Re: [PATCH v1 2/4] PCI: Mark invalid BARs as unassigned

[Bjorn Helgaas <bhelgaas@google.com>]

Hi Tony,

On Tue, Apr 14, 2015 at 6:14 PM, Tony Luck <tony.luck@gmail.com> wrote:
> On Thu, Mar 12, 2015 at 10:35 AM, Bjorn Helgaas <bhelgaas@google.com> wrote:
>> If a BAR is not inside any upstream bridge window, or if it conflicts with
>> another resource, mark it as IORESOURCE_UNSET so we don't try to use it.
>> We may be able to assign a different address for it.
>
> This went into Linux tree - and broke my ia64 box in a weird way. It only
> sees one of the two disks.  git bisect points to this commit, and reverting
> it fixes the problem.
>
> Attached are serial logs from the "bad" boot (tip of Linus tree = b79013b2449c)
> and for a "good" boot with this commit reverted.

Thanks for the report, and sorry for breaking you.  I hate to ask for
more after all your work already, but would you mind collecting logs
with "ignore_loglevel"?  We only print the BAR information at
KERN_DEBUG, so it doesn't make it to the console unless you use
ignore_loglevel.

I don't really see how the mptsas device at 83:00.0 worked before;
ACPI didn't tell us about any MMIO apertures in the PCI1 host bridge.
But I'll look closer.

Bjorn