Re: [PATCH 1/2] PCI: Setup bridge resources earlier

["Ilpo Järvinen" <ilpo.jarvinen@linux.intel.com>]

On Sat, 18 Oct 2025, Bhanu Seshu Kumar Valluri wrote:

> On 18/10/25 00:22, Bjorn Helgaas wrote:
> > On Fri, Oct 17, 2025 at 11:52:58PM +0530, Bhanu Seshu Kumar Valluri wrote:
> >>
> >> I want to report that this PATCH also break PCI RC port on TI-AM64-EVM.
> >>
> >> I did git bisect and it pointed to the a43ac325c7cb ("PCI: Set up bridge resources earlier")
> >>
> >> Happy to help if any testing or logs are required.
> > 
> > Thanks for the report!  Can you test this patch?
> > 
> >   https://patch.msgid.link/20251014163602.17138-1-ilpo.jarvinen@linux.intel.com
> > 
> > That patch is queued up as
> > https://git.kernel.org/pub/scm/linux/kernel/git/pci/pci.git/commit/?id=469276c06aff
> > and should appear in v6.18-rc2 on Sunday if all goes well.
> > 
> > If that doesn't work, let us know and we'll debug this further.
> 
> I applied above patch on top of commit f406055cb18c ("Merge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux")
> 
> Did pci rescan and run kselftest (pci_endpoint_test). It is working.
> 
> Thanks for the patch.

Thanks for testing the revert.

> Happy to help if any testing or logs are required.

I'd be interested to understand what goes wrong with the change I was 
trying to make as I want to attempt the same change later, but with all 
known issues solved by supporting changes, obviously :-).

The log snippets you provided are unfortunately too short to contain all 
the necessary information (missing e.g. root bus resources and possibly 
other helpful details).

So if you could provide dmesg and /proc/iomem contents from broken and
working (with the revert) cases to let me easily compare them, that would 
help. Please take the dmesg with dyndbg="file drivers/pci/*.c +p" on 
kernel's cmdline.

No further actions needed beyond that until later if I need to test some 
of those supporting changes before retrying all this in the mainline. It 
may take some time, even more than one kernel cycle as there have been 
quite many regressions.


-- 
 i.