[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH RFC 2/2] AMD IOMMU: allow command line overrides for broken IVRS tables

Tuesday, August 27, 2013, 4:11:51 PM, you wrote:

>>>> On 27.08.13 at 14:04, "Jan Beulich" <JBeulich@xxxxxxxx> wrote:
>>>>> On 27.08.13 at 13:09, Sander Eikelenboom <linux@xxxxxxxxxxxxxx> wrote:
>>>> Would the syntax be the same as linux ?
>>>> f.e. ivrs_ioapic[6]=00:14.0 for my case ?
>> Yes.
>>>> I'm asking because using it on the command line seems to enable the iommu 
>>> fine (which it shouldn't without the override), but i don't seem to see the
>>>> "IVHD: Command line override present for IO-APIC %#x" in my xl dmesg... 
>>> Hrrrmm could it be due to the fact that in my case it actually shouldn't 
>>> override a present IVRS entry, but add a extra one ?
>> Yes. Does Linux print something even in that case?

> Just looked at their code once more, and I can't see how they
> would print something in that case.

> Jan

Ah yeah that was smart from me :-)

Hmm looking at the code, could it be a implementation difference with the linux 
variant ?

Linux seems to first make a list by looping through the special devices, 
skipping what will be overridden.
And after that add the mappings parsed from the command line.
It seems to use that list to for walking the parse_ivhd_special function 
mapping them.

This allows for not only overriding but also adding missing entries.

Where as your implementation seems to only walk the list present from bios and 
override a specific entry when it's mapping them in parse_ivhd_special ?

Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.