From mboxrd@z Thu Jan 1 00:00:00 1970 From: Claude Code Review Bot To: dri-devel-reviews@example.com Subject: Claude review: Virtio-GPU S4 (hibernation) support Date: Tue, 05 May 2026 11:00:34 +1000 Message-ID: In-Reply-To: <20260429204729.993669-1-dongwon.kim@intel.com> References: <20260429204729.993669-1-dongwon.kim@intel.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: Virtio-GPU S4 (hibernation) support Author: dongwon.kim@intel.com Patches: 4 Reviewed: 2026-05-05T11:00:34.590670 --- This is a well-structured 3-patch series adding S4 (hibernation) support to= the virtio-gpu driver. The approach is sound: patch 1 adds the virtio free= ze/restore plumbing, patch 2 tracks GPU objects for restoration, and patch = 3 wires up the PM notifier to orchestrate unref-before-suspend and re-creat= e-on-resume. The series has been through significant review iteration (v8) = with Dmitry Osipenko and Nirmoy Das, and many earlier issues have been addr= essed.=20 However, several issues remain: - The `no_cb` parameter added to `virtio_gpu_cmd_unref_resource` has a late= nt use-after-free if the virtqueue submission fails. - `wait_event_timeout` return values are silently ignored during freeze, me= aning queue drain failures are undetected. - Minor style/type issues: `int no_cb` should be `bool`, unnecessary `#incl= ude`, and missing braces in a macro loop body. The overall design is reasonable =E2=80=94 virgl objects are correctly excl= uded (non-recoverable), imported dma-buf objects get special restoration ha= ndling, and the PM notifier cleanly scopes the work to S4 only. --- --- Generated by Claude Code Patch Reviewer