ath10k: reduce htt tx/rx spinlock overhead
authorMichal Kazior <michal.kazior@tieto.com>
Thu, 27 Feb 2014 16:50:05 +0000 (18:50 +0200)
committerKalle Valo <kvalo@qca.qualcomm.com>
Fri, 28 Feb 2014 09:59:55 +0000 (11:59 +0200)
It is inefficient to grab irqsave spinlocks for
skb lists for each queue/dequeue action.

Using rx_ring.lock and tx_lock allows to use less
heavy bh spinlock functions and moving locking
upwards allows to toggle spinlocks less often.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>

No differences found