iwlcore: Allow skb allocation from tasklet.

If RX queue becomes empty then we need to restock the queue from tasklet to prevent
ucode from starving. A caller to iwl_rx_allocate will decide if allocated buffer should
come from GFP_ATOMIC or GFP_KERNEL.

Signed-off-by: Mohamed Abbas <mohamed.abbas@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This commit is contained in:
Mohamed Abbas
2009-05-22 11:01:51 -07:00
committed by John W. Linville
parent ef850d7cb3
commit 4752c93c30
4 changed files with 34 additions and 10 deletions

View File

@@ -145,12 +145,14 @@ int iwl_rx_queue_update_write_ptr(struct iwl_priv *priv, struct iwl_rx_queue *q)
goto exit_unlock;
}
iwl_write_direct32(priv, rx_wrt_ptr_reg, q->write & ~0x7);
q->write_actual = (q->write & ~0x7);
iwl_write_direct32(priv, rx_wrt_ptr_reg, q->write_actual);
/* Else device is assumed to be awake */
} else {
/* Device expects a multiple of 8 */
iwl_write32(priv, rx_wrt_ptr_reg, q->write & ~0x7);
q->write_actual = (q->write & ~0x7);
iwl_write_direct32(priv, rx_wrt_ptr_reg, q->write_actual);
}
q->need_update = 0;
@@ -212,7 +214,7 @@ int iwl_rx_queue_restock(struct iwl_priv *priv)
/* If we've added more space for the firmware to place data, tell it.
* Increment device's write pointer in multiples of 8. */
if (write != (rxq->write & ~0x7)) {
if (rxq->write_actual != (rxq->write & ~0x7)) {
spin_lock_irqsave(&rxq->lock, flags);
rxq->need_update = 1;
spin_unlock_irqrestore(&rxq->lock, flags);
@@ -232,7 +234,7 @@ EXPORT_SYMBOL(iwl_rx_queue_restock);
* Also restock the Rx queue via iwl_rx_queue_restock.
* This is called as a scheduled work item (except for during initialization)
*/
void iwl_rx_allocate(struct iwl_priv *priv)
void iwl_rx_allocate(struct iwl_priv *priv, gfp_t priority)
{
struct iwl_rx_queue *rxq = &priv->rxq;
struct list_head *element;
@@ -254,7 +256,8 @@ void iwl_rx_allocate(struct iwl_priv *priv)
/* Alloc a new receive buffer */
rxb->skb = alloc_skb(priv->hw_params.rx_buf_size + 256,
GFP_KERNEL);
priority);
if (!rxb->skb) {
IWL_CRIT(priv, "Can not allocate SKB buffers\n");
/* We don't reschedule replenish work here -- we will
@@ -289,7 +292,7 @@ void iwl_rx_replenish(struct iwl_priv *priv)
{
unsigned long flags;
iwl_rx_allocate(priv);
iwl_rx_allocate(priv, GFP_KERNEL);
spin_lock_irqsave(&priv->lock, flags);
iwl_rx_queue_restock(priv);
@@ -297,6 +300,14 @@ void iwl_rx_replenish(struct iwl_priv *priv)
}
EXPORT_SYMBOL(iwl_rx_replenish);
void iwl_rx_replenish_now(struct iwl_priv *priv)
{
iwl_rx_allocate(priv, GFP_ATOMIC);
iwl_rx_queue_restock(priv);
}
EXPORT_SYMBOL(iwl_rx_replenish_now);
/* Assumes that the skb field of the buffers in 'pool' is kept accurate.
* If an SKB has been detached, the POOL needs to have its SKB set to NULL
@@ -352,6 +363,7 @@ int iwl_rx_queue_alloc(struct iwl_priv *priv)
/* Set us so that we have processed and used all buffers, but have
* not restocked the Rx queue with fresh buffers */
rxq->read = rxq->write = 0;
rxq->write_actual = 0;
rxq->free_count = 0;
rxq->need_update = 0;
return 0;
@@ -390,6 +402,7 @@ void iwl_rx_queue_reset(struct iwl_priv *priv, struct iwl_rx_queue *rxq)
/* Set us so that we have processed and used all buffers, but have
* not restocked the Rx queue with fresh buffers */
rxq->read = rxq->write = 0;
rxq->write_actual = 0;
rxq->free_count = 0;
spin_unlock_irqrestore(&rxq->lock, flags);
}