Re: [Xen-devel] abstract model of IOMMU unmaping/mapping failures

[George Dunlap <george.dunlap@xxxxxxxxxx>]

On 31/03/16 10:06, Xu, Quan wrote:
> All,
> 
> Here is a summary of my investigation of the abstract model:
> 
> Below policies are adopted when deciding whether to rollback a callchain:
> 
> 1. Domain will be crashed immediately within iommu_{,un}map_page, treated as 
> a fatal error (with the exception of the hardware one). Whether to rollback 
> depends on the need of hardware domain;
> 
> 2. For hardware domain, roll back on a best effort basis. When rollback is 
> not feasible (in early initialization phase or trade-off of complexity), at 
> least unmap upon maps error and then throw out error message;
> 
> Below are a detail analysis of all existing callers on IOMMU interfaces (8-11 
> needs more discussions):

Hey Quan,

I only reviewed the p2m-related ones (5-10), but I agree with your and
Jan's conclusions.  Thanks for doing this legwork.

 -George


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel