Re: [PATCH v1 1/1] lib/crypto: tests: Add KUnit tests for AES

[Eric Biggers <ebiggers@kernel.org>]

On Thu, Jan 15, 2026 at 10:51:25PM +0100, Holger Dengler wrote:
> On 15/01/2026 21:43, Eric Biggers wrote:
> > On Thu, Jan 15, 2026 at 07:38:31PM +0100, Holger Dengler wrote:
> >> Add a KUnit test suite for AES library functions, including KAT and
> >> benchmarks.
> >>
> >> Signed-off-by: Holger Dengler <dengler@linux.ibm.com>
> > 
> > The cover letter had some more information.  Could you put it in the
> > commit message directly?  Normally cover letters aren't used for a
> > single patch: the explanation should just be in the patch itself.
> 
> Ok, I'll move the explanation to the commit message. I assume that the example
> output of the test can be dropped?

Yes, that's fine.

> > 10000000 iterations is too many.  That's 160 MB of data in each
> > direction per AES key length.  Some CPUs without AES instructions can do
> > only ~20 MB AES per second.  In that case, this benchmark would take 16
> > seconds to run per AES key length, for 48 seconds total.
> > 
> > hash-test-template.h and crc_kunit.c use 10000000 / (len + 128)
> > iterations.  That would be 69444 in this case (considering len=16),
> > which is less than 1% of the iterations you've used.  Choosing a number
> > similar to that would seem more appropriate.
> > 
> > Ultimately these are just made-up numbers.  But I think we should aim
> > for the benchmark test in each KUnit test suite to take less than a
> > second or so.  The existing tests roughly achieve that, whereas it seems
> > this one can go over it by quite a bit due to the 10000000 iterations.
> 
> As we have a fixed length, I would go stay with a fix value for the iterations
> (instead of calculating it based on len).
> 
> The benchmark has a separate loop for encrypt and decrypt, so I will do the
> half iterations on encrypt and the other half on decrypt. I will also reduce
> the iterations for the warm-ups.
> 
> What about 100 iterations for each warm-up and 500.000 iterations for each
> real measurement? Means processing 2x 8MiB with preemption disabled.

I'd suggest 50000 for each direction as well as the warm-up loop.

- Eric