smsc9220 with omap3

smsc9220 with omap3

[Mike Rapoport]

Hi,
I'm trying to enable smsc9220 chip with OMAP3EVM board. Till now I was able to
get as far as "eth1: smsc911x_mii_probe: Failed Loop Back Test" error message
after a bunch of "eth1: smsc911x_phy_check_loopbackpkt: Failed to transmit
during loopback test".
Does anybody have a clue what's going wrong?

-- 
Sincerely yours,
Mike.


-------------------------------------------------------------------
List admin: http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm-kernel
FAQ:        http://www.arm.linux.org.uk/mailinglists/faq.php
Etiquette:  http://www.arm.linux.org.uk/mailinglists/etiquette.php

Re: smsc9220 with omap3

[Daniel Mack]

On Tue, Jun 23, 2009 at 07:09:22PM +0300, Mike Rapoport wrote:
> I'm trying to enable smsc9220 chip with OMAP3EVM board. Till now I was able to
> get as far as "eth1: smsc911x_mii_probe: Failed Loop Back Test" error message
> after a bunch of "eth1: smsc911x_phy_check_loopbackpkt: Failed to transmit
> during loopback test".
> Does anybody have a clue what's going wrong?

We're using that chip connected to a PXA300 and it works well, so I
wouldn't suspect the driver. Does it work from any other scenario like
the bootloader? And I don't know OMAP, but it might be helpful to others
if you posted your platform_data config.

Daniel

Re: smsc9220 with omap3

[Mike Rapoport]

Daniel Mack wrote:
> On Tue, Jun 23, 2009 at 07:09:22PM +0300, Mike Rapoport wrote:
>> I'm trying to enable smsc9220 chip with OMAP3EVM board. Till now I was able to
>> get as far as "eth1: smsc911x_mii_probe: Failed Loop Back Test" error message
>> after a bunch of "eth1: smsc911x_phy_check_loopbackpkt: Failed to transmit
>> during loopback test".
>> Does anybody have a clue what's going wrong?
> 
> We're using that chip connected to a PXA300 and it works well, so I
> wouldn't suspect the driver. Does it work from any other scenario like
> the bootloader? And I don't know OMAP, but it might be helpful to others
> if you posted your platform_data config.

The platform_data I use is:
static struct smsc911x_platform_config smsc911x_config = {
	.irq_polarity = SMSC911X_IRQ_POLARITY_ACTIVE_LOW,
	.irq_type = SMSC911X_IRQ_TYPE_PUSH_PULL,
	.flags = SMSC911X_USE_32BIT | SMSC911X_FORCE_INTERNAL_PHY,
};

The chip is properly detected, and if I tweak the loopback test to return 0, the
 driver does not complain any more. Moreover, it seems that RX work, at least
ifconfig reports non-zero values for RX packets. Still, there is no even single
packet transmitted from the smsc9220 :(
I suspect that there's some problem with the hardware, and probably someone
encountered similar problems and may have found a solution.

> Daniel
> 

-- 
Sincerely yours,
Mike.

Re: smsc9220 with omap3

[Daniel Mack]

On Wed, Jun 24, 2009 at 05:52:16PM +0300, Mike Rapoport wrote:
> > We're using that chip connected to a PXA300 and it works well, so I
> > wouldn't suspect the driver. Does it work from any other scenario like
> > the bootloader? And I don't know OMAP, but it might be helpful to others
> > if you posted your platform_data config.
> 
> The platform_data I use is:
> static struct smsc911x_platform_config smsc911x_config = {
> 	.irq_polarity = SMSC911X_IRQ_POLARITY_ACTIVE_LOW,
> 	.irq_type = SMSC911X_IRQ_TYPE_PUSH_PULL,
> 	.flags = SMSC911X_USE_32BIT | SMSC911X_FORCE_INTERNAL_PHY,
> };
> 
> The chip is properly detected, and if I tweak the loopback test to return 0, the
>  driver does not complain any more. Moreover, it seems that RX work, at least
> ifconfig reports non-zero values for RX packets. Still, there is no even single
> packet transmitted from the smsc9220 :(
> I suspect that there's some problem with the hardware, and probably someone
> encountered similar problems and may have found a solution.

For the electrical part, there is a reference schematic from SMSC. And
on PXA, you need some dumb CMOS logic or a CPLD for proper interfacing
of the digital bus. Don't know if the latter is also necessary for OMAP.

Daniel

Re: smsc9220 with omap3

[Mike Rapoport]

Daniel Mack wrote:
> On Wed, Jun 24, 2009 at 05:52:16PM +0300, Mike Rapoport wrote:
>>> We're using that chip connected to a PXA300 and it works well, so I
>>> wouldn't suspect the driver. Does it work from any other scenario like
>>> the bootloader? And I don't know OMAP, but it might be helpful to others
>>> if you posted your platform_data config.
>> The platform_data I use is:
>> static struct smsc911x_platform_config smsc911x_config = {
>> 	.irq_polarity = SMSC911X_IRQ_POLARITY_ACTIVE_LOW,
>> 	.irq_type = SMSC911X_IRQ_TYPE_PUSH_PULL,
>> 	.flags = SMSC911X_USE_32BIT | SMSC911X_FORCE_INTERNAL_PHY,
>> };
>>
>> The chip is properly detected, and if I tweak the loopback test to return 0, the
>>  driver does not complain any more. Moreover, it seems that RX work, at least
>> ifconfig reports non-zero values for RX packets. Still, there is no even single
>> packet transmitted from the smsc9220 :(
>> I suspect that there's some problem with the hardware, and probably someone
>> encountered similar problems and may have found a solution.
> 
> For the electrical part, there is a reference schematic from SMSC.

Yeah, I know. We actually copied the design from the reference.

> And on PXA, you need some dumb CMOS logic or a CPLD for proper interfacing
> of the digital bus. Don't know if the latter is also necessary for OMAP.

Do you mean that the logic is necessary to demux address and data? If yes, OMAP
does not need it.
Besides, we have no problem in CPU <-> lan9220 communications, there's no
network traffic from the chip. :(

> Daniel
> 

-- 
Sincerely yours,
Mike.

Re: smsc9220 with omap3

[Tony Lindgren]

* Mike Rapoport <mike@compulab.co.il> [090625 08:58]:
> Daniel Mack wrote:
> > On Wed, Jun 24, 2009 at 05:52:16PM +0300, Mike Rapoport wrote:
> >>> We're using that chip connected to a PXA300 and it works well, so I
> >>> wouldn't suspect the driver. Does it work from any other scenario like
> >>> the bootloader? And I don't know OMAP, but it might be helpful to others
> >>> if you posted your platform_data config.
> >> The platform_data I use is:
> >> static struct smsc911x_platform_config smsc911x_config = {
> >> 	.irq_polarity = SMSC911X_IRQ_POLARITY_ACTIVE_LOW,
> >> 	.irq_type = SMSC911X_IRQ_TYPE_PUSH_PULL,
> >> 	.flags = SMSC911X_USE_32BIT | SMSC911X_FORCE_INTERNAL_PHY,
> >> };
> >>
> >> The chip is properly detected, and if I tweak the loopback test to return 0, the
> >>  driver does not complain any more. Moreover, it seems that RX work, at least
> >> ifconfig reports non-zero values for RX packets. Still, there is no even single
> >> packet transmitted from the smsc9220 :(
> >> I suspect that there's some problem with the hardware, and probably someone
> >> encountered similar problems and may have found a solution.
> > 
> > For the electrical part, there is a reference schematic from SMSC.
> 
> Yeah, I know. We actually copied the design from the reference.
> 
> > And on PXA, you need some dumb CMOS logic or a CPLD for proper interfacing
> > of the digital bus. Don't know if the latter is also necessary for OMAP.
> 
> Do you mean that the logic is necessary to demux address and data? If yes, OMAP
> does not need it.
> Besides, we have no problem in CPU <-> lan9220 communications, there's no
> network traffic from the chip. :(

In addition to checking the pin muxing.. Do you have the GPMC configured for
the timings? See arch/arm/mach-omap2/gpmc.c.

Regards,

Tony