armv8crypto: Canonicalize document description

[ziaee]

Summary

I think arm is calling the (sub?)devices implementing these extensions
"Arm Security Algorithm Accelerators", which sounds good. We should do
the same. Also the previous description was incomplete.

We may want to still keep CPU as a keyword, but I don't really think so.

https://support.arm.com/downloads/-/arm-security-algorithm-accelerators

MFC after: 3 days

[fuz]

Arm Security Algorithm Accelerators

These are separate IP blocks SoC vendors can use in conjunction with the cryptographic instructions. Unrelated.

Note that there is not only the AES extension, but also the PMULL, SHA1, and SHA2 extensions, which may or may not end up being used by this driver (didn't check). For hashing within the kernel, we use the SHA1 and SHA2 extensions regardless of whether arvm8crypto(4) is loaded or not.

share/man/man4/man4.aarch64/armv8crypto.4

32

This matches the Arm Architecture Reference Manual for A-profile Architecture, § A2.12 “The Armv8 Cryptographic Extension.”

[ziaee]

what about without the "v8" since it's in the name?
i'm looking at this [0] and it seems to call it "Arm Cryptographic Extension" more often than not.

https://support.arm.com/documentation/ddi0487/mc/?lang=en

[fuz]

In D59730#1370879, @ziaee wrote:

what about without the "v8" since it's in the name?
i'm looking at this [0] and it seems to call it "Arm Cryptographic Extension" more often than not.

https://support.arm.com/documentation/ddi0487/mc/?lang=en

We still support Armv7, which doesn't have this. Armv8 (both in 32 and 64 bit mode) has this extension, so it sounds like a good idea to stick with armv8.

[ziaee]

okay, sounds good!

[fuz]

Let's wait for @andrew for final feedback as he works at ARM and probably has the best knowledge about this.