Live data from Hacker News

Microsoft's Linux Kernel

github.com

531–532 of 532 posts

Re: Microsoft's Linux Kernel

#531

Earlier quoted context omitted.

So, you buy Windows to get better hardware compatibility for your Linux system on top of whatever Windows ecosystem brings you. I did something like that with non-Windows VM's. Although I couldn't find the link, there was an academic prototype you might have liked that used virtualization to reuse Windows drivers in other OS's or VM's. Seems like there could be a market for that where it's combined with recent enhanc…

In general I think it is an interesting idea. There has always been an economic issue with hardware manufacturers who were unwilling to expose too much information in an open source driver, and security issues of running a binary blob at ring level 0 in the kernel. Microsoft puts itself in the middle of that by certifying a driver through a vendor sign up / certification process, and since it doesn't require the vend…

There's another potential advantage. Thanks to SLAM, the Windows drivers have some strong verification against that interface. If one implements it well, then they get both benefits of Windows hardware in general and extra robustness of their driver verification. As in, drivers might be better than Linux drivers without such interface verification.

The drawback being whatever in middle interfaces the Windows interface to the non-Windows VM's might introduce bugs. It must be correct and compatible.

Post reply on HN