Using hardware Input Output Memory Management Unit (IOMMU) to troubleshoot memory corruption

Products

• VMware vSphere ESXi

Issue/Introduction

This article provides information about using hardware Input Output Memory Management Unit (IOMMU) to troubleshoot memory corruption.

 

Environment

VMware vSphere ESXi 8.0
VMware vSphere ESXi 7.0
VMware vSphere ESXi 6.7
VMware vSphere ESXi 6.0
VMware vSphere ESXi 6.5
VMware vSphere ESXi 5.5

 

 

Resolution

From vSphere ESXi 5.5 supports hardware input/output memory management unit (IOMMU) features that may assist in identifying the cause of memory corruption.

Faulty firmware, device hardware,and drivers can cause stray direct memory address (DMA) into memory addresses that may result in memory data corruption. Since memory corruption manifests as purple diagnostic screens or data integrity failures, it is difficult to find the source of the problem. If memory corruption is suspected, especially due to devices or drivers, turn on this feature and try to reproduce the issue. This causes the system to generate a purple diagnostic screen at the point of occurrence of faulty DMA operations, which increases the chances of identifying the cause of the corruption.

 

To set a vmkernel boot option:

 

Caution: There is a performance overhead associated to enabling this feature and it should not be used in production.

1. Reboot the ESXi host you want to troubleshoot.
2. When prompted, press SHIFT + O.
3. Add this text to the boot options:
   
   dmaMapperPolicy=debug
   
   Note: This feature is only available during boot and is not persistent.