Claude review: drm/amdgpu: Wire up dmem cgroup reclaim for VRAM manager

[Claude Code Review Bot <claude-review@example.com>]

Patch Review

**Verdict: Good, with one observation on the NULL place guard.**

**NULL place guard in `amdgpu_ttm_bo_eviction_valuable`:**

```c
 		if (amdkfd_fence_check_mm(f, current->mm) &&
-		    !(place->flags & TTM_PL_FLAG_CONTIGUOUS))
+		    !(place && (place->flags & TTM_PL_FLAG_CONTIGUOUS)))
 			return false;
```

When `place` is NULL (cgroup drain mode), `!(place && ...)` evaluates to `true`, so the condition becomes `amdkfd_fence_check_mm(f, current->mm) && true` → returns `false` (BO not evictable). This means KFD BOs belonging to `current->mm` are protected from cgroup reclaim.

In practice this is likely fine: the cgroup `max` write is typically done by a cgroup management process (systemd, container runtime), not by the KFD process itself, so `amdkfd_fence_check_mm(f, current->mm)` would be false and the check would pass. The conservative behavior (protect KFD BOs from the writing process) is the safe choice here.

**Cgroup registration wiring** mirrors patch 4 and uses the same `ttm_resource_manager_set_dmem_region` helper. Clean.

---
Generated by Claude Code Patch Reviewer