From mboxrd@z Thu Jan 1 00:00:00 1970 From: Claude Code Review Bot To: dri-devel-reviews@example.com Subject: Claude review: Support sparse mappings in Panthor Date: Thu, 23 Apr 2026 07:41:44 +1000 Message-ID: In-Reply-To: <20260422122538.3117380-1-adrian.larumbe@collabora.com> References: <20260422122538.3117380-1-adrian.larumbe@collabora.com> 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: Support sparse mappings in Panthor Author: =3D?UTF-8?q?Adri=3DC3=3DA1n=3D20Larumbe?=3D Patches: 7 Reviewed: 2026-04-23T07:41:44.304387 --- This v9 series adds sparse mapping support to the Panthor (Mali CSF) GPU dr= iver, allowing userspace (Vulkan) to bind virtual address ranges over a ker= nel-allocated dummy BO in a cyclic fashion. The approach of per-file-contex= t dummy BOs for information isolation is sound, and the series is well-stru= ctured with clean preparatory patches. However, there are several issues that need attention: 1. **uAPI enum ordering break** (Patch 1) =E2=80=94 inserting `DRM_PANTHOR_= DEV_QUERY_MMU_INFO` in the middle of an auto-numbered enum changes the valu= es of all subsequent members, breaking existing userspace. 2. **Memory leak on error path** (Patch 5) =E2=80=94 `panthor_vm_pool_creat= e` leaks memory if `panthor_dummy_bo_create` fails. 3. **Wrong address in hugepage alignment check** (Patch 5) =E2=80=94 the `o= p->next` branch of `unmap_hugepage_align` now passes `*unmap_start` instead= of the original `unmap_end - 1`, which is incorrect for non-sparse mapping= s. --- --- Generated by Claude Code Patch Reviewer