Bugfix: Separate login and message replay timestamps to allow companion RTC differs from app clock

[usrflo]

Summary

Properly fixes the silent-message-rejection issue that was also addressed in #1551 and #1889.
It does so by tracking login replay and message replay in two separate per-client fields instead of sharing one to fix clock/timestamp differences at the client. This uniformly applies to room-server, repeater and sensor.

Fix

Split the two replay tracks:

• Add a transient field ClientInfo::last_login_timestamp.
• The login replay check and the login timestamp assignment use last_login_timestamp.
• The message/REQ handlers are unchanged and keep using last_timestamp.

Result:

• Login replay now runs purely on the companion RTC. Because getCurrentTimeUnique() is monotonic per device, every login's timestamp strictly exceeds the previous one → repeated logins work again (regression fixed).
• Message replay still runs on last_timestamp, which is no longer poisoned by the login's RTC value → messages are no longer rejected when the RTC runs ahead.
• The forward-only RTC sync (clock sync / time refuse to go backwards) no longer matters, because the two replay tracks are independent.

Backward compatibility

No wire-protocol change. Old clients keep working against a patched server (the server just tracks login and messages separately), and patched clients work against an unpatched server.

[usrflo]

Proof by test case "companion RTC in the future"

Stock firmware (dev-Branch) denies messages after login with the debug log statement "possible replay attack detected"

Bug: the login-timestamp from the rtc clock is compared with the message-timestamp from the app clock, login_ts>msg_ts, error

Companion with future time (+3 months)

(venv) $ meshcli -s /dev/meshl1new
INFO:meshcore:Serial Connection started
INFO:meshcore:Connected to meshl1new running on a v1.16.0 fw.
meshl1new|* clock
Current time : 2026-10-14 21:26:08 (1792005968)
meshl1new|* quit
(venv) $ meshcli -s /dev/meshl1new login meshhtv3 [...]
INFO:meshcore:Serial Connection started
INFO:meshcore:Connected to meshl1new running on a v1.16.0 fw.
Login success
(venv) $ meshcli -s /dev/meshl1new msg meshhtv3 "A0 baseline (RTC incorrect)" wait_ack
INFO:meshcore:Serial Connection started
INFO:meshcore:Connected to meshl1new running on a v1.16.0 fw.
Timeout waiting ack
(venv) $ 

Repeater-Serial-Log

DEBUG: RadioLibWrapper: noise_floor = -92
10:52:54 - 15/5/2024 U RAW: 1E00B71C1944EF77CBD5B639451A11EDFE0039F5BEA9CFC83899D45E77C0C71F98CE947AABFC268A17EF0EE6B1C1134A404D37FF0D
10:52:54 - 15/5/2024 U: RX, len=53 (type=7, route=D, payload_len=51) SNR=11 RSSI=-24 score=1000 time=467 hash=C326FAC717A35C86
DEBUG: Login success!
DEBUG: RadioLibWrapper: noise_floor = -92
10:52:55 - 15/5/2024 U: TX, len=22 (type=1, route=F, payload_len=20) [B7 -> 1C]
10:52:57 - 15/5/2024 U RAW: 0501921CB7BEB08B637928A1CC46D89FAE45F491DB6A0B
10:52:57 - 15/5/2024 U: RX, len=23 (type=1, route=F, payload_len=20) SNR=12 RSSI=-43 score=1000 time=324 hash=BD2021EAC9EB7553 [B7 -> 1C]
DEBUG: 10:52:57 - 15/5/2024 U Dispatcher::checkRecv(), score delay below threshold (0)
DEBUG: RadioLibWrapper: noise_floor = -93
10:52:58 - 15/5/2024 U RAW: 2200B71C2CF7EB49D0E271A90242AEF0278AB00F695D
10:52:58 - 15/5/2024 U: RX, len=22 (type=8, route=D, payload_len=20) SNR=12 RSSI=-23 score=1000 time=304 hash=C8340E778B8C42DB [1C -> B7]
DEBUG: PATH to client, path_len=0
DEBUG: RadioLibWrapper: noise_floor = -92
DEBUG: RadioLibWrapper: noise_floor = -93
DEBUG: RadioLibWrapper: noise_floor = -92
10:53:03 - 15/5/2024 U RAW: 0A00B71CDE1891FF919B689CA45E060A87358F2B8CAB07877ED44ECB672C387225CF36EF08D7
10:53:03 - 15/5/2024 U: RX, len=38 (type=2, route=D, payload_len=36) SNR=11 RSSI=-24 score=1000 time=386 hash=8E180F4B9C366CE4 [1C -> B7]
DEBUG: onPeerDataRecv: possible replay attack detected
DEBUG: RadioLibWrapper: noise_floor = -93
DEBUG: RadioLibWrapper: noise_floor = -92

Firmware with this Pull-Request 2834

Comparison of login_ts with login_ts from RTC clock and msg_ts with msg_ts from app clock

Companion with future time (+3 months)

(venv) $ meshcli -s /dev/meshl1new
INFO:meshcore:Serial Connection started
INFO:meshcore:Connected to meshl1new running on a v1.16.0 fw.
meshl1new|* clock
Current time : 2026-10-14 21:19:06 (1792005546)
meshl1new|* quit
(venv) $ meshcli -s /dev/meshl1new login meshhtv3 [...]
INFO:meshcore:Serial Connection started
INFO:meshcore:Connected to meshl1new running on a v1.16.0 fw.
Login success
(venv) $ meshcli -s /dev/meshl1new msg meshhtv3 "A0 baseline (RTC correct)" wait_ack
INFO:meshcore:Serial Connection started
INFO:meshcore:Connected to meshl1new running on a v1.16.0 fw.
Msg acked
(venv) $ 

Repeater-Serial-Log

DEBUG: RadioLibWrapper: noise_floor = -91
10:53:20 - 15/5/2024 U RAW: 1E00B71C1944EF77CBD5B639451A11EDFE0039F5BEA9CFC83899D45E77C0C71F98CE94F1C0189C1D6CAB186EBCA537FA9991ACA4AD
10:53:20 - 15/5/2024 U: RX, len=53 (type=7, route=D, payload_len=51) SNR=11 RSSI=-27 score=1000 time=467 hash=3682E6825BC1DD86
DEBUG: Login success!
DEBUG: RadioLibWrapper: noise_floor = -92
10:53:21 - 15/5/2024 U: TX, len=22 (type=1, route=F, payload_len=20) [B7 -> 1C]
DEBUG: RadioLibWrapper: noise_floor = -92
10:53:24 - 15/5/2024 U RAW: 2200B71CC541AC889F86C4139451502BB8B04831BB5C
10:53:24 - 15/5/2024 U: RX, len=22 (type=8, route=D, payload_len=20) SNR=11 RSSI=-27 score=1000 time=304 hash=0749A2FCBA2A929D [1C -> B7]
DEBUG: PATH to client, path_len=0
DEBUG: RadioLibWrapper: noise_floor = -92
DEBUG: RadioLibWrapper: noise_floor = -91
DEBUG: RadioLibWrapper: noise_floor = -91
DEBUG: RadioLibWrapper: noise_floor = -91
DEBUG: RadioLibWrapper: noise_floor = -92
10:53:34 - 15/5/2024 U RAW: 0A00B71C5D10DD60488C01E467F65A8D8AC0380BF25907877ED44ECB672C387225CF36EF08D7
10:53:34 - 15/5/2024 U: RX, len=38 (type=2, route=D, payload_len=36) SNR=11 RSSI=-26 score=1000 time=386 hash=7EE044590288F8EC [1C -> B7]
DEBUG: RadioLibWrapper: noise_floor = -91
10:53:35 - 15/5/2024 U: TX, len=6 (type=3, route=D, payload_len=4)
10:53:35 - 15/5/2024 U: TX, len=38 (type=2, route=D, payload_len=36) [B7 -> 1C]
DEBUG: RadioLibWrapper: noise_floor = -91
DEBUG: RadioLibWrapper: noise_floor = -91

--> the wrong evaluation "possible replay attack detected" doesn't occur anymore with the bugfix

[usrflo]

It is always a pleasure to see how smoothly repeater/room-server logins work with this pull request.
This PR ist part of the ufo firmware, latest build at time of writing: v0.6