Re: [PATCH 3/6] gpu: nova-core: add NVKV decoder

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] NVKV decoder uses host-endian memory pointer casts and byte casting, breaking big-endian compatibility.
--

--- Patch [3]: [PATCH 3/6] gpu: nova-core: add NVKV decoder ---
commit 462ea319dececf0891b941e125a3a7cd586a70f8
Author: Eliot Courtney <[email protected]>

gpu: nova-core: add NVKV decoder

This commit adds a decoder for NVKV to process GMCAPI call messages from
the GSP. It introduces a Schema visitor trait to process the (key, index,
value) sequences from the encoded stream.

> diff --git a/drivers/gpu/nova-core/gsp/nvkv/decode.rs b/drivers/gpu/nova-core/gsp/nvkv/decode.rs
> new file mode 100644
> index 0000000000000..ee8b6ab5a3a43
> --- /dev/null
> +++ b/drivers/gpu/nova-core/gsp/nvkv/decode.rs
[ ... ]
> +    fn take_u8s(&mut self, count: usize) -> Result<&[u8]> {
> +        let values = self.take_u64s(count.div_ceil(8))?;
> +        values.as_bytes().get(..count).ok_or(EINVAL)

[Severity: High]
Does casting a native [u64] to [u8] using as_bytes() expose the native
byte order of the host? If the GSP sends data in little-endian format, won't
this yield reversed 8-byte chunks on big-endian architectures?

> +    }
> +
> +    fn take_u32s(&mut self, count: usize) -> Result<&[u32]> {
> +        let values = self.take_u64s(count.div_ceil(2))?;
> +        // SAFETY: `values` is 8 byte aligned and only 4 byte alignment is required. All bit
> +        // patterns are valid for `u32`.
> +        Ok(unsafe { core::slice::from_raw_parts(values.as_ptr().cast::<u32>(), count) })

[Severity: High]
Could this raw pointer cast from u64 to u32 cause issues on big-endian
systems? This appears to change the order of the 32-bit words compared to
the original little-endian GSP message layout.

> +    }
> +
> +    fn take_u64s(&mut self, count: usize) -> Result<&[u64]> {

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=3