Internet-Draft Symantec
Intended status: Informational H. Tschofenig
Expires: September 12, 2019 Arm Limited
D. Wheeler
A. Atyeo
L. Dapeng
Alibaba Group
March 11, 2019

Trusted Execution Environment Provisioning (TEEP) Architecture


A Trusted Execution Environment (TEE) is designed to provide a hardware-isolation mechanism to separate a regular operating system from security-sensitive application components.

This architecture document motivates the design and standardization of a protocol for managing the lifecycle of trusted applications running inside a TEE.

Status of This Memo

This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.

Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.

Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."

This Internet-Draft will expire on September 12, 2019.

Copyright Notice

Copyright (c) 2019 IETF Trust and the persons identified as the document authors. All rights reserved.

This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.

This document may contain material from IETF Documents or IETF Contributions published or made publicly available before November 10, 2008. The person(s) controlling the copyright in some of this material may not have granted the IETF Trust the right to allow modifications of such material outside the IETF Standards Process. Without obtaining an adequate license from the person(s) controlling the copyright in such materials, this document may not be modified outside the IETF Standards Process, and derivative works of it may not be created outside the IETF Standards Process, except to format it for publication as an RFC or to translate it into languages other than English.

Table of Contents

1. Introduction

Applications executing in a device are exposed to many different attacks intended to compromise the execution of the application, or reveal the data upon which those applications are operating. These attacks increase with the number of other applications on the device, with such other applications coming from potentially untrustworthy sources. The potential for attacks further increase with the complexity of features and applications on devices, and the unintended interactions among those features and applications. The danger of attacks on a system increases as the sensitivity of the applications or data on the device increases. As an example, exposure of emails from a mail client is likely to be of concern to its owner, but a compromise of a banking application raises even greater concerns.

The Trusted Execution Environment (TEE) concept is designed to execute applications in a protected environment that separates applications inside the TEE from the regular operating system and from other applications on the device. This separation reduces the possibility of a successful attack on application components and the data contained inside the TEE. Typically, application components are chosen to execute inside a TEE because those application components perform security sensitive operations or operate on sensitive data. An application component running inside a TEE is referred to as a Trusted Application (TA), while a normal application running in the regular operating system is referred to as an Untrusted Application (UA).

The TEE uses hardware to enforce protections on the TA and its data, but also presents a more limited set of services to applications inside the TEE than is normally available to UA’s running in the normal operating system.

But not all TEEs are the same, and different vendors may have different implementations of TEEs with different security properties, different features, and different control mechanisms to operate on TAs. Some vendors may themselves market multiple different TEEs with different properties attuned to different markets. A device vendor may integrate one or more TEEs into their devices depending on market needs.

To simplify the life of developers and service providers interacting with TAs in a TEE, an interoperable protocol for managing TAs running in different TEEs of various devices is needed. In this TEE ecosystem, there often arises a need for an external trusted party to verify the identity, claims, and rights of Service Providers(SP), devices, and their TEEs. This trusted third party is the Trusted Application Manager (TAM).

This protocol addresses the following problems:

2. Terminology

The key words “MUST”, “MUST NOT”, “REQUIRED”, “SHALL”, “SHALL NOT”, “SHOULD”, “SHOULD NOT”, “RECOMMENDED”, “NOT RECOMMENDED”, “MAY”, and “OPTIONAL” in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.

The following terms are used:

This document uses the following abbreviations:

3. Assumptions

This specification assumes that an applicable device is equipped with one or more TEEs and each TEE is pre-provisioned with a device-unique public/private key pair, which is securely stored.

A TEE uses an isolation mechanism between Trusted Applications to ensure that one TA cannot read, modify or delete the data and code of another TA.

4. Use Cases

4.1. Payment

A payment application in a mobile device requires high security and trust about the hosting device. Payments initiated from a mobile device can use a Trusted Application to provide strong identification and proof of transaction.

For a mobile payment application, some biometric identification information could also be stored in a TEE. The mobile payment application can use such information for authentication.

A secure user interface (UI) may be used in a mobile device to prevent malicious software from stealing sensitive user input data. Such an application implementation often relies on a TEE for user input protection.

4.2. Authentication

For better security of authentication, a device may store its sensitive authentication keys inside a TEE, providing hardware-protected security key strength and trusted code execution.

4.3. Internet of Things

The Internet of Things (IoT) has been posing threats to networks and national infrastructures because of existing weak security in devices. It is very desirable that IoT devices can prevent malware from manipulating actuators (e.g., unlocking a door), or stealing or modifying sensitive data such as authentication credentials in the device. A TEE can be the best way to implement such IoT security functions.

TEEs could be used to store variety of sensitive data for IoT devices. For example, a TEE could be used in smart door locks to store a user’s biometric information for identification, and for protecting access the locking mechanism.

4.4. Confidential Cloud Computing

A tenant can store sensitive data in a TEE in a cloud computing server such that only the tenant can access the data, preventing the cloud hosting provider from accessing the data. A tenant can run TAs inside a server TEE for secure operation and enhanced data security. This provides benefits not only to tenants with better data security but also to cloud hosting provider for reduced liability and increased cloud adoption.

5. Architecture

5.1. System Components

The following are the main components in the system. Full descriptions of components not previously defined are provided below. Interactions of all components are further explained in the following paragraphs.

   | Device                                    |
   |                          +--------+       |        Service Provider
   |                          |        |----------+               |
   |    +-------------+       | TEEP   |---------+|               |
   |    | TEE-1       |<------| Broker |       | ||   +--------+  |
   |    |             |       |        |<---+  | |+-->|        |<-+
   |    |             |       |        |    |  | |  +-|  TAM-1 |
   |    |             |       |        |<-+ |  | +->| |        |<-+
   |    | +---+ +---+ |       +--------+  | |  |    | +--------+  |
   |    | |TA1| |TA2| |                   | |  |    | TAM-2  |    |
   |  +-->|   | |   | |        +-------+  | |  |    +--------+    |
   |  | | |   | |   |<---------| App-2 |--+ |  |                  |
   |  | | +---+ +---+ |    +-------+   |    |  |    Device Administrator
   |  | +-------------+    | App-1 |   |    |  |
   |  |                    |       |   |    |  |
   |  +--------------------|       |---+    |  |
   |                       |       |--------+  |
   |                       +-------+           |

Figure 1: Notional Architecture of TEEP

5.2. Different Renditions of TEEP Architecture

There is nothing prohibiting a device from implementing multiple TEEs. In addition, some TEEs ( for example, SGX) present themselves as separate containers within memory without a controlling manager within the TEE. In these cases, the rich operating system hosts multiple TEEP brokers, where each broker manages a particular TEE or set of TEEs. Enumeration and access to the appropriate broker is up to the rich OS and the applications. Verification that the correct TA has been reached then becomes a matter of properly verifying TA attestations, which are unforgeable. The multiple TEE approach is shown in the diagram below. For brevity, TEEP Broker 2 is shown interacting with only one TAM and UA, but no such limitation is intended to be implied in the architecture.

   | Device                                    |
   |                          +--------+       |        Service Provider
   |                          |        |----------+               |
   |    +-------------+       | TEEP   |---------+|               |
   |    | TEE-1       |<------| Broker |       | ||   +--------+  |
   |    |             |       | 1      |<---+  | |+-->|        |<-+
   |    | +---+ +---+ |       |        |    |  | |  +-|  TAM-1 |
   |    | |TA1| |TA2| |       |        |<-+ |  | +->| |        |<-+
   |  +-->|   | |   |<---+    +--------+  | |  |    | +--------+  |
   |  | | +---+ +---+ |  |                | |  |    | TAM-2  |    |
   |  | |             |  |     +-------+  | |  |    +--------+    |
   |  | +-------------+  +-----| App-2 |--+ |  |       ^          |
   |  |                    +-------+   |    |  |       |        Device       
   |  +--------------------| App-1 |   |    |  |       |   Administrator
   |                +------|       |   |    |  |       |
   |    +-----------|-+    |       |---+    |  |       |
   |    | TEE-2     | |    |       |--------+  |       |
   |    |           | |    |       |------+    |       |
   |    | +---+     | |    +-------+      |    |       |
   |    | |TA3|<----+ |    +----------+   |    |       |
   |    | |   |       |    | TEEP     |<--+    |       |
   |    | +---+       |<---| Broker   |----------------+
   |    |             |    | 2        |        |
   |    |             |    |          |        |
   |    +-------------+    +----------+        |
   |                                           |

Figure 2: Notional Architecture of TEEP wtih multiple TEEs

In the diagram above, TEEP Broker 1 controls interactions with the TA’s in TEE-1, and TEEP Broker 2 controls interactions with the TA’s in TEE-2. This presents some challenges for a TAM in completely managing the device, since a TAM may not interact with all the TEEP Brokers on a particular platform. In addition, since TEE’s may be physically separated, with wholly different resources, there may be no need for TEEP Brokers to share information on installed TAs or resource usage. However, the architecture guarantees that the TAM will receive all the relevant information from the TEEP Broker to which it communicates.

5.3. Multiple TAMs and Relationship to TAs

As shown in Figure 2, the TEEP Broker provides connections from the TEE and the Client App to one or more TAMs. The selection of which TAM to communicate with is dependent on information from the Client App and is directly related to the TA.

When a SP offers a service which requires a TA, the SP associates that service with a specific TA. The TA itself is digitally signed, protecting its integrity, but the signature also links the TA back to the signer. The signer is usually the SP, but in some cases may be another party that the SP trusts. The SP selects one or more TAMs through which to offer their service, and communicates the information of the service and the specific client apps and TAs to the TAM.

The SP chooses TAMs based upon the markets into which the TAM can provide access. There may be TAMs that provide services to specific types of mobile devices, or mobile device operating systems, or specific geographical regions or network carriers. A SP may be motivated to utilize multiple TAMs for its service in order to maximize market penetration and availability on multiple types of devices. This likely means that the same service will be available through multiple TAMs.

When the SP publishes the Client App to an app store or other app repositories, the SP binds the Client App with a manifest that identifies what TAMs can be contacted for the TA. In some situations, an SP may use only a single TAM - this is likely the case for enterprise applications or SPs serving a closed community. For broad public apps, there will likely be multiple TAMs in the manifest - one servicing one brand of mobile device and another servicing a different manufacturer, etc. Because different devices and different manufacturers trust different TAMs, the manifest will include different TAMs that support this SP’s client app and TA. Multiple TAMs allow the SP to provide thier service and this app (and TA) to multiple different devices.

When the TEEP Broker recieves a request to contact the TAM for a Client App in order to install a TA, a list of TAMs may be provided. The TEEP Broker selects a single TAM that is consistent with the list of trusted TAMs (trust anchors) provisioned on the device. For any client app, there should be only a single TAM for the TEEP Broker to contact. This is also the case when a Client App uses multiple TAs, or when one TA depends on anther TA in a software dependency (see section TBD). The reason is that the SP should provide each TAM that it places in the Client App’s manifest all the TAs that the app requires. There is no benefit to going to multiple different TAMs, and there is no need for a special TAM to be contacted for a specific TA.

[Note: This should always be the case. When a particular device or TEE supports only a special proprietary attestation mechanism, then a specific TAM will be needed that supports that attestation scheme. The TAM should also support standard atttestation signatures as well. It is highly unlikely that a set of TAs would use different proprietary attestation mechanisms since a TEE is likley to support only one such proprietary scheme.]

[Note: This situation gets more complex in situations where a Client App expects another application or a device to already have a specific TA installed. This situation does not occur with SGX, but could occur in situations where the secure world maintains an trusted operating system and runs an entire trusted system with multiple TAs running. This requires more discussion.]

5.4. Client Apps, Trusted Apps, and Personalization Data

In TEEP, there is an explicit relationship and dependence between the client app in the REE and one or more TAs in the TEE, as shown in Figure 2. From the perspective of a device user, a client app that uses one or more TA’s in a TEE appears no different from any other untrusted application in the REE. However, the way the client app and its corresponding TA’s are packaged, delivered, and installed on the device can vary. The variations depend on whether the client app and TA are bundled together or are provided separately, and this has implications to the management of the TAs in the TEE. In addition to the client app and TA, the TA and/or TEE may require some additional data to personalize the TA to the service provider or the device user. This personalization data is dependent on the TEE, the TA and the SP; an example of personalization data might be username and password of the device user’s account with the SP, or a secret symmetric key used to by the TA to communicate with the SP. The personalization data must be encrypted to preserve the confidentiality of potentially sensitive data contained within it. Other than this requirement to support confidentiality, TEEP place no limitations or requirements on the personalization data.

There are three possible cases for bundling of the Client App, TA, and personalizaiton data:

  1. The Client App, TA, and personnalization data are all bundled together in a single package by the SP and provided to the TEEP Broker through the TAM.
  2. The Client App and the TA are bundled together in a single binary, which the TAM or a publicly accessible app store maintains in repository, and the personalization data is separately provided by the SP. In this case, the personalization data is collected by the TAM and included in the InstallTA message to the TEEP Broker.
  3. All components are independent. The device user installs the Client App through some independent or device-specific mechanism, and the TAM provides the TA and personalization data from the SP. Delivery of the TA and personalization data may be combined or separate.

5.5. Examples of Application Delivery Mechanisms in Existing TEEs

In order to better understand these cases, it is helpful to review actual implementations of TEEs and their application delivery mechanisms.

In Intel Software Guard Extensions (SGX), the Client App and TA are typically bound into the same binary (Case 2). The TA is compiled into the Client App binary using SGX tools, and exists in the binary as a shared library (.so or .dll). The Client App loads the TA into an SGX enclave when the client needs the TA. This organization makes it easy to maintain compatibility between the Client App and the TA, since they are updated together. It is entirely possible to create a Client App that loads an external TA into an SGX enclave and use that TA (Case 3). In this case, the Client App would require a reference to an external file or download such a file dynamically, place the contents of the file into memory, and load that as a TA. Obviously, such file or downloaded content must be properly formatted and signed for it to be accepted by the SGX TEE. In SGX, for Case 2 and Case 3, the personalization data is normally loaded into the SGX enclave (the TA) after the TA has started. Although Case 1 is possible with SGX, there are no instances of this known to be in use at this time, since such a construction would required a special installation program and SGX TA to recieve the encrypted binary, decrypt it, separate it into the three different elements, and then install all three. This installation is complex, because the Client App decrypted inside the TEE must be passed out of the TEE to an installer in the REE which would install the Client App; this assumes that the Client App binary includes the TA code also, otherwise there is a significant problem in getting the SGX encalve code (the TA) from the TEE, through the installer and into the Client App in a trusted fashion. Finally, the personalization data would need to be sent out of the TEE (encrypted in an SGX encalve-to-enclave manner) to the REE’s installation app, which would pass this data to the installed Client App, which would in turn send this data to the SGX enclave (TA). This complexity is due to the fact that each SGX enclave is separate and does not have direct communication to one another.

[NOTE: Need to add an equivalent discussion for an ARM/TZ implementation]

5.6. TEEP Architectural Support for Client App, TA, and Personalization Data Delivery

This section defines TEEP support for the three different cases for delivery of the Client App, TA, and personalization data.

[Note: discussion of format of this single binary, and who/what is responsible for splitting these things apart, and installing the client app into the REE, the TA into the TEE, and the personalization data into the TEE or TA. Obviously the decryption must be done by the TEE but this may not be suported by all TAs.]

5.7. Entity Relations

This architecture leverages asymmetric cryptography to authenticate a device to a TAM. Additionally, a TEE in a device authenticates a TAM and TA signer. The provisioning of Trust Anchors to a device may different from one use case to the other. A device administrator may want to have the capability to control what TAs are allowed. A device manufacturer enables verification of the TA signers and TAM providers; it may embed a list of default Trust Anchors that the signer of an allowed TA’s signer certificate should chain to. A device administrator may choose to accept a subset of the allowed TAs via consent or action of downloading.

PKI    CA    -- CA                                 CA --
        |    |                                         |
        |    |                                         |
        |    |                                         |
Device  |    |   ---    Agent / Client App   ---       |
SW      |    |   |                             |       |
        |    |   |                             |       |
        |    |   |                             |       |
        |    -- TEE                           TAM-------

Figure 3: Entities

 (App Developer)    (App Store)    (TAM)     (Device with TEE)  (CAs)
        |                                            |
        |                               --> (Embedded TEE cert) <--
        |                                            |
        | <------------------------------  Get an app cert ----- |
        |                           | <--  Get a TAM cert ------ |
1. Build two apps:
    Client App
   Client App -- 2a. --> | ----- 3. Install -------> |
      TA ------- 2b. Supply ------> | 4. Messaging-->|
        |                |          |                |

Figure 4: Developer Experience

Figure 4 shows an application developer building two applications: 1) a rich Client Application; 2) a TA that provides some security functions to be run inside a TEE. At step 2, the application developer uploads the Client Application (2a) to an Application Store. The Client Application may optionally bundle the TA binary. Meanwhile, the application developer may provide its TA to a TAM provider that will be managing the TA in various devices. 3. A user will go to an Application Store to download the Client Application. The Client Application will trigger TA installation by initiating communication with a TAM. This is the step 4. The Client Application will get messages from TAM, and interacts with device TEE via an Agent.

The following diagram shows a system diagram about the entity relationships between CAs, TAMs, SPs and devices.

        ------- Message Protocol  -----
        |                             |
        |                             |
 --------------------           ---------------   ----------
 |  REE   |  TEE    |           |    TAM      |   |  SP    |
 |  ---   |  ---    |           |    ---      |   |  --    |
 |        |         |           |             |   |        |
 | Client | SD (TAs)|           |   SD / TA   |   |  TA    |
 |  Apps  |         |           |     Mgmt    |   |        |
 |   |    |         |           |             |   |        |
 |   |    | List of |           |  List of    |   |        |
 |        | Trusted |           |  Trusted    |   |        |
 | Agent  |  TAM/SP |           |   FW/TEE    |   |        |
 |        |   CAs   |           |    CAs      |   |        |
 |        |         |           |             |   |        |
 |        |TEE Key/ |           |  TAM Key/   |   |SP Key/ |
 |        |  Cert   |           |    Cert     |   | Cert   |
 |        | FW Key/ |           |             |   |        |
 |        |  Cert   |           |             |   |        |
 --------------------           ---------------   ----------
              |                        |              |
              |                        |              |
        -------------              ----------      ---------
        | TEE CA    |              | TAM CA |      | SP CA |
        -------------              ----------      ---------

Figure 5: Keys

In the previous diagram, different CAs can be used for different types of certificates. Messages are always signed, where the signer key is the message originator’s private key such as that of a TAM, the private key of trusted firmware (TFW), or a TEE’s private key.

The main components consist of a set of standard messages created by a TAM to deliver device SD and TA management commands to a device, and device attestation and response messages created by a TEE that responds to a TAM’s message.

It should be noted that network communication capability is generally not available in TAs in today’s TEE-powered devices. The networking functionality must be delegated to a rich Client Application. Client Applications will need to rely on an agent in the REE to interact with a TEE for message exchanges. Consequently, a TAM generally communicates with a Client Application about how it gets messages that originate from a TEE inside a device. Similarly, a TA or TEE generally gets messages from a TAM via some Client Application, namely, an agent in this protocol architecture, not directly from the network.

It is imperative to have an interoperable protocol to communicate with different TAMs and different TEEs in different devices. This is the role of the agent, which is a software component that bridges communication between a TAM and a TEE. The agent does not need to know the actual content of messages except for the TEE routing information.

5.8. Trust Anchors in TEE

Each TEE comes with a trust store that contains a whitelist of Trust Anchors that are used to validate a TAM’s certificate. A TEE will accept a TAM to create new Security Domains and install new TAs on behalf of an SP only if the TAM’s certificate is chained to one of the root CA certificates in the TEE’s trust store.

A TEE’s trust store is typically preloaded at manufacturing time. It is out of the scope in this document to specify how the trust anchors should be updated when a new root certificate should be added or existing one should be updated or removed. A device manufacturer is expected to provide its TEE trust anchors live update or out-of-band update to Device Administrators.

When trust anchor update is carried out, it is imperative that any update must maintain integrity where only authentic trust anchor list from a device manufacturer or a Device Administrator is accepted. This calls for a complete lifecycle flow in authorizing who can make trust anchor update and whether a given trust anchor list are non-tampered from the original provider. The signing of a trust anchor list for integrity check and update authorization methods are desirable to be developed. This can be addressed outside of this architecture document.

Before a TAM can begin operation in the marketplace to support a device with a particular TEE, it must obtain a TAM certificate from a CA that is listed in the trust store of the TEE.

5.9. Trust Anchors in TAM

The Trust Anchor store in a TAM consists of a list of CA certificates that sign various device TEE certificates. A TAM will accept a device for TA management if the TEE in the device uses a TEE certificate that is chained to a CA that the TAM trusts.

5.10. Keys and Certificate Types

This architecture leverages the following credentials, which allow delivering end-to-end security without relying on any transport security.

| Key Entity  | Location | Issuer | Checked Against   | Cardinality |
| Name        |          |        |                   |             |
| 1. TFW key  | Device   | FW CA  | A whitelist of    | 1 per       |
| pair and    | secure   |        | FW root CA        | device      |
| certificate | storage  |        | trusted by TAMs   |             |
|             |          |        |                   |             |
| 2. TEE key  | Device   | TEE CA | A whitelist of    | 1 per       |
| pair and    | TEE      | under  | TEE root CA       | device      |
| certificate |          | a root | trusted by TAMs   |             |
|             |          | CA     |                   |             |
|             |          |        |                   |             |
| 3. TAM key  | TAM      | TAM CA | A whitelist of    | 1 or        |
| pair and    | provider | under  | TAM root CA       | multiple    |
| certificate |          | a root | embedded in TEE   | can be used |
|             |          | CA     |                   | by a TAM    |
|             |          |        |                   |             |
| 4. SP key   | SP       | SP     | A SP uses a TAM.  | 1 or        |
| pair and    |          | signer | TA is signed by a | multiple    |
| certificate |          | CA     | SP signer. TEE    | can be used |
|             |          |        | delegates trust   | by a TAM    |
|             |          |        | of TA to TAM. SP  |             |
|             |          |        | signer is         |             |
|             |          |        | associated with a |             |
|             |          |        | SD as the owner.  |             |

Figure 6: Key and Certificate Types

  1. TFW key pair and certificate: A key pair and certificate for evidence of trustworthy firmware in a device. This key pair is optional for TEEP architecture. Some TEE may present its trusted attributes to a TAM using signed attestation with a TFW key. For example, a platform that uses a hardware based TEE can have attestation data signed by a hardware protected TFW key.
  2. TEE key pair and certificate: It is used for device attestation to a remote TAM and SP.
  3. TAM key pair and certificate: A TAM provider acquires a certificate from a CA that a TEE trusts.
  4. SP key pair and certificate: An SP uses its own key pair and certificate to sign a TA.

5.11. Scalability

This architecture uses a PKI. Trust Anchors exist on the devices to enable the TEE to authenticate TAMs, and TAMs use Trust Anchors to authenticate TEEs. Since a PKI is used, many intermediate CA certificates can chain to a root certificate, each of which can issue many certificates. This makes the protocol highly scalable. New factories that produce TEEs can join the ecosystem. In this case, such a factory can get an intermediate CA certificate from one of the existing roots without requiring that TAMs are updated with information about the new device factory. Likewise, new TAMs can join the ecosystem, providing they are issued a TAM certificate that chains to an existing root whereby existing TEEs will be allowed to be personalized by the TAM without requiring changes to the TEE itself. This enables the ecosystem to scale, and avoids the need for centralized databases of all TEEs produced or all TAMs that exist.

5.12. Message Security

Messages created by a TAM are used to deliver device SD and TA management commands to a device, and device attestation and messages created by the device TEE to respond to TAM messages.

These messages are signed end-to-end and are typically encrypted such that only the targeted device TEE or TAM is able to decrypt and view the actual content.

5.13. Security Domain Hierarchy and Ownership

The primary job of a TAM is to help an SP to manage its trusted applications. A TA is typically installed in an SD. An SD is commonly created for an SP.

When an SP delegates its SD and TA management to a TAM, an SD is created on behalf of a TAM in a TEE and the owner of the SD is assigned to the TAM. An SD may be associated with an SP but the TAM has full privilege to manage the SD for the SP.

Each SD for an SP is associated with only one TAM. When an SP changes TAM, a new SP SD must be created to associate with the new TAM. The TEE will maintain a registry of TAM ID and SP SD ID mapping.

From an SD ownership perspective, the SD tree is flat and there is only one level. An SD is associated with its owner. It is up to the TEE implementation how it maintains SD binding information for a TAM and different SPs under the same TAM.

It is an important decision in this architecture that a TEE doesn’t need to know whether a TAM is authorized to manage the SD for an SP. This authorization is implicitly triggered by an SP Client Application, which instructs what TAM it wants to use. An SD is always associated with a TAM in addition to its SP ID. A rogue TAM isn’t able to do anything on an unauthorized SP’s SD managed by another TAM.

Since a TAM may support multiple SPs, sharing the same SD name for different SPs creates a dependency in deleting an SD. An SD can be deleted only after all TAs associated with the SD are deleted. An SP cannot delete a Security Domain on its own with a TAM if a TAM decides to introduce such sharing. There are cases where multiple virtual SPs belong to the same organization, and a TAM chooses to use the same SD name for those SPs. This is totally up to the TAM implementation and out of scope of this specification.

5.14. SD Owner Identification and TAM Certificate Requirements

There is a need of cryptographically binding proof about the owner of an SD in a device. When an SD is created on behalf of a TAM, a future request from the TAM must present itself as a way that the TEE can verify it is the true owner. The certificate itself cannot reliably used as the owner because TAM may change its certificate.

** need to handle the normal key roll-over case, as well as the less frequent key compromise case

To this end, each TAM will be associated with a trusted identifier defined as an attribute in the TAM certificate. This field is kept the same when the TAM renew its certificates. A TAM CA is responsible to vet the requested TAM attribute value.

This identifier value must not collide among different TAM providers, and one TAM shouldn’t be able to claim the identifier used by another TAM provider.

The certificate extension name to carry the identifier can initially use SubjectAltName:registeredID. A dedicated new extension name may be registered later.

One common choice of the identifier value is the TAM’s service URL. A CA can verify the domain ownership of the URL with the TAM in the certificate enrollment process.

A TEE can assign this certificate attribute value as the TAM owner ID for the SDs that are created for the TAM.

An alternative way to represent an SD ownership by a TAM is to have a unique secret key upon SD creation such that only the creator TAM is able to produce a proof-of-possession (PoP) data with the secret.

5.15. Service Provider Container

A sample Security Domain hierarchy for the TEE is shown in Figure 7.

       |  TEE   |
           |          ----------
           |----------| SP1 SD1 |
           |          ----------
           |          ----------
           |----------| SP1 SD2 |
           |          ----------
           |          ----------
           |----------| SP2 SD1 |

Figure 7: Security Domain Hierarchy

The architecture separates SDs and TAs such that a TAM can only manage or retrieve data for SDs and TAs that it previously created for the SPs it represents.

5.16. A Sample Device Setup Flow

Step 1: Prepare Images for Devices

  1. [TEE vendor] Deliver TEE Image (CODE Binary) to device OEM
  2. [CA] Deliver root CA Whitelist
  3. [Soc] Deliver TFW Image

Step 2: Inject Key Pairs and Images to Devices

  1. [OEM] Generate TFW Key Pair (May be shared among multiple devices)
  2. [OEM] Flash signed TFW Image and signed TEE Image onto devices (signed by TFW Key)

Step 3: Set up attestation key pairs in devices

  1. [OEM] Flash TFW Public Key and a bootloader key.
  2. [TFW/TEE] Generate a unique attestation key pair and get a certificate for the device.

Step 4: Set up Trust Anchors in devices

  1. [TFW/TEE] Store the key and certificate encrypted with the bootloader key
  2. [TEE vendor or OEM] Store trusted CA certificate list into devices

6. TEEP Broker

A TEE and TAs do not generally have the capability to communicate to the outside of the hosting device. For example, GlobalPlatform [GPTEE] specifies one such architecture. This calls for a software module in the REE world to handle the network communication. Each Client Application in the REE might carry this communication functionality but such functionality must also interact with the TEE for the message exchange.
The TEE interaction will vary according to different TEEs. In order for a Client Application to transparently support different TEEs, it is imperative to have a common interface for a Client Application to invoke for exchanging messages with TEEs.

A shared agent comes to meet this need. An agent is an application running in the REE of the device or an SDK that facilitates communication between a TAM and a TEE. It also provides interfaces for Client Applications to query and trigger TA installation that the application needs to use.

It isn’t always that a Client Application directly calls such an agent to interact with a TEE. A REE Application Installer might carry out TEE and TAM interaction to install all required TAs that a Client Application depends. A Client Application may have a metadata file that describes the TAs it depends on and the associated TAM that each TA installation goes to use. The REE Application Installer can inspect the application metadata file and installs TAs on behalf of the Client Application without requiring the Client Application to run first.

This interface for Client Applications or Application Installers may be commonly an OS service call for an REE OS. A Client Application or an Application Installer interacts with the device TEE and the TAMs.

6.1. Role of the Agent

An agent abstracts the message exchanges with the TEE in a device. The input data is originated from a TAM or the first initialization call to trigger a TA installation.

The agent may internally process a message from a TAM. At least, it needs to know where to route a message, e.g., TEE instance. It does not need to process or verify message content.

The agent returns TEE / TFW generated response messages to the caller. The agent is not expected to handle any network connection with an application or TAM.

The agent only needs to return an agent error message if the TEE is not reachable for some reason. Other errors are represented as response messages returned from the TEE which will then be passed to the TAM.

6.2. Agent Implementation Consideration

A Provider should consider methods of distribution, scope and concurrency on devices and runtime options when implementing an agent. Several non-exhaustive options are discussed below. Providers are encouraged to take advantage of the latest communication and platform capabilities to offer the best user experience.

6.2.1. Agent Distribution

The agent installation is commonly carried out at OEM time. A user can dynamically download and install an agent on-demand.

It is important to ensure a legitimate agent is installed and used. If an agent is compromised it may drop messages and thereby introduce a denial of service.

6.2.2. Number of Agents

We anticipate only one shared agent instance in a device. The device’s TEE vendor will most probably supply one agent.

With one shared agent, the agent provider is responsible to allow multiple TAMs and TEE providers to achieve interoperability. With a standard agent interface, each TAM can implement its own SDK for its SP Client Applications to work with this agent.

Multiple independent agent providers can be used as long as they have standard interface to a Client Application or TAM SDK. Only one agent is expected in a device.

TAM providers are generally expected to provide an SDK for SP applications to interact with an agent for the TAM and TEE interaction.

7. Attestation

Attestation is the process through which one entity (an attestor) presents a series of claims to another entity (a verifier), and provides sufficient proof that the claims are true. Different verifiers may have different standards for attestation proofs and not all attestations are acceptable to every verifier. TEEP attestations are based upon the use of an asymmetric key pair under the control of the TEE to create digital signatures across a well-defined claim set.

In TEEP, the primary purpose of an attestation is to allow a device to prove to TAMs and SPs that a TEE in the device has particular properities, was built by a particular manufacturer, or is executing a particular TA. Other claims are possible; this architecture specification does not limit the attestation claims, but defines a minimal set of claims required for TEEP to operate properly. Extensions to these claims are possible, but are not defined in the TEEP specifications. Other standards or groups may define the format and semantics of extended claims. The TEEP specification defines the claims format such that these extended claims may be easily included in a TEEP attestation message.

As of the writing of this specification, device and TEE attestations have not been standardized across the market. Different devices, manufacturers, and TEEs support different attestation algorithms and mechanisms. In order for TEEP to be inclusive, the attestation format shall allow for both proprietary attestation signatures, as well as a standardized form of attestation signature. Either form of attesation signature may be applied to a set of TEEP claims, and both forms of attestation shall be considered conformant with TEEP. However, it should be recognized that not all TAMs or SPs may be able to process all proprietary forms of attestations. All TAMs and SPs MUST be able to process the TEEP standard attestation format and attached signature.

The attestation formats and mechanisms described and mandated by TEEP shall convey a particular set of cryptographic properties based on minimal assumptions. The cryptographic properties are conveyed by the attestation; however the assumptions are not conveyed within the attestation itself.

The assumptions which may apply to an attestation have to do with the quality of the attestation and the quality and security provided by the TEE, the device, the manufacturer, or others involved in the device or TEE ecosystem. Some of the assumptions that might apply to an attestations include (this may not be a comprehensive list):

TAMs and SPs must be comfortable with the assumptions that are inherently part of any attestation they accept. Alternatively, any TAM or SP may choose not to accept an attestation generated from a particular manufacturer or device’s TEE based on the inherent assumptions. The choice and policy decisions are left up to the particular TAM/SP.

Some TAMs or SPs may require additional claims in order to properly authorize a device or TEE. These additional claims may help clear up any assumptions for which the TAM/SP wants to alleviate. The specific format for these additional claims are outside the scope of this specification, but the OTrP protocol SHALL allow these additional claims to be included in the attestation messages.

The following sub-sections define the cryptographic properties conveyed by the TEEP attestation, the basic set of TEEP claims required in a TEEP attestation, the TEEP attestation flow between the TAM the device TEE, and some implementation examples of how an attestation key may be realized in a real TEEP device.

7.1. Attestation Cryptographic Properties

The attestation constructed by TEEP must convey certain cryptographic properties from the attestor to the verifier; in the case of TEEP, the attestation must convey properties from the device to the TAM and/or SP. The properties required by TEEP include:

Standard public key algorithms such as RSA and ECDSA digital signatures convey these properties. Group public key algorithms such as EPID can also convey these properties, if the attestation includes a unique device identifier and an identifier for the TEE. Other cryptographic operations used in other attestation schemes may also convey these properties.

The TEEP standard attestation format SHALL use one of the following digital signature formats:

All TAMs and SPs MUST be able to accept attestations using these algorithms, contingent on their acceptance of the assumptions implied by the attestations.

7.2. TEEP Attestation Structure

For a TEEP attestation to be useful, it must contain an information set allowing the TAM and/or SP to assess the attestation and make a related security policy decision. The structure of the TEEP attestation is shown in the diagram below.

                      +------(Signed By)-----------+
                      |                            |
        /--------------------------\               V
      | Attestation   | The         | The                      |
      | Header        | Claims      | Attestation Signature(s) |
             |            |                  |                 |              |
             V            V                  V                 V              V
      +-------------+  +-------------+  +----------+   +-----------------+  +------------+
      | Device      |  | TEE         |  |          |   | Action or       |  | Additional |
      | Identifying |  | Identifying |  | Liveness |   | Operation       |  | or optional|
      | Info        |  | Info        |  | Proof    |   | Specific claims |  | Claims     |
      +-------------+  +-------------+  +----------+   +-----------------+  +------------+

Figure 8: Structure of TEEP Attestation

The Attestation Header SHALL identify the “Attestation Type” and the “Attestation Signature Type” along with an “Attestation Format Version Number.” The “Attestation Type” identifies the minimal set of claims that MUST be included in the attestation; this is an identifier for a profile that defines the claims that should be included in the attestation as part of the “Action or Operation Specific Claims.” The “Attestation Signature Type” identifies the type of attestation signature that is attached. The type of attestation signature SHALL be one of the standard signatures types identified by an IANA number, a proprietary signature type identified by an IANA number, or the generic “Proprietary Signature” with an accompanying proprietary identifier. Not all TAMs may be able to process proprietary signatures.

The claims in the attestation are set of mandatory and optional claims. The claims themselves SHALL be defined in an attestation claims dictionary. See the next section on TEEP Attestation Claims. Claims are grouped in profiles under an identifier (Attestation Type), however all attestations require a minimal set of claims which includes:

7.3. TEEP Attestation Claims

TEEP requires a set of attestation claims that provide sufficient evidence to the TAM and/or SP that the device and its TEE meet certain minimal requirements. Because attestation formats are not yet broadly standardized across the industry, standardization work is currently ongoing, and it is expected that extensions to the attestation claims will be required as new TEEs and devices are created, the set of attestation claims required by TEEP SHALL be defined in an IANA registry. That registry SHALL be defined in the OTrP protocol with sufficient elements to address basic TEEP claims, expected new standard claims (for example from https://www.ietf.org/id/draft-mandyam-eat-01.txt), and proprietary claim sets.

7.4. TEEP Attestation Flow

Attesations are required in TEEP under the following flows:

7.5. Attestation Key Example

The attestation hierarchy and seed required for TAM protocol operation must be built into the device at manufacture. Additional TEEs can be added post-manufacture using the scheme proposed, but it is outside of the current scope of this document to detail that.

It should be noted that the attestation scheme described is based on signatures. The only decryption that may take place is through the use of a bootloader key.

A boot module generated attestation can be optional where the starting point of device attestation can be at TEE certificates. A TAM can define its policies on what kinds of TEE it trusts if TFW attestation is not included during the TEE attestation.

7.5.1. Attestation Hierarchy Establishment: Manufacture

During manufacture the following steps are required:

  1. A device-specific TFW key pair and certificate are burnt into the device. This key pair will be used for signing operations performed by the boot module.
  2. TEE images are loaded and include a TEE instance-specific key pair and certificate. The key pair and certificate are included in the image and covered by the code signing hash.
  3. The process for TEE images is repeated for any subordinate TEEs, which are additional TEEs after the root TEE that some devices have.

7.5.2. Attestation Hierarchy Establishment: Device Boot

During device boot the following steps are required:

  1. The boot module releases the TFW private key by decrypting it with the bootloader key.
  2. The boot module verifies the code-signing signature of the active TEE and places its TEE public key into a signing buffer, along with its identifier for later access. For a TEE non-compliant to this architecture, the boot module leaves the TEE public key field blank.
  3. The boot module signs the signing buffer with the TFW private key.
  4. Each active TEE performs the same operation as the boot module, building up their own signed buffer containing subordinate TEE information.

7.5.3. Attestation Hierarchy Establishment: TAM

Before a TAM can begin operation in the marketplace, it must obtain a TAM certificate from a CA that is registered in the trust store of devices. In this way, the TEE can check the intermediate and root CA and verify that it trusts this TAM to perform operations on the TEE.

8. Algorithm and Attestation Agility

RFC 7696 [RFC7696] outlines the requirements to migrate from one mandatory-to-implement algorithm suite to another over time. This feature is also known as crypto agility. Protocol evolution is greatly simplified when crypto agility is already considered during the design of the protocol. In the case of the Open Trust Protocol (OTrP) the diverse range of use cases, from trusted app updates for smart phones and tablets to updates of code on higher-end IoT devices, creates the need for different mandatory-to-implement algorithms already from the start.

Crypto agility in the OTrP concerns the use of symmetric as well as asymmetric algorithms. Symmetric algorithms are used for encryption of content whereas the asymmetric algorithms are mostly used for signing messages.

In addition to the use of cryptographic algorithms in OTrP there is also the need to make use of different attestation technologies. A Device must provide techniques to inform a TAM about the attestation technology it supports. For many deployment cases it is more likely for the TAM to support one or more attestation techniques whereas the Device may only support one.

9. Security Considerations

9.1. TA Trust Check at TEE

A TA binary is signed by a TA signer certificate. This TA signing certificate/private key belongs to the SP, and may be self-signed (i.e., it need not participate in a trust hierarchy). It is the responsibility of the TAM to only allow verified TAs from trusted SPs into the system. Delivery of that TA to the TEE is then the responsibility of the TEE, using the security mechanisms provided by the protocol.

We allow a way for an (untrusted) application to check the trustworthiness of a TA. An agent has a function to allow an application to query the information about a TA.

An application in the Rich O/S may perform verification of the TA by verifying the signature of the TA. The GetTAInformation function is available to return the TEE supplied TA signer and TAM signer information to the application. An application can do additional trust checks on the certificate returned for this TA. It might trust the TAM, or require additional SP signer trust chaining.

9.2. One TA Multiple SP Case

A TA for multiple SPs must have a different identifier per SP. A TA will be installed in a different SD for each respective SP.

9.3. Agent Trust Model

An agent could be malware in the vulnerable REE. A Client Application will connect its TAM provider for required TA installation. It gets command messages from the TAM, and passes the message to the agent.

The architecture enables the TAM to communicate with the device’s TEE to manage SDs and TAs. All TAM messages are signed and sensitive data is encrypted such that the agent cannot modify or capture sensitive data.

9.4. Data Protection at TAM and TEE

The TEE implementation provides protection of data on the device. It is the responsibility of the TAM to protect data on its servers.

9.5. Compromised CA

A root CA for TAM certificates might get compromised. Some TEE trust anchor update mechanism is expected from device OEMs. A compromised intermediate CA is covered by OCSP stapling and OCSP validation check in the protocol. A TEE should validate certificate revocation about a TAM certificate chain.

If the root CA of some TEE device certificates is compromised, these devices might be rejected by a TAM, which is a decision of the TAM implementation and policy choice. Any intermediate CA for TEE device certificates SHOULD be validated by TAM with a Certificate Revocation List (CRL) or Online Certificate Status Protocol (OCSP) method.

9.6. Compromised TAM

The TEE SHOULD use validation of the supplied TAM certificates and OCSP stapled data to validate that the TAM is trustworthy.

Since PKI is used, the integrity of the clock within the TEE determines the ability of the TEE to reject an expired TAM certificate, or revoked TAM certificate. Since OCSP stapling includes signature generation time, certificate validity dates are compared to the current time.

9.7. Certificate Renewal

TFW and TEE device certificates are expected to be long lived, longer than the lifetime of a device. A TAM certificate usually has a moderate lifetime of 2 to 5 years. A TAM should get renewed or rekeyed certificates. The root CA certificates for a TAM, which are embedded into the Trust Anchor store in a device, should have long lifetimes that don’t require device Trust Anchor update. On the other hand, it is imperative that OEMs or device providers plan for support of Trust Anchor update in their shipped devices.

10. IANA Considerations

This document does not require actions by IANA.

11. Acknowledgements

The authors thank Dave Thaler for his very thorough review and many important suggestions. Most content of this document is split from a previously combined OTrP protocol document [I-D.ietf-teep-opentrustprotocol]. We thank the former co-authors Nick Cook and Minho Yoo for the initial document content, and contributors Brian Witten, Tyler Kim, and Alin Mutu.

12. References

12.1. Normative References

[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, DOI 10.17487/RFC2119, March 1997.
[RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, May 2017.

12.2. Informative References

[GPTEE] Global Platform, "GlobalPlatform Device Technology: TEE System Architecture, v1.1", Global Platform GPD_SPE_009, January 2017.
[I-D.ietf-teep-opentrustprotocol] Pei, M., Atyeo, A., Cook, N., Yoo, M. and H. Tschofenig, "The Open Trust Protocol (OTrP)", Internet-Draft draft-ietf-teep-opentrustprotocol-02, October 2018.
[RFC6024] Reddy, R. and C. Wallace, "Trust Anchor Management Requirements", RFC 6024, DOI 10.17487/RFC6024, October 2010.
[RFC7696] Housley, R., "Guidelines for Cryptographic Algorithm Agility and Selecting Mandatory-to-Implement Algorithms", BCP 201, RFC 7696, DOI 10.17487/RFC7696, November 2015.

Appendix A. History


IETF Drafts

draft-00: - Initial working group document

Authors' Addresses

Mingliang Pei Symantec EMail: mingliang_pei@symantec.com
Hannes Tschofenig Arm Limited EMail: hannes.tschofenig@arm.com
David Wheeler Intel EMail: david.m.wheeler@intel.com
Andrew Atyeo Intercede EMail: andrew.atyeo@intercede.com
Liu Dapeng Alibaba Group EMail: maxpassion@gmail.com