netdev - Re: [net-next 01/16] net/mlx5: E-Switch, let user to enable disable metadata

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20210413155712.48d97d95@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
Date:   Tue, 13 Apr 2021 15:57:12 -0700
From:   Jakub Kicinski <kuba@...nel.org>
To:     Saeed Mahameed <saeed@...nel.org>
Cc:     "David S. Miller" <davem@...emloft.net>, netdev@...r.kernel.org,
        Tariq Toukan <tariqt@...dia.com>,
        Parav Pandit <parav@...dia.com>, Roi Dayan <roid@...dia.com>,
        Vu Pham <vuhuong@...dia.com>
Subject: Re: [net-next 01/16] net/mlx5: E-Switch, let user to enable disable
 metadata

On Tue, 13 Apr 2021 15:40:24 -0700 Saeed Mahameed wrote:
> > > Hence, allow user to disable metadata using driver specific devlink
> > > parameter.
> > > 
> > > Example to show and disable metadata before changing eswitch mode:
> > > $ devlink dev param show pci/0000:06:00.0 name esw_port_metadata
> > > pci/0000:06:00.0:
> > >   name esw_port_metadata type driver-specific
> > >     values:
> > >       cmode runtime value true
> > > 
> > > $ devlink dev param set pci/0000:06:00.0 \
> > >           name esw_port_metadata value false cmode runtime
> > > 
> > > $ devlink dev eswitch set pci/0000:06:00.0 mode switchdev  
> > 
> > Is this something that only gets enabled when device is put into
> > switchdev mode? That needs to be clarified in the documentation IMO 
> > to give peace of mind to all users who don't enable switchdev.  
> 
> Currently this is always enabled when switchdev is turned on, it
> affects the whole operation mode of the FDB and the offloaded flows so
> it can't be dynamic, it must be decided before user enables switchdev,
> it is needed only to allow LAG use cases, hence we add a disable knob
> for those who don't want LAG and could use some more packet rate.
> 
> Some documentation was pushed as part of this patch:
> please let me know if it needs improvement. (maybe we should add the
> benefit of packet rate ?)

Right, I didn't see the info on when the performance impact is felt 
in the documentation either.

AFAIU we have 3 categories of users:
1 - those who use esw|roce + LAG etc and therefore must leave this
    enabled;
2 - those who don't use LAG but use switchdev so they can turn it off;
3 - those who don't use switchdev or RoCE..

While the documentation should be clear for 1 and 2, group 3 does not
know (a) if they will be impacted; or (b) may not even know what kind 
of stacked devices documentation is referring to.