fix: tests: SRC tests catch aplay warnings and refactor

[ekurdybx]

What's changed:

• check for aplay warnings (ex. about wrong sample rate)
• check aplay and arecord exit codes (these tests do not exit on errors automatically, because we want to go through all the tests in the loop, even if previous ones have failed)
• refactor the check_soundfiles_for_glitches function
• add logs for cleaner and more readable results
• move generated audio files used for testing from logs dir to avoid "littering" in the test artifacts
• add default parameters

[marc-hb]

https://unix.stackexchange.com/questions/65803/why-is-printf-better-than-echo

[ekurdybx]

Fixed

[marc-hb]

Suggested change

	check_for_aplay_warnings()
	check_for_warnings()

[ekurdybx]

Done

[marc-hb]

$? will never be useful when you background a command. To get the status of a backgrounded command you need to wait $PID

[ekurdybx]

Oh yeah you're right, I've missed that. Fixed :)

[marc-hb]

Why not just:

aplay_output=$(aplay $SOF_ALSA_OPTS $SOF_APLAY_OPTS $2 2>&1) || errors=$((errors+1))

[ekurdybx]

Fixed

[marc-hb]

Capturing output loses "real-time" feedback. You can (and should) have both with something like this:

set -o pipefail
aplay $SOF_ALSA_OPTS $SOF_APLAY_OPTS $2 2>&1 | tee _.log || errors=$((errors+1))

check_warnings _.log

There are other ways:

https://stackoverflow.com/questions/1221833/pipe-output-and-capture-exit-status-in-bash

[ekurdybx]

Okay I've refactored the whole func a little bit again, now it prints log to the console and also captures it for further analysis

[marc-hb]

Forgot to mention this:

• Use aplay --fatal-errors to stop on underruns or errors #489

The "real" fix should be in aplay itself, not in sof-test. As in: aplay --fatal-errors --fatal-warnings. @kv2019i is it something that ALSA could implement in the (distant...) future?