net: ll_temac: fix checksum offload logic

The current checksum offload code does not work and this corrects that functionality. It also updates the interrupt coallescing initialization so than there are fewer interrupts and performance is increased. Signed-off-by: Brian Hill <brian.hill@xilinx.com> Signed-off-by: John Linn <john.linn@xilinx.com> Signed-off-by: David S. Miller <davem@davemloft.net>

net: ll_temac: fix checksum offload logic
The current checksum offload code does not work and this corrects that functionality. It also updates the interrupt coallescing initialization so than there are fewer interrupts and performance is increased. Signed-off-by: Brian Hill <brian.hill@xilinx.com> Signed-off-by: John Linn <john.linn@xilinx.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Brian Hill · David S. Miller
1 parent 755fae0ac4
Showing 2 changed files with 63 additions and 24 deletions Side-by-side Diff
drivers/net/ll_temac.h
drivers/net/ll_temac_main.c
@@ -295,6 +295,10 @@
  
 #define MULTICAST_CAM_TABLE_NUM 4
  
+/* TEMAC Synthesis features */
+#define TEMAC_FEATURE_RX_CSUM  (1 << 0)
+#define TEMAC_FEATURE_TX_CSUM  (1 << 1)
+
 /* TX/RX CURDESC_PTR points to first descriptor */
 /* TX/RX TAILDESC_PTR points to last descriptor in linked list */
  
@@ -353,6 +357,7 @@
 	struct mutex indirect_mutex;
 	u32 options;			/* Current options word */
 	int last_link;
+	unsigned int temac_features;
  
 	/* Buffer descriptors */
 	struct cdmac_bd *tx_bd_v;
@@ -245,7 +245,7 @@
 					  CHNL_CTRL_IRQ_COAL_EN);
 	/* 0x10220483 */
 	/* 0x00100483 */
-	lp->dma_out(lp, RX_CHNL_CTRL, 0xff010000 |
+	lp->dma_out(lp, RX_CHNL_CTRL, 0xff070000 |
 					  CHNL_CTRL_IRQ_EN |
 					  CHNL_CTRL_IRQ_DLY_EN |
 					  CHNL_CTRL_IRQ_COAL_EN |
@@ -574,6 +574,10 @@
 		if (cur_p->app4)
 			dev_kfree_skb_irq((struct sk_buff *)cur_p->app4);
 		cur_p->app0 = 0;
+		cur_p->app1 = 0;
+		cur_p->app2 = 0;
+		cur_p->app3 = 0;
+		cur_p->app4 = 0;
  
 		ndev->stats.tx_packets++;
 		ndev->stats.tx_bytes += cur_p->len;
@@ -589,6 +593,29 @@
 	netif_wake_queue(ndev);
 }
  
+static inline int temac_check_tx_bd_space(struct temac_local *lp, int num_frag)
+{
+	struct cdmac_bd *cur_p;
+	int tail;
+
+	tail = lp->tx_bd_tail;
+	cur_p = &lp->tx_bd_v[tail];
+
+	do {
+		if (cur_p->app0)
+			return NETDEV_TX_BUSY;
+
+		tail++;
+		if (tail >= TX_BD_NUM)
+			tail = 0;
+
+		cur_p = &lp->tx_bd_v[tail];
+		num_frag--;
+	} while (num_frag >= 0);
+
+	return 0;
+}
+
 static int temac_start_xmit(struct sk_buff *skb, struct net_device *ndev)
 {
 	struct temac_local *lp = netdev_priv(ndev);
@@ -603,7 +630,7 @@
 	start_p = lp->tx_bd_p + sizeof(*lp->tx_bd_v) * lp->tx_bd_tail;
 	cur_p = &lp->tx_bd_v[lp->tx_bd_tail];
  
-	if (cur_p->app0 & STS_CTRL_APP0_CMPLT) {
+	if (temac_check_tx_bd_space(lp, num_frag)) {
 		if (!netif_queue_stopped(ndev)) {
 			netif_stop_queue(ndev);
 			return NETDEV_TX_BUSY;
  
  
@@ -613,29 +640,14 @@
  
 	cur_p->app0 = 0;
 	if (skb->ip_summed == CHECKSUM_PARTIAL) {
-		const struct iphdr *ip = ip_hdr(skb);
-		int length = 0, start = 0, insert = 0;
+		unsigned int csum_start_off = skb_transport_offset(skb);
+		unsigned int csum_index_off = csum_start_off + skb->csum_offset;
  
-		switch (ip->protocol) {
-		case IPPROTO_TCP:
-			start = sizeof(struct iphdr) + ETH_HLEN;
-			insert = sizeof(struct iphdr) + ETH_HLEN + 16;
-			length = ip->tot_len - sizeof(struct iphdr);
-			break;
-		case IPPROTO_UDP:
-			start = sizeof(struct iphdr) + ETH_HLEN;
-			insert = sizeof(struct iphdr) + ETH_HLEN + 6;
-			length = ip->tot_len - sizeof(struct iphdr);
-			break;
-		default:
-			break;
-		}
-		cur_p->app1 = ((start << 16) | insert);
-		cur_p->app2 = csum_tcpudp_magic(ip->saddr, ip->daddr,
-						length, ip->protocol, 0);
-		skb->data[insert] = 0;
-		skb->data[insert + 1] = 0;
+		cur_p->app0 |= 1; /* TX Checksum Enabled */
+		cur_p->app1 = (csum_start_off << 16) | csum_index_off;
+		cur_p->app2 = 0;  /* initial checksum seed */
 	}
+
 	cur_p->app0 |= STS_CTRL_APP0_SOP;
 	cur_p->len = skb_headlen(skb);
 	cur_p->phys = dma_map_single(ndev->dev.parent, skb->data, skb->len,
@@ -699,6 +711,15 @@
 		skb->protocol = eth_type_trans(skb, ndev);
 		skb->ip_summed = CHECKSUM_NONE;
  
+		/* if we're doing rx csum offload, set it up */
+		if (((lp->temac_features & TEMAC_FEATURE_RX_CSUM) != 0) &&
+			(skb->protocol == __constant_htons(ETH_P_IP)) &&
+			(skb->len > 64)) {
+
+			skb->csum = cur_p->app3 & 0xFFFF;
+			skb->ip_summed = CHECKSUM_COMPLETE;
+		}
+
 		netif_rx(skb);
  
 		ndev->stats.rx_packets++;
@@ -883,6 +904,7 @@
 	struct temac_local *lp;
 	struct net_device *ndev;
 	const void *addr;
+	__be32 *p;
 	int size, rc = 0;
  
 	/* Init network device structure */
@@ -925,6 +947,18 @@
 		dev_err(&op->dev, "could not map temac regs.\n");
 		goto nodev;
 	}
+
+	/* Setup checksum offload, but default to off if not specified */
+	lp->temac_features = 0;
+	p = (__be32 *)of_get_property(op->dev.of_node, "xlnx,txcsum", NULL);
+	if (p && be32_to_cpu(*p)) {
+		lp->temac_features |= TEMAC_FEATURE_TX_CSUM;
+		/* Can checksum TCP/UDP over IPv4. */
+		ndev->features |= NETIF_F_IP_CSUM;
+	}
+	p = (__be32 *)of_get_property(op->dev.of_node, "xlnx,rxcsum", NULL);
+	if (p && be32_to_cpu(*p))
+		lp->temac_features |= TEMAC_FEATURE_RX_CSUM;
  
 	/* Find the DMA node, map the DMA registers, and decode the DMA IRQs */
 	np = of_parse_phandle(op->node, "llink-connected", 0);
...	...	@@ -295,6 +295,10 @@
295	295
296	296	#define MULTICAST_CAM_TABLE_NUM 4
297	297
	298	+/* TEMAC Synthesis features */
	299	+#define TEMAC_FEATURE_RX_CSUM (1 << 0)
	300	+#define TEMAC_FEATURE_TX_CSUM (1 << 1)
	301	+
298	302	/* TX/RX CURDESC_PTR points to first descriptor */
299	303	/* TX/RX TAILDESC_PTR points to last descriptor in linked list */
300	304
...	...	@@ -353,6 +357,7 @@
353	357	struct mutex indirect_mutex;
354	358	u32 options; /* Current options word */
355	359	int last_link;
	360	+ unsigned int temac_features;
356	361
357	362	/* Buffer descriptors */
358	363	struct cdmac_bd *tx_bd_v;
...	...	@@ -245,7 +245,7 @@
245	245	CHNL_CTRL_IRQ_COAL_EN);
246	246	/* 0x10220483 */
247	247	/* 0x00100483 */
248		- lp->dma_out(lp, RX_CHNL_CTRL, 0xff010000 \|
	248	+ lp->dma_out(lp, RX_CHNL_CTRL, 0xff070000 \|
249	249	CHNL_CTRL_IRQ_EN \|
250	250	CHNL_CTRL_IRQ_DLY_EN \|
251	251	CHNL_CTRL_IRQ_COAL_EN \|
...	...	@@ -574,6 +574,10 @@
574	574	if (cur_p->app4)
575	575	dev_kfree_skb_irq((struct sk_buff *)cur_p->app4);
576	576	cur_p->app0 = 0;
	577	+ cur_p->app1 = 0;
	578	+ cur_p->app2 = 0;
	579	+ cur_p->app3 = 0;
	580	+ cur_p->app4 = 0;
577	581
578	582	ndev->stats.tx_packets++;
579	583	ndev->stats.tx_bytes += cur_p->len;
...	...	@@ -589,6 +593,29 @@
589	593	netif_wake_queue(ndev);
590	594	}
591	595
	596	+static inline int temac_check_tx_bd_space(struct temac_local *lp, int num_frag)
	597	+{
	598	+ struct cdmac_bd *cur_p;
	599	+ int tail;
	600	+
	601	+ tail = lp->tx_bd_tail;
	602	+ cur_p = &lp->tx_bd_v[tail];
	603	+
	604	+ do {
	605	+ if (cur_p->app0)
	606	+ return NETDEV_TX_BUSY;
	607	+
	608	+ tail++;
	609	+ if (tail >= TX_BD_NUM)
	610	+ tail = 0;
	611	+
	612	+ cur_p = &lp->tx_bd_v[tail];
	613	+ num_frag--;
	614	+ } while (num_frag >= 0);
	615	+
	616	+ return 0;
	617	+}
	618	+
592	619	static int temac_start_xmit(struct sk_buff skb, struct net_device ndev)
593	620	{
594	621	struct temac_local *lp = netdev_priv(ndev);
...	...	@@ -603,7 +630,7 @@
603	630	start_p = lp->tx_bd_p + sizeof(lp->tx_bd_v) lp->tx_bd_tail;
604	631	cur_p = &lp->tx_bd_v[lp->tx_bd_tail];
605	632
606		- if (cur_p->app0 & STS_CTRL_APP0_CMPLT) {
	633	+ if (temac_check_tx_bd_space(lp, num_frag)) {
607	634	if (!netif_queue_stopped(ndev)) {
608	635	netif_stop_queue(ndev);
609	636	return NETDEV_TX_BUSY;
610	637
611	638
...	...	@@ -613,29 +640,14 @@
613	640
614	641	cur_p->app0 = 0;
615	642	if (skb->ip_summed == CHECKSUM_PARTIAL) {
616		- const struct iphdr *ip = ip_hdr(skb);
617		- int length = 0, start = 0, insert = 0;
	643	+ unsigned int csum_start_off = skb_transport_offset(skb);
	644	+ unsigned int csum_index_off = csum_start_off + skb->csum_offset;
618	645
619		- switch (ip->protocol) {
620		- case IPPROTO_TCP:
621		- start = sizeof(struct iphdr) + ETH_HLEN;
622		- insert = sizeof(struct iphdr) + ETH_HLEN + 16;
623		- length = ip->tot_len - sizeof(struct iphdr);
624		- break;
625		- case IPPROTO_UDP:
626		- start = sizeof(struct iphdr) + ETH_HLEN;
627		- insert = sizeof(struct iphdr) + ETH_HLEN + 6;
628		- length = ip->tot_len - sizeof(struct iphdr);
629		- break;
630		- default:
631		- break;
632		- }
633		- cur_p->app1 = ((start << 16) \| insert);
634		- cur_p->app2 = csum_tcpudp_magic(ip->saddr, ip->daddr,
635		- length, ip->protocol, 0);
636		- skb->data[insert] = 0;
637		- skb->data[insert + 1] = 0;
	646	+ cur_p->app0 \|= 1; /* TX Checksum Enabled */
	647	+ cur_p->app1 = (csum_start_off << 16) \| csum_index_off;
	648	+ cur_p->app2 = 0; /* initial checksum seed */
638	649	}
	650	+
639	651	cur_p->app0 \|= STS_CTRL_APP0_SOP;
640	652	cur_p->len = skb_headlen(skb);
641	653	cur_p->phys = dma_map_single(ndev->dev.parent, skb->data, skb->len,
...	...	@@ -699,6 +711,15 @@
699	711	skb->protocol = eth_type_trans(skb, ndev);
700	712	skb->ip_summed = CHECKSUM_NONE;
701	713
	714	+ /* if we're doing rx csum offload, set it up */
	715	+ if (((lp->temac_features & TEMAC_FEATURE_RX_CSUM) != 0) &&
	716	+ (skb->protocol == __constant_htons(ETH_P_IP)) &&
	717	+ (skb->len > 64)) {
	718	+
	719	+ skb->csum = cur_p->app3 & 0xFFFF;
	720	+ skb->ip_summed = CHECKSUM_COMPLETE;
	721	+ }
	722	+
702	723	netif_rx(skb);
703	724
704	725	ndev->stats.rx_packets++;
...	...	@@ -883,6 +904,7 @@
883	904	struct temac_local *lp;
884	905	struct net_device *ndev;
885	906	const void *addr;
	907	+ __be32 *p;
886	908	int size, rc = 0;
887	909
888	910	/* Init network device structure */
...	...	@@ -925,6 +947,18 @@
925	947	dev_err(&op->dev, "could not map temac regs.\n");
926	948	goto nodev;
927	949	}
	950	+
	951	+ /* Setup checksum offload, but default to off if not specified */
	952	+ lp->temac_features = 0;
	953	+ p = (__be32 *)of_get_property(op->dev.of_node, "xlnx,txcsum", NULL);
	954	+ if (p && be32_to_cpu(*p)) {
	955	+ lp->temac_features \|= TEMAC_FEATURE_TX_CSUM;
	956	+ /* Can checksum TCP/UDP over IPv4. */
	957	+ ndev->features \|= NETIF_F_IP_CSUM;
	958	+ }
	959	+ p = (__be32 *)of_get_property(op->dev.of_node, "xlnx,rxcsum", NULL);
	960	+ if (p && be32_to_cpu(*p))
	961	+ lp->temac_features \|= TEMAC_FEATURE_RX_CSUM;
928	962
929	963	/* Find the DMA node, map the DMA registers, and decode the DMA IRQs */
930	964	np = of_parse_phandle(op->node, "llink-connected", 0);