you are viewing a single comment's thread
view the rest of the comments
[–] 111 points 1 year ago (23 children)

Window's kernel has been a monstrosity for decades now, but they're so invested in it they can't change. I've worked directly with guys who worked on the NT kernel and they all agree it's absolute shit, but they can't break away from it either. A handheld has to be one of the worst places for it to run

So for us this is hilarious

  • source
  • hideshow 23 child comments
  • [–] 73 points 1 year ago (20 children)

    I've also worked with people that worked on the NT kernel. I couldn't agree more. Im just waiting for the day someone exploits these anticheat kernal hooks to create the ultimate rootkit. It'll make crowdstrike look like nothing.

  • source
  • parent
  • hideshow 20 child comments
  • [–] 17 points 1 year ago (2 children)

    The funny thing is even though it has been done, there's not even that much of an incentive to do it because Windows on consumer side has so little defense that most attackers opt for lazy premade viruses sold on the darkweb, and Windows on enterprise side is so insanely insecure that the only groups that make high end rootkit level software are usually government backed APTs.

    Microsoft also very conveniently avoided making a new filesystem from old ass NTFS because SSDs started popping up around the time Window's IO operations were clogging every old machine with HDDs.

    I remember upgrading from 7 to 8 and the disk IO just sat at a solid 100% at idle lol.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 3 points 1 year ago (1 child)

    I'm curious, is this due to many unnecessary files operations (due to Windows 8 bloat), or because the file system sucks at scaling up file operations?

  • source
  • parent
  • hideshow 1 child comment
  • [–] 6 points 1 year ago

    Both. WIndows 8 added a ton of unnecessary operations, part in due to the horrendous new PWA system they made to replace all the proven software.

    NTFS meanwhile functionally reflects FAT32. It has no proper block allocation algorithm, so files get fragmented and placed in poor locations all over the physical disk. Tools like defraggler became super popular because they provided serious and visible IO gains from defragging your drives.

    Compare that to ext4 which only begins to fragment once you hit something like 95%+ capacity.

  • source
  • parent
  • [–] 9 points 1 year ago (15 children)

    Every piece of software is vulnerable (or likely vulnerable I guess), but kernel level anti cheat has been around for a while right? Why hasn't it been exploited yet?

  • source
  • parent
  • hideshow 15 child comments
  • [–] 35 points 1 year ago (1 child)
  • [–] 22 points 1 year ago

    The driver/module, "mhypro2.sys," doesn't need the target system to have the game installed, and it can operate independently or even embedded in malware, offering the threat actors a powerful vulnerability that can disable security software.

    I will never stop highlighting this because it's just too funny

  • source
  • parent
  • [–] 6 points 1 year ago (12 children)

    Yeah i mean on a level where breaks the kernel and MS is forced to stop allowing these kinds of kernel modifications like they were talking about doing after the crowdstrike incident.

  • source
  • parent
  • hideshow 12 child comments
  • [–] 11 points 1 year ago (10 children)

    Too much money. I worked on the Windows kernel from minkernel to onekernel. There were massive rewrites with the switch of the CE kernel out for minkernel when Windows Phone was in development. minkernel used to chew through eMMC memory in a few weeks on the first Windows Phone internal dev devices. Microsoft could, rewrite onekernel (I'm assuming they are still on onekernel), if they wanted. I think Windows is a dead man walking.

    Microsoft keeps building up Azure Linux. Also they push Windows 365, the cloud based Windows OS for businesses (if I understand correctly). If I'm reading the tea leaves, Windows runs like shit in the cloud and is very expensive. Because of this, companies are switching to Linux containerization for their servers. Even on Azure, Linux is on 60% of the servers. Even I work exclusively on services containerized with Linux, never Windows. If Windows was so good, you'd think it would be the opposite.

    Also, Microsoft makes all their money from Cloud, i.e. Linux. Which again is why Azure Linux is getting more and more development. So, imagine if you will, Windows 365 instances suddenly become Azure with a Windows userland ( Windows/Linux, not GNU/Linux). Most users wouldn't even know. If you had problems, running your software, Microsoft could allow you to drop back to Full Windows. For every Azure Linux instance running as Windows 365, that would be a significant cost savings to Microsoft, especially when everybody does everything in Chrome. If that's how it all unfolds, why would Microsoft want to put any major engineering dollars towards a kernel rewrite? They do have the money. I just don't see Microsoft every fixing the kernel root kit situation. It's 100% in their wheel house though.

  • source
  • parent
  • hideshow 10 child comments
  • [–] 9 points 1 year ago (9 children)

    There were rumours that windows would become a Linux desktop environment for a while, I can see the business case for it but the migration seems impossible

  • source
  • parent
  • hideshow 9 child comments
  • [–] 5 points 1 year ago (6 children)

    Maybe it'd be a new "Windows S Mode" situation.
    Got a new cheapo laptop? Enjoy our Secure Windows Home Basic (Linux + Windows DE) and install your apps ONLY from the Windows Store (that we made sure run in the new environment)
    Need full Windows? Upgrade to Pro.

  • source
  • parent
  • hideshow 6 child comments
  • [–] 3 points 1 year ago (5 children)

    I wouldn't hate a closed windows ecosystem on Linux. We would get the kernel patches and more software would work. Even if we didn't get kernel patches because windows is scummy and ignores the gpl, a common abi would still be amazing

  • source
  • parent
  • hideshow 5 child comments
  • [–] 4 points 1 year ago (4 children)

    "common abi"? Like embracing?

  • source
  • parent
  • hideshow 4 child comments
  • [–] 1 point 1 year ago (3 children)

    Abi means application binary interface. Iirc it is a convention basically on how the operating system expects system calls to be made.

    That is a fancy way of saying it's how programs can use the operating system.

    Linux does it one way and windows does it another. It is apart of why we can't run windows programs on Linux.

    Executable format is another big reason, I'm sure there are many.

  • source
  • parent
  • hideshow 3 child comments
  • [–] 2 points 1 year ago (2 children)

    Oh, thanks... I meant to say that a common ABI sounds like the first stage of embrace, extend and extinguish

    I, actually, teach OSs at a university 😁

    I truly appreciate, tough, your kindness on teaching to a lemmy fellow 🙏🏼

  • source
  • parent
  • hideshow 2 child comments
  • [–] 1 point 1 year ago* (1 child)

    Oh God I didn't think of that. I feel like Linus is ornery enough to stop that, but they won't live forever.

    Edit: also what did I get wrong!? I haven't taken an OS class in years and my only exposure to binary formats has been reverse engineering toy files

  • source
  • parent
  • hideshow 1 child comment
  • [–] 2 points 1 year ago

    Yeah the big question is what's gonna happen to the project after Linus...

    you got it pretty much right. An ABI, depending on the context, could include just the app/OS interface or also the across-apps, across-apps-modules interface too. Things like calling convention, register usage, stack usage, etc.

    when you distribute compiled libraries, you want clients to know how to invoke your functions and know how to retrieve your returned values. That's part of the ABI too.

    The ABI also defines the type translation from the language (say, C) to asm undertood by a processor (say riscv64g) so, you map types. Following that example, you may instruct the assembler and linker to use abi "lp64" that maps longs and pointers to 64 bits, and integers (int) to 32 bits. This abi also emulates floating point operations (n the other hand, lp64d would make use of dedicated hardware for "double precision" floats)

  • source
  • parent
  • [–] 3 points 1 year ago (1 child)

    I wonder about that. I'm probably not thinking of some very important things. Edge, Office, Active Directory, Co-Pilot, a Windows DE, userland programs(could even be GNU+Windows, don't want to forget notepad and minesweeper), Powershell, DirectX and SDKs. I think they could do it in a year or two. I just figure, if they could improve Windows in the cloud, they would have done it. And they've already got a massive head start with Azure Linux.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 2 points 1 year ago

    Right? They can still have their closed ecosystem and get their monthly check from your bank, but they could get out of the operating system business. I bet that is an expensive one with low revenue

  • source
  • parent
  • [–] 17 points 1 year ago (1 child)

    I was a mobile developer and worked on Windows Mobile and Windows CE (un-ironically called "WinCE" by Microsoft themselves) applications some twenty years ago. It was basically just Windows with a lot of unnecessary cruft stripped out. The basic UI was indeed absurd, with the standard Start menu and utterly dependent on the fucking stylus to work. But for applications it wasn't actually necessary to even use that shit. You could actually write applications that ran in kiosk mode and had nice big buttons so that users never had to deal with the Start menu or use the stylus at all. And in that mode it was actually extremely powerful -- you could do anything that you needed to do programmatically. I never once encountered a situation where something that I needed to do programmatically wasn't still available in the stripped-down WinCE API.

  • source
  • parent
  • hideshow 1 child comment