Re: [RFC BUG] pci: Freescale i.MX6 PCIe controller: revert mainline regression

[Sven Van Asbroeck]

Thank you Trent !

This patch was intended as a bug report, rather than an actual
solution to be merged.
I chose this format so people could easily verify the change in
behaviour, just by
applying the patch locally.

On Tue, Dec 18, 2018 at 6:06 PM Trent Piepho <tpiepho@xxxxxxxxxx> wrote:
>
>
> It's actually commit f3fdfc4ac3a2 ("PCI: Remove host driver Kconfig
> selection of CONFIG_PCIEPORTBUS") that made this problem apparent.
>
> But the real bug is not in that commit, but a problem with imx6/7 not
> configuring a setting in the RC correctly without pcieportbus, which
> should not be necessary.
>
> See "[v4] PCI: imx: make msi work without CONFIG_PCIEPORTBUS=y" by
> Richard Zhu for a real fix.

Awesome! I can verify this patch on our i.MX6 boards, if Richard Zhu
(or anyone else) is interested.

>
> There another bug that makes this driver sometimes lose an MSI, but

Is it the one you alerted me to before?
https://patchwork.kernel.org/patch/10657987/

We are planning to use the PCIe controller in a production environment -
to communicate with a tg3 based Broadcom ethernet NIC.

What are the practical consequences of this bug?
What is the state of the patch that fixes it?
Last patch activity seems to have been in mid-November?

Apologies for the trivial questions, I am by no means a PCIe expert.

> that regression was in v4.14 and would not be fixed by your patch, so
> this tigon3 problem is almost certainly caused by the previously
> mentioned issue with pcieportbus.
>