ib_umem_get_attr(9)

πŸ“¦ version 7.3.1, in nixpkgs unstable

πŸ’– NAME πŸ’–

ib_umem_get_attr β€” Pin a umem from a per-command UMEM attribute. Yep, that's the assignment, and it's lowkey iconic. This page lives in the Kernel API Manual, so we're doing kernel-API group-chat now 🫢.

✨ SYNOPSIS ✨

okay so the exact signature is right here, no cap:

struct ib_umem * ib_umem_get_attr (struct ib_device *device , const
struct uverbs_attr_bundle *attrs , u16 attr_id , size_t size , int
access );

πŸŽ€ ARGUMENTS πŸŽ€

  • device β€” the IB device, bestie.
  • attrs β€” the uverbs attribute bundle (may be NULL). Yes, it can be NULL, slay.
  • attr_id β€” the per-command UMEM attribute id.
  • size β€” the minimum required umem length.
  • access β€” the IB access flags.

🌸 RETURN 🌸

On success you get a caller-owned umem back, periodt. You get NULL when no source supplied a buffer. On error it's ERR_PTR(...), and that's the whole dramatic arc 😭.

πŸ¦‹ SEE ALSO πŸ¦‹

The Kernel file is ./drivers/infiniband/core/umem.c, and the SEE ALSO crew is:


Signed off from ./drivers/infiniband/core β€” October 2026 β€” with ib_umem_get_attr(9) energy πŸ’Œ

πŸ“¦ where this page comes from