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: media: rockchip-rga: attach dma_resv release fence at buf_queue
Date: Tue, 05 May 2026 11:08:37 +1000
Message-ID: <review-patch3-20260429195306.239666-4-mfritsche@reauktion.de>
In-Reply-To: <20260429195306.239666-4-mfritsche@reauktion.de>
References: <3d8deeb15581b754e4c061d4c4a13657aa08bc3c.camel@ndufresne.ca> <20260429195306.239666-4-mfritsche@reauktion.de>
X-Mailer: Claude Code Patch Reviewer
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 7bit
MIME-Version: 1.0

Patch Review

**Same OUTPUT/CAPTURE issue as patch 2**

RGA is an M2M image processor. `rga_buf_queue` handles both source (OUTPUT) and destination (CAPTURE) buffers. The fence is attached unconditionally:
```c
v4l2_m2m_buf_queue(ctx->fh.m2m_ctx, vbuf);
...
(void)vb2_buffer_attach_release_fence(vb);
```

Same fix needed: guard with `V4L2_TYPE_IS_CAPTURE()`.

**Same verbose comment issue**

The 7-line comment block is excessive for a single function call.

---

**Summary of required fixes before this can move past RFC:**

1. Handle the requeue path (`VB2_BUF_STATE_QUEUED`) -- signal or clean up the old fence so re-attach doesn't WARN and leak.
2. Add defensive fence cleanup in `__vb2_queue_cancel`.
3. Guard fence attachment with `V4L2_TYPE_IS_CAPTURE()` in both driver patches (or move the check into the helper itself).
4. Optionally skip allocation when no planes have dmabufs.

---
Generated by Claude Code Patch Reviewer