Re: [PATCH v6 0/3] Rust abstractions for the SPI driver subsystem

[Gary Guo]

On Tue Oct 27, 2026 at 5:42 AM GMT, Shuai Liu wrote:
> Changes since v5:
>   - validate in `write_then_read()` that the sum of the two buffer
>     lengths fits in `u32`: the C core's spi_write_then_read() sums the
>     two lengths as `unsigned int` when selecting and sizing its bounce
>     buffer, so two ~2GiB slices would wrap the sum to zero and select
>     the preallocated `SPI_BUFSIZ` buffer, which then overflows on the
>     copy. Found by Karl Mehltretter.
>
> The Rust abstractions for several buses are already present in the
> kernel (PCI, platform, I2C, USB, serdev, ...), but the SPI driver
> subsystem is still missing such abstractions. This series adds them,
> along with a sample driver demonstrating their usage.
>
> To my knowledge, nobody is currently working on Rust SPI abstractions;
> if someone already has work in progress, please let me know and I will
> happily drop or rebase this series.
>
> The abstractions follow the same layout as the existing I2C and serdev
> abstractions: a `Driver` trait for drivers to implement, an `Adapter`
> for driver registration, and a `SpiDevice` wrapping `struct spi_device`.
> This first installment provides synchronous `read`, `write` and
> `write_then_read` operations; message-based transfers (`spi_message` /
> `spi_transfer`) can be added on top in a follow-up series.
>
> One deviation from the I2C abstractions: `spi_match_id()` is not
> exported to modules, so device id matching by modalias uses the
> exported `spi_get_device_id()` instead, falling back to the OF and ACPI
> tables.
>
> The series is based on the current mainline tree (7.4 merge window) and
> has been build-tested with CONFIG_RUST=y, including the KUnit tests and
> doctests for the new abstractions. I do not have access to real SPI
> hardware to run-test this series, but the included sample driver can be
> used for that purpose.
>
> Signed-off-by: Shuai Liu <847807867@xxxxxx>

Your patches are not replies to the cover letter, as your email provider
rewrites the Message-ID. This break lore (and email clients).

You can try "--outlook-id-fix" (but it's not guaranteed to work if your email
provider doesn't report the reassigned Message-ID in the same way as outlook).
Alternatively, submit your patches via B4.

Best,
Gary

>
> Shuai Liu (3):
>   rust: helpers: add SPI helpers
>   rust: spi: add SPI driver abstractions
>   samples: rust: add SPI driver sample
>
>  rust/bindings/bindings_helper.h |   1 +
>  rust/helpers/helpers.c          |   1 +
>  rust/helpers/spi.c              |  25 ++
>  rust/kernel/Kconfig.test        |  11 +
>  rust/kernel/lib.rs              |   2 +
>  rust/kernel/spi.rs              | 570 ++++++++++++++++++++++++++++++++
>  samples/rust/Kconfig            |  11 +
>  samples/rust/Makefile           |   1 +
>  samples/rust/rust_driver_spi.rs |  66 ++++
>  9 files changed, 688 insertions(+)
>  create mode 100644 rust/helpers/spi.c
>  create mode 100644 rust/kernel/spi.rs
>  create mode 100644 samples/rust/rust_driver_spi.rs