[nvptx, libgomp] Fix memleak in GOMP_OFFLOAD_fini_device

I wrote a test-case:
...
int
main (void)
{
  for (unsigned i = 0; i < 128; ++i)
    {
      acc_init (acc_device_nvidia);
      acc_shutdown (acc_device_nvidia);
    }

  return 0;
}
...
and ran it under valgrind.  The only leak location reported with a frequency
of 128, was the allocation of ptx_devices in nvptx_init.

Fix this by freeing ptx_devices in GOMP_OFFLOAD_fini_device, once
instantiated_devices drops to 0.

2019-01-24  Tom de Vries  <tdevries@suse.de>

	* plugin/plugin-nvptx.c (GOMP_OFFLOAD_fini_device): Free ptx_devices
	once instantiated_devices drops to 0.

From-SVN: r268237

libgomp/ChangeLog

+2019-01-24  Tom de Vries  <tdevries@suse.de>
+
+	* plugin/plugin-nvptx.c (GOMP_OFFLOAD_fini_device): Free ptx_devices
+	once instantiated_devices drops to 0.
+
 2019-01-23  Tom de Vries  <tdevries@suse.de>
 
 	PR target/PR88946

libgomp/plugin/plugin-nvptx.c

@@ -1936,6 +1936,12 @@ GOMP_OFFLOAD_fini_device (int n)
       instantiated_devices--;
     }
 
+  if (instantiated_devices == 0)
+    {
+      free (ptx_devices);
+      ptx_devices = NULL;
+    }
+
   pthread_mutex_unlock (&ptx_dev_lock);
   return true;
 }