Re: [PATCH 5/5] This patch extends the VT-d driver to support KVM

["Yang, Sheng" <sheng.yang@intel.com>]

On Tuesday 29 July 2008 00:26:28 Ben-Ami Yassour wrote:
> [Ben: fixed memory pinning]
>
>  	kvm_free_physmem_slot(&old, &new);
> +
> +	/* map the pages in iommu page table */
> +	if (intel_iommu_found())
> +		kvm_iommu_map_pages(kvm, base_gfn, npages);
> +

Realize one serious problem here: this line pinned all memory 
regardless of if VT-d device have been assigned. That's means if VT-d 
engine is there(even without assigned device), KVM would pin all 
memory, then disable swapping capability. It's very undesired... I 
don't know if Avi aware of that.

I think we should check assigned device list here to avoid this kind 
of thing happen. Also, we may need a lock here to prevent racy with 
assign_device().

--
regards
Yang, Sheng