From mboxrd@z Thu Jan 1 00:00:00 1970 From: Claude Code Review Bot To: dri-devel-reviews@example.com Subject: Claude review: drm/amd/display: Change dither policy for 10 bpc output back to dithering Date: Sun, 22 Mar 2026 03:13:39 +1000 Message-ID: In-Reply-To: <20260321052033.23472-1-mario.kleiner.de@gmail.com> References: <20260321052033.23472-1-mario.kleiner.de@gmail.com> <20260321052033.23472-1-mario.kleiner.de@gmail.com> X-Mailer: Claude Code Patch Reviewer Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable MIME-Version: 1.0 Patch Review **Correctness:** The change at `dc_resource.c:5057-5058` from `DITHER_OPTIO= N_TRUN10` to `DITHER_OPTION_SPATIAL10` is correct. Looking at the downstrea= m handling: - `DITHER_OPTION_SPATIAL10` is handled at lines 5125-5133, which sets `SPAT= IAL_DITHER_ENABLED =3D 1`, `SPATIAL_DITHER_DEPTH =3D 2`, along with `HIGHPA= SS_RANDOM` and conditional `RGB_RANDOM`. This is the proper spatial ditheri= ng path. - The old `DITHER_OPTION_TRUN10` at lines 5076-5086 only set truncation wit= h `TRUNCATE_MODE =3D 1` (rounding), losing the precision benefit of spatial= dithering. - The pattern is now consistent with `COLOR_DEPTH_666 =E2=86=92 DITHER_OPTI= ON_SPATIAL6` and `COLOR_DEPTH_888 =E2=86=92 DITHER_OPTION_SPATIAL8`. **Commit message quality:** Excellent. The author provides thorough technic= al justification for the revert, explains the real-world impact (scientific= applications, fp16/rgba16 framebuffers, the 8 bpc vs 10 bpc absurdity), na= mes specific software that benefits (Mesa 25.3, GNOME 50 mutter), and docum= ents actual hardware testing (Polaris DCE 11.2 and Raven Ridge DCN 1.0 with= photometer measurements). **Minor observations:** - The `Fixes:` tag and `Cc: stable` are appropriate since this is restoring= a regression in precision for users of > 10 bpc pipelines. - The only concern would be whether there exists some hardware where spatia= l dithering to 10 bpc causes visual artifacts =E2=80=94 but the author expl= icitly tested on two different generations (Polaris and Raven Ridge) with n= o issues, and the original commit that introduced truncation did not cite a= ny such hardware problems either. **No issues found.** This is a clean, well-justified fix. --- Generated by Claude Code Patch Reviewer