Re: [PATCHv4] blk-integrity: support arbitrary buffer alignment

[Eric Biggers <ebiggers@kernel.org>]

On Thu, Nov 13, 2025 at 03:11:56PM -0500, Keith Busch wrote:
> On Thu, Nov 13, 2025 at 08:02:37PM +0000, Eric Biggers wrote:
> > On Thu, Nov 13, 2025 at 02:48:43PM -0500, Keith Busch wrote:
> > > Like on real hardware? I'm a bit at a loss as to how, I've never seen
> > > anything subscribe to this format, not even in emulation. The only thing
> > > I can readily do to test this is run random data through the old code,
> > > print the result, then run the same data through the new code and see if
> > > they're the same. That test is successful. Not good enough?
> > 
> > ip_compute_csum() returns a folded 16-bit checksum, whereas
> > csum_partial() returns an unfolded 32-bit checksum.
> 
> Sorry, I must be missing something. do_csum() returns an unfolded 32-bit
> result

Nope.  It returns a folded 16-bit result.

> In any case, I find that running any random data through it at block
> interval lengths (4k or 512b) is always producing results that fit in 16
> bits anyway, so maybe that's why it appears to be working?

Your kernel build may be using the generic csum_partial(), which folds
the checksum more frequently than the x86 one (and maybe some of the
other arch-optimized implementations too).  Try the x86 one.

- Eric