aiden/h8-536-decoder
1
0
Fork 0
Code Activity

traces

Browse Source
This commit is contained in:
Aiden
2026-05-27 11:50:10 +10:00
parent 0d099235c5
commit c0304c575c
55 changed files with 26035 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files

83
docs/pt2-menu-state-machine.md
View File

@@ -102,6 +102,52 @@ Relevant ROM behavior:
- The FRT timer path decrements `F726`; when it expires, it clears `F713.6` and `F711.4-F711.7`.
- The OTHERS/COPY branch sets `F711.7` and `F726=H'64` to keep the local key/display state alive briefly.
## OTHERS Action Latch Trace
The missing OTHERS mechanism appears to be `F770`, not another simple command-0 selector.
Root OTHERS/SHUTTER handler `H'6EE4`:
- Reads `F770`, clears it, and only keeps low bits `0x01/0x02`.
- If the first action path is present and `E400[0x008F].11` is enabled, writes `E800[0x008F]=0x0800` and queues selector `0x008F`.
- If the second action path is present and `E400[0x008F].12` is enabled, writes `E800[0x008F]=0x1000` and queues selector `0x008F`.
COPY TO SLAVES handler `H'6FF0`:
- Reads `F770`, clears it, and only keeps bit `0x04`.
- If `F770.2` is clear, it only displays the OTHERS/COPY page.
- If `F770.2` is set and `F791.7` is set, it enters the local copy-start branch.
- If `F770.2` is set and `F791.7` is clear, it diverts to `SET RCP` / `MASTER` and stages restore selector `F734=0x0101`.
Static direct writers found so far only write high bits into `F770`:
| Writer family | Value | Meaning candidate |
| --- | --- | --- |
| `H'4394` | `0x80` | value/dial page redraw or change latch |
| `H'4457` | `0x40` | value/dial page redraw or change latch |
| `H'451A` | `0x20` | value/dial page redraw or change latch |
No decoded direct instruction has been found that writes `F770=0x01`, `0x02`, or `0x04`.
Updated interpretation:
- `E000[0x008F]` and `E400[0x0015]` can make OTHERS-adjacent UI state visible, but they do not synthesize the local OTHERS action.
- The physical OTHERS/COPY/menu action likely arrives through a separate local key latch or indirect panel-chip path that produces the low `F770` bits only when the page machinery is in the right context.
- Pressing OTHERS under the generic queued-report gate can therefore legitimately produce no new serial report.
## Adjacent F109 Menu/Shutter Key Trace
A focused trace of the `F109 -> F6D0` local-key handlers found a neighboring menu/shutter cluster. These are useful context because they share the same local-page style, but they do not directly prove the OTHERS low-bit latch.
| Source bit | Handler | Selector/page effect |
| --- | --- | --- |
| `F6D0.7` | `H'24E8` | Writes `E800[0x008F]=0x8000` or `0x0000`; if `E000[0x0088].14` is set, diverts to timed page `F732=0x1C01` |
| `F6D0.6` | `H'252E` | Writes `E800[0x008F]=0x2000` or `0x0000`; if `E000[0x0088].14` is set, diverts to timed page `F732=0x1C01` |
| `F6D0.4/F6D0.5` | `H'2574/H'25D4` | Uses selectors `0x0091/0x0092` and modal byte `F6F6`; if `E000[0x0088].14` is set, diverts to timed page `F732=0x1C01` |
| `F6D0.1/F6D0.2/F6D0.3` | `H'2390/H'2408/H'24A9` | Uses selector `0x0083`; if `E000[0x0088].15` is set, diverts to timed page `F732=0x1C01` |
`F732=0x1C01` is in the DETAIL/KNEE neighborhood seen in earlier traces, not the OTHERS page `0x01`. This suggests selector `0x0088` can retarget adjacent local keys into a timed menu overlay, while OTHERS itself still depends on the page-1 wrapper and the `F770` low-bit action latch.
## Bench Implications
To make the local COPY path available from the panel, the fake CCU probably needs to:
@@ -191,3 +237,40 @@ Manual correlation:
- The same RCP-TX7 manual lists `EVS/ECS` under OTHERS for DXC-D30/D30P normal settings.
- A later CCU/RCP manual states that when EVS is on, `EVS` is displayed; when the shutter switch is off, `OFF` is displayed.
- Therefore `E000[0x008F].11` is best labeled `shutter_evs_display_or_mode`, and `E000[0x008F].12` is best labeled `shutter_off_display_or_mode` until ROM traces split display-only status from actual camera setting.
## Bench Observation: OTHERS Common-Gate Button Press
After the queued-report button tests found BARS and IRIS AUTO reports, the physical OTHERS button was tested with the same common queue-service gate.
Result:
- `button-common-OTHERS-result.json` matched `button-common-baseline-result.json`.
- No extra ACK-target frames appeared.
- The queued selector stream still included baseline selector `0x0015 = 0x0000`, but OTHERS did not turn that into a visible/new report.
Interpretation:
- OTHERS does not behave like BARS or IRIS AUTO under the simple common queue-service gate.
- This supports the ROM model that OTHERS is a local menu/display key whose useful effect depends on additional page/menu state, not just the generic report queue being serviced.
- The next fair comparison needs an OTHERS-specific no-button baseline because the OTHERS seeds themselves alter display/report state.
New focused scenarios:
```powershell
.\.venv\Scripts\python.exe scripts\serial_scenario.py scenarios\others-menu-gated-baseline.json --parity E --quiet-console --log captures\others-menu-gated-baseline.txt --result-json captures\others-menu-gated-baseline-result.json
.\.venv\Scripts\python.exe scripts\serial_scenario.py scenarios\others-menu-gated-press.json --parity E --quiet-console --log captures\others-menu-gated-press.txt --result-json captures\others-menu-gated-press-result.json
.\.venv\Scripts\python.exe scripts\serial_scenario_compare.py captures\others-menu-gated-baseline-result.json captures\others-menu-gated-press-result.json --show-labels
```
Result of the first OTHERS-gated comparison:
- The current baseline rerun overwrote the accidental first baseline capture; `serial_scenario.py` writes logs/results fresh when explicit paths are reused.
- `others-menu-gated-baseline-result.json` and `others-menu-gated-press-result.json` were identical at the ACK-target level.
- Both had `rx_frames=153`, `ack_sent=64`, and no label-count increase in the press run.
- The seeded state was visible in the queued stream as `0x008F = 0x1800`, but pressing OTHERS still did not create an extra serial report.
Updated interpretation:
- The simple common queue gate is not enough for OTHERS.
- The current OTHERS-adjacent seeds are also not enough for OTHERS to emit a report.
- OTHERS is increasingly likely to be a local menu/page navigation key whose effect is visible only if the ROM has selected the right page/context (`F732/F733/F72C/F731/F770`), rather than a host-visible selector-report button like BARS or IRIS AUTO.