Gravis UltraSound emulation

[Copilot]

Description of Changes

Adds Gravis Ultrasound sound card emulation to Spice86.

Rationale Behind Changes

The Gravis UltraSound (GUS) sound card is used by real-mode DOS games and demos.

Suggested Testing Steps

Gravis Ultrasound Tests should pass

• Fixes Feature request: Gravis Ultrasound sound card emulation #319

[maximilien-noal]

@copilot I want the full gravis ultrasound, not a skeleton.

https://github.com/dosbox-staging/dosbox-staging see gus.cpp and related files

[Copilot]

Copilot wasn't able to review any files in this pull request.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[Copilot]

Pull request overview

Copilot reviewed 8 out of 8 changed files in this pull request and generated 1 comment.

Suppressed comments (3)

src/Spice86.Core/Emulator/Devices/Sound/GravisUltraSound.cs:701

• CheckVoiceIrqs sets IrqVolStateBit/IrqWaveStateBit based on the full _voiceIrq bitmasks, but totalMask is computed using activeVoiceMask. If a program raises IRQ bits on an inactive voice index (>= _activeVoices), the status byte can incorrectly report a pending wave/vol IRQ. Mask the individual wave/vol bitmasks before setting the status bits.

        if (_voiceIrq.VolState != 0) {
            _irqStatus |= IrqVolStateBit;
        }
        if (_voiceIrq.WaveState != 0) {
            _irqStatus |= IrqWaveStateBit;

src/Spice86.Core/Emulator/Devices/Sound/GusVoice.cs:77

• RenderFrames' early-return checks the intersection of CtrlDisabled bits across WaveState and VolState. If WaveState and VolState are both disabled but with different disabled bits set (e.g., Reset vs Stopped), rendering will incorrectly proceed. This should instead check each control state independently for any disabled bit.

    public void RenderFrames(byte[] ram, float[] volScalars, AudioFrame[] panScalars, AudioFrame[] frames, int count) {
        // Skip rendering only when BOTH wave AND vol controls are independently disabled,
        // matching the GF1 hardware behaviour documented in the UltraSound SDK.
        if ((WaveState & VolState & CtrlDisabled) != 0) {
            return;

src/Spice86.Core/Emulator/Devices/Sound/GravisUltraSound.cs:193

• This PR adds a new hardware device with a non-trivial port map, IRQ, and DMA behavior, but there are no integration tests validating GUS detection/probing or basic IRQ/DMA flows. There is already a sound integration test suite (e.g., SoundIntegrationTests) for SoundBlaster; a similar test that runs a small COM fixture which probes 0x240 and/or checks ULTRASND-related behavior would help prevent regressions and ensure the original crash scenario is covered.

        InitPortHandlers(ioPortDispatcher, portBase);

        // Mixer setup
        mixer.RegisterQueueNotifier(this);
        mixer.LockMixerThread();

[Copilot]

🟡 Changes recommended

The GUS timer implementation is currently one-shot (does not reschedule), which breaks expected periodic timer behavior and can impact real driver/program operation.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Review details

• Files reviewed: 9/9 changed files
• Comments generated: 3
• Review effort level: Lite

[Copilot]

🔵 Needs a closer look

It introduces a large new hardware emulation device and scheduler/DMA/timer logic, and the review found correctness and performance issues that should be addressed and re-validated with broader testing.

Review details

• Files reviewed: 15/15 changed files
• Comments generated: 3
• Review effort level: Lite