e1000: Add performance enahancement by balancing TX and RX
authorJeff Kirsher <jeffrey.t.kirsher@intel.com>
Fri, 3 Mar 2006 02:20:43 +0000 (18:20 -0800)
committerroot <root@jk-desktop.jf.intel.com>
Fri, 3 Mar 2006 02:20:43 +0000 (18:20 -0800)
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: John Ronciak <john.ronciak@intel.com>
drivers/net/e1000/e1000_main.c

index 58561a4955e3bc6140c6f906942bc883074477a2..e80378a220d4c99e19f67b0320cdec97bde5e215 100644 (file)
@@ -3349,6 +3349,9 @@ e1000_clean_tx_irq(struct e1000_adapter *adapter,
        struct e1000_tx_desc *tx_desc, *eop_desc;
        struct e1000_buffer *buffer_info;
        unsigned int i, eop;
+#ifdef CONFIG_E1000_NAPI
+       unsigned int count = 0;
+#endif
        boolean_t cleaned = FALSE;
 
        i = tx_ring->next_to_clean;
@@ -3370,6 +3373,11 @@ e1000_clean_tx_irq(struct e1000_adapter *adapter,
 
                eop = tx_ring->buffer_info[i].next_to_watch;
                eop_desc = E1000_TX_DESC(*tx_ring, eop);
+#ifdef CONFIG_E1000_NAPI
+#define E1000_TX_WEIGHT 64
+               /* weight of a sort for tx, to avoid endless transmit cleanup */
+               if (count++ == E1000_TX_WEIGHT) break;
+#endif
        }
 
        tx_ring->next_to_clean = i;