[RESEND PATCH 2/2] NVMe: Add nvme subsystem reset IOCTL

[Keith Busch]

On Tue, 11 Aug 2015, J Freyensee wrote:
> On Mon, 2015-08-10 at 15:20 -0600, Keith Busch wrote:
>> From: Jon Derrick <jonathan.derrick at intel.com>
>>
>> Controllers can perform optional subsystem resets as introduced in NVMe
>> 1.1. This patch adds an IOCTL to trigger the subsystem reset by writing
>> "NVMe" to the NSSR register.
>
> I just took a Linux Foundation training class and one of the points the
> Foundation made was the Linux community wanting to move off of IOCTLs
> to other userspace-to-kernelspace mechanisms (like sysfs for example).
>
> I was wondering if it's a good thing to be continuing to add IOCTLs if
> it is true that the Linux community would like to move off the reliance
> of IOCTLs?

Maybe some developers have stronger opinions on adding ioctls. I don't
see a big problem with it. The proposed ioctl doesn't transfer data with
userspace, so we're not committing to maintain a new ABI.

We have sysfs and IOCTL for NVMe controller reset, so maybe we should
have both for subsystem resets too. I like sysfs for shell scripting,
but ioctl seems easier/cleaner to use within applications. I can add a
sysfs entry if this series gets applied.