Claude review: cgroup,cgroup/dmem: Add (dmem_)cgroup_common_ancestor helper

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

Patch Review

**Purpose:** Adds `cgroup_common_ancestor()` (generic cgroup helper) and `dmem_cgroup_common_ancestor()` (dmem-specific wrapper), plus refactors `get_cg_pool_unlocked()` into a lookup-only `find_cg_pool_unlocked()`.

**Bug — doc comment copy-paste error:**
```c
/**
 * dmem_cgroup_common_ancestor(): Find the first common ancestor of two pools.
 * @a: First pool to find the common ancestor of.
 * @b: First pool to find the common ancestor of.
```
`@b` should say **"Second pool"**, not "First pool."

**`cgroup_common_ancestor()` correctness:** The algorithm is correct — walking from `min(a->level, b->level)` down to 0, comparing the `ancestors[]` arrays. Since all cgroups in the same hierarchy share the root at level 0, this will always find an ancestor (never return NULL) for cgroups in the same subsystem.

**Reference counting concern:** `dmem_cgroup_common_ancestor()` calls `find_cg_pool_unlocked()` which performs an RCU lookup. If this returns a pool pointer without grabbing a reference, then patch 6's `dmem_cgroup_pool_state_put(ancestor)` would be unbalanced. The diff context doesn't show enough of the original `get_cg_pool_unlocked()` to confirm whether `find_cg_pool_unlocked()` grabs a `css_tryget()` reference. Please verify that `find_cg_pool_unlocked()` returns a properly referenced pool, or add explicit reference counting in `dmem_cgroup_common_ancestor()`.

The refactoring of `get_cg_pool_unlocked()` into `find_cg_pool_unlocked()` + `get_cg_pool_unlocked()` is clean — the "find" variant does the lookup without allocation, and the "get" variant retains the allocate-on-miss behavior.

---
Generated by Claude Code Patch Reviewer