Kernel Bug Tracker – Bug 37742
iwlagn with IOMMU enabled causes DMA Write faults
Last modified: 2012-04-01 10:31:48 UTC
Created attachment 62492 [details]
My wifi device causes DMA faults during initialization and association with an AP. This is on a ThinkPad X201s with the 6000AGN wifi (see below). IOMMU is enabled via intel_iommu=igfx_off commandline option. As far as I can tell the wifi works, although it might take longer to associate to an AP than without the IOMMU enabled.
(complete log is attached)
[ 62.238298] DRHD: handling fault status reg 2
[ 62.238305] DMAR:[DMA Write] Request device [02:00.0] fault addr ffe82000
[ 62.238307] DMAR:[fault reason 05] PTE Write access is not set
[ 62.642904] DRHD: handling fault status reg 2
[ 62.642912] DMAR:[DMA Write] Request device [02:00.0] fault addr ffe30000
[ 62.642914] DMAR:[fault reason 05] PTE Write access is not set
[ 64.006425] wlan0: direct probe to 08:17:35:82:b0:2d (try 1/3)
[ 64.265438] DRHD: handling fault status reg 2
[ 64.265445] DMAR:[DMA Write] Request device [02:00.0] fault addr ffea0000
[ 64.265447] DMAR:[fault reason 05] PTE Write access is not set
02:00.0 Network controller: Intel Corporation WiFi Link 6000 Series (rev 35)
Subsystem: Intel Corporation WiFi Link 6000 Series 3x3 6000AGN
Flags: bus master, fast devsel, latency 0, IRQ 45
Memory at f2400000 (64-bit, non-prefetchable) [size=8K]
Capabilities: [c8] Power Management version 3
Capabilities: [d0] MSI: Enable+ Count=1/1 Maskable- 64bit+
Capabilities: [e0] Express Endpoint, MSI 00
Capabilities:  Advanced Error Reporting
Capabilities:  Device Serial Number 00-24-d7-ff-ff-5f-f1-64
Kernel driver in use: iwlagn
Kernel modules: iwlagn
Does it still happen on new kernel? If so, can you compile driver with CONFIG_IWLWIFI_DEBUG and load module with gather logs as described in
(use iwlwifi module instead of iwl4965 of course) Then we will see where exactly DMAR problems happen.
Hmm Wey @ intel is CCed here and ignore that bug since more that half a year, that's kinda disappointing.
(In reply to comment #2)
> Hmm Wey @ intel is CCed here and ignore that bug since more that half a year,
> that's kinda disappointing.
I forgot bugzilla was down, shame on me, sorry Wey!
no problem :-), I am also very interest on root cause the problem here.
Can't reproduce this with 3.2.5 right now. I'll leave the IOMMU enabled for now and report back if it pops up again.
(Did I say that VT-d support on laptops rocks? Thanks Intel!)
This problem is fixed.