From mboxrd@z Thu Jan  1 00:00:00 1970
From: Claude Code Review Bot <claude-review@example.com>
To: dri-devel-reviews@example.com
Subject: Claude review: drm/amdgpu: hook up ZONE_DEVICE registration in device init and reset
Date: Sat, 16 May 2026 12:15:39 +1000
Message-ID: <review-patch5-20260513095734.69598-6-Junhua.Shen@amd.com>
In-Reply-To: <20260513095734.69598-6-Junhua.Shen@amd.com>
References: <20260513095734.69598-1-Junhua.Shen@amd.com> <20260513095734.69598-6-Junhua.Shen@amd.com>
X-Mailer: Claude Code Patch Reviewer
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0

Patch Review

**Unnecessary `#if IS_ENABLED` guards:**
```c
#if IS_ENABLED(CONFIG_DRM_AMDGPU_SVM)
	amdgpu_svm_migration_init(adev);
#endif
```
The header already provides a static inline stub returning 0 when `CONFIG_D=
RM_AMDGPU_SVM` is disabled. The `#if` guard is redundant and adds visual no=
ise. Just call the function directly.

**Return value of `amdgpu_svm_migration_init` is ignored** in both callsite=
s. If initialization fails (e.g., `devm_memremap_pages` fails), the device =
will continue without VRAM migration support. The cover letter says `adev->=
apagemap` is NULL when disabled, and the helper functions check for this, s=
o this is likely intentional =E2=80=94 but a `dev_warn` on failure would he=
lp debugging.

**In the reset path:**
```c
if (!tmp_adev->kfd.init_complete) {
#if IS_ENABLED(CONFIG_DRM_AMDGPU_SVM)
	amdgpu_svm_migration_init(tmp_adev);
#endif
	kgd2kfd_init_zone_device(tmp_adev);
```
`amdgpu_svm_migration_init` has an early-return check `if (adev->apagemap &=
& adev->apagemap->initialized) return 0;` =E2=80=94 so calling it again aft=
er reset is safe. Good.

---
Generated by Claude Code Patch Reviewer