[PATCH 2/2] nvme: Don't use a stack buffer for keep-alive command

[Sagi Grimberg]

> We needn't worry about the DNR case since driver allocated commands are
> flagged "REQ_FAILFAST_DRIVER" and are never retried.

Yea.

> Allocating the full 64-byte NVMe command for each request seems a bit
> excessive.

Why?

> Since it only really applies to async driver commands, and
> there are so few of those as it is, I'm okay with having special cases
> for these as Roland suggests. Sound okay?

I'm fine with fixing this locally, btw isn't this exact same stale sqe
happen in pci's nvme_delete_queue?