Fix crashing

captures the block
returned by the last emulate_i128_* call (loop_body_exit_bb) and uses it for
the bit_phi, quot_phi, and rem_phi incoming edges.

This keeps the PHIs consistent with the optimized control flow that those
helpers build, so the codegen no longer crashes.

Author: LegNeato

crates/rustc_codegen_nvvm/src/builder.rs

@@ -1677,10 +1677,11 @@ impl<'a, 'll, 'tcx> Builder<'a, 'll, 'tcx> {
         let one_shifted = body_bx.emulate_i128_shl(one_i128, bit_next);
         let quot_candidate = body_bx.or(quot_phi, one_shifted);
         let quot_new = body_bx.select(rem_ge, quot_candidate, quot_phi);
+        let loop_body_exit_bb = body_bx.llbb();
         body_bx.br(loop_header_bb);
-        header_bx.add_incoming_to_phi(bit_phi, bit_next, loop_body_bb);
-        header_bx.add_incoming_to_phi(quot_phi, quot_new, loop_body_bb);
-        header_bx.add_incoming_to_phi(rem_phi, rem_new, loop_body_bb);
+        header_bx.add_incoming_to_phi(bit_phi, bit_next, loop_body_exit_bb);
+        header_bx.add_incoming_to_phi(quot_phi, quot_new, loop_body_exit_bb);
+        header_bx.add_incoming_to_phi(rem_phi, rem_new, loop_body_exit_bb);
 
         *self = exit_bx;
         (quot_res, rem_res)

deploy-cuda.sh

@@ -0,0 +1,167 @@
+#!/bin/bash
+set -e
+
+# Remote connection details
+REMOTE_HOST="[email protected]"
+REMOTE_PORT="24527"
+REMOTE_DIR="/root/Rust-CUDA"
+SSH_OPTS=(-o StrictHostKeyChecking=no -L 8080:localhost:8080)
+
+echo "🚀 Deploying Rust GPU Chimera Demo to vast.ai (Native CUDA Setup)"
+echo "=============================================================="
+
+# Create remote directory
+echo "Creating remote directory..."
+ssh "${SSH_OPTS[@]}" -p "$REMOTE_PORT" "$REMOTE_HOST" "mkdir -p $REMOTE_DIR"
+
+# Sync files
+echo "Syncing project files..."
+rsync -avz --delete \
+    --exclude 'target/' \
+    --exclude '.git/' \
+    --exclude '*.swp' \
+    --exclude '.DS_Store' \
+    --exclude 'deploy-*.sh' \
+    -e "ssh ${SSH_OPTS[*]} -p $REMOTE_PORT" \
+    ./ "$REMOTE_HOST:$REMOTE_DIR/"
+
+echo "Files synced successfully!"
+
+# SSH into the machine and set up environment
+echo -e "\n📦 Setting up CUDA development environment..."
+ssh "${SSH_OPTS[@]}" -p "$REMOTE_PORT" "$REMOTE_HOST" << 'EOF'
+# Clear MOTD
+echo "" > /etc/motd
+
+cd /root/Rust-CUDA
+
+# Check GPU
+echo -e "\n🎮 GPU Information:"
+nvidia-smi --query-gpu=name,driver_version,memory.total --format=csv
+
+# Check CUDA version
+echo -e "\n🚀 CUDA Version:"
+nvcc --version 2>/dev/null || echo "NVCC not found, checking nvidia-smi..."
+nvidia-smi | grep "CUDA Version" || echo "CUDA info not found"
+
+# Install build dependencies if not present
+echo -e "\n📦 Installing build dependencies..."
+if ! command -v clang &> /dev/null; then
+    apt-get update
+    DEBIAN_FRONTEND=noninteractive apt-get -qq -y install \
+        build-essential \
+        clang \
+        curl \
+        libssl-dev \
+        libtinfo-dev \
+        pkg-config \
+        xz-utils \
+        zlib1g-dev
+fi
+
+# Install Rust if not present
+if ! command -v cargo &> /dev/null; then
+    echo -e "\n🦀 Installing Rust..."
+    curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+    source $HOME/.cargo/env
+fi
+
+# Show Rust version
+echo -e "\n🦀 Rust version:"
+cargo --version
+rustc --version
+
+# Check if LLVM 7 is needed and not present
+if ! command -v llvm-config-7 &> /dev/null && ! command -v llvm-config &> /dev/null; then
+    echo -e "\n⚠️  LLVM 7 not found. Installing LLVM for CUDA support..."
+    
+    # Install LLVM dependencies
+    apt-get update
+    DEBIAN_FRONTEND=noninteractive apt-get -qq -y install \
+        libffi-dev \
+        libedit-dev \
+        libncurses5-dev \
+        libxml2-dev \
+        python3 \
+        ninja-build
+    
+    # Download and build LLVM 7.1.0
+    mkdir -p /tmp/llvm7
+    cd /tmp/llvm7
+    
+    echo "Downloading LLVM 7.1.0..."
+    curl -sSf -L -O https://github.com/llvm/llvm-project/releases/download/llvmorg-7.1.0/llvm-7.1.0.src.tar.xz
+    tar -xf llvm-7.1.0.src.tar.xz
+    cd llvm-7.1.0.src
+    mkdir build && cd build
+    
+    echo "Building LLVM 7.1.0 (this will take a while)..."
+    cmake -G Ninja \
+        -DCMAKE_BUILD_TYPE=Release \
+        -DLLVM_TARGETS_TO_BUILD="X86;NVPTX" \
+        -DLLVM_BUILD_LLVM_DYLIB=ON \
+        -DLLVM_LINK_LLVM_DYLIB=ON \
+        -DLLVM_ENABLE_ASSERTIONS=OFF \
+        -DLLVM_ENABLE_BINDINGS=OFF \
+        -DLLVM_INCLUDE_EXAMPLES=OFF \
+        -DLLVM_INCLUDE_TESTS=OFF \
+        -DLLVM_INCLUDE_BENCHMARKS=OFF \
+        -DLLVM_ENABLE_ZLIB=ON \
+        -DLLVM_ENABLE_TERMINFO=ON \
+        -DCMAKE_INSTALL_PREFIX=/usr \
+        ..
+    
+    ninja -j$(nproc)
+    ninja install
+    
+    # Create symlink
+    ln -s /usr/bin/llvm-config /usr/bin/llvm-config-7
+    
+    # Cleanup
+    cd /
+    rm -rf /tmp/llvm7
+    
+    echo "LLVM 7 installed successfully!"
+fi
+
+# Set up environment variables
+export LD_LIBRARY_PATH="/usr/local/cuda/nvvm/lib64:${LD_LIBRARY_PATH}"
+export LLVM_LINK_STATIC=1
+export RUST_LOG=info
+export PATH="$HOME/.cargo/bin:${PATH}"
+
+# Go back to project directory
+cd /root/Rust-CUDA
+
+# Show toolchain info
+echo -e "\n🔧 Rust toolchain:"
+rustup show
+
+# Build and run
+echo -e "\n🔨 Building project..."
+
+# Test CPU backend first
+echo -e "\n📊 Testing CPU backend..."
+cargo test --release
+cargo run --release
+
+# Test CUDA backend
+echo -e "\n🎮 Building and testing CUDA backend..."
+if cargo build --release --features cuda; then
+    echo "CUDA build successful!"
+    echo -e "\n🚀 Running CUDA demo..."
+    cargo test --release --features cuda
+    cargo run --release --features cuda
+else
+    echo "CUDA build failed. Checking environment..."
+    echo "LD_LIBRARY_PATH: $LD_LIBRARY_PATH"
+    echo "CUDA NVVM libraries:"
+    ls -la /usr/local/cuda/nvvm/lib64/ 2>/dev/null || echo "NVVM lib64 not found"
+    echo "LLVM config:"
+    llvm-config --version 2>/dev/null || echo "llvm-config not found"
+fi
+
+echo -e "\n✅ Setup complete!"
+EOF
+
+echo -e "\n✅ Deployment complete!"