[SUNGEM] LLTX support

[SUNGEM] LLTX support

[Eric Lemoine]

[-- Attachment #1: Type: text/plain, Size: 125 bytes --]

The attached patch adds LLTX support to SunGEM NAPI. The tx_lock patch
I just sent must be applied first.

Thanks,

-- 
Eric

[-- Attachment #2: patch-rset-sungem_lltx-2-6-9-rc1 --]
[-- Type: application/octet-stream, Size: 1765 bytes --]

# This is a BitKeeper generated diff -Nru style patch.
#
# ChangeSet
#   2004/09/15 08:32:25+02:00 eric.lemoine@gmail.com 
#   [SUNGEM]: add LLTX support to SunGEM
#   
#   Signed-off-by: Eric Lemoine <eric.lemoine@gmail.com>
# 
# drivers/net/sungem.c
#   2004/09/15 08:32:17+02:00 eric.lemoine@gmail.com +10 -4
#   [SUNGEM]: add LLTX support to SunGEM
#   
#   Signed-off-by: Eric Lemoine <eric.lemoine@gmail.com>
# 
diff -Nru a/drivers/net/sungem.c b/drivers/net/sungem.c
--- a/drivers/net/sungem.c	2004-09-15 08:38:20 +02:00
+++ b/drivers/net/sungem.c	2004-09-15 08:38:20 +02:00
@@ -942,6 +942,7 @@
 	struct gem *gp = dev->priv;
 	int entry;
 	u64 ctrl;
+	unsigned long flags;
 
 	ctrl = 0;
 	if (skb->ip_summed == CHECKSUM_HW) {
@@ -955,12 +956,17 @@
 			(csum_stuff_off << 21));
 	}
 
-	spin_lock_irq(&gp->tx_lock);
+	local_irq_save(flags);
+	if (!spin_trylock(&gp->tx_lock)) {
+		/* Tell upper layer to requeue */
+		local_irq_restore(flags);
+		return -1;
+	}
 
 	/* This is a hard error, log it. */
 	if (TX_BUFFS_AVAIL(gp) <= (skb_shinfo(skb)->nr_frags + 1)) {
 		netif_stop_queue(dev);
-		spin_unlock_irq(&gp->tx_lock);
+		spin_unlock_irqrestore(&gp->tx_lock, flags);
 		printk(KERN_ERR PFX "%s: BUG! Tx Ring full when queue awake!\n",
 		       dev->name);
 		return 1;
@@ -1047,7 +1053,7 @@
 		       dev->name, entry, skb->len);
 	mb();
 	writel(gp->tx_new, gp->regs + TXDMA_KICK);
-	spin_unlock_irq(&gp->tx_lock);
+	spin_unlock_irqrestore(&gp->tx_lock, flags);
 
 	dev->trans_start = jiffies;
 
@@ -2960,7 +2966,7 @@
 	pci_set_drvdata(pdev, dev);
 
 	/* GEM can do it all... */
-	dev->features |= NETIF_F_SG | NETIF_F_HW_CSUM;
+	dev->features |= NETIF_F_SG | NETIF_F_HW_CSUM | NETIF_F_LLTX;
 	if (pci_using_dac)
 		dev->features |= NETIF_F_HIGHDMA;
 

Re: [SUNGEM] LLTX support

[David S. Miller]

On Wed, 15 Sep 2004 08:45:21 +0200
Eric Lemoine <eric.lemoine@gmail.com> wrote:

> The attached patch adds LLTX support to SunGEM NAPI. The tx_lock patch
> I just sent must be applied first.

Also applied, thanks Eric.

I remember you mentioning yesterday that Harald's sungem NAPI
patch had netpoll support, can you cook up a patch that adds
it for current sungem?

Thanks.

Re: [SUNGEM] LLTX support

[Eric Lemoine]

On Wed, 15 Sep 2004 08:35:59 -0700, David S. Miller <davem@davemloft.net> wrote:
> On Wed, 15 Sep 2004 08:45:21 +0200
> Eric Lemoine <eric.lemoine@gmail.com> wrote:
> 
> > The attached patch adds LLTX support to SunGEM NAPI. The tx_lock patch
> > I just sent must be applied first.
> 
> Also applied, thanks Eric.
> 
> I remember you mentioning yesterday that Harald's sungem NAPI
> patch had netpoll support, can you cook up a patch that adds
> it for current sungem?

Can't do it right now but will do it soon...

-- 
Eric

Re: [SUNGEM] LLTX support

[Eric Lemoine]

[-- Attachment #1: Type: text/plain, Size: 647 bytes --]

On Wed, 15 Sep 2004 08:35:59 -0700, David S. Miller <davem@davemloft.net> wrote:
> On Wed, 15 Sep 2004 08:45:21 +0200
> Eric Lemoine <eric.lemoine@gmail.com> wrote:
> 
> > The attached patch adds LLTX support to SunGEM NAPI. The tx_lock patch
> > I just sent must be applied first.
> 
> Also applied, thanks Eric.
> 
> I remember you mentioning yesterday that Harald's sungem NAPI
> patch had netpoll support, can you cook up a patch that adds
> it for current sungem?

Patch attached. Unlike Harald's, this patch doesn't disable_irq before
calling gem_interrupt because gem_interrupt is safe to reentrance.
Hopefully this is correct...

-- 
Eric

[-- Attachment #2: rset-sungem_netpoll-2-6-9-rc1.patch --]
[-- Type: application/octet-stream, Size: 1486 bytes --]

# This is a BitKeeper generated diff -Nru style patch.
#
# ChangeSet
#   2004/09/15 23:37:24+02:00 eric.lemoine@gmail.com 
#   [SUNGEM] netpoll support
#   
#   Signed-off-by: Eric Lemoine <eric.lemoine@gmail.com>
# 
# drivers/net/sungem.c
#   2004/09/15 23:37:15+02:00 eric.lemoine@gmail.com +16 -0
#   [SUNGEM] netpoll support
#   
#   Signed-off-by: Eric Lemoine <eric.lemoine@gmail.com>
# 
diff -Nru a/drivers/net/sungem.c b/drivers/net/sungem.c
--- a/drivers/net/sungem.c	2004-09-15 23:38:12 +02:00
+++ b/drivers/net/sungem.c	2004-09-15 23:38:12 +02:00
@@ -5,6 +5,9 @@
  * 
  * Support for Apple GMAC and assorted PHYs by
  * Benjamin Herrenscmidt (benh@kernel.crashing.org)
+ *
+ * NAPI and NETPOLL support
+ * (C) 2004 by Eric Lemoine (eric.lemoine@gmail.com)
  * 
  * TODO: 
  *  - Get rid of all those nasty mdelay's and replace them
@@ -898,6 +901,16 @@
 	return IRQ_HANDLED;
 }
 
+#ifdef CONFIG_NET_POLL_CONTROLLER
+static void gem_poll_controller(struct net_device *dev)
+{
+	/* gem_interrupt is safe to reentrance so no need
+	 * to disable_irq here.
+	 */
+	gem_interrupt(dev->irq, dev, NULL);
+}
+#endif
+
 static void gem_tx_timeout(struct net_device *dev)
 {
 	struct gem *gp = dev->priv;
@@ -2934,6 +2947,9 @@
 	dev->change_mtu = gem_change_mtu;
 	dev->irq = pdev->irq;
 	dev->dma = 0;
+#ifdef CONFIG_NET_POLL_CONTROLLER
+        dev->poll_controller = gem_poll_controller;
+#endif
 
 	if (register_netdev(dev)) {
 		printk(KERN_ERR PFX "Cannot register net device, "

Re: [SUNGEM] LLTX support

[David S. Miller]

On Wed, 15 Sep 2004 23:43:16 +0200
Eric Lemoine <eric.lemoine@gmail.com> wrote:

> On Wed, 15 Sep 2004 08:35:59 -0700, David S. Miller <davem@davemloft.net> wrote:
> > On Wed, 15 Sep 2004 08:45:21 +0200
> > Eric Lemoine <eric.lemoine@gmail.com> wrote:
> > 
> > > The attached patch adds LLTX support to SunGEM NAPI. The tx_lock patch
> > > I just sent must be applied first.
> > 
> > Also applied, thanks Eric.
> > 
> > I remember you mentioning yesterday that Harald's sungem NAPI
> > patch had netpoll support, can you cook up a patch that adds
> > it for current sungem?
> 
> Patch attached. Unlike Harald's, this patch doesn't disable_irq before
> calling gem_interrupt because gem_interrupt is safe to reentrance.

Looks fine to me.  Patch applied, thanks Eric.