Re: pq2_find_bridges hangs system

Re: pq2_find_bridges hangs system

[Alex BASTOS]

Vitaly,

So, I have arrived to some conclusions.

1) With my previous kernel version (2.6.11) I had no problems
because it had CONFIG_PCI_8260 and CONFIG_PPC_INDIRECT_PCI unset.
So I think no effective read/write operation was executed, and so
that, no Machine Check.

In fact, this is the configuration I would prefer for my board. I have
to set CONFIG_PCI for enabling USB, but I don't want all the HW stuff
of the PCI. I can get these also with 2.6.15 if I modify Kconfig, to
allow me to unset PCI_8260 (and then PPC_INDIRECT_PCI) for my board.
With that, the problem dissapears and I can boot 2.6.15.

Shouldn't this (PCI_8260) be visible from xconfig for those like me
who wants USB (=> CONFIG_PCI) but don't really have any PCI device?

2) Although I am using Uboot 1.1.4, it is not top of git. I have
found your changes to support PCI on 8272ADS for u-boot are not
still applied on the version i am using. So, could this be causing
the problem? Is there any configuration done in u-boot required by
the PCI on the linux kernel to boot (BR3,OR3, EMR, ...)?

I should say that, on ADS, I have an older version of u-boot, 1.1.2,
and it boots OK.

3) I have seen that from u-boot, a reset occurs when I read Internal
Memory at offset 0x10904 (PCI CFG_DATA). From BDI, when I do the same
all PCI section on IM becomes zero. Is this a known behaviour? May this
reflect a hardware problem?

In conclusion, 1) solves my "must", a working board (without PCI).
But I still would like to know what am I doing so wrong with this.

Best regards,

Alex

Citando Alex BASTOS <alebas@televes.com>:

> Vitaly,
>
> It didn't work. So I will check pq2ads_setup_pci to check if
> some board specific issue is affected.
>
> I will say you if I find anything
>
> Thanks
>
> Alex.
>

Re: pq2_find_bridges hangs system

[Vitaly Bordug]

Alex BASTOS wrote:
> Vitaly,
> 
> So, I have arrived to some conclusions.
> 
> 1) With my previous kernel version (2.6.11) I had no problems
> because it had CONFIG_PCI_8260 and CONFIG_PPC_INDIRECT_PCI unset.
> So I think no effective read/write operation was executed, and so
> that, no Machine Check.
> 
> In fact, this is the configuration I would prefer for my board. I have
> to set CONFIG_PCI for enabling USB, but I don't want all the HW stuff
> of the PCI. I can get these also with 2.6.15 if I modify Kconfig, to
> allow me to unset PCI_8260 (and then PPC_INDIRECT_PCI) for my board.
> With that, the problem dissapears and I can boot 2.6.15.
> 
> Shouldn't this (PCI_8260) be visible from xconfig for those like me
> who wants USB (=> CONFIG_PCI) but don't really have any PCI device?
> 
Maybe, but I guess more correct is to deal with USB->PCI dependency.. 
Is it really needed(I am not USB expert)?
> 2) Although I am using Uboot 1.1.4, it is not top of git. I have
> found your changes to support PCI on 8272ADS for u-boot are not
> still applied on the version i am using. So, could this be causing
> the problem? Is there any configuration done in u-boot required by
> the PCI on the linux kernel to boot (BR3,OR3, EMR, ...)?
> 
> I should say that, on ADS, I have an older version of u-boot, 1.1.2,
> and it boots OK.
> 
My changes do not required for kernel to deal with PCI, that patch just allows access to PCI in 
U-Boot. Hence I might be useful if you'll head for resolving this issue which I guess unlikely to happen
> 3) I have seen that from u-boot, a reset occurs when I read Internal
> Memory at offset 0x10904 (PCI CFG_DATA). From BDI, when I do the same
> all PCI section on IM becomes zero. Is this a known behaviour? May this
> reflect a hardware problem?
> 
Sounds weird 
> In conclusion, 1) solves my "must", a working board (without PCI).
> But I still would like to know what am I doing so wrong with this.
> 
IMHO ability to disable PCI_8260 while PCI is on might be confusing, at least for upstream.



> Best regards,
> 
> Alex
> 
> Citando Alex BASTOS <alebas@televes.com>:
> 
>> Vitaly,
>>
>> It didn't work. So I will check pq2ads_setup_pci to check if
>> some board specific issue is affected.
>>
>> I will say you if I find anything
>>
>> Thanks
>>
>> Alex.
>>
> 
> 


-- 
Sincerely, 
Vitaly

[SOLVED] Re: pq2_find_bridges hangs system

[Alex BASTOS]

Vitaly,

> Maybe, but I guess more correct is to deal with USB->PCI dependency..
> Is it really needed(I am not USB expert)?

Neither am I. I guess for this Host Controler it is not needed at
HW level, but I am not sure why it is still required by the SW.


> > 3) I have seen that from u-boot, a reset occurs when I read Internal
> > Memory at offset 0x10904 (PCI CFG_DATA). From BDI, when I do the same
> > all PCI section on IM becomes zero. Is this a known behaviour? May this
> > reflect a hardware problem?
> >
> Sounds weird

Here is the problem. The cause for this weird behaviour is that I
had PCI_ARBITER_EN_  disabled. So, a HW problem at the end.
I had disabled it trying to reduce power compsumption due to the PCI,
as I am not going to use it. It seems it is not possible that way.
Any idea how to get this done?

> > In conclusion, 1) solves my "must", a working board (without PCI).
> > But I still would like to know what am I doing so wrong with this.
> >
> IMHO ability to disable PCI_8260 while PCI is on might be confusing, at least
> for upstream.

You are right.

Thanks so much for your help.

Best regards,

Alex

Re: [SOLVED] Re: pq2_find_bridges hangs system

[Vitaly Bordug]

Alex BASTOS wrote:
> Vitaly,
> 
>> Maybe, but I guess more correct is to deal with USB->PCI dependency..
>> Is it really needed(I am not USB expert)?
> 
> Neither am I. I guess for this Host Controler it is not needed at
> HW level, but I am not sure why it is still required by the SW.
> 
> 
>>> 3) I have seen that from u-boot, a reset occurs when I read Internal
>>> Memory at offset 0x10904 (PCI CFG_DATA). From BDI, when I do the same
>>> all PCI section on IM becomes zero. Is this a known behaviour? May this
>>> reflect a hardware problem?
>>>
>> Sounds weird
> 
> Here is the problem. The cause for this weird behaviour is that I
> had PCI_ARBITER_EN_  disabled. So, a HW problem at the end.
> I had disabled it trying to reduce power compsumption due to the PCI,
> as I am not going to use it. It seems it is not possible that way.
> Any idea how to get this done?
> 
Well, I guess the best you can do now - just use custom tree. I'll try to find 
good solution for such a case (PCI required by deps but not hw) next time I'll touch PCI.

>>> In conclusion, 1) solves my "must", a working board (without PCI).
>>> But I still would like to know what am I doing so wrong with this.
>>>
>> IMHO ability to disable PCI_8260 while PCI is on might be confusing, at least
>> for upstream.
> 
> You are right.
> 
> Thanks so much for your help.
> 
> Best regards,
> 
> Alex
> 
> 


-- 
Sincerely, 
Vitaly

Re: [SOLVED] Re: pq2_find_bridges hangs system

[Kumar Gala]

On Dec 14, 2005, at 8:14 AM, Vitaly Bordug wrote:

> Alex BASTOS wrote:
>> Vitaly,
>>> Maybe, but I guess more correct is to deal with USB->PCI  
>>> dependency..
>>> Is it really needed(I am not USB expert)?
>> Neither am I. I guess for this Host Controler it is not needed at
>> HW level, but I am not sure why it is still required by the SW.
>>>> 3) I have seen that from u-boot, a reset occurs when I read  
>>>> Internal
>>>> Memory at offset 0x10904 (PCI CFG_DATA). From BDI, when I do the  
>>>> same
>>>> all PCI section on IM becomes zero. Is this a known behaviour?  
>>>> May this
>>>> reflect a hardware problem?
>>>>
>>> Sounds weird
>> Here is the problem. The cause for this weird behaviour is that I
>> had PCI_ARBITER_EN_  disabled. So, a HW problem at the end.
>> I had disabled it trying to reduce power compsumption due to the PCI,
>> as I am not going to use it. It seems it is not possible that way.
>> Any idea how to get this done?

I'm guessing you dont have anything else doing arbitration on the bus.

- kumar

Re: [SOLVED] Re: pq2_find_bridges hangs system

[Alex BASTOS]

Kumar,

> >> Here is the problem. The cause for this weird behaviour is that I
> >> had PCI_ARBITER_EN_  disabled. So, a HW problem at the end.
> >> I had disabled it trying to reduce power compsumption due to the PCI,
> >> as I am not going to use it. It seems it is not possible that way.
> >> Any idea how to get this done?
>
> I'm guessing you dont have anything else doing arbitration on the bus.

You're right. I wasn't using the PCI at all.

>
> - kumar
>


Best regards,

Alex