mirror of
https://github.com/espressif/openthread.git
synced 2026-09-07 09:40:12 +00:00
Sven Kirmess
a61151a5d0
The radio URL can configure the CCA threshold, the FEM LNA gain, the region and a per-channel power cap, but not the power the radio actually transmits at. `ot-ctl txpower` sets it at runtime only, so a deployment cannot be brought up from its configuration alone -- something has to issue a CLI command after every start. `max-power-table` does not cover this. It is a cap: the effective power is `min(txPower, channelMax)`, so it can lower the power but never raise it above whatever the RCP happens to be at. Nothing on the host sets that power during initialisation, and OPENTHREAD_CONFIG_DEFAULT_TRANSMIT_POWER is consumed by the platform layer rather than here, so an RCP-based deployment runs at whatever constant its RCP firmware was built with. This adds `tx-power[=dbm]`, applied during radio initialisation next to `cca-threshold` and `fem-lnagain`. The two combine rather than override: the radio transmits at `min(tx-power, max-power-table)`, which the help text spells out -- conditionally, since `max-power-table` is itself behind OPENTHREAD_POSIX_CONFIG_MAX_POWER_TABLE_ENABLE and a build without it would otherwise point at a parameter its own help does not list. `tests/scripts/expect/posix-tx-power.exp` covers it, following posix-ccathreshold.exp: two valid values, the two int8 boundaries that must be rejected, and a run without the parameter that must report 0 dBm -- without that last case the test could not tell a value that came from the radio URL from one the radio happened to be at. Because it goes through `RadioSpinel::SetTransmitPower()`, it also lands in the restoration bookkeeping and survives an RCP reset. Measured against a simulated RCP: tx-power=11 ot-ctl txpower -> 11 dBm no parameter ot-ctl txpower -> 0 dBm and against an SLZB-06M (EFR32, SL-OPENTHREAD/2.4.2.0), with the radio deliberately set to 5 dBm beforehand so neither the default nor a retained value could explain the result: tx-power=11 ot-ctl txpower -> 10 dBm 10 rather than 11 because the EFR32 selects the nearest step at or below the requested power; 5 reads back as 4 for the same reason. Closes #13573