[BUG] Selector-race VM fixture bypasses native exact-entry acquisition #8
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/ramfsd#8
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
Native exact-entry acquisition bypassed Ramfsd's selector-race hook. The VM case
could pass after an ordinary
mkdirwithout injecting the intended race.Reproduction
At Ramfsd
d159c5a3, compareACQUIRE_EXACT_ENTRY_V1with the legacyOPEN_EXACT_ENTRYdispatch. Only the latter calls the hook. The originalappliance-disk-image-exact-selector-race-negativecase acceptsSELECTOR0without proving that injection ran.
Expected and actual behavior
Inject rename/replacement after acquiring the original parent, preserve that
parent's identity, and reject an uninjected run. The old hook missed native
acquisition and expected
BUSYfrom rename despite retained-object lifetime.Failed injection must retire its original acquisition before refusing.
Environment and identity
Originally found by source inspection at Ramfsd
d159c5a3, Vfsd4f18e43eandIntegration
914650c7. The fixture targets x86-64 QEMU and uses the existingprivate filesystem endpoint and physical consent route. No image or toolchain
identity was recorded for that inspection; the last working revision is unknown.
Evidence and prior checks
Original native dispatch
and original scenario.
The repaired fixture requires unique, ordered
RACESETUP0,SELECTOR0andREPLACEMENT0alongside the existing exit, input and cleanup gates.Tracking
Ramfsd review,
Integration review,
native ownership,
fixture contract,
manual review,
Phase 6 tracker.
erikinkinen referenced this issue2026-10-01 23:51:42 +02:00
erikinkinen referenced this issue from erix/integration2026-10-01 23:51:45 +02:00
erikinkinen referenced this issue2026-10-02 00:08:40 +02:00
erikinkinen referenced this issue from erix/integration2026-10-02 00:08:40 +02:00
erikinkinen referenced this issue from erix/docs2026-10-02 00:08:44 +02:00
erikinkinen referenced this issue from erix/vfsd2026-10-02 00:08:44 +02:00
CI 2825 at
f4ff9683exercises the repaired native fixture with Ramfsd29acd773and passes.The run builds the selector-race feature artifact and requires all three ordered
zero-status markers, along with the original input, exit and cleanup gates.
This confirms the fixture repair. The historical observer exception in
Integration #64 remains
separate.