consomme/tcp: use memcpy for ring buffer payload copy

[benhillis]

Replace byte-by-byte iterator copy with copy_from_slice on the ring buffer's contiguous slices. This compiles to memcpy and avoids per-byte iterator overhead when constructing outbound TCP packets.

Add View::copy_to_slice() to encapsulate the two-slice copy pattern and remove the now-unused View::iter() method.

[Copilot]

Pull request overview

Optimizes Consommé’s TCP transmit path by replacing per-byte payload copying with slice-based copying that compiles down to memcpy, and centralizes the ring-buffer two-slice copy logic in View.

Changes:

• Add View::copy_to_slice() to copy ring-buffer views into a destination buffer using copy_from_slice.
• Remove the previous byte-by-byte iterator-based payload copy in tcp.rs and use copy_to_slice() instead.
• Remove the now-unused View::iter() API (replaced by the new copy helper).

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
vm/devices/net/net_consomme/consomme/src/tcp/ring.rs	Replaces View::iter() with View::copy_to_slice() to enable efficient contiguous/two-slice copies.
vm/devices/net/net_consomme/consomme/src/tcp.rs	Switches TCP payload construction from byte iteration to View::copy_to_slice() for better performance.

---

You can also share your feedback on Copilot code review. Take the survey.

[benhillis]

I'm going to confirm the old code isn't being optimized to use memcpy already.

[benhillis]

I'm going to confirm the old code isn't being optimized to use memcpy already.

I've confirmed that the previous code generated code that does not use memcpy, but this does.

[benhillis]

Holding off on merging this for now until I can reason about some odd perf behavior we are seeing with this change.

[benhillis]

Ok some final testing, this doesn't make a difference in the send path, but the receive path seems ~20% improvement in throughput.