fix(firmware): the node log's edge frame counters have always reported zero

firmware/esp32-csi-node/main/edge_processing.c

@@ -382,6 +382,14 @@ static uint8_t  s_presence_below_count;  /**< Consecutive frames below low thres
 static bool     s_fall_detected;
 static int8_t   s_latest_rssi;
 static uint32_t s_frame_count;
+/** Frames process_frame() refused. Counted separately from s_frame_count so
+ *  the two together distinguish "the edge stage is idle" from "the edge stage
+ *  is rejecting everything" -- states that look identical from the outside
+ *  and have completely different causes. EDGE_MAX_SUBCARRIERS is
+ *  target-conditional, so a build that selects the pre-HE 128 on an HE part
+ *  rejects every frame here while the task still starts and logs its
+ *  banner. */
+static uint32_t s_frame_rejected;
 
 /** Previous phase velocity for fall detection (acceleration). */
 static float s_prev_phase_velocity;
@@ -1062,7 +1070,10 @@ static void send_feature_vector(void)
 static void process_frame(const edge_ring_slot_t *slot)
 {
     uint16_t n_subcarriers = slot->iq_len / 2;
-    if (n_subcarriers == 0 || n_subcarriers > EDGE_MAX_SUBCARRIERS) return;
+    if (n_subcarriers == 0 || n_subcarriers > EDGE_MAX_SUBCARRIERS) {
+        s_frame_rejected++;
+        return;
+    }
 
     s_frame_count++;
     s_latest_rssi = slot->rssi;
@@ -1389,6 +1400,16 @@ void edge_get_variances(float *out_variances, uint16_t n_subcarriers)
     }
 }
 
+uint32_t edge_processing_get_frames_processed(void)
+{
+    return s_frame_count;
+}
+
+uint32_t edge_processing_get_frames_rejected(void)
+{
+    return s_frame_rejected;
+}
+
 esp_err_t edge_processing_init(const edge_config_t *cfg)
 {
     if (cfg == NULL) {
@@ -1421,6 +1442,7 @@ esp_err_t edge_processing_init(const edge_config_t *cfg)
     s_fall_detected = false;
     s_latest_rssi = 0;
     s_frame_count = 0;
+    s_frame_rejected = 0;
     s_sample_rate_hz = EDGE_CONFIGURED_SAMPLE_RATE_HZ;
     s_filter_design_fs = EDGE_CONFIGURED_SAMPLE_RATE_HZ;
     s_rate_window_start_us = 0;

firmware/esp32-csi-node/main/edge_processing.h

@@ -343,4 +343,21 @@ void edge_get_phase_history(const float **out_buf, uint16_t *out_len,
  */
 void edge_get_variances(float *out_variances, uint16_t n_subcarriers);
 
+/**
+ * Frames the edge pipeline accepted, cumulative since init.
+ *
+ * Read with edge_processing_get_frames_rejected(): the pair is what makes an
+ * idle edge stage distinguishable from one rejecting every frame.
+ */
+uint32_t edge_processing_get_frames_processed(void);
+
+/**
+ * Frames process_frame() refused, cumulative since init.
+ *
+ * Non-zero and climbing with frames_processed stuck at 0 means every frame is
+ * failing the EDGE_MAX_SUBCARRIERS guard -- the signature of a build that
+ * selected the pre-HE 128 on an HE-capable part.
+ */
+uint32_t edge_processing_get_frames_rejected(void);
+
 #endif /* EDGE_PROCESSING_H */

firmware/esp32-csi-node/main/node_log.c

@@ -16,6 +16,7 @@
 
 #include "c6_sync_espnow.h"
 #include "csi_collector.h"
+#include "edge_processing.h"
 #include "nvs_config.h"
 #include "thermal.h"
 
@@ -179,6 +180,14 @@ static void fill_health(node_log_periodic_t *h)
     h->csi_fps_x100  = csi_collector_get_pkt_yield_per_sec() * 100;
     h->tx_send_fail  = csi_collector_get_send_fail_count();
 
+    /* node_log_periodic_t has declared these two since the log shipped and
+     * node_log_read.py has always rendered them, but nothing assigned them,
+     * so every record on every node reported 0/0 -- the memset, not a
+     * measurement. On the one question this log exists to answer about an
+     * unreachable node, it returned a plausible constant zero. */
+    h->frames_processed = edge_processing_get_frames_processed();
+    h->frames_rejected  = edge_processing_get_frames_rejected();
+
     h->disconnect_count  = s_disconnect_count;
     h->last_disc_reason  = s_last_disc_reason;
     h->last_disc_rssi    = s_last_disc_rssi;