Rust Programming Language FOSDEM Rust Devroom 2018, 6 Feb 2018

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Frank Rehberger - Rust - embedding WebAssembly for scripting

Description

WebAssembly in non-Javascript environments

Rust is associated with performance, memory safety and control of memory usage. Embedding dynamic runtimes such as for Lua or Javascript for dynamic scripting within the Rust-App would introduce an huge overhead. A WebAssembly-engine seems to be a good choice as compact and portable runtime environment. JIT-compiler may be used in future to transform WASM files to native code.

The talk will present WebAssembly technology and the benefits and pitfalls integrating it into a Rust-app. Small routines are implemented in C++/Rust and compiled to Wasm. The Rust-App is loading the wasm-code as plugin at runtime to execute dynamic tasks.

Frank is Software Consultant with the focus onto back ends, messaging, security, automotive, embedded and CI. Already having experience with various programming languages in small and large development projects, Frank is enthusiastic about the combination of features of the programming language of Rust. Hands-on projects permit the evaluation of these features.

Rust devroom @ FOSDEM 2018
https://fosdem.org/2018/schedule/event/rust_embedding_wasm/

These talks have been recorded at FOSDEM (https://fosdem.org/)
This work is licensed under the Creative Commons Attribution 2.0 Belgium Licence.
To view a copy of this licence, visit http://creativecommons.org/licenses/by/2.0/be/deed.en

A

B

B

Okay, so welcome to everybody. My name is Frank Rebecca, so I'm living in Berlin, I'm software consultant and some time ago, I got attracted by rust, and you assume everybody here knows rust or just to repeat it. So its key features are support for concurrency memory, safety performance and especially memory management, deterministic management of resources. There is no garbage collector built in which is a very important feature. Its favor stack allocation, no implicit boxing, so there is no hidden location of heap memory and causing delays in processing so and I asked myself.

B

So if we want to- or did you ask yourself, did you ever ask if I've got such a cool system? How could I integrate customizable features that are performed some scripting and so allowing me to change behavior during runtime without restarts, without reprogramming so and I I thought about a small message broker, which should be customizable, so we would need byte level, manipulation, deterministic, management of resources, of course, because we don't want to get a step back back to the system.

B

So, if I want to, if I want to have for heap allocation I, could you use Java whatever or JavaScript from the beginning and implement everything in JavaScript? So this is a very important feature for me. Small footprint, security safety and customizable, and so I had to look around so what's available. What could I use and so I had a look at Lua, Python and JavaScript as interpreters, and there is always a high complexity dynamic heap allocation, it's always involved, so we can't get a get out of it.

B

There's an execution overhead and similar with JavaScript to the age. So there is even a more higher complete city, because it's everything during start-up and then the the benefit is there was a lower lower overhead execution overhead during runtime. But none of these really allow me to do the domestic resource management they all kick in. They all bring in a garbage collector and in fact they destroy the benefits of the rust environment or the application executing this.

B

This execution engine and so I had a look finally at forth and I stumbled across webassembly and which is very low. Level has low overhead or almost no overhead, and the difference is that the other concepts they always incorporate. A full powers are pausing, abstract, syntax tree, optimizing the code getting intermediate representation and then, if it's treated getting some execution code, but still JavaScript still is based. So it's not. It doesn't favor stack allocation, javascript favors heap allocation, so we always end up with a system.

B

Allocating a lot of memory and I don't want this, so that was my main red line so and so having a look at bread assembly.

B

So where does it come from and what is it intended for so weapon sembly has came up in 2015 and it's not a replacement of javascript. So it's a complement of JavaScript. So every time the the problem with JavaScript is that it's always heap oriented and it's doing a lot of even mathematical operations requiring refrain and keep operations and so depend the ADEA. Is we introduced some kind of assembly code which will call by the JavaScript to do to to speed up perform, perform critical operations?

B

And so, if we've got a JavaScript code- and we have got the flow the flow here so there might be in operation which kicks into the web assembly code during the performance, critical computation and then getting back and continuing JavaScript and so web assemblies incorporates a portable stack machine.

B

The notation is it's a binary notation and but it also provides a text text-based or a representation, and you can you can use C++ or rust to create a binary code, webassembly binary code, which will be glued with JavaScript into an HTML page and then execute it in the browser and the idea or the question now boss would I be able to get away with JavaScript, and would it be a solution just to use what assembly just to use the the execution engine here and instead of JavaScript interacting with the execution engine directly interacting from grass to to webassembly engine and getting all the benefits.

B

So it's a it's widespread. So there is a community using that assembly. There are good compilers I can do my compilation on a high level language compiler too, but assembly and execute it in an in a small lightweight stack machine, so the built so for C++ and rust web assembly is a build target. So we've got the binary form a format webassembly the text representation as expressions are called ray st it.

B

What assembly supports only few scalar types, guy, scalar types key, so integer, 3264, floating 32 and floating 64, and there was no support. So there was no native support of unsigned integers, so everything is mapped onto these onto these types.

B

The webassembly stack machine supports less than or has fewer than 256 operations or instructions. So it's very compact and these operations can be met or could be mapped easily onto up-to-date architecture, GPU architecture. So it's sandboxed it's using linear memory, so it hasn't direct access to the global memory of your application. It's using tables and all their memory access. Every memory accesses a bond checked if we stay in the linear memory and so on.

B

So pointers are just indices into the linear memory, and so when we well, when we do it just in time compilation the there are it's possible to get a one point: five factor of native code speed.

B

So how would it look like so we saw we have. We have seen the picture before, with a JavaScript and and but assembly execution engine so with rust. Would look like like like this one. We've got the controller setting up or allocating some memory which will be assigned to the web assembly instance. We've got a web assembly file which will be loaded into into the context of the rust application executable and which represents the module.

B

We have got a table where we can register callbacks, which could be which these callbacks are there. The functions the module has got depends on has got to import. So anytime, we've got an external function here and loaded as a module. We have got provide a table which will provide the functionality of a function, to step, to call in indirect call to call this rust implementation.

B

We've got a stack machine, and so this is the Box web assemblies running in, and so it's quite secure execution environment.

B

So how does it look like so I tried with as a set we can use rust or C++ to create web assembly code binary code, and so, for example, if we use rust few thrusts, we declare a function and we declare some parameters in a return point. A return value, as you can see here.

B

This is an integer 32 and in fact this one would be a pointer into the into linear memory, and this would be mapped onto an index of type integer 32 as well, and I can show you later, and so we would do there.

B

I would iterate over over the memory over this buffer and check for all characters, e I and replace it by O so compiling this was cargo nightly I, get a file, transform, wasn't of size, 50, 50, 57k kilobytes, and which is quite big, so rust, compiled into a lot of additional meter information for the panic panicking.

B

So in case there is an Fault in the application that would there's also all the the integers and Aramis is inside that are written into the console and so far I didn't manage to get this to optimize it and get rid of these strings so name mangling, so the within the module the operation would be called transform.

B

That would be the method name, that's important, because if we look at the C++ implementation, similar return, type, integer, integer, first parameter, we've got a pointer to an array, the same logic, and so this will be compiled with with her what M script right and the result is just not is 296 bytes because it doesn't contain all the error. Strings rust is compelling. The the drawback is that the compiler is doing name mangling and the name within the module would be underlined, transform, but that's just so, and I didn't manage to get this way.

B

So I didn't manage to get the underline away. So if anybody has got a clue, the binary code looks like this. This jar, the 296 bytes and their textual form, would look like this, and here you can see their first parameters. This is the first parameter, and this is the second parameter pointing to the location of the array within the linear memory and calling it we would just we would just.

B

We would create a linear memory place our our message, byte message into the linear memory at index, 1, 0 and then calling this function with the length of the array and 0 index 0. And this way we would tell the web assembly where, in the linear memory, the error is located, so the credits to this they well based. No, this is the implementation prototype at.

B

It is based on rep assembly interpreter done by Nick Worf, formerly at parity wasn't github, and this one has been renamed now to bottom interpreter: Alex Christian for Tokyo IO and Karl Asia. For me, oh, these are the three important modules my prototypes, based on just as a proof of concept.

B

So I've got their the code here and, as you could see here, can you read it this one? Is the the method I declare? This is the file I'm loading, then I declare environment, very environment, object here and.

B

Then I create the memory linear memory. This one is the place where the buffer, the message buffer, is placed into the linear memory, and then this execution is performed.

B

These are the parameters.

B

And here is the invoke export that I invoke an exported operation of the module and I get a return value and then I've got to evaluate the return value.

B

So ten minutes left: okay I've got a small showcase, so I'm starting I'm starting the application.

B

So it's listening at the udp-based application. Listening for UDP packets.

B

So I've got this client and and if I send a hello e, it responds so it receives the message: it's processed by web assembly, replacing all the I's bios and sending back the message just to demonstrate. So so it works even with longer strings and that's it. So what I can do now is I can change.

B

So all I shall be replaced by wise.

B

B

Reckon filed the bottom.

B

And now, when I sent.

B

So I get a why Hey hooray it works so.

B

So I did not check for performance. I did not. So it's it's a proof of concept. So every time every message it loads, the the wasn't file: okay, creating the assembly instance the module instance executing it and then throwing it away. So don't ask me about performance so, and another aspect is it's it's interpreted, so the bottom eye is interpreting webassembly.

B

So there is the would be possible to create some kind of JIT compiler, compiling the module directly to native code, and probably there would be much faster now, but that would be the next step and maybe I could make use of existing modules for for Mozilla, Firefox or other web browsers and just reusing those webassembly execution engines.

B

Okay, are there any questions.

B

This interpreter well, the question is the question: is: how much code does it require to create a web assembly interpreter.

B

I've know I can't tell you number of lines, but it's a small project at github and so yeah I can't I can't answer the question. So sorry it's more it's a small, it's a small module and when I strip this application, which is not ideal.

B

B

B

So this is so it's 800 K, so the application containing all the Tokyo logic as inference IO and the and the interpreter book wasn't interpreter is 800 K statically linked against the local libraries. So that's not much and probably could be even reduced.

B

B

Yes, you can. The question is: is it possible to declare to access functions in the rust environment from the web assembly? Yes, there is a good. There is.

B

This one, this is the callback table for or the table this table here this contains. So you you define in your C++ code or US code, you define a signature of a method and which is external. You use it in your code, you call it in your code. The compiler will realize how it's an external external function and it will be declared as external function here, which needs to be imported and these functions are accessed using the table.

B

So when you, when you load there wasn't filed module and resolve all the dependencies, it will check for the existence of specific functions, and so it will call the function which is registered here and call a call in here. Yes,.

B

I don't know: well it's a question: if, if we can stack up so I would I.

B

I didn't I, didn't try this, but I don't see I, don't see why this shouldn't be possible, so it might be well I, don't know how well I've got to check the the web assembly interpreter how this the the calling the exported function. If there is an a restriction that there should be only one call to an exported function at any time, I don't know so good question. I didn't I didn't check this.

B

That's so right now strings are problem because strings. So if we talk about web assembly, webassembly is something which has been established in a JavaScript world and so javascript strings look different to rust strings. But there is some. There is a there are.

B

What is it there are? Some people are working right to get some kind of together, it's kind of standard to access strings. So the idea is to use to register strings in here. So if, let's assume, we've got a function, we want to call the function and hand over a string, so the string could be would be registered here and the state machine or the module could verify what kind of string is it and then performing the correct interpretation of the string?

B

So it could be 0, terminated, string, it could be a length and coded string. It could be whatever string so so that that's the problem. Some people are using the linear memory more effect, more or less. They serialize the data into the zero into linear memory and then call the sec machine. So right now a stack machine should be so. If you define a boundary between rust and represent me, you should use the the boundary where you deal with binary data.

B

For example, if you receive a message from the network and you want to process it, that would be easy to to put into the linear memory and then process it using web assembly if you've got a complex, Dom tree internally. That's not a good idea so, but people are working on this and JavaScript.

B

The JavaScript community is working on this too. So getting Dom trees handed over to the web assembly engine and we've got to check how we would be able from rust to do the same, but we always have got to first of all, it's a it's a JavaScript standard or web standard, it's not defined by the rest community. So we are. We are always second and we have got to check out how we could incorporate our using the technology.

B

A

B