Re: [PATCH v11 2/3] rust: add dma coherent allocator abstraction.

[Alice Ryhl]

On Mon, Jan 27, 2025 at 12:00 PM Daniel Almeida
<daniel.almeida@xxxxxxxxxxxxx> wrote:
>
>
> >
> > You were asked to rename this function because it returns a slice, but
> > I wonder if it's better to take an `&mut [T]` argument and to have
> > this function copy data into that argument. That way, we could make
> > the function itself safe. Perhaps the actual copy needs to be
> > volatile?
> >
> > Well ... I understand that we did this previously and that we want to
> > avoid it because it causes too much reading if T is a struct and we
> > just want to read one of its fields. How about an API like this?
> >
> > dma_read!(my_alloc[7].foo)
> >
> > which expands to something that reads the value of the foo field of
> > the 7th element, and
> >
> > dma_write!(my_alloc[7].foo = 13);
> >
> > That expands to something that writes 13 to field foo of the 7th element.
> >
> > Thoughts? I'm proposing this to avoid going in circles between the
> > same solutions.
> >
> > Alice
> >
>
>
> I think I missed something here. How is this any different from the *safe*
> functions we currently have?
>
> Anything that involves a copy is basically what we have already, although
> your version seems to reduce the amount of data copied from sizeof(T) to
> sizeof(T.field)?

The difference is exactly that - you reduce the amount of data copied
to just the size of the field.

Alice