On Thu, 2009-07-16 at 15:55 -0700, Luis R. Rodriguez wrote:
> + * NL80211_WOW_TRIGGER_DISABLE_ALL: Disables all WoW triggers on
> + * the device.
I don't understand the need for this. How is that different from
userspace passing a 0 value for the attribute?
> + triggers_requested =
> + nla_get_u32(info->attrs[NL80211_ATTR_WOW_TRIGGERS_ENABLED]);
> +
> + if (triggers_requested & NL80211_WOW_TRIGGER_DISABLE_ALL) {
> + rdev->wow.triggers_enabled = 0;
> + goto out;
> + }
I mean, what do these extra lines of code buy us?
> + /* We only support magic packet right now */
> + if (!triggers_requested ||
> + !(triggers_requested & NL80211_WOW_TRIGGER_MAGIC_PACKET)) {
> + err = -EOPNOTSUPP;
> + goto out;
> + }
Oi that seems like a roundabout way of saying
if (triggers_requested & ~NL80211_WOW_TRIGGER_MAGIC_PACKET)
...
> + mutex_lock(&rdev->devlist_mtx);
> + /*
> + * It only makes sense to enable WoW if we're associated as a STA,
> + * the AP should be buffering frames for us. We'll discard all frames
> + * and only process the frames which will trigger us on in hardware.
> + */
> + list_for_each_entry(wdev, &rdev->netdev_list, list) {
I don't understand. You already have a netdev pointer from
get_rdev_dev_by_info_ifindex(info->attrs, &rdev, &dev);
why not just check the corresponding wdev for that netdev? I would store
this netdev in the cfg80211_wow struct too, so you know on
__cfg80211_disconnected() whether it was this netdev. And reject wow
configuration if a wow configuration is already active for a different
netdev?
That way you can configure wow 'per netdev' but it's really only for one
at a time.
johannes
Attachment:
signature.asc
Description: This is a digitally signed message part
