netdev - Re: [PATCH net 0/2] lantiq

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]

Message-ID: <aNNxC7-b3hduosIh@pidgin.makrotopia.org>
Date: Wed, 24 Sep 2025 05:18:19 +0100
From: Daniel Golle <daniel@...rotopia.org>
To: Jakub Kicinski <kuba@...nel.org>
Cc: Vladimir Oltean <vladimir.oltean@....com>, netdev@...r.kernel.org,
	Hauke Mehrtens <hauke@...ke-m.de>, Andrew Lunn <andrew@...n.ch>,
	"David S. Miller" <davem@...emloft.net>,
	Eric Dumazet <edumazet@...gle.com>, Paolo Abeni <pabeni@...hat.com>
Subject: Re: [PATCH net 0/2] lantiq_gswip fixes

On Mon, Sep 22, 2025 at 11:34:52AM -0700, Jakub Kicinski wrote:
> On Mon, 22 Sep 2025 14:07:17 +0300 Vladimir Oltean wrote:
> > - I don't think your local_termination.sh exercises the bug fixed by
> >   patch "[1/2] net: dsa: lantiq_gswip: move gswip_add_single_port_br()
> >   call to port_setup()". The port has to be initially down before
> >   joining a bridge, and be brought up afterwards. This can be tested
> >   manually. In local_termination.sh, although bridge_create() runs
> >   "ip link set $h2 up" after "ip link set $h2 master br0", $h2 was
> >   already up due to "simple_if_init $h2".
> 
> Waiting for more testing..

I've added printk statements to illustrate the function calls to
gswip_port_enable() and gswip_port_setup(), and tested both the current
'net' without (before.txt) and with (after.txt) patch
"net: dsa: lantiq_gswip: move gswip_add_single_port_br() call to port_setup()"
applied. This makes it obvious that gswip_port_enable() calls
gswip_add_single_port_br() even though the port is at this point
already a member of another bridge.

I'm ready to do more testing or spray for printk over it, just let me
know.

> 
> > - If the vast majority of users make use of this driver through OpenWrt,
> >   and if backporting to the required trees is done by OpenWrt and the
> >   fixes' presence in linux-stable is not useful, I can offer to resend
> >   this set plus the remaining patches all together through the net-next
> >   tree, and avoid complications such as merge conflicts.
> 
> FWIW I don't even see a real conflict when merging this. git seems to
> be figuring things out on its own.

My concern here was the upcoming merge of the 'net' tree with the
'net-next' tree which now already contains the splitting of the driver
into .h and .c file, and moved both into a dedicated folder.
This may result in needing (trivial) manual intervention.

It would be great if all of Vladimir's patches can be merged without
a long delay, so more patches adding support for newer hardware can
be added during the next merge window. Especially the conversion of
the open-coded register access functions to be replaced by regmap_*
calls should only be committed after Vladimir's fixes.

View attachment "before.txt" of type "text/plain" (19488 bytes)

View attachment "after.txt" of type "text/plain" (19773 bytes)