public inbox for drm-ai-reviews@public-inbox.freedesktop.org
 help / color / mirror / Atom feed
From: Claude Code Review Bot <claude-review@example.com>
To: dri-devel-reviews@example.com
Subject: Claude review: vfio/pci: Add mmap() attributes to DMABUF feature
Date: Thu, 28 May 2026 12:28:53 +1000	[thread overview]
Message-ID: <review-patch9-20260527102319.100128-10-mattev@meta.com> (raw)
In-Reply-To: <20260527102319.100128-10-mattev@meta.com>

Patch Review

Adds `VFIO_DEVICE_FEATURE_DMA_BUF_MEMATTR` for setting UC/WC memory attributes on exported DMABUFs.

**Bug -- wrong errno**:
```c
+			default:
+				ret = -ENOTSUPP;
```
`-ENOTSUPP` is `524` (kernel-internal, `EOPNOTSUPP` equivalent for in-kernel use only). For a UAPI-facing path that returns to userspace, this should be `-EOPNOTSUPP` (`95`). The docstring even says `ENOTSUPP` but userspace will see the raw `524` value which is non-standard. Fix:
```c
				ret = -EOPNOTSUPP;
```

**GET+SET combined semantics**: When both `VFIO_DEVICE_FEATURE_GET` and `VFIO_DEVICE_FEATURE_SET` are specified, the code applies SET first, then returns the *old* value via GET:
```c
+		uint32_t old_attr = priv->memattr;
+		if (flags & VFIO_DEVICE_FEATURE_SET) { ... }
+		db_attr.memattr = old_attr;
+	}
+	if (!ret && (flags & VFIO_DEVICE_FEATURE_GET)) {
+		if (copy_to_user(arg, &db_attr, sizeof(db_attr)))
```
This "swap" semantic (set new, return old) is unusual and should be documented in the UAPI header. Users would typically expect GET to return the current (post-SET) value. Consider whether this is the intended behavior or a bug.

**`memattr` initialization for BAR mmap DMABUFs**: In `vfio_pci_core_mmap_prep_dmabuf()`, `priv->memattr` is not explicitly set, relying on `kzalloc` zero-initialization to give `VFIO_DEVICE_FEATURE_DMA_BUF_MEMATTR_NC = 0`. This is fine but fragile -- if the enum values ever change, this would break. The explicit `VFIO_DEVICE_FEATURE_DMA_BUF_MEMATTR_NC` enum name is used in patch 9 for the explicit export path, so it would be consistent to also set it explicitly in the mmap path. (Low priority.)

**Naming inconsistency**: The enum uses `_NC` (noncached) but the UAPI docstring says "The default ... is NC." Meanwhile patch 7 uses `pgprot_noncached`. The cover letter and commit message say "UC" (uncached). These are the same thing in practice, but the UAPI should pick one term and stick with it. `UC` is more conventional for MMIO.

---
Generated by Claude Code Patch Reviewer

  reply	other threads:[~2026-05-28  2:28 UTC|newest]

Thread overview: 25+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-05-27 10:23 [PATCH v2 0/9] vfio/pci: Add mmap() for DMABUFs Matt Evans
2026-05-27 10:23 ` [PATCH v2 1/9] PCI/P2PDMA: Add CONFIG_PCI_P2PDMA_CORE Matt Evans
2026-05-27 16:07   ` Logan Gunthorpe
2026-05-27 17:13     ` Matt Evans
2026-05-27 21:09   ` Alex Williamson
2026-05-28  2:28   ` Claude review: " Claude Code Review Bot
2026-05-27 10:23 ` [PATCH v2 2/9] vfio/pci: Add a helper to look up PFNs for DMABUFs Matt Evans
2026-05-27 22:38   ` Alex Williamson
2026-05-28  2:28   ` Claude review: " Claude Code Review Bot
2026-05-27 10:23 ` [PATCH v2 3/9] vfio/pci: Add a helper to create a DMABUF for a BAR-map VMA Matt Evans
2026-05-27 22:59   ` Alex Williamson
2026-05-28  2:28   ` Claude review: " Claude Code Review Bot
2026-05-27 10:23 ` [PATCH v2 4/9] vfio/pci: Convert BAR mmap() to use a DMABUF Matt Evans
2026-05-28  2:28   ` Claude review: " Claude Code Review Bot
2026-05-27 10:23 ` [PATCH v2 5/9] vfio/pci: Provide a user-facing name for BAR mappings Matt Evans
2026-05-28  2:28   ` Claude review: " Claude Code Review Bot
2026-05-27 10:23 ` [PATCH v2 6/9] vfio/pci: Clean up BAR zap and revocation Matt Evans
2026-05-28  2:28   ` Claude review: " Claude Code Review Bot
2026-05-27 10:23 ` [PATCH v2 7/9] vfio/pci: Support mmap() of a VFIO DMABUF Matt Evans
2026-05-28  2:28   ` Claude review: " Claude Code Review Bot
2026-05-27 10:23 ` [PATCH v2 8/9] vfio/pci: Permanently revoke a DMABUF on request Matt Evans
2026-05-28  2:28   ` Claude review: " Claude Code Review Bot
2026-05-27 10:23 ` [PATCH v2 9/9] vfio/pci: Add mmap() attributes to DMABUF feature Matt Evans
2026-05-28  2:28   ` Claude Code Review Bot [this message]
2026-05-28  2:28 ` Claude review: vfio/pci: Add mmap() for DMABUFs Claude Code Review Bot

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=review-patch9-20260527102319.100128-10-mattev@meta.com \
    --to=claude-review@example.com \
    --cc=dri-devel-reviews@example.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox