[BUG] Malformed materialization reply preserves a success error code #4
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/loaderd#4
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?
Summary
Loaderd's materialization reply helper returns an error carrying result code OK when its transport reports OK with a response length other than the required 36 bytes. The graph path still enters failure cleanup, but its error status contradicts the failure and can prevent a canonical failure reply.
Reproduction
At Loaderd
490438e17be2b299f0f62e5803a00a9a82c0c3c9, inspectcall_materializationinsrc/handoff.rs. Supply a transport resultOk((IPC_RESULT_OK, 35))orOk((IPC_RESULT_OK, 37)). The wrong-length branch constructsHandoffError::handoff(transport_result), therefore itsresult_codeis zero. This is a deterministic source-level reproducer; the new whole-graph regression controls are written but have not yet run against the coherent successor dependencies.Expected and actual behavior
Expected: malformed success framing produces a nonzero protocol failure, and the original active materialization is retained through acknowledged abort. Existing nonzero transport failures remain lossless.
Actual: the helper returns
Errwithresult_code == IPC_RESULT_OK. This report does not claim that malformed output was observed in a guest or that cleanup was bypassed.Environment and identity
490438e17be2b299f0f62e5803a00a9a82c0c3c9,feature/posix-compat.Evidence and prior checks
Independent inspection found the wrong-length branch while reviewing the new authenticated realm-layout binding. The correction maps only an OK transport status with invalid framing to INTERNAL. Controls cover short and oversized OK replies through the actual graph handoff, exact original transaction abort, linker abort and scratch erasure. Source, strict local checks and original CI remain pending before correction acceptance; no failed input has been rerun to claim success.
Tracking
Related: Loaderd authority and quality work, realm lifecycle, and the phase completion record. Keep this issue open until the corrected source, fault controls and required validation are recorded.
erikinkinen referenced this issue2026-09-24 22:45:42 +02:00
erikinkinen referenced this issue from erix/integration2026-09-24 22:45:45 +02:00
Fixed in e3e3d192. Wrong-length replies carrying OK now produce INTERNAL; existing nonzero transport errors are preserved. The graph regression exercises short and oversized replies, verifies abort of the original transaction and linker work, and checks scratch erasure. It passes in the complete warning-free run 286, alongside all 163 unit tests. Closing the reported reply-framing defect.