On Thu, Nov 12, 2020 at 02:59:51PM -0400, Jason Gunthorpe wrote: > On Wed, Nov 04, 2020 at 04:40:03PM +0200, Leon Romanovsky wrote: > > > Leon Romanovsky (5): > > RDMA/core: Allow drivers to disable restrack DB > > This stuff is never used It is in use in mlx4/mlx5 QPs. https://lore.kernel.org/linux-rdma/20201104144008.3808124-2-leon@xxxxxxxxxx/ https://lore.kernel.org/linux-rdma/20201104144008.3808124-2-leon@xxxxxxxxxx/#Z30drivers:infiniband:hw:mlx4:qp.c https://lore.kernel.org/linux-rdma/20201104144008.3808124-2-leon@xxxxxxxxxx/#Z30drivers:infiniband:hw:mlx5:qp.c > > > RDMA/cma: Be strict with attaching to CMA device > > This adds a return 0 which is pointless.. It will be used after I will resubmit patch "RDMA/restrack: Add error handling while adding restrack object" and I'm working on it right now. Can you please add this patch anyway and save me need to do constant rebases? Thanks > > > RDMA/counter: Combine allocation and bind logic > > RDMA/restrack: Store all special QPs in restrack DB > > RDMA/cma: Add missing error handling of listen_id > > Took these three to for-next > > Thanks, > Jason
