White Paper

White Paper: Hidden Risks of UEFI Secure Boot—and How RelianceOne™ Offers a Stronger Security Foundation

Issue link: https://read.uberflip.com/i/1544840

Contents of this Issue

Navigation

Page 3 of 4

WHITE PAPER The Hidden Risk of UEFI Secure Boot mrcy.com 4 Confidentiality and Key Management The key material unsealed by the TPM during measured boot ser ves dual purposes: authenticating your OS kernel, initramfs, and kernel command-line parameters, and encr ypting their contents. This encr yption protects sensitive key material, while ensuring its release only within a trusted environment, and ser ves to obscure the OS configuration from adversaries – making it significantly harder to develop and deploy targeted exploits. This capability enables the secure release of credentials and key material to the RelianceOne for Linux runtime, Hardware Security Modules (HSMs), Self-Encr ypting Drives (SEDs), or software-based Full-Disk Encr yption (FDE) solutions. The approach is ideal for customers with both secure boot and data-at- rest requirements who need automatic unlocking for unsupervised systems, eliminating the operational burden of manual intervention while maintaining security. The combination of TPM boot measurements creates a powerful binding – with encrypted drive contents tied not only to the specific host hardware, but also to the verified OS and security configuration. This ensures that both sensitive data and key material remain protected even if the drive is physically removed or the system configuration is altered in a way which might compromise runtime security. Secure Update with Rollback Prevention A rollback attack occurs when an adversary attacker reinstalls an older, vulnerable version of software that has already been superseded replaced by a security update. By doing so, the attacker can exploit known weaknesses that were previously patched. These attacks pose a significant danger, especially when they target high-privilege components such as OS kernels or hypervisors. By bypassing updates in this manner, an attacker effectively undoes important security measures and jeopardizes the entire platform. While it is true that there are some implementations of UEFI secure boot that have options for preventing rollback, these are uncommon and these implementations suffer from the flaws discussed in the sections above. In general, if a signed older kernel continues to generate an expected signature – it will be allowed to execute. Using hardware-enforced monotonic counters, RelianceOne Trusted Platform permanently invalidates any older system software version once a newer version is deployed. This eliminates the shortcomings of software-only rollback prevention, maintaining a continuous state of defense as systems evolve. Because the version tracking resides securely inside the TPM, attackers with elevated privileges or even physical access cannot reset it. This makes RelianceOne Trusted Platform especially valuable for any environment that applies periodic security updates to critical components, including hypervisors, OS kernels, the initramfs, and other configuration items. Each update not only deploys fresh security patches but also permanently excludes outdated versions, preserving the integrity of the trusted computing base. BEYOND SECURE BOOT – MAINTAINING TRUST DURING RUNTIME Traditionally, an OS kernel enforces security boundaries between user-space applications. The Linux kernel uses a monolithic architecture, meaning that all kernel code operates with full, unrestricted control over the entire system. Technologies such as secure boot and kernel module signature validation ensure the integrity of kernel code primarily during the boot process. However, as we explored when discussing the DRTM, these protections do not prevent runtime attacks. An attacker may exploit kernel vulnerabilities during runtime to achieve privilege escalation and execute arbitrary kernel code from user space. If successful, the attacker can modify kernel code to conceal their activities and subvert other applications. These attacks can be executed entirely in memory, leaving no modified files or forensic evidence on disk. The Linux kernel employs several hardware-based technologies to protect itself and user-space processes. Key runtime-configurable CPU features include: ▪ WP (Write Protection) Marks read-only kernel code and data as immutable to prevent unintentional or malicious modification ▪ SMAP/SMEP (Supervisor Mode Access/Execution Prevention) Prevents the kernel from unintentionally executing user- space code or accessing user-space data ▪ CFI (Control Flow Integrity) Prevents exploitation techniques such as Return-Oriented Programming (ROP) attacks On x86-based systems, these technologies are configured through bits in the CR0 and CR4 control registers. Under normal operation, once enabled during boot, these features remain active throughout runtime. However, an attacker who gains kernel-level access can disable these protections by modifying the control registers, eliminating critical security barriers.

Articles in this issue

Links on this page

view archives of White Paper - White Paper: Hidden Risks of UEFI Secure Boot—and How RelianceOne™ Offers a Stronger Security Foundation