vxge: improve tx performance by using mmiowb() instead of wmb()

PATCH In review Benjamin LaHaise 1 participant 1 message Aug 4, 2009

Subsystems (2)

Networking Network drivers

lore

Share & tools

lore

Benjamin LaHaise

author

2009-08-04 20:39 UTC #0 lore raw

PATCH 3/6 vxge: improve tx performance by using mmiowb() instead of wmb()

wmb() is extremely heavy on x86. The semantics required in the driver are provided by mmiowb(), so use that and improve tx performance on P4 Xeons by 5-10%.

Signed-off-by Benjamin LaHaise <[email protected]>

Signed-off-by Sreenivasa Honnur <[email protected]>

Signed-off-by Ramkrishna Vepa <[email protected]>

---

drivers/net/vxge/vxge-traffic.c +2 -2

index 370f55c..0efb769 100644
@@ -1070,11 +1070,11 @@ static void __vxge_hw_non_offload_db_post(struct __vxge_hw_fifo *fifo,
 		VXGE_HW_NODBW_GET_NO_SNOOP(no_snoop),
 		&fifo->nofl_db->control_0);
 
-	wmb();
+	mmiowb();
 
 	writeq(txdl_ptr, &fifo->nofl_db->txdl_ptr);
-	wmb();
 
+	mmiowb();
 }
 
 /**