amdgpu_bo_ge...erred_domain(9)

πŸ“¦ version 7.3.1, in nixpkgs unstable

πŸ’… NAME

ok so this is amdgpu_bo_ge...erred_domain(9) from the Kernel API Manual, and bestie, the whole personality of this page is literally just: amdgpu_bo_get_preferred_domain β€” get preferred domain. that's it. that's the vibe. short king energy πŸ‘‘


✨ SYNOPSIS ✨

uint32_t amdgpu_bo_get_preferred_domain (struct amdgpu_device *adev ,
uint32_t domain );

lowkey the cleanest signature ever β€” you hand it a device and a domain mask, you get back a uint32_t, no drama πŸŽ€


🌸 ARGUMENTS 🌸

  • adev β€” the amdgpu device object πŸ–₯️ like, the actual GPU you're talking about
  • domain β€” the allowed memory domains, i.e. the memory domains <amdgpu_memory_domains> ref. think of it as the menu of places the buffer could live, and this is the list of what's on the menu πŸ“

πŸ¦‹ RETURN πŸ¦‹

it tells you which of the allowed domains is the preferred one for allocating the BO. so like, out of everything you said was legal, which one does the kernel actually want? that one. slay πŸ’…βœ¨

ngl the return value is only as good as the domain mask you feed it β€” the allowed domains are exactly what you passed in, no cap.

πŸ”₯ SEE ALSO πŸ”₯

the kernel file is right here, go read it:

./drivers/gpu/drm/amd/amdgpu/amdgpu_object.c

and here's the whole friend group (yes, all of them, i'm not skipping anyone, that would be rude πŸ™ˆ):


ok that's the whole page, go allocate something iconic πŸ’œπŸ«Ά

./drivers/gpu/drm/amd/amdgpu Β· October 2026 Β· amdgpu_bo_ge...erred_domain(9) β€” mwah, that's a wrap, stream it, cite it, love it 🌈⭐

πŸ“¦ where this page comes from