Post by @etchedpixels@mastodon.social

2026-09-20 10:54 UTC

@qwertydotplus@mastodon.social The no overcommit feature is disabled on Linux by default because it meant people had to write software properly. All the swapping/paging behaviour on Linux is fixable. There are algorithms that deal with it and they are well understood. Just nobody cares enough to bother. Firstly you don't allow more anonymous address space to exist that you have swap for. This fixes much of it but upsets all the sad "but I just mmap 10GB and figure it out later" people

Replies (2)

@etchedpixels@mastodon.social 2026-09-20 10:57

@qwertydotplus@mastodon.social The second set of things are much cleverer. When your paging rate becomes high you start moving to a more swap like behaviour. You target paging in from a small set of tasks and paging out others, when they have made some progress (actual CPU time doing useful work) you move onto the next ones. This sharply reduces the percentage of time the machine spends thrashing at the expense of responsiveness. In classic (8MB RAM) days you simply switched from paging to swapping.

Open ##5093876
@qwertydotplus@mastodon.social 2026-09-20 13:56

@etchedpixels@mastodon.social the no overcommit feature is disabled on linux because it means EVERY piece of software has to be built properly to avoid faulting every OTHER piece of software. if some application misbehaves, allocates memory, and then doesn't use it, that memory becomes unusable for everyone else. the only way for other applications to work around this is to preallocate and hoard a large chunk of memory. which just means fewer things can start running before stuff breaks.

Open ##5093877