Claude review: drm/amdgpu: Move test for fbdev GEM object into generic helper

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

Patch Review

**Overall**: Good patch. The new helper correctly generalizes the amdgpu-specific check and is more flexible by also supporting the `fb_helper->fb` path (via `drm_gem_fb_get_obj`) when `buffer` isn't set.

**EXPORT_SYMBOL_GPL vs EXPORT_SYMBOL**: Every other export in `drm_fb_helper.c` uses `EXPORT_SYMBOL` (non-GPL). The new function uses:

```c
EXPORT_SYMBOL_GPL(drm_fb_helper_gem_is_fb);
```

This should be `EXPORT_SYMBOL` for consistency with the rest of the file. While both amdgpu and radeon are GPL-licensed so this works today, using `EXPORT_SYMBOL_GPL` here is inconsistent and could trip up any hypothetical non-GPL consumer that uses the other fb_helper functions.

**Typo in kdoc**: Minor typo in the documentation comment:

```c
 * Call drm_fb_helper_gem_is_fb to test is a DRM device's fbdev emulation
```

"test is" should be "test if".

**Logic correctness**: The helper's fallback logic is sound:

```c
if (fb_helper->buffer && fb_helper->buffer->gem)
    gem = fb_helper->buffer->gem;
else if (fb_helper->fb)
    gem = drm_gem_fb_get_obj(fb_helper->fb, 0);
```

This correctly handles the two paths: drivers using client buffers (amdgpu's original path checked `buffer->gem`) and drivers using `fb` directly (radeon's original path used `drm_gem_fb_get_obj(fb_helper->fb, 0)`). The original amdgpu code only checked `buffer->gem`, and the new helper adds the `fb` fallback, which is a strict superset — safe because if `buffer` is set, it will be checked first.

**Scoping improvement**: Nice touch moving `robj` into the inner block where it's actually used:

```c
-		robj = gem_to_amdgpu_bo(fb->obj[0]);
-		if (!amdgpu_display_robj_is_fb(adev, robj)) {
+		if (!drm_fb_helper_gem_is_fb(dev->fb_helper, fb->obj[0])) {
+			struct amdgpu_bo *robj = gem_to_amdgpu_bo(fb->obj[0]);
```

This avoids the unnecessary conversion when the object *is* the fbdev framebuffer.

**Header placement**: The `#else` static inline stub is correctly placed — the existing header had no `#else` for `CONFIG_DRM_FBDEV_EMULATION`, and this adds one cleanly.

---
Generated by Claude Code Patch Reviewer