[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <20060914195344.GC27531@curie-int.orbis-terrarum.net>
Date: Thu, 14 Sep 2006 12:53:44 -0700
From: "Robin H. Johnson" <robbat2@...too.org>
To: linux-kernel@...r.kernel.org
Cc: robbat2@...too.org
Subject: x86_64 (EM64T Core2 Duo, DG965RY) PCI mmconfig regression - complete hangs on boot - ACPI interaction?
[Please CC me, I'm not subscribed to LKML.]
I recently picked up a new server of really new hardware, and I've run
into a number of issues with the latest git HEAD kernels. I'll make
separate posts for each of the issues.
CPU: Intel Core 2 Duo E6400
Motherboard: Intel DG965RY
BIOS revisions: July/07/2006 AND September/06/2006 (tried both)
The most significant of them - is that somewhere between 2.6.17.11 and
2.6.18-rc6-git4, MMCONFIG is now being used, but leads to a complete
hardware hang (SYSRQ does not respond).
As of the current HEAD, I need to boot with pci=nommconf OR acpi=off,
otherwise the machine completely freezes a short while into the PCI
detection routine. Using acpi=off disables SMP, which is not desirable
on this box, but otherwise does work.
All of the following boot params lead to a failure:
pci=assign-busses
acpi=strict
pci=noacpi
acpi=ht
pci=nomsi
At the following URL, I have photos and transcriptions of the kernel
boots, as well as dmesg output from working 2.6.17.11 and .18-rc7-git1
with pci=nommconf.
http://www.orbis-terrarum.net/~robbat2/x86_64-mmconfig-failure/
--
Robin Hugh Johnson
E-Mail : robbat2@...too.org
GnuPG FP : 11AC BA4F 4778 E3F6 E4ED F38E B27B 944E 3488 4E85
Content of type "application/pgp-signature" skipped
Powered by blists - more mailing lists