Re: [LTP] [PATCH] syscalls/fsync02: restore runtime to 5m

[Cyril Hrubis <chrubis@suse.cz>]

Hi!
> Test allows up to 240 seconds for PASS result (depending if its VM or not),
> but on slower systems library now kills it after a minute. Restore
> runtime to 5 minutes.

Looking at the test itself it's a bit messed up too.

The test uses rand(); to initialize the buffer size but without
initializing the seed which is not random at all. It also uses number of
available disk blocks as a upper limit, which makes the test runtime
completely unpredictable.

I guess that it would make sense to randomize the buffer sizes but in
certain bounds to make the test more predictable and print the numbers
we are going to use too. Maybe run the test with a few different sizes
and time limits. Maybe the size of the buffers can be function of the
test runtime.

All in all I think that we should really rething what we are doing here
since the current code does not make that much sense to me.

> Signed-off-by: Jan Stancek <jstancek@redhat.com>
> ---
>  testcases/kernel/syscalls/fsync/fsync02.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/testcases/kernel/syscalls/fsync/fsync02.c b/testcases/kernel/syscalls/fsync/fsync02.c
> index e13ba89f1b63..55c7a71c1d65 100644
> --- a/testcases/kernel/syscalls/fsync/fsync02.c
> +++ b/testcases/kernel/syscalls/fsync/fsync02.c
> @@ -114,5 +114,6 @@ static struct tst_test test = {
>  	.test_all = run,
>  	.setup = setup,
>  	.cleanup = cleanup,
> -	.needs_tmpdir = 1
> +	.needs_tmpdir = 1,
> +	.max_runtime = 300,
>  };



-- 
Cyril Hrubis
chrubis@suse.cz

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