[PATCH] [NET] uli526x: add support for netpoll

[PATCH] [NET] uli526x: add support for netpoll

[Anton Vorontsov]

This patch adds netpoll support for the uli526x ethernet driver --
simply call the interrupt handler for polling.

To do this without disable_irq()/enable_irq() pair we should fully
protect the handler. Luckily, it's already using irqsave spinlock,
the only unprotected place is interrupts re-enabling write. It was
safe to re-enable interrupts without holding the spinlock, but with
netpoll possibility now it doesn't seem so.

Patch was tested using netconsole and KGDBoE.

Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
---
 drivers/net/tulip/uli526x.c |   14 +++++++++++++-
 1 files changed, 13 insertions(+), 1 deletions(-)

diff --git a/drivers/net/tulip/uli526x.c b/drivers/net/tulip/uli526x.c
index 2511ca7..6a04a95 100644
--- a/drivers/net/tulip/uli526x.c
+++ b/drivers/net/tulip/uli526x.c
@@ -225,6 +225,7 @@ static void uli526x_set_filter_mode(struct net_device *);
 static const struct ethtool_ops netdev_ethtool_ops;
 static u16 read_srom_word(long, int);
 static irqreturn_t uli526x_interrupt(int, void *);
+static void uli526x_poll(struct net_device *dev);
 static void uli526x_descriptor_init(struct uli526x_board_info *, unsigned long);
 static void allocate_rx_buffer(struct uli526x_board_info *);
 static void update_cr6(u32, unsigned long);
@@ -339,6 +340,9 @@ static int __devinit uli526x_init_one (struct pci_dev *pdev,
 	dev->get_stats = &uli526x_get_stats;
 	dev->set_multicast_list = &uli526x_set_filter_mode;
 	dev->ethtool_ops = &netdev_ethtool_ops;
+#ifdef CONFIG_NET_POLL_CONTROLLER
+	dev->poll_controller = &uli526x_poll;
+#endif
 	spin_lock_init(&db->lock);
 
 
@@ -681,8 +685,9 @@ static irqreturn_t uli526x_interrupt(int irq, void *dev_id)
 	db->cr5_data = inl(ioaddr + DCR5);
 	outl(db->cr5_data, ioaddr + DCR5);
 	if ( !(db->cr5_data & 0x180c1) ) {
-		spin_unlock_irqrestore(&db->lock, flags);
+		/* Restore CR7 to enable interrupt mask */
 		outl(db->cr7_data, ioaddr + DCR7);
+		spin_unlock_irqrestore(&db->lock, flags);
 		return IRQ_HANDLED;
 	}
 
@@ -715,6 +720,13 @@ static irqreturn_t uli526x_interrupt(int irq, void *dev_id)
 	return IRQ_HANDLED;
 }
 
+#ifdef CONFIG_NET_POLL_CONTROLLER
+static void uli526x_poll(struct net_device *dev)
+{
+	/* ISR grabs the irqsave lock, so this should be safe */
+	uli526x_interrupt(dev->irq, dev);
+}
+#endif
 
 /*
  *	Free TX resource after TX complete
-- 
1.5.5.1

Re: [PATCH] [NET] uli526x: add support for netpoll

[Florian Fainelli]

Hello Anton,

Le Thursday 15 May 2008 17:12:55 Anton Vorontsov, vous avez écrit :
> +#ifdef CONFIG_NET_POLL_CONTROLLER
> +static void uli526x_poll(struct net_device *dev)
> +{
> +	/* ISR grabs the irqsave lock, so this should be safe */
> +	uli526x_interrupt(dev->irq, dev);
> +}
> +#endif

You do not need to wrap this function into the ifdef, since setting the poll 
callback is between ifdef. Otherwise patch looks good.
-- 
Cordialement, Florian Fainelli
------------------------------

Re: [PATCH] [NET] uli526x: add support for netpoll

[Anton Vorontsov]

On Thu, May 15, 2008 at 10:51:23PM +0200, Florian Fainelli wrote:
> Hello Anton,
> 
> Le Thursday 15 May 2008 17:12:55 Anton Vorontsov, vous avez écrit :
> > +#ifdef CONFIG_NET_POLL_CONTROLLER
> > +static void uli526x_poll(struct net_device *dev)
> > +{
> > +	/* ISR grabs the irqsave lock, so this should be safe */
> > +	uli526x_interrupt(dev->irq, dev);
> > +}
> > +#endif
> 
> You do not need to wrap this function into the ifdef, since setting the poll 
> callback is between ifdef. Otherwise patch looks good.

Nope, I need this #ifdef, otherwise gcc will warn about unused function
if netpoll is not selected.

Of course, I could do __maybe_unused, but #ifdef is what most (all?)
drivers are doing.

-- 
Anton Vorontsov
email: cbouatmailru@gmail.com
irc://irc.freenode.net/bd2

[PATCH v2] [NET] uli526x: add support for netpoll

[Anton Vorontsov]

This patch adds netpoll support for the uli526x ethernet driver --
simply call the interrupt handler for polling.

To do this without disable_irq()/enable_irq() pair we should fully
protect the handler. Luckily, it's already using irqsave spinlock,
the only unprotected place is interrupts re-enabling write. It was
safe to re-enable interrupts without holding the spinlock, but with
netpoll possibility now it doesn't seem so.

Patch was tested using netconsole and KGDBoE.

Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
---

On Fri, May 16, 2008 at 03:13:49PM +0400, Anton Vorontsov wrote:
> On Thu, May 15, 2008 at 10:51:23PM +0200, Florian Fainelli wrote:
> > Hello Anton,
> > 
> > Le Thursday 15 May 2008 17:12:55 Anton Vorontsov, vous avez écrit :
> > > +#ifdef CONFIG_NET_POLL_CONTROLLER
> > > +static void uli526x_poll(struct net_device *dev)
> > > +{
> > > +	/* ISR grabs the irqsave lock, so this should be safe */
> > > +	uli526x_interrupt(dev->irq, dev);
> > > +}
> > > +#endif
> > 
> > You do not need to wrap this function into the ifdef, since setting the poll 
> > callback is between ifdef. Otherwise patch looks good.
> 
> Nope, I need this #ifdef, otherwise gcc will warn about unused function
> if netpoll is not selected.

Even more, I need #ifdef the declaration, because gcc will warn too. :-)
Oops. Here is v2.

 drivers/net/tulip/uli526x.c |   16 +++++++++++++++-
 1 files changed, 15 insertions(+), 1 deletions(-)

diff --git a/drivers/net/tulip/uli526x.c b/drivers/net/tulip/uli526x.c
index 2511ca7..e9e6286 100644
--- a/drivers/net/tulip/uli526x.c
+++ b/drivers/net/tulip/uli526x.c
@@ -225,6 +225,9 @@ static void uli526x_set_filter_mode(struct net_device *);
 static const struct ethtool_ops netdev_ethtool_ops;
 static u16 read_srom_word(long, int);
 static irqreturn_t uli526x_interrupt(int, void *);
+#ifdef CONFIG_NET_POLL_CONTROLLER
+static void uli526x_poll(struct net_device *dev);
+#endif
 static void uli526x_descriptor_init(struct uli526x_board_info *, unsigned long);
 static void allocate_rx_buffer(struct uli526x_board_info *);
 static void update_cr6(u32, unsigned long);
@@ -339,6 +342,9 @@ static int __devinit uli526x_init_one (struct pci_dev *pdev,
 	dev->get_stats = &uli526x_get_stats;
 	dev->set_multicast_list = &uli526x_set_filter_mode;
 	dev->ethtool_ops = &netdev_ethtool_ops;
+#ifdef CONFIG_NET_POLL_CONTROLLER
+	dev->poll_controller = &uli526x_poll;
+#endif
 	spin_lock_init(&db->lock);
 
 
@@ -681,8 +687,9 @@ static irqreturn_t uli526x_interrupt(int irq, void *dev_id)
 	db->cr5_data = inl(ioaddr + DCR5);
 	outl(db->cr5_data, ioaddr + DCR5);
 	if ( !(db->cr5_data & 0x180c1) ) {
-		spin_unlock_irqrestore(&db->lock, flags);
+		/* Restore CR7 to enable interrupt mask */
 		outl(db->cr7_data, ioaddr + DCR7);
+		spin_unlock_irqrestore(&db->lock, flags);
 		return IRQ_HANDLED;
 	}
 
@@ -715,6 +722,13 @@ static irqreturn_t uli526x_interrupt(int irq, void *dev_id)
 	return IRQ_HANDLED;
 }
 
+#ifdef CONFIG_NET_POLL_CONTROLLER
+static void uli526x_poll(struct net_device *dev)
+{
+	/* ISR grabs the irqsave lock, so this should be safe */
+	uli526x_interrupt(dev->irq, dev);
+}
+#endif
 
 /*
  *	Free TX resource after TX complete
-- 
1.5.5.1

Re: [PATCH v2] [NET] uli526x: add support for netpoll

[Jeff Garzik]

Anton Vorontsov wrote:
> This patch adds netpoll support for the uli526x ethernet driver --
> simply call the interrupt handler for polling.
> 
> To do this without disable_irq()/enable_irq() pair we should fully
> protect the handler. Luckily, it's already using irqsave spinlock,
> the only unprotected place is interrupts re-enabling write. It was
> safe to re-enable interrupts without holding the spinlock, but with
> netpoll possibility now it doesn't seem so.
> 
> Patch was tested using netconsole and KGDBoE.
> 
> Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>

applied