[Xen-devel] Changeset 20209 causes an issue in xen_in_range()

["Cui, Dexuan" <dexuan.cui@xxxxxxxxx>]

Currently PERCPU_SIZE is 2 4K-pages and only 1 page is used actually.

>From xen 20209 on, the second unused page is freed and returned to domheap in 
>debug=n build (MEMORY_GUARD is not defined):
percpu_free_unused_areas() -> free_xen_data() -> init_xenheap_pages().
Later the returned pages could be allocated to dom0 and dom0 could use them as 
DMA buffer.

However, in iommu_set_dom0_mapping(), xen_in_range() is still True for the 
freed pages above, so devices in Dom0 can meet with DMA fault.

We may have 2 options:
1) enhance xen_in_range() and check the freed pages. This means we should refer 
to the logic of percpu_free_unused_areas().
2) don't free the unused per_cpu pages: we will waste NR_CPUS pages. e.g., if 
NR_CPUS is 256, we'll waste 1M memory. Looks this is not a big issue?

I personally perfer option 2 as it's simple :-)
And actually long ago we also had frame_table checked in xen_in_range(), but  
gave up the checking later as it's not easy to implement the exact checking.

Any suggestions?

PS,  in percpu_free_unused_areas():
  data_size = (data_size + PAGE_SIZE + 1) & PAGE_MASK;
here "PAGE_SIZE + 1" should be "PAGE_SIZE - 1"?
 
Thanks,
-- Dexuan
 

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel