fix(radio): restore upstream token-bucket duty-cycle enforcement

Commit 22eb9b87 ("Revert 'Merge remote-tracking branch origin/dev...'")
reverted an entire upstream merge to escape a bad merge state, dropping
860 lines across 66 files. Among the collateral never reconciled on a
later re-merge was eb4fa032's token-bucket duty-cycle enforcement — the
mechanism that keeps nodes under a configured airtime budget (and EU
868 MHz nodes under the legally-mandated duty cycle). The fork had fallen
back to fixed per-packet spacing (getAirtimeBudgetFactor reverted to 2.0),
losing the windowed enforcement.

This was never an intentional design choice, so restoring it re-aligns the
fork with upstream and REDUCES the merge-conflict surface: Dispatcher.{h,cpp}
now diverge from upstream by watchdog additions only (77 insertions, 0
deletions) instead of rewriting checkSend()/loop().

Restored from upstream: updateTxBudget/tx_budget_ms/duty_cycle_window_ms/
getRemainingTxBudget/getDutyCycleWindowMs and the windowed budget logic in
Dispatcher; getOutboundTotal() and the 0xFFFFFFFF count-all sentinel in
StaticPoolPacketManager; the getOutboundTotal() call in StatsFormatHelper.
Re-applied the fork's MQTT radio-watchdog on top as pure additions
(#ifdef WITH_MQTT_BRIDGE), keeping formatRadioDiag.

Stored airtime_factor settings keep their meaning: fork's t*factor spacing
and upstream's 1/(1+factor) windowed budget yield the same steady-state
duty cycle; upstream additionally allows short bursts within the window.

Phase 2 (CAD / radio_fem_rxgain, which touch NodePrefs persistence and
per-board FEM wiring) is documented in RESTORE_UPSTREAM_NOTES.md, not done
here. Builds: Heltec_v3 observer + plain repeater. NEEDS ON-DEVICE
duty-cycle validation before merge.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
agessamanandClaude Fable 5 committed 2026-07-09 10:08:14 -07:00
1 parent 58b9cb66a6
commit 7076ff5b31
6 files changed
+147 -11

No files matched your search

+67
View File
@@ -0,0 +1,67 @@
# Restoring accidentally-reverted upstream features
## Background
On 2026-03-20, commit `22eb9b87` — *Revert "Merge remote-tracking branch 'origin/dev' into mqtt-bridge-implementation"* — reverted an entire upstream merge to escape a bad merge state, deleting 860 lines across 66 files. That was not intentional feature removal; it wholesale dropped a batch of upstream progress. When upstream was later re-merged, some collateral came back (MicroNMEA `claim()/release()`, the GAT562 board) but several upstream features were never reconciled and remained missing.
This branch restores them. They are **pure upstream code the fork accidentally dropped**, so restoring re-aligns the fork with upstream and *reduces* the future merge-conflict surface rather than adding to it.
## Phase 1 — duty-cycle enforcement (done on this branch)
Restored the token-bucket duty-cycle enforcement and its cluster:
- `src/Dispatcher.{h,cpp}` — restored upstream's `updateTxBudget()`, `tx_budget_ms`,
`duty_cycle_window_ms`, `getRemainingTxBudget()`, `getDutyCycleWindowMs()`, and the
windowed `checkSend()`/`loop()` budget logic. The fork's MQTT **radio watchdog** was
re-applied on top as pure additions (behind `#ifdef WITH_MQTT_BRIDGE`).
- `src/helpers/StaticPoolPacketManager.{h,cpp}` — restored `getOutboundTotal()` and the
`0xFFFFFFFF` count-all sentinel in `countBefore()`.
- `src/helpers/StatsFormatHelper.h` — restored the `getOutboundTotal()` call; kept the
fork's `formatRadioDiag` template.
Net effect: these files now diverge from upstream by **watchdog additions only (0 deletions)**
instead of rewriting upstream's TX logic.
### Why this is safe (no reinterpretation of stored settings)
`getAirtimeBudgetFactor()` changed *meaning* between fork and upstream, but the resulting
duty cycle is the **same formula**:
- Fork: `next_tx = t · factor` → duty ≈ `1/(1+factor)`
- Upstream: `duty = 1/(1+factor)` enforced over a rolling window
So a device's stored `airtime_factor` (a `NodePrefs` field, unchanged) keeps its meaning.
The only behavioral difference is upstream enforces it over a 1-hour window (allowing
short bursts) instead of rigid per-packet spacing — a strict improvement, and the
mechanism that keeps EU 868 MHz nodes under the legally-mandated duty cycle.
### Must be validated on-device before merge
This touches core TX timing. Confirm on hardware:
- Normal traffic still flows (repeater forwards, observer uplinks).
- Sustained TX is throttled to the configured duty cycle (watch `get` airtime/queue stats;
verify `next_tx_time` spacing under load).
- The observer radio watchdog still recovers a stuck radio (`radio_watchdog_minutes`).
## Phase 2 — CAD and FEM RX gain (NOT done; needs care + device testing)
Still missing at HEAD, also dropped by `22eb9b87`, still present upstream:
- **`cad_enabled`** — hardware Channel Activity Detection (listen-before-talk) before TX.
The `Dispatcher`/`Radio` interface (`setCADEnabled`/`getCADEnabled`) is restored by
Phase 1, so CAD currently stays **off by default** (unchanged behavior). To make it
configurable again, restore:
- `NodePrefs.cad_enabled` field, its CLI get/set, and its persistence in
`CommonCLI.cpp` (`loadPrefsInt`/`savePrefs`). **Offset care:** this branch's
`/com_prefs` layout is carefully managed — add `cad_enabled` following the same
append-and-size-guard pattern used for `rx_boosted_gain`/`flood_max_*`, and add a
host-side round-trip test (see `scratchpad/migtest`).
- The `MyMesh::getCADEnabled()` override returning `_prefs.cad_enabled`.
- `RadioLibWrappers::setCADEnabled()` override so the hardware CAD is actually driven
(the fork's wrapper currently doesn't override it).
- **`radio_fem_rxgain`** — LoRa front-end-module RX gain. Restore `NodePrefs.radio_fem_rxgain`
+ CLI + persistence (same offset care), plus the per-board FEM wiring reverted across
~20 `variants/*/target.cpp` and the `heltec_tracker_v2/LoRaFEMControl.{cpp,h}` files.
This is board-specific and only affects FEM-equipped hardware.
Phase 2 is lower urgency than duty-cycle enforcement (CAD/FEM are capability gaps, not a
compliance regression) and is best done as its own change with per-board hardware testing.
+57 -8
View File
@@ -8,7 +8,9 @@
namespace mesh {
#define MAX_RX_DELAY_MILLIS 32000 // 32 seconds
#define MAX_RX_DELAY_MILLIS 32000 // 32 seconds
#define MIN_TX_BUDGET_RESERVE_MS 100 // min budget (ms) required before allowing next TX
#define MIN_TX_BUDGET_AIRTIME_DIV 2 // require at least 1/N of estimated airtime as budget before TX
#ifndef NOISE_FLOOR_CALIB_INTERVAL
#define NOISE_FLOOR_CALIB_INTERVAL 2000 // 2 seconds
@@ -20,12 +22,34 @@ void Dispatcher::begin() {
_err_flags = 0;
radio_nonrx_start = _ms->getMillis();
duty_cycle_window_ms = getDutyCycleWindowMs();
float duty_cycle = 1.0f / (1.0f + getAirtimeBudgetFactor());
tx_budget_ms = (unsigned long)(duty_cycle_window_ms * duty_cycle);
last_budget_update = _ms->getMillis();
_radio->begin();
prev_isrecv_mode = _radio->isInRecvMode();
}
float Dispatcher::getAirtimeBudgetFactor() const {
return 2.0; // default, 33.3% (1/3rd)
return 1.0;
}
void Dispatcher::updateTxBudget() {
unsigned long now = _ms->getMillis();
unsigned long elapsed = now - last_budget_update;
float duty_cycle = 1.0f / (1.0f + getAirtimeBudgetFactor());
unsigned long max_budget = (unsigned long)(getDutyCycleWindowMs() * duty_cycle);
unsigned long refill = (unsigned long)(elapsed * duty_cycle);
if (refill > 0) {
tx_budget_ms += refill;
if (tx_budget_ms > max_budget) {
tx_budget_ms = max_budget;
}
last_budget_update = now;
}
}
int Dispatcher::calcRxDelay(float score, uint32_t air_time) const {
@@ -48,6 +72,7 @@ uint32_t Dispatcher::getRadioWatchdogMillis() const {
void Dispatcher::loop() {
if (millisHasNowPassed(next_floor_calib_time)) {
_radio->triggerNoiseFloorCalibrate(getInterferenceThreshold());
_radio->setCADEnabled(getCADEnabled());
next_floor_calib_time = futureMillis(NOISE_FLOOR_CALIB_INTERVAL);
}
_radio->loop();
@@ -93,11 +118,24 @@ void Dispatcher::loop() {
if (outbound) { // waiting for outbound send to be completed
if (_radio->isSendComplete()) {
long t = _ms->getMillis() - outbound_start;
total_air_time += t; // keep track of how much air time we are using
total_air_time += t;
//Serial.print(" airtime="); Serial.println(t);
// will need radio silence up to next_tx_time
next_tx_time = futureMillis(t * getAirtimeBudgetFactor());
updateTxBudget();
if (t > tx_budget_ms) {
tx_budget_ms = 0;
} else {
tx_budget_ms -= t;
}
if (tx_budget_ms < MIN_TX_BUDGET_RESERVE_MS) {
float duty_cycle = 1.0f / (1.0f + getAirtimeBudgetFactor());
unsigned long needed = MIN_TX_BUDGET_RESERVE_MS - tx_budget_ms;
next_tx_time = futureMillis((unsigned long)(needed / duty_cycle));
} else {
next_tx_time = _ms->getMillis();
}
_radio->onSendFinished();
last_radio_active_ms = _ms->getMillis(); // TX success → radio is alive
@@ -268,9 +306,20 @@ void Dispatcher::processRecvPacket(Packet* pkt) {
}
void Dispatcher::checkSend() {
if (_mgr->getOutboundCount(_ms->getMillis()) == 0) return; // nothing waiting to send
if (!millisHasNowPassed(next_tx_time)) return; // still in 'radio silence' phase (from airtime budget setting)
if (_radio->isReceiving()) { // LBT - check if radio is currently mid-receive, or if channel activity
if (_mgr->getOutboundCount(_ms->getMillis()) == 0) return;
updateTxBudget();
uint32_t est_airtime = _radio->getEstAirtimeFor(MAX_TRANS_UNIT);
if (tx_budget_ms < est_airtime / MIN_TX_BUDGET_AIRTIME_DIV) {
float duty_cycle = 1.0f / (1.0f + getAirtimeBudgetFactor());
unsigned long needed = est_airtime / MIN_TX_BUDGET_AIRTIME_DIV - tx_budget_ms;
next_tx_time = futureMillis((unsigned long)(needed / duty_cycle));
return;
}
if (!millisHasNowPassed(next_tx_time)) return;
if (_radio->isReceiving()) {
if (cad_busy_start == 0) {
cad_busy_start = _ms->getMillis(); // record when CAD busy state started
}
+15 -2
View File
@@ -68,6 +68,8 @@ public:
virtual void triggerNoiseFloorCalibrate(int threshold) { }
virtual void setCADEnabled(bool enable) { }
virtual void resetAGC() { }
virtual uint8_t getRadioState() const { return 0; }
@@ -102,6 +104,7 @@ public:
virtual void queueOutbound(Packet* packet, uint8_t priority, uint32_t scheduled_for) = 0;
virtual Packet* getNextOutbound(uint32_t now) = 0; // by priority
virtual int getOutboundCount(uint32_t now) const = 0;
virtual int getOutboundTotal() const = 0;
virtual int getFreeCount() const = 0;
virtual Packet* getOutboundByIdx(int i) = 0;
virtual Packet* removeOutboundByIdx(int i) = 0;
@@ -141,8 +144,12 @@ class Dispatcher {
bool prev_isrecv_mode;
uint32_t n_sent_flood, n_sent_direct;
uint32_t n_recv_flood, n_recv_direct;
unsigned long tx_budget_ms;
unsigned long last_budget_update;
unsigned long duty_cycle_window_ms;
void processRecvPacket(Packet* pkt);
void updateTxBudget();
protected:
PacketManager* _mgr;
@@ -155,12 +162,15 @@ protected:
{
outbound = NULL;
total_air_time = rx_air_time = 0;
next_tx_time = 0;
next_tx_time = ms.getMillis();
cad_busy_start = 0;
next_floor_calib_time = next_agc_reset_time = 0;
_err_flags = 0;
radio_nonrx_start = 0;
prev_isrecv_mode = true;
tx_budget_ms = 0;
last_budget_update = 0;
duty_cycle_window_ms = 3600000;
last_watchdog_recovery = 0;
last_radio_active_ms = 0;
}
@@ -179,7 +189,9 @@ protected:
virtual uint32_t getCADFailRetryDelay() const;
virtual uint32_t getCADFailMaxDuration() const;
virtual int getInterferenceThreshold() const { return 0; } // disabled by default
virtual bool getCADEnabled() const { return false; } // hardware CAD disabled by default
virtual int getAGCResetInterval() const { return 0; } // disabled by default
virtual unsigned long getDutyCycleWindowMs() const { return 3600000; }
#ifdef WITH_MQTT_BRIDGE
virtual uint32_t getRadioWatchdogMillis() const; // observer-only radio recovery
#endif
@@ -192,8 +204,9 @@ public:
void releasePacket(Packet* packet);
void sendPacket(Packet* packet, uint8_t priority, uint32_t delay_millis=0);
unsigned long getTotalAirTime() const { return total_air_time; } // in milliseconds
unsigned long getTotalAirTime() const { return total_air_time; }
unsigned long getReceiveAirTime() const {return rx_air_time; }
unsigned long getRemainingTxBudget() const { return tx_budget_ms; }
uint32_t getNumSentFlood() const { return n_sent_flood; }
uint32_t getNumSentDirect() const { return n_sent_direct; }
uint32_t getNumRecvFlood() const { return n_recv_flood; }
+6
View File
@@ -9,6 +9,8 @@ PacketQueue::PacketQueue(int max_entries) {
}
int PacketQueue::countBefore(uint32_t now) const {
if (now == 0xFFFFFFFF) return _num; // sentinel: count all entries regardless of schedule
int n = 0;
for (int j = 0; j < _num; j++) {
if ((int32_t)(_schedule_table[j] - now) > 0) continue; // scheduled for future... ignore for now
@@ -97,6 +99,10 @@ int StaticPoolPacketManager::getOutboundCount(uint32_t now) const {
return send_queue.countBefore(now);
}
int StaticPoolPacketManager::getOutboundTotal() const {
return send_queue.count();
}
int StaticPoolPacketManager::getFreeCount() const {
return unused.count();
}
+1
View File
@@ -29,6 +29,7 @@ public:
void queueOutbound(mesh::Packet* packet, uint8_t priority, uint32_t scheduled_for) override;
mesh::Packet* getNextOutbound(uint32_t now) override;
int getOutboundCount(uint32_t now) const override;
int getOutboundTotal() const override;
int getFreeCount() const override;
mesh::Packet* getOutboundByIdx(int i) override;
mesh::Packet* removeOutboundByIdx(int i) override;
+1 -1
View File
@@ -14,7 +14,7 @@ public:
board.getBattMilliVolts(),
ms.getMillis() / 1000,
err_flags,
mgr->getOutboundCount(0xFFFFFFFF)
mgr->getOutboundTotal()
);
}