Conversation

Notices

1. Embed this notice
   Haelwenn /элвэн/ :triskell: (lanodan@queer.hacktivis.me)'s status on Friday, 17-Jan-2025 12:08:54 JST Haelwenn /элвэн/ :triskell:

   in reply to

   • miauz genyau
   • nina
   • ninee!

   @q66 @mia @0x9E01 Even just vm.overcommit_memory being a setting per cgroup or something like pid namespace would be great.
   At least here only glib-dependant software needs overcommitting, rest typically can handle malloc returning NULL.

   • Embed this notice
     nina (q66@blahaj.social)'s status on Friday, 17-Jan-2025 12:08:55 JST nina

     • miauz genyau
     • ninee!

     @mia @0x9E01 the whole idea behind overcommit is scuffed and it's unlike what basically every other os does and it makes it impossible to trust the allocator to reliably fail

     you can get less sucky *behavior* out of it with careful tweaking (and ideally a cgroups-aware userspace oom daemon) but it won't make it any less busted