Fix dead condition, division-by-zero, and uninitialized members in LLM stats (#18819)

[kirklandsign]

Summary:

Three issues fixed:

1. text_llm_runner.cpp: The condition num_generated_tokens == max_new_tokens
   was always false because TextTokenGenerator::generate() receives
   max_new_tokens - 1. Fixed to compare against max_new_tokens - 1.

2. stats.h print_report(): Division by zero when inference/prefill/decode
   time is zero (e.g., during very fast warmup runs). Added guards matching
   the pattern already used in stats_to_json_string().

3. stats.h Stats: Added default initializers (= 0) to all timestamp and
   counter members to prevent undefined behavior from uninitialized reads.

Reviewed By: manuelcandales

Differential Revision: D99708774

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/18819

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❗ 1 Active SEVs

There are 1 currently active SEVs. If your PR is affected, please view them below:

• Rolling out OSDC (ARC) runners on pull workflow for PyTorch trunk commits

❌ 3 New Failures, 18 Cancelled Jobs, 5 Unrelated Failures

As of commit 7a0d529 with merge base 5e8a0df ():

NEW FAILURES - The following jobs have failed:

• Cadence Build & Test / cpu-test / test-aot / test-aot (gh)
  backends/cadence/aot/tests/test_replace_ops_passes.py::TestReplaceOpsPasses::test_replace_conv2d_with_linear
• Lint / lintrunner (gh)
  >>> Lint for extension/llm/runner/stats.h:
• pull / unittest / macos / macos-job (gh)
  export/tests/test_target_recipes.py::TestTargetRecipes::test_mv3_model

CANCELLED JOBS - The following jobs were cancelled. Please retry:

• pull / test-eval_llama-wikitext-linux / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-llama-runner-qnn-linux (fp32, qnn_16a16w, qnn) / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-lora-linux / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-lora-multimethod-linux / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-models-linux (ic4, portable, linux.4xlarge.memory) / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-models-linux (ic4, xnnpack-quantization-delegation, linux.4xlarge.memory) / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-models-linux (mobilebert, portable, linux.2xlarge) / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-models-linux (mobilebert, xnnpack-quantization-delegation, linux.2xlarge) / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-models-linux (phi_4_mini, portable, linux.4xlarge.memory) / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-moshi-linux / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-multimodal-linux (gemma3-4b) / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-parakeet-xnnpack-linux / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-phi-3-mini-runner-linux / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-samsung-models-linux / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-samsung-quantmodels-linux / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-sqnr-static-llm-qnn-linux (smollm2_135m) / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-voxtral-realtime-xnnpack-linux / linux-job (gh)
  ##[error]The operation was canceled.
• pull / test-vulkan-models-linux / linux-job (gh)
  ##[error]The operation was canceled.

BROKEN TRUNK - The following jobs failed but were present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

• pull / unittest / linux / linux-job (gh) (trunk failure)
  ##[error]The operation was canceled.
• pull / unittest / windows / windows-job (gh) (trunk failure)
  ##[error]The operation was canceled.
• pull / unittest-arm-backend-with-no-deps (test_pytest_models_tosa) / linux-job (gh) (trunk failure)
  ##[error]The operation was canceled.
• pull / unittest-editable / linux / linux-job (gh) (trunk failure)
  ##[error]The operation was canceled.
• pull / unittest-editable / windows / windows-job (gh) (trunk failure)
  ##[error]The operation was canceled.

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[meta-codesync]

@kirklandsign has exported this pull request. If you are a Meta employee, you can view the originating Diff in D99708774.

[github-actions]

This PR needs a release notes: label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[Copilot]

Pull request overview

Fixes correctness/robustness issues in the LLM runner stats/reporting path, improving runtime diagnostics and avoiding undefined behavior in edge cases (warmup/fast runs and uninitialized reads).

Changes:

• Fix off-by-one “max new tokens reached” condition in TextLLMRunner::generate().
• Guard token-rate computations in print_report() against division-by-zero.
• Add default initializers to Stats timestamp/counter members to avoid uninitialized reads.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
extension/llm/runner/text_llm_runner.cpp	Fixes a dead/incorrect condition by aligning the comparison with the max_new_tokens - 1 passed to the generator.
extension/llm/runner/stats.h	Initializes Stats members to safe defaults and prevents division-by-zero in printed token-rate metrics.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

stats.num_generated_tokens is int64_t, but the format string uses PRIu64. This is a signed/unsigned format mismatch and can lead to undefined behavior or incorrect output on some platforms. Use the matching PRId64 (or cast the value to uint64_t if it’s guaranteed non-negative) to align the format specifier with the argument type.

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Consider adding a small regression test that calls print_report() with Stats timestamps set such that inference_time_ms/prompt_eval_time/eval_time are 0 (or negative) to ensure the rate calculations don’t reintroduce division-by-zero/inf outputs. The LLM runner already has gtest coverage (e.g., extension/llm/runner/test/test_text_llm_runner.cpp), so this should be straightforward to exercise in CI.