SiFive Gives Its WorldGuard Security Model to the RISC-V Community | Hacker News Reader

SiFive Gives Its WorldGuard Security Model to the RISC-V Community

allaboutcircuits.com

89 pointshasheddan28 comments

I'd like to get started with RISC-V chips. Can anyone reccommend a good dev board similarly priced or spec'd to the Pi Pico? Would be nice if the standalone SoC are easy to acquire as well.

Which is the best cost wise? Which is the best performance wise? Power consumption wise?

I don't need WiFi/Bluetooth though it would be nice. Low cost is preferrable over sheer performance. Ideally I get a fair amount of GPIO (>30) and it supports 3V or 5V.

I have really enjoyed the ESP32-C3 (https://www.espressif.com/en/products/socs/esp32-c3)

I have written a few posts that may help you get started with it: - https://danielmangum.com/posts/risc-v-bytes-exploring-custom... - https://danielmangum.com/posts/risc-v-bytes-zephyr-on-esp32/ - https://danielmangum.com/posts/risc-v-bytes-zephyr-before-ma... - https://danielmangum.com/posts/risc-v-bytes-semihosting-zeph...

https://takahirox.github.io/riscv-rust/

This is really cool. This is riscv running under wasm in your browser and else where. Paired with a web repl and some JS peripherals, should should probably be the top answer.

This is much more usable, Full Fedora, with everything you'd expect:

https://bellard.org/jslinux/vm.html?cpu=riscv64&url=fedora33...

WCH has a bunch of new ones that have come out recently, like the CH32V003 and CH32V307 [1]. I would start there, they are close enough and could probably run micropython since they have some that have reasonable amounts of SRAM.

I think they are mostly looking to replace STM32s though.

[1] http://www.wch-ic.com/products/CH32V003.html

Pine64 also has a RISC-V SBC. I like their products, but I did not buy a Star64 yet.

https://pine64.com/product-category/star64/

I don't know why this comment was dead. Pine64 is legit and the Star64 looks like a viable option.

Anyone with experience on the Star64? I've had some rough times in the past with Pine64 products, from boards dying to the pinebook pro suddenly unable to boot from sd card. Better documentation would have helped. I didn't have hours to comb through forums and discords/matrices.

Still have a major warm place in my heart for Pine64 though, and overall have felt like they're a good value.

I guess the question was about board with specs similar to Pi Pico (~264kB of RAM) and the Star64 is more similar to RPi4 (~4GB of RAM).

Right, the Star64 is like the full-size Pi; the equivalent to the Pico would be the Ox64 line.

I've been happy with my mango pi mq pro, but that might be more powerful than what you are looking for. It's very similar to the raspberry pi zero w.

Probably the closest to a Pico in terms of being symmetrical would be the $9 Milk-V duo with two 1 GHz C906 cores (with vector units), and 64 MB RAM. It's obviously far more powerful than the Pico. And isn't available outside China yet.

Taking a step down to things that are actually available, the Sipeed M1s Dock and Pine64 Ox64 use the BL808 chip which has one 320 MHz 32 bit core, one 480 MHz 64 bit core (with vector processor), and a 3rd smaller 32 bit core that is intended to be dedicated to running the WIFI/BT. This chip also has 64 MB of on-chip RAM. The Ox64 is $6 or $8 depending on how big a flash chip (2 MB or 16 MB) it has. The M1s is a little more.

Going back to 2019, the Kendryte K210 chip has two 400 MHz 64 bit cores and 8 MB of RAM. There have been a ton of boards using it, especially from Sipeed but it's a bit more expensive than the newer chips.

There are tons of cheaper boards, but they tend to be single core and only a few KB to few hundred KB of RAM.

You of course want to go with RUST + RISCV

https://www.mouser.com/ProductDetail/Espressif-Systems/ESP32...

ESP32-C3-DevKit-RUST-1

$21 shipped from Mouser.

https://github.com/esp-rs/esp-rust-board

I'm using the StarFive VisionFive 2 and its working quite well, there are debain and ubuntu images available as well as some other distributions, generally most of the boards features are working fine and there is a community around it actively improving software support.

For your requirements it might be a bit overkill in performance as its more in the neighborhood of a RasperryPi 3/4 in terms of size, power consumption and performance.

Esp32-C3.

Arduino, micropython, rust, and bog standard C/C++ are all well supported.

BLE/Wifi support, dev boards < $5usd are common.

https://www.espressif.com/en/products/socs/esp32-c3

There seems to be a good explanation of WorldGuard here: https://youtu.be/Pj4YHJjcjvI

I believe this is equivalent to Intel SGX. Having this open source is encouraging.

Not specifically talking about WorldGuard, but this video comparing Intel SGX, Arm TrustZone, and RISC-V PMP is quite good: https://youtu.be/MREwcSo0uz4

Not even remotely comparable. A trivial analogy would be TrustZone with more than one secure world per core.

It’s not a terrible idea, but it’s far from a complete solution.

Does anyone have a "Getting started" with Worldguard? I've wanted to experiment with it, but: 1. Didn't find any easily available hardware with it 2. Didn't see actually how to use it.

> In closed-source designs, the lack of access to information acts as a barrier for adversaries

ROFL. Acts as a barrier for script kiddies, until the tool is released on Reddit, then you're just screwt.

They further elaborate that security by obscurity is a poor approach though

We see time and again that more eyes on open source doesn't count for a ton either. We had a relatively simple use after free in the kernel just a couple months ago that had been there for a long time. The benefit of open source is that we could demand and verify good testing and analysis for foundational bits of code, but we don't.

> we could demand and verify good testing and analysis for foundational bits of code, but we don't.

Who would you take these demands to? Most foss software is written and managed by volunteer maintainers, if you start demanding things from them they will rightfully tell you to jog on.

Even when they are not volunteers, they are likely employees of companies that fund this development largely because they require and rely on the software. In which case, they likely have priorities set that are aligned with that company and the needs of other users are low priority at best, but more likely largely irreverent.

The people who donthe testing and analysis don't have to be the authors. If I do something for fun, and a faang makes a double digit percentage of the world gdp depend on it's security, then I would hope we demand the company who benefits ensure that security in an open way or choose other software. But that would cost more.

> RISC-V benefits from a lack of obscurity in this sense since there is more visibility into the system design, resulting in more known and addressed vulnerabilities and hence a more robust system.

Ditto. What a terrible statement to open with. Not instilling me with a lot of confidence. Shannon's Maxim states that the enemy knows the system...

As the cracking and modding scene says: "Source code? We don't need no stinkin' source code!"

Reply on news.ycombinator.com