## Summary
Adds a unified mechanism for locating third-party IO DLLs through environment
variables, on top of the existing per-game `[xxxio] path=` INI setting.
## Changes
For each IO DLL, the path is resolved in this order:
1. The section-specific INI value, e.g. `[mai2io] path=`.
2. The section-specific environment variable, e.g. `SEGATOOLS_MAI2IO_PATH`.
3. `SEGATOOLS_IO_ROOT` joined with the default IO DLL filename, e.g.
`%SEGATOOLS_IO_ROOT%\mai2io.dll`.
4. Empty path, which keeps the built-in IO implementation (unchanged default).
INI and environment-variable paths are run through Windows environment-variable
expansion, so values such as `%SEGATOOLS_IO_ROOT%\mai2io.dll` are valid in the
INI directly.
Reviewed-on: https://gitea.tendokyu.moe/TeamTofuShop/segatools/pulls/119
Co-authored-by: グローランプ <130208311+Gl0w1amp@users.noreply.github.com>
Co-committed-by: グローランプ <130208311+Gl0w1amp@users.noreply.github.com>
## Overview
This PR addresses the severe memory leak and performance issues observed in `mai2`, while also introducing improvements to the touch emulation logic.
## Root Cause Analysis
As discussed previously, the root cause of the `mai2` memory leak is not a global `segatools` buffer bug. Instead, the game aggressively spams overlapped empty reads specifically on the `LED 15070` UART. On real hardware, the serial driver naturally throttles this. Under emulation, without a throttle, it hits approximately **260kHz of empty async reads**, which causes the memory usage to explode.
## The Fix
Instead of introducing complex locking mechanisms and condition variables globally in `uart.c`, this PR applies a targeted fix:
* Added a local `Sleep(1)` directly in `common/board/led15070.c` to rate-limit empty reads on the LED path.
* Because this is isolated to LED communications, it completely resolves the memory leak without introducing any lag, livelocks, or overhead to other critical inputs.
## Additional Changes in this PR
Alongside the memory leak fix, this PR includes a few touch-related improvements (as touch emulation was reviewed during the debugging process):
* Enhanced touch input handling and improved auto-scan state management.
* Implemented IOCTL handling for touch input to properly manage communication status.
## Testing
* **mai2:** Tested successfully on multiple machines. The memory leak is completely gone, and the game runs smoothly.
* **chusan:** Tested to ensure no regressions. Sliders and inputs work flawlessly without the lag.
Reviewed-on: https://gitea.tendokyu.moe/TeamTofuShop/segatools/pulls/101
Co-authored-by: Gl0w1amp <gl0w1amp@noreply.gitea.tendokyu.moe>
Co-committed-by: Gl0w1amp <gl0w1amp@noreply.gitea.tendokyu.moe>
This adds full support for the Taisen series of games, namely Sangokushi Taisen and Eiketsu Taisen.
Games added:
* Sangokushi Taisen (SDDD)
* Eiketsu Taisen (SDGY)
Devices added:
* CHC-320 printer (SGT)
* "Printer camera" (SGT, unsure what this actually really is)
* CX-7000 printer (EKT)
* Y3CR BD SIE F720MM (SGT, EKT)
Notable changes in the codebase:
* Renamed everything printer specific to seperate between CHC and CX.
* Many new function and registry hooks were added across the board.
* An error is now logged when segatools.ini (or the path in `SEGATOOLS_CONFIG_PATH`) cannot be found.
* Netenv now redirects UDP broadcasts targeted at the subnet that is specified in the keychip configuration. The terminal announces it's presence by broadcasting UDP to 192.168.189.255, this will be redirected to 255.255.255.255.
* Vfs now seperates between absolute and relative paths in `vfs_fixup_path` via an environment variable called `SEGATOOLS_VFS_RELATIVE_PATH`. This is needed because amcapture accesses files as workingdirectory-relative.
* The Y3 board emulation has support for external Y3 I/O dlls. The default implementation (y3ws) that comes with this is a websocket implementation. The docs are available under `doc\y3ws.txt` and a sample card player .html file is under `dist\ekt\card_player.html`. I already know one person that is hosting a massively improved version of it.
* For websockets, my own websocket implementation is used as a subproject (MIT license): https://github.com/akechi-haruka/cwinwebsocket
* For JSON, cJSON was embedded (MIT license): https://github.com/DaveGamble/cJSON
* y3ws reads all printed cards from `DEVICE\print` by default including card back sides. It's up to the client to merge or skip them.
Remarks:
* SGT takes ~8 minutes to load. This seems to be intentional.
* SGT uses some weird TCP network implementation like IDZ. I have not bothered reversing that yet and I have confirmed everything working from the test menu.
* EKT will throw a network error if no terminal is found. You must run the terminal on another computer to be able to launch the satellite.
* EKT has a very bizzare speed glitch that will speed up the ingame unit movement by several 1000% and also slows down cutscene animations by 90%. When this effect is active, you also take a ton more damage than usual. I do not know what causes this and it seems PC specific.
* EKT is very stutter sensitive and will throw error 6401 (I/O timeout) at random when trying to alt+tab or have other things running.
* EKT features a livestream system called Enbu (or "Dojo Upload"). While you are in a match, regardless of vs. AI or another player, the game will record your screen and live-stream it to the Enbu server as defined by the game server's startup response. The application responsible for that, "AM Capture" will not limit it's recording to the game window, but also anything that overlays the game window (notifications, popups, alt+tabbed windows, web browsers, etc). Since this is live-streamed, killing the process will have no effect afterwards, as the frames showing unwanted things will already have been transmitted. To make people aware of this, a one-time dialog message will pop up when starting EKT. The flag for that is stored in the DEVICE folder.
Closes#25.
Co-authored-by: Dniel97 <Dniel97@noreply.gitea.tendokyu.moe>
Reviewed-on: https://gitea.tendokyu.moe/TeamTofuShop/segatools/pulls/85
Co-authored-by: kyoubate-haruka <46010460+kyoubate-haruka@users.noreply.github.com>
Co-committed-by: kyoubate-haruka <46010460+kyoubate-haruka@users.noreply.github.com>
This PR implements the missing NFC, FeliCa, and Mifare command handlers in `aimeio`. This expands the reader's capability to support games and tools that utilize these specific radio functions.
**Changes:**
* Implemented NFC radio control functions.
* Added MIFARE and FeliCa command handling.
* Updated `aimeio` DLL exports and hook definitions.
**Testing Status:**
* I have verified the implementation using a virtual `aimeio` test environment, and the command handling works as expected.
But I currently do not have access to a physical card reader, so haven't been able to verify the behavior on actual hardware yet.
Reviewed-on: https://gitea.tendokyu.moe/TeamTofuShop/segatools/pulls/94
Co-authored-by: グローランプ <130208311+Gl0w1amp@users.noreply.github.com>
Co-committed-by: グローランプ <130208311+Gl0w1amp@users.noreply.github.com>
## Summary
- feat(aimeio): add VFD text forwarding API and wire through DLL/VFD backend
- feat(aimeio): add VFD state callback and export symbols across hooks
- refactor(aimeio): rename VFD callbacks and align VFD parsing
- feat(aimeio): extend VFD state fields for richer reporting
## Description
This PR expands the AIME IO VFD interface by adding text forwarding and a VFD state callback, then wires them through the AIME DLL and VFD backend.
It also renames VFD callbacks to align parsing behavior, updates all hook exports accordingly, and extends VFD state fields to carry more detailed state data.
Reviewed-on: https://gitea.tendokyu.moe/TeamTofuShop/segatools/pulls/90
Co-authored-by: グローランプ <130208311+Gl0w1amp@users.noreply.github.com>
Co-committed-by: グローランプ <130208311+Gl0w1amp@users.noreply.github.com>
## Summary
- feat(io4): implement PWM/GPIO support for Billboard and Code Reader lights
- feat(mai2io): expose LED control APIs to support new io4 features
- fix(led15070): correct multi-LED color logic and add fade support
## Description
**io4 & mai2io Updates**
Introduced PWM control and GPIO write support in `io4`. These additions allow `mai2io` to properly drive the **Billboard LEDs** and **Code Reader / Player Camera lights**. The APIs have been exposed through `mai2hook` to facilitate scriptable control of these peripherals.
**led15070 Improvements**
Fixed an issue with the multi-LED color changing logic in the `led15070` driver. This ensures correct color output sequence and includes improvements for fade calculation.
Reviewed-on: https://gitea.tendokyu.moe/TeamTofuShop/segatools/pulls/89
Co-authored-by: グローランプ <130208311+Gl0w1amp@users.noreply.github.com>
Co-committed-by: グローランプ <130208311+Gl0w1amp@users.noreply.github.com>
This adds a setting to the global configuration to allow only input when the game is focused. This prevents issues, such as triggering game buttons or service buttons while being tabbed out. Turned off by default.
The implementation is not the best, but in the context how these games are structured, the most acceptable.
It will keep scanning if the foreground window title fully matches (some games partially match, like FGO because due to the revision number in the title), and if found, will grab the HWND of that window and compare that from then on, to not tank performance due to constant string operations.
Tested with FGO, chusan and ongeki.
Reviewed-on: https://gitea.tendokyu.moe/TeamTofuShop/segatools/pulls/83
Reviewed-by: Dniel97 <dniel97@noreply.gitea.tendokyu.moe>
Co-authored-by: kyoubate-haruka <46010460+kyoubate-haruka@users.noreply.github.com>
Co-committed-by: kyoubate-haruka <46010460+kyoubate-haruka@users.noreply.github.com>
グローランプ