Re: [gentoo-user] Any fix yet for PTE Read access is not set error?

[Bill Kenworthy]

On 04/02/17 03:40, Mick wrote:
> On Friday 03 Feb 2017 09:03:10 Peter Humphrey wrote:
>> On Thursday 02 Feb 2017 05:12:17 Dale wrote:
>>> Peter Humphrey wrote:
>>>> Hello list,
>>>>
>>>> I was looking through dmesg this morning and found 1231 instances (!) of
>>>> this error:
>>>>
>>>> DMAR: DMAR:[DMA Read] Request device [01:00.0] fault addr f400156000
>>>> DMAR: [fault reason 06] PTE Read access is not set
>>>>
>>>> This is an Intel i7 box with an NVMe SSD and an AMD/ATI Amethyst XT
>>>> Radeon R9 M295X display card. The kernel is gentoo-sources-4.4.39. I've
>>>> tried setting intel_iommu=on or =pt as kernel parameter to no effect. I
>>>> don't want to set it =off because I've read that it slows the machine
>>>> dramatically.
>>>>
>>>> Googling has shown nothing, other than playing with that kernel
>>>> parameter. Does anyone here know what the problem is?
>>>
>>> I'm not sure if it applies but I found this:
>>>
>>> https://forums.gentoo.org/viewtopic-p-7904584.html#7904584
>>>
>>> It should scroll down a bit to the post.  The last comment seems to have
>>> a solution.
>>>
>>> Hope that helps.  If not, maybe someone else has a better idea.
>>
>> Thanks, but no, it makes no difference here.
>
> Peter, I recall a similar problem shown in the logs of an i7.   and all I did
> was to enable IOMMU in the kernel, just as a message in the logs recommended.
> From memory I built in the kernel something like CONFIG_INTEL_IOMMU_DEFAULT_ON
> as well as some Intel related IOMMU options. I did not pass any options on the
> kernel line at boot time.  The error messages stopped as a result.
>

This my kernel args for a MS surface 4 - acpi_osi=Linux
i915.semaphores=1 intel_iommu=off pcie_port_pm=off

stops it.

BillK