On Thu, Dec 10, 2020 at 03:07:32AM +0000, Chaitanya Kulkarni wrote:
> I think something like following (totally untested) will help to avoid the
> scenarios like this for ZNS drives so we can rejects the buggy controllers
> early to make sure we are spec compliant :-
>
> # git diff# git diff
> diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
> index d9b152bae19d..7b196299c9b7 100644
> --- a/drivers/nvme/host/core.c
> +++ b/drivers/nvme/host/core.c
> @@ -2166,6 +2166,11 @@ static int nvme_update_ns_info(struct nvme_ns
> *ns, struct nvme_id_ns *id)
> nvme_set_queue_limits(ns->ctrl, ns->queue);
>
> if (ns->head->ids.csi == NVME_CSI_ZNS) {
> + if (!(NVME_CAP_CSS(ns->ctrl->cap) & NVME_CAP_CSS_CSI)) {
> + pr_err("zns ns found with ctrl support for CSI");
> + goto out_unfreeze;
> + }
I think you want to use nvme_multi_css() for the 'if()' check, but
otherwise this looks like a valid sanity check.
