Release Notes

NMI message is displayed on Dell EMC AMD systems
with NVMe SSDs
Description:
While running high intensity I/O on NVMe drives connected to Dell EMC AMD system with EPYC
processor, the system displays the following message:
Uhhuh. NMI received for unknown
reason 2d on CPU 59.
Applies to: Red Hat Enterprise Linux 7.4
Cause: The AMD iommu driver code used in RHEL 7.4 is old and does not have fixes and enhancements that are
available in the upstream kernel. The high intensity I/O to NVMe disks results in huge number of requests
to add and remove regions in iommu and the unnecessary regions are flushed. The flushing process
occurs very often with the iommu code in RHEL 7.4 and each flush takes longer than expected resulting
in a soft lockup scenario.
Workaround: Pass kernel boot parameter "iommu=pt"
Booting into OS with the QLogic FastLinQ QL41000
card setup with iscsi offload logs multiple error
messages related to iSCSI session
Description:
While booting into the server operating system which has the FastLinQ QL41000 card installed with the
iscsi offload configuration, displays the following error message, and those error messages are logged by
iscsistart during dracut initqueue stage.
dracut-initqueue: iscsistart: initiator reported error (15 - session
exists)
Applies to: Red Hat Enterprise Linux 7.5 and 7.6
Cause: NPAR setting has to be enabled for setting up iSCSI offload on QLogic FastLinQ QL41000. During dracut-
initqueue stage, iscsistart is trying to establish the iSCSI session for all the partitions that come online,
but the session is already established with one partition, hence the dracut-initqueue stage error messages
are logged.
Workaround: These error messages can be ignored because there is no functionality loss.
Kernel panic occurs when FCoE function is enabled in
QL41000 series adapters
Description:
During Red Hat Enterprise Linux 7.4 installation on Dell EMC's yx4x PowerEdge R7425, R7415 and R6415
servers, kernel panic occurs when FCoE function is enabled in QL41000 series adapters.
Applies to: Red Hat Enterprise Linux 7.4 on Dell EMC's yx4x PowerEdge R7425, R7415 and R6415 servers
Cause: This occurs due to an issue with the AMD/iommu driver.
Workaround: Boot with kernel boot parameter iommu=pt.
22 Known issues