rx_err vs rx_drop

[Duncan Sands]

In the usbatm (i.e. speedtch) driver, we don't support OAM cells
right now, so we do:

                if (pti == ATM_PTI_E2EF5) {
                        atm_warn(instance, "%s: OAM not supported (vpi %d, vci %d)!\n", __func__, vpi, vci);
                        atomic_inc(&vcc->stats->rx_err);
                        continue;
                }

Is it correct to increment rx_err, or should we be incrementing
rx_drop?  The comment on rx_drop is:

    __HANDLE_ITEM(rx_drop);             /* RX out of memory */

We didn't run out of memory, but we do drop the packet...

Thanks for your help,

Duncan.