gianfar generic PHY assumptions.

gianfar generic PHY assumptions.

[David Updegraff]

[-- Attachment #1: Type: text/plain, Size: 324 bytes --]

Hi.

This patch lets gianfar work on 83xx (or 85xx??) boards that have 
Generic PHYs on them that do _NOT_ happen to have an IRQ line wired to 
external_15.  Surely reasonable that if we can't ID the PHY that its 
probably not irq-wired the Freescale-ADS "way"...

It was critical for us; perhaps usefull for others.

-dbu.

[-- Attachment #2: x --]
[-- Type: text/plain, Size: 1071 bytes --]

diff --git a/drivers/net/gianfar_phy.c b/drivers/net/gianfar_phy.c
index 7c965f2..1c08bd0 100644
--- a/drivers/net/gianfar_phy.c
+++ b/drivers/net/gianfar_phy.c
@@ -234,6 +234,19 @@ static int marvell_config_aneg(struct gf
 
 	return 0;
 }
+/* Init for GENERIC: just here to make sure the gainfar driver does not assume
+ * that it has an interrupt. Since don't even know what model it is, seems 
+ * unlikely that they have bothered to wire in an IRQ line exactly like the 
+ * the platform/ files have assumed.
+ */
+static int genmii_init(struct gfar_mii_info *mii_info)
+{
+	struct gfar_private *priv = netdev_priv(mii_info->dev);
+	if (priv)
+		priv->einfo->board_flags &= ~FSL_GIANFAR_BRD_HAS_PHY_INTR;
+	return 0;
+}
+
 static int genmii_config_aneg(struct gfar_mii_info *mii_info)
 {
 	if (mii_info->autoneg) {
@@ -585,6 +598,7 @@ static struct phy_info phy_info_genmii= 
 	.phy_id_mask	= 0x00000000,
 	.name		= "Generic MII",
 	.features	= MII_BASIC_FEATURES,
+	.init		= genmii_init,
 	.config_aneg	= genmii_config_aneg,
 	.read_status	= genmii_read_status,
 };

Re: gianfar generic PHY assumptions.

[Kumar Gala]

On Thu, 17 Nov 2005, David Updegraff wrote:
> Hi.
> 
> This patch lets gianfar work on 83xx (or 85xx??) boards that have 
> Generic PHYs on them that do _NOT_ happen to have an IRQ line wired to 
> external_15.  Surely reasonable that if we can't ID the PHY that its 
> probably not irq-wired the Freescale-ADS "way"...
> 
> It was critical for us; perhaps usefull for others.


The way PHYs are handle in 2.6.15 is completely different and the patch 
is not valid against the current tree.

- kumar

Re: gianfar generic PHY assumptions.

[Matt Porter]

On Thu, Nov 17, 2005 at 03:50:59PM -0600, David Updegraff wrote:
> Hi.
> 
> This patch lets gianfar work on 83xx (or 85xx??) boards that have 
> Generic PHYs on them that do _NOT_ happen to have an IRQ line wired to 
> external_15.  Surely reasonable that if we can't ID the PHY that its 
> probably not irq-wired the Freescale-ADS "way"...
> 
> It was critical for us; perhaps usefull for others.

Aside from the fact that the patch doesn't apply against the
current kernel due to the change of gianfar to use the phy
layer...why wouldn't you just set the board_flags field
appropriately in your board port? If you don't use a phy interrupt
then don't set the flag claiming that you do.  stx_gp3.c doesn't
make use of the phy interrupt functionality, as an example.

I looked at an older kernel and I'm not sure where you see external
15 being used. It seems to be EXT5 on all the Fscale boards.

-Matt