mi_xfer: Added nvme_mi_mi_xfer API

[chorkin]

Exposed an nvme_mi_mi_xfer for MI messages. This is similar to the Admin passthrough, nvme_mi_admin_xfer.

Added new API and test cases in test-mi
Signed-off-by: Chuck Horkin chorkin@microsoft.com

[igaw]

@jk-ozlabs could you have a look? Thanks!

[jk-ozlabs]

I'm fine with the concept here, but could you elaborate on what the intended use-case is here? If it's just for coverage of specified MI commands, then we may want to add those (rather than leaking the implementations out the caller). Or is this for vendor-defined MI interactions?

It may be helpful to include a bit of a rationale/use-case in the commit message, if that's applicable - your call on that.

I have a couple of comments inline. You'll also need to update the .map file, otherwise the new function will not be visible.

[chorkin]

I'm fine with the concept here, but could you elaborate on what the intended use-case is here? If it's just for coverage of specified MI commands, then we may want to add those (rather than leaking the implementations out the caller). Or is this for vendor-defined MI interactions?

It may be helpful to include a bit of a rationale/use-case in the commit message, if that's applicable - your call on that.

I have a couple of comments inline. You'll also need to update the .map file, otherwise the new function will not be visible.

Thanks for the review. Regarding the use-case, we have a service we're writing and it's meant to be mostly pass through to allow multiple processes to talk to the device we are working on. The device will also have some custom vendor-defined MI interactions. For both of those reasons I wanted to expose this pass through API.

Noted on the .map. I will fix this.

[jk-ozlabs]

We're at the stage of me doing minor tweaks in comments, so all looks pretty good to me with the .map fix that @igaw has mentioned.

[igaw]

Thanks!