netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* RFC: Convert printks with net_device to dev_<level>
@ 2010-01-29 20:01 Joe Perches
  2010-01-29 20:25 ` Ben Hutchings
  0 siblings, 1 reply; 3+ messages in thread
From: Joe Perches @ 2010-01-29 20:01 UTC (permalink / raw)
  To: netdev; +Cc: David Miller, Greg Kroah-Hartman

I've been playing with a cocci script and some additional
scripts to automate a conversion of printk and pr_<level>
logging messages with a reference to a struct net_device
to dev_<level>(&net_device->dev, ...)

A sample conversion:

-	printk(KERN_INFO "%s: Using MII transceiver %d, status %4.4x.\n",
-	       dev->name, tp->phys[0], tulip_mdio_read(dev, tp->phys[0], 1));
+	dev_info(&dev->dev, "Using MII transceiver %d, status %04x\n",
+		 tp->phys[0], tulip_mdio_read(dev, tp->phys[0], 1));

I submitted conversions for drivers/net/tulip/.
I did not convert any calls with KERN_DEBUG.
http://patchwork.ozlabs.org/patch/43889/

The logging messages are a bit more verbose/complete.
Code size increases a small amount.

Is this sort of conversion useful?
Should more conversions be submitted?
Would it be useful to convert the KERN_DEBUG calls?
Any automated conversion could use:

	printk(KERN_DEBUG "%s: ...", dev->name
	dev_printk(KERN_DEBUG, &dev->dev, "...


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: RFC: Convert printks with net_device to dev_<level>
  2010-01-29 20:01 RFC: Convert printks with net_device to dev_<level> Joe Perches
@ 2010-01-29 20:25 ` Ben Hutchings
  2010-01-30  2:26   ` Joe Perches
  0 siblings, 1 reply; 3+ messages in thread
From: Ben Hutchings @ 2010-01-29 20:25 UTC (permalink / raw)
  To: Joe Perches; +Cc: netdev, David Miller, Greg Kroah-Hartman

On Fri, 2010-01-29 at 12:01 -0800, Joe Perches wrote:
> I've been playing with a cocci script and some additional
> scripts to automate a conversion of printk and pr_<level>
> logging messages with a reference to a struct net_device
> to dev_<level>(&net_device->dev, ...)
> 
> A sample conversion:
> 
> -	printk(KERN_INFO "%s: Using MII transceiver %d, status %4.4x.\n",
> -	       dev->name, tp->phys[0], tulip_mdio_read(dev, tp->phys[0], 1));
> +	dev_info(&dev->dev, "Using MII transceiver %d, status %04x\n",
> +		 tp->phys[0], tulip_mdio_read(dev, tp->phys[0], 1));
[...]

My understanding is that dev_* should be given a bus device (pci_device,
usb_device, ...), not a class device (net_device).  Of course, the net
device name is rather useful as well.  In sfc we print both once the net
device is registered.

It might be useful to add netdev_* print macros along the lines of:

#define netdev_name(dev)					\
	(((dev)->reg_state == NETREG_REGISTERED) ? (dev)->name : "")

#define netdev_info(dev, fmt, ...)				\
	dev_info((dev)->dev.parent, "%s " fmt, 			\
		 netdev_name(dev), __VA_ARGS__)

Ben.

-- 
Ben Hutchings, Senior Software Engineer, Solarflare Communications
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: RFC: Convert printks with net_device to dev_<level>
  2010-01-29 20:25 ` Ben Hutchings
@ 2010-01-30  2:26   ` Joe Perches
  0 siblings, 0 replies; 3+ messages in thread
From: Joe Perches @ 2010-01-30  2:26 UTC (permalink / raw)
  To: Ben Hutchings; +Cc: netdev, David Miller, Greg Kroah-Hartman

On Fri, 2010-01-29 at 20:25 +0000, Ben Hutchings wrote:
> On Fri, 2010-01-29 at 12:01 -0800, Joe Perches wrote:
> > I've been playing with a cocci script and some additional
> > scripts to automate a conversion of printk and pr_<level>
> > logging messages with a reference to a struct net_device
> > to dev_<level>(&net_device->dev, ...)
> > 
> > A sample conversion:
> > 
> > -   printk(KERN_INFO "%s: Using MII transceiver %d, status %4.4x.\n",
> > -          dev->name, tp->phys[0], tulip_mdio_read(dev, tp->phys[0], 1));
> > +   dev_info(&dev->dev, "Using MII transceiver %d, status %04x\n",
> > +            tp->phys[0], tulip_mdio_read(dev, tp->phys[0], 1));
> [...]
> 
> My understanding is that dev_* should be given a bus device (pci_device,
> usb_device, ...), not a class device (net_device).  Of course, the net
> device name is rather useful as well.  In sfc we print both once the net
> device is registered.
> 
> It might be useful to add netdev_* print macros along the lines of:
> 
> #define netdev_name(dev)                                      \
>       (((dev)->reg_state == NETREG_REGISTERED) ? (dev)->name : "")
> 
> #define netdev_info(dev, fmt, ...)                            \
>       dev_info((dev)->dev.parent, "%s " fmt,                  \
>                netdev_name(dev), __VA_ARGS__)

Something like this?

 include/linux/netdevice.h |   59 +++++++++++++++++++++++++++++++++++++++++++++
 1 files changed, 59 insertions(+), 0 deletions(-)

diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 93a32a5..1997b2d 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -2077,6 +2077,65 @@ static inline u32 dev_ethtool_get_flags(struct net_device *dev)
 		return 0;
 	return dev->ethtool_ops->get_flags(dev);
 }
+
+/* debugging and troubleshooting/diagnostic helpers. */
+
+static inline const char *netdev_name(const struct net_device *dev)
+{
+	if (dev->reg_state != NETREG_REGISTERED)
+		return "";
+	return dev->name;
+}
+
+#define netdev_printk(level, dev, format, arg...)	\
+	dev_printk(level, (dev)->dev.parent,		\
+		   "%s:" format, netdev_name(dev), ##arg)
+
+#define netdev_emerg(dev, format, arg...)		\
+	netdev_printk(KERN_EMERG, dev, format, ##arg)
+#define netdev_alert(dev, format, arg...)		\
+	netdev_printk(KERN_ALERT, dev, format, ##arg)
+#define netdev_crit(dev, format, arg...)		\
+	netdev_printk(KERN_CRIT, dev, format, ##arg)
+#define netdev_err(dev, format, arg...)		\
+	netdev_printk(KERN_ERR, dev, format, ##arg)
+#define netdev_warn(dev, format, arg...)		\
+	netdev_printk(KERN_WARNING, dev, format, ##arg)
+#define netdev_notice(dev, format, arg...)		\
+	netdev_printk(KERN_NOTICE, dev, format, ##arg)
+#define netdev_info(dev, format, arg...)		\
+	netdev_printk(KERN_INFO, dev, format, ##arg)
+
+#if defined(DEBUG)
+#define netdev_dbg(dev, format, arg...)			\
+	netdev_printk(KERN_DEBUG, dev, format, ##arg)
+#elif defined(CONFIG_DYNAMIC_DEBUG)
+#define netdev_dbg(dev, format, arg...) do {			\
+	dynamic_dev_dbg((dev)->dev.parent,			\
+			"%s:" format, netdev_name(dev), ##arg); \
+	} while (0)
+#else
+#define netdev_dbg(dev, format, arg...)		\
+	({ if (0) netdev_printk(KERN_DEBUG, dev, format, ##arg); 0; })
+#endif
+
+#ifdef VERBOSE_DEBUG
+#define netdev_vdbg	netdev_dbg
+#else
+
+#define netdev_vdbg(dev, format, arg...)		\
+	({ if (0) netdev_printk(KERN_DEBUG, dev, format, ##arg); 0; })
+#endif
+
+/*
+ * netdev_WARN() acts like dev_printk(), but with the key difference
+ * of using a WARN/WARN_ON to get the message out, including the
+ * file/line information and a backtrace.
+ */
+#define netdev_WARN(dev, format, arg...) \
+	WARN(1, "netdevice: %s\n" format, netdev_name(dev), ##arg);
+
+
 #endif /* __KERNEL__ */
 
 #endif	/* _LINUX_NETDEVICE_H */



^ permalink raw reply related	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2010-01-30  2:26 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2010-01-29 20:01 RFC: Convert printks with net_device to dev_<level> Joe Perches
2010-01-29 20:25 ` Ben Hutchings
2010-01-30  2:26   ` Joe Perches

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).