3 ms·
https://news.ycombinator.com/item?id=16068363 https://news.ycombinator.com/item?id=16068363 > This could be an extinction level event for less popular OSes. I
by cryptonector 9y ago
https://news.ycombinator.com/item?id=16068363 https://news.ycombinator.com/item?id=16068363
> This could be an extinction level event for less popular OSes. Intel, and to a lesser extent AMD, may end up causing enormous damage for which they'll pay nothing much.
- anonacct37 9y agoThis is true, but like other extinction level events it might actually open up the space for some new ideas. Someone on my twitter put it this way: microkernels could see a resurgence if the cost of a context switch into kernel is just as high as the cost of context switching between 2 processes.
- nly 9y agoMicrokernels you say? Yay, more complexity! In all seriousness, the Nintendo Switch users a microkernel and just got completely p3wned. It's extremely fanciful that we know how to build robust modular microkernels when we can't even get the simple stuff right.
- willtim 9y agoI think you have it backwards. Monolithic kernels are by their very nature more complex to reason about due to all the communication backchannels made possible when everything sits in one process.
- SOLAR_FIELDS 9y agoThis discussion seems vaguely familiar for some reason...[1] https://en.wikipedia.org/wiki/Tanenbaum%E2%80%93Torvalds_debate https://en.wikipedia.org/wiki/Tanenbaum%E2%80%93Torvalds_deb...
- marcosdumay 9y agoMicrokernels mean more insulation and standardization. Yes, it's a form of complexity; just not what tend to relate to problems.
- derekp7 9y agoThe complexity comes from when those standards don't let you do something, and you have to do weird things to break through them. Such as leaky abstractions.
- marcosdumay 9y agoMicrokernels IPC ought to be enough for anybody. Seriously, completely generic IPC is not something impossible. You may have some to create some large drivers, but the insulation is still there, even for those.
- anonacct37 9y agoIt's definitely a tradeoff. Distributed systems are hard. Isolation is good. One of the best tools we have for isolation is processes and MMUs. I'm not saying we should all switch, it's just that it's possible, given my understanding of how we're dealing with the intel bug, that one of the arguments against microkernels just went way. So they might make sense in places they previously didn't.
- lloeki 9y agoLISP machines!
- starsinspace 9y agoDepends on the target audience. For example, Haiku most probably won't get around to implement KPTI any time soon. But then again, most people consider it more of a fun/hobby OS[0] anyway, so it wouldn't matter that much. [0] And that isn't meant negatively at all, I actually think the computing landscape would be way more fun again if there were many more projects like it again. Not every OS needs to try and be an everything-OS, and not every OS needs to follow the "always online, everything through the internet" paradigm. If the most interesting use-cases are entirely offline, most of today's popular attack vectors cease to exist.
- userbinator 9y ago...and don't forget TempleOS, which follows the philosophy/religion of deliberately running everything in ring0 and one address space, because it believes the user should have full control over everything in the system.
- waddlesplash 9y agoHaiku has run all GUI applications as root since forever, so as soon as you got RCE you could install a rogue keylogging driver silently. Obviously fixing that is higher up on the priority list than KPTI... and it's unclear when we'll get around to that, even. :(