multiple-pipeline: add delay before ps_checks

[fredoh9]

Sometimes last aplay or arecord process might delay to start. This causes
the test case to fail due to number of PIDs mistmatch. This is not 100%
reproducible but slower platform has more chance to fail.

Signed-off-by: Fred Oh fred.oh@linux.intel.com

fixes: #672

[marc-hb]

I think we unfortunately need this because of my tentative explanation in #672. If you agree with it then please summarize it in the comment here and maybe add a link to it.

[fredoh9]

@marc-hb left a detailed explanation about the problem. We will find better management of background processes... later.
#672 (comment)

[aiChaoSONG]

@keqiaozhang @fredoh9 @marc-hb As I observed from Keqiao's demo, aplay/arecord process from last iteration may not be fully killed at the current iteration, so are you sure it is due to delay to start, rather than not fully killed?

[marc-hb]

Which demo? "Not fully killed" sounds like another issue.

[marc-hb]

BTW I would like to close this PR, superseded by PR #676. So can we keep the discussion in bug #672?

[aiChaoSONG]

Which demo? "Not fully killed" sounds like another issue.

When Keqiao found the issue days before, he ran this case once, and I checked the log.

Without the sleep here, for example, in iteration 0, four aplay/arecord processes are started and killed, PIDs = [0 1 2 3], In the second iteration, we start four aplay/arecord again, PIDs=[4 5 6 7], but in the first ps check of the second iteration, six aplay/arecord processes are detected, PIDs=[0 1 4 5 6 7], two are from iteration 0, in the second ps check of the second iteration, only four pipelines are detected(PIDs=[ 4 5 6 7]), and test case thinks it started 6 process, but only 4 were alive at the end, so it failed.

That's why I said not fully killed, resources are released (we can run aplay on the same pcm), but process not fully cleaned. Hope I have made me clear to you.

[fredoh9]

#676 is merged

[marc-hb]

Thanks @aiChaoSONG , that really sounds like a different issue to me. Let's keep an eye on it and if it happens again let's please discuss it in a better place than this abandoned PR :-)

[fredoh9]

@marc-hb @keqiaozhang @aiChaoSONG Yes that is different issue. If you have seen that problem, please open a new issue. I remember the same error in long time ago. I think that will be not so easy to reproduce the problem.