From mboxrd@z Thu Jan 1 00:00:00 1970 From: Claude Code Review Bot To: dri-devel-reviews@example.com Subject: Claude review: Introduce cold reset recovery method Date: Sat, 16 May 2026 13:17:48 +1000 Message-ID: In-Reply-To: <20260512132614.1793083-7-mallesh.koujalagi@intel.com> References: <20260512132614.1793083-7-mallesh.koujalagi@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: Introduce cold reset recovery method Author: Mallesh Koujalagi Patches: 12 Reviewed: 2026-05-16T13:17:48.673793 --- This v5 series introduces cold reset recovery for Intel Xe GPUs through the= DRM wedge infrastructure. The architecture is: system controller reports h= ardware errors (PUNIT, memory, compute) via RAS, and when a PUNIT error is = detected that persists across warm resets, the driver wedges the device wit= h `DRM_WEDGE_RECOVERY_COLD_RESET`, notifying userspace to power-cycle the s= lot. **Patch 1** is a dependency blob from a separate series (explicitly marked = "DO NOT REVIEW"), so I'll focus on patches 2-5. The overall design of addin= g a new recovery method to the DRM wedge framework is straightforward and c= lean. However, there are several issues ranging from a compile-breaking bug= to a duplicate `#define`. **Key issues:** 1. **Critical bug (Patch 1):** `handle_page_offline()` uses `ret` uninitial= ized in a `switch` statement =E2=80=94 the call that should set it is comme= nted out as a TODO. This is undefined behavior and will be flagged by the c= ompiler. 2. **Duplicate define (Patch 1):** `XE_SYSCTRL_FLOOD` is defined in both `x= e_sysctrl_event_types.h` and `xe_sysctrl_mailbox_types.h`. 3. **Patch 5 naming:** Static functions use the `pcie_` prefix which could = be confused with PCI core API functions. --- --- Generated by Claude Code Patch Reviewer