Add test script to validate USB UAC#266
Conversation
achauras-qcom
force-pushed
the
feature/usb-uac
branch
2 times, most recently
from
6877bdd to
9f1040f
Compare
Srikanth Muppandam (smuppand)
left a comment
Srikanth Muppandam (smuppand)
left a comment
There was a problem hiding this comment.
Commit ordering / path inconsistency (important for clean history)
Create files directly in Runner/suites/Kernel/Baseport/USB/usb_uac/ from patch 1 onward, and drop patch 4/4 entirely.
Alternative:
Move patch 4/4 earlier (as 1/4), then add script/README/YAML into the final location.
achauras-qcom
force-pushed
the
feature/usb-uac
branch
from
9f1040f to
402263a
Compare
| audio_device_count=$( | ||
| for f in /sys/bus/usb/devices/*/bInterfaceClass; do | ||
| [ -r "$f" ] || continue | ||
| if grep -qx '01' "$f"; then |
There was a problem hiding this comment.
Validate more than bInterfaceClass == 01
| # Count uniques devices with bInterfaceClass = 01 (UAC) under /sys/bus/usb/devices | ||
| audio_device_count=0 | ||
| log_info "=== USB Audio device Detection ===" | ||
| audio_device_count=$( |
There was a problem hiding this comment.
Add ALSA/device-node checks (/dev/snd, /proc/asound/cards, aplay -l, arecord -l where available)
Print detected UAC device details (VID:PID, product, device path)
| log_info "=== Test Initialization ===" | ||
|
|
||
| # Check if grep is installed, else skip test | ||
| deps_list="grep sed sort wc" |
There was a problem hiding this comment.
Add missing dependency tr
| done 2>/dev/null | sed 's/:.*$//' | sort -u | wc -l | tr -d '[:space:]' | ||
| ) | ||
|
|
||
| log_info "Number of USB audio devices found: $audio_device_count" |
There was a problem hiding this comment.
Distinguish playback-only vs capture-only devices and also log bound driver info from sysfs
|
|
||
| - Connect USB Audio peripheral(s) to USB port(s) on DUT. | ||
| - Only applicable for USB ports that support Host Mode functionality. | ||
| - USB Audio peripherals examples: USB headset, microphone, sound card, etc. |
There was a problem hiding this comment.
Improve README to clarify exact validation scope
| echo "$TESTNAME PASS" > "$RES_FILE" | ||
| exit 0 | ||
| else | ||
| log_fail "$TESTNAME : Test Failed - No 'USB Audio Device' found" |
There was a problem hiding this comment.
Consider whether “no UAC device attached” should be FAIL or SKIP
achauras-qcom
force-pushed
the
feature/usb-uac
branch
from
402263a to
63b743c
Compare
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
achauras-qcom
force-pushed
the
feature/usb-uac
branch
from
63b743c to
2431fc7
Compare
|
Output for UAC device connected: DEVICE VID:PID PRODUCT1-1 04e8:a05e Samsung USB C Earphones [INFO] ALSA cards (/proc/asound/cards): |
Srikanth Muppandam (smuppand)
left a comment
Srikanth Muppandam (smuppand)
left a comment
There was a problem hiding this comment.
squash it to one commit.
|
|
||
| # Default result file (works even before functestlib is available) | ||
| # shellcheck disable=SC2034 | ||
| RES_FILE="$SCRIPT_DIR/${TESTNAME}.res" |
There was a problem hiding this comment.
current default is ./${TESTNAME}.res, so early SKIP paths before cd "$test_path" can write the result file into an unexpected working directory.
There was a problem hiding this comment.
Skip paths before cd "$test_path" could reliably write result file in "$SCRIPT_DIR", please correct me if my understanding is wrong. Does default result file need to be changed to "./${TESTNAME}.res" instead of "$SCRIPT_DIR/${TESTNAME}.res"?
There was a problem hiding this comment.
Skip paths before cd "$test_path" could reliably write result file in "$SCRIPT_DIR", please correct me if my understanding is wrong. Does default result file need to be changed to "./${TESTNAME}.res" instead of "$SCRIPT_DIR/${TESTNAME}.res"?
Yes, your understanding is correct.
Using:
RES_FILE="$SCRIPT_DIR/${TESTNAME}.res"
is better for the early SKIP paths before cd "$test_path" because the script may be launched from any working directory. If we used:
RES_FILE="./${TESTNAME}.res"
then early failures such as missing init_env could write the result into the caller's current directory instead of the testcase directory.
There was a problem hiding this comment.
Using RES_FILE="$SCRIPT_DIR/${TESTNAME}.res" for early skip paths in the script
| sys="/sys/bus/usb/devices/$dev" | ||
| vid=$([ -r "$sys/idVendor" ] && tr -d '[:space:]' < "$sys/idVendor" || echo -) | ||
| pid=$([ -r "$sys/idProduct" ] && tr -d '[:space:]' < "$sys/idProduct" || echo -) | ||
| if [ -r "$sys/product" ]; then |
There was a problem hiding this comment.
when a UAC interface enumerates but ALSA integration is incomplete, we still lose useful debug context in CI logs. Add a DRIVER column and populate it from the relevant interface driver symlink in sysfs.
DEVICE VID:PID DRIVER PRODUCT
1-1 04e8:a05e snd-usb-audio Samsung USB C Earphones
There was a problem hiding this comment.
Added driver info as well for each UAC device.
| missing_nodes=1 | ||
| fi | ||
| if [ "$has_pcm_p" -ne 1 ] && [ "$has_pcm_c" -ne 1 ]; then | ||
| log_fail "Missing ALSA PCM device(s) for card $c (no playback or capture node found)" |
There was a problem hiding this comment.
tighten PASS to require all detected UAC-backed ALSA cards to have playback/capture nodes, or
at least log an explicit ratio such as usable ALSA USB cards: X/Y so partial failures are visible.
There was a problem hiding this comment.
Added missing nodes info for each UAC device as part of debug logs and also logging explicit ratio of usable ALSA USB cards.
The shell script verifies the enumeration of USB Audio Class devices. Signed-off-by: Aanchal Chaurasia <achauras@qti.qualcomm.com>
achauras-qcom
force-pushed
the
feature/usb-uac
branch
from
2431fc7 to
eaf38ac
Compare
|
@ualcomm/qualcomm-linux-testing.triage This pull request has been marked as stale due to 30 days of inactivity and will automatically close after an additional 5 days. |
|
Srikanth Muppandam (@smuppand) May you please review the latest changes? |
| # Check if dependencies are installed, else skip test | ||
| deps_list="grep sed sort wc tr readlink" | ||
| if ! check_dependencies "$deps_list"; then | ||
| log_skip "$TESTNAME SKIP - missing dependencies: $deps_list" |
There was a problem hiding this comment.
deps_list does not include every external command used by the script. For example, head is used later when extracting vidpid, driver_info, and product_info, but it is not part of the dependency check.
| deps_list="grep sed sort wc tr readlink" | ||
| if ! check_dependencies "$deps_list"; then | ||
| log_skip "$TESTNAME SKIP - missing dependencies: $deps_list" | ||
| echo "$TESTNAME SKIP" >"$RES_FILE" |
There was a problem hiding this comment.
check_dependencies from functestlib.sh exits by default when dependencies are missing, so the if ! check_dependencies "$deps_list"; then ... block will normally not run.
Recommended fix: either call check_dependencies "$deps_list" directly, or set CHECK_DEPS_NO_EXIT=1 before the call if this script wants to handle the failure itself.
| # Identify ALSA cards that correspond to USB | ||
| usb_alsa_card_nums="$(sed -n 's/^[[:space:]]*\([0-9][0-9]*\)[[:space:]]\{1,\}\[[^]]*\]:[[:space:]]\{1,\}USB.*/\1/p' /proc/asound/cards | sort -u)" | ||
| usb_alsa_card_count="$(printf "%s\n" "$usb_alsa_card_nums" | sed '/^$/d' | wc -l | tr -d '[:space:]')" | ||
| log_info "Number of ALSA USB sound cards: $usb_alsa_card_count" |
There was a problem hiding this comment.
USB ALSA card discovery depends on parsing /proc/asound/cards with a regex that only accepts card descriptions beginning with USB.
Recommended fix: discover candidate cards from /sys/class/sound/card*/device first, resolve each card’s parent USB interface/device, and match that against the detected UAC device list; use /proc/asound/cards only for logging.
| card_map="" | ||
| while IFS= read -r c; do | ||
| [ -n "$c" ] || continue | ||
| link="$(readlink "/sys/class/sound/card${c}/device" 2>/dev/null || true)" |
There was a problem hiding this comment.
The card-to-USB mapping uses readlink without -f, then extracts only the basename. if the symlink target shape differs across kernels or ends in an intermediate sound object rather than the USB interface name, parent="${base%%:*}" can produce the wrong USB device id and mark a valid card as unmapped.
Recommended fix: use readlink -f "/sys/class/sound/card${c}/device" and walk upward until a directory with idVendor/idProduct or a USB interface pattern is found, then map that stable parent.
| # Look up device details for debug messages | ||
| vidpid="$(printf "%b" "$dev_info_db" | sed -n "s/^${dev}|\\([^|]*\\)|.*/\\1/p" | head -n1)" | ||
| driver_info="$(printf "%b" "$dev_info_db" | sed -n "s/^${dev}|[^|]*|\\([^|]*\\)|.*/\\1/p" | head -n1)" | ||
| product_info="$(printf "%b" "$dev_info_db" | sed -n "s/^${dev}|[^|]*|[^|]*|\\(.*\\)$/\\1/p" | head -n1)" |
There was a problem hiding this comment.
The script interpolates raw USB device ids into sed regexes when looking up dev_info_db. USB device ids can contain dots, which are regex metacharacters. A device such as 1-2.3 can match unintended rows if similarly named devices exist, producing wrong VID/PID/product logs and potentially confusing failure triage.
Recommended fix: avoid regex interpolation for structured data; use awk -F'|' -v dev="$dev" '$1 == dev {print $2; exit}' or another literal-field comparison.
| # Cards mapped to this USB device | ||
| cards_for_dev="$(printf "%b" "$card_map" | sed -n "s/^${dev}|\\([0-9][0-9]*\\)$/\\1/p" | tr '\n' ' ' | sed 's/[[:space:]]*$//')" | ||
|
|
||
| if [ -z "$cards_for_dev" ]; then |
There was a problem hiding this comment.
The same raw-device-id regex issue is used to derive cards_for_dev from card_map. This is more than a logging issue; a wrong match here can map the wrong ALSA card to a UAC device or fail to map a valid one.
Recommended fix: replace the sed lookup with literal field matching, for example awk -F'|' -v dev="$dev" '$1 == dev {print $2}', then join results.
| # Check for PCM devices (playback/capture) | ||
| pcm_found=0 | ||
|
|
||
| for pcm in "$card_path"/pcmC"${c}"D*p "$card_path"/pcmC"${c}"D*c; do |
There was a problem hiding this comment.
PCM node validation manually scans "$card_path"/pcmC"${c}"D*p and *c. This assumes sysfs PCM naming/layout and duplicates ALSA PCM enumeration logic that could be checked more directly through /proc/asound/pcm or aplay -l/arecord -l.
Recommended fix: for the card being validated, confirm playback/capture PCMs from /proc/asound/pcm or ALSA tooling, then check matching /dev/snd/pcmC* nodes only as the final device-node check.
| ------------------------------------------------------------------------------- | ||
| <dev> <vid:pid> <snd-usb-audio> <product> | ||
| ``` | ||
| The test PASS requires all detected UAC devices to have associated ALSA nodes. |
There was a problem hiding this comment.
The README says PASS requires all detected UAC devices to have associated ALSA nodes, but it does not describe the expected behavior when no USB audio peripheral is connected. The script currently reports FAIL in that case. lab users may schedule this test on boards without an attached UAC peripheral and interpret the result incorrectly.
Recommended fix: document “no UAC device connected” as an expected FAIL, or add a YAML/script parameter such as USB_UAC_REQUIRED=1 so labs can choose FAIL vs SKIP.
3 participants
The shell script verifies the enumeration of USB Audio Devices connected to DUT.