Re: [PATCH v1] usb: hcd: Try MSI interrupts on PCI devices

["andriy.shevchenko@linux.intel.com" <andriy.shevchenko@linux.intel.com>]

On Sun, Mar 13, 2022 at 09:44:36PM +0000, micklorain wrote:
> On Wednesday, March 9th, 2022 at 18:47, andriy.shevchenko@linux.intel.com <andriy.shevchenko@linux.intel.com> wrote:
> > On Sun, Mar 06, 2022 at 04:09:49PM +0000, micklorain wrote:
> > > On Monday, February 28th, 2022 at 21:59, andriy.shevchenko@linux.intel.com andriy.shevchenko@linux.intel.com wrote:
> > > > On Mon, Feb 28, 2022 at 08:12:47PM +0000, micklorain wrote:

> > ...
> >
> > > > Last time I have got something similar it becomes that PCI bridge which is used
> > > >
> > > > to connect USB controller to the PCI Root Bridge was not capable of MSI, while
> > > >
> > > > advertising that capability. I.o.w. HW bug.
> > > >
> > > > To understand if it's something similar, please run (under the root) each of
> > > >
> > > > the following commands:
> > > >
> > > > lspci -nk -vvv
> > > >
> > > > cat /proc/interrupts
> > > >
> > > > in both cases, i.e. working and non-working.
> > > >
> > > > And then share the output (all 4 files).
> >
> > > Thanks for your reply.
> > >
> > > This is the results of the commands you requested :
> > >
> > > * When things work (commit dcb85f85fa6f142aae1fe86f399d4503d49f2b60 with commit 306c54d0edb6ba94d39877524dddebaad7770cf2 reverted)
> > >
> > > - lspci -nk -vvv :
> > >
> > > https://paste.debian.net/hidden/77d92dc9/
> > >
> > > - cat /proc/interrupts
> > >
> > > https://paste.debian.net/hidden/67208c8e/
> > >
> > > * When things are broken (commit dcb85f85fa6f142aae1fe86f399d4503d49f2b60)
> > >
> > > - lspci -nk -vvv :
> > >
> > > https://paste.debian.net/hidden/121362b3/
> > >
> > > - cat /proc/interrupts :
> > >
> > > https://paste.debian.net/hidden/dbe8d1bb/
> > >
> > > Hope this can help.
> >
> > Thank you for sharing. Are you able to compile a kernel and boot it? If so,
> >
> > can you try the following patch?
> 
> I tried the patch, but it didn't help...
> 
> >
> > diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
> >
> > index 65f7f6b0576c..149742aa5f4d 100644
> >
> > --- a/drivers/pci/quirks.c
> >
> > +++ b/drivers/pci/quirks.c
> >
> > @@ -3041,6 +3041,13 @@ DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_BROADCOM,
> >
> > PCI_DEVICE_ID_TIGON3_5715S,
> >
> > quirk_msi_intx_disable_bug);
> >
> > +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x4386, quirk_msi_intx_disable_ati_bug);
> >
> > +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x4387, quirk_msi_intx_disable_ati_bug);
> >
> > +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x4388, quirk_msi_intx_disable_ati_bug);
> >
> > +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x4389, quirk_msi_intx_disable_ati_bug);
> >
> > +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x438a, quirk_msi_intx_disable_ati_bug);
> >
> > +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x438b, quirk_msi_intx_disable_ati_bug);
> >
> > +
> >
> > DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x4390,
> >
> > quirk_msi_intx_disable_ati_bug);
> >
> > DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x4391,
> >
> > In case it doesn't help, remove 'ati_' part from the function name and try again.
> 
> ... but removing 'ati_' from the function name indeed fixed my issue.
> 
> Will this fix be upstreamed now ?

I will do it ASAP. Thanks for report and testing!

-- 
With Best Regards,
Andy Shevchenko