asmjit VS durin

Some other instruction sets in some JSON: https://github.com/asmjit/asmjit/tree/master/db

The attentive reader may notice that this code is from a third-party library. So, we didn't want to include it in the article at first. However, we found something interesting. In 2017, somebody opened an issue in the asmjit project: the GCC 7.2 compiler issued a warning to the code above. The project authors fixed it:

You can use something like https://asmjit.com/ to generate and call x64 code at runtime.

But an easy to create a JIT would be to use https://github.com/asmjit/asmjit, which is used in RPCS3.

You could go the way of https://asmjit.com (or forth) and make it your assembler DSL on top of the low-level call.

But if you're willing to get closer to the hardware is https://github.com/bitfunnel/nativejit/ and https://asmjit.com/

https://github.com/asmjit/asmjit is nice. But using a JIT seems like a sledgehammer working around a lacking design.

So I think I may be one of the few people in the world who has actually implemented a GC using LLVM's statepoint infrastructure. It's poorly documented and there are some gotchas, but I'd say it's definitely usable, and it works with basically any collector design, including moving collectors (I'm using Immix) and has no runtime bookkeeping overhead and allows LLVM to optimize the code without worrying about GC, which is nice. It's actually gotten a bit better with LLVM 13, too. If you're curious what a LLVM-based GC looks like, mine is in this folder. Of course, if you just want some sort of GC, you can also just link it with Boehm which is quite easy and has pretty good performance - this is what e.g. Crystal does, although they're talking about switching.

I also fixed lots of bugs in the GC and backend, so it should be a lot more stable now.

Recently, I've been working on adding garbage collection to Pika. I've successfully written an Immix-based garbage collector that works with the LLVM GC support infrastructure, and I'm currently working on integrating the GC with Pika, or really Durin, the dependently-typed intermediate representation that Pika compiles to. Because types are passed around at runtime, objects of unknown type and size can be stored unboxed in polymorphic data structures; but that makes keeping track of type information for heap allocations somewhat harder, because type information needs to be allocated and constructed at runtime in some cases. It's an interesting design problem, because you want constructing type information to be fast; but the GC will run much more often, so maximizing tracing speed by avoiding e.g. indirection in type information is important; and you also want to construct as much type information as possible at compile time and embed it as constants.