[PATCH V3 00/10] nvme: use xarray for ns tracking

[Chaitanya Kulkarni]

Christoph,

On 7/15/20 00:03, Christoph Hellwig wrote:
>> Following are the performance numbers with NVMeOF (nvme-loop) backed by
>> null_blk devices mapped 1:1 on NVMeOF target backend :-
> Can you make sure the nvmet patch is the first in the series?  That one
> is obviously useful, so as soon as everything is sorted out I'm happy
> to apply it.  I'm not quite as sold on the host side.
> 

Sure.

Can you please elaborate on what is wrong with host side patch ? any
potential bug/race/deadlock/performance regression ?

Cover-letter has basic explanation for host patch. See [1].

In case that is not sufficient at least I'd know what kind of more data
I need to collect to prove the usefulness of the patch.

I've also offered to forward port the passthru and get the performance
numbers with nvme_find_get_ns() with XArray version, but didn't get any
response on that.


[1] Host side patch cover letter reference :-

"For host even though nvme_find_get_ns() doesn't fall into the fast path
yet, it does for NVMeOF passthru. This prepares us to improve 
performance for future NVMeOF passthru backend which is under review 
that uses the similar data structure in host-core as target-core."