From mboxrd@z Thu Jan 1 00:00:00 1970 From: Claude Code Review Bot To: dri-devel-reviews@example.com Subject: Claude review: accel/qda: Add DMA-backed GEM objects and memory manager integration Date: Tue, 24 Feb 2026 18:20:12 +1000 Message-ID: In-Reply-To: <20260224-qda-firstpost-v1-10-fe46a9c1a046@oss.qualcomm.com> References: <20260224-qda-firstpost-v1-0-fe46a9c1a046@oss.qualcomm.com> <20260224-qda-firstpost-v1-10-fe46a9c1a046@oss.qualcomm.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 - Dmitry asked: "Why are you not using drm_gem_dma_helper?" =E2=80=94 this = is a key question. The DRM subsystem provides `drm_gem_dma_object` and help= ers specifically for DMA-backed GEM objects. Reimplementing this is unjusti= fied without a clear explanation. - **Namespace pollution**: Functions are not consistently named with the `q= da_` prefix. - **PID-based IOMMU device assignment**: The logic assigns IOMMU devices ba= sed on PID: - Dmitry asked: "What if there are two different FastRPC sessions within = the same PID?" This is a real scenario (multi-threaded DSP applications). - **Over-verbose logging** with debug enabled =E2=80=94 Dmitry noted this i= ndependently. - **Kernel-doc placement**: "Move the kerneldoc from the headers to the dri= ver code, otherwise they are mostly ignored by the automatic validators." --- Generated by Claude Code Patch Reviewer