[2/3] via-rhine: de-isolate PHY

[2/3] via-rhine: de-isolate PHY

[Roger Luethi]

PHYs may come up isolated. Make sure we can send data to them. This code
section needs a clean-up, but I prefer to merge this fix in isolation.

Report and suggested fix by Tam, Ming Dat (Tommy).

Signed-off-by: Roger Luethi <rl@hellgate.ch>

--- linux-2.6.8-rc3-mm1/drivers/net/via-rhine.c.01	2004-08-08 12:36:03.440855262 +0200
+++ linux-2.6.8-rc3-mm1/drivers/net/via-rhine.c	2004-08-08 13:15:24.527527919 +0200
@@ -896,7 +896,10 @@ static int __devinit rhine_init_one(stru
 	pci_set_drvdata(pdev, dev);
 
 	{
+		u16 mii_cmd;
 		int mii_status = mdio_read(dev, phy_id, 1);
+		mii_cmd = mdio_read(dev, phy_id, MII_BMCR) & ~BMCR_ISOLATE;
+		mdio_write(dev, phy_id, MII_BMCR, mii_cmd);
 		if (mii_status != 0xffff && mii_status != 0x0000) {
 			rp->mii_if.advertising = mdio_read(dev, phy_id, 4);
 			printk(KERN_INFO "%s: MII PHY found at address "

Re: [2/3] via-rhine: de-isolate PHY

[Manfred Spraul]

Roger wrote:

> PHYs may come up isolated. Make sure we can send data to them. This code
> section needs a clean-up, but I prefer to merge this fix in isolation.
>
What was the phyid value for the isolated PHYs?

I know that PHYs go into isolate mode if the startup id is wired to 0, 
but I haven't figured out what's necessary to initialize them: Just 
clear the isolate bit or is it necessary to set the id to a nonzero value.

--
    Manfred

Re: [2/3] via-rhine: de-isolate PHY

[Roger Luethi]

[ BCCed reporter in case he wants to weigh in on the issue ]

On Sun, 08 Aug 2004 21:53:57 +0200, Manfred Spraul wrote:
> Roger wrote:
> 
> >PHYs may come up isolated. Make sure we can send data to them. This code
> >section needs a clean-up, but I prefer to merge this fix in isolation.
> >
> What was the phyid value for the isolated PHYs?

I suspect it was 0, because another suggestion in the same message was to
start scanning at 0 rather than 1 (obsolete with current via-rhine code).

> I know that PHYs go into isolate mode if the startup id is wired to 0, 

Wouldn't that be s/go/can go/ ?

> but I haven't figured out what's necessary to initialize them: Just 
> clear the isolate bit or is it necessary to set the id to a nonzero value.

The proposed fix apparently worked for the PHY in question (LSI Logic
80225).

Roger

Re: [2/3] via-rhine: de-isolate PHY

[Manfred Spraul]

Roger Luethi wrote:

>  
>
>>I know that PHYs go into isolate mode if the startup id is wired to 0, 
>>    
>>
>
>Wouldn't that be s/go/can go/ ?
>
>  
>
I don't have the MII standard, my knowledge is from the DP83840A specs:
The pin description contains a section about the phy ids:
During power up five pins are latched to determine the initial phy address.
Then the following sentence in bold: "An address selection of all zeros 
(00000) will result in a PHY isolation condition".

I've reread the DP specs and I now think that your current patch is 
sufficient:
The isolate state is independant from the phy address - a non-zero phy 
can be in isolate mode and the phy zero can be non-isolated. The phy id 
just sets the power-up value of the isolate bit: 0 means start isolated, 
non-zero means start non-isolated.

If this is really true then handling phy 0 is trivial:
First scan 1-31. If nothing found: try 0. If a phy is found: clear the 
isolate bit and then use phy 0.

--
    Manfred

Re: [2/3] via-rhine: de-isolate PHY

[Roger Luethi]

On Sun, 08 Aug 2004 23:04:06 +0200, Manfred Spraul wrote:
> Roger Luethi wrote:
> >>I know that PHYs go into isolate mode if the startup id is wired to 0, 
> >
> >Wouldn't that be s/go/can go/ ?
> >
> I don't have the MII standard, my knowledge is from the DP83840A specs:
> The pin description contains a section about the phy ids:
> During power up five pins are latched to determine the initial phy address.
> Then the following sentence in bold: "An address selection of all zeros 
> (00000) will result in a PHY isolation condition".

I suppose all PHYs do that. Even if they don't, though, I should be
safe as long as I de-isolate unconditionally (instead of testing for
phy_id==0).

> I've reread the DP specs and I now think that your current patch is 
> sufficient:
> The isolate state is independant from the phy address - a non-zero phy 
> can be in isolate mode and the phy zero can be non-isolated. The phy id 

Stands to reason. A PHY that can't get out of isolation wouldn't be
very useful.

> If this is really true then handling phy 0 is trivial:
> First scan 1-31. If nothing found: try 0. If a phy is found: clear the 
> isolate bit and then use phy 0.

Makes sense. The Rhine is actually pretty neat in that regard, I've
been able to drop the PHY scanning entirely.

Roger