On Wed, Sep 05, 2018 at 03:32:03PM +0200, Hannes Reinecke wrote:
> On 09/05/2018 03:18 PM, Christoph Hellwig wrote:
> > On Wed, Sep 05, 2018 at 09:00:50AM +0200, Hannes Reinecke wrote:
> >> We should be registering the ns_id attribute as default sysfs
> >> attribute groups, otherwise we have a race condition between
> >> the uevent and the attributes appearing in sysfs.
> >
> > Please give Bart credit for his work, as the lightnvm bits are almost
> > bigger than the rest.
> >
> Okay, will be doing so.
>
> >> +static umode_t nvm_dev_attrs_visible(struct kobject *kobj,
> >> + struct attribute *attr, int index)
> >> {
> >> + struct device *dev = container_of(kobj, struct device, kobj);
> >> + struct gendisk *disk = dev_to_disk(dev);
> >> + struct nvme_ns *ns = disk->private_data;
> >> struct nvm_dev *ndev = ns->ndev;
> >> + struct device_attribute *dev_attr =
> >> + container_of(attr, typeof(*dev_attr), attr);
> >>
> >> + if (dev_attr->show == nvm_dev_attr_show)
> >> + return attr->mode;
> >>
> >> + switch (ndev ? ndev->geo.major_ver_id : 0) {
> >
> > How could ndev be zero here?
> >
> For 'normal' NVMe devices (ie non-lightnvm). As we now register all
> sysfs attributes (including the lightnvm ones) per default we'll need to
> blank them out for non-lightnvm devices.
But then we need to exit early at the beginning of the function,
as we should not register attributes using nvm_dev_attr_show (or
anything else for that matter) either.
