Reply to thread

Message

<blockquote data-quote="Deleted Member 308817310" data-source="post: 822043" data-attributes="member: 80647">I&#039;m right about all of it.The answer is as simple as I&#039;ve been making it out. There&#039;s no need to over-complicate this.Microsoft&#039;s sandbox technology is using a privilege level that&#039;s reserved by the firmware for Type 2 hyper-visors (ring -1) which behaves identically to ring 0 in the eyes of the code running under it, but is controlled by the host environment. The host environment&#039;s kernel is running with a Current Privilege Level (CPL) 0 and the guest environment&#039;s kernel is running with a CPL of -1. Whenever the guest environment&#039;s kernel executes a privileged instruction, the host environment is allowed to control what happens, and many instructions from the x86/AMD64 instruction set which are not privileged instructions can also be controlled.Sandboxie is entirely dependent on the host environment. There is no real isolation between the program being put in the sandbox and the rest of the host environment, it&#039;s merely an illusion.This isn&#039;t an issue and it is no different to if you had done this on a VM using VMware.Sandboxie might deny the installation of drivers, but what about vulnerable drivers that might already be on the machine?Windows Sandbox is literally the same as a VM except it&#039;s been designed to be more convenient; there&#039;s no need to install OS media because it takes the system files from your host environment.This conversation can go on for decades but the obsession of Sandboxie being more powerful than Microsoft&#039;s sandbox technology when Microsoft use dedicated CPU features designed for isolation is shocking.</blockquote>

[QUOTE="Deleted Member 308817310, post: 822043, member: 80647"] I'm right about all of it. The answer is as simple as I've been making it out. There's no need to over-complicate this. Microsoft's sandbox technology is using a privilege level that's reserved by the firmware for Type 2 hyper-visors (ring -1) which behaves identically to ring 0 in the eyes of the code running under it, but is controlled by the host environment. The host environment's kernel is running with a Current Privilege Level (CPL) 0 and the guest environment's kernel is running with a CPL of -1. Whenever the guest environment's kernel executes a privileged instruction, the host environment is allowed to control what happens, and many instructions from the x86/AMD64 instruction set which are not privileged instructions can also be controlled. Sandboxie is entirely dependent on the host environment. There is no real isolation between the program being put in the sandbox and the rest of the host environment, it's merely an illusion. This isn't an issue and it is no different to if you had done this on a VM using VMware. Sandboxie might deny the installation of drivers, but what about vulnerable drivers that might already be on the machine? Windows Sandbox is literally the same as a VM except it's been designed to be more convenient; there's no need to install OS media because it takes the system files from your host environment. This conversation can go on for decades but the obsession of Sandboxie being more powerful than Microsoft's sandbox technology when Microsoft use dedicated CPU features designed for isolation is shocking. [/QUOTE]

Verification