Claude review: drm: Rename drm_atomic_state

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

Overall Series Review

Subject: drm: Rename drm_atomic_state
Author: Maxime Ripard <mripard@kernel.org>
Patches: 2
Reviewed: 2026-04-28T14:55:34.132450

---

This is a 2-patch series by Maxime Ripard (v4, rebased on 7.1-rc1) to rename `struct drm_atomic_state` to `struct drm_atomic_commit` across the entire DRM subsystem. The motivation is well-established: the term "state" at the device level is confusing because `drm_atomic_state` doesn't represent the full device state — it represents a *commit* (a delta/update to a subset of objects). Per-object structs like `drm_crtc_state` genuinely represent the full state of that object, so the naming inconsistency causes confusion and bugs.

**Critical issue: Patch 1/2 is missing from the mbox.** The mbox only contains the cover letter (0/2) and patch 2/2. Patch 1/2 — "drm: Rename struct drm_atomic_state to drm_atomic_commit" — which is the actual 365-file, ~3500-line mechanical rename, is absent. This makes a full review of the series impossible. Only patch 2/2 (the documentation todo cleanup) can be reviewed.

**Naming concern with the existing `drm_atomic_commit()` function.** After the rename, both the struct and the existing function `drm_atomic_commit()` (at `drm_atomic.c:1774`) would share the same base name. Post-rename, the function signature becomes:
```c
int drm_atomic_commit(struct drm_atomic_commit *state)
```
While there's a certain elegance to "commit a commit," this creates ambiguity — `drm_atomic_commit` could refer to either the type or the function depending on context. In C this isn't a compiler issue (struct tags and function names occupy different namespaces), but it could confuse developers reading code or grepping the tree. This was presumably discussed in earlier revisions (v1 used a different name), but it's worth noting.

**Merge conflict risk.** Touching 365 files in one patch guarantees conflicts with virtually any in-flight DRM work. The cover letter correctly notes this should go through drm-next directly to minimize the window of pain.

**Skips file concern.** The vkms CI skip file (`vkms-none-skips.txt`) contains commented-out stack traces with function names like `drm_atomic_state_default_clear` and `__drm_atomic_state_free`. If patch 1/2 mechanically renames these, the skip patterns may no longer match actual kernel symbols, potentially changing CI behavior. These are in comments (prefixed with `#`), so the impact depends on whether the CI tooling parses them.

---
Generated by Claude Code Patch Reviewer