race between nvme device creation and discovery?

[Daniel Wagner]

On Mon, Feb 05, 2024 at 04:57:24PM +0900, Hannes Reinecke wrote:
> I thought something along these lines:
> 
> diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
> index 85ab0fcf9e88..3ed706f595fe 100644
> --- a/drivers/nvme/host/core.c
> +++ b/drivers/nvme/host/core.c
> @@ -2845,6 +2845,7 @@ static int nvme_init_subsystem(struct nvme_ctrl *ctrl,
> struct nvme_id_ctrl *id)
>                         goto out_put_subsystem;
>                 }
>         } else {
> +               dev_set_uevent_suppress(&subsys->dev, 1);
>                 ret = device_add(&subsys->dev);
>                 if (ret) {
>                         dev_err(ctrl->device,
> @@ -2869,6 +2870,10 @@ static int nvme_init_subsystem(struct nvme_ctrl
> *ctrl, struct nvme_id_ctrl *id)
>         ctrl->subsys = subsys;
>         list_add_tail(&ctrl->subsys_entry, &subsys->ctrls);
>         mutex_unlock(&nvme_subsystems_lock);
> +       if (!found) {
> +               dev_set_uevent_suppress(&subsys->dev, 0);
> +               kobject_uevent(&subsys->dev.kobj, KOBJ_ADD);
> +       }
>         return 0;

Unfortunatly, this fixes a different race. Not the one I reported here:

nvme/030 (ensure the discovery generation counter is updated appropriately) [failed]
    runtime  0.602s  ...  0.542s
    --- tests/nvme/030.out      2023-08-30 10:39:08.428409596 +0200
    +++ /home/wagi/work/blktests/results/nodev/nvme/030.out.bad 2024-02-05 09:44:28.768208105 +0100
    @@ -1,2 +1,3 @@
     Running nvme/030
    +failed to lookup subsystem for controller nvme0
     Test complete