Re: TLS 1.3 HTTPS project for C64

From: Justin <shadow_at_darksideresearch.com>
Date: Sun, 19 Apr 2026 11:46:48 -0500
Message-Id: <39421089-9979-43E1-A046-6AA23AC7222E_at_darksideresearch.com>
I have pushed an update today after discovering and fixing a bug that made it not work properly with IP65/RR-net. I have now validated that it completes the TLS 1.3 handshake and an HTTPS GET at 1 MHz in VICE and on the U64E. I do not have a real RR-net to test with on a real 64. It think it takes ~1h 37m to complete at 1 Mhz. If anyone chooses to try it on real metal or in VICE you should expect a very long wait after the third PROC symbol is displayed.

Thanks,

Justin

> On Apr 18, 2026, at 10:11, Justin <shadow_at_darksideresearch.com> wrote:
> 
> Well, I have this in an initial working state if you run using UCI networking on the U64E so I figured I’d make the repo public while I work on the ridiculously slow process of validating it for 1Mhz C64 with RR-net using VICE. The debug stream feature in the U64E was invaluable in landing this as there were a lot of issue with how packet data flowed into and through the TLS stack that would otherwise have been hard to diagnose.
> 
> In its current state the U64E running at 48Mhz using UCI networking can complete an HTTPS GET using TLS 1.3, speaking to a python HTTPS listener (it is very patient). It is not doing a complete certificate validation etc as I have stubbed out P-384. I have a working P-384 but it is a speed barrier and was blocking validation of the rest of TLS because of how it impacted test cycles. I have other faster math libraries in flight but wanted to land this in a working state before adding chrome. This is built more or less entirely with Claude Code agent teams in 6502 assembly so I am confident there is a lot of inefficient assembly in here, but I’ll point back to adding chrome and then polishing it all. I do intend to add PQC to it as well but right now that is optional in TLS 1.3 and therefore not in this program.  The faster math will almost certainly require an REU.
> 
> I was inspired to do this after reading about trying to add a network based authentication system for C64OS. I figured it would be an interesting challenge to try to land modern crypto on the C64 and it would also be an interesting project to learn to herd coding agents on. I had AES256-GCM-SIV working months ago but then pivoted to try to land completely modern crypto and an over the wire protocol. Doing all of this with Claude Code has definitely met my interesting challenge/learning experience expectations.
> 
> https://github.com/JC-000/c64-https
> 
> Justin


Received on 2026-04-19 18:00:02

Archive generated by hypermail 2.4.0.