News › security
By Zayden R., August 23, 2026
Linux 7.3 introduces new AES encryption APIs, promising performance optimizations. This update, led by Google's Eric Biggers, addresses inefficiencies in existing crypto APIs, paving the way for future enhancements.
Linux kernel version 7.3 is making waves with the introduction of new AES encryption APIs, a development led by Eric Biggers from Google. These new library APIs are designed to replace the existing crypto_skcipher and crypto_aead APIs, which have been criticized for their inefficiency and difficulty in use. This is a significant step towards enhancing performance and reducing code duplication within the kernel.
The update encompasses a wide range of AES encryption modes, including ECB, CBC, CBC-CTS, CTR, XCTR, XTS, GCM, and CCM. Biggers highlighted in his pull request that the existing APIs lacked proper library support, creating a gap that these new APIs aim to fill. The implementation involves several key steps: building the new APIs on top of existing single-block AES support, fully documenting them, and transitioning users from the old AES-GCM API to the new, more flexible version.
One of the primary objectives of this change is to integrate the new APIs with the traditional crypto API by adding crypto_skcipher and crypto_aead algorithms. This ensures that the new APIs undergo rigorous self-tests, making them viable for real-world use, especially on systems lacking architecture-optimized code for these modes. The ultimate goal is to migrate architecture-optimized code for these AES modes into the library, which will eventually eliminate redundant code.
While this release lays the groundwork, the full benefits, such as performance improvements and code reduction, are expected to materialize in future cycles. Engineers can anticipate a more streamlined and efficient crypto library as these changes are gradually implemented. The Linux 7.3 update represents a meaningful progression in the ongoing effort to optimize kernel crypto operations.
The Linux Camp teaches these topics as hands-on labs on real virtual machines, verified as you type.