On Tue, 2012-05-22 at 11:26 -0700, David Daney wrote:

On 05/22/2012 11:17 AM, Joe Perches wrote:

quoted

On Tue, 2012-05-22 at 10:59 -0700, David Daney wrote:

quoted

From: David Daney<redacted>

trivia:

As long as we are splitting hairs...

and zooming in and enhancing...

quoted

quoted

diff --git a/drivers/net/phy/bcm87xx.c b/drivers/net/phy/bcm87xx.c

[]

quoted

@@ -0,0 +1,237 @@

quoted

+static int bcm87xx_of_reg_init(struct phy_device *phydev)
+{
+	const __be32 *paddr;
+	int len, i, ret;
+
+	if (!phydev->dev.of_node)
+		return 0;
+
+	paddr = of_get_property(phydev->dev.of_node,
+				"broadcom,c45-reg-init",&len);
+	if (!paddr || len<  (4 * sizeof(*paddr)))
+		return 0;
+
+	ret = 0;
+	len /= sizeof(*paddr);
+	for (i = 0; i<  len - 3; i += 4) {
+		u16 devid = be32_to_cpup(paddr + i);
+		u16 reg = be32_to_cpup(paddr + i + 1);
+		u16 mask = be32_to_cpup(paddr + i + 2);
+		u16 val_bits = be32_to_cpup(paddr + i + 3);
+		int val;

These might read better as

	len /= 4;

Where did the magic value of 4 come from?

equivalence to the original for loop

	for (i = 0; i < len - 3; i += 4) {

quoted

	for (i = 0; i<  len; i++) {

quoted

		u16 devid	= be32_to_cpu(*paddr++);
		u16 reg		= be32_to_cpu(*paddr++);
		u16 mask	= be32_to_cpu(*paddr++);
		u16 val_bits	= be32_to_cpu(*paddr++);

Is the main problem that they didn't align, or that the index was 
explicit instead of implicit?

There's no real problem, it's just that
i++, be32_to_cpu and *addr++ is a bit
more common and perhaps more easily read.

The alignment's just a visual nicety.

Ignore it if you choose.

cheers, Joe