3 ms·

Here's some context - the actual fix's commit message itself https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2c66ca3949dc701da7f4

Here's some context - the actual fix's commit message itself https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2c66ca3949dc701da7f4c9407f2140ae425683a5 https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/lin... Some security modules that depend on cpu features, the example is camellia_aesni_avx_x86_64, failed to load and caused the performance regression. So you'd be affected if you use something like that.

arthur2e5 3y ago
> camellia_aesni_avx_x86_64 An interesting point here is that AES-NI can be used to accelerate a host of things other than AES. In this case, it's because the S-box can take advantage of the AES S-Box (SubBytes) instruction because of affine isomorphism or something: https://github.com/jkivilin/camellia-simd-aesni https://github.com/jkivilin/camellia-simd-aesni; https://kernel.googlesource.com/pub/scm/linux/kernel/git/shawnguo/linux/+/imx/soc/arch/x86/crypto/camellia-aesni-avx2-asm_64.S https://kernel.googlesource.com/pub/scm/linux/kernel/git/sha.... Similar acceleration has been done with SM4, the Chinese analogue of AES. https://github.com/mjosaarinen/sm4ni https://github.com/mjosaarinen/sm4ni