Brian Smith

> @briansmith I'm completly open to make the PR. It would just be very helpful if you could show me a single location where action is required and what needs...

Great! In terms of testing, look at how qemu is used in CI to test on older CPUs. Then you can run `cargo test` under QEMU with a very old...

After PR #2295 is merged, it will be very clear what needs to be done to resolve this properly: * Review `ChaCha20_ctr32_nohw` to see if it uses SIMD instructions. *...

> [@briansmith](https://github.com/briansmith) would it help if a PR was made with the patch that debian is using? It looks okay to me. > > Here is the comparison, it needs...

PR #2346 lays the foundation for adding dynamic detection of SSE/SSE2 to *ring* to targets that don't enable SSE/SSE2 by default. In that PR (and the ones leading up to...

Hi @plugwash, thank you for your work adapting *ring* to work with Debian's platform support policies. I saw your patch that you linked above. I think it would be great...

Also, since you're a Debian maintainer, this is just as good of a place as any to note that I intend for the 0.17.9 release to also have MSRV 1.63,...

> ``` > } else if cfg!(target_feature = "sse2") { > chacha20_ctr32_ffi!( > unsafe { (1, (), &mut [u8]) => ChaCha20_ctr32_nohw }, > self, counter, in_out.copy_within(), ()) > } else...

AFAICT, montgomery.rs is the only place that needs to be patched for SSE2, under the very reasonable assumption that every AES-NI-capable CPU will support SSE2 (the CPU and OS must...

To make the patching easier, I've submitted PR #2358.