Re: RTNL: assertion failed at net/core/dev.c

[Joe Damato <jdamato@fastly.com>]

On Tue, Oct 22, 2024 at 09:27:03AM -0700, Joe Damato wrote:
> On Tue, Oct 22, 2024 at 11:24:45AM +0300, Dmitry Antipov wrote:
> > Hello,
> > 
> > running around https://syzkaller.appspot.com/bug?extid=b390c8062d8387b6272a
> > with net-next and linux-next, I've noticed the following:

[...]

> 
> Thanks for reporting this.
> 
> The issue is that in the path highlighted above, rtnl is not held
> before e1000_down is called.
> 
> I believe this will fix the issue you are seeing, but I am not sure
> if a similar change needs to be added for power management's suspend
> (which also eventually calls e1000_down):
> 
> diff --git a/drivers/net/ethernet/intel/e1000/e1000_main.c b/drivers/net/ethernet/intel/e1000/e1000_main.c
> index 4de9b156b2be..ebbd3fa3a5c8 100644
> --- a/drivers/net/ethernet/intel/e1000/e1000_main.c
> +++ b/drivers/net/ethernet/intel/e1000/e1000_main.c
> @@ -5074,7 +5074,9 @@ static int __e1000_shutdown(struct pci_dev *pdev, bool *enable_wake)
>                         usleep_range(10000, 20000);
> 
>                 WARN_ON(test_bit(__E1000_RESETTING, &adapter->flags));
> +               rtnl_lock()

As you can see, I didn't compile test this as I'm missing a
semicolon (sorry).

I am still reading through the source to see what other spots may
need rtnl_lock.