Skip to content

Move radio TX to a dedicated task so the HA WebSocket stays connected - #35

Merged
cjkas merged 53 commits into
mainfrom
fix/radio-tx-task
Oct 12, 2026
Merged

Move radio TX to a dedicated task so the HA WebSocket stays connected#35
cjkas merged 53 commits into
mainfrom
fix/radio-tx-task

Conversation

@cjkas

@cjkas cjkas commented Oct 10, 2026 •
edited
Loading

Copy link
Copy Markdown
Owner

Problem

Home Assistant's WebSocket (port 8080, ping_timeout=15) disconnects during radio operations, especially when several covers are moved at once.

Causes:

  • Web::queueCommand runs on the async_tcp task and waited up to 3 s for loop() to finish transmitting. async_tcp also serves the WebSocket, so pongs stalled. Back-to-back HA commands added up past 15 s.
  • /setMyPosition (35 repeats, about 5 s) and /unpairShade transmitted directly on async_tcp. They also raced loop() on the CC1101.
  • Transceiver::sendFrame busy-waits about 140 ms per frame, with all repeats sent in one call.
  • API repeat was unbounded (up to 127, about 18 s).

Changes

  • Radio TX task: new radio_tx task (priority 5) fed by a 32-slot job queue.
    • SomfyShadeController::sendFrame, SomfyRemote::repeatFrame and repeater frames now queue a copy of the frame instead of transmitting inline.
    • The 80-bit repeat counter is encoded exactly as before.
    • The task waits 100 ms between jobs, which replaces the delay() spacing in checkMovement.
  • Radio mutex: a recursive mutex serializes CC1101/pin access: TX task, config.apply(), Transceiver::end(), frequency scan, and the RSSI read (try-lock only). apply()/end() flush pending jobs.
  • Web: queueCommand no longer waits for the command to run (semaphore removed, queue 32 deep). /setMyPosition and /unpairShade go through the command queue.
  • Limits: repeats capped at MAX_TX_REPEATS (40).
  • platformio.ini: CONFIG_ASYNC_TCP_RUNNING_CORE=0 on dual-core envs (esp32dev, esp32devdbg, esp32s3).

Behaviour changes

  • HTTP command responses return before the radio transmits. Updated state still arrives over the WebSocket.
  • loop() is still paused while a job transmits on its core (at most about 6 s per job with the cap). The 100 ms gap between jobs keeps the watchdog fed.

Testing

Draft: not yet built or tested on hardware.

  • Builds: esp32devdbg, esp32c3, esp32s3, esp32c6
  • Up/Down/My on a 56-bit and an 80-bit shade, tilt, set My position, unpair, sensor command, repeater
  • TX waveform compared on a logic analyser before/after
  • HA automation moving all covers repeatedly: no WebSocket reconnects
  • repeat=127 is capped; WebSocket stays up
  • Radio settings save and frequency scan during a long transmit

🤖 Generated with Claude Code

cjkas and others added 30 commits October 10, 2026 20:13
@cjkas @claude
fix: move radio TX to a dedicated task so HA WebSocket stays connected
9a9afa5 
Web command handlers ran on the async_tcp task and blocked up to 3s
waiting for loop() to bit-bang radio frames. async_tcp also services the
WebSocket on port 8080, so pongs stalled and Home Assistant
(ping_timeout=15) disconnected when several covers were commanded.

- Add a radio_tx FreeRTOS task with a job queue; sendFrame/repeatFrame
  and repeater frames enqueue instead of transmitting inline.
- Serialize CC1101 access with a recursive mutex (TX task, apply, end,
  frequency scan, RSSI read).
- Make Web::queueCommand fire-and-forget; drop the done semaphore.
- Route /setMyPosition and /unpairShade through the command queue.
- Clamp repeats to MAX_TX_REPEATS (40).
- Pin AsyncTCP to core 0 on dual-core targets.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: address radio TX task review findings
9d16d6c 
- Abort in-flight transmissions on flush so config save, frequency scan
  and OTA never wait on the radio lock for a whole job
- Suspend TX during frequency scan and OTA
- Refuse commands the radio queue cannot take before consuming a rolling
  code or moving simulated shade state; shift movement start when a job
  goes on air late
- Apply radio config changes under the radio lock
- Re-check receiver idle before repeater transmissions
- Return 503 from web handlers when the command queue is full
- Restore 200 ms settle delay before the set-My command
- Report RSSI 0 instead of a stale value when the radio is busy
- Encode each frame once and centralise the repeat clamp

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: make radio TX queue flush-safe and report job outcomes
a426efe 
- Serialize queueTx and flushTx so a queued job can't be lost to a flush.
- Settings changes preempt only the job in flight; the queue survives.
- Flush drains the queue and reports each job as dropped.
- Track suspend reasons (scan, OTA) separately.
- Report started/dropped for every job on a single event queue.
- Add priority jobs that go to the front and may use reserved slots.
- Remove the MAX_TX_REPEATS clamp.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: propagate radio TX result to callers instead of pre-checking
ad15eca 
- Send functions return whether the radio took the command.
- Shade, group and tilt state only change when the command was queued.
- Rolling code is committed only after the frame is queued.
- repeatFrame advances the repeat counter only on success.
- Remove canTransmit/canQueueTx; queueTx is the single authority.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: keep shade simulation in step with what goes on air
e55d9e4 
- Hold a shade's simulated movement until its command is on air.
- Send stops at the target as priority jobs.
- If a stop can't be sent or is dropped, simulate the run to the limit.
- Roll back targets when a movement command is dropped.
- Only store a favorite position when SETMY was queued.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: never leave radio TX suspended after a scan or aborted OTA
2a54740 
- endFrequencyScan always resumes TX; applying radio settings ends a scan.
- Frequency scans time out after 5 minutes.
- Reboot when an OTA upload is abandoned mid-transfer.
- /sendRemoteCommand returns 503 when the radio can't take the frame.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: hold simulation only for shades a command starts moving
130b5b6 
markTxPending ran inside sendCommandAfter before callers set the new
target, so isAtTarget() kept idle shades from ever being held, while
shades already moving were frozen although their motor kept running.

Hold a shade only when it was idle before the command and release the
hold when the command didn't set it moving.  The single mark in
sendCommandAfter now covers group members too, so the second pass in
SomfyGroup::sendCommand is gone.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: keep rolling codes in order when a stop is queued
44227bf 
A priority stop always went to the front of the radio queue, ahead of
an earlier frame of the same remote that was still waiting.  The older
rolling code then went out last and the motor rejected it.

Split "may use the reserved slots" (priority) from "goes to the front"
(front) and only send a stop to the front when its remote has nothing
queued at the back.  The radio task and flushes track which back jobs
have been taken so Transceiver::isQueued() can tell.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: cap repeats per radio job to keep loop under the WDT
2a45dba 
The radio task bit-bangs a whole job on loop()'s core at a higher
priority.  A job with 127+ repeats ran for 18-36s without loop() ever
running and the 15s task watchdog rebooted the device.

Limit a job to 50 repeats (about 7s) in encodeTxJob, which every send
path goes through, and advance the 80-bit repeat counter by the number
of frames actually queued.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: commit favorite only once SETMY is fully on air
352f36e 
The favorite was persisted as soon as the long My press was queued.  A
flush or radio reconfiguration could drop or cut it short and the app
kept a favorite the motor never stored.

The radio task now reports finished/aborted for every job.  Setting a
favorite goes through sendSetMy, which commits it on finished and puts
the previous one back otherwise.  The new favorite is set before the
My is sent so the simulation no longer heads for the old one.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: swap restored radio config under the radio lock
00a0840 
Restoring a backup wrote the transceiver config in place from the async
task while the radio task could be transmitting with it.

Read the record into a copy and swap it in through the new
Transceiver::setConfig, which Transceiver::fromJSON now uses as well.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: back-pressure web commands and report radio failures
1f394ca 
Web commands are answered once queued, but the main loop handed all of
them to the radio at once and anything past its free slots was dropped
silently.  Unpair also cleared paired even when the Prog wasn't sent.

- processQueue leaves commands queued until the radio has room.
- Commands that transmit get a 503 straight away during a frequency
  scan instead of being dropped later.
- Unpair restores paired when the Prog can't be sent.
- Failed web commands are logged.
- Note that /setPositions and /setSensor answer with the state before
  the command; the socket events carry the result.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
test: add native Unity env with Utils tests
52c67b1 
Adds an [env:native] for host-side unit tests (pio test -e native) with
a minimal Arduino.h stub, and covers the Utils helpers: _trim,
toBoolean, SETCHARPROP, sortArray and the Timestamp conversions.

The firmware envs now extend a shared [esp32] section instead of [env]
so the native env doesn't inherit the Arduino framework, the ESP32
libraries or the build scripts.

The half-hour calcTZOffset cases fail: +05:30 comes back as 2300.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
ci: run native unit tests before firmware builds
aa23b46 
Adds a "Unit tests" job running pio test -e native on ubuntu-latest.
The firmware build matrix waits for it, so a failing test also stops
release uploads.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: report half-hour time zones correctly in ISO timestamps
02df2da 
calcTZOffset added the leftover seconds instead of minutes to the hhmm
offset, so +05:30 came out as 2300 and ISO timestamps showed +2300.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
refactor: move the Somfy frame codec into SomfyFrame.cpp
bb16c90 
The command and protocol enums, somfy_frame_t and its encode/decode
methods, and both translateSomfyCommand overloads move verbatim out of
Somfy.cpp/Somfy.h, which now include SomfyFrame.h.  The codec doesn't
depend on the radio, storage or network, so on its own it can be unit
tested on the PC.

decodeFrame(somfy_rx_t*) stays in Somfy.cpp since it reads the RSSI
from the CC1101.  No behaviour change.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
test: cover the frame codec and command names
3774aba 
Builds SomfyFrame.cpp in the native env, with a minimal String and
no-op ESP_LOGx in the stubs, and tests:

- a fixed 56-bit frame, byte for byte, so what goes on air is pinned
- encode/decode round trips for 56-bit, 80-bit (repeats, Favorite,
  Stop, steps, Toggle), RTW and RTV frames
- rejection of bad checksums, addresses, rolling codes, keys and
  80-bit parity
- repeat/synonym tracking in copy()
- parsing and naming of commands

The command-name round trip fails: names the firmware reports, such as
"Step Up" and "My+Up", parse back to the wrong command.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: parse the command names the firmware reports
60419a3 
shadeCommand events (and the Home Assistant events built from them)
name commands "Step Up", "Sun Flag", "My+Up" and so on, but parsing
those names fell through to the prefix shortcuts: "Step Up" became
SunFlag, "Sun Flag" became StepUp and "My+Up" became My.  Match the
reported names exactly before the shortcuts.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
refactor: move JsonFormatter into JsonFormatter.cpp
d723c10 
JsonFormatter and BufferedJsonFormatter move verbatim out of
WResp.cpp/WResp.h, which now include JsonFormatter.h.  They don't need
the web server, so on their own they can be unit tested on the PC; the
native env builds the new file.  AsyncJsonResp and JsonSockEvent stay
in WResp.  No behaviour change.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
test: cover the JSON formatter
48676b0 
Tests BufferedJsonFormatter output: nesting and commas, number and bool
values, skipped null strings and escaping.  A canary after the buffer
catches writes past its end.

Three fail: control characters other than \b\f\n\r\t aren't escaped,
calcEscapedLength skips the first character, and so a value starting
with an escaped character can be written one byte past the buffer.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: escape control characters and count JSON string lengths correctly
b2294ce 
calcEscapedLength looped from strlen(raw) down to 1, counting the
terminator instead of the first character.  A value starting with a
character that needs escaping was counted one byte short, so the fit
check could let it be written one byte past the buffer.  Control
characters other than \b\f\n\r\t were copied raw, which isn't valid
JSON.

Both now go through one escapeChar helper, which also escapes the
remaining control characters as \u00XX.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: never write past a JSON buffer or send a broken socket event
8881289 
When something didn't fit, the formatters skipped it silently, closing
brackets included, and clients got malformed JSON.  Two writes could
also go past the buffer:

- AsyncJsonResp flushed when full and then wrote the value without
  checking again, so a value bigger than the whole buffer overflowed.
  It is now streamed straight out instead.
- JsonSockEvent::closeEvent added the closing ] when only one byte was
  left, putting the terminator one byte past the buffer.

The formatters now set an overflowed() flag and log an error, and a
socket event that overflowed is logged and not sent.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
refactor: share the shade travel maths in ShadeMotion.cpp
3a8736f 
checkMovement worked out the position four times, nearly copy-pasted,
for lift down, lift up, tilt down and tilt up.  All four now call
travelPosition(), which keeps the same arithmetic, including how the
elapsed time wraps with millis().  Tilt up with no tilt time still
stops the tilt direction in the caller.

transformPosition and transformCommand keep their behaviour through
the new displayPosition() and flipCommand() helpers.  The native env
builds ShadeMotion.cpp so all of this can be unit tested on the PC.
No behaviour change.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
test: cover shade travel and position/command flipping
1d1f685 
Tests travelPosition (part-way positions both ways, separate up and
down times, stopping at the ends, zero travel time, steady progress
and staying within 0-100), displayPosition with and without flipping,
and flipCommand.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: send wifiStrength only for real signal changes
05c3d30 
At -70 dBm the RSSI moves a dBm or two constantly, so the >1 dBm check
every 1.5s sent wifiStrength to every client about 270 times an hour.
That is extra traffic on an already lossy link and a state write in
Home Assistant for each one.

Broadcast on a channel change, or for a change of at least 3 dBm no
more than every 10s. A newly connected client still gets the current
value straight away.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
feat: log timings for slow work, socket clients and boot crashes
dd47c04 
Add Timing.h and use it to log what holds up the main loop, how long a
new socket client's setup takes, slow socket sends, and how long each
web request took and from which client. Socket connects and disconnects
are logged with how long the client stayed, and late pings as warnings.
Routine sends and pings are not logged, as each line costs several ms
on a 115200 baud console.

Report a stored crash once with the crashed and running build IDs, then
erase it, since it was reported on every boot and its backtrace only
decodes against the build that crashed. Mount the file system once.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: print each log line whole
6b19d24 
The Arduino core's log_printf() has its UART lock commented out and
writes with ets_printf(), so lines logged at the same time from
async_tcp and loop() came out interleaved character by character. It
also formats short lines into one static buffer shared by every task.

Wrap log_printf at link time and hold a mutex around the core's
log_printfv(). It never blocks before the scheduler runs, in an ISR or
in a critical section, and waits at most 50ms before printing anyway.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: report the real flipPosition for linked shades
f1eac31 
serializeShadeRef sent flipCommands as flipPosition, so group linked
shades and the /shadeCommand reply showed the wrong setting.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
feat: log AP signal levels and roam for a 6 dB gain
bd7ed17 
The roaming scan found two APs for the SSID every 10 minutes but the log
never showed their signal, so there was no way to tell whether the
device at -70 dBm had a better AP to move to. Log each AP on the
roaming scans, show roams at info level, and switch for a 6 dB gain
instead of 10.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: broadcast memStatus every 30s
66d264e 
It went out whenever free heap moved by 1.5 KB and 7s had passed,
which with normal traffic was about 240 times an hour. Send it every
30s instead. It is also the one event always sent, so 30s keeps Home
Assistant well inside its 60s receive timeout. Per-client sends and
the web UI's room are unchanged.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
cjkas and others added 23 commits October 11, 2026 20:50
@cjkas @claude
feat: log where /discovery spends its time
0610dc4 
The /discovery handler takes about 200ms on the async_tcp task on every
call with no obvious cause. When it takes 50ms or more, log how long
the header, memory stats, rooms, shades, groups and sending took.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: hook the log lock into log_printfv
b6ed24f 
The Arduino core already links with --wrap=log_printf for the ESP
Insights hook in libesp_diagnostics, so defining __wrap_log_printf
failed to link with a duplicate definition. That hook prints through
log_printfv, so wrap that instead and take the lock there.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: stop warning about late pings
55b1db4 
Home Assistant's aiohttp client only pings after 20s without a message
from the device, so with memStatus every 30s pings arrive every 30-60s
and every one was logged as late. The gap depends on our own traffic,
not the link, so log pings at debug level and keep the time of the
last one for the disconnect line.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: track the JSON length instead of recounting it
a8bb377 
Every append found the end of the buffer with strlen(), up to four
times per append and four appends per field, so building a response
cost the square of its size. /discovery fills a 4 KB buffer and spent
about 195 ms of its ~210 ms there (shades 126 ms, groups 69 ms), all on
the async_tcp task that serves every socket and web request.

Keep the length in the formatter and write every append at the end
through one helper. The web response and socket event subclasses and
ControllerChunker use the tracked length too. Output is unchanged.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: stop the log lock lifting loop() above the radio task
6cec411 
The log lock was a mutex, so when async_tcp waited on it loop() inherited
async_tcp's priority and could preempt the radio task mid-frame.  Use a
binary semaphore, which has no priority inheritance, track the holder so
a nested log doesn't wait on itself, and cap the wait at 20ms.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: log only timings of 100ms or more by default
77d3e18 
Every HTTP request logged two timing lines and every socket connect five,
all from the async_tcp task, at several ms each on a 115200 baud console.
Raise the default TIMING_LOG_MIN_MS to 100 and drop the unused ScopeTimer.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: keep the core dump in flash after reporting it
1cc55e0 
Erasing the dump on the first boot after a crash lost it on headless
devices where nobody saw the serial summary.  Leave it until the next
crash replaces it and use the reset reason to tell a fresh crash (WARN)
from one reported on an earlier boot (INFO).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: let a priority stop overtake a delayed radio job
03eb3b3 
The radio task took a delayed job (the My after a shade reaches its
target) off the queue and slept until it was due, so a stop queued at the
front waited for the sleep plus the whole delayed job.  Wait on a task
notification instead and send front jobs while the delayed one is held.

A held job still counts as queued for isQueued(), so a stop for the same
remote goes behind it and the rolling codes stay in order.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: leave out /controller items that overflow instead of breaking th…
ab0eba2 
…e JSON

The formatter drops a value that doesn't fit but keeps appending the
commas and braces after it, so an oversized shade or group streamed
invalid JSON.  Check overflowed() after each unit, log the item and skip
it, and keep producing until a unit or the end so a skip never ends the
stream early.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: don't hold position updates back while the radio is full
a49fa5d 
processQueue stopped draining web commands whenever the radio queue was
near full, including set_positions, which never transmits.  Look at the
command at the head and only hold it back if it needs the radio, keeping
the commands in order.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: read the rolling code from NVS only until it has been written
08d2cd7 
Splitting getNextRollingCode into peek and commit opened the ShadeCodes
namespace twice per command.  Once a code has been written for an
address lastRollingCode matches NVS, so peek uses it and only commit
opens NVS.  Changing the address drops back to reading NVS.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: send memStatus once per network broadcast
b4d480c 
emitSockets() sent memStatus and then emitSockets(255) sent it again, so
room 0 got two identical events in one pass after every /discovery.
Only the per-client path sends it from emitSockets(num).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
refactor: share shade remote matching in SomfyShade::findRemote
2ee7af5 
markTxPending copied the shade and linked-remote address matching from
SomfyShade::processFrame.  Both now use findRemote() so they can't drift.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: compare the coredump build SHA without reading past the array
e1c76bc 
app_elf_sha256 has no terminator (IDF 4.4 fills all 17 bytes), so strcmp
never matched the running build and %s printed whatever followed it.  Copy
it into a terminated buffer first.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: keep a repeater frame until the radio queue takes it
7e9687c 
The frame was popped before queueTx(), so whenever the radio queue was busy
it was lost.  Peek it instead, only remove it once it is queued (or TX is
suspended), and don't try while the queue has no room.  The repeater buffer
moves to TxBuffer.h/.cpp so it can be unit tested.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: send frequency scan events from loop() only
a212d56 
beginFrequencyScan, endFrequencyScan and apply() run on async_tcp from the
web handlers and emitted into the socket event buffer loop() also writes,
so an event could be mixed up or dropped.  They now only flag the change
and Transceiver::loop() sends it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
refactor: pin which radio job holds a shade's simulated movement
0ba5260 
Review flagged the outer My/Favorite losing to the nested Up/Down it queues.
Every such nested send is from a simMy shade, where the Up/Down is what
starts the motor, so the behaviour stays.  The hold moves into TxHold in
ShadeMotion with a comment explaining the order and unit tests pinning it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: set the radio up for a new config before anything transmits on it
c4472ba 
setConfig() swapped the config in and released the radio lock, and apply()
only ran later from save(), so the radio task could send queued jobs with
the new TX pin and settings on a CC1101 that wasn't set up for them.
setConfig() now swaps, saves and applies under one lock and save() goes.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: only let wifiStrength broadcasts move the RSSI baseline
e47381b 
The emit to a newly connected client also stored the RSSI and time, which
restarted the 10s hold-off and hid a real change from the clients already
connected.  The baseline moves into RssiBroadcast, which ignores
per-client emits, with unit tests.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: don't wait on a log lock holder that can't run
9988a2b 
The log lock is a binary semaphore, so when the radio task preempted loop()
mid-log every log from async_tcp waited the full 20ms for a holder that
couldn't let go until the radio job ended.  Waiting now stops as soon as
the holder is preempted on another core.  The rule lives in
LogLockPolicy.h with unit tests.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
fix: skip formatting request timings that won't be logged
6a37694 
timeRequest built both 112-byte descriptions (with a String copy of the URL)
on async_tcp for every request, and logElapsed threw them away for anything
under 100ms.  Check worthLogging() first.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
test: cover the radio queue rules, shade TX state, rolling codes and …
1962778 
…web command hold

Move the pieces of the radio TX path that don't need FreeRTOS or NVS out of
Somfy.cpp and Web.cpp so they build in the native test env: the TX job and
event types and job encoding (TxBuffer), the queue rules (TxQueuePolicy.h),
a shade's pending radio jobs (ShadeTxState), the next rolling code
(RollingCode.h) and the web command queue types (WebCommand.h).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas @claude
refactor: let one RadioScheduler own the radio job queue
4135f21 
The order radio jobs go out in was spread across callers, file statics and
the radio task: each remote tracked its last back job so a stop could decide
whether it may jump the queue, and the task kept a separate count of the back
jobs it had taken.  RadioScheduler now owns the waiting jobs, their tickets,
the stop-goes-first rule, delayed jobs, flush and suspend, and the outcome of
every job.  It is plain C++, with the time passed in and the caller locking,
so all of it is covered by native tests.  Transceiver keeps the FreeRTOS task,
the locks and the CC1101.

sendCommandAfter returns the job's ticket instead of leaving it in lastTxSeq.

Two stops for the same remote now go out in the order they were sent; queued
to the front they used to come out reversed, putting the rolling codes out of
order.

Adds GLOSSARY.md for the radio terms.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cjkas
cjkas marked this pull request as ready for review October 12, 2026 19:00
@cjkas
cjkas merged commit 125a758 into main Oct 12, 2026
5 checks passed
@cjkas
cjkas deleted the fix/radio-tx-task branch October 18, 2026 11:31
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

1 participant

@cjkas