Re: [Linux-ATM-General] RX/close vcc race with solos/atmtcp/usbatm/he

[Stanislaw G. <st...@wp...>]

On Wed, May 26, 2010 at 12:16:24PM +0100, David Woodhouse wrote:
> The problem is that the 'find_vcc' functions in these drivers are
> returning a vcc with the ATM_VF_READY bit cleared, because it's already
> in the process of being destroyed. If we fix that simple oversight,
> there's still a race condition because the socket can still be closed
> (and completely freed, afaict) between our call to find_vcc() and our
> call to vcc->push() in the RX tasklet.
> 
> Here's a patch for solos-pci which should fix it. We prevent the race by
> making the dev->ops->close() function wait for the RX tasklet to
> complete, so it can't still be using the vcc in question.

I do not think this is problem with usbatm. We use custom vcc_list with
custom usbatm_vcc_data entries. We remove entry on atmdev_ops->close()
within tasklet_disable()/tasklet_enable() section. After that
rx tasklet will not find usbatm_vcc_data for corresponding vpi, vpi.

Stanislaw