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/vmwgfx: Fix hrtimer interrupt storm due to 0-period vblank Date: Mon, 25 May 2026 18:01:51 +1000 Message-ID: In-Reply-To: <20260518071741.441794-1-w15303746062@163.com> References: <20260518071741.441794-1-w15303746062@163.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/vmwgfx: Fix hrtimer interrupt storm due to 0-period vblank Author: w15303746062@163.com Patches: 3 Reviewed: 2026-05-25T18:01:51.532748 --- This is a v1=E2=86=92v2 iteration of a single patch fixing a real DoS vulne= rability: userspace can craft a display mode with huge `crtc_clock` and tin= y `htotal*vtotal`, causing `drm_calc_timestamping_constants()` to truncate = `framedur_ns` to 0. This leads to an infinite hrtimer hard-IRQ loop that lo= cks up the CPU. The v2 correctly moves the fix from the vmwgfx driver into the DRM core (`d= rm_crtc_vblank_start_timer()`), which is the right location =E2=80=94 the t= imer setup logic was refactored into the core, so the fix there protects al= l drivers using software vblank simulation, not just vmwgfx. The v2 also co= rrectly drops `WARN_ON_ONCE()` to avoid `panic_on_warn` issues from unprivi= leged userspace. The fix is a reasonable defensive bandaid. A more thorough long-term fix wo= uld reject these pathological modes earlier in the pipeline (e.g., in `atom= ic_check` or mode validation), but as a targeted DoS prevention this is app= ropriate for stable backport. --- Generated by Claude Code Patch Reviewer