Re: [PATCH] rust: dma: update safety comments for volatile memory access

[Andreas Hindborg]

"Gary Guo" <gary@xxxxxxxxxxx> writes:

> On Fri Jan 30, 2026 at 2:59 PM GMT, Andreas Hindborg wrote:
>> At the time `CoherentAllocation::read_field` and
>> `CoherentAllocation::write_field` was merged, `ptr::{read,write}_volatile`
>> was under specified. The documentation for these functions have been
>> updated and we can now formulate a proper safety comment for the calls.
>>
>> Update safety comments in `CoherentAllocation::{read,write}_field`.
>>
>> Link: https://doc.rust-lang.org/stable/core/ptr/fn.read_volatile.html
>> Link: https://doc.rust-lang.org/stable/core/ptr/fn.write_volatile.html
>> Signed-off-by: Andreas Hindborg <a.hindborg@xxxxxxxxxx>
>> ---
>>  rust/kernel/dma.rs | 25 +++++++++----------------
>>  1 file changed, 9 insertions(+), 16 deletions(-)
>>
>> diff --git a/rust/kernel/dma.rs b/rust/kernel/dma.rs
>> index acc65b1e0f245..0b55671a94faf 100644
>> --- a/rust/kernel/dma.rs
>> +++ b/rust/kernel/dma.rs
>> @@ -593,14 +593,12 @@ pub fn item_from_index(&self, offset: usize) -> Result<*mut T> {
>>      pub unsafe fn field_read<F: FromBytes>(&self, field: *const F) -> F {
>>          // SAFETY:
>>          // - By the safety requirements field is valid.
>> -        // - Using read_volatile() here is not sound as per the usual rules, the usage here is
>> -        // a special exception with the following notes in place. When dealing with a potential
>> -        // race from a hardware or code outside kernel (e.g. user-space program), we need that
>> -        // read on a valid memory is not UB. Currently read_volatile() is used for this, and the
>> -        // rationale behind is that it should generate the same code as READ_ONCE() which the
>> -        // kernel already relies on to avoid UB on data races. Note that the usage of
>> -        // read_volatile() is limited to this particular case, it cannot be used to prevent
>> -        // the UB caused by racing between two kernel functions nor do they provide atomicity.
>> +        // - `field` points to memory outside any Rust allocation.
>
> Hmm, this isn't actually correct, as the memory behind `CoherentAllocation`
> isn't MMIO (they're just also accessible by device). We provide `as_slice()`
> which exposes these memory directly to the Rust memory model.

I would not get too hung on the MMIO term. But if we provide `as_slice`,
this implementation of `read_field` and `write_field` is not sound at
all. We should fix that.


Best regards,
Andreas Hindborg