On Tue, 2008-12-02 at 22:04 +0200, Kalle Valo wrote:
> +void __ieee80211_stop_queues(struct ieee80211_hw *hw,
> + enum queue_stop_reason reason)
> {
> int i;
>
> for (i = 0; i < ieee80211_num_queues(hw); i++)
> - ieee80211_stop_queue(hw, i);
> + __ieee80211_stop_queue(hw, i, reason);
> +}
> +
> +void ieee80211_stop_queues(struct ieee80211_hw *hw)
> +{
> + __ieee80211_stop_queues(hw, IEEE80211_QUEUE_STOP_REASON_DRIVER);
> }
> EXPORT_SYMBOL(ieee80211_stop_queues);
>
> @@ -367,12 +409,18 @@ int ieee80211_queue_stopped(struct ieee80211_hw *hw, int queue)
> }
> EXPORT_SYMBOL(ieee80211_queue_stopped);
>
> -void ieee80211_wake_queues(struct ieee80211_hw *hw)
> +void __ieee80211_wake_queues(struct ieee80211_hw *hw,
> + enum queue_stop_reason reason)
> {
> int i;
>
> for (i = 0; i < hw->queues + hw->ampdu_queues; i++)
> - ieee80211_wake_queue(hw, i);
> + __ieee80211_wake_queue(hw, i, reason);
> +}
> +
> +void ieee80211_wake_queues(struct ieee80211_hw *hw)
> +{
> + __ieee80211_wake_queues(hw, IEEE80211_QUEUE_STOP_REASON_DRIVER);
> }
> EXPORT_SYMBOL(ieee80211_wake_queues);
I wonder if we should optimise the spinlock here?
Also, I'd prefer to first add this with just a single DRIVER reason
first, and then add the PS patches that add and use the second reason.
Other than that, it looks good to me, thanks!
johannes
Attachment:
signature.asc
Description: This is a digitally signed message part
