[U-Boot-Users] PPC4xx enet crossover

[U-Boot-Users] PPC4xx enet crossover

[Mike Nuss]

I'm using a PPC440EPx with a Marvell PHY (same as on Sequoia). If I connect to a PC with a crossover Ethernet cable, U-Boot sees a link but the PC does not. The PC (running Linux) reports "Autonegotiation failed (remote fault)" and ethtool reports no link.

However if I let the PowerPC boot up into Linux, all of a sudden the PC sees link and pings start working. In fact, it works in Linux with a standard Ethernet cable, too (gigabit ports should be autosensing).

It seems U-Boot is not initializing the PHY correctly to handle a direct connection, but Linux is... any ideas? Everything works fine through a switch.

Thanks,
Mike

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.denx.de/pipermail/u-boot/attachments/20080303/a9450161/attachment.htm 

[U-Boot-Users] PPC4xx enet crossover

[Stefan Roese]

On Monday 03 March 2008, Mike Nuss wrote:
> I'm using a PPC440EPx with a Marvell PHY (same as on Sequoia). If I connect
> to a PC with a crossover Ethernet cable, U-Boot sees a link but the PC does
> not. The PC (running Linux) reports "Autonegotiation failed (remote fault)"
> and ethtool reports no link.

Did you really try to communicate with the PC in this situation? Please note 
that the PHY is only initialized upon ethernet usage. So if you just turn on 
the Sequoia and stop on the U-Boot prompt, no real PHY configuration will 
happen.

> However if I let the PowerPC boot up into Linux, all of a sudden the PC
> sees link and pings start working. In fact, it works in Linux with a
> standard Ethernet cable, too (gigabit ports should be autosensing).

Understood. You should not need a crossover cable, since the PHY supports 
automatic MDI/MDIX crossover at all speeds of operation.

Best regards,
Stefan

=====================================================================
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: +49-8142-66989-0 Fax: +49-8142-66989-80  Email: office at denx.de
=====================================================================

[U-Boot-Users] PPC4xx enet crossover

[Mike Nuss]

Stefan Roese wrote:
> On Monday 03 March 2008, Mike Nuss wrote:
> > I'm using a PPC440EPx with a Marvell PHY (same as on 
> Sequoia). If I connect
> > to a PC with a crossover Ethernet cable, U-Boot sees a link 
> but the PC does
> > not. The PC (running Linux) reports "Autonegotiation failed 
> (remote fault)"
> > and ethtool reports no link.
> 
> Did you really try to communicate with the PC in this 
> situation? Please note 
> that the PHY is only initialized upon ethernet usage. So if 
> you just turn on 
> the Sequoia and stop on the U-Boot prompt, no real PHY 
> configuration will 
> happen.

Yes, I set the ipaddr variable and tried to do a ping. U-Boot reported the PHY autonegotiation succeeding but reported the remote host as down, and the PC logged the autonegotiation error.

Oddly enough, I tried a different PC (with a different NIC) and it worked there. But I'm reluctant to blame the NIC on the first machine, since it worked fine once Linux was running on the PowerPC.

Mike

[U-Boot-Users] PPC4xx enet crossover

[Stefan Roese]

On Monday 03 March 2008, Mike Nuss wrote:
> > Did you really try to communicate with the PC in this
> > situation? Please note
> > that the PHY is only initialized upon ethernet usage. So if
> > you just turn on
> > the Sequoia and stop on the U-Boot prompt, no real PHY
> > configuration will
> > happen.
>
> Yes, I set the ipaddr variable and tried to do a ping. U-Boot reported the
> PHY autonegotiation succeeding but reported the remote host as down, and
> the PC logged the autonegotiation error.

Hmmm. Then it really could be a problem in the U-Boot PHY configuration.

> Oddly enough, I tried a different PC (with a different NIC) and it worked
> there. But I'm reluctant to blame the NIC on the first machine, since it
> worked fine once Linux was running on the PowerPC.

Yes, sounds like a U-Boot related PHY init problem. Perhaps you could take a 
look at how the PHY registers are configured in U-Boot and in Linux and 
search for differences.

Best regards,
Stefan

=====================================================================
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: +49-8142-66989-0 Fax: +49-8142-66989-80  Email: office at denx.de
=====================================================================

[U-Boot-Users] PPC4xx enet crossover

[Wolfgang Denk]

In message <2C7DE72B9BD00F44BAECA5B0CBB87395072980@hermes.terascala.com> you wrote:
>
> Oddly enough, I tried a different PC (with a different NIC) and it worked there. But I'm reluctant to blame the NIC on the first machine, since it worked fine once Linux was running on the PowerPC.

[Please restrict your line length to < 70 characters or so.]

It seems every now and then a report pops up that the  Sequoia  board
(or  now on equivalent hardware) has problems connecting to a Gigabit
capable network interface. In our case,  it's  when  the  Sequoia  is
attached to a HP Procurve switch.

AMCC Technical Support was not able  to  reproduce  the  problem,  so
officially it does not exist.

Could you please let me know which exact network card you  are  using
on  your PC, and which Linux distro / lenrel version / driver version
you are running on it?

Best regards,

Wolfgang Denk

-- 
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd@denx.de
Hacking's just another word for nothing left to kludge.

[U-Boot-Users] PPC4xx enet crossover

[Mike Nuss]

Wolfgang Denk wrote:

> It seems every now and then a report pops up that the  Sequoia  board
> (or  now on equivalent hardware) has problems connecting to a Gigabit
> capable network interface. In our case,  it's  when  the  Sequoia  is
> attached to a HP Procurve switch.

There were differences in the PHY setup between Linux and U-Boot.

This patch fixes the problem. Apparently "remote fault" is being set,
which signals to some devices (on the other end of the cable) that a
fault has occurred, while other devices ignore it. I believe the RF bit
was causing the issue, but I removed T4 also, to match up with Linux.

Mike

diff --git a/cpu/ppc4xx/miiphy.c b/cpu/ppc4xx/miiphy.c
index 4216f0b..f48e68d 100644
--- a/cpu/ppc4xx/miiphy.c
+++ b/cpu/ppc4xx/miiphy.c
@@ -143,7 +143,7 @@ int phy_setup_aneg (char *devname, unsigned char addr)
        u16 adv;
 
        miiphy_read (devname, addr, PHY_ANAR, &adv);
-       adv |= (PHY_ANLPAR_ACK | PHY_ANLPAR_RF | PHY_ANLPAR_T4 |
+       adv |= (PHY_ANLPAR_ACK |
                PHY_ANLPAR_TXFD | PHY_ANLPAR_TX | PHY_ANLPAR_10FD |
                PHY_ANLPAR_10);
        miiphy_write (devname, addr, PHY_ANAR, adv);

[U-Boot-Users] PPC4xx enet crossover

[Wolfgang Denk]

Dear Mike,

in message <2C7DE72B9BD00F44BAECA5B0CBB873950729A0@hermes.terascala.com> you wrote:
> 
> There were differences in the PHY setup between Linux and U-Boot.
> 
> This patch fixes the problem. Apparently "remote fault" is being set,
> which signals to some devices (on the other end of the cable) that a
> fault has occurred, while other devices ignore it. I believe the RF bit
> was causing the issue, but I removed T4 also, to match up with Linux.

Excellent! thanks a lot.

Stefan, can you please test this tomorrow morning in the VL setup? If
possible, I would like to see this go into the 1.3.2 release.

Best regards,

Wolfgang Denk

-- 
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
When choosing between two evils, I always like to take the  one  I've
never tried before.                     -- Mae West, "Klondike Annie"

[U-Boot-Users] PPC4xx enet crossover

[Wolfgang Denk]

In message <2C7DE72B9BD00F44BAECA5B0CBB873950729A0@hermes.terascala.com> you wrote:
>
> There were differences in the PHY setup between Linux and U-Boot.
> 
> This patch fixes the problem. Apparently "remote fault" is being set,
> which signals to some devices (on the other end of the cable) that a
> fault has occurred, while other devices ignore it. I believe the RF bit
> was causing the issue, but I removed T4 also, to match up with Linux.

I verified that it fixes the problem with our setup (Sequoia board
with HP Procurve switch), too. I still see a timeout for the very
first packet after a reset, but after that it's working fine.

Thanks.

Could you just please add your Signed-off-by: line?

Best regards,

Wolfgang Denk

-- 
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
At the source of every error which is blamed on the computer you will
find at least two human errors, including the error of blaming it  on
the computer.