PCI: Work around PCIe link training failures

[Bjorn Helgaas]

On Mon, Aug 05, 2024 at 06:06:59PM -0600, Matthew W Carlis wrote:
> Hello again. I just realized that my first response to this thread two weeks
> ago was not actually starting from the end of the discussion. I hope I found
> it now... Must say sorry for this I am still figuring out how to follow these
> threads.
> I need to ask if we can either revert this patch or only modify the quirk to
> only run on the device in mention (ASMedia ASM2824). We have now identified
> it as causing devices to get stuck at Gen1 in multiple generations of our
> hardware & across product lines on ports were hot-plug is common. To be a
> little more specific it includes Intel root ports and Broadcomm PCIe switch
> ports and also Microchip PCIe switch ports.
> The most common place where we see our systems getting stuck at Gen1 is with
> device power cycling. If a device is powered on and then off quickly then the
> link will of course fail to train & the consequence here is that the port is
> forced to Gen1 forever. Does anybody know why the patch will only remove the
> forced Gen1 speed from the ASMedia device?

Thanks for keeping this thread alive.  I don't know the fix, but it
does seem like this series made wASMedia ASM2824 work better but
caused regressions elsewhere, so maybe we just need to accept that
ASM2824 is slightly broken and doesn't work as well as it should.

Bjorn