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/amdgpu: clamp user gartsize against device capacity Date: Sat, 16 May 2026 10:27:15 +1000 Message-ID: In-Reply-To: <20260514194937.35649-1-en22ue@gmail.com> References: <20260514194937.35649-1-en22ue@gmail.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: drm/amdgpu: clamp user gartsize against device capacity Author: Ahmed Elmetwally Patches: 1 Reviewed: 2026-05-16T10:27:15.093189 --- **This patch is based on a stale tree and will not apply.** The upstream co= de has already been refactored: the inline gart size logic in each `gmc_v*_= 0.c` backend was consolidated into a shared `amdgpu_gmc_set_gart_size()` he= lper (see `amdgpu_gmc.c:283`). The patch's diff context in `gmc_v11_0.c` ex= pects the old inline code (`if (amdgpu_gart_size =3D=3D -1) ... else ...`) = which no longer exists =E2=80=94 the current tree has a single call to `amd= gpu_gmc_set_gart_size(adev, SZ_512M)` at line 712. The underlying problem (an absurdly large `amdgpu.gartsize` parameter causi= ng probe failure) is real and the commit message is well-written with a cle= ar reproducer. However, the correct fix should modify the existing `amdgpu_= gmc_set_gart_size()` function rather than introducing a parallel `amdgpu_gm= c_validate_gart_size()` helper. This would fix all gmc backends (v6 through= v12) in one shot rather than requiring per-backend follow-up patches. Additionally, the patch has several technical issues detailed below. --- --- Generated by Claude Code Patch Reviewer