Problems with PCI-E devices not being detected with switch

[Stefan Roese]

On Thursday 16 October 2008, Benjamin Herrenschmidt wrote:
> > pci 0001:02:1c.0: ignoring class 5d01 (doesn't match header type 02)
> > pci 0001:02:1d.0: unknown header type 41, ignoring device
> > pci 0001:02:1e.0: unknown header type 03, ignoring device
> > pci 0001:02:1f.0: unknown header type 41, ignoring device
>
> Stefan, do you reckon it could be that we aren't leaving enough time
> for the things behind the switch to initialize ? Or could there be
> a subtle kernel bug here ? It looks to me that config space
> access behind the switch is broken.

Yes, this could be possible. We (Ayman) should at least test with the delay as 
you mentioned below to see if this helps.

> Ayman, can you try adding a long delay (such as msleep(5000), ie 5s)
> at the beginning of pcibios_init() in arch/powerpc/kernel/pci_32.c ?
> This will add 5s delay between the init/reset of the port and the
> probing by linux. Do that help ?
>
> Stefan, shouldn't we find a nice way to avoid the whole port reset and
> reconfiguration of the HW also when uboot already did a good enough job,
> maybe via some device-tree property ? It would also significantly speed
> up boot times.

Doing this unconditionally is not a good idea since we could have an old 
(buggy) firmware which didn't configure the PCIe controller correctly. But I 
really like your idea with the device-tree property to optionally skip this 
re-configuration. Now we only need to find some "volunteer" to do this 
job... ;)

Best regards,
Stefan