OFFLINE DEVELOPER REPORT
WaitLens
Follow QNX IPC waits from a thread to its destination process.
pidin queries processes sequentially; this is not an atomic snapshot. Process-level cycle candidates do not prove deadlock. Destination threads and timing guarantees cannot be inferred. PID/TID reuse can affect comparisons.
Source labels: synthetic
cycle.txt
Input: synthetic
Input provenance is supplied by the caller, not independently verified.
IPC waits · thread to process
| Source PID:TID | State | Destination PID | Coverage | Source line |
|---|---|---|---|---|
| 200:1 | SEND | 300 | observed | 2 |
| 300:1 | REPLY | 200 | observed | 3 |
Process cycle candidates
200, 300
Multiple server threads can make process cycles harmless. Investigate with a kernel trace.
All parsed threads
| PID:TID | Name | Priority | State | Blocked | Source line |
|---|---|---|---|---|---|
| 200:1 | demo_client | 10r | SEND | 300 | 2 |
| 300:1 | demo_service | 10r | REPLY | 200 | 3 |
| 300:2 | demo_service | 10r | RECEIVE | 1 | 4 |
| 400:1 | idle_service | 10r | RECEIVE | 200 | 5 |
Normalized input SHA-256: 8190d7910f301bc00fa4530afc783521d46bd951fef7902414b3775a9bc58b76
Read the evidence
- SEND
- The client is waiting for its message to be received.
- REPLY
- The message has been received; the client is waiting for a reply.
- RECEIVE
- The server is waiting on a local channel. Its blocked value is not a process edge.
Next: inspect the referenced source rows, check destination coverage, and recapture if a phase is missing. Use a kernel trace to investigate persistence or scheduling. Two samples do not measure wait duration.