On Sun, Dec 31, 2017 at 3:05 PM, Yishai Hadas <yishaih@xxxxxxxxxxxxxxxxxx> wrote: > On 12/31/2017 2:37 PM, Or Gerlitz wrote: >>> It doesn't make sense to overwrite the input QP type (i.e. IB_QPT_DRIVER) >>> which was supplied by the core layer to some internal sub-type. >> why? > A driver code doesn't expect to change an input IB core property (i.e > qp_type). Specifically the DC QP type does not expect to be recognized out > of the mlx5 driver, for that reason a sub qp type will be used as part of V2 > in the mlx5 driver code. sub qp type sound better from qp type stored in two places... I think we should still strongly aim to have the qp type in one place... I guess it would be correct for this to be a driver maintainer decision. >> So what you are saying is the core logic being good for all >> QP types X all drivers X all transports and only mlx5/DC needs this >> different treatment? interesting, so what is the source for this misalignment? > We believe that it doesn't justify a core refactoring for that specific mask > checking. ok, just remember that C&Ps have the tendency to add bugs, so look carefully on the code.. -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
