Bug 9701

Summary: hostap_cs_{resume,suspend}(): inconsequent NULL checking
Product: Drivers Reporter: Adrian Bunk (bunk)
Component: network-wirelessAssignee: drivers_network-wireless (drivers_network-wireless)
Status: CLOSED CODE_FIX    
Severity: normal CC: linville, protasnb
Priority: P1    
Hardware: All   
OS: Linux   
Kernel Version: 2.6.* Subsystem:
Regression: --- Bisected commit-id:

Description Adrian Bunk 2008-01-05 19:42:06 UTC
The Coverity checker spotted the following in drivers/net/wireless/hostap/hostap_cs.c:

<--  snip  -->

...
static int hostap_cs_suspend(struct pcmcia_device *link)
{
        struct net_device *dev = (struct net_device *) link->priv;
        int dev_open = 0;
        struct hostap_interface *iface = NULL;

        if (dev)
                iface = netdev_priv(dev);

        PDEBUG(DEBUG_EXTRA, "%s: CS_EVENT_PM_SUSPEND\n", dev_info);
        if (iface && iface->local)
                dev_open = iface->local->num_dev_open > 0;
        if (dev_open) {
                netif_stop_queue(dev);
                netif_device_detach(dev);
        }
        prism2_suspend(dev);

        return 0;
}

static int hostap_cs_resume(struct pcmcia_device *link)
{
        struct net_device *dev = (struct net_device *) link->priv;
        int dev_open = 0;
        struct hostap_interface *iface = NULL;

        if (dev)
                iface = netdev_priv(dev);

        PDEBUG(DEBUG_EXTRA, "%s: CS_EVENT_PM_RESUME\n", dev_info);

        if (iface && iface->local)
                dev_open = iface->local->num_dev_open > 0;

        prism2_hw_shutdown(dev, 1);
        prism2_hw_config(dev, dev_open ? 0 : 1);
        if (dev_open) {
                netif_device_attach(dev);
                netif_start_queue(dev);
        }

        return 0;
}
...

<--  snip  -->

If the "if (dev)" is false then there's a guaranteed NULL dereference later in the "prism2_suspend(dev)" resp. "prism2_hw_config(dev, dev_open ? 0 : 1)".
Comment 1 Natalie Protasevich 2008-06-05 00:09:00 UTC
Ugh, yes indeed! John, was this fixed somewhere? still same in the mainline...
Comment 2 Adrian Bunk 2008-07-10 23:20:18 UTC
fixed by commit fcee7a01ad7516eeb8dfdd0a17ef04cd2ee30757