Re: [PATCH v2 1/1] PCI: brcmstb: Fix regression regarding missing PCIe linkup

[Jim Quinlan]

On Thu, Jun 2, 2022 at 3:18 PM Cyril Brulebois <kibi@xxxxxxxxxx> wrote:
>stm32mp157c-ev1.dts
> Hi Stefan,
>
> Stefan Wahren <stefan.wahren@xxxxxxxx> (2022-05-30):
> > i think we should better trust the results based on the v5.18 tag. During
> > the merge window, regressions from other subsystems are possible.
>
> Alright, that looks like a great plan.
>
> Before getting your answer, I had tried taking the reboot part out of
> the equation, but I found out that even cold boots might fail with the
> mmc storage.

Hi Cyril,

FWIW, I can deliberately reproduce the errors you observed by using an
incorrect "interrupts" property for
the sdhci device's DT node.  It can also be triggered by removing its
"clocks" property.

However, if I do one of the above,  the error will occur on every
boot, but in your case (I think) you are seeing it
sporadically.  So that is peculiar.  I've looked at the recent
upstream commits for changes in the  sdhci driver
and also the relevant DT node and do not see anything obvious that
might cause this.

BTW,  when you observe this error, can you please do a  "cat
/proc/interrupts" and post the results?

Thanks & regards,
Jim Quinlan
Broadcom STB


>
> I haven't been able to conduct a systematic testing of the patch on top sdhci driver or the relevant
DT node
> of the v5.18 tag yet (say, 10 or 20 cold boots, and the same with
> reboots) due to strong work constraints these past few days, but that's
> definitely still on my short term todo list (hopefully before the end of
> the week).
>
> Sorry I didn't manage to get that lined up before Bjorn's pull request.
>
>
> Cheers,
> --
> Cyril Brulebois (kibi@xxxxxxxxxx)            <https://debamax.com/>
> D-I release manager -- Release team member -- Freelance Consultant