[LTP] [PATCH] openposix: sem_timedwait: replace sleep with tst_process_state_wait3

[LTP] [PATCH] openposix: sem_timedwait: replace sleep with tst_process_state_wait3

[Jan Stancek]

This makes the test faster and also avoids a small race - when the
time is very close to starting new second, this makes the test use
timeout that's effectively just slightly over one second, because
nanosecond portion is always set to 0. And previously child would
sleep roughly one second.

Signed-off-by: Jan Stancek <jstancek@redhat.com>
---
 .../conformance/interfaces/sem_timedwait/2-1.c                | 4 +++-
 .../conformance/interfaces/sem_timedwait/9-1.c                | 4 +++-
 2 files changed, 6 insertions(+), 2 deletions(-)

diff --git a/testcases/open_posix_testsuite/conformance/interfaces/sem_timedwait/2-1.c b/testcases/open_posix_testsuite/conformance/interfaces/sem_timedwait/2-1.c
index d9012cc5ed1a..61603cee56b1 100644
--- a/testcases/open_posix_testsuite/conformance/interfaces/sem_timedwait/2-1.c
+++ b/testcases/open_posix_testsuite/conformance/interfaces/sem_timedwait/2-1.c
@@ -22,6 +22,7 @@
 #include <time.h>
 #include <sys/mman.h>
 #include "posixtest.h"
+#include "proc.h"
 
 #define TEST "2-1"
 #define SHM_NAME "/posixtest_2-1"
@@ -79,7 +80,8 @@ int main(void)
 	} else if (pid > 0)	// parent to unlock semaphore
 	{
 		int i;
-		sleep(1);
+
+		tst_process_state_wait3(pid, 'S', 1);
 		if (sem_post(mysemp) == -1) {
 			perror(ERROR_PREFIX "sem_post");
 			return PTS_FAIL;
diff --git a/testcases/open_posix_testsuite/conformance/interfaces/sem_timedwait/9-1.c b/testcases/open_posix_testsuite/conformance/interfaces/sem_timedwait/9-1.c
index f9175839dab3..ee7ad7aefe26 100644
--- a/testcases/open_posix_testsuite/conformance/interfaces/sem_timedwait/9-1.c
+++ b/testcases/open_posix_testsuite/conformance/interfaces/sem_timedwait/9-1.c
@@ -21,6 +21,7 @@
 #include <sys/wait.h>
 #include <time.h>
 #include "posixtest.h"
+#include "proc.h"
 
 #define TEST "9-1"
 #define FUNCTION "sem_timedwait"
@@ -79,7 +80,8 @@ int main(void)
 
 	} else {		// parent to send a signal to child
 		int i;
-		sleep(1);
+
+		tst_process_state_wait3(pid, 'S', 1);
 		(void)kill(pid, SIGABRT);	// send signal to child
 		if (wait(&i) == -1) {
 			perror("Error waiting for child to exit\n");
-- 
2.43.0


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH] openposix: sem_timedwait: replace sleep with tst_process_state_wait3

[Petr Vorel]

Hi Jan,

> This makes the test faster and also avoids a small race - when the
> time is very close to starting new second, this makes the test use
> timeout that's effectively just slightly over one second, because
> nanosecond portion is always set to 0. And previously child would
> sleep roughly one second.

LGTM.
Reviewed-by: Petr Vorel <pvorel@suse.cz>

I also wonder if other tests with sleep(1) are relevant. e.g.

conformance/interfaces/pthread_rwlock_timedrdlock/6-2.c
conformance/interfaces/pthread_rwlock_timedwrlock/6-2.c

(There are more sleep() uses, but I suppose others aren't relevant.)

Also, some tests (e.g conformance/interfaces/clock_nanosleep/1-3.c updated few
years ago) did not updated sleep() in docs comment.

Kind regards,
Petr

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH] openposix: sem_timedwait: replace sleep with tst_process_state_wait3

[Cyril Hrubis]

Hi!
Reviewed-by: Cyril Hrubis <chrubis@suse.cz>

-- 
Cyril Hrubis
chrubis@suse.cz

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH] openposix: sem_timedwait: replace sleep with tst_process_state_wait3

[Jan Stancek]

On Thu, Feb 27, 2025 at 2:39 PM Petr Vorel <pvorel@suse.cz> wrote:
>
> Hi Jan,
>
> > This makes the test faster and also avoids a small race - when the
> > time is very close to starting new second, this makes the test use
> > timeout that's effectively just slightly over one second, because
> > nanosecond portion is always set to 0. And previously child would
> > sleep roughly one second.
>
> LGTM.
> Reviewed-by: Petr Vorel <pvorel@suse.cz>

Thanks, pushed.

>
> I also wonder if other tests with sleep(1) are relevant. e.g.
>
> conformance/interfaces/pthread_rwlock_timedrdlock/6-2.c
> conformance/interfaces/pthread_rwlock_timedwrlock/6-2.c
>
> (There are more sleep() uses, but I suppose others aren't relevant.)

I suspect there are still many testcases where sleep() could be replaced.

>
> Also, some tests (e.g conformance/interfaces/clock_nanosleep/1-3.c updated few
> years ago) did not updated sleep() in docs comment.
>
> Kind regards,
> Petr
>


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp