VFIO device registration fix + IOMMUFD callbacks + VFIO ioctl handlers

- Removed IOMMU group addition before vfio_register_emulated_iommu_dev() in nv_vgpu_mdev_create()
- Added IOMMUFD callbacks (bind_iommufd, unbind_iommufd, attach_ioas, detach_ioas) using kernel's vfio_iommufd_emulated_* functions
- Implemented VFIO ioctl handlers: GET_INFO, GET_REGION_INFO, GET_IRQ_INFO, SET_IRQS, MMAP
- Added config space read/write handlers at offset 0x70000000000
- Mmap handler for config space, BAR0 (emulation shadow), BAR1-5 (physical VRAM/registers)
- Interrupt handler for VFIO_DEVICE_SET_IRQS (DATA_EVENTFD, DATA_NONE, DATA_BOOL)
- Added udev rules for vfio group permissions
- Created stack_manager.sh unified tool
- Fixed mdev remove crash with deferred cleanup in vfio_device_release callback
- Kernel 6.8+ IOMMUFD compatibility
- QEMU launches successfully with vGPU passthrough

1 files changed, 172 insertions, 0 deletions

diff --git a/scripts/debug_guest_init.sh b/scripts/debug_guest_init.sh
new file mode 100755
index 0000000..324f403
--- /dev/null
+++ b/scripts/debug_guest_init.sh
@@ -0,0 +1,172 @@
+#!/bin/bash
+# debug_guest_init.sh - One-command debug launch for guest NVIDIA driver initialization
+# Usage: ./debug_guest_init.sh [options]
+
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+BASE_DIR="$(cd "$SCRIPT_DIR/.." && pwd)"
+
+# Defaults
+GDB_PORT=1234
+FREEZE_AT_START=true
+MONITOR_GUEST=true
+GUEST_SERIAL="/dev/ttyS0"
+DURATION=300
+OUTPUT_DIR="/tmp/gsp_emu_debug_$(date +%Y%m%d_%H%M%S)"
+
+# Colors
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+YELLOW='\033[1;33m'
+BLUE='\033[0;34m'
+NC='\033[0m'
+
+log_info() { echo -e "${BLUE}[INFO]${NC} $1"; }
+log_warn() { echo -e "${YELLOW}[WARN]${NC} $1"; }
+log_error() { echo -e "${RED}[ERROR]${NC} $1"; }
+log_ok() { echo -e "${GREEN}[OK]${NC} $1"; }
+
+usage() {
+    cat <<EOF
+Usage: $0 [options]
+
+Debug guest NVIDIA driver initialization with full observability.
+
+Options:
+  --gdb-port PORT         GDB server port (default: 1234)
+  --no-freeze             Don't freeze QEMU at startup (-S flag)
+  --no-monitor            Disable guest init monitor
+  --guest-serial DEV      Guest serial device (default: /dev/ttyS0)
+  --duration SEC          Monitor duration in seconds (default: 300)
+  --output-dir DIR        Output directory (default: /tmp/gsp_emu_debug_TIMESTAMP)
+  --vmlinux PATH          Path to guest vmlinux with debug symbols
+  --help                  Show this help
+
+This script:
+  1. Enables kernel tracepoints for GSP emulation
+  2. Launches QEMU with GDB server
+  3. Starts guest init monitor (correlates host traces + guest serial)
+  4. Opens GDB with nvidia_init.gdb script
+  5. Saves all logs to output directory
+
+Example:
+  $0 --vmlinux /path/to/guest/vmlinux
+  $0 --no-freeze --duration 600
+EOF
+}
+
+# Parse arguments
+while [[ $# -gt 0 ]]; do
+    case $1 in
+        --gdb-port) GDB_PORT="$2"; shift 2 ;;
+        --no-freeze) FREEZE_AT_START=false; shift ;;
+        --no-monitor) MONITOR_GUEST=false; shift ;;
+        --guest-serial) GUEST_SERIAL="$2"; shift 2 ;;
+        --duration) DURATION="$2"; shift 2 ;;
+        --output-dir) OUTPUT_DIR="$2"; shift 2 ;;
+        --vmlinux) VMLINUX="$2"; shift 2 ;;
+        --help) usage; exit 0 ;;
+        *) log_error "Unknown option: $1"; usage; exit 1 ;;
+    esac
+done
+
+mkdir -p "$OUTPUT_DIR"
+log_info "Output directory: $OUTPUT_DIR"
+
+# Check prerequisites
+log_info "Checking prerequisites..."
+
+# Check for mdev
+MDEV_UUID=$(ls /sys/bus/mdev/devices/ 2>/dev/null | head -1)
+if [[ -z "$MDEV_UUID" ]]; then
+    log_error "No mdev device found. Run create_mdev.sh first."
+    exit 1
+fi
+log_info "Using mdev: $MDEV_UUID"
+
+# Build QEMU command
+QEMU_CMD=(
+    qemu-system-x86_64
+    -enable-kvm
+    -cpu host
+    -smp 4
+    -m 8G
+    -device vfio-pci,sysfsdev=/sys/bus/mdev/devices/$MDEV_UUID,display=off
+    -drive file=/var/lib/libvirt/images/ubuntu-vgpu.qcow2,format=qcow2,if=virtio
+    -netdev user,id=net0,hostfwd=tcp::2222-:22
+    -device virtio-net-pci,netdev=net0
+    -nographic
+    -serial mon:stdio
+)
+
+if [[ "$FREEZE_AT_START" == "true" ]]; then
+    QEMU_CMD+=(-s -S)
+else
+    QEMU_CMD+=(-s)
+fi
+
+if [[ "$GDB_PORT" != "1234" ]]; then
+    QEMU_CMD+=(-gdb "tcp::$GDB_PORT")
+fi
+
+# Start guest init monitor in background
+MONITOR_PID=""
+if [[ "$MONITOR_GUEST" == "true" ]]; then
+    log_info "Starting guest init monitor..."
+    python3 "$BASE_DIR/tools/guest_init_monitor.py" \
+        --guest-serial "$GUEST_SERIAL" \
+        --duration "$DURATION" \
+        > "$OUTPUT_DIR/guest_monitor.log" 2>&1 &
+    MONITOR_PID=$!
+    log_ok "Guest monitor started (PID: $MONITOR_PID)"
+fi
+
+# Start host trace capture
+log_info "Starting host trace capture..."
+sudo cat /sys/kernel/debug/tracing/trace_pipe > "$OUTPUT_DIR/host_trace.log" 2>&1 &
+TRACE_PID=$!
+log_ok "Host trace capture started (PID: $TRACE_PID)"
+
+# Save QEMU command
+echo "${QEMU_CMD[@]}" > "$OUTPUT_DIR/qemu_cmd.txt"
+
+# Launch QEMU in background
+log_info "Launching QEMU with GDB server on port $GDB_PORT..."
+"${QEMU_CMD[@]}" > "$OUTPUT_DIR/qemu.log" 2>&1 &
+QEMU_PID=$!
+log_ok "QEMU started (PID: $QEMU_PID)"
+
+# Give QEMU time to start
+sleep 3
+
+# Launch GDB if vmlinux provided
+if [[ -n "${VMLINUX:-}" && -f "$VMLINUX" ]]; then
+    log_info "Launching GDB with nvidia_init.gdb..."
+    log_info "GDB will connect to localhost:$GDB_PORT"
+    log_info "In GDB: (gdb) target remote localhost:$GDB_PORT"
+    log_info "Then: (gdb) continue"
+    
+    gdb -x "$BASE_DIR/tools/gdb_scripts/nvidia_init.gdb" "$VMLINUX" \
+        -ex "target remote localhost:$GDB_PORT" \
+        -ex "set logging file $OUTPUT_DIR/gdb.log" \
+        -ex "set logging on"
+else
+    log_warn "No vmlinux provided. Connect manually:"
+    log_warn "  gdb -x $BASE_DIR/tools/gdb_scripts/nvidia_init.gdb /path/to/vmlinux"
+    log_warn "  (gdb) target remote localhost:$GDB_PORT"
+    log_warn ""
+    log_warn "Press Ctrl+C to stop QEMU and monitoring..."
+    
+    # Wait for QEMU or interrupt
+    wait $QEMU_PID
+fi
+
+# Cleanup
+log_info "Cleaning up..."
+[[ -n "$MONITOR_PID" ]] && kill $MONITOR_PID 2>/dev/null || true
+[[ -n "$TRACE_PID" ]] && kill $TRACE_PID 2>/dev/null || true
+kill $QEMU_PID 2>/dev/null || true
+
+log_info "Debug session complete. Logs in: $OUTPUT_DIR"
+ls -la "$OUTPUT_DIR"
\ No newline at end of file