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: Re: [PATCH] drm/appletbdrm: Use kvzalloc for big allocations
Date: Thu, 23 Apr 2026 09:55:24 +1000
Message-ID: <review-patch1-302f22e3-5390-4c0a-aa06-bfd02a3175c0@suse.de>
In-Reply-To: <302f22e3-5390-4c0a-aa06-bfd02a3175c0@suse.de>
References: <302f22e3-5390-4c0a-aa06-bfd02a3175c0@suse.de> <302f22e3-5390-4c0a-aa06-bfd02a3175c0@suse.de>
X-Mailer: Claude Code Patch Reviewer
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0

Patch Review

**NAK =E2=80=94 vmalloc memory cannot be used as a USB DMA transfer buffer.=
**

The allocation site:
```c
-	appletbdrm_state->request =3D kzalloc(request_size, GFP_KERNEL);
+	appletbdrm_state->request =3D kvzalloc(request_size, GFP_KERNEL);
```

This buffer is subsequently passed to USB bulk transfer at `appletbdrm.c:44=
9`:
```c
ret =3D appletbdrm_send_request(adev, &request->header, request_size);
```

Which calls `usb_bulk_msg()` at line 160:
```c
ret =3D usb_bulk_msg(udev, usb_sndbulkpipe(udev, adev->out_ep),
		   request, size, &actual_size, APPLETBDRM_BULK_MSG_TIMEOUT);
```

The USB HCD layer DMA-maps this buffer via `dma_map_single()` (`drivers/usb=
/core/hcd.c:1488`), and the DMA subsystem explicitly rejects vmalloc addres=
ses (`include/linux/dma-mapping.h:513-515`). When `kvzalloc` falls back to =
vmalloc under memory pressure =E2=80=94 the exact scenario this patch targe=
ts =E2=80=94 the USB transfer will fail.

The free-side change is correct in isolation (matching `kvzalloc` with `kvf=
ree`), but is moot given the allocation change is wrong:
```c
-	kfree(appletbdrm_state->request);
+	kvfree(appletbdrm_state->request);
```

**Suggestion:** Replace `kvzalloc` with `kzalloc` using `__GFP_RETRY_MAYFAI=
L`:
```c
appletbdrm_state->request =3D kzalloc(request_size, GFP_KERNEL | __GFP_RETR=
Y_MAYFAIL);
```

This tells the page allocator to try harder (compacting memory, reclaiming =
pages) before giving up, while keeping the allocation physically contiguous=
 and safe for DMA. No change needed on the free side since `kfree` remains =
correct.

---
Generated by Claude Code Patch Reviewer