xe_observat...ranoid_check(9)

📦 version 7.3.1, in nixpkgs unstable

💅 xe_observation_paranoid_check — Kernel API Manual, bestie edition ✨

okay so before we even start: this page lives in the Kernel API Manual, and yes the title is the whole mouthful xe_observation_paranoid_check(9) — we do not shorten it, we respect her government name 🫶

🎀 NAME 🎀

xe_observation_paranoid_check — literally the bouncer at the club that is xe observation streams. she checks the list, she decides who gets in 💅

✨ SYNOPSIS ✨

int xe_observation_paranoid_check (void );

that's it. that's the whole vibe. no arguments, no drama 🌸

🦋 ARGUMENTS 🦋

  • void — no arguments, bestie. none. zero. the function said "I work alone" and honestly? main character energy 👑

💖 DESCRIPTION 💖

okay so here's the tea 🍓

There's this xe-specific sysctl called observation_paranoid, and it's enabled by default (iconic, cautious, we love her). When it's on, this function defers to perf_allow_cpu — which means access gets governed by the exact same policy as system-wide perf CPU events. And that policy is two things, no cap:

  • kernel.perf_event_paranoid
  • the security_perf_event_open LSM hook

so like... the same rules, the same gatekeeping, the same energy. consistency ate 🔥

But — and this is the plot twist — when a privileged user has cleared that sysctl, observation is just open to all callers. doors wide open, everyone's invited, no velvet rope 😭

🌈 RETURN 🌈

  • 0 — access is permitted, come on in, the water's fine 🫶
  • a negative errno — access denied bestie, it's giving "not on the list" 💀

👑 SEE ALSO 👑

ok so real talk: the paranoid sysctl is the default for a reason. if a privileged user clears it, observation is open to all callers — that's the documented behaviour, not me being dramatic. know what you're doing before you touch it 🙈

⭐ that's the whole page, bestie! this one comes from ./drivers/gpu/drm/xe, dated October 2026 — slay, see you in the next manpage 💜🎀

📦 where this page comes from