[FEATURE] Implement original independent FAT directory anchors #12
Labels
No labels
alpha-1
bug
ci
docs
duplicate
enhancement
help wanted
invalid
performance
phase-6
question
refactor
security
wontfix
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set
Reference
erix/fatd#12
Loading…
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Problem and motivation
Vfsd prepares independent root/cwd scopes through the existing conditional provider-anchor protocol. Fatd currently implements mounted-root acquisition and component custody but not the independent-anchor operation. Completing object-based cwd and launch/exact/redirection migration requires original directory cloning with its own lifetime and an explicit navigation ceiling.
Proposed behavior and scope
Implement the existing
FsProviderAcquireAnchorRequestV1family in Fatd without changing wire layouts. Consume fresh original source evidence, reserve a mount-scoped cleanup key and counted row before publication, and clone the same actual directory independently of its source handle. An optional held ceiling permits bounded parent navigation; omission narrows the clone to the source directory itself.Selected catalogs prove actual ancestry by original serials. Imported media uses original retained directory chains where those prove the selected ceiling; absent proof must refuse without a pathname or mounted-root fallback. Share parent selection, directory observations and ordinary-custody admission. Update original root retirement to retain all dependent ceilings and retry keys.
Authority, security and reliability
The original startup provider/storage route remains the only authority. Directory names, serials, mount IDs and permission fields do not select a new route. Both source and explicit ceiling must be actual live directories on that original mount. Freshness is one-use; replay fences never reset. Clones do not inherit source cleanup ownership. Bound descendants retain their exact original ceiling; its retirement disables ordinary use before waiting for descendants. Lost replies retire by original mount/key independently of source liveness. Unknown cleanup preserves disabled custody and the existing fail-stop boundary.
Acceptance criteria
Checklist
Full compatibility cwd migration and guest self-hosting remain separate unfinished requirements.
Alternatives and tradeoffs
Reopening the source spelling cannot preserve original identity through moves or removal. Promoting a bounded current directory to the mounted root would widen authority. Original catalog ancestry supports selected formats without a depth ceiling; imported media requires counted retained ancestry and keeps its existing mutation guards. Deployment storage limits remain explicit capacity outcomes rather than hidden fallback behavior.
Tracking and rollout
Prerequisite within the namespace/lifetime feature, building on current parent observations. Provider review, Integration review, phase requirements. Keep legacy spelling guards until all Vfsd consumers are migrated; this provider operation alone does not close the full cwd/lifetime criteria.
Current independent FAT directory anchors — 1 October 2026
Imported clones copy only required retained ancestors below their ceiling, carrying no source cleanup keys or broader original roots.
Full Vfsd cwd/launch/exact/redirection migration and complete cross-provider lifetime remain open; legacy spelling guards stay until their consumers migrate.
Both full EriX guest build generations and external Rust/LLVM rebuilding remain required and unperformed.
Related records and contracts
erikinkinen referenced this issue2026-10-01 10:05:54 +02:00
erikinkinen referenced this issue from erix/docs2026-10-01 10:05:54 +02:00
erikinkinen referenced this issue from erix/integration2026-10-01 10:05:55 +02:00