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/panthor: Lazily allocate pages on mmap() Date: Tue, 31 Mar 2026 17:24:46 +1000 Message-ID: In-Reply-To: <20260330094848.2169422-7-boris.brezillon@collabora.com> References: <20260330094848.2169422-1-boris.brezillon@collabora.com> <20260330094848.2169422-7-boris.brezillon@collabora.com> X-Mailer: Claude Code Patch Reviewer Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit MIME-Version: 1.0 Patch Review Good refactoring of the fault handler into non-blocking/blocking paths. One concern: In `panthor_gem_any_fault`, when `fault_flag_allow_retry_first` is true and `FAULT_FLAG_RETRY_NOWAIT` is not set, the code does: ```c drm_gem_object_get(&bo->base); mmap_read_unlock(vmf->vma->vm_mm); ret = blocking_page_setup(vmf, order, bo, page_offset, false); drm_gem_object_put(&bo->base); return ret; ``` This correctly grabs a reference before releasing the mmap lock. In `blocking_page_setup` with `mmap_lock_held=false`, on error, it returns `VM_FAULT_RETRY` rather than `VM_FAULT_SIGBUS`, which is correct since the mmap lock isn't held. No issues. --- Generated by Claude Code Patch Reviewer