On 08/14/2018 05:20 PM, Bart Van Assche wrote:
> On Tue, 2018-08-14 at 09:33 +0200, Hannes Reinecke wrote:
>> +const struct attribute_group *nvme_ns_id_attr_groups[] = {
>> + &nvme_ns_id_attr_group,
>> + NULL, /* Will be filled in by lightnvm if present */
>> + NULL,
>> +};
> [ ... ]
>> -void nvme_nvm_unregister_sysfs(struct nvme_ns *ns)
>> -{
>> - struct nvm_dev *ndev = ns->ndev;
>> - struct nvm_geo *geo = &ndev->geo;
>> + return;
>>
>> switch (geo->major_ver_id) {
>> case 1:
>> - sysfs_remove_group(&disk_to_dev(ns->disk)->kobj,
>> - &nvm_dev_attr_group_12);
>> + nvme_ns_id_attr_groups[1] = &nvm_dev_attr_group_12;
>> break;
>> case 2:
>> - sysfs_remove_group(&disk_to_dev(ns->disk)->kobj,
>> - &nvm_dev_attr_group_20);
>> + nvme_ns_id_attr_groups[1] = &nvm_dev_attr_group_20;
>
> This patch introduces a really ugly dependency between the NVMe core code and
> the lightnvm code, namely that the lightnvm code has to know at which position
> in the nvme_ns_id_attr_groups it can fill in its attribute group pointer. Have
> you considered to make nvme_nvm_register_sysfs() return an attribute group
> pointer such that the nvme_ns_id_attr_groups can be changed from a global into
> a static array?
>
Wouldn't help much, as the 'nvme_ns_id_attr_groups' needs to be a global
pointer anyway as it's references from drivers/nvme/host/core.c and
drivers/nvme/host/multipath.c.
While I have considered having nvme_nvm_register_sysfs() returning a
pointer I would then have to remove the 'static' declaration from the
nvm_dev_attr_group_12/20.
Which I didn't really like, either.
Cheers,
Hannes
--
Dr. Hannes Reinecke Teamlead Storage & Networking
hare@xxxxxxx +49 911 74053 688
SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: F. Imendörffer, J. Smithard, J. Guild, D. Upmanyu, G. Norton
HRB 21284 (AG Nürnberg)
