Potential missed guarantee for byte-order in byte slice casting

[shinmao]

rv6/kernel-rs/src/fs/lfs/imap.rs

Lines 32 to 38 in f55cf81

	impl<'s> From<&'s mut BufData> for &'s mut DImapBlock {
	fn from(b: &'s mut BufData) -> Self {
	const_assert!(mem::size_of::<DImapBlock>() <= mem::size_of::<BufData>());
	const_assert!(mem::align_of::<BufData>() % mem::align_of::<DImapBlock>() == 0);
	unsafe { &mut *(b.as_mut_ptr() as *mut DImapBlock) }
	}
	}

In From implementation above, we can cast from [u8, BSIZE] to [u32; NENTRY]. In addition to checking size and alignment, the results of casting around bytes between data types are endian dependent, and I couldn't find any specification about byte-order here. What's more, the u32 byte slice is mutable, which means we can change the u32 byte slice value and affect the u8 byte slice by different order again.

Maybe using byteorder could help solving this kind of issues.

Please correct me if I mistake for something;)

reference: https://docs.rs/bytemuck/latest/bytemuck/trait.Pod.html