[E1000] Added Interrupt Throttle Rate tuning support

* Added Interrupt Throttle Rate tuning support

Documentation/networking/e1000.txt

@@ -123,6 +123,20 @@ Default: Read flow control settings from the EEPROM
     This parameter controls the automatic generation(Tx) and response(Rx) to 
     Ethernet PAUSE frames.
 
+InterruptThrottleRate
+Valid Range: 100-100000 (0=off, 1=dynamic)
+Default Value: 1
+    This value represents the maximum number of interrupts per second the 
+    controller generates. InterruptThrottleRate is another setting used in 
+    interrupt moderation. Dynamic mode uses a heuristic algorithm to adjust 
+    InterruptThrottleRate based on the current traffic load.
+
+    NOTE: InterruptThrottleRate takes precedence over the TxAbsIntDelay and 
+          RxAbsIntDelay parameters. In other words, minimizing the receive 
+          and/or transmit absolute delays does not force the controller to 
+          generate more interrupts than what the Interrupt Throttle Rate 
+          allows.
+
 RxDescriptors
 Valid Range: 80-256 for 82542 and 82543-based adapters
              80-4096 for 82540, 82544, 82545, and 82546-based adapters

drivers/net/e1000/e1000.h

@@ -187,6 +187,7 @@ struct e1000_adapter {
 	uint32_t txd_cmd;
 	uint32_t tx_int_delay;
 	uint32_t tx_abs_int_delay;
+	uint32_t gotcl;
 	uint32_t tx_fifo_head;
 	uint32_t tx_head_addr;
 	uint32_t tx_fifo_size;
@@ -199,6 +200,10 @@ struct e1000_adapter {
 	uint32_t rx_int_delay;
 	uint32_t rx_abs_int_delay;
 	boolean_t rx_csum;
+	uint32_t gorcl;
+
+	/* Interrupt Throttle Rate */
+	uint32_t itr;
 
 	/* OS defined structs */
 	struct net_device *netdev;

drivers/net/e1000/e1000_main.c

@@ -937,12 +937,9 @@ e1000_configure_rx(struct e1000_adapter *adapter)
 
 	if(adapter->hw.mac_type >= e1000_82540) {
 		E1000_WRITE_REG(&adapter->hw, RADV, adapter->rx_abs_int_delay);
-
-		/* Set the interrupt throttling rate.  Value is calculated
-		 * as DEFAULT_ITR = 1/(MAX_INTS_PER_SEC * 256ns) */
-#define MAX_INTS_PER_SEC        8000
-#define DEFAULT_ITR             1000000000/(MAX_INTS_PER_SEC * 256)
-		E1000_WRITE_REG(&adapter->hw, ITR, DEFAULT_ITR);
+		if(adapter->itr > 1)
+			E1000_WRITE_REG(&adapter->hw, ITR,
+				1000000000 / (adapter->itr * 256));
 	}
 
 	/* Setup the Base and Length of the Rx Descriptor Ring */
@@ -1391,6 +1388,18 @@ e1000_watchdog(unsigned long data)
 		}
 	}
 
+	/* Dynamic mode for Interrupt Throttle Rate (ITR) */
+	if(adapter->hw.mac_type >= e1000_82540 && adapter->itr == 1) {
+		/* Symmetric Tx/Rx gets a reduced ITR=2000; Total
+		 * asymmetrical Tx or Rx gets ITR=8000; everyone
+		 * else is between 2000-8000. */
+		uint32_t goc = (adapter->gotcl + adapter->gorcl) / 10000;
+		uint32_t dif = (adapter->gotcl > adapter->gorcl ? 
+			adapter->gotcl - adapter->gorcl :
+			adapter->gorcl - adapter->gotcl) / 10000;
+		uint32_t itr = goc > 0 ? (dif * 6000 / goc + 2000) : 8000;
+		E1000_WRITE_REG(&adapter->hw, ITR, 1000000000 / (itr * 256));
+	}
 
 	/* Cause software interrupt to ensure rx ring is cleaned */
 	E1000_WRITE_REG(&adapter->hw, ICS, E1000_ICS_RXDMT0);
@@ -1811,7 +1820,8 @@ e1000_update_stats(struct e1000_adapter *adapter)
 
 	adapter->stats.crcerrs += E1000_READ_REG(hw, CRCERRS);
 	adapter->stats.gprc += E1000_READ_REG(hw, GPRC);
-	adapter->stats.gorcl += E1000_READ_REG(hw, GORCL);
+	adapter->gorcl = E1000_READ_REG(hw, GORCL);
+	adapter->stats.gorcl += adapter->gorcl;
 	adapter->stats.gorch += E1000_READ_REG(hw, GORCH);
 	adapter->stats.bprc += E1000_READ_REG(hw, BPRC);
 	adapter->stats.mprc += E1000_READ_REG(hw, MPRC);
@@ -1842,7 +1852,8 @@ e1000_update_stats(struct e1000_adapter *adapter)
 	adapter->stats.xofftxc += E1000_READ_REG(hw, XOFFTXC);
 	adapter->stats.fcruc += E1000_READ_REG(hw, FCRUC);
 	adapter->stats.gptc += E1000_READ_REG(hw, GPTC);
-	adapter->stats.gotcl += E1000_READ_REG(hw, GOTCL);
+	adapter->gotcl = E1000_READ_REG(hw, GOTCL);
+	adapter->stats.gotcl += adapter->gotcl;
 	adapter->stats.gotch += E1000_READ_REG(hw, GOTCH);
 	adapter->stats.rnbc += E1000_READ_REG(hw, RNBC);
 	adapter->stats.ruc += E1000_READ_REG(hw, RUC);

drivers/net/e1000/e1000_param.c

@@ -183,6 +183,15 @@ E1000_PARAM(RxIntDelay, "Receive Interrupt Delay");
 
 E1000_PARAM(RxAbsIntDelay, "Receive Absolute Interrupt Delay");
 
+/* Interrupt Throttle Rate (interrupts/sec)
+ *
+ * Valid Range: 100-100000 (0=off, 1=dynamic)
+ *
+ * Default Value: 1
+ */
+
+E1000_PARAM(InterruptThrottleRate, "Interrupt Throttling Rate");
+
 #define AUTONEG_ADV_DEFAULT  0x2F
 #define AUTONEG_ADV_MASK     0x2F
 #define FLOW_CONTROL_DEFAULT FLOW_CONTROL_FULL
@@ -213,6 +222,10 @@ E1000_PARAM(RxAbsIntDelay, "Receive Absolute Interrupt Delay");
 #define MAX_TXABSDELAY            0xFFFF
 #define MIN_TXABSDELAY                 0
 
+#define DEFAULT_ITR                    1
+#define MAX_ITR                   100000
+#define MIN_ITR                      100
+
 struct e1000_option {
 	enum { enable_option, range_option, list_option } type;
 	char *name;
@@ -422,6 +435,27 @@ e1000_check_options(struct e1000_adapter *adapter)
 		adapter->rx_abs_int_delay = RxAbsIntDelay[bd];
 		e1000_validate_option(&adapter->rx_abs_int_delay, &opt);
 	}
+	{ /* Interrupt Throttling Rate */
+		struct e1000_option opt = {
+			.type = range_option,
+			.name = "Interrupt Throttling Rate (ints/sec)",
+			.err  = "using default of " __MODULE_STRING(DEFAULT_ITR),
+			.def  = DEFAULT_ITR,
+			.arg  = { .r = { .min = MIN_ITR,
+					 .max = MAX_ITR }}
+		};
+
+		adapter->itr = InterruptThrottleRate[bd];
+		if(adapter->itr == 0) {
+			printk(KERN_INFO "%s turned off\n", opt.name);
+		} else if(adapter->itr == 1 || adapter->itr == -1) {
+			/* Dynamic mode */
+			adapter->itr = 1;
+		} else {
+			e1000_validate_option(&adapter->itr, &opt);
+		}
+	}
+
 	switch(adapter->hw.media_type) {
 	case e1000_media_type_fiber:
 		e1000_check_fiber_options(adapter);