On Sun, Nov 12, 2017 at 11:41:42PM +0200, Yishai Hadas wrote:
> @@ -319,6 +319,9 @@ struct mlx5_buf {
> struct mlx5_pd {
> struct ibv_pd ibv_pd;
> uint32_t pdn;
> + int is_parent_domain;
> + struct ibv_td *td;
> + struct ibv_pd *protection_domain;
is_parent_domain can just be protection_domain != NULL
> +struct ibv_pd *mlx5_alloc_parent_domain(struct ibv_context *context,
> + struct ibv_parent_domain_init_attr *attr)
> +{
> + struct mlx5_pd *pd;
> +
> + pd = calloc(1, sizeof *pd);
> + if (!pd)
> + return NULL;
> +
> + pd->is_parent_domain = 1;
> + pd->td = attr->td;
> + pd->protection_domain= attr->pd;
Don't we need some kind of ref counting here?
What is the intention for the final version of this patch?
Are you going to do
pd->ibv_pd = *attr->pd;
During create?
Or change every call site very roughly like like..
inline struct mlx5_pd *resolve_pd(struct ibv_pd *pd)
{
if (pd->protection_domain)
return pd->protection_domain;
return pd;
}
mlx5_foo(struct ibv_pd *arg_pd)
{
struct mlx5_pd *pd = resolve_pd(arg_pd);
Jason
--
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
