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: vfio/pci: Add a helper to look up PFNs for DMABUFs
Date: Thu, 28 May 2026 12:28:52 +1000
Message-ID: <review-patch2-20260527102319.100128-3-mattev@meta.com>
In-Reply-To: <20260527102319.100128-3-mattev@meta.com>
References: <20260527102319.100128-1-mattev@meta.com> <20260527102319.100128-3-mattev@meta.com>
X-Mailer: Claude Code Patch Reviewer
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 7bit
MIME-Version: 1.0

Patch Review

Adds `vfio_pci_dma_buf_find_pfn()` for translating fault addresses to PFNs via the DMABUF's `phys_vec[]` ranges.

**The comment block is quite long (lines 438-470)**. A 30+ line comment preceding the function duplicates what the code already communicates. Consider trimming to just the non-obvious parts (the `vma_pgoff_adjust` semantics and the hugepage/range-boundary limitation).

**The `vma_pgoff_adjust` field and its WARN**: 
```c
+	if (vpdmabuf->vma_pgoff_adjust != 0 &&
+	    vpdmabuf->vma_pgoff_adjust != (vma->vm_pgoff &
+					   (VFIO_PCI_OFFSET_MASK >> PAGE_SHIFT))) {
+		WARN(1, "Unexpected vma_pgoff_adjust 0x%lx (vm_pgoff 0x%lx)\n",
```
This WARN fires if `vma_pgoff_adjust` is nonzero and doesn't match the VMA's pgoff. The check is a debugging assertion that should arguably be a `WARN_ON_ONCE` rather than `WARN(1, ...)` to avoid spamming dmesg.

**Struct movement**: Moving `struct vfio_pci_dma_buf` to the header and adding `vma_pgoff_adjust` is clean. Good that `vma_pgoff_adjust` is `unsigned long` to match `vm_pgoff`.

---
Generated by Claude Code Patch Reviewer