Date:   Fri, 14 Jan 2022 04:52:24 +0000
From:   Parav Pandit <parav@...dia.com>
To:     Jakub Kicinski <kuba@...nel.org>
CC:     Sunil Sudhakar Rani <sunrani@...dia.com>,
        Saeed Mahameed <saeedm@...dia.com>,
        Jiri Pirko <jiri@...dia.com>,
        "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
        "davem@...emloft.net" <davem@...emloft.net>,
        Bodong Wang <bodong@...dia.com>
Subject: RE: [PATCH net-next 1/2] devlink: Add support to set port function as
 trusted



> From: Jakub Kicinski <kuba@...nel.org>
> Sent: Friday, January 14, 2022 10:12 AM
> 
> On Thu, 13 Jan 2022 03:37:47 +0000 Parav Pandit wrote:
> > > > The fairness among VFs is present via the QoS knobs. Hence it
> > > > doesn't hogg
> > > the entire crypto path.
> 
> Could you please fix your email client? It's incorrectly wrapping the quotes and
> at the same time not wrapping your replies at all. :( What client is this?
>
I will fix the client.
 
> > > Why do you want to disable it, then?
> > Each enabled feature consumes
> > (a) driver level memory resource such as querying ip sec capabilities
> > and more later,
> > (b) time in querying those capabilities,
> 
> These are on the VM's side, it's not hypervisors responsibility to help the client
> by stripping features.
> 
HV is composing the device before giving it to the VM.
VM can always disable certain feature if it doesn't want to use by ethtool or other means.
But here we are discussing about offering/not offering the feature to the VF from HV.
HV can choose to not offer certain features based on some instruction received from orchestration.

> > (c) device level initialization in supporting this capability
> >
> > So for light weight devices which doesn't need it we want to keep it disabled.
> 
> You need to explain this better. We are pretty far from "trust"
> settings, which are about privilege and not breaking isolation.
>
We split the abstract trust to more granular settings, some related to privilege and some to capabilities.
 
> "device level initialization" tells me nothing.
>
Above one belongs to capabilities bucket. Sw_steering belongs to trust bucket.
 
> > > > It is the internal mlx5 implementation of how to do steering,
> > > > triggered by
> > > netdev ndo's and other devices callback.
> > > > There are multiple options on how steering is done.
> > > > Such as sw_steering or dev managed steering.
> > > > There is already a control knob to choose sw vs dev steering as
> > > > devlink
> > > param on the PF at [1].
> > > > This [1] device specific param is only limited to PF. For VFs, HV
> > > > need to
> > > enable/disable this capability on selected VF.
> > > > API wise nothing drastic is getting added here, it's only on different
> object.
> > > (instead of device, it is port function).
> > > >
> > > > [1]
> > > > https://www.kernel.org/doc/html/v5.8/networking/device_drivers/mel
> > > > lano
> > > > x/mlx5.html#devlink-parameters
> > >
> > > Ah, that thing. IIRC this was added for TC offloads, VFs don't own
> > > the eswitch so what rules are they inserting to require "high
> > > insertion rate"? My suspicion is that since it's not TC it'd be
> > > mostly for the "DR" feature you have hence my comment on it not being
> netdev.
> > No it is limited to tc offloads.
> > A VF netdev inserts flow steering rss rules on nic rx table.
> > This also uses the same smfs/dmfs when a VF is capable to do so.
> 
> Given the above are you concerned about privilege or also just resources use
> here? Do VFs have SMFS today?
Privilege.
VFs have SMFS today, but by default it is disabled. The proposed knob will enable it.

Powered by blists - more mailing lists