On Fri, Aug 25, 2017 at 05:05:25PM +0300, Anton Volkov wrote:
> Hello.
>
> While searching for races in the Linux kernel I've come across
> "drivers/net/irda/nsc-ircc.ko" module. Here is a question that I came up
> with while analyzing results. Lines are given using the info from Linux
> v4.12.
>
> Consider the following case:
>
> Thread 1: Thread 2:
> nsc_ircc_init
> ->nsc_ircc_open
> self = netdev_priv(dev)
> register_netdev(dev)
> nsc_ircc_net_ioctl
> ->nsc_ircc_change_speed
> self->dongle_id = ... <READ self->io.dongle_id>
> (nsc-ircc.c: line 485) (nsc-ircc.c: line 1318)
> platform_device_register_simple
>
> Before the initialization of self->dongle_id in msc_ircc_open() its value is
> 0. Thus if read access to its value in nsc_ircc_change_speed occurs before
> the initialization there will be an attempt to change speed of dongle with
> undesired id (if the dongle with id 0 exists). Is this case feasible from
> your point of view?
>
> Thank you for your time.
>
> -- Anton Volkov
A first glance, that seems like a valid race. I'm not sure if
there is a netdev lock/status to protect the driver, because it looks
like doing any operation on a device before "open" has completed would
be dangerous for most drivers.
I don't have time to check the code paths, as I have not
looked at that code in ages.
Good luck !
Jean
