Claude review: drm: add edid_overridden flag to drm_connector

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

Patch Review

**Approach concern:** The flag is set only inside `_drm_do_get_edid()`, which is the legacy EDID read path. While `drm_edid_read_ddc()` currently chains through `_drm_do_get_edid()`, this is an implementation detail that could change. A more robust approach would be to check at query time whether an override exists (e.g. via `drm_edid_override_get()` or checking `connector->edid_override`), rather than caching a flag that can become stale.

**Staleness risk:** The flag is set during EDID read, but nothing clears it if the override is removed between reads. If userspace removes an override via debugfs without triggering a new EDID read, `edid_overridden` remains `true`. The `connector->edid_overridden = false` at line 2376 only runs when `_drm_do_get_edid()` is called again.

**Missing Cc:** The patch modifies core DRM infrastructure but doesn't Cc `linux-kernel@vger.kernel.org` in the commit (though it is in the email headers from get_maintainer). More importantly, it would benefit from Cc'ing the DRM misc maintainers explicitly since this touches `drm_connector.h`.

**Naming:** `edid_overridden` is reasonable, but `edid_override_active` might be clearer since it describes current state rather than a past-tense event.

---
Generated by Claude Code Patch Reviewer