feat(repeater): report the TX-failure counters in the RF status reply
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
This commit is contained in:
1 parent
e06d95ed22
commit
0d8ca69d69
2 files changed
+22
No files matched your search
@@ -497,6 +497,8 @@ int MyMesh::handleRequest(ClientInfo *sender, uint32_t sender_timestamp, uint8_t
|
||||
stats.n_flood_dups = ((SimpleMeshTables *)getTables())->getNumFloodDups();
|
||||
stats.total_rx_air_time_secs = getReceiveAirTime() / 1000;
|
||||
stats.n_recv_errors = radio_driver.getPacketsRecvErrors();
|
||||
stats.n_tx_start_fail = getNumTxStartFail(); // CUSTOM (TeTeHacko)
|
||||
stats.n_tx_timeout = getNumTxTimeout(); // CUSTOM (TeTeHacko)
|
||||
memcpy(&reply_data[4], &stats, sizeof(stats));
|
||||
|
||||
return 4 + sizeof(stats); // reply_len
|
||||
|
||||
@@ -57,8 +57,28 @@ struct RepeaterStats {
|
||||
uint16_t n_direct_dups, n_flood_dups;
|
||||
uint32_t total_rx_air_time_secs;
|
||||
uint32_t n_recv_errors;
|
||||
// CUSTOM (TeTeHacko): appended at the END on purpose. Clients parse this
|
||||
// struct by fixed offsets, so fields added here are simply invisible to older
|
||||
// ones instead of shifting everything after them -- the same way upstream
|
||||
// grew its companion stats frame (docs/stats_binary_frames.md documents a
|
||||
// 26-byte "legacy" and a 30-byte frame that appends recv_errors).
|
||||
//
|
||||
// These two come from the Dispatcher requeue counters. Without them a
|
||||
// repeater that keeps failing to get its transmits out looks perfectly
|
||||
// healthy in every other field: sent/airtime just stop growing, and there is
|
||||
// nothing in the status that says why.
|
||||
uint32_t n_tx_start_fail, n_tx_timeout;
|
||||
};
|
||||
|
||||
// This struct goes on the air byte-for-byte, so its size is part of the
|
||||
// protocol: 56 bytes up to n_recv_errors (what upstream clients expect and
|
||||
// gate on -- meshcore-py's parse_status() reads recv_errors only
|
||||
// `if len(data) >= offset + 56`), plus the 8 appended here. Reordering a field
|
||||
// or letting the compiler insert padding would silently shift every offset
|
||||
// after it, so pin the number instead of trusting review.
|
||||
static_assert(sizeof(RepeaterStats) == 64,
|
||||
"RepeaterStats is a wire format: append only, keep it padding-free");
|
||||
|
||||
#ifndef MAX_CLIENTS
|
||||
#define MAX_CLIENTS 32
|
||||
#endif
|
||||
|
||||
Reference in new issue
Block a user