Project

General

Profile

Actions

Bug #192

open

KCMA-D8 iommu INVALID_DEVICE_REQUEST events

Added by Sean Rhone almost 6 years ago. Updated about 5 years ago.

Status:
New
Priority:
Normal
Assignee:
-
Category:
os_handoff
Target version:
-
Start date:
01/20/2019
Due date:
% Done:

0%

Estimated time:
Affected versions:
Needs backport to:
Affected hardware:
Affected OS:

Description

In dmesg, I get the following iommu event logs about INVALID_DEVICE_REQUEST:

[    3.651111] iommu ivhd0: Event logged [INVALID_DEVICE_REQUEST device=00:02.0 pasid=0x00000 address=0xfdf8210300 flags=0x0a00]
[    6.075105] iommu ivhd0: Event logged [INVALID_DEVICE_REQUEST device=00:04.0 pasid=0x00000 address=0xfdf8210400 flags=0x0a00]
[    6.075311] iommu ivhd0: Event logged [INVALID_DEVICE_REQUEST device=00:09.0 pasid=0x00000 address=0xfdf8210500 flags=0x0a00]
[    6.075492] iommu ivhd0: Event logged [INVALID_DEVICE_REQUEST device=00:0a.0 pasid=0x00000 address=0xfdf8210600 flags=0x0a00]
[    6.075665] iommu ivhd0: Event logged [INVALID_DEVICE_REQUEST device=00:0b.0 pasid=0x00000 address=0xfdf8210700 flags=0x0a00]

The 4 devices in question are (lspci):

00:02.0 PCI bridge: Advanced Micro Devices, Inc. [AMD/ATI] RD890/RD9x0/RX980 PCI to PCI bridge (PCI Express GFX port 0)
00:04.0 PCI bridge: Advanced Micro Devices, Inc. [AMD/ATI] RD890/RD9x0/RX980 PCI to PCI bridge (PCI Express GPP Port 0)
00:09.0 PCI bridge: Advanced Micro Devices, Inc. [AMD/ATI] RD890/RD9x0/RX980 PCI to PCI bridge (PCI Express GPP Port 4)
00:0a.0 PCI bridge: Advanced Micro Devices, Inc. [AMD/ATI] RD890/RD9x0/RX980 PCI to PCI bridge (PCI Express GPP Port 5)
00:0b.0 PCI bridge: Advanced Micro Devices, Inc. [AMD/ATI] RD890/RD990 PCI to PCI bridge (PCI Express GFX2 port 0)

In Coreboot's config, I noticed some option for an early PCI bridge; I wonder if that could be relevant in some way?

  • This happens with or without CPU microcode being included in the ROM
  • I haven't tested anything with iommu or virtualization to know if this is an actual problem
  • If I build Coreboot and disable IOMMU from nvram (along with having it automatically not used from the Linux OS), there's no errors related to those PCI devices
  • I attached a full dmesg and lspci dump

Files

dmesg.txt (74.6 KB) dmesg.txt dmesg Sean Rhone, 01/20/2019 08:36 PM
lspci.txt (3.51 KB) lspci.txt lspci Sean Rhone, 01/20/2019 08:36 PM
Actions #1

Updated by Henry-Nicolas Tourneur about 5 years ago

Hello Sean,

I'v exactly the same issue as described here with LibreBoot on a KCMA-D8, plus the other bug which you reported.
I do have some DMA errors with a PCI-e card which I believe are related to these IOMMU issues.

If not using virtualisation technologies, what would be the downside of disabling IOMMU as per your suggestion?
Any advise in order to progress this toward resolution?

Actions

Also available in: Atom PDF