[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20250424130813.GZ1213339@ziepe.ca>
Date: Thu, 24 Apr 2025 10:08:13 -0300
From: Jason Gunthorpe <jgg@...pe.ca>
To: Abhijit Gangurde <abhijit.gangurde@....com>
Cc: shannon.nelson@....com, brett.creeley@....com, davem@...emloft.net,
edumazet@...gle.com, kuba@...nel.org, pabeni@...hat.com,
corbet@....net, leon@...nel.org, andrew+netdev@...n.ch,
allen.hubbe@....com, nikhil.agarwal@....com,
linux-rdma@...r.kernel.org, netdev@...r.kernel.org,
linux-doc@...r.kernel.org, linux-kernel@...r.kernel.org,
Andrew Boyer <andrew.boyer@....com>
Subject: Re: [PATCH 08/14] RDMA/ionic: Register auxiliary module for ionic
ethernet adapter
On Wed, Apr 23, 2025 at 03:59:07PM +0530, Abhijit Gangurde wrote:
> +static int ionic_aux_probe(struct auxiliary_device *adev,
> + const struct auxiliary_device_id *id)
> +{
> + struct ionic_aux_dev *ionic_adev;
> + struct net_device *ndev;
> + struct ionic_ibdev *dev;
> +
> + ionic_adev = container_of(adev, struct ionic_aux_dev, adev);
> + ndev = ionic_api_get_netdev_from_handle(ionic_adev->handle);
It must not do this, the net_device should not go into the IB driver,
like this that will create a huge complex tangled mess.
The netdev(s) come in indirectly through the gid table and through the
net notifiers and ib_device_set_netdev() and they should only be
touched in paths dealing with specific areas.
So don't use things like netdev_err, we have ib_err/dev_err and
related instead for IB drivers to use.
> +struct ionic_ibdev {
> + struct ib_device ibdev;
> +
> + struct device *hwdev;
> + struct net_device *ndev;
Same here, this member should not exist, and it didn't hold a
refcount for this pointer.
Jason
Powered by blists - more mailing lists