DNS resolver LRU cache used []string slice for access ordering:
- moveToFrontLocked: O(n) linear scan + slice重组 per cache update
- storeCache held exclusive write lock during entire O(n) operation,
blocking all concurrent reads
Replace with container/list + map[string]*list.Element:
- storeCache: O(1) MoveToFront via list.Element pointer
- evictLRULocked: O(1) Back() + Remove()
- DeleteCacheEntry: O(1) instead of O(n) scan
This matches the LRU pattern already used by FileCache and
FileInfoCache. Write lock duration reduced from O(n) to O(1),
significantly reducing read contention under mixed workloads.
Conflict: sendfile.go (!linux build tag) was incorrectly modified to
include linuxSendfile and getSocketFd functions which already exist
in sendfile_linux.go.
Resolution: Keep HEAD version (simple fallback returning ENOTSUP) as
Linux implementation is properly separated in sendfile_linux.go.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
