mpt fails to find disks on LSILogic 1068

[John Baldwin]

On Friday, March 23, 2012 2:13:43 am Dmitry Sivachenko wrote:
> On 3/23/12 3:42 AM, John Baldwin wrote:
> > On 3/22/12 7:05 PM, Marius Strobl wrote:
> >> On Thu, Mar 22, 2012 at 11:09:38PM +0400, Dmitry Sivachenko wrote:
> >>> On 3/22/12 2:01 PM, Marius Strobl wrote:
> >>>> On Thu, Mar 22, 2012 at 09:15:52AM +0400, Dmitry Sivachenko wrote:
> >>>>> Hello!
> >>>>>
> >>>>> I have a server with LSILogic 1068-IR Controller with 2 disks connected.
> >>>>> This system was running FreeBSD-7.3 without a problem and recently I tried
> >>>>> to upgrade to 9-STABLE (build on 15 of February).
> >>>>>
> >>>>> New version of FreeBSD fails to initialize this controller:
> >>>>>
> >>>>> mpt0:<LSILogic SAS/SATA Adapter>   port 0xd000-0xd0ff irq 26 at device 3.0
> >>>>> on pci6
> >>>>> pcib3: attempting to grow memory window for (0-0xffffffff,0x100000)
> >>>>>          front candidate range: 0xfdd00000-0xfddfffff
> >>>>>          back candidate range: 0cfe000000-0xfe0fffff
> >>>>> pcib7: failed to allocate initial memory window (0-0xffffffff,0x100000)
> >>>>> mpt0: 0x4000 bytes of rid 0x14 res 3 failed (0, 0xffffffffffffffff).
> >>>>> mpt0: Unable to memory map registers.
> >>>>> mpt0: Giving Up.
> > Can you get a full log from a verbose boot?  The root problem is
> > probably earlier when mpt's BARs are first examined and probably has to
> > do with an issue with the decoding windows on pcib7.
> >
> >
> 
> Please find the full log below:
> 
> pcib0: <ACPI Host-PCI bridge> port 0xcf8-0xcff on acpi0
> pcib0: decoding 4 range 0-0xcf7
> pcib0: decoding 4 range 0xd00-0xffff
> pcib0: decoding 3 range 0xa0000-0xbffff
> pcib0: decoding 3 range 0xd0000-0xdffff
> pcib0: decoding 3 range 0xf0000000-0xfebfffff

This is ok.

> pcib3: <ACPI PCI-PCI bridge> irq 56 at device 9.0 on pci0
> pcib0: allocated type 4 (0xd000-0xefff) for rid 1c of pcib3
> pcib3: failed to allocate initial memory window: 0xfdb00000-0xfebfffff

And here is where it blows up.  That range is in the range of the
parent bridge (pcib0), so there must be some sort of conflict.

Can you try booting with 'debug.acpi.disabled=sysres'?  Also, can
you obtain the output of devinfo -rv from a pre-9 kernel?

-- 
John Baldwin