15/04/2021 17:45, Ferruh Yigit: > Not exactly related to this patch, but related to the API input, some input > that > is filled by API is memset first before passing it to the PMD, which makes > sense. And for this we have two different ordering with dev_ops check: > > 1. > memset(input) > check dev_ops, return error if not supported > call dev_ops > > 2. > check dev_ops, return error if not supported > memset(input) > call dev_ops > > Connor, Thomas, Andrew, > Do you think does it have any benifit to unify it, and is one better than > other?
Yes good catch. It deserves a separate patch to not memset if not needed. We must check if supported before memset.
