[RFC 0/3] nvme uring passthrough diet

[Keith Busch]

On Wed, May 03, 2023 at 12:57:17PM +0530, Kanchan Joshi wrote:
> On Mon, May 01, 2023 at 08:33:03AM -0700, Keith Busch wrote:
> > From: Keith Busch <kbusch at kernel.org>
> > 
> > When you disable all the optional features in your kernel config and
> > request queue, it looks like the normal request dispatching is just as
> > fast as any attempts to bypass it. So let's do that instead of
> > reinventing everything.
> > 
> > This doesn't require additional queues or user setup. It continues to
> > work with multiple threads and processes, and relies on the well tested
> > queueing mechanisms that track timeouts, handle tag exhuastion, and sync
> > with controller state needed for reset control, hotplug events, and
> > other error handling.
> 
> I agree with your point that there are some functional holes in
> the complete-bypass approach. Yet the work was needed to be done
> to figure out the gain (of approach) and see whether the effort to fill
> these holes is worth.
> 
> On your specific points
> - requiring additional queues: not a showstopper IMO.
>  If queues are lying unused with HW, we can reap more performance by
>  giving those to application. If not, we fall back to the existing path.
>  No disruption as such.

The current way we're reserving special queues is bad and should
try to not extend it futher. It applies to the whole module and
would steal resources from some devices that don't want poll queues.
If you have a mix of device types in your system, the low end ones
don't want to split their resources this way.

NVMe has no problem creating new queues on the fly. Queue allocation
doesn't have to be an initialization thing, but you would need to
reserve the QID's ahead of time.

> - tag exhaustion: that is not missing, a retry will be made. I actually
>  wanted to do single command-id management at the io_uring level itself,
>  and that would have cleaned things up. But it did not fit in
>  because of submission/completion lifetime differences.
> - timeout and other bits you mentioned: yes, those need more work.
> 
> Now with the alternate proposed in this series, I doubt whether similar
> gains are possible. Happy to be wrong if that happens.

One other thing: the pure-bypass does appear better at low queue
depths, but utilizing the plug for aggregated sq doorbell writes
is a real win at higher queue depths from this series. Batching
submissions at 4 deep is the tipping point on my test box; this
series outperforms pure bypass at any higher batch count.

Part of the problem with the low queue depth for this mode is the
xarray lookup on each bio_poll, but I think that we can fix that.

> Please note that for some non-block command sets, passthrough is the only
> usable interface. So these users would want some of the functionality
> bits too (e.g. cgroups). Cgroups is broken for the passthrough at the
> moment, and I wanted to do something about that too.
> 
> Overall, the usage model that I imagine with multiple paths is this -
> 
> 1. existing block IO path: for block-friendly command-sets
> 2. existing passthrough IO path: for non-block command sets
> 3. new pure-bypass variant: for both; and this one deliberately trims all
> the fat at the expense of some features/functionality.
> 
> #2 will not have all the features of #1, but good to have all that are
> necessary and do not have semantic troubles to fit in. And these may
> grow over time, leading to a kernel that has improved parity between block
> and non-block io.
> Do you think this makes sense?