| Previous | Next |
| ERROR_INTERRUPT_VECTOR_ALREADY_CONNECTED | ERROR_WAIT_FOR_OPLOCK |
ERROR_INTERRUPT_STILL_CONNECTED
an interrupt connection remains active when teardown expected it to be gone.
ERROR_INTERRUPT_STILL_CONNECTED identifies the condition that this condition This status reveals incomplete interrupt cleanup. A device or driver is progressing through stop, removal, unload, or reconfiguration while an interrupt object is still connected.
Where the result appears
- PnP stop and remove paths.
- driver unload.
- failed-start rollback.
- device reset or resource rebalance.
How to interpret the result
Interpret the numeric value in the Win32/LRESULT domain and in the context of the API that returned it. Preserve this result before wrappers convert it to an HRESULT, exception, Boolean failure, or provider-specific message. When recording it, note whether the operation produced usable output; this determines whether the value is completion information, control flow, or a terminal failure.
Diagnostic evidence to collect
- interrupt object and vector.
- last successful connect and disconnect call sites.
- in-flight DPC and ISR counters.
- PnP state transition and device power state.
A diagnostic record for it should contain the API name, input flags, relevant handles or object identifiers, thread and process identity, timestamp, and the immediately preceding state transition. Those fields are more useful than a generic screenshot because they show which contract and lifecycle phase produced the value.
Handling and recovery
Quiesce the device, disable interrupt generation, synchronize with the ISR, drain deferred work, and disconnect the interrupt before releasing resources. Cleanup must also execute after partial initialization failures.
Retry after this result only when an external state relevant to this operation can actually change. When it is rooted in object state, unsupported capability, ownership, or invalid sequencing, an unchanged retry repeats the defect and can obscure the first useful diagnostic record.
Common misinterpretation
Forcing resource release while the interrupt remains connected can create use-after-free crashes. The correct fix is lifecycle ordering, not suppression of the status.
Guidance for developers
Keep the original this result value and its Win32 domain in structured telemetry. If it crosses COM or another HRESULT-based boundary, store both the source Win32 value and the converted result. If it crosses JSON, RPC, or a database boundary, include error_domain, error_code, operation, and object_state so the receiver does not apply a second conversion.
An automated test should construct the specific state that leads to this result, assert the exact returned value, and verify cleanup on both the normal and exceptional path. The result test should additionally prove that the caller does not collapse this status into an unrelated generic error and does not retry forever.
References
Looking for a different code? Search another status or error code.