Open-IOV: initial public release

Open-IOV virtualizes NVIDIA GPUs so that unmodified guest drivers run with
real GPU functionality - CUDA first, 3D and video to follow - on unmodified
hosts, without a proprietary vGPU license and without patching any driver,
firmware or hypervisor.

Core principle: emulate the interface, pass through the work.

  kernel/     open-iov-vfio.ko - host VFIO mdev driver: virtual GPU device
              (PCI config space, BAR shadows, MSI-X), GSP boot emulation,
              GSP-RM message-queue/RPC emulation, memory model (GPU-VA to
              guest-PA discovery) and the semaphore-completion service that
              advances guest UVM progress.
  userspace/  libiov.so management API and examples.
  tools/      guest diagnostics, firmware analysis wrappers, RM probes.
  scripts/    stack manager, VM harness, build/load/test automation.
  config/     hardware/version profiles (the portability surface).
  docs/       documentation tree - start at docs/README.md.
  wiki/       long-form firmware and protocol deep dives.

Status: initialization works end to end for a stock 570.144 Linux guest on
Turing (TU104): driver probe, nvidia-smi, the full GSP-RM control battery,
object creation and the early CUDA/UVM start-up path. GPU work forwarding -
the step that makes CUDA kernels execute on real hardware - is the active
work item; see docs/04-implementation/01-known-knowns-and-unknowns.md.

License: BSD-3-Clause for the project; GPL-2.0 for kernel modules that link
GPL-only kernel interfaces. See LICENSE and docs/09-reference/legal-and-licensing.md.

1 files changed, 220 insertions, 0 deletions

diff --git a/scripts/windows_guest_automation.sh b/scripts/windows_guest_automation.sh
new file mode 100755
index 0000000..f95471e
--- /dev/null
+++ b/scripts/windows_guest_automation.sh
@@ -0,0 +1,220 @@
+#!/bin/bash
+# windows_guest_automation.sh - Windows guest test automation via WinRM/SSH
+
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+BASE_DIR="$(cd "$SCRIPT_DIR/.." && pwd)"
+
+RESULTS_DIR="/tmp/gsp_emu_windows/$(date +%Y%m%d_%H%M%S)"
+mkdir -p "$RESULTS_DIR"
+
+log_info() { echo -e "\033[0;34m[INFO]\033[0m $1"; }
+log_ok() { echo -e "\033[0;32m[OK]\033[0m $1"; }
+log_error() { echo -e "\033[0;31m[ERROR]\033[0m $1"; }
+
+# Configuration
+WIN_USER="${WIN_USER:-Administrator}"
+WIN_IP="${WIN_IP:-192.168.122.100}"
+WIN_PASSWORD="${WIN_PASSWORD:-}"
+WINRM_PORT="${WINRM_PORT:-5985}"
+SSH_PORT="${SSH_PORT:-22}"
+
+check_winrm() {
+    log_info "Checking WinRM connectivity to $WIN_IP..."
+    if command -v winrm &>/dev/null; then
+        winrm identify -r:"http://$WIN_IP:$WINRM_PORT/wsman" -u:"$WIN_USER" -p:"$WIN_PASSWORD" -auth:basic 2>/dev/null
+    elif command -v python3 &>/dev/null; then
+        python3 -c "
+import winrm
+session = winrm.Session('$WIN_IP:$WINRM_PORT', auth=('$WIN_USER', '$WIN_PASSWORD'), transport='ntlm')
+result = session.run_cmd('echo test')
+print('WinRM OK' if result.status_code == 0 else 'WinRM FAILED')
+"
+    else
+        log_error "No WinRM client found. Install winrm or python3 with pywinrm"
+        return 1
+    fi
+}
+
+run_winrm_cmd() {
+    local cmd="$1"
+    local output_file="$2"
+    
+    if command -v winrm &>/dev/null; then
+        winrm invoke -r:"http://$WIN_IP:$WINRM_PORT/wsman" -u:"$WIN_USER" -p:"$WIN_PASSWORD" \
+            -auth:basic Create "http://schemas.microsoft.com/wbem/wsman/1/windows/shell/Command" \
+            "Command=cmd.exe" "Arguments=/c $cmd" > "$output_file" 2>&1
+    else
+        python3 -c "
+import winrm, sys
+session = winrm.Session('$WIN_IP:$WINRM_PORT', auth=('$WIN_USER', '$WIN_PASSWORD'), transport='ntlm')
+result = session.run_cmd('$cmd')
+with open('$output_file', 'w') as f:
+    f.write(result.std_out.decode() + result.std_err.decode())
+sys.exit(result.status_code)
+"
+    fi
+}
+
+run_ssh_cmd() {
+    local cmd="$1"
+    local output_file="$2"
+    sshpass -p "$WIN_PASSWORD" ssh -o StrictHostKeyChecking=no -p "$SSH_PORT" "$WIN_USER@$WIN_IP" "$cmd" > "$output_file" 2>&1
+}
+
+test_nvidia_smi() {
+    log_info "Testing nvidia-smi..."
+    run_winrm_cmd "nvidia-smi" "$RESULTS_DIR/nvidia_smi.txt"
+    if grep -q "NVIDIA" "$RESULTS_DIR/nvidia_smi.txt"; then
+        log_ok "nvidia-smi: PASS"
+        return 0
+    else
+        log_error "nvidia-smi: FAIL"
+        cat "$RESULTS_DIR/nvidia_smi.txt"
+        return 1
+    fi
+}
+
+test_cuda_vectoradd() {
+    log_info "Testing CUDA vectorAdd..."
+    run_winrm_cmd "cd C:\\cuda_samples\\1_Utilities\\vectorAdd && .\\vectorAdd.exe" "$RESULTS_DIR/cuda_vectoradd.txt"
+    if grep -q "Test passed" "$RESULTS_DIR/cuda_vectoradd.txt"; then
+        log_ok "CUDA vectorAdd: PASS"
+        return 0
+    else
+        log_error "CUDA vectorAdd: FAIL"
+        cat "$RESULTS_DIR/cuda_vectoradd.txt"
+        return 1
+    fi
+}
+
+test_cuda_devquery() {
+    log_info "Testing CUDA deviceQuery..."
+    run_winrm_cmd "cd C:\\cuda_samples\\1_Utilities\\deviceQuery && .\\deviceQuery.exe" "$RESULTS_DIR/cuda_devquery.txt"
+    if grep -q "Result = PASS" "$RESULTS_DIR/cuda_devquery.txt"; then
+        log_ok "CUDA deviceQuery: PASS"
+        return 0
+    else
+        log_error "CUDA deviceQuery: FAIL"
+        cat "$RESULTS_DIR/cuda_devquery.txt"
+        return 1
+    fi
+}
+
+test_vulkan() {
+    log_info "Testing Vulkan..."
+    run_winrm_cmd "vulkaninfo.exe" "$RESULTS_DIR/vulkaninfo.txt"
+    if grep -q "GPU id" "$RESULTS_DIR/vulkaninfo.txt"; then
+        log_ok "Vulkan: PASS"
+        return 0
+    else
+        log_error "Vulkan: FAIL"
+        cat "$RESULTS_DIR/vulkaninfo.txt"
+        return 1
+    fi
+}
+
+test_dx12() {
+    log_info "Testing DirectX 12..."
+    run_winrm_cmd "dxdiag /t $RESULTS_DIR/dxdiag.txt" "$RESULTS_DIR/dxdiag_cmd.txt"
+    # Parse dxdiag output
+    if grep -q "Feature Level" "$RESULTS_DIR/dxdiag.txt"; then
+        log_ok "DirectX 12: PASS"
+        return 0
+    else
+        log_error "DirectX 12: FAIL"
+        return 1
+    fi
+}
+
+test_ffmpeg_nvenc() {
+    log_info "Testing FFmpeg NVENC..."
+    run_winrm_cmd "ffmpeg -encoders" "$RESULTS_DIR/ffmpeg_encoders.txt"
+    if grep -q "h264_nvenc\|hevc_nvenc" "$RESULTS_DIR/ffmpeg_encoders.txt"; then
+        log_ok "FFmpeg NVENC: PASS"
+        return 0
+    else
+        log_error "FFmpeg NVENC: FAIL"
+        return 1
+    fi
+}
+
+test_pytorch() {
+    log_info "Testing PyTorch..."
+    run_winrm_cmd "python -c \"import torch; print('CUDA:', torch.cuda.is_available()); print('Device:', torch.cuda.get_device_name(0) if torch.cuda.is_available() else 'N/A')\"" "$RESULTS_DIR/pytorch.txt"
+    if grep -q "CUDA: True" "$RESULTS_DIR/pytorch.txt"; then
+        log_ok "PyTorch: PASS"
+        return 0
+    else
+        log_error "PyTorch: FAIL"
+        cat "$RESULTS_DIR/pytorch.txt"
+        return 1
+    fi
+}
+
+test_windows_driver_signature() {
+    log_info "Checking driver signature status..."
+    run_winrm_cmd "pnputil /enum-drivers | findstr nvidia" "$RESULTS_DIR/driver_signature.txt"
+    cat "$RESULTS_DIR/driver_signature.txt"
+}
+
+collect_event_logs() {
+    log_info "Collecting Windows event logs..."
+    run_winrm_cmd "wevtutil qe System /q:\"*[System[Provider[@Name='NVIDIA']]]\" /f:text /c:50" "$RESULTS_DIR/nvidia_events.txt"
+    run_winrm_cmd "wevtutil qe System /q:\"*[System[Provider[@Name='Display']]]\" /f:text /c:50" "$RESULTS_DIR/display_events.txt"
+}
+
+run_full_suite() {
+    log_info "=== Windows Guest Test Suite ==="
+    local failed=0
+    
+    test_nvidia_smi || ((failed++))
+    test_cuda_devquery || ((failed++))
+    test_cuda_vectoradd || ((failed++))
+    test_vulkan || ((failed++))
+    test_dx12 || ((failed++))
+    test_ffmpeg_nvenc || ((failed++))
+    test_pytorch || ((failed++))
+    test_windows_driver_signature
+    collect_event_logs
+    
+    log_info "=== Results ==="
+    log_info "Failed tests: $failed"
+    
+    # Generate summary
+    cat > "$RESULTS_DIR/summary.json" <<EOF
+{
+    "timestamp": "$(date -Iseconds)",
+    "guest_ip": "$WIN_IP",
+    "failed_tests": $failed,
+    "tests": {
+        "nvidia_smi": $(test -f "$RESULTS_DIR/nvidia_smi.txt" && grep -q "NVIDIA" "$RESULTS_DIR/nvidia_smi.txt" && echo true || echo false),
+        "cuda_devquery": $(test -f "$RESULTS_DIR/cuda_devquery.txt" && grep -q "Result = PASS" "$RESULTS_DIR/cuda_devquery.txt" && echo true || echo false),
+        "cuda_vectoradd": $(test -f "$RESULTS_DIR/cuda_vectoradd.txt" && grep -q "Test passed" "$RESULTS_DIR/cuda_vectoradd.txt" && echo true || echo false),
+        "vulkan": $(test -f "$RESULTS_DIR/vulkaninfo.txt" && grep -q "GPU id" "$RESULTS_DIR/vulkaninfo.txt" && echo true || echo false),
+        "dx12": $(test -f "$RESULTS_DIR/dxdiag.txt" && grep -q "Feature Level" "$RESULTS_DIR/dxdiag.txt" && echo true || echo false),
+        "ffmpeg_nvenc": $(test -f "$RESULTS_DIR/ffmpeg_encoders.txt" && grep -q "h264_nvenc\|hevc_nvenc" "$RESULTS_DIR/ffmpeg_encoders.txt" && echo true || echo false),
+        "pytorch": $(test -f "$RESULTS_DIR/pytorch.txt" && grep -q "CUDA: True" "$RESULTS_DIR/pytorch.txt" && echo true || echo false)
+    }
+}
+EOF
+    
+    log_ok "Results saved to $RESULTS_DIR"
+    cat "$RESULTS_DIR/summary.json"
+}
+
+main() {
+    case "${1:-full}" in
+        check) check_winrm ;;
+        nvidia-smi) test_nvidia_smi ;;
+        cuda) test_cuda_devquery && test_cuda_vectoradd ;;
+        graphics) test_vulkan && test_dx12 ;;
+        video) test_ffmpeg_nvenc ;;
+        pytorch) test_pytorch ;;
+        full) run_full_suite ;;
+        *) echo "Usage: $0 {check|nvidia-smi|cuda|graphics|video|pytorch|full}"; exit 1 ;;
+    esac
+}
+
+main "$@"
\ No newline at end of file