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/msm/dsi: Restore widebus calculation for CMDMode panels Date: Fri, 13 Mar 2026 14:40:21 +1000 Message-ID: In-Reply-To: <20260311-dsi-dsc-regresses-again-v1-1-6a422141eeea@somainline.org> References: <20260311-dsi-dsc-regresses-again-v1-1-6a422141eeea@somainline.org> X-Mailer: Claude Code Patch Reviewer Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable MIME-Version: 1.0 Overall Series Review Subject: drm/msm/dsi: Restore widebus calculation for CMDMode panels Author: Marijn Suijten Patches: 4 Reviewed: 2026-03-13T14:40:21.566414 --- This is a single-patch fix for a regression introduced by commit ac47870fd7= 95 ("drm/msm/dsi: fix hdisplay calculation when programming dsi registers")= . That commit removed the command-mode widebus path that set `bits_per_pclk= =3D 48`, incorrectly claiming CMD mode doesn't use this code path. The res= ult is that all DSI DSC command-mode panels on SM8350+ (with widebus suppor= t) broke =E2=80=94 the `hdisplay` calculation became wrong because `mipi_ds= i_pixel_format_to_bpp()` returns 24 for RGB888, identical to the non-widebu= s case, so widebus effectively did nothing. The fix is logically correct and restores the original behavior for CMD mod= e while preserving the video-mode change from ac47870fd795. **Note:** The current drm-next tree (line 1030 of `dsi_host.c`) already has= a different refactoring of this area that uses `bytes_per_pclk` (3/6) inst= ead of `bits_per_pclk` (24/48), and already includes the `!(msm_host->mode_= flags & MIPI_DSI_MODE_VIDEO)` guard. This means the regression may have alr= eady been fixed differently on drm-next, and this patch targets a tree wher= e ac47870fd795 exists but that subsequent refactoring hasn't landed. The ma= intainers should verify which tree this needs to go into (likely a stable b= ackport target). --- Generated by Claude Code Patch Reviewer