> > What we could do is have the user specify an explicit flag saying > that they do not care about the output IV. You could then skip the > output IV step in your driver. > That would work for me, if the maintainers would be OK with adding such flags. Also, as a heads up - just to get other peoples opinion here - I might prefer responding to such a flag by selecting the fallback cipher instead of actually implementing a workaround for my hardware in case such a workaround would be detrimental to the performance.
> But you'll have to code this up, including the bit on the user-side > to actually set the flag. > Since that user side probably lives in the kernel tree too, I could do that. This would then apply to testmgr as well though, as I need to ensure it disables features for tests that don't require them. (otherwise you'd be verifying only the fallback cipher) Regards, Pascal
