Skip to content

Draft: Fix trace not masking from delayed SetMasked()

Arran Walker requested to merge ajwalker/trace-fix into master

What does this MR do?

Allows SetMasked() and SetLimit() to be called even after writing data to the trace.

Why was this MR needed?

Trace tests pass due to the assumption that SetMasked is called closer to the initialization of creating a new trace.

Due to a race condition found within review, we opted to make SetMasked be a no-op if the writer had already been set up.

However, in practice, we do write data to the trace before using SetMasked().

What's the best way to test this MR?

The job:

job:
  stage: test
  script:
    - env

Should display CI_BUILD_TOKEN=[MASKED] and not CI_BUILD_TOKEN=xxxxxxx.

Additional tests were added to check for delayed SetMasked and SetLimits.

What are the relevant issue numbers?

Edited by Arran Walker

Merge request reports