[PATCH v3 0/2] rust: drm: gpuvm: implement Send and Sync for refcounted handles

[Sami Tolvanen <samitolvanen@google.com>]

Hi folks,

GpuVaAlloc and GpuVmBo implement neither Send or Sync, so drivers
that move them between threads need their own unsafe impls. GpuVm
implements both unconditionally, which Sashiko points out is unsound.

Patch 1 adds Send and Sync for GpuVaAlloc (unconditionally) and
GpuVmBo (similarly to Arc). Patch 2 gives GpuVm the same bounds as
it can alias types using obtain() and drop them across the threads
via deferred_cleanup().

Tested with downstream Tyr as there don't appear to be any in-tree
users yet.

	Sami

---
v3:
- Added a patch for GpuVm, and changed GpuVmBo bounds based on
  another Sashiko analysis.

v2:
- Added a missing T::Object: Send + Sync bound pointed out by
  Sashiko.

---
Sami Tolvanen (2):
  rust: drm: gpuvm: implement Send and Sync for GpuVaAlloc and GpuVmBo
  rust: drm: gpuvm: require Send and Sync for GpuVm's shared data

 rust/kernel/drm/gpuvm/mod.rs   | 22 +++++++++++++++++++---
 rust/kernel/drm/gpuvm/va.rs    |  8 ++++++++
 rust/kernel/drm/gpuvm/vm_bo.rs | 22 ++++++++++++++++++++++
 3 files changed, 49 insertions(+), 3 deletions(-)


base-commit: fea3a2dd7d3fc1936211ced5f84420e610435730
-- 
2.54.0.1099.g489fc7bff1-goog