Re: [PATCH v3] rust: transmute: Add implementation for FromBytes trait

[Christian]

Hi, Daniel! I didn't know. Thank you in advance, I'll improve in v4.

Best regards,
Christian

On Sat, Nov 9, 2024 at 8:33 PM Daniel Almeida
<daniel.almeida@xxxxxxxxxxxxx> wrote:
>
> +cc Abdiel
>
> Hi Christian,
>
>
> > On 9 Nov 2024, at 02:54, Christian dos Santos de Lima <christiansantoslima21@xxxxxxxxx> wrote:
> >
> > Add implementation and documentation for FromBytes trait.
> >
> > Add new feature block in order to allow using ToBytes
> > and bound to from_bytes_mut function. I'm adding this feature
> > because is possible create a value with disallowed bit pattern
> > and as_byte_mut could create such value by mutating the array and
> > accessing the original value. So adding ToBytes this can be avoided.
>
> I’ve read this a couple of times. It’s hard to understand what you’re trying to say.
>
> There are some grammar errors, but the even the key idea is hard to understand. IOW,
> I had to go on Github to get the extra context needed to understand this part below:
>
> > possible create a value with disallowed bit pattern
> > and as_byte_mut could create such value by mutating the array and
> > accessing the original value. So adding ToBytes this can be avoided.
>
>
> Also, core::simd::ToBytes? Maybe Bjorn meant a new kernel::transmute::ToBytes trait?
>
> Christian, are you aware of Abdiel’s dma_alloc_coherent patch [0]? How is this patch
> an improvement over the simpler code below, for example?
>
> > + fn cpu_buf(&self) -> &[T]
> > + {
> > + // SAFETY: The pointer is valid due to type invariant on `CoherentAllocation` and
> > + // is valid for reads for `self.count * size_of::<T>` bytes.
> > + unsafe { core::slice::from_raw_parts(self.cpu_addr, self.count) }
> > + }
> > +
> > + fn cpu_buf_mut(&mut self) -> &mut [T]
> > + {
> > + // SAFETY: The pointer is valid due to type invariant on `CoherentAllocation` and
> > + // is valid for reads for `self.count * size_of::<T>` bytes.
> > + unsafe { core::slice::from_raw_parts_mut(self.cpu_addr, self.count) }
> > + }
>
> I believe that most users would have a pointer to memory allocated on the C side, as you can
> see from Abdiel’s excerpt above. Without a user, it’s hard to know why we need this, specially
> given the simd::ToBytes bound.
>
> Just my 2c.
>
> — Daniel
>
>
> [0] https://lore.kernel.org/rust-for-linux/20241105104726.3111058-3-abdiel.janulgue@xxxxxxxxx/