Claude review: drm/vmwgfx: Determine lock-waiting timeout from vblank state

[Claude Code Review Bot <claude-review@example.com>]

Patch Review

**Purpose:** Switches `vmw_vkms_lock_max_wait_ns()` from reading `du->vkms.period_ns` (the hrtimer period) to reading `vblank->framedur_ns` from the DRM vblank core. This decouples the lock timeout from the timer implementation details.

**Assessment: Looks good.**

The change from:
```c
s64 nsecs = ktime_to_ns(du->vkms.period_ns);
return (nsecs > 0) ? nsecs : 16666666;
```
to:
```c
struct drm_vblank_crtc *vblank = drm_crtc_vblank_crtc(crtc);
if (!vblank || !vblank->framedur_ns)
    return NSEC_PER_SEC / 60;
return vblank->framedur_ns;
```

is semantically equivalent. The fallback value `NSEC_PER_SEC / 60 = 16666666` matches the old hardcoded `16666666`.

**Minor observation:** `drm_crtc_vblank_crtc()` is a simple index into `dev->vblank`, which is allocated by `drm_vblank_init()`. The `!vblank` check is defensive — in practice, if `drm_vblank_init()` succeeded (which is required for `vkms_enabled` to be true), this shouldn't be NULL. But it's harmless and provides safety if the function is called before vblank init, which is reasonable.

No issues.

---

---
Generated by Claude Code Patch Reviewer