Rust – Reducing support for 32-bit Apple targets(blog.rust-lang.org)
blog.rust-lang.org
Rust – Reducing support for 32-bit Apple targets
https://blog.rust-lang.org/2020/01/03/reducing-support-for-32-bit-apple-targets.html
2 comments
People have complained a lot about this decision by Apple, but I respect it. They're using their leverage over the ecosystem to cut off a huge piece of cruft from not only their own codebase, but many other codebases like Rust's that can now point to their decision in the face of criticism. And the only cost is that old binaries (not old code, old binaries) will no longer run without being rebuilt. Plus, this will probably never need to happen again because 64 bits can address 18 million terabytes of main memory.
> And the only cost is that old binaries (not old code, old binaries) will no longer run without being rebuilt.
Not always true. There's plenty of 32bit code that can't trivially be ported over to 64bit, even if you have the full source for all the dependencies. e.g. maybe the file formats have padding or pointer width assumptions - now you need a compatibility shim whenever loading/saving the files. Maybe the code has implicit padding assumptions, which are not documented anywhere - now you'll get occasional random functionality bugs that are a huge effort to catch and debug.
It is not always a matter of flipping a compile-time switch for the developers, even if they are still around, and have all the source, and all the libraries and whatnot.
Not always true. There's plenty of 32bit code that can't trivially be ported over to 64bit, even if you have the full source for all the dependencies. e.g. maybe the file formats have padding or pointer width assumptions - now you need a compatibility shim whenever loading/saving the files. Maybe the code has implicit padding assumptions, which are not documented anywhere - now you'll get occasional random functionality bugs that are a huge effort to catch and debug.
It is not always a matter of flipping a compile-time switch for the developers, even if they are still around, and have all the source, and all the libraries and whatnot.
I feel embarrassed asking this, but better that than not to. Is memory address space literally the _only_ difference between 32 and 64 bit architectures?
Not neccessarily. While the size of pointers and the size of general purpose registers does change between 32 and 64-bit architectures, there can be other significant differences.
For example, on X86_64, the ISA provides additional GPRs that aren't just 64-bit versions of the 32-bit register. For example, rax is the 64-bit version of eax, but IIRC you also have r0-7, for which there is no 32-bit equivalent. Furthermore, the ABI (at least on Windows) specifies a different function calling convention versus X86.
Additionally, I imagine there additional instructions/ops on x86_64 over x86. I dont know how else the cpu would distinguish add %eax, 1 from add %rax, 1, both of which are legal when the cpu is running in 64-bit mode.
I recall seeing a cool talk on how to confuse or crash many debuggers by doing something clever in assembly. The idea is you would write a block of polyglot 32 and 64-bit x86/64 assembly (i.e. binary that is both a valid x86 and x86_64 instruction sequence), switch the cpu from 32 to 64 bit mode at the end of the sequence, then branch back to the start of the block and reinterpret the same instructions as 64-bit rather than 32. You could use this technique to frustrate reverse engineering.
For example, on X86_64, the ISA provides additional GPRs that aren't just 64-bit versions of the 32-bit register. For example, rax is the 64-bit version of eax, but IIRC you also have r0-7, for which there is no 32-bit equivalent. Furthermore, the ABI (at least on Windows) specifies a different function calling convention versus X86.
Additionally, I imagine there additional instructions/ops on x86_64 over x86. I dont know how else the cpu would distinguish add %eax, 1 from add %rax, 1, both of which are legal when the cpu is running in 64-bit mode.
I recall seeing a cool talk on how to confuse or crash many debuggers by doing something clever in assembly. The idea is you would write a block of polyglot 32 and 64-bit x86/64 assembly (i.e. binary that is both a valid x86 and x86_64 instruction sequence), switch the cpu from 32 to 64 bit mode at the end of the sequence, then branch back to the start of the block and reinterpret the same instructions as 64-bit rather than 32. You could use this technique to frustrate reverse engineering.
But pretty much all of those differences get handled at the compiler level, right?
If you thought about portability when you were writing the code, then yes, you just flip a compiler switch and you're good to go.
A lot of the older 32bit software didn't ever consider the need to switch to 64bit in the future, so there is plenty of implicit assumptions made. These are sometimes very obvious (e.g. hand coded 32bit assembly), but sometimes very hard to detect, and can have a lot of logic built on top of it. For example, consider these two structures:
There's plenty of ways the code could assume that sizeof(A) == sizeof(B). For example, it could use memcmp() to verify the structures are equal, or it could be allocating them from the same slab allocator to minimize fragmentation. These kinds of bugs will not be caught by the compiler, and they might not be easily noticeable at runtime. It takes significant effort to port a, say, 20 year old codebase written for 32bit over to 64bit.
Yes, it's easy to say "that's bad code, you shouldn't have written it like that in the first place", but programming 20 years ago was drastically different than today, and those "hacks" could make or break a product back then. And, at the time some 32bit software was written, it was not even obvious what 64bit would look like, so it was not obvious how you'd prepare for it even if you wanted to.
A lot of the older 32bit software didn't ever consider the need to switch to 64bit in the future, so there is plenty of implicit assumptions made. These are sometimes very obvious (e.g. hand coded 32bit assembly), but sometimes very hard to detect, and can have a lot of logic built on top of it. For example, consider these two structures:
struct A { int n; void *p; float f; };
struct B { int a, b, c; };
There's plenty of ways the code could assume that sizeof(A) == sizeof(B). For example, it could use memcmp() to verify the structures are equal, or it could be allocating them from the same slab allocator to minimize fragmentation. These kinds of bugs will not be caught by the compiler, and they might not be easily noticeable at runtime. It takes significant effort to port a, say, 20 year old codebase written for 32bit over to 64bit.
Yes, it's easy to say "that's bad code, you shouldn't have written it like that in the first place", but programming 20 years ago was drastically different than today, and those "hacks" could make or break a product back then. And, at the time some 32bit software was written, it was not even obvious what 64bit would look like, so it was not obvious how you'd prepare for it even if you wanted to.
I'm not an expert, but technically what it defines is the size of a "word", which is a basic unit of memory handled by the processor, used for - among other things - addresses. I think this might affect number precision too but I might be wrong. But regardless, most higher-level languages don't care about word size. Even C/++ code can be written such that it doesn't care about word size, though it can also be written such that it does.
Seems fair. Dropping support doesn’t mean it won’t work anymore. Just that it’s not as tested (reading the RFC).
32bit MacOS was never used on x86 as far as I know. And even PPC 32bit was a very small number of machines a very long time ago. Not sure about 32bit iPads and iPhones.
But, I’ve never seen a project using rust on iOS. Does anyone have an example of this?
32bit MacOS was never used on x86 as far as I know. And even PPC 32bit was a very small number of machines a very long time ago. Not sure about 32bit iPads and iPhones.
But, I’ve never seen a project using rust on iOS. Does anyone have an example of this?
There’s a non trivial amount of 32 bit audio processing code for MacOS in the wild due to the longevity of snow leopard among professionals.
OS X Lion was the first version to use an x64 kernel, and there were definitely x86 (32-bit) versions of the OS and apps.
In particular, the quicktime runtime / codecs (until quicktime x) was only x86, and various VFX apps on OS X had to write RPC code to communicate with it as a 32-bit process from 64-bit land.
In particular, the quicktime runtime / codecs (until quicktime x) was only x86, and various VFX apps on OS X had to write RPC code to communicate with it as a 32-bit process from 64-bit land.
32-bit Apple mobile devices stopped with the launch of iPhone 5s (~5 years ago). Apple dropped support for 32-bit only desktop CPUs ~7 years ago (Lion) and 32-bit only mobile CPUs ~2 years ago (iOS 11). Apple dropped support for 32-bit apps completely in iOS 11 and in macOS 10.15 (this latest release).
TLDR: No recent Apple OSes support running 32-bit at all.
TLDR: No recent Apple OSes support running 32-bit at all.
FWIW the iPhone 5s was September 2013, i.e. 6.25 years ago.
I only mention it because the timing was a surprise to everyone in the industry.
I only mention it because the timing was a surprise to everyone in the industry.
> TLDR: No recent Apple OSes support running 32-bit at all.
macOS Mojave 10.14 (released 2018) supports running 32-bit apps.
iOS 10 (released 2016) supports running 32 bit apps.
I would agree that iOS 10 is not recent, in the context of the iOS timeline. I would disagree that macOS Mojave is not recent, given that Apple still has numerous show-stopper bugs in the upgrade process to Catalina.
macOS Mojave 10.14 (released 2018) supports running 32-bit apps.
iOS 10 (released 2016) supports running 32 bit apps.
I would agree that iOS 10 is not recent, in the context of the iOS timeline. I would disagree that macOS Mojave is not recent, given that Apple still has numerous show-stopper bugs in the upgrade process to Catalina.
Cloudflare’s 1.1.1.1 app uses Rust, on both iOS and Android. (I work there but not on it.)
Oh interesting. I wonder what the workflow actually is for that then.
I guess it’s not open source. :(
I guess it’s not open source. :(
The app isn’t, but the rust bit is: https://github.com/cloudflare/boringtun
The workflow shouldn’t be anything too intense, you build a library, compile to the right arch, and treat it like any C library.
The workflow shouldn’t be anything too intense, you build a library, compile to the right arch, and treat it like any C library.
Yeah there was one generation of Mac's that used the Intel Core Duo which 32bit only. They are long not supported. https://everymac.com/mac-answers/snow-leopard-mac-os-x-faq/m...
PPC 32 bit is still used extensively in the embedded space. It's understandable that it's not the main target for newer languages as LLVM support seems to be sparse (and practically non-existent for some dialects such as Freescale PPC-VLE) but it would be great to enable safer embedded development.