[PATCH v3] pci: Work around PCIe link training failures

[Maciej W. Rozycki]

On Wed, 12 Jan 2022, Tom Rini wrote:

> >  I believe this version has addressed all concerns raised in the review 
> > thus far.  With the nature of a problem better understood now I'm sending 
> > a corresponding update for Linux as well.
> 
> What as the feedback to your Linux change?  Is this essentially the path
> forward still?  Thanks!

 There has been no response so far, perhaps due to unfavourable timing, 
the festive season, etc.  I have rebased my original change and posted a 
regenerated version a bit more than a week ago:

<https://lore.kernel.org/linux-pci/alpine.DEB.2.21.2201022100430.56863@angie.orcam.me.uk/>

and a few of my other resubmissions in the PCI area have been partially 
reviewed last week.  So things have been progressing, but with Linux 5.16 
released last Sun we're in the merge window for 5.17 now, so people are 
surely busy with that.  We shall see.

 NB as I previously noted the Linux change has to be different, because 
you cannot busy-loop polling a bit in a device register in an OS, unlike 
in firmware, and therefore the Linux version has to rely on the data link 
layer active reporting capability, which may not be there in PCIe 2.0 
devices.  Conceptually the two changes remain similar though, so I guess 
input from the Linux side will still be valuable.

 Thank you for your attention to my proposal.

  Maciej