[PATCH v2] fm10k: Add skb->xmit_more support

[PATCH v2] fm10k: Add skb->xmit_more support

[alexander.duyck]

From: Alexander Duyck <alexander.h.duyck@redhat.com>

This change adds support for skb->xmit_more based on the changes that were
made to igb to support the feature.  The main changes are moving up the
check for maybe_stop_tx so that we can check netif_xmit_stopped to determine
if we must write the tail because we can add no further buffers.

Acked-by: Matthew Vick <matthew.vick@intel.com>
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
---

v2: Minor fix to patch description for typo.

 drivers/net/ethernet/intel/fm10k/fm10k_main.c |   65 +++++++++++++------------
 1 file changed, 34 insertions(+), 31 deletions(-)

diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_main.c b/drivers/net/ethernet/intel/fm10k/fm10k_main.c
index 6c800a3..8ad7ff4 100644
--- a/drivers/net/ethernet/intel/fm10k/fm10k_main.c
+++ b/drivers/net/ethernet/intel/fm10k/fm10k_main.c
@@ -930,6 +930,30 @@ static bool fm10k_tx_desc_push(struct fm10k_ring *tx_ring,
 	return i == tx_ring->count;
 }
 
+static int __fm10k_maybe_stop_tx(struct fm10k_ring *tx_ring, u16 size)
+{
+	netif_stop_subqueue(tx_ring->netdev, tx_ring->queue_index);
+
+	smp_mb();
+
+	/* We need to check again in a case another CPU has just
+	 * made room available. */
+	if (likely(fm10k_desc_unused(tx_ring) < size))
+		return -EBUSY;
+
+	/* A reprieve! - use start_queue because it doesn't call schedule */
+	netif_start_subqueue(tx_ring->netdev, tx_ring->queue_index);
+	++tx_ring->tx_stats.restart_queue;
+	return 0;
+}
+
+static inline int fm10k_maybe_stop_tx(struct fm10k_ring *tx_ring, u16 size)
+{
+	if (likely(fm10k_desc_unused(tx_ring) >= size))
+		return 0;
+	return __fm10k_maybe_stop_tx(tx_ring, size);
+}
+
 static void fm10k_tx_map(struct fm10k_ring *tx_ring,
 			 struct fm10k_tx_buffer *first)
 {
@@ -1023,13 +1047,18 @@ static void fm10k_tx_map(struct fm10k_ring *tx_ring,
 
 	tx_ring->next_to_use = i;
 
+	/* Make sure there is space in the ring for the next send. */
+	fm10k_maybe_stop_tx(tx_ring, DESC_NEEDED);
+
 	/* notify HW of packet */
-	writel(i, tx_ring->tail);
+	if (netif_xmit_stopped(txring_txq(tx_ring)) || !skb->xmit_more) {
+		writel(i, tx_ring->tail);
 
-	/* we need this if more than one processor can write to our tail
-	 * at a time, it synchronizes IO on IA64/Altix systems
-	 */
-	mmiowb();
+		/* we need this if more than one processor can write to our tail
+		 * at a time, it synchronizes IO on IA64/Altix systems
+		 */
+		mmiowb();
+	}
 
 	return;
 dma_error:
@@ -1049,30 +1078,6 @@ dma_error:
 	tx_ring->next_to_use = i;
 }
 
-static int __fm10k_maybe_stop_tx(struct fm10k_ring *tx_ring, u16 size)
-{
-	netif_stop_subqueue(tx_ring->netdev, tx_ring->queue_index);
-
-	smp_mb();
-
-	/* We need to check again in a case another CPU has just
-	 * made room available. */
-	if (likely(fm10k_desc_unused(tx_ring) < size))
-		return -EBUSY;
-
-	/* A reprieve! - use start_queue because it doesn't call schedule */
-	netif_start_subqueue(tx_ring->netdev, tx_ring->queue_index);
-	++tx_ring->tx_stats.restart_queue;
-	return 0;
-}
-
-static inline int fm10k_maybe_stop_tx(struct fm10k_ring *tx_ring, u16 size)
-{
-	if (likely(fm10k_desc_unused(tx_ring) >= size))
-		return 0;
-	return __fm10k_maybe_stop_tx(tx_ring, size);
-}
-
 netdev_tx_t fm10k_xmit_frame_ring(struct sk_buff *skb,
 				  struct fm10k_ring *tx_ring)
 {
@@ -1117,8 +1122,6 @@ netdev_tx_t fm10k_xmit_frame_ring(struct sk_buff *skb,
 
 	fm10k_tx_map(tx_ring, first);
 
-	fm10k_maybe_stop_tx(tx_ring, DESC_NEEDED);
-
 	return NETDEV_TX_OK;
 
 out_drop:

Re: [PATCH v2] fm10k: Add skb->xmit_more support

[Jeff Kirsher]

[-- Attachment #1: Type: text/plain, Size: 888 bytes --]

On Fri, 2014-10-10 at 14:30 -0700, alexander.duyck@gmail.com wrote:
> From: Alexander Duyck <alexander.h.duyck@redhat.com>
> 
> This change adds support for skb->xmit_more based on the changes that
> were
> made to igb to support the feature.  The main changes are moving up
> the
> check for maybe_stop_tx so that we can check netif_xmit_stopped to
> determine
> if we must write the tail because we can add no further buffers.
> 
> Acked-by: Matthew Vick <matthew.vick@intel.com>
> Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
> ---
> 
> v2: Minor fix to patch description for typo.
> 
>  drivers/net/ethernet/intel/fm10k/fm10k_main.c |   65
> +++++++++++++------------
>  1 file changed, 34 insertions(+), 31 deletions(-)

Acked-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Its like you never left Intel Alex... :-)
Dave, please pull this.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

Re: [PATCH v2] fm10k: Add skb->xmit_more support

[David Miller]

From: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Date: Sat, 11 Oct 2014 04:14:43 -0700

> On Fri, 2014-10-10 at 14:30 -0700, alexander.duyck@gmail.com wrote:
>> From: Alexander Duyck <alexander.h.duyck@redhat.com>
>> 
>> This change adds support for skb->xmit_more based on the changes that
>> were
>> made to igb to support the feature.  The main changes are moving up
>> the
>> check for maybe_stop_tx so that we can check netif_xmit_stopped to
>> determine
>> if we must write the tail because we can add no further buffers.
>> 
>> Acked-by: Matthew Vick <matthew.vick@intel.com>
>> Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
>> ---
>> 
>> v2: Minor fix to patch description for typo.
>> 
>>  drivers/net/ethernet/intel/fm10k/fm10k_main.c |   65
>> +++++++++++++------------
>>  1 file changed, 34 insertions(+), 31 deletions(-)
> 
> Acked-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
> 
> Its like you never left Intel Alex... :-)
> Dave, please pull this.

Applied, thanks everyone.