Claude review: drm/colorop: Preserve bypass value in duplicate_state()

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

Patch Review

**Verdict: Looks good.**

The fix removes the unconditional `state->bypass = true` assignment after `memcpy()` in `__drm_atomic_helper_colorop_duplicate_state()`:

```c
 	if (state->data)
 		drm_property_blob_get(state->data);
-
-	state->bypass = true;
```

This is clearly correct. The `memcpy()` on line 465 already copies the entire state including `bypass`. Overwriting it to `true` afterwards defeats the purpose of duplicating the current hardware state — exactly the same pattern used for plane and CRTC state duplication, where no fields are unconditionally reset to non-hardware values after the copy.

As the cover letter notes, this is also a problem for colorops where bypass is immutably false — the forced `true` would be contradicting the hardware capabilities.

One question: was there an original reason this line existed? Looking at the commit message for `8c5ea1745f4c` ("drm/colorop: Add BYPASS property") might reveal if this was intentional to ensure bypass-by-default semantics for userspace-initiated commits. If so, the change is still correct (userspace should set bypass explicitly), but it's worth understanding the original intent to ensure no regressions.

---
Generated by Claude Code Patch Reviewer