Claude review: drm/atomic: track individual colorop updates

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

Patch Review

This is the core patch. It sets `plane_state->color_mgmt_changed` whenever a colorop property value actually changes.

**Change to `drm_atomic_set_colorop_for_plane`**: Now returns `bool` indicating whether the color pipeline pointer actually changed. The early-out when `plane_state->color_pipeline == colorop` avoids unnecessary state dirtying when userspace re-sets the same pipeline. The only call site at line 615 uses the return value correctly:

```c
state->color_mgmt_changed |= drm_atomic_set_colorop_for_plane(state, colorop);
```

The function is `EXPORT_SYMBOL` so theoretically out-of-tree callers exist, but since the return value is ignorable (void→bool), this is ABI-safe.

**Change to `drm_atomic_colorop_set_property`**: Each property branch now does a compare-before-assign and sets `*replaced = true` only on actual value change. This is correct for all scalar properties (bypass, lut1d_interpolation, curve_1d_type, multiplier, lut3d_interpolation). The `data_property` case passes `replaced` through to `drm_atomic_color_set_data_property` which delegates to `drm_property_replace_blob_from_id` — that function already computes `replaced` correctly for blobs.

**Change to `drm_atomic_set_property` (COLOROP case)**: After detecting a change, it pulls in the plane state and sets `color_mgmt_changed`:

```c
ret = drm_atomic_colorop_set_property(colorop, colorop_state,
                                      file_priv, prop, prop_value,
                                      &replaced);
if (ret || !replaced)
    break;

plane_state = drm_atomic_get_plane_state(state, colorop->plane);
...
plane_state->color_mgmt_changed |= replaced;
```

**Minor observation**: At line 1325, `plane_state->color_mgmt_changed |= replaced;` is guarded by `if (ret || !replaced)` above (line 1317), so `replaced` is always `true` when line 1325 is reached. The `|=` is harmless (and consistent with the style used elsewhere), but `= true` would be equivalent. This is not a bug — the existing code is fine for consistency and future-proofing.

**`#include <linux/types.h>`** added to `drm_atomic_uapi.h` for the `bool` return type — correct.

No issues.

---

---
Generated by Claude Code Patch Reviewer