (void)
     if (getaffinity (sizeof (set), &set) < 0 && errno == ENOSYS)
       {
 	puts ("warning: getaffinity not supported, test cannot run");
-	return 0;
+	return EXIT_UNSUPPORTED;
       }
     if (sched_getcpu () < 0 && errno == ENOSYS)
       {
 	puts ("warning: sched_getcpu not supported, test cannot run");
-	return 0;
+	return EXIT_UNSUPPORTED;
       }
   }