Skip to content

send-to-lava: protect LAVA signals from printk corruption#478

Open
smuppand wants to merge 1 commit into
qualcomm-linux:mainfrom
smuppand:lava-emit-result
Open

send-to-lava: protect LAVA signals from printk corruption#478
smuppand wants to merge 1 commit into
qualcomm-linux:mainfrom
smuppand:lava-emit-result

Conversation

@smuppand
Copy link
Copy Markdown
Contributor

@smuppand smuppand commented Jun 3, 2026

Update [send-to-lava.sh](http://send-to-lava.sh/) to make LAVA testcase signal emission more robust on noisy serial consoles.

Kernel printk messages can be interleaved into userspace stdout and corrupt LAVA protocol lines, causing valid test results to be missed or misparsed by LAVA. This change protects only the tiny signal emission critical section without suppressing kernel logs during test execution.

Why

On some platforms, asynchronous kernel printk output can be injected into the middle of a LAVA signal line, for example:

<<<LAVA_SIGNAL_TESTCASE TEST_CASE_ID=OpenCV [ 192.9] qcom,... RESULT=PASS>>>

Kernel printk messages can be interleaved into userspace stdout on noisy
systems and corrupt LAVA TESTCASE signal lines. This can cause LAVA to
miss or misparse otherwise valid test results.

Update send-to-lava.sh to validate and buffer result-file entries before
emitting signals, sanitize testcase IDs, and briefly lower the kernel
console loglevel only around the individual LAVA signal printf when
/proc/sys/kernel/printk is writable. The original printk settings are
saved and restored immediately after each signal line.

Signed-off-by: Srikanth Muppandam <smuppand@qti.qualcomm.com>
@smuppand smuppand added the bug Something isn't working label Jun 3, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

bug Something isn't working

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant