Re: [3.9-rc1] irq 16: nobody cared (was [3.9-rc1] very poorinterrupt responses)

[Peter Hurley]

On Thu, 2013-03-14 at 18:22 +0100, Rafael J. Wysocki wrote:
> On Thursday, March 14, 2013 01:06:04 PM Peter Hurley wrote:
> > On Thu, 2013-03-14 at 17:46 +0100, Rafael J. Wysocki wrote:
> > > On Thursday, March 14, 2013 05:09:59 PM Jiri Kosina wrote:
> > > > On Thu, 14 Mar 2013, Jiri Kosina wrote:
> > > > 
> > > > > > > I don't think I have seen this message on rc1+ (8343bce, to be precise), 
> > > > > > > but I have definitely seen sluggish system response on that kernel as 
> > > > > > > well.
> > > > > > > 
> > > > > > > Attaching lspci, /proc/interrupts and dmesg. 
> > > > > > 
> > > > > > Can you try to do a git bisect for this?  Is the sluggish system 
> > > > > > response clear enough that you can tell reliably when it is present and 
> > > > > > when it isn't?
> > > > > 
> > > > > That was my first thought, but unfortunately I am afraid there will be 
> > > > > point at which I will easily make a bisection mistake, as the 
> > > > > responsiveness of the system varies over time, so it's not really a 
> > > > > 100% objective measure.
> > > > 
> > > > So I will try a bisect, but it'll take some time so that I could claim it 
> > > > to be trustworthy.
> > > > 
> > > > Therefore in case anyone has any idea in parallel, I am all ears.
> > > 
> > > This one is a candidate to focus on I think:
> > > 
> > > commit 181380b702eee1a9aca51354d7b87c7b08541fcf
> > > Author: Yinghai Lu <yinghai@xxxxxxxxxx>
> > > Date:   Sat Feb 16 11:58:34 2013 -0700
> > > 
> > >     PCI/ACPI: Don't cache _PRT, and don't associate them with bus numbers
> > 
> > This patch __fixed__ this problem for me in linux-next back in February.
> > 
> > Rafael, did you hold back some ACPI patches from 3.9 that would have
> > made fix no longer applicable?
> 
> No, I didn't.
> 
> I'm afraid, though, that the fix might not be effective on some systems for a
> reason that's unclear at the moment.
> 
> So in fact the one to check is commit 4f535093cf ("PCI: Put pci_dev in device
> tree as early as possible") and if the problem doesn't appear before that, we
> need to figure out why the fix may not be sufficient.

I agree. 

Commit 4f535093cf ("PCI: Put pci_dev in device tree as early as
possible") is the likely culprit, and "Don't cache _PRT..." is probably
an insufficient fix.

Not so sure about the other reporters though because they had active
devices on those USB ports.

Regards,
Peter Hurley


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/