Conversation

Notices

1. Embed this notice
   Haelwenn /элвэн/ :triskell: (lanodan@queer.hacktivis.me)'s status on Friday, 29-Mar-2024 07:26:21 JST Haelwenn /элвэн/ :triskell:
   @chjara Or like memory poisoning.
   
   inb4:
   - Nooo, my performance.
   - Well if you'd care about performance, you also wouldn't be duplicating lists/strings to make sublists/substrings all the time.
   • Embed this notice
     niconiconi (niconiconi@mk.absturztau.be)'s status on Friday, 29-Mar-2024 08:18:59 JST niconiconi

     in reply to

     @lanodan@queer.hacktivis.me @chjara@akko.wtf There are many well-known mitigation techniques, like memory poisoning, and have already existed for 10+ years. But many are very aggressive (to them, Linux is exploitable despite some mitigations because they're too half-assed). Meanwhile kernel developers don't like them and think they're either too invasive or too paranoid (e.g. In PaX, you must call pax_open_kernel() whenever you need to change a critical kernel data structure). Linus in particular, hates security people and think they're mostly impractical jerks.

     Haelwenn /элвэн/ :triskell: likes this.
   • Embed this notice
     niconiconi (niconiconi@mk.absturztau.be)'s status on Friday, 29-Mar-2024 08:32:44 JST niconiconi

     in reply to

     @lanodan@queer.hacktivis.me @chjara@akko.wtf For the note, PaX supported zero-poison after free (PAX_MEMORY_SANITIZE) since the year 2010. But of course LKML and PaX Team hate each other so it took 7 years before the Google / Kees Cook gang pushed it for upstreaming.
     
     How it started: https://lwn.net/Articles/386706/
     How it is going: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=1414c7f4f7d72d138fff35f00151d15749b5beda

     Haelwenn /элвэн/ :triskell: likes this.
   • Embed this notice
     Haelwenn /элвэн/ :triskell: (lanodan@queer.hacktivis.me)'s status on Friday, 29-Mar-2024 08:34:12 JST Haelwenn /элвэн/ :triskell:

     in reply to

     • niconiconi
     @niconiconi @chjara heh checked and I have CONFIG_PAGE_POISONING=y on my gentoo machines. ^^
     
     (tfw alpine doesn't have /proc/config.gz enabled so it's annoying to check)