Post AboPPJe8mLDszbc41o by jjdavis@infosec.exchange
 (DIR) More posts by jjdavis@infosec.exchange
 (DIR) Post #AboO1uHio45rijOtSC by mttaggart@infosec.town
       2023-11-14T22:55:18.938Z
       
       0 likes, 0 repeats
       
       Is this legit? I certainly couldn't say for myself. https://thehftguy.com/2023/11/14/the-linux-kernel-has-been-accidentally-hardcoded-to-a-maximum-of-8-cores-for-nearly-20-years/
       
 (DIR) Post #AboPPJe8mLDszbc41o by jjdavis@infosec.exchange
       2023-11-14T23:04:33Z
       
       1 likes, 0 repeats
       
       @mttaggart The code snippet cited does seem to lock the number of CPUs to no more than 8.  But... the linux kernel is big.  You won't believe how vastly, hugely, mind-bogglingly big it is.  You may think it's a long way down the road to the chemists, err, sorry, got distracted, but in any event the linux kernel is really big and looking at any one snippet may not tell the whole story.
       
 (DIR) Post #AboRPp28HZuQFJCiae by indrora@furry.engineer
       2023-11-14T23:32:24Z
       
       1 likes, 0 repeats
       
       @mttaggart It passes the first sniff test, but I'd need someone who was Very aware of the state of the linux scheduler to really dig under the hood. At first blush, it would appear that only up to 8 cpus are taken into consideration.
       
 (DIR) Post #Abom6lZzXXgsVCOdBQ by bonealisa@infosec.town
       2023-11-15T02:45:06.884Z
       
       1 likes, 1 repeats
       
       @mttaggart So from what I understand this title is pretty clickbait. I'm no kernel expert, jfyi.I believe this piece of code is essentially the minimum amount of time a process should live, on low core systems you want to be killing processes as quickly as possible for performance. The kernel will scale up the minimum time a process will live as core counts increase, but only to a maximum of 8 cores. This is good because you don't want a processes minimum lifetime being in the milliseconds on a 64 core system.
       
 (DIR) Post #Abom7znQAVagsrGQfQ by mttaggart@infosec.town
       2023-11-15T03:25:21.431Z
       
       0 likes, 0 repeats
       
       @bonealisa This makes a lot of sense. Thank you!
       
 (DIR) Post #AbpjLY5bs9T4IqVDoe by Paxxi@hachyderm.io
       2023-11-15T06:11:29Z
       
       1 likes, 0 repeats
       
       @bonealisa @mttaggart the idea that a bug like the blog describes would survive for years don't work. Everyone is chasing performance, enough so that Google project 0 are finding cpu bugs. If scheduling was broken in Linux one of the larger users would run into it pretty quickly