Re: [PATCH v6 02/16] chardev: introduce cdev_get_by_path()

[Logan Gunthorpe]

On 2019-07-25 12:10 p.m., Greg Kroah-Hartman wrote:
> On Thu, Jul 25, 2019 at 11:53:20AM -0600, Logan Gunthorpe wrote:
>>
>>
>> On 2019-07-25 11:40 a.m., Greg Kroah-Hartman wrote:
>>> On Thu, Jul 25, 2019 at 11:23:21AM -0600, Logan Gunthorpe wrote:
>>>> cdev_get_by_path() attempts to retrieve a struct cdev from
>>>> a path name. It is analagous to blkdev_get_by_path().
>>>>
>>>> This will be necessary to create a nvme_ctrl_get_by_path()to
>>>> support NVMe-OF passthru.
>>>
>>> Ick, why?  Why would a cdev have a "pathname"?
>>
>> So we can go from "/dev/nvme0" (which points to a char device) to its
>> struct cdev and eventually it's struct nvme_ctrl. Doing it this way also
>> allows supporting symlinks that might be created by udev rules.
>>
>> This is very similar to blkdev_get_by_path() that lets regular NVMe-OF
>> obtain the struct block_device from a path.
>>
>> I didn't think this would be all that controversial.
>>
>>> What is "NVMe-OF passthru"?  Why does a char device node have anything
>>> to do with NVMe?
>>
>> NVME-OF passthru is support for NVME over fabrics to directly target a
>> regular NVMe controller and thus export an entire NVMe device to a
>> remote system. We need to be able to tell the kernel which controller to
>> use and IMO a path to the device file is the best way as it allows us to
>> support symlinks created by udev.
> 
> open() in userspace handles symlinks just fine, what crazy interface
> passes a string to try to find a char device node that is not open()?

configfs. Which I'm stuck with seeing nvme-of already uses that for
configuration and I don't think that's going to change...

> And why do you need a char device at all anyway?  Is this just the
> "normal" nvme controller's character device node?

Yes.

Logan