Fix race conditions in file cache and block cache

[jfantinhardesty]

What type of Pull Request is this? (check all applicable)

• Refactor
• Feature
• Bug Fix
• Optimization
• Documentation Update

Describe your changes in brief

This fixes most races that can be detected using the race detector. (There are still some races that would require a fix in the upstream tlru). In the future, we could hopefully enable the race detector in our CI/CD as well. This also adds a few tests for race conditions that were found while running cloudfuse but were not in our test suite.

Checklist

• Tested locally
• Added new dependencies
• Updated documentation
• Added tests

Related Issues

• Related Issue #
• Closes #

[Copilot]

🟡 Changes recommended

Worker shutdown can still accept lost work, and block-cache workers still race with handle-path renames.

2 open findings

• Queued work races with rename and uses stale file paths · New
• Schedule races with shutdown and loses queued work · New

What changed in this PR

Fixes concurrency issues across cache, streaming, loopback, and xload components, with race-focused tests.

Changes:

• Synchronizes shared cache and handle state.
• Improves worker shutdown and async-operation cleanup.
• Adds regression tests for concurrent reads, writes, renames, expiry, and uploads.

File	Description
component/​xload/​xload_test.go	Reorders configuration cleanup.
component/​xload/​threadpool.go	Revises worker shutdown.
component/​xload/​threadpool_test.go	Removes timing-dependent assertion.
component/​xload/​stats_manager.go	Synchronizes exporter shutdown.
component/​xload/​splitter_test.go	Stops all test components.
component/​xload/​lister.go	Localizes goroutine error state.
component/​stream/​read.go	Atomically reads cached-object count.
component/​stream/​read_test.go	Avoids unsafe mock comparisons.
component/​loopback/​loopback_fs.go	Locks handle-path updates and reads.
component/​loopback/​loopback_fs_test.go	Tests rename/read concurrency.
component/​file_cache/​lru_policy.go	Protects expiry-list inspection.
component/​file_cache/​lru_policy_test.go	Tests concurrent expiry updates.
component/​file_cache/​file_cache.go	Synchronizes workers, handles, and scheduling.
component/​file_cache/​file_cache_test.go	Adds race regression coverage.
component/​file_cache/​async.go	Locks upload-window transitions.
component/​block_cache/​threadpool.go	Snapshots scheduled file size.
component/​block_cache/​threadpool_test.go	Uses eventual async assertions.
component/​block_cache/​block_cache.go	Coordinates uploads and overwrites.
component/​block_cache/​block_cache_test.go	Tests upload/write races.
component/​attr_cache/​cacheMap.go	Uses copy-on-write attributes.
component/​attr_cache/​attr_cache_test.go	Tests immutable cached results.

🧠 Review effort: Balanced

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.