The Windows integrity mechanism extends the security architecture by defining a new access control entry (ACE) type to represent an integrity level in an object’s security descriptor. The new ACE represents the object integrity level. An integrity level is also assigned to the security access token when the access token is initialized. The integrity level in the access token represents a subject integrity level. The integrity level in the access token is compared against the integrity level in the security descriptor when the security reference monitor performs an access check. Windows Vista uses the AccessCheck function to determine what access rights are allowed to a securable object. Windows restricts the allowed access rights depending on whether the subject‘s integrity level is higher or lower than the object, and depending on the integrity policy flags in the new access control ACE. The security subsystem implements the integrity level as a mandatory label to distinguish it from the discretionary access under user control that ACLs provide.

（Windows完整性机制通过在对象的安全描述符中定义一个代表完整性级别的ACE来扩展安全体系。新增加的ACE代表了对象的完整性级别。当访问令牌初始化的时候，完整性级别也会分配给访问令牌。访问令牌中的完整性级别代表了主体的完整性级别。当安全参考监视器执行访问检查的时候，访问令牌中的完整性级别与安全描述符中的完整性级别进行比较。Windows Vista使用AccessCheck函数来决定哪些访问可以被允许访问对象。Windows根据主体的完整性级别高于或者低于对象的完整性级别，以及根据新增的ACE中的完整性策略标志位来限制访问权限。安全子系统将完整性级别作为强制性标签，与ACL中的自主访问ACE区分出来。）

The Windows integrity mechanism enables a number of important scenarios in Windows Vista. In order to address the requirements, the Windows integrity mechanism‘s design had to meet the following goals.

• Integrity levels must be assigned automatically to every security access token during access token creation, so that every process and thread has an effective integrity level for access control.
• The security subsystem automatically assigns mandatory labels to specific object types.
• The system must use as few integrity levels as possible, to keep the basic architecture simple to understand and use.
• Integrity policy must be flexible to meet the access requirements of different object resource managers, and to allow for future extensibility.
• Integrity mechanism must integrate with existing security architecture to minimize impact to the large legacy of system and application code that depends on Windows security.
• There is no requirement for administrators or users to configure integrity levels for the enforcement mechanism to work correctly.

（Windows完整性机制在Vista中实现了一些重要的场景。为了表达需求，Windows完整性机制的设计不得不满足以下目标

• 在访问令牌创建的时候，完整性级别必须自动分配给每个访问令牌，这样每个进程和线程可以拥有一个有效的完整级别，实现访问控制
• 安全子系统自动分配强制性标签给特定的对象类型
• 系统必须使用尽可能少的完整性级别，来保证基本的安全体系便于理解和使用
• 完整性策略必须灵活，来满足不同类型对象的资源管理器的访问需求，并且允许以后的扩展
• 完整性机制必须与现存安全体系整合，对于大规模系统和基于Windows安全的代码的应用程序的影响最小化
• Windows管理员或者用户不需要其他操作就可以保证完整性机制正确工作）

The Windows integrity mechanism meets these goals by defining a new mandatory label ACE type for assigning an integrity level to objects. Details of this structure are described in a later section of this paper. However, the mandatory label ACE defines an object integrity level without changes to the existing security descriptor data structure definition or to the commonly used discretionary access control list.

The Windows integrity mechanism is based on a mandatory label that the operating system assigns in order to differentiate it from discretionary access under user control. Discretionary access control allows the object owner, or the group that is granted permission, to change the object‘s access permissions. Windows provides a graphical user interface (UI) for advanced users to view and modify the security permissions (represented by the discretionary ACL) on objects, such as files and registry keys. Mandatory labels are always assigned to specific objects, and there are controls on how the object creator can set or initialize the label on object creation. No graphical UI for managing integrity labels was implemented for Windows Vista because label management is available or necessary for relatively few areas.

The purpose of the Windows integrity mechanism is to restrict the access permissions of applications that are running under the same user account and that are less trustworthy. Unknown, potentially malicious code that is downloaded from the Internet must be prevented from modifying system state, changing user data files, or manipulating the behavior of other application programs. The Windows security subsystem assigns a simple hierarchy of integrity levels to code running at different privilege levels for the same user. Previous versions of Windows can adjust the security access token privileges of an application process, although such adjustment is not common. Before Windows Vista, most applications ran using an administrative account with full administrator rights. Windows Vista incorporates the concept of least privilege by enabling broader use of standard user accounts. User Account Control (UAC) in Admin Approval Mode for administrator accounts means that multiple applications on the same desktop are running with different privilege levels. For example, Protected Mode Internet Explorer uses the integrity mechanism to run the Web browser in a process with limited access permissions.

The primary security problem that the Windows integrity mechanism addresses is unauthorized tampering with user data and, indirectly, with system state. A secondary problem the integrity mechanism helps with is information disclosure. However, information disclosure is prevented only with respect to access to process address space. Information sharing is very common between Windows applications, and convenient information sharing between applications is fundamental to the user experience. One example is copy and paste. Rigid security boundaries on information sharing between applications running under the same user account can severely affect application compatibility and user experience.

Untrustworthy code can try to modify user data in many ways. Some attacks may try to manipulate data directly by creating, modifying, or deleting files. Other attacks target another process running at higher privilege, with the goal of getting arbitrary code to execute in another application that does have the required level of access. There are many types of cross-process attacks. Because of the wide range of application design and implementation, the integrity mechanism cannot provide a complete isolation barrier. The Windows integrity mechanism is not intended as an application sandbox. However, it can be one of the security tools that application developers use to restrict the behavior of less trustworthy applications.

Some traits of the Windows integrity mechanism are similar to earlier integrity models for computer security. However, the Windows integrity mechanism is designed primarily to address tampering or elevation of privilege in the highly collaborative Windows application environment. Previous integrity models were more concerned with maintaining integrity of trustworthy processes by enforcing policies that prevent the reading of untrusted data.

The Biba security model is based on a hierarchy of integrity labels and the access policies that are allowed when a subject integrity level dominates the object integrity level. The Windows integrity mechanism resembles the Biba model in the following ways:

• It uses a hierarchy of integrity labels (integrity labels are not the same as security labels in the Bell-LePadula model).
  
• The system uses a set of ordered subjects, objects, and integrity levels.
  
• The subject’s integrity level dominates (is greater than or equal to) the object’s integrity level.
  
• Integrity policies inhibit access to objects but are not used primarily to limit the flow of information.
  
• Preventing information disclosure is not a goal of the integrity mechanism in Windows Vista.
  

The Windows integrity levels are ordered so that a lower value indicates less trustworthiness, and a higher value indicates greater trustworthiness. A lower-level subject cannot modify a higher-level object. The subject’s integrity level is not dynamic. For example, the integrity level of a subject does not change to a lower value if the process reads data from a low-integrity object. The strict integrity model in Biba does not allow a higher-integrity process to read lower-integrity data. This is sometimes called a “no-read-down” integrity policy. The Windows integrity policies, which are described in more detail below, do not inhibit or prevent higher-integrity subjects from reading or executing lower-integrity objects. There are many examples of attacks where reading malformed, untrusted input data results in an exploit of a vulnerability in an application and arbitrary code execution. The Windows integrity mechanism does not inhibit or prevent reading data at any level. Windows does not enforce a strict integrity policy described in the Biba model. The integrity design assumes that processes that are designed to handle untrusted data from an unknown or untrusted source are running at a lower integrity level, or that untrusted data is verified before use. However, the Windows integrity mechanism does not enforce that constraint.

The Windows integrity mechanism does not implement a dynamic, or "Low-Water-Mark," policy. A dynamic policy changes the integrity level of the subject as the subject opens lower-integrity objects. An issue with dynamic integrity is when a high-integrity process obtains open handles to many objects based on high integrity, and then suddenly becomes a lower-integrity subject after it opens a particular low-integrity file. Forcing all open handles to higher-integrity objects to close when the integrity level changes significantly affects the application‘s behavior. The dynamic lower-integrity process itself becomes a target object of other processes at the lower integrity level. Such processes might now be able to modify the behavior of the application (at the same lower level) that has open handles to higher-integrity objects.

The Windows integrity mechanism is not designed for integrity protection of data to support military or commercial requirements as described in the Clark-Wilson model. The Windows implementation of integrity controls does not build on the concepts of Constrained or Unconstrained Data Items and certified Transformation Procedures. However, these concepts are useful for application designers when they consider information flow from untrusted sources into higher-integrity processes.

Although the Windows integrity mechanism is similar to earlier integrity models in computer security, Windows Vista does not try to implement any of the models. Instead, the Windows integrity mechanism limits access permissions that are available to processes running with different privilege or trust levels. For more information about earlier work on integrity models by Biba and Clark-Wilson and dynamic integrity policies, see Windows Integrity Mechanism Resources.