Rust is secretly taking over chip development

[@letsgetrusty]

Get your *FREE Rust training* : letsgetrusty.com/bootcamp

[@muudus_tv]

Join the Advent of Code 2024. Let's get the advent rusty in 2024.

[@maxwellflitton3973]

I'm a researcher at Kings College London in the surgical robotics department. Kings College Is the largest bioengineering department outside of the USA. We are using rust in our new research

[@franciscogerardohernandezR1979]

Hi there, would you mind sharing a little on how you use rust? I want to switch to rust for numerical optimización but It seems a bit green. Edit: I use NAG libraries, mathpluplus and vcpkg in visual studio.

[@rasputindasilva858]

Don't mess up the dpt! Don't touch any robot or device you have only permission to observe.

[@abacaabaca8131]

Sounds interesting.. a robotic surgeon. Will it replace doctors then?

[@rumplstiltztinkerstein]

@@abacaabaca8131 usually robots like these start as support. On the initial stages, there must be constant monitoring by someone that can be blamed if something goes wrong.

[@boredstudent9468]

​@@abacaabaca8131 No, or at least not in the foreseeable future, human bodies and especially their illnesses are just too individual and messy to properly automate procedures.

[@Slamy4096]

3:30 I work now for quite some time in the automotive industry of Germany. I really like Rust and try to promote it to my manager as often as I can. But Rust really has some issues when it comes to embedded hardware. You need to have a PAC and it helps to have a HAL for embedded Rust. Not every automotive qualified microcontroller has these offered by either a community or by the manufacturer. Not even all relevant commercially grade processors have a good Rust HAL. For example I tried to build something which uses the USB Host Controller of a RP2040. A USB driver is available in C but not in Rust. Delicate interfacing is required. Just my 50 cents. Embedded Rust is not as developed as on a standard operating system.

[@marcelkblanckhema]

It's not that hard. As long as there is a compiler, build your own HAL. It does not have to be as versatile and full featured as open source or vendor provided support packages, just fit the needs of your project.

[@valoubambou4950]

Maybe I'm wrong but I remember when using embassy for a rp2040 that a embassy-usb existed. Have you tried it ?

[@Slamy4096]

@@valoubambou4950 embassy-usb is only handling as a USB device but not as a host.

[@Slamy4096]

@@marcelkblanckhema Developing for a company, resources are not limitless. If the microcontroller already comes with a big library for C, we of course would like to use it to be more efficient. Writing a unique HAL comes also with some risks of introduced bugs you usually won't have.

[@GTLugo]

@@Slamy4096 I'm honestly very curious, please don't take this the wrong way. Why would the drivers being only available in C stop you from consuming it in Rust? Doesn't Rust support ffi with C? Sure you won't be able to guarantee the driver itself won't crash, but at least you can safely reason about your own code while making safe(ish) abstractions around the driver.

[@luiscarlosjayk]

Maaaan, I'd love any of these companies using Rust hire me. I'd like to escape Node.js :'(

[@adrian_sp6def]

Have you tried to get a job there?

[@luiscarlosjayk]

@@adrian_sp6def I've tried in a couple I've found in RustJobs sites, LinkedIn and Discords. But I'm in Mexico, some have discarded me due to the timezone diff.

[@debuti]

I am far off. I have to use C!!!!

[@luiscarlosjayk]

​@@adrian_sp6def I've tried indeed: LinkedIn, RustJobs sites, Discords, a few more.

[@luiscarlosjayk]

@@adrian_sp6def I've tried, yes

[@meanmole3212]

I rewrote my political ideology in Rust, I am now in 100% safe echo chamber

[@swastikacharyya]

also now you have memory safety, congo 😅

[@meanmole3212]

@@swastikacharyya I assume that means I cannot forget the propaganda I consume.

[@swastikacharyya]

@@meanmole3212 exactly 🤣

[@meanmole3212]

@@swastikacharyya cargo check, my mate

[@letsgetrusty]

touché

[@JorgetePanete]

The Rust github is like a newspaper for me, I don't understand most of it but I like that it progresses.

[@basedfacistman]

how brain-dead do you have to be to struggle with the newspaper

[@michaelschnell5633]

Re your question: I am eager to use Rust with my next embedded project. The problem is that support libraries provided by the chip makers (yet) mostly are only available in C. So I would have to port them to Rust even before starting the actual project.

[@franciscogerardohernandezR1979]

@@michaelschnell5633 It Is a bloody hassle, such ports are expert friendly.

[@Arniores]

Thats the Big problem when using rust, most api are thought to be used in c or c++

[@igiona]

True, but look at embassy. It already covers a ton of drivers and quite some chips. And the good news is, that you'll be able to port your code to a different MCU in a much easier way with respect to C. And for me the painless toolchain of Rust is already a winner 🏆 argument. If I could choose, I would ditch CMake.

[@franciscogerardohernandezR1979]

@@igiona I dropped Cmake for vcpkg. Horrible tool.

[@redcrafterlppa303]

That's actually more of an ffi problem than a binding problem.

[@xxportalxx.]

Sketchy thing about the industry getting together to collab the security is that instead of reading about a security vulnerability leading to certain devices being vulnerable, suddenly it will be 'oh shit, EVERYTHING is vulnerable...'

[@lawrencemanning]

“Chip development” in the title is of course misleading. This is low level embedded firmware development.

[@ProgrammingWIthRiley]

Still great progress

[@lombeelo8780]

Thx for saving my time. I sincerely wondered how they replaced SystemVerilog with rust

[@gkvikram]

what is the difference.pls elaborate?

[@lombeelo8780]

@@gkvikram chips are developed using hardware definition languages SystemVerilog/SystemC/HDL/etc, not generic programming languages. (btw my comment got deleted lol)

[@lombeelo8780]

@@gkvikram chips are developed using hardware definition languages SystemVerilog/SystemC/HDL/etc, not generic programming languages. (btw my comment got deleted lol)

[@bobby9568]

Firefox devs made Rust, yet Firefox just had one of the biggest flaws exposed xD

[@dangelgeek]

Awesome, Rust is the future :D

[@KRATOSMORT]

0:40 how do you know that when you were downloading checksums they were tempered?

[@benspencer1163]

This is something I'm very excited about and that's why I transitioned to Rust from JavaScript

[@Xerox482]

@1:25 what is this draw tool ? can u share link

[@yanushkowalsky1402]

is this the same as intel TDX (previously names SGX)?

[@JulianBolivarGaleno74]

On embedded systems, one of the Rust flaws is that doesn't exists one RTOS implemented purely on Rust, I only could find FreeRTOs wrappers

[@franciscogerardohernandezR1979]

@@JulianBolivarGaleno74 wrappers take a whole lot to develop property, I'd say about 40% of the time to completely develop the driver, asuming no skill issues.

[@troylee4171]

Tock?

[@okhsunrog]

RTIC, Tock. Also embassy (not really an rtos, an async framework for embedded, but I really like it)

[@victorpinasarnault9135]

Not yet

[@joshuarowe8410]

You didn't look very hard

[@erick_falker]

Is Rust being used in the robotics development as well?

[@igiona]

Yes

[@mannycalavera121]

Looking suave as hell bro

[@PaulFisher]

The thing you call “intellectual property” isn’t quite what the rest of us mean when we talk about “IP”. Within the semiconductor industry, an “IP” typically refers to essentially a library-a block you can put into your design that performs a certain function. One prominent kind of IP block is a video codec, which is designed by an outside firm and you can embed into your own project to encode or decode video in your chosen format (often many formats). They come in various formats, from VHDL/Verilog source to netlists to raw silicon layouts, which licensees and integrators may be able to modify to varying degrees depending upon the format and the license it is offered under.

[@letsgetrusty]

Thank you for clarifying!

[@LDRTHM]

I wonder, how will this be different from OpenTitan?

[@Kiyuja]

Today I saw a comment on a clip stating that embedded Rust somehow requires more storage than typical C code and thus might not be viable for many devices, I have no idea but maybe it means more sophisticated code. Other than that I see Rusts next big breakthrough with Tauri apps and maybe even video games

[@ethancooper8850]

from memory, most of it is doing runtime checks for a panic. I wouldnt be too eager about Tauri either, the maintainers have left the Linux version pretty far in the past with rather old bindings to WebKitGTK which drastically slows a Tauri app on Linux. Iced or a native UI library is a better alternative if you have the choice.

[@Kiyuja]

@@ethancooper8850 I personally dont care for Linux so that doesnt bother me at all. Creating an agnostic solution for that ecosystem is painful anyway due to the high fragmentation. They offer Chromium as a fallback bundle for Linux. My only "critizism" is the reliance on a local Webview. But hey, maybe they ship Servo in the near future instead. That should improve things and make it more independent while also integrating better in the Rust ecosystem.

[@lawrencemanning]

I suspect, but no data etc, that Rust wouldn’t be that useful on very small embedded devices. Talking in the region of double and triple digit KByte RAM. But since the use of such controllers is diminishing as the years go by it’s no big thing.

[@AbhishekDas-c5u]

How did you get the news?

[@Mike-tm7iq]

Trust him bro 😆

[@Game7Red]

I like your content, but i am seeing less rust programming than blog post reading, i agree this channel covers all rust related talks.There are many awesome rust packages out there, may be we don't know, make video about that or advance rust programing.

[@datascience7920]

what do you think about rust future in blockchain? can it be the same popular like solidity for example?

[@FineWine-v4.0]

Well there's polkadot

[@debuti]

Thanks for all this research effort!!

[@doords]

This is huge. Every time you buy some hardware and connect to your computer in someway, you have no idea what is in the firmware and your antivirus cant scan any of those firmware. So all of those things you buy on ebay, you have no idea what is in it.

[@tui3264]

This means RISC-V is also winning as Caliptra uses it

[@bmx666bmx666]

Unfortunately, what you mentioned about automotive companies isn't entirely true. Some people have dreams of migrating to Rust or starting new projects. I also tried to find support for starting a Rust-based project, but many employees are either unwilling to learn new technologies or languages and are completely fine with C/C++. I was a nobody in one of these companies and couldn't find any support from colleagues. I'm also pretty sure it will take Qualcomm more than 10 years to adopt Rust.

[@flippert0]

I don't like the acronym 'rot' for something you are supposed to trust. But then again, we are using 'rust' for 'iron clad' security.

[@alst4817]

I don’t know much about Rust so please be polite when you correct me! I thought that anytime Rust has to deal with actual hardware, physical memory addresses, it has to be unsafe? Because the underlying architecture doesn’t have Rust types, ownership etc? Doesn’t that make it just as unsafe as C or assembly?

[@wojtekkrupski8583]

In my experience, at the lowest level, it's possible to "shoot yourself in the foot" when writing Rust code. For example, I implemented a custom protocol on top of MODBUS, and if noise creeps into the communication, panics or even UB can occur-despite using safe code and safe libraries. If you don’t sanitize incoming bytes first, issues like serializer panics or out-of-bounds reads/writes can happen. However, I would classify these as program logic errors, since no compiler can protect against random bit flips at the hardware level and programmer should know that. Rust's type system can also be leveraged to manage hardware access and physical memory operations effectively. By wrapping peripherals in types that prevent incorrect operations, you can enforce users of library to safe interaction, making it a solid choice for low-level programming. So, Rust significantly reduces the likelihood of many bugs, and it's generally a pleasure to work with. However, if you have no idea what you're doing, you can always fuck up.

[@benspencer1163]

Believe me, safe codes can still be written in C, just that C doesn't guide you when you're writing unsafe codes... Similar to unsafe Rust, You're expected to know how to implement all the safety rules of Rust on your own without the help of the compiler and the borrow checker before using unsafe Rust.

[@deoradh]

The safety features are more of a code review aimed at being honest about handling memory and exceptional behaviors. It won’t magically make hardware or I/O issues go away, and it won’t stop you when you say “trust me”. It’s good training that you can apply toward how you write in other languages (after you’re done cringing at it).

[@complexity5545]

This is basically true. They usually use rust for API stuff. The underlying drivers can always be hacked and are unsafe|flawed. Most hardware drivers do not need OOP source code. Usually the I/O pinout of hardware is finite and you can statically compile a driver for each permutation or combination which C is the best language for. This is why real programmers hate rust community because they lie about the safety aspect. Its more about creating FFI and function calls that allow other programming languages to embed object files (with restrictions and constraints). The Rust community wants the compiler to basically remove deeply nested pointers and restrict certain calls. But sometimes a hardware guy will have to use a chain of pointers to fit the program within a certain memory size. Basically, the rust community is ushering in a bunch of sorry programmers that know nothing about the lower level programming and they want it that way. It pushes unqualified programmers in positions of power based on phony wisdom. Its basically like hiring a python programmer to make embedded software. Its the mindset behind these (mediocre) guys that destroys individualism and excellent programs/programming. Its going to make most system mediocre by default. It will also create a monopoly on certain industries via licenses and other B.S. that C and C++ don't do. They're opening up a "can of worms." Basically since ~2005 with python, the industry is over-saturated with crappy programmers and unimportant software. All the major software are usually still written in C and C++. With all that being said. Rust is good because its forcing other compilers (like gnu, delphi or whatever) to warn and cry about nested pointers and possible memory allocation problems. Rust is good for APIs and for herding in horrible programmers and that's what the Rust God's want: a language that traps certain people. I don't like the fact that the industry is moving into OOP programming with Rust as the first go to language. Rust sucks at certain applications like graphics and hardware because it won't let you point to a pointer to a pointer without turning off the borrow checker. We're basically wrapping such applications with a huge unsafe block or using some compiler macro to turn if all off. By the end of it all, I expect microsoft to own the core of Rust (ironically through aliases (just like pointers)). It reminds me of Sun > Java > Oracle and Oracle now trying to 100% own Java.

[@nafuranyleniwiec5885]

@complexity5545 Everything has its place. Rust is (probably) the best choice for writing various types of middleware and backends. In embedded its use at the application layer is.... cool. But if I had to write a simple driver for a simple, e.g., thermometer, writing it in C would be faster, simpler and probably as (un)safe as implementing it in Rust. Still, Rust makes it POSSIBLE to write that driver and then use it in an application that will probably be safer and better written than a similar app in C (via a tighter compiler and better, homogeneous tooling). Rust encourages the use of patterns such as data driven development (how cool it works with Rust's enums and error handling) which is also very positive. I doubt that Rust will cause the “Python-effect” that you described, this language requires that the user knows the basics of memory management, so modern programmers will be the beneficiaries of this knowledge.

[@mahor1221]

Thank you for the video!

[@stanrock8015]

Wow. Hadn’t heard of thos

[@Jia-Tan]

So "taking over" was clickbait?

[@RustIsWinning]

Something tells me that you really like these words but I'm not sure why...

[@stefanalecu9532]

​@@RustIsWinning you didn't disprove it's clickbait ;)

[@knnk4000]

Jia Cheong Tan!!

[@thingsiplay]

Linux Why was it developed in secret with Microsoft Windows in mind? If Linux does not support this, how will it affect Linux users?

[@stdprocedure]

"If Linux does not support this": what is "this"?

[@thingsiplay]

@@stdprocedure "this" subject of the video, what was developed secretly.

[@stdprocedure]

@@thingsiplay I'm not sure which subject of the video is being developed secretly if you are talking about the title, I guess it is meant to be "Rust is doing really well in chip development, and no one is talking about it or no one is giving attention to it despite being good", "silently", "underrated"

[@thingsiplay]

​@@stdprocedure Caliptra consists of IP and firmware. Does it only work with specific hardware? Is there a signing process? I hope this is not a situation like Secure Boot, which you often need to disable in Linux.

[@-boiadeiro-]

ну да, корпорации так хотят для нас безопасности, верим... потом не удивляйтесь когда в один день все девайсы выйдут из строя, продолжайте доверять этим ребятам...

[@Singlton]

This means R.I.B installing linux

[@kirakorunov5407]

A comment to promote the channel

[@rezah336]

this is 100% a big brother thing, all chips will be monitored

[@alexisfibonacci]

As long as it is connected, it is monitored. How do you think you get firmware updates? They only trying to make the process more secure and keep out bad actors - aka typical memory issues with C and C++.

[@rezah336]

@@alexisfibonacci big business are the big actors, why do you trust them?

[@RustIsWinning]

I looove big brother business companies

[@Takatou__Yogiri]

Bro. now there will be more rust hater. i can't imagine how many more will spawn in the future.

[@romanstingler435]

But Rust is like an avalanche, it is too late to stop it. They will be drowned because everything will be written in Rust :P

[@psi4j]

@@romanstingler435I can’t wait. Eventually the last boomer engineer will give up C and C++.

[@yuan.pingchen3056]

@@romanstingler435 except linux

[@RustIsWinning]

They can all come because I'm fighting every single one of them until they mald lmao

[@FineWine-v4.0]

The best part is watching C & C++ cultists cope & seethe & call chip-development woke just because it uses Rust😂

[@JoJoDramo-ih7qk]

yeah, sounds about right having memory safety without much perf penalty for a security thingy., but linux c-ansi devs would say its not necessary if you are good enough, hahaaaaaaa

[@readwrite-p5g]

Better better and better improve large strong

[@mohit84604]

This man is going to do everything just to advertise RUST.

[@AdamFiregate]

Correct. Because this is a channel about the Rust language.

[@ev3rybodygets177]

with people installing to rootkits so they can "get good and shoot noobs" who gives a shit

[@shifureisaikyou2055]

The fact that Rust has to be named seems redundant. if rust is good for that use it, if it isnt dont. Rust feels like down your throat kind of software

[@RustIsWinning]

Did not watch the video lol

[@vantadaga]

I will never understand the rust cult, why do people build their entire identity around a singular language

[@Polynuttery]

👍👍👍✝✝✝🇳🇿🇳🇿🇳🇿