This version introduces enhancements and fixes for the AArch64 architecture. Changes were made to correctly handle the Scalable Matrix Extension (SME) state by clearing the processor's ZA state before clone and clone3 system calls to conform to the procedure call standard. The vector math library was updated, with fixes for instability in the AdvSIMD implementations of sinh and tan, corrections for SVE tanpi and pow functions, and optimizations to SVE scalar callbacks.

On the ppc64le architecture, optimized versions of strcmp and strncmp for Power10 processors were restored. This change resolves recent security issues by avoiding the use of non-volatile vector registers. A bug in the rawmemchr function that modified a non-volatile register was also corrected.

For the x86 architecture, support was added for new Intel processors, including Nova Lake and Wildcat Lake. A performance bug in wmemset was fixed; an incorrect conditional check caused the function to fall back to a slower SSE2 implementation on modern CPUs. This has been corrected to allow the use of faster AVX2 and AVX512 code paths.

Several general bug fixes were also included. The threading library now correctly manages stack guard pages for cached threads, preventing potential memory advisory failures. The Name Service Switch (NSS) functionality was updated to properly handle errors during group entry merges. I/O library headers were updated to align certain constants with definitions in the Linux kernel, resolving preprocessor warnings.