NVIDIA Attestation

https://docs.nvidia.com/attestation
Admin
NVIDIA Attestation provides tools and cloud services to cryptographically verify the authenticity...
Tokens:122,276
Snippets:984
Trust Score:10
Update:4 weeks ago
Show doc for...
Tokens:
Raw
### Build and run examples with make run-examples

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-c/development.html

Builds and installs the SDK using make, then runs the examples.

```makefile
# build and install sdk using make
make run-examples

```

--------------------------------

### Example: help(verify_measurement_signature)

Source: https://docs.nvidia.com/attestation/attestation-client-tools-sdk/latest/local-verifier/usage.html

Example of using help() to get information about the verify_measurement_signature function.

```python
e.g. help(verify_measurement_signature)
```

--------------------------------

### Install the SDK

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-c/development.html

Install the built SDK.

```bash
make install-sdk
```

--------------------------------

### Install PPCIE Verifier via Installer Script

Source: https://docs.nvidia.com/attestation/attestation-client-tools-ppcie-sdk-cpp/latest/ppcie_quickstart_guide.html

Installs PPCIE Verifier using the installer script from the nvtrust repository.

```bash
1. git clone https://github.com/NVIDIA/nvtrust
2. cd nvtrust/guest_tools/ppcie-verifier-sdk-cpp/install
3. source ppcie-installer.sh

```

--------------------------------

### Run examples

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-rust/development.html

Run the remote and local attestation examples using cargo.

```bash
# Remote attestation example
cargo run -p nv-attestation-sdk --example basic_remote_attestation

# Local attestation example
cargo run -p nv-attestation-sdk --example basic_local_attestation

```

--------------------------------

### Install from source and run verifier

Source: https://docs.nvidia.com/attestation/attestation-client-tools-sdk/latest/local-verifier/installation.html

Installs the local GPU verifier from source by cloning the NVIDIA nvtrust repository, navigating to the local_gpu_verifier directory, installing the package, and running the verifier module.

```bash
git clone https://github.com/NVIDIA/nvtrust
cd nvtrust/guest_tools/gpu_verifiers/local_gpu_verifier
pip3 install .
python3 -m verifier.cc_admin

```

--------------------------------

### Install SDK from Source

Source: https://docs.nvidia.com/attestation/quick-start-guide/latest/install.html

Clones the nvtrust repository and installs the Attestation SDK from source. Requires nv-local-gpu-verifier as a pre-requisite.

```bash
git clone https://github.com/NVIDIA/nvtrust
cd nvtrust/guest_tools/attestation_sdk
pip3 install .

```

--------------------------------

### Build and install the nvattest CLI

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-cli/introduction.html

Build the CLI using make and then install it.

```bash
make build-cli
make install-cli
```

--------------------------------

### Install SDK from PyPI

Source: https://docs.nvidia.com/attestation/quick-start-guide/latest/install.html

Installs the Attestation SDK directly from PyPI. Use within the activated virtual environment.

```bash
pip3 install nv-attestation-sdk

```

--------------------------------

### Install from PyPI and run verifier

Source: https://docs.nvidia.com/attestation/attestation-client-tools-sdk/latest/local-verifier/installation.html

Installs the local GPU verifier from PyPI and runs the verifier module.

```bash
pip3 install nv-local-gpu-verifier
python3 -m verifier.cc_admin

```

--------------------------------

### Build and install the CLI from source

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/overview.html

Clone the repository and use CMake to build and install the nvattest CLI.

```bash
git clone https://github.com/NVIDIA/attestation-sdk.git
cd attestation-sdk/nv-attestation-cli
cmake -S . -B build -DCMAKE_BUILD_TYPE=Release
cmake --build build
cmake --install build
sudo ldconfig
```

--------------------------------

### Create virtual environment and install PPCIE Verifier

Source: https://docs.nvidia.com/attestation/attestation-client-tools-ppcie/latest/ppcie_quickstart_guide.html

Creates a Python virtual environment, activates it, and installs PPCIE Verifier version 1.x from PyPi.

```bash
python3 -m venv venv

source venv/bin/activate

# This installs version 1.x 
pip3 install "nv-ppcie-verifier>=1.0,<2.0"

```

--------------------------------

### Install build dependencies

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-rust/development.html

Commands to install required build tools and libraries on Ubuntu. Run with sudo apt-get.

```bash
# Update package list
sudo apt-get update

# Install build tools and dependencies
sudo apt-get install -y \
    cmake \
    git \
    pkg-config \
    clang \
    libclang-dev \
    libcurl4-openssl-dev \
    libssl-dev \
    libxml2-dev \
    libxmlsec1-dev \
    libxmlsec1-openssl

```

--------------------------------

### Install PPCIE Verifier via PyPI

Source: https://docs.nvidia.com/attestation/attestation-client-tools-ppcie-sdk-cpp/latest/ppcie_quickstart_guide.html

Installs PPCIE Verifier from PyPI, requiring a Python virtual environment.

```bash
1. python3 -m venv venv
2. source venv/bin/activate
3. pip3 install nv-ppcie-verifier

```

--------------------------------

### Install build dependencies for CLI

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/overview.html

Install the necessary build dependencies on Ubuntu before compiling the CLI from source.

```bash
apt update && \
apt install cmake git pkg-config clang \
    libcurl4-openssl-dev libssl-dev libxml2-dev \
    libxmlsec1-dev libxmlsec1-openssl
```

--------------------------------

### Create and Activate a Virtual Environment

Source: https://docs.nvidia.com/attestation/quick-start-guide/latest/install.html

Creates and activates a Python virtual environment before installing the SDK. Recommended to avoid permission errors and system-wide installation issues.

```bash
python3 -m venv venv
source venv/bin/activate

```

--------------------------------

### Build and install

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-c/introduction.html

CMake build, build, install, and update the linker cache. Note that find_package(nvat) is not supported.

```bash
cmake -S . -B build
cmake --build build
cmake --install build --strip
sudo ldconfig
```

--------------------------------

### Subprocess error output during installation

Source: https://docs.nvidia.com/attestation/attestation-client-tools-sdk/latest/local-verifier/installation.html

Example error output when subprocess-related issues occur during installation, indicating a problem with updating the timestamp of the egg-info directory.

```text
   Installing build dependencies ... done
   Getting requirements to build wheel ... error
   error: subprocess-exited-with-error
   × Getting requirements to build wheel did not run successfully.
   │ exit code: 1
   ╰─> [2 lines of output]
       running egg_info
   error: Cannot update time stamp of directory 'src/nv_local_gpu_verifier.egg-info'
    [end of output]
   note: This error originates from a subprocess and is likely not a problem with pip.
   error: subprocess-exited-with-error
   × Getting requirements to build wheel did not run successfully.
   │ exit code: 1
   ╰─> See above for output.

```

--------------------------------

### Complete configuration example using nv_attestation_sdk

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-rust/configuration.html

Complete Rust example for the NVIDIA Attestation SDK. Initializes logging, configures the SDK with SdkOptions and Logger, sets HTTP options (max retry 5, connection timeout 10s, request timeout 30s), creates a filesystem RimStore and default OcspClient with the NVIDIA OCSP endpoint and an API key, then configures an AttestationContext for GPU device type and local verifier before performing attestation and printing the claims JSON.

```rust
use nv_attestation_sdk::{
    NvatSdk, SdkOptions, Logger, AttestationContext,
    HttpOptions, RimStore, OcspClient,
    DeviceType, VerifierType
};

fn main() -> Result<(), Box<dyn std::error::Error>> {
    // 1. Initialize logging
    env_logger::init();

    // 2. Configure SDK
    let mut sdk_opts = SdkOptions::new()?;
    let logger = Logger::new()?;
    sdk_opts.set_logger(logger);
    let _client = NvatSdk::init(sdk_opts)?;

    // 3. Configure HTTP
    let mut http_opts = HttpOptions::default_options()?;
    http_opts.set_max_retry_count(5);
    http_opts.set_connection_timeout_ms(10000);
    http_opts.set_request_timeout_ms(30000);

    // 4. Set up verification components
    let _rim_store = RimStore::create_filesystem("/var/lib/nvat/rims")?;
    let _ocsp_client = OcspClient::create_default(
        Some("https://ocsp.attestation.nvidia.com"),
        Some("your-api-key"),
        Some(&http_opts)
    )?;

    // 5. Configure attestation context
    let mut ctx = AttestationContext::new()?;
    ctx.set_device_type(DeviceType::Gpu)?;
    ctx.set_verifier_type(VerifierType::Local)?;

    // 6. Perform attestation
    let result = ctx.attest_device(None)?;

    println!("Attestation successful!");
    println!("Claims: {}", result.claims_json()?);

    Ok(())
}

```

--------------------------------

### Example NRAS Token (Remote Verifier) - Multi-GPU

Source: https://docs.nvidia.com/attestation/quick-start-guide/latest/attestation-examples/blackwell_multi_gpu.html

Shows the structure of an NRAS token (JWT) for remote attestation in a multi-GPU setup. The token contains an outer JWT and a nested 'REMOTE_GPU_CLAIMS' object with per-GPU attestation reports.

```json
[
    [
        "JWT", "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJOVi1BdHRlc3RhdGlvbi1TREsiLCJpYXQiOjE3NjEyODU0ODgsImV4cCI6MTc2MTI4OTA4OCwibmJmIjoxNzYxMjg1MzY4LCJqdGkiOiI1YWQ2M2NlNC1iYTk3LTQ5NzYtOThmZi04ZDgyMDI3NGNhY2YifQ.KmNy9wu2H3eNpliBadX58zdAeAVYP7jQscg_azPc7n4"
    ], 
    {
        "REMOTE_GPU_CLAIMS": 
        [
            [
                "JWT", "eyJraWQiOiJudi1lYXQta2lkLXN0Zy0yMDI1MTAyMzA3NDcyMzk1MS1hZTk1ZWY5ZS0wMTUzLTQ3ZDUtOTEyNy00ZjkwMTU1MTI2Y2YiLCJhbGciOiJFUzM4NCJ9.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.QpluUrkJkS9owNGB7KCYcchfChdhYrg_Wln4bx53iwAKFU_6zHDUf_iIie8v8aLWEWD0ERGWHJZ5YyUrsJ_ELzam9ZQpiOsLqDvpaUq9FnpUGFvU6o8XdHymhZYOwg3t"
            ], 
            {
                "GPU-0": "eyJraWQiOiJudi1lYXQta2lkLXN0Zy0yMDI1MTAyMzA3NDcyMzk1MS1hZTk1ZWY5ZS0wMTUzLTQ3ZDUtOTEyNy00ZjkwMTU1MTI2Y2YiLCJhbGciOiJFUzM4NCJ9.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.DL_i1GEBU5Znz_W8awCxs7TgaSfGYy_W-uGSo1pt0xUAtr-MEexUjoA0x0yjhWZE3fCUa4ekp2ZikdyVabC38MIc4mhrJwFUJJzD5E-VZPLJF1jeSegfbCHz6yK4BwP-",
                "GPU-1": "eyJra...",
                "GPU-2": "eyJra..."
            }
        ]
    }
]

```

--------------------------------

### Remote GPU attestation with nv_attestation_sdk

Source: https://docs.nvidia.com/attestation/quick-start-guide/latest/attestation-examples/hopper_single_gpu.html

Complete remote attestation flow for a single Hopper H100 GPU using the NVIDIA attestation SDK. Sets the API key, adds the remote NRAS verifier, obtains evidence, performs attestation, retrieves the JWT token, and decodes it. Requires completing the Installation Guide first.

```python
from nv_attestation_sdk import attestation

# Create attestation client
client = attestation.Attestation()

# Set API key (if not set via environment variable)
client.set_service_key("YOUR_API_KEY")

# Add remote GPU verifier
NRAS_URL = "https://nras.attestation.nvidia.com/v4/attest/gpu"
client.add_verifier(attestation.Devices.GPU, attestation.Environment.REMOTE, NRAS_URL, "")

# Get evidence and perform attestation
evidence_list = client.get_evidence()
result = client.attest(evidence_list)

# Get the attestation token (JWT Format)
token = client.get_token()
print("[RemoteGPUTest] token : " + str(token))

# Decode the token
client.decode_token(token)
```

--------------------------------

### Uninstall Improperly Installed Package

Source: https://docs.nvidia.com/attestation/quick-start-guide/latest/install.html

Removes any improperly installed versions of the package before retrying installation.

```bash
pip uninstall nvidia-attestation-sdk

```

--------------------------------

### Typical development workflow

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-c/development.html

A standard sequence of commands to set up, start, enter, and stop the development container.

```bash
./dev-env.sh setup
./dev-env.sh start
./dev-env.sh shell
# do some work
./dev-env.sh stop

# to nuke the container
./dev-env.sh clean # need to run setup again after this
```

--------------------------------

### Verify nvattest Installation

Source: https://docs.nvidia.com/attestation/attestation-client-tools-ppcie-sdk-cpp/latest/ppcie_quickstart_guide.html

Verifies that the NVIDIA Attestation CLI (nvattest) is installed and on the PATH before installing PPCIE Verifier.

```bash
nvattest version

```

--------------------------------

### Build with system-installed library

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-rust/introduction.html

Use the system-installed NVIDIA Attestation SDK library. The library will be found in standard system paths such as /usr/lib and /usr/local/lib.

```bash
export NVAT_USE_SYSTEM_LIB=1
cargo build

```

--------------------------------

### Install build dependencies for nvattest from source

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-cli/introduction.html

Install required packages for building the CLI from source. Includes Rust installation via rustup.

```bash
sudo apt-get update && \
sudo apt-get install -y cmake git pkg-config clang \
    libcurl4-openssl-dev libssl-dev libxml2-dev \
    libxmlsec1-dev libxmlsec1-openssl

# rust is a build dependency
# install from here - https://rust-lang.org/tools/install/
# For Linux/Unix or macOS, install Rust with:
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
```

--------------------------------

### Install GTest for testing

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-cli/development.html

Clones and installs GoogleTest v1.16.0 from source. Required for building the tests. The command chains git clone, build, and install steps.

```bash
git clone --depth 1 https://github.com/google/googletest.git -b v1.16.0 
&& cd googletest 
&& mkdir build 
&& cd build 
&& cmake .. 
&& cmake --build . --target install 
&& cd / 
&& rm -rf googletest

```

--------------------------------

### attest example output

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-cli/command-reference.html

Example JSON output from attest, truncated.

```json
{
  "claims": [
    {
      "x-nvidia-device-type": "gpu",
      "secboot": true,
      "x-nvidia-gpu-driver-version": "575.32"
      // ...
    }
  ],
  "detached_eat": [
  	[
  		"JWT",
  		"the-overall-jwt"
  	],
  	{
  		"GPU-0": "jwt-for-this-submodule", 
  		"GPU-1": "jwt-for-this-submodule",
  		// ...
  	}
  ],
  "result_code": 0,
  "result_message": "Ok"
}

```

--------------------------------

### collect-evidence example output

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-cli/command-reference.html

Example JSON output from collect-evidence, truncated.

```json
{
  "evidences": [
    {
      "arch": "HOPPER",
      "nonce": "123..",
      "evidence": "EeAB..",
      "certificate": "LS0t.."
    },
    {
      "evidence": "RfdT..",
      // ...
    }
  ],
  "result_code": 0,
  "result_message": "Ok"
}

```

--------------------------------

### Build the SDK

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-c/development.html

Build the SDK to generate the libnvat.so library.

```bash
make build-sdk # see make help for more info
```

--------------------------------

### Install Clang/LLVM for bindgen

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-rust/introduction.html

Install Clang/LLVM, required by bindgen for generating FFI bindings.

```bash
sudo apt-get install clang libclang-dev

```

--------------------------------

### Attest using evidence files

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-cli/development.html

Uses pre-existing JSON evidence files instead of live collection. Specify the evidence source as 'file' and provide the path to the evidence file for the GPU or NVSwitch.

```bash
nvattest attest --device gpu --gpu-evidence-sourc file --gpu-evidence-file ./path/to/gpu_evidence.json
nvattest attest --device nvswitch --nvswitch-evidence-source file --nvswitch-evidence-file ./path/to/switch_evidence.json

```

--------------------------------

### Build from source: C++ SDK, Rust workspace, and runtime path

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-rust/introduction.html

Build the C++ SDK, then build the Rust workspace, and set the runtime library path. These steps are part of the same workflow for building from source.

```bash
cd nv-attestation-sdk-cpp
cmake -S . -B build -DCMAKE_BUILD_TYPE=Release
cmake --build build
cmake --install build
sudo ldconfig

```

```bash
cd ../nv-attestation-sdk-rust
cargo build --workspace

```

```bash
export LD_LIBRARY_PATH=$(pwd)/../nv-attestation-sdk-cpp/build:$LD_LIBRARY_PATH

```

--------------------------------

### Get the source

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-c/introduction.html

Clone the attestation-sdk repository and navigate to the C++ SDK directory.

```bash
git clone https://github.com/NVIDIA/attestation-sdk.git
cd attestation-sdk/nv-attestation-sdk-cpp
```

--------------------------------

### Set NVAT_BUILD_DIR or install SDK system-wide for bindgen errors

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-rust/development.html

Use when bindgen fails to find headers. Set the NVAT_BUILD_DIR environment variable to the C++ SDK build directory, or install the SDK system-wide using cmake --install.

```bash
# Set C++ SDK build directory
export NVAT_BUILD_DIR=/path/to/nv-attestation-sdk-cpp/build

# Or install SDK system-wide
cd nv-attestation-sdk-cpp
sudo cmake --install build

```

--------------------------------

### nvat_sdk_init()

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/api/group__options.html

Initializes the NVIDIA Attestation SDK with the provided options.

```APIDOC
## nvat_sdk_init()

### Description
Initializes the NVIDIA Attestation SDK with the provided options.

### Syntax
```c
nvat_status_t nvat_sdk_init(const nvat_sdk_opts_t opts);
```

### Parameters
- **opts** (const nvat_sdk_opts_t) - Required - Options used to initialize the SDK.

### Returns
- **nvat_status_t** - Status code indicating success or failure.
```

--------------------------------

### Local GPU Attestation: Python to C Migration

Source: https://docs.nvidia.com/attestation/attestation-client-tools-sdk/latest/migration_guide.html

Shows the migration from the Python SDK's local GPU attestation flow to the C SDK's equivalent. The C example initializes the SDK, calls nvat_attest_device() with no verifier configuration (local is default), and prints whether attestation passed. Cleanup must be done in reverse order of creation.

```python
from nv_attestation_sdk import attestation

client = attestation.Attestation()
client.set_name("myNode")
client.add_verifier(attestation.Devices.GPU,
                    attestation.Environment.LOCAL, "", "")

evidence = client.get_evidence()
result = client.attest(evidence)
token = client.get_token()
print("Attestation passed:", result)

```

```c
#include <nvat.h>
#include <stdio.h>

int main(void) {
    nvat_sdk_opts_t opts = NULL;
    nvat_attestation_ctx_t ctx = NULL;
    nvat_str_t eat = NULL;
    nvat_claims_collection_t claims = NULL;

    // Initialize the SDK
    nvat_sdk_opts_create(&opts);
    nvat_sdk_init(opts);
    nvat_attestation_ctx_create(&ctx);

    // Attest (local GPU is the default)
    nvat_rc_t rc = nvat_attest_device(ctx, NULL, &eat, &claims);
    printf("Attestation %s\n", rc == NVAT_RC_OK ? "passed" : "failed");

    // Cleanup — always free in reverse order of creation
    nvat_str_free(&eat);
    nvat_claims_collection_free(&claims);
    nvat_attestation_ctx_free(&ctx);
    nvat_sdk_opts_free(&opts);
    nvat_sdk_shutdown();

    return rc == NVAT_RC_OK ? 0 : 1;
}

```

--------------------------------

### Build the C++ SDK

Source: https://docs.nvidia.com/attestation/nv-attestation-sdk-cpp/latest/sdk-rust/development.html

Clone the attestation-sdk repository, build the C++ SDK with CMake, install it system-wide, and update the linker cache.

```bash
# Clone the repository
git clone https://github.com/NVIDIA/attestation-sdk.git
cd attestation-sdk

# Build the C++ SDK
cd nv-attestation-sdk-cpp
cmake -S . -B build -DCMAKE_BUILD_TYPE=Release
cmake --build build
cmake --install build
sudo ldconfig

```
Tokens:
Raw