Comment by asveikau
3 days ago
The explanation they give is they need to put their trademark, Windows, before Linux. Sometimes they say this is advice from the legal department.
I still think they could fulfill that requirement and call it the "Windows Linux subsystem" or something, but what do I know?
Unrelated, but I think the WSL2 design is kind of stupid. It's just a VM. I think the WSL1 design, where it was a syscall layer, is a better call. But that was slower, IIRC chiefly because the NT filesystem syscalls are slower than Linux's VFS. Rather than improve that problem, they side-step it by running Linux in a VM.
The design of WSL(1) makes more sense when you think of its original design goal of being a compatibility layer for Android apps. Android is "a Linux", but it is (1) a relatively unique one, and (2) everything between the Android kernel and Android apps really isolates the application layer from kernel-level details. Given this separation, it makes a lot of sense to leverage the existing NT flexibility and emulate a Linux kernel at the syscall layer. Sure, you'll have to patch some parts of the WSL(1) Android system components, but MS was already going to have to do that to work around Google-only components. In many ways, this route is no more complex than what Blackberry did to run Android apps atop their QNX-based OS.
But once you give up the specialization for Android and want WSL to be a "real Linux" (i.e. behave like a specific Ubuntu/Fedora/etc distribution) now you no longer can get away with being Linux-like, you have to be Linux (i.e. you need the syscall layer to directly mirror all kernel development and features). It's actually fairly impressive how much worked with WSL(1) given how different the internals were, but you didn't have to go that far to find tools/services/etc that just wouldn't work.
Instead, once you consider how long MS had been working on Hyper-V, and how interested they are in using it to apply additional security boundaries/isolation (e.g. VBS) within what outwardly appears to be a single Windows OS instance to the user, it makes a lot of sense to leverage that same approach to just run a real Linux kernel atop Hyper-V. In that world, you no longer have to match Linux kernel development, you just need to develop/maintain the kernel drivers used to interact with Hyper-V - and MS already had a lot of experience and need to do that given how much of Azure is running Linux VMs.
Full VMs make much more sense indeed, because they allow you to run fundamentally different OS while still keeping the host mostly the same. And they basically show that with the FreeBSD addition.
The syscall way is just a form of emulation that you have to contain and it becomes a pain to keep up to date. VMs will use more ressource but at least they are disposable and only require a good virtualization layer on the host.
Funnily enough, with time, Microsoft might be able to run all the OSs inside their own OS. Of course, that won't happen for something like macOS but that would be hilarious.
IO on many little files is dramatically faster in Linux on ext4 vs Windows thanks to NTFS' journaling overhead. So if you're doing development, you really want to do it inside wsl2.
I think it's more complicated just than NTFS's design.
In my original comment I said that the difference is the Linux VFS for a reason. The slow part in NT is when you go from a filename to a handle. Doing things like caching lookups by name is, IIRC, the responsibility of the individual drivers. Linux does better at this by having a heavily optimized layer sitting between the filesystem driver and the caller. Doing tons of open(2)s is faster on Linux because of the overall kernel design.
Use a ReFS drive,
https://learn.microsoft.com/en-us/windows/dev-drive/
Journaling isn't the issue, small files go into the $MFT which is the fast-path. The issue is the file system filter overhead.
The bigger problem was how fast Linux evolves. Windows kernel development is glacial by comparison. Keeping up with every new thing in Linux was tantamount to maintaining a second operating system.
IIRC, there was an article, whose author said that improving NT kernel without blessing from higher ups is even frowned upon.
So making a better WSL on syscall layer, which NT kernel is designed for, is not only behind a technical effort wall, but also behind a big red tape.
That's a good thing. The kernel is supposed to be the most stable thing.
I'd argue that the breaking point was Docker. Running typical Linux userland applications was doable - and they did it - but making Docker work would have been effectively impossible, as it depends on a ton of highly Linux-specific systems (namespaces, cgroups, bind mounts, etc) with no direct Windows equivalents.
Windows has native support for containers.
3 replies →
Improving that problem probably would've been a massive undertaking. That aside, there's the problem that implementing kernel mechanics is a lot more than faking syscalls: the various types of namespaces, FUSE, random edge cases that applications do expect, kernel modules, etc. At the end of the day, users don't want to stumble into some weird compatibility issue because they're running not-quite-Linux; it's a better UX to just offer normal Linux with better integration.
The WSL2 design isn't stupid, it's practical. What I will give you is that it's not elegant in an "ivory tower of ideal computing" sense.
When people talk about improved compatibility or higher practicality I wonder why they don’t just run Linux on metal at that point. You can either run it on your laptop, or connect to a networked computer.
Can I even use a usb serial port yet after how many years? (Possibly by now but how long did it take, and does it actually work well?)
It is stupid in that it's not really any kind of subsystem, it's just a vm. VMs have their uses, but it's basically just an app.
The reason hardware such as my usb serial example (or any serial) worked on wsl1 was because it actually was a subsystem.
Your serial might have worked, but your docker didn’t. (And someone else’s other drivers didn’t, and mmapping had ever-so-slightly different semantics causing rare and hard to reproduce issues).
WSL2, on the whole, is much more compatible. If you want 100% Linux compatibility, just run Linux.
2 replies →
There is no native USB passthrough support, but you can use USB/IP to access them via network.
https://github.com/dorssel/usbipd-win
>Unrelated, but I think the WSL2 design is kind of stupid
Yet in practice works very well
There's saying if something is stupid, but works, then it aint stupid
True, but I love the instant boot of WSL1 when I need to do something quickly. And, most of the time, if you don't do IO-heavy stuff, it works quite well.
You can still use WSL1 if you want. (maybe you know this).
> I still think they could fulfill that requirement and call it the "Windows Linux subsystem" or something, but what do I know?
Your name is bizarrely better considering how small the difference is.
> The explanation they give is they need to put their trademark, Windows, before Linux. Sometimes they say this is advice from the legal department.
It feels like they have some strange internal naming policy. Maybe it is called the “Policy Product for Naming.”
Their naming is a serious problem that spreads confusion instead of clarity but what do they know, they’re like a headless chicken that has enough power to stay alive no matter what..
Windows Subsystem for Linux Usage or something would have been clearer.
"Windows Subsystem for Linux applications"