netdev - Re: [RFC PATCH net-next] net: create a NETDEV_ETH_IOCTL notifier for DSA to reject PTP on DSA master

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: Windows password security audit tool. GUI, reports in PDF.

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

Date:   Mon, 21 Mar 2022 13:28:29 -0700
From:   Jakub Kicinski <kuba@...nel.org>
To:     Vladimir Oltean <vladimir.oltean@....com>
Cc:     netdev@...r.kernel.org, Paolo Abeni <pabeni@...hat.com>,
        Florian Fainelli <f.fainelli@...il.com>,
        Andrew Lunn <andrew@...n.ch>,
        Vivien Didelot <vivien.didelot@...il.com>,
        Vladimir Oltean <olteanv@...il.com>,
        "David S. Miller" <davem@...emloft.net>,
        Richard Cochran <richardcochran@...il.com>
Subject: Re: [RFC PATCH net-next] net: create a NETDEV_ETH_IOCTL notifier
 for DSA to reject PTP on DSA master

On Fri, 18 Mar 2022 00:50:35 +0200 Vladimir Oltean wrote:
> The fact that PTP 2-step TX timestamping is deeply broken on DSA
> switches if the master also timestamps the same packets is well
> documented by commit f685e609a301 ("net: dsa: Deny PTP on master if
> switch supports it"). We attempt to help the users avoid shooting
> themselves in the foot by making DSA reject the timestamping ioctls on
> an interface that is a DSA master, and the switch tree beneath it
> contains switches which are aware of PTP.
> 
> The only problem is that there isn't an established way of intercepting
> ndo_eth_ioctl calls, so DSA creates avoidable burden upon the network
> stack by creating a struct dsa_netdevice_ops with overlaid function
> pointers that are manually checked from the relevant call sites. There
> used to be 2 such dsa_netdevice_ops, but now, ndo_eth_ioctl is the only
> one left.

Remind me - are the DSA CPU-side interfaces linked as lower devices 
of the ports?

> In fact, the underlying reason which is prompting me to make this change
> is that I'd like to hide as many DSA data structures from public API as
> I can. But struct net_device :: dsa_ptr is a struct dsa_port (which is a
> huge structure), and I'd like to create a smaller structure. I'd like
> struct dsa_netdevice_ops to not be a part of this, so this is how the
> need to delete it arose.

Isn't it enough to move the implementation to a C source instead 
of having it be a static inline?

> The established way for unrelated modules to react on a net device event
> is via netdevice notifiers. These have the advantage of loose coupling,
> i.e. they work even when DSA is built as module, without resorting to
> static inline functions (which cannot offer the desired data structure
> encapsulation).
> 
> Signed-off-by: Vladimir Oltean <vladimir.oltean@....com>
> ---
> I'd mostly like to take this opportunity to raise a discussion about how
> to handle this. It's clear that calling the notifier chain is less
> efficient than having some dev->dsa_ptr checks, but I'm not sure if the
> ndo_eth_ioctl can tolerate the extra performance hit at the expense of
> some code cleanliness.
> 
> Of course, what would be great is if we didn't have the limitation to
> begin with, but the effort to add UAPI for multiple TX timestamps per
> packet isn't proportional to the stated goal here, which is to hide some
> DSA data structures.

Was there a reason we haven't converted the timestamping to ndos?
Just a matter of finding someone with enough cycles to go thru all 
the drivers?