US20250390443
2025-12-25
Physics
G06F12/145
The patent application outlines a data processing system leveraging virtualization-based platform protection technology (PPT) to safeguard software modules' code and data. This system utilizes a virtual machine monitor (VMM) to run both untrusted and trusted applications on a single operating system (OS), ensuring the untrusted application cannot access the memory allocated to the trusted application. The VMM employs distinct extended page tables (EPTs) to translate guest physical addresses (GPAs) into host physical addresses (HPAs) for each application, allowing the same GPA to map to different HPAs. This innovation addresses security concerns by isolating trusted applications from potential threats posed by untrusted applications.
In traditional data processing systems, applications use logical addresses to access memory, which the OS translates into linear addresses. A memory management unit (MMU) within the CPU then converts these into physical addresses using page tables. However, if an OS is compromised by malware, it can manipulate page tables to redirect memory access, potentially leading to unauthorized data access or code injection. This vulnerability is a significant concern, as it can result in critical data breaches.
The proposed system enhances security by implementing a platform protection technology (PPT) that operates below the guest OS level. It uses a hypervisor to control memory access, preventing the guest OS from altering page tables. This approach involves a shadow page table (SPT) that the host manages, ensuring that even if the guest OS is compromised, it cannot modify the SPT. By using second-level address translation (SLAT) through extended page tables (EPTs), the system improves efficiency over traditional shadow paging methods, reducing the performance impact associated with multiple address translations.
PPT introduces a trusted execution environment (TEE) that allows dynamic memory allocation and isolation between trusted applications and untrusted OS components. Managed by the hypervisor, PPT facilitates memory sharing without data duplication and ensures memory cleanup after application crashes or exits. This model supports a secure environment by maintaining trusted components within a trust control boundary, excluding the OS and system libraries to mitigate risks of malicious installations.
The architecture comprises multiple levels, starting with a trusted security engine at the hardware level to ensure secure boot processes. Above this, a trusted PPT VMM operates in the hypervisor space, while untrusted drivers and interrupt handlers function in the kernel space. In the user space, untrusted applications can utilize both untrusted and trusted PPT libraries to initiate trusted applications within a secure TEE. This layered approach enhances security by ensuring that only authenticated components are included in the trusted computing base, thereby protecting against potential threats from compromised OS or system libraries.