RHEL 6.7을 Basic Server package로 설치 후 콘솔 화면에 아래와 유사/동일 메시지를 지속 기록하는 현상이 확인됨
----------------------------------------------------
dmar: DRHD: handling fault status reg 2
dmar: INTR-REMAP: Request device [[00:00.0] fault index 14
INTR-REMAP:[fault reason 38] Blocked an interrupt request due to source-id verification failure
----------------------------------------------------
[00:00.0] 장치는 Intel의 Skylake host bus controller이고,
메시지는, IOMMU(input–output memory management unit)의 interrupt redirector로 부터 발생된다.
VM등의 가상 머신에서 IO 장치를 보다 원활히 사용하기 위해, IO 가상화 과정에,
물리적인 장치의 메모리 영역과 가상화된 장치의 메모리를 mapping하는 것이 interrupt remapping(IR)으로 이해할 수 있다.
이 remapping이 문제가 있다는 알림 메시지이나, 일부 OS와 HW의 Interrupt remapping(IR)이 제대로 연동되지 않아 발생될 수 있다.
BIOS에서 IOMMU-Interrupt remapping 기능을 H/W에서 비활성화 하거나, OS에서 intremap(Interrupt remapping 기능을 비활성화하여, 해소/회피할 수 있다.
Test Environment
DL20 Gen9 + RHEL 6.7
-------------------------------------------------------------------
- OS install with Basic Server packages
- Default
# dmesg
dmar: DRHD: handling fault status reg 2
dmar: INTR-REMAP: Request device [[00:00.0] fault index 14
INTR-REMAP:[fault reason 38] Blocked an interrupt request due to source-id verification failure
-------------------------------------------------------------------
-------------------------------------------------------------------
- Disable Virtualization Technology from BIOS
>> Symptom persist presents
-------------------------------------------------------------------
- Disable Intel VT-d from BIOS
>> Symptom disappeared
-------------------------------------------------------------------
- Disable Shared Memory from each NIC on BIOS
>> Symptom persist presents
-------------------------------------------------------------------
- Add kernel parameter to GRUB tail (1)
-------------------------------------------
# vim /boot/grub/grub.conf (Legacy)
# vim /boot/efi/EFI/redhat/grub.conf (UEFI)
...
kernel /vmlinuz-2.6.32-573.el6.x86_64 ... rhgb quiet
-------------------------------------------
- Add kernel parameter to GRUB tail (2)
-------------------------------------------
# vim /etc/grub.conf
...
kernel /vmlinuz-2.6.32-573.el6.x86_64 ... rhgb quiet
-------------------------------------------
a) intel_iommu=off
b) intremap=no_x2apic_optout
>> Symptom persist presents
c) intremap=off
>> Symptom disappeared (1 or 2)
-------------------------------------------------------------------
참고문서.
Installed CentOS6.7 to DL20Gen9. However, strange log is output.
https://community.hpe.com/t5/ProLiant-Servers-ML-DL-SL/Installed-CentOS6-7-to-DL20Gen9-However-strange-log-is-output/td-p/6860807
Advisory: (Revision) Linux - DMAR Fault on Network Adapter With HP NC-Series Broadcom 1GbE Multifunction Driver for Linux Driver (bnx2x) When Linux "intel_iommu=on" Kernel Boot Parameter Is Used on HPE Servers
https://support.hpe.com/hpsc/doc/public/display?docId=emr_na-c04565693
Advisory: HPE Moonshot - Error Messages Indicating Interrupt Failures in dmesg May Be Displayed After Successful Deployment of Red Hat Enterprise Linux 6 on a ProLiant m710x Server Cartridge
https://support.hpe.com/hpsc/doc/public/display?docId=emr_na-a00019143en_us
RHEL6: "dmar: DRHD: handling fault status reg 2" messages are printed by the kdump kernel
https://access.redhat.com/solutions/1480103
