On 2020-02-14 17:11, Sebastian Huber wrote:
On 14/02/2020 04:41, chr...@rtems.org wrote:
From: Chris Johns <chr...@rtems.org>
Update #3870
---
testsuites/libtests/newlib01/init.c | 16 +++++++++++++++-
1 file changed, 15 insertions(+), 1 deletion(-)
diff --git a/testsuites/libtests/newlib01/init.c
b/testsuites/libtests/newlib01/init.c
index 74e648799e..1bff052363 100644
--- a/testsuites/libtests/newlib01/init.c
+++ b/testsuites/libtests/newlib01/init.c
@@ -23,6 +23,8 @@
#include <rtems/imfs.h>
#include <rtems/libcsupport.h>
+#include <rtems/bspIo.h>
+
Not needed?
#include "tmacros.h"
const char rtems_test_name[] = "NEWLIB 1";
@@ -63,12 +65,24 @@ static void wait(void)
static void worker_task(rtems_task_argument arg)
{
test_context *ctx = &test_instance;
+ struct _reent *reent = __getreent();
I think you should include <sys/reent.h> and use
struct _reent *reent = _REENT;
Thanks. I was looking at the path the code took with the debugger but
now you point this out it is how newlib accesses the pointer.
+ FILE *output;
char buf[1] = { 'x' };
size_t n;
- stdout = fopen(&file_path[0], "r+");
+ rtems_test_assert(reent->__sdidinit == 0);
+
+ output = stdout = fopen(&file_path[0], "r+");
rtems_test_assert(stdout != NULL);
+ /*
+ * Check newlib's __sinit does not touch our assigned file pointer.
+ */
+ rtems_test_assert(reent->__sdidinit == 0);
+ fflush(stdout);
fflush() should return 0 here.
I will add this and send v2.
Thanks for the review.
Chris
+ rtems_test_assert(reent->__sdidinit != 0);
+ rtems_test_assert(stdout == output);
+
n = fwrite(&buf[0], sizeof(buf), 1, stdout);
rtems_test_assert(n == 1);
_______________________________________________
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel
