Re: [PATCH 2/4] x86,xen: introducex86_init.mapping.pagetable_reserve

[Stefano Stabellini]

On Wed, 13 Apr 2011, H. Peter Anvin wrote:
> On 04/13/2011 03:35 AM, Stefano Stabellini wrote:
> > Yes, that would be ideal, but we cannot do that because we don't know
> > exactly where is pgt_buf_end before allocating the pagetable pages and
> > the pagetable pages need to be marked RO before being hooked into the
> > pagetable. This is why we mark the whole range RO and after the
> > pagetable allocation when we know for sure where is pgt_buf_end we
> > modify the range pgt_buf_end-pgt_buf_top to RW.
> 
> The hell?  You have to fill the pages before you hook them into the page
> tables anyway (this means writing!) and then you have to mark them RO as
> you add them to the page tables... anything else doesn't make any sense
> at all.

Right.
The problem is that at some point init_memory_mapping is going reach the
pagetable pages area and map those pages too (I don't mean hooking the
pagetable pages in the pagetable, I mean mapping them as normal memory
that falls in the range of addresses passed to init_memory_mapping as
argument).
Some of those pages are already pagetable pages (they are in the range
pgt_buf_start-pgt_buf_end) therefore they are going to be mapped RO and
everything is fine.
Some of these pages are not pagetable pages yet (they fall in the range
pgt_buf_end-pgt_buf_top; for example the page at pgt_buf_end) so they
are going to be mapped RW.  When these pages become pagetable pages and
are hooked into the pagetable, xen will find that the guest has already
a RW mapping of them somewhere and fail the operation.

In order to fix the issue I could mark all the pages in the entire range
pgt_buf_start-pgt_buf_top as RO, but then once the pagetable allocation
is completed only the range pgt_buf_start-pgt_buf_end is reserved by
init_memory_mapping therefore the kernel is going to crash as soon as
one of the pages in the range pgt_buf_end-pgt_buf_top is reused.


Initially I suggested to add two hooks: one to allocate the pagetable
pages memory and one to reserve the pagetable pages memory after the
allocation:

http://marc.info/?l=linux-kernel&m=130141955626268

Following Yinghai's suggestion I removed the first hook (currently
unnecessary because we would use the same implementation on native and
on xen) and modified the second one, that became
x86_init.mapping.pagetable_reserve.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/