# Reproduce `waitcheck-gpt-oss`

Sanitizer case `waitcheck-gpt-oss` from run `2026-10-31T140437-41200155808` of the AORTA sanitizer nightly. This directory is the run area for that one case: its report, the sanitizer output the verdict came from, and the provenance below. See `index.html` for every file actually published here.

## Run

- Commit: `d22f6116235c05395a0241a83d6f7e952a988454`
- Date: 2026-10-31 14:04:37 UTC
- Target: `gfx950`
- Class: survey (observed-only, non-gating)
- Workflow run: https://github.com/ROCm/aorta/actions/runs/41200155808
- Container image: `rocm/pytorch:rocm10.0_ubuntu26.04_py3.14_pytorch_release_2.13.0@sha256:3174cb7061d94c427da96c0edef4adea28046fa3f3b2ff3948dc4e995665ff8c`
- rocjitsu bundle: `d03c6fe410eed62b211d56bb7878dd1f494269fe` (https://github.com/ROCm/rocm-systems/actions/runs/37841600188)

## Observed

- Verdict: `pass`
- Execution: `complete`
- Findings: 0

## Run it yourself

First, the checkout this run used:

```
git clone https://github.com/ROCm/aorta && cd aorta
git checkout d22f6116235c05395a0241a83d6f7e952a988454
pip install -e .
```

Set `ROCJITSU_PREBUILT` (unpacked rocjitsu bundle supplying rj_waitcheck and the ConSan hook) before running.

Finally, the sweep itself:

```
python scripts/sanitizers/model_survey.py run --inventory recipes/sanitizers/models/gpt-oss.yaml --out results
```

## Recorded digests

| Key | Value |
|---|---|
| `code_object:dispatched.hsaco` | `02170daa86f629b0ea705f94033d775cf4d939c72e40c87b9744bb783ada493d` |
| `path` | `/workspace/aorta/.sanitizer-nightly/rocjitsu-prebuilt/bin/rj_waitcheck` |
| `sha256` | `2c406bcc3be32c48ae0ac46d0ca197a7b9a6173e0d25f5b5226fa972d8e31e77` |

## Files here

See `index.html` for the browsable list. Logs are gzipped; `sanitizer_report.json` is the full `aorta.sanitizer_report/0.1` document the dashboard renders from.