linux-kernel - Re: Kernel panic with intel

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20110812160846.29fd9149@laptop.anywhere.org>
Date:	Fri, 12 Aug 2011 16:08:46 +0200
From:	"Udo A. Steinberg" <udo@...ervisor.org>
To:	Dave Airlie <airlied@...il.com>
Cc:	Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
	iommu@...ts.linux-foundation.org,
	David Woodhouse <dwmw2@...radead.org>
Subject: Re: Kernel panic with intel_iommu=on

On Fri, 12 Aug 2011 23:51:47 +1000 Dave Airlie (DA) wrote:

DA> Does intel_iommu=igfx_off work as well?
DA> 
DA> I've been seeing a few of these with Ironlake chipsets, the reason
DA> firefox does it is probably because it causes the 3D driver to be
DA> used.

With intel_iommu=igfx_off it works as well. Looks like that keeps IOMMU 1
disabled. So the bug seems to be in the register-based invalidation of mappings
in the DMAR engine that remaps igfx. And it must be something that firefox 5.0
doesn't trigger, whereas 6.0 does.

For completeness, here's the relevant dmesg output with intel_iommu=igfx_off.

DMAR: Host address width 36
DMAR: DRHD base: 0x000000feb00000 flags: 0x0
IOMMU 0: reg_base_addr feb00000 ver 1:0 cap c9008020a30270 ecap 1000
DMAR: DRHD base: 0x000000feb01000 flags: 0x0
IOMMU 1: reg_base_addr feb01000 ver 1:0 cap c0000020230270 ecap 1000
DMAR: DRHD base: 0x000000feb03000 flags: 0x1
IOMMU 2: reg_base_addr feb03000 ver 1:0 cap c9008020230270 ecap 1000
DMAR: RMRR base: 0x000000000e0000 end: 0x000000000effff
DMAR: RMRR base: 0x0000003e600000 end: 0x0000003effffff
DMAR: No ATSR found
IOMMU 0 0xfeb00000: using Register based invalidation
IOMMU 2 0xfeb03000: using Register based invalidation
IOMMU: Setting RMRR:
IOMMU: Setting identity map for device 0000:00:1d.0 [0xe0000 - 0xeffff]
IOMMU: Setting identity map for device 0000:00:1d.1 [0xe0000 - 0xeffff]
IOMMU: Setting identity map for device 0000:00:1d.2 [0xe0000 - 0xeffff]
IOMMU: Setting identity map for device 0000:00:1d.7 [0xe0000 - 0xeffff]
IOMMU: Setting identity map for device 0000:00:1a.0 [0xe0000 - 0xeffff]
IOMMU: Setting identity map for device 0000:00:1a.1 [0xe0000 - 0xeffff]
IOMMU: Setting identity map for device 0000:00:1a.2 [0xe0000 - 0xeffff]
IOMMU: Setting identity map for device 0000:00:1a.7 [0xe0000 - 0xeffff]
IOMMU: Prepare 0-16MiB unity mapping for LPC
IOMMU: Setting identity map for device 0000:00:1f.0 [0x0 - 0xffffff]
PCI-DMA: Intel(R) Virtualization Technology for Directed I/O

Cheers,

	- Udo

Download attachment "signature.asc" of type "application/pgp-signature" (199 bytes)