[2/2] aiodio: Set timeout 30 min

Originally only ADSP075 failed (dio_sparse  -w 518192k -s 518192k -n 1000)
but just in case increase timeout for all newly rewritten tests.

Suggested-by: Cyril Hrubis <chrubis@suse.cz>
Signed-off-by: Petr Vorel <pvorel@suse.cz>
---
 testcases/kernel/io/ltp-aiodio/aiodio_append.c | 1 +
 testcases/kernel/io/ltp-aiodio/dio_append.c    | 1 +
 testcases/kernel/io/ltp-aiodio/dio_read.c      | 1 +
 testcases/kernel/io/ltp-aiodio/dio_sparse.c    | 1 +
 testcases/kernel/io/ltp-aiodio/dio_truncate.c  | 1 +
 5 files changed, 5 insertions(+)

Hi!
> Originally only ADSP075 failed (dio_sparse  -w 518192k -s 518192k -n 1000)
> but just in case increase timeout for all newly rewritten tests.

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

Hi,
only dio_sparse really needs higher timeout because we use it for a slow
stress test in a runfile. dio_read should get a lower default number of
child processes instead (8 or 16 to keep the default memory requirements
under 512MB).

The other programs finish in less than 30 seconds so if anybody wants to
use them for a custom stress test, it's their responsibility to set
LTP_TIMEOUT_MUL accordingly.

On 21. 01. 22 11:30, Petr Vorel wrote:
> Originally only ADSP075 failed (dio_sparse  -w 518192k -s 518192k -n 1000)
> but just in case increase timeout for all newly rewritten tests.
> 
> Suggested-by: Cyril Hrubis <chrubis@suse.cz>
> Signed-off-by: Petr Vorel <pvorel@suse.cz>
> ---
>  testcases/kernel/io/ltp-aiodio/aiodio_append.c | 1 +
>  testcases/kernel/io/ltp-aiodio/dio_append.c    | 1 +
>  testcases/kernel/io/ltp-aiodio/dio_read.c      | 1 +
>  testcases/kernel/io/ltp-aiodio/dio_sparse.c    | 1 +
>  testcases/kernel/io/ltp-aiodio/dio_truncate.c  | 1 +
>  5 files changed, 5 insertions(+)
> 
> diff --git a/testcases/kernel/io/ltp-aiodio/aiodio_append.c b/testcases/kernel/io/ltp-aiodio/aiodio_append.c
> index 46cc74ee4e..3c4add66ec 100644
> --- a/testcases/kernel/io/ltp-aiodio/aiodio_append.c
> +++ b/testcases/kernel/io/ltp-aiodio/aiodio_append.c
> @@ -183,6 +183,7 @@ static struct tst_test test = {
>  		"tmpfs",
>  		NULL
>  	},
> +	.timeout = 1800,
>  };
>  #else
>  TST_TEST_TCONF("test requires libaio and its development packages");
> diff --git a/testcases/kernel/io/ltp-aiodio/dio_append.c b/testcases/kernel/io/ltp-aiodio/dio_append.c
> index c099793f6c..dd044325bd 100644
> --- a/testcases/kernel/io/ltp-aiodio/dio_append.c
> +++ b/testcases/kernel/io/ltp-aiodio/dio_append.c
> @@ -97,4 +97,5 @@ static struct tst_test test = {
>  		"tmpfs",
>  		NULL
>  	},
> +	.timeout = 1800,
>  };
> diff --git a/testcases/kernel/io/ltp-aiodio/dio_read.c b/testcases/kernel/io/ltp-aiodio/dio_read.c
> index 67a28147fd..f48414ec01 100644
> --- a/testcases/kernel/io/ltp-aiodio/dio_read.c
> +++ b/testcases/kernel/io/ltp-aiodio/dio_read.c
> @@ -181,4 +181,5 @@ static struct tst_test test = {
>  		"tmpfs",
>  		NULL
>  	},
> +	.timeout = 1800,
>  };
> diff --git a/testcases/kernel/io/ltp-aiodio/dio_sparse.c b/testcases/kernel/io/ltp-aiodio/dio_sparse.c
> index 39fc895d65..0039daa8d1 100644
> --- a/testcases/kernel/io/ltp-aiodio/dio_sparse.c
> +++ b/testcases/kernel/io/ltp-aiodio/dio_sparse.c
> @@ -135,4 +135,5 @@ static struct tst_test test = {
>  		"tmpfs",
>  		NULL
>  	},
> +	.timeout = 1800,
>  };
> diff --git a/testcases/kernel/io/ltp-aiodio/dio_truncate.c b/testcases/kernel/io/ltp-aiodio/dio_truncate.c
> index 1fbf83de06..80445c48a3 100644
> --- a/testcases/kernel/io/ltp-aiodio/dio_truncate.c
> +++ b/testcases/kernel/io/ltp-aiodio/dio_truncate.c
> @@ -169,4 +169,5 @@ static struct tst_test test = {
>  		"tmpfs",
>  		NULL
>  	},
> +	.timeout = 1800,
>  };

> Hi,
> only dio_sparse really needs higher timeout because we use it for a slow
> stress test in a runfile. dio_read should get a lower default number of
> child processes instead (8 or 16 to keep the default memory requirements
> under 512MB).
Thanks!

I also noticed dio_read, but as you pointed out it's already fixed by
f3caabe44c ("Reduce memory footprint of the dio_read test")

> The other programs finish in less than 30 seconds so if anybody wants to
> use them for a custom stress test, it's their responsibility to set
> LTP_TIMEOUT_MUL accordingly.
+1

Kind regards,
Petr

On 21. 01. 22 14:33, Petr Vorel wrote:
>> Hi,
>> only dio_sparse really needs higher timeout because we use it for a slow
>> stress test in a runfile. dio_read should get a lower default number of
>> child processes instead (8 or 16 to keep the default memory requirements
>> under 512MB).
> Thanks!
> 
> I also noticed dio_read, but as you pointed out it's already fixed by
> f3caabe44c ("Reduce memory footprint of the dio_read test")

Runfile usage is fixed but running the program manually with default
setting would still result in extreme memory use. We should make the
defaults more sensible as well.

I agree with Martin that some tests in the aiodio testing suite should 
handle default value a bit better. To reduce the amount of memory/time 
required by manual execution with default values could be a task for the 
next future.

On 1/21/22 14:38, Martin Doucha wrote:
> On 21. 01. 22 14:33, Petr Vorel wrote:
>>> Hi,
>>> only dio_sparse really needs higher timeout because we use it for a slow
>>> stress test in a runfile. dio_read should get a lower default number of
>>> child processes instead (8 or 16 to keep the default memory requirements
>>> under 512MB).
>> Thanks!
>>
>> I also noticed dio_read, but as you pointed out it's already fixed by
>> f3caabe44c ("Reduce memory footprint of the dio_read test")
> Runfile usage is fixed but running the program manually with default
> setting would still result in extreme memory use. We should make the
> defaults more sensible as well.
>

> On 21. 01. 22 14:33, Petr Vorel wrote:
> >> Hi,
> >> only dio_sparse really needs higher timeout because we use it for a slow
> >> stress test in a runfile. dio_read should get a lower default number of
> >> child processes instead (8 or 16 to keep the default memory requirements
> >> under 512MB).
> > Thanks!

> > I also noticed dio_read, but as you pointed out it's already fixed by
> > f3caabe44c ("Reduce memory footprint of the dio_read test")

> Runfile usage is fixed but running the program manually with default
> setting would still result in extreme memory use. We should make the
> defaults more sensible as well.
+1. So do we want in the end to increase dio_read just for now before release?
That'd make sense to me.

Kind regards,
Petr