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

Re: [Xen-devel] [PATCH 1/1] keep iommu disabled until iommu_setup is called

>>> On 19.10.12 at 02:35, Ronny Hegewald <ronny.hegewald@xxxxxxxxx> wrote:
>>  The concept is plausible, but I'm not convinced that there aren't
>> any users of iommu_enabled that actually depend on it having
>> its final value after command line parsing, yet before
>> iommu_setup() gets called. 
> But that is currently only true for systems with a iommu. For the ones 
> without 
> one have iommu_enabled=1 first, until iommu_setup() is called and 
> iommu_enabled is set to 0 because of the not detectable iommu.

That's a valid argument.

> So, with or without my patch, a similiar problem seems to be present, just 
> for different systems.
>> Did you fully audit all users?
> No, because there is just no way i can make a meaningfull judgement how the 
> iommu code interdepends with all the other parts that check iommu_enabled. 

I did a brief check, and it seems to be okay. I guess we'll have to
deal with eventual fallout then.

>> As to the patch itself, ...
> I will send a 2. version which addresses your comments as soon its clear if 
> the approach in the patch is the right one to fix the crash.

Please do.


Xen-devel mailing list



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