Hi,
I have a Supermicro X8DTH-6F motherboard and have been successfully running ESXi 4.0 for some time.
Recently, I upgraded to ESXi 4.1 and was face with a PSOD on boot. After a bit of debugging, I have found that turning of VT-d support in the BIOS is what is causing it to PSOD.
This is puzzling as when i had VT-d enabled in 4.0, there were no issues. After upgrading, it has suddenly caused ESX to crash.
What's more is that when i upgraded 4.0 to 4.1 using vSphere Host Update Utility (this was possible a few weeks ago) it seemed to work fine as well.
After some further digging, I have noticed this in the release notes :
-
Supported Hardware Issues
PCI device mapping errors on 370 G6 HP server
When you run I/O operations on the 370 G6 HP server, you might encounter a purple screen or see alerts about 'Lint1 Interrupt' or NMI on the console. The HP ProLiant DL370 G6 server has two Intel I/O hub (IOH)s and a BIOS defect in the ACPI Direct Memory Access remapping (DMAR) structure definitions, which causes some PCI devices to be incorrectly described under the wrong DMA remapping unit. Any DMA access by such incorrectly described PCI devices trigger an IOMMU fault and the device will receive an I/O error. Depending on the device, this I/O error might result in an Lint1 Interrupt or NMI alert message on the console or the system might freeze with a purple screen.
Workaround: Check the HP Web site for availability of a BIOS patch for the HP ProLiant DL370 G6 server. If a patch is not available, disable 'Intel(R) VT-d' in the BIOS so that the BIOS will not publish DMAR structures in the ACPI tables.
-
As my motherboard also has two IOH, I was wondering whether this could potentially be related.
Can anyone provide any further insight into the problem with the DL370 G6 and also whether or not it experienced the same issues in ESX 4.0?
Thanks in advance!
Best regards
Leonard
