Repository navigation
Discover optional Lerwatt advice without manual integration setup - #8
Merged
Merged
Conversation
Default new services and foreground helpers to best-effort discovery. Keep scheduling models, CPUperf selection and saved service choices independent of peer presence. Preserve existing pins in automatic mode and disable BPF advice consumption if publication fails. Validate real kernel startup orders, expiry, reconnection, recovery and conflicting-pin fallback.
Raise the test-only subprocess budget from 15 to 45 seconds and include recent stub events on timeout. Preserve the failed CI evidence and all transaction assertions; production watchdog and thermal limits are unchanged.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
New Lertick services and normal foreground helper runs use best-effort Lerwatt discovery without integration on. Automatic mode preserves the governor-based CPUperf default, refuses to clear existing pins and disables BPF advice consumption if publication fails, while local scheduling continues. Strict legacy diagnostics remain available.
New installations discover a compatible running peer automatically while each project keeps its own model, controls and lifecycle. Package/file presence is not treated as a connected peer. Version, topology, session and freshness checks remain ABI 2.
The counterpart is not installed, started, stopped or retrained. Either startup order works; fresh advice reconnects after independent restarts and expires safely when a publisher stops. Saved settings, including off opt-outs, remain preserved. Prediction selection is independent of discovery; enabling the link does not imply better model accuracy.
Validation: native ARM64 Ubuntu 24.04.5 VM, kernel 7.0.0-38-generic, eight vCPUs and real sched_ext/BPF plus PID 1. Both projects' deployment/ABI tests passed. The paired test proves real contract consumption, both startup orders, expiry/resumption, independent model-method changes, restoration and a conflicting test-owned pin preserved during independent scheduling. Hardware controls use simulated files. Development failures are retained in results/automatic-discovery.
Known limits: two prior cooperative finite-work pairs still regress by 0.02/0.07 seconds; no energy or accuracy improvement is claimed. A conflicting/unavailable local publication environment needs local repair/restart, while an absent peer joins later without restart. Existing saved off configurations remain off on upgrade.