[josephnef]

A small read-only diagnostic that answers "can this radio run a cell node?" for any USB or PCIe radio, the probe used to characterise the Kestrel caps in #579.

What

examples/cellcaps opens the radio named in the environment and prints, as one JSON line, the caps that examples/cell caps_ok checks — plus the ACK-responder and timestamp caps and a cell_node_ok verdict:

DEVOURER_VID=0x0bda DEVOURER_PID=0xc85a build/cellcaps      # USB
DEVOURER_PCIE_BDF=0000:85:00.0        build/cellcaps        # vfio PCIe (-DDEVOURER_PCIE=ON)

{"ev":"cellcaps","radio":"usb 0bda:c85a","chip":"RTL8852C","variant":"C8852C","transport":"usb",
 "tsf_write_ok":0,"tbtt_follows_tsf":0,"beacon_backoff_ctl":1,"hw_beacon_txtsf":1,
 "ack_responder_ok":1,"hw_rx_timestamp":1,"hw_injected_mgmt_txtsf":1,"cell_node_ok":0}

It reuses examples/common/radio_open.h, so the one binary serves both transports, and does nothing beyond the normal open path — no beacon, no TX.

Why

caps_ok only prints its four gate caps on failure, and only from inside the cell node. When qualifying a new radio or bus for a cell role (e.g. the USB-Kestrel question in #579, or the moving-UE node in #563) it's useful to read the full cell-gate verdict directly, without standing up a DU/UE. This is the probe that produced the hardware readout in #579.

Notes

• Guarded like the other radio_open probes; the PCIe path compiles only under -DDEVOURER_PCIE=ON (via DEVOURER_HAVE_PCIE).
• No behaviour change to any shipping path; new target only.

Refs #579.

🤖 Generated with Claude Code

[qodo-free-for-open-source-projects]

PR Summary by Qodo

Add a USB and PCIe cell-capability probe

✨ Enhancement 🕐 Less than 10 minutes

AI Description

• Add a read-only probe for checking whether an opened radio meets the cell-node capability gates.
• Print radio identity, relevant capabilities, and a cell-node verdict as one JSON line.
• Build the probe for USB and, when enabled, vfio PCIe without changing existing radio paths.

Diagram

graph TD
  Env["Radio environment"] --> Open["Shared radio open"] --> Caps["Adapter caps"] --> Gate{"Cell gates pass?"} -->|yes| Output["JSON verdict"]
  Gate -->|no| Output

Loading

High-Level Assessment

A standalone probe is appropriate: it reuses the existing cross-transport open helper and exposes the cell-specific verdict without coupling the diagnostic to the cell executable. Reusing the broader adapter-capability event was considered, but it does not provide this verdict or the focused output.

Files changed (2) +33 / -0

Enhancement (1) +27 / -0

main.cppReport cell-node capability gates as JSON +27/-0

Report cell-node capability gates as JSON

• Opens the environment-selected radio and reports its identity, four cell-node gate capabilities, ACK-responder and timestamp capabilities, and a derived cell_node_ok verdict. Returns an error if opening the radio fails; it does not beacon or transmit.

examples/cellcaps/main.cpp

Other (1) +6 / -0

CMakeLists.txtBuild the cellcaps diagnostic executable +6/-0

Build the cellcaps diagnostic executable

• Adds a cellcaps target with the shared environment configuration, devourer and libusb links, and access to the common radio-opening header. PCIe support follows the existing build configuration.

CMakeLists.txt

[qodo-free-for-open-source-projects]

Code Review by Qodo

🐞 Bugs (0) 📘 Rule violations (0) 📎 Requirement gaps (0) 🎨 UX issues (0) 🔗 Cross-repo conflicts (0) 📜 Skill insights (0)

1. Probe builds on Windows despite POSIX-only convention ✓ Resolved

Description

The cellcaps add_executable is declared unconditionally in CMakeLists.txt, while every other
radio_open.h-based probe sits under if(NOT WIN32) (the m0probe loop and the 1687/1711 blocks), and
the file's own comment there says these bench tools are POSIX-only. On a Windows configure the new
target is therefore built anyway, which can fail the build or ship a tool that the surrounding
convention says should not exist there.

---

2. Probe's cell verdict can drift from caps_ok ✓ Resolved

Description

cell_ok in the probe re-implements the four-cap conjunction from caps_ok in
examples/cell/main.cpp instead of sharing it, so the JSON cell_node_ok verdict is a copy. If the
gate in examples/cell changes (a fifth cap, or a transport-specific rule), the probe will keep
reporting the old verdict and the hardware readouts used in #579 will silently stop matching what
the cell node enforces.

---

Tip of the day

💡 Did you know, you can ask Qodo to dismiss a finding you disagree with, with your reason on record

More tips ↗ | Customize Qodo ↗ | Qodo docs ↗

[josephnef]

Both fixed in fc0c3ea2:

1. POSIX-only. The cellcaps target is now wrapped in if(NOT WIN32), like the other radio_open bench probes (libusb + _exit).
2. Verdict drift. Factored the cell-gate predicate into examples/common/cell_caps.h (cell_node_caps_ok(const AdapterCaps&)). Both examples/cell caps_ok and the cellcaps probe call it now, so the probe's cell_node_ok can't drift from the gate the cell node actually applies. ctest 95/95.