Hi Alex, On Thu, 14 Jan 2021 at 09:09, Alex G. <mr.nuke...@gmail.com> wrote: > > On 1/13/21 10:10 AM, Simon Glass wrote: > > Hi Alexandru, > > > > On Mon, 11 Jan 2021 at 08:41, Alexandru Gagniuc <mr.nuke...@gmail.com> > > wrote: > >> > >> Define a UCLASS API for verifying ECDSA signatures. Unlike > >> UCLASS_MOD_EXP, which focuses strictly on modular exponentiation, > >> the ECDSA class focuses on verification. This is done so that it > >> better aligns with mach-specific implementations, such as stm32mp. > >> > >> Signed-off-by: Alexandru Gagniuc <mr.nuke...@gmail.com> > >> --- > >> include/crypto/ecdsa-uclass.h | 39 +++++++++++++++++++++++++++++++++++ > >> include/dm/uclass-id.h | 1 + > >> 2 files changed, 40 insertions(+) > >> create mode 100644 include/crypto/ecdsa-uclass.h > > > > This needs a test, as do all uclasses in U-Boot. If it isn't easy to > > implement the algorithm then I suppose you could fake it by using an > > easy algorithm like md5, but it does need a test. > > I agree. I couldn't find a test for UCLASS_MOD_EXP (for guidance), so > I'm not sure where to even start.
How about dm_test_rng_read()? That is pretty simple and does a few basic things. Re UCLASS_MOD_EXP this is tested by the vboot tests. It should really have a simple unit test too, though. This is an example of how to add a software implementation of a hardware algorithm. See mod_exp_sw.c Regards, Simon
