From mboxrd@z Thu Jan 1 00:00:00 1970 From: Claude Code Review Bot To: dri-devel-reviews@example.com Subject: Claude review: soc/qcom/ubwc: rework UBWC configuration database Date: Mon, 25 May 2026 21:41:56 +1000 Message-ID: In-Reply-To: <20260520-ubwc-rework-v5-0-72f2749bc807@oss.qualcomm.com> References: <20260520-ubwc-rework-v5-0-72f2749bc807@oss.qualcomm.com> X-Mailer: Claude Code Patch Reviewer Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit MIME-Version: 1.0 Overall Series Review Subject: soc/qcom/ubwc: rework UBWC configuration database Author: Dmitry Baryshkov Patches: 29 Reviewed: 2026-05-25T21:41:56.176616 --- This is a well-structured 28-patch series that reworks the Qualcomm UBWC (Universal Bandwidth Compression) configuration database. The core thesis is sound: most UBWC configuration parameters are derivable from the UBWC encoder version, so they should be computed by inline helper functions rather than stored per-SoC in a data table. This eliminates redundancy, reduces the chance of copy-paste errors when adding new SoCs, and makes the relationship between UBWC version and configuration parameters explicit in code. **Architecture:** The series follows a clean three-phase approach: 1. **Patches 01-03:** Add new UBWC version constants and helper functions 2. **Patches 04-21:** Migrate all driver consumers to use the new helpers 3. **Patches 22-28:** Remove the now-redundant fields from the database and deduplicate entries This ordering is correct -- helpers are introduced before use, consumers are migrated before data is removed. Each patch is individually bisectable within its phase. **Strengths:** - The final `qcom_ubwc_cfg_data` struct is dramatically simplified: only `ubwc_enc_version`, `highest_bank_bit`, and `flags` remain, down from 6+ fields - Patch 28's deduplication is a satisfying payoff -- 30+ per-SoC structs collapse to ~15 generic `ubwc_X_Y_hbbZ` entries - All patches have Reviewed-by tags from Konrad Dybcio, and several from Akhil P Oommen - The version-based derivation logic is well-documented in the helper functions **Concerns:** - The introduction of UBWC_3_1 in patches 01/25 warrants scrutiny -- it reclassifies several SoCs (sc7280, sc8180x, sar2130p) from UBWC_3_0 to UBWC_3_1, which affects the macrotile_mode derivation. This is the highest-risk change in the series. - Patch 18 converts a switch statement to if-else chains that mix `ubwc_version` (local) and `cfg->ubwc_enc_version` (direct field access) inconsistently. - The `amsbc` helper in patch 19 changes behavior for the MDSS driver: the old code checked `== UBWC_3_0`, but the helper returns `>= UBWC_3_0`. This is a functional change for MDSS that depends on patches 14/19 being on the same base. **Verdict:** The series is in good shape for a v5. The design is clean, the ordering is correct, and the end result is clearly superior to the current state. A few minor nits noted per-patch below, but nothing blocking. --- --- Generated by Claude Code Patch Reviewer