Audio: Add YAML configuration files for AudioPlayback and AudioRecord test cases by tmoida · Pull Request #262 · qualcomm-linux/qcom-linux-testkit

tmoida · 2026-01-28T06:44:51Z

Created 10 AudioPlayback configuration files (Config1-Config10)
Created 10 AudioRecord configuration files (Config1-Config10)
Organized files into AudioPlayback_Configs and AudioRecord_Configs directories
Updated README files with new directory structure and usage instructions
Added documentation for YAML configuration files and LAVA test commands

smuppand

Fix Config mapping mismatch (README vs YAML vs clip names).
Standardize names:
- Use Config01 everywhere (metadata, RES_SUFFIX, result files, docs).
Remove || true from send-to-lava.sh step
Make TIMEOUT types consistent (strings with units or numeric everywhere).
Shorten metadata.description and move long text to README.
Convert README config listing into a proper markdown table (and correct ordering).

tmoida · 2026-01-29T08:23:16Z

Hi @smuppand, As per your suggestions, I have fixed all the issues in the PR.
Here's what I have addressed:

1. Standardized all config naming to use leading zeros (Config01 format) in all 20 YAML files.

2. Updated metadata.name, RES_SUFFIX, CLIP_NAMES, and CONFIG_NAMES to use consistent format with leading zeros.

3. Removed all instances of || true from send-to-lava.sh calls to allow proper error propagation.

4. For TIMEOUT types, I kept the different formats for each test type:

- AudioPlayback: string format with units ("10s")
- AudioRecord: numeric format (0)

AudioRecord doesn't need a timeout since recording automatically stops after the RECORD_SECONDS duration. Setting timeout to 0 prevents premature termination. AudioPlayback needs a timeout to prevent hanging if playback gets stuck.
Let me know if you want me to standardize both to use the same format instead.

5. Shortened all metadata descriptions to single concise lines in all YAML files.

6. Converted config listings in both README files to proper markdown tables for better readability.

Please review the latest changes and let me know if any further modifications are required.

smuppand

Keep YAMLs in the same directory as run.sh, named:
AudioPlayback_Config01.yaml … AudioPlayback_Config10.yaml
AudioRecord_Config01.yaml … AudioRecord_Config10.yaml

smuppand · 2026-02-02T16:20:28Z

Runner/suites/Multimedia/Audio/AudioPlayback/AudioPlayback_Config01.yaml

@@ -1,7 +1,7 @@
 metadata:
-  name: audio-playback
+  name: audio-playback-config01


Backward compatibility break you moved AudioPlayback.yaml

AudioPlayback.yaml → AudioPlayback_Configs/AudioPlayback_Config01.yaml

This can break:
existing LAVA jobs referencing the old YAML path
docs / CI scripts expecting .../AudioPlayback
Request: keep a compatibility YAML at the old location:
Runner/suites/Multimedia/Audio/AudioPlayback/AudioPlayback.yaml

Those can simply be copies of Config01 with legacy metadata name (or same, doesn’t matter much), but the file path must exist.

Fixed. I've kept AudioPlayback.yaml at its original location to maintain backward compatibility with existing LAVA jobs and CI scripts.

smuppand · 2026-02-02T16:21:42Z

Runner/suites/Multimedia/Audio/AudioRecord/AudioRecord_Config01.yaml

@@ -1,7 +1,7 @@
 metadata:
-  name: audio-record
+  name: audio-record-config01


Backward compatibility break: moved AudioRecord.yaml
They renamed:

AudioRecord.yaml → AudioRecord_Configs/AudioRecord_Config01.yaml
This can break:
existing LAVA jobs referencing the old YAML path
docs / CI scripts expecting ...AudioPlayback.yaml
Request: keep a compatibility YAML at the old location:

Runner/suites/Multimedia/Audio/AudioRecord/AudioRecord.yaml
Those can simply be copies of Config01 with legacy metadata name (or same, doesn’t matter much), but the file path must exist.

Fixed. I've kept AudioPlayback.yaml at its original location to maintain backward compatibility with existing LAVA jobs and CI scripts.

smuppand · 2026-02-02T16:23:18Z

Runner/suites/Multimedia/Audio/AudioRecord/AudioRecord_Configs/AudioRecord_Config01.yaml

    - cd Runner/suites/Multimedia/Audio/AudioRecord/
-    - ./run.sh --backend "${AUDIO_BACKEND}" --source "${SOURCE_CHOICE}" --config-name "${CONFIG_NAMES}" --config-filter "${CONFIG_FILTER}" --record-seconds "${RECORD_SECONDS}" --loops "${LOOPS}" --timeout "${TIMEOUT}" --strict "${STRICT}" --res-suffix "${RES_SUFFIX}" || true
-    - $REPO_PATH/Runner/utils/send-to-lava.sh AudioRecord${RES_SUFFIX:+_${RES_SUFFIX}}.res || true
+    - ./run.sh --backend "${AUDIO_BACKEND}" --source "${SOURCE_CHOICE}" --config-name "${CONFIG_NAMES}" --config-filter "${CONFIG_FILTER}" --record-seconds "${RECORD_SECONDS}" --loops "${LOOPS}" --timeout "${TIMEOUT}" --strict "${STRICT}" --res-suffix "${RES_SUFFIX}"


risk of missing results if run.sh returns non-zero
keep ./run.sh ... || true

Fixed. I've restored the || true for run.sh calls in all YAML files while keeping send-to-lava.sh without it.

smuppand · 2026-02-02T16:25:46Z

Runner/suites/Multimedia/Audio/AudioPlayback/Read_me.md

+```  
+| Config   | Config Name       | Sample Rate | Bit Rate | Channels |
+|----------|-------------------|-------------|----------|----------|
+| Config01 | playback_config1  | 8 KHz       | 8-bit    | 1ch      |


Config01 → playback_config1 (no zero)
But the YAML uses:
CLIP_NAMES: playback_config01
Same for record side.

Fixed. I've updated all YAML files to use config names without leading zeros (playback_config1, record_config1) to match the source code in audio_common.sh. File names still use the Config01 format (with leading zero) for better sorting and organization.

smuppand · 2026-02-02T16:28:22Z

Runner/utils/audio_common.sh

  case "$config_name" in
+    record_config0*)
+      # Extract number and remove leading zero for internal processing
+      config_num="$(printf '%s' "$config_name" | sed -n 's/^record_config0*\([0-9][0-9]*\)$/\1/p')"


sed returns empty (unexpected input), normalized_name becomes record_config which won’t match any case labels and could change behavior subtly.

config_num="$(...)"
if [ -n "$config_num" ]; then
normalized_name="record_config$config_num"
fi

Same pattern applies anywhere they normalize playback_config0*.

Fixed. I've added validation to the normalization code in audio_common.sh to handle cases where sed extraction fails. The code now checks if config_num is non-empty before creating the normalized name, preventing silent failures with unexpected input formats.

… test cases - Created 10 AudioPlayback configuration files (Config01-Config10) with standardized naming - Created 10 AudioRecord configuration files (Config01-Config10) with standardized naming - Reorganized YAML files from subdirectories to parent directories - Added legacy files (AudioPlayback.yaml and AudioRecord.yaml) for backward compatibility - Updated README files with proper markdown tables and usage instructions - Removed error masking (|| true) from send-to-lava.sh calls for proper error propagation - Shortened metadata descriptions for better readability - Updated audio_common.sh to handle standardized naming conventions - Added documentation for YAML configuration files and LAVA test commands Signed-off-by: Teja Swaroop Moida <tmoida@qti.qualcomm.com>

tmoida · 2026-02-04T13:59:05Z

Hi @smuppand,
As per your suggestions, I have fixed the issues mentioned above.

Moved all YAML files to the same directory as run.sh

Kept AudioPlayback.yaml and AudioRecord.yaml at their original locations for backward compatibility

Restored || true for run.sh calls to ensure results are always uploaded

Fixed config name mismatch by using playback_config1/record_config1 format (without leading zeros) in YAML parameters

Added validation to audio_common.sh to handle empty sed results

Updated documentation to clarify the naming conventions

Please review the latest changes and let me know if any further modifications are required.

smuppand · 2026-02-04T16:30:16Z

Runner/utils/audio_common.sh

+# Input: record_config1 or record_config01
 # Output: record_8KHz_1ch
 # Returns: 0=success, 1=invalid config
 generate_record_testcase_name() {


generate_record_testcase_name() sets normalized_name="record_config$config_num" even if config_num is empty (it doesn’t guard like the other function). That’s minor but could create "record_config" and fall into invalid case anyway. Not a big deal, but if you want it fully consistent, add the same guard.

This can be addressed in a separate PR. For now, I am merging the change. Please take note.

smuppand

LGTM

smuppand requested changes Jan 28, 2026

View reviewed changes

tmoida force-pushed the main branch from 75bc042 to a25ccf0 Compare January 29, 2026 08:02

smuppand requested changes Feb 2, 2026

View reviewed changes

tmoida force-pushed the main branch from a25ccf0 to 78fd4c8 Compare February 4, 2026 13:40

smuppand requested changes Feb 4, 2026

View reviewed changes

smuppand approved these changes Feb 5, 2026

View reviewed changes

smuppand merged commit 8322bc9 into qualcomm-linux:main Feb 5, 2026
8 checks passed

Conversation

tmoida commented Jan 28, 2026

Uh oh!

smuppand left a comment

Choose a reason for hiding this comment

Uh oh!

tmoida commented Jan 29, 2026

Uh oh!

smuppand left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tmoida Feb 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tmoida commented Feb 4, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

smuppand left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Comments

tmoida Feb 4, 2026 •

edited

Loading