[Dev] message during boot on Libreboot system: "DMAR: Failed to map dmar2"

[Bruno Dantas]

Hello, Parabola hackers! This is probably more of a Libreboot question, but I figure enough of you are using Parabola + Libreboot that at least a handful of you have also seen this error.

I am running Parabola on a Libreboot T400 that I purchased from Ministry of Freedom. I have not made any changes to Libreboot.

Everytime I boot into my Parabola installation, I see this message: "DMAR: Failed to map dmar2". I booted into Parabola's "MATE desktop ISO" just to see what would happen, and the same message scrolls by during the boot process. Once Parabola is fully booted and I'm at the desktop, everything seems to be working fine.

Here are the relevant lines from dmesg:

bruno at T400 ~ $ dmesg | grep -i dmar
[    0.000000] ACPI: DMAR 0x00000000BDAB9F80 0000A8 (v01 CORE   COREBOOT 00000000 CORE 00000000)
[    0.000000] WARNING: CPU: 0 PID: 0 at drivers/iommu/dmar.c:829 warn_invalid_dmar+0x79/0x90()
[    0.000000] Your BIOS is broken; DMAR reported at address fed92000 returns all ones!
[    0.000000]  [<ffffffff813bcb69>] warn_invalid_dmar+0x79/0x90
[    0.000000]  [<ffffffff81568105>] dmar_validate_one_drhd+0xa5/0xe0
[    0.000000]  [<ffffffff813bcd12>] dmar_walk_remapping_entries+0x82/0x190
[    0.030126] DMAR: Host address width 40
[    0.030131] DMAR: DRHD base: 0x000000fed90000 flags: 0x0
[    0.030140] DMAR: dmar0: reg_base_addr fed90000 ver 1:0 cap c9008020e30260 ecap 1000
[    0.030142] DMAR: DRHD base: 0x000000fed91000 flags: 0x0
[    0.030149] DMAR: dmar1: reg_base_addr fed91000 ver 1:0 cap c0000020630260 ecap 1000
[    0.030151] DMAR: DRHD base: 0x000000fed92000 flags: 0x0
[    0.030163] DMAR: Failed to map dmar2
[    0.030169] DMAR: Parse DMAR table failure.
[    0.134662] DMAR: Forcing write-buffer flush capability
[    0.134662] DMAR: Disabling IOMMU for graphics on this chipset

What does this DMAR failure mean? Is it fixable? If not, is there a way to make the error stop showing up during boot? I read somewhere that adding iommu=soft to the kernel boot parameters makes the error go away, but that didn't have any effect.

Thanks,
Bruno