xe_pt_zap_ptes_range(9)

πŸ“¦ version 7.3.1, in nixpkgs unstable

πŸ’… xe_pt_zap_ptes_range(9) β€” like, the whole vibe

okay so bestie, welcome to the Kernel API Manual ✨ this one lives in section 9 and it's all about zapping stuff (don't worry, it's a good kind of zapping, zeroing gpu ptes, very main character energy πŸ’…)


✨ NAME ✨

xe_pt_zap_ptes_range β€” zap (zero) gpu ptes of a SVM range 🧚

Like literally, it zeroes out gpu ptes across a whole SVM range. That's the whole job. Slay.


🌸 SYNOPSIS 🌸

bool xe_pt_zap_ptes_range (struct xe_tile *tile , struct xe_vm *vm ,
struct xe_svm_range *range );

πŸŽ€ ARGUMENTS πŸŽ€

  • tile β€” the tile we're zapping for, bestie. that's the one.
  • vm β€” the VM we're zapping for. yes, also that one.
  • range β€” the SVM range we're zapping for. it's giving very consistent, honestly 🫢

πŸ¦‹ DESCRIPTION πŸ¦‹

okay so here's the lore πŸ‘‘ β€” SVM invalidation needs to be able to zap the gpu ptes of a given address range. like, it's non-negotiable, it's the whole plot. and to actually pull that off, this function needs access to the shared page-table entries, so it can either clear the leaf PTEs or clear the pointers to lower-level page-tables. those are the two moves it's got, and it ate with both πŸ”₯

⚠️ but listen β€” the caller is required to hold the SVM notifier lock. that's not a suggestion, that's not a vibe, that's a hard requirement. don't come for this function without it, fr fr 😭

⭐ RETURN ⭐

it returns whether ptes were actually updated and a TLB invalidation is required. so like, a bool that tells you "did I do something and do you need to go flush things now" πŸ’€ useful.


🌈 SEE ALSO 🌈


and that's the page, bestie πŸ’– go forth and zap responsibly ✨

this whole thing came to you from ./drivers/gpu/drm/xe, dated October 2026 β€” mwah, take care, iykyk πŸ“πŸ’œ

πŸ“¦ where this page comes from