Re: [BUG] PCIe Stall on USB_PCI=y with SoCFPGA Root Port

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

[-- Attachment #1: Type: text/plain, Size: 2215 bytes --]

On Tue, 3 Mar 2026, Sune Brian wrote:

> On Tue, Mar 3, 2026 at 7:51 AM Bjorn Helgaas <helgaas@kernel.org> wrote:
> >
> > [+to Joyce, pcie-altera.c maintainer]
> >
> > On Sat, Feb 28, 2026 at 11:05:17PM +0800, Sune Brian wrote:
> > > Dear PCI maintainers,
> > >
> > > Root Port: Cyclone V HPS Root Port PCIe
> > > Endpoint: uPD720202 Card
> > > Kernel version: 6.12.33-g3234b1ed8956-dirty
> >
> > v6.12 is fairly old; can you reproduce a problem with v6.19 or
> > v7.0-rc1?
> 
> Had tried 6.19 same situation USB_PCI=y simply repeatable hang.
> No message shown or tells what happened.

Hi,

Has it worked with some kernel at some point in the past (in which case 
git bisect might be possible to pinpoint what broke it)?

-- 
 i.

> > > Sanity confirmation of PCIe RP:
> > > Both NVMe and Realtek WIFI cards are tested
> > > and show no major issue with MSI/MSI-X enabled.
> > >
> > > Repeatable Result:
> > > When uPD720202 card is applied onto the PCIe slot.
> > > Either module or included driver could stall during kernel boot.
> > > The stall can be resolved by setting USB_PCI=n and
> > > after boot under distro lspci -vvv can disable the card info [2].
> > > All nosmi noaspm etc are tested and state the same stall.
> > >
> > > The root port info is shown in [1].
> > > Based on the above behavior we had also tried to reflash the
> > > ROM of the uPD720202 card and no difference was shown
> > > on the stall issue.
> > > The uPD720202 card is tested on another ARM 64 board and
> > > shows no stall issue.
> >
> > Do you have any details about what the actual issue is?  I guess it's
> > some kind of hang, but you don't know exactly where in the code it
> > happens?
> 
> I can't have tried so many basic dmesg and possible runtime debug methods.
> It just simply hangs or stalls.
> Maybe more guidance on how to turn on more debug messages could help.
> 
> The best I can do to know what triggered is USB_PCI=y.
> 
> Thanks,
> Brian
> 
> >
> > > With USB_PCI=n it is sure this is driver issue between
> > > pcie-altera.c & pcie-altera-msi.c
> > >
> > > Additional boot log [3] with USB_PCI=n
> > >
> > > Thanks,
> > > Brian
> > >
>