i915_vma_revoke_fence(9)

πŸ“¦ version 7.3.1, in nixpkgs unstable

πŸ’… i915_vma_revoke_fence(9) πŸ’…

okay so welcome to the i915_vma_revoke_fence(9) page, straight out of the Kernel API Manual bestie. this one's short and sweet but it's literally load-bearing, so pay attention πŸ’–


✨ NAME ✨

i915_vma_revoke_fence β€” force-remove fence for a VMA

like, force-remove. not "ask nicely", not "politely suggest" β€” we are evicting that fence 🫢


πŸ¦‹ SYNOPSIS πŸ¦‹

void i915_vma_revoke_fence (struct i915_vma *vma );

yes that's the whole thing. void return, one argument, no drama (the drama is in the fence itself, iykyk) ✨


πŸŽ€ ARGUMENTS πŸŽ€

  • vma β€” the vma to map linearly (not through a fence reg). like, we're doing the direct route here, no fence-reg middleman, no detour, just linear mapping energy πŸ’…

🌸 DESCRIPTION 🌸

okay so what does it actually do: this function force-removes any fence from the given object. every single one. no survivors.

and why would you want that? it's useful if the kernel wants to do untiled GTT access β€” because a fence is basically sitting there in the way being like "um actually i'm sorting the tiling" and the kernel's like "not today bestie" 😭

so yeah, the fence gets revoked. it's giving main character energy for the kernel, honestly πŸ‘‘


🧚 SEE ALSO 🧚

the whole fence friend group fr fr, go read them next, they all ate πŸ”₯


ngl there's nothing dangerous documented here, but remember: you're force-removing something. if some other part of your code was lowkey relying on that fence still being there... that's a you problem, bestie πŸ™ˆ

k bye, hope this helped!! this page came from ./drivers/gpu/drm/i915/gt and it's dated October 2026 β€” mwah, slay, drink some water, and remember to git blame responsibly πŸ“πŸ’œβ­

πŸ“¦ where this page comes from