Linux PCI subsystem development
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "Gary Guo" <gary@garyguo.net>
Cc: "Danilo Krummrich" <dakr@kernel.org>,
	linux-pci@vger.kernel.org, dri-devel@lists.freedesktop.org,
	ojeda@kernel.org, "Alexandre Courbot" <acourbot@nvidia.com>
Subject: Re: [PATCH v3 05/19] rust: io: generalize `MmioRaw` to pointer to arbitrary type
Date: Mon, 08 Jun 2026 20:14:46 +0000	[thread overview]
Message-ID: <20260608201447.023961F00893@smtp.kernel.org> (raw)
In-Reply-To: <20260608-io_projection-v3-5-c5cde13a5ec4@garyguo.net>

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] Strict `usize` alignment check in `Region::ptr_try_from_raw_parts_mut` breaks driver probes for valid 4-byte aligned MMIO mappings.
--

commit facc65605656f1d5585f26a25f200941e6ae7d73
Author: Gary Guo <gary@garyguo.net>

rust: io: generalize MmioRaw to pointer to arbitrary type

This commit generalizes MmioRaw to represent an untyped I/O region using
the dynamically-sized Region<SIZE> type. It renames the maxsize method
to size and implements Clone, Copy, Send, and Sync manually.

> diff --git a/rust/kernel/io.rs b/rust/kernel/io.rs
> index d57df2a072a0e..c9533d3f003ba 100644
> --- a/rust/kernel/io.rs
> +++ b/rust/kernel/io.rs

[ ... ]

> @@ -90,37 +90,67 @@ fn size(p: *const Self) -> usize {

[ ... ]

> +impl<const SIZE: usize> MmioRaw<Region<SIZE>> {
> +    /// Create a `MmioRaw` representing a I/O region with given size.
> +    ///
> +    /// The size is checked against the minimum size specified via const generics.
> +    #[inline]
> +    pub fn new_region(addr: usize, size: usize) -> Result<Self> {
> +        Ok(Self {
> +            ptr: Region::ptr_try_from_raw_parts_mut(core::ptr::without_provenance_mut(addr), size)?,

[Severity: High]
Does this alignment check break driver probes for standard 4-byte aligned
MMIO mappings on 64-bit architectures?

Looking at Region::ptr_try_from_raw_parts_mut(), it enforces a strict
usize alignment on the base address:

rust/kernel/io.rs:Region::ptr_try_from_raw_parts_mut() {
    ...
    if size < SIZE || base.align_offset(size_of::<usize>()) != 0 {
        return Err(EINVAL);
    }
    ...
}

On 64-bit systems, this forces an 8-byte alignment requirement.

However, PCI 32-bit BARs and SoC IP block MMIO regions frequently have
4-byte aligned physical base addresses. When mapped with ioremap() or
pci_iomap(), the resulting virtual address retains this 4-byte alignment
offset, which would fail the 8-byte alignment check here.

Consequently, standard hardware mapping operations in devres::IoMem::new(),
IoMem::ioremap(), and pci::Bar::new() will unconditionally fail at runtime
with -EINVAL during driver probe for these devices, making them unusable.

> +        })
>      }
> +}

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260608-io_projection-v3-0-c5cde13a5ec4@garyguo.net?part=5

  reply	other threads:[~2026-06-08 20:14 UTC|newest]

Thread overview: 31+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-06-08 19:58 [PATCH v3 00/19] rust: I/O type generalization and projection Gary Guo
2026-06-08 19:58 ` [PATCH v3 01/19] rust: io: add dynamically-sized `Region` type Gary Guo
2026-06-08 23:52   ` Gary Guo
2026-06-08 19:58 ` [PATCH v3 02/19] rust: io: add missing safety requirement in `IoCapable` methods Gary Guo
2026-06-08 20:13   ` sashiko-bot
2026-06-08 19:59 ` [PATCH v3 03/19] rust: io: restrict untyped IO access and `register!` to `Region` Gary Guo
2026-06-08 19:59 ` [PATCH v3 04/19] rust: io: implement `Io` on reference types instead Gary Guo
2026-06-08 19:59 ` [PATCH v3 05/19] rust: io: generalize `MmioRaw` to pointer to arbitrary type Gary Guo
2026-06-08 20:14   ` sashiko-bot [this message]
2026-06-08 19:59 ` [PATCH v3 06/19] rust: io: rename `Mmio` to `MmioOwned` Gary Guo
2026-06-08 19:59 ` [PATCH v3 07/19] rust: io: implement `Mmio` as view type Gary Guo
2026-06-08 20:15   ` sashiko-bot
2026-06-08 19:59 ` [PATCH v3 08/19] rust: pci: io: make `ConfigSpace` a view Gary Guo
2026-06-08 20:11   ` sashiko-bot
2026-06-08 19:59 ` [PATCH v3 09/19] rust: io: use view types instead of addresses for `Io` Gary Guo
2026-06-08 19:59 ` [PATCH v3 10/19] rust: io: remove `MmioOwned` Gary Guo
2026-06-08 20:12   ` sashiko-bot
2026-06-08 19:59 ` [PATCH v3 11/19] rust: io: move `Io` methods to extension trait Gary Guo
2026-06-08 19:59 ` [PATCH v3 12/19] rust: io: add projection macro and methods Gary Guo
2026-06-08 20:13   ` sashiko-bot
2026-06-08 19:59 ` [PATCH v3 13/19] rust: io: add I/O backend for system memory with volatile access Gary Guo
2026-06-08 20:09   ` sashiko-bot
2026-06-08 19:59 ` [PATCH v3 14/19] rust: io: implement a view type for `Coherent` Gary Guo
2026-06-08 20:18   ` sashiko-bot
2026-06-08 19:59 ` [PATCH v3 15/19] rust: io: add `read_val` and `write_val` function on `Io` Gary Guo
2026-06-08 19:59 ` [PATCH v3 16/19] gpu: nova-core: use I/O projection for cleaner encapsulation Gary Guo
2026-06-08 19:59 ` [PATCH v3 17/19] rust: dma: drop `dma_read!` and `dma_write!` API Gary Guo
2026-06-08 19:59 ` [PATCH v3 18/19] rust: io: add copying methods Gary Guo
2026-06-08 20:20   ` sashiko-bot
2026-06-08 19:59 ` [PATCH v3 19/19] rust: io: implement `Io` for `Either` Gary Guo
2026-06-08 21:22 ` [PATCH v3 00/19] rust: I/O type generalization and projection Danilo Krummrich

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=20260608201447.023961F00893@smtp.kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=acourbot@nvidia.com \
    --cc=dakr@kernel.org \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=gary@garyguo.net \
    --cc=linux-pci@vger.kernel.org \
    --cc=ojeda@kernel.org \
    --cc=sashiko-reviews@lists.linux.dev \
    /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